Influences of the Powder Size and Process Parameters on the Quasi-Stability of Molten Pool Shape in Powder Bed Fusion-Laser Beam of Molybdenum

Influences of the Powder Size and Process Parameters on the Quasi-Stability of Molten Pool Shape in Powder Bed Fusion-Laser Beam of Molybdenum

몰리브덴 분말층 융합-레이저 빔의 용융 풀 형태의 준안정성에 대한 분말 크기 및 공정 매개변수의 영향

Abstract

Formation of a quasi-steady molten pool is one of the necessary conditions for achieving excellent quality in many laser processes. The influences of distribution characteristics of powder sizes on quasi-stability of the molten pool shape during single-track powder bed fusion-laser beam (PBF-LB) of molybdenum and the underlying mechanism were investigated.

The feasibility of improving quasi-stability of the molten pool shape by increasing the laser energy conduction effect and preheating was explored. Results show that an increase in the range of powder sizes does not significantly influence the average laser energy conduction effect in PBF-LB process. Whereas, it intensifies fluctuations of the transient laser energy conduction effect.

It also leads to fluctuations of the replenishment rate of metals, difficulty in formation of the quasi-steady molten pool, and increased probability of incomplete fusion and pores defects. As the laser power rises, the laser energy conduction effect increases, which improves the quasi-stability of the molten pool shape. When increasing the laser scanning speed, the laser energy conduction effect grows.

However, because the molten pool size reduces due to the decreased heat input, the replenishment rate of metals of the molten pool fluctuates more obviously and the quasi-stability of the molten pool shape gets worse. On the whole, the laser energy conduction effect in the PBF-LB process of Mo is low (20-40%). The main factor that affects quasi-stability of the molten pool shape is the amount of energy input per unit length of the scanning path, rather than the laser energy conduction effect.

Moreover, substrate preheating can not only enlarge the molten pool size, particularly the length, but also reduce non-uniformity and discontinuity of surface morphologies of clad metals and inhibit incomplete fusion and pores defects.

준안정 용융 풀의 형성은 많은 레이저 공정에서 우수한 품질을 달성하는 데 필요한 조건 중 하나입니다. 몰리브덴의 단일 트랙 분말층 융합 레이저 빔(PBF-LB) 동안 용융 풀 형태의 준안정성에 대한 분말 크기 분포 특성의 영향과 그 기본 메커니즘을 조사했습니다.

레이저 에너지 전도 효과와 예열을 증가시켜 용융 풀 형태의 준안정성을 향상시키는 타당성을 조사했습니다. 결과는 분말 크기 범위의 증가가 PBF-LB 공정의 평균 레이저 에너지 전도 효과에 큰 영향을 미치지 않음을 보여줍니다. 반면, 과도 레이저 에너지 전도 효과의 변동이 강화됩니다.

이는 또한 금속 보충 속도의 변동, 준안정 용융 풀 형성의 어려움, 불완전 융합 및 기공 결함 가능성 증가로 이어집니다. 레이저 출력이 증가함에 따라 레이저 에너지 전도 효과가 증가하여 용융 풀 모양의 준 안정성이 향상됩니다. 레이저 스캐닝 속도를 높이면 레이저 에너지 전도 효과가 커집니다.

그러나 열 입력 감소로 인해 용융 풀 크기가 줄어들기 때문에 용융 풀의 금속 보충 속도의 변동이 더욱 뚜렷해지고 용융 풀 형태의 준안정성이 악화됩니다.

전체적으로 Mo의 PBF-LB 공정에서 레이저 에너지 전도 효과는 낮다(20~40%). 용융 풀 형상의 준안정성에 영향을 미치는 주요 요인은 레이저 에너지 전도 효과보다는 스캐닝 경로의 단위 길이당 입력되는 에너지의 양입니다.

또한 기판 예열은 용융 풀 크기, 특히 길이를 확대할 수 있을 뿐만 아니라 클래드 금속 표면 형태의 불균일성과 불연속성을 줄이고 불완전한 융합 및 기공 결함을 억제합니다.

References

  1. M. Sharifitabar, F.O. Sadeq, and M.S. Afarani, Synthesis and Kinetic Study of Mo (Si, Al)2 Coatings on the Surface of Molybdenum Through Hot Dipping into a Commercial Al-12 wt.% Si Alloy Melt, Surf. Interfaces, 2021, 24, p 101044.Article CAS Google Scholar 
  2. Z. Zhang, X. Li, and H. Dong, Response of a Molybdenum Alloy to Plasma Nitriding, Int. J. Refract. Met. Hard Mater., 2018, 72, p 388–395.Article CAS Google Scholar 
  3. C. Tan, K. Zhou, M. Kuang, W. Ma, and T. Kuang, Microstructural Characterization and Properties of Selective Laser Melted Maraging Steel with Different Build Directions, Sci. Technol. Adv. Mater., 2018, 19(1), p 746–758.Article CAS Google Scholar 
  4. C. Tan, F. Weng, S. Sui, Y. Chew, and G. Bi, Progress and Perspectives in Laser Additive Manufacturing of Key Aeroengine Materials, Int. J. Mach. Tools Manuf, 2021, 170, p 103804.Article Google Scholar 
  5. S.A. Khairallah and A. Anderson, Mesoscopic Simulation Model of Selective Laser Melting of Stainless Steel Powder, J. Mater. Process. Technol., 2014, 214(11), p 2627–2636.Article CAS Google Scholar 
  6. S.A. Khairallah, A.T. Anderson, A. Rubenchik, and W.E. King, Laser Powder-Bed Fusion Additive Manufacturing: Physics of Complex Melt Flow and Formation Mechanisms of Pores, Spatter, and Denudation Zones, Acta Mater., 2016, 108, p 36–45.Article CAS ADS Google Scholar 
  7. K.Q. Le, C. Tang, and C.H. Wong, On the Study of Keyhole-Mode Melting in Selective Laser Melting Process, Int. J. Therm. Sci., 2019, 145, p 105992.Article Google Scholar 
  8. M. Bayat, A. Thanki, S. Mohanty, A. Witvrouw, S. Yang, J. Thorborg, N.S. Tiedje, and J.H. Hattel, Keyhole-Induced Porosities in Laser-Based Powder Bed Fusion (L-PBF) of Ti6Al4V: High-Fidelity Modelling and Experimental Validation, Addit. Manuf., 2019, 30, p 100835.CAS Google Scholar 
  9. B. Liu, G. Fang, L. Lei, and X. Yan, Predicting the Porosity Defects in Selective Laser Melting (SLM) by Molten Pool Geometry, Int. J. Mech. Sci., 2022, 228, p 107478.Article Google Scholar 
  10. W. Ge, J.Y.H. Fuh, and S.J. Na, Numerical Modelling of Keyhole Formation in Selective Laser Melting of Ti6Al4V, J. Manuf. Process., 2021, 62, p 646–654.Article Google Scholar 
  11. W. Ge, S. Han, S.J. Na, and J.Y.H. Fuh, Numerical Modelling of Surface Morphology in Selective Laser Melting, Comput. Mater. Sci., 2021, 186, p 110062.Article Google Scholar 
  12. Y.-C. Wu, C.-H. San, C.-H. Chang, H.-J. Lin, R. Marwan, S. Baba, and W.-S. Hwang, Numerical Modeling of Melt-Pool Behavior In Selective Laser Melting with Random Powder Distribution and Experimental Validation, J. Mater. Process. Technol., 2018, 254, p 72–78.Article Google Scholar 
  13. C. Tang, J.L. Tan, and C.H. Wong, A Numerical Investigation on the Physical Mechanisms of Single Track Defects in Selective Laser Melting, Int. J. Heat Mass Transf., 2018, 126, p 957–968.Article CAS Google Scholar 
  14. X. Zhou, X. Liu, D. Zhang, Z. Shen, and W. Liu, Balling Phenomena in Selective Laser Melted Tungsten, J. Mater. Process. Technol., 2015, 222, p 33–42.Article CAS Google Scholar 
  15. J.D.K. Monroy and J. Ciurana, Study of the Pore Formation on CoCrMo Alloys by Selective Laser Melting Manufacturing Process, Procedia Eng., 2013, 63, p 361–369.Article CAS Google Scholar 
  16. L. Kaserer, J. Braun, J. Stajkovic, K.H. Leitz, B. Tabernig, P. Singer, I. Letofsky-Papst, H. Kestler, and G. Leichtfried, Fully Dense and Crack Free Molybdenum Manufactured by Selective Laser Melting Through Alloying with Carbon, Int. J. Refract. Met. Hard Mater., 2019, 84, p 105000.Article CAS Google Scholar 
  17. T.B.T. Majumdar, E.M.C. Ribeiro, J.E. Frith, and N. Birbilis, Understanding the Effects of PBF Process Parameter Interplay on Ti-6Al-4V Surface Properties, PLoS ONE, 2019, 14, p e0221198.Article CAS PubMed PubMed Central Google Scholar 
  18. A.K.J.-R. Poulin, P. Terriault, and V. Brailovski, Long Fatigue Crack Propagation Behavior of Laser Powder Bed-Fused Inconel 625 with Intentionally- Seeded Porosity, Int. J. Fatigue, 2019, 127, p 144–156.Article CAS Google Scholar 
  19. P. Rebesan, M. Ballan, M. Bonesso, A. Campagnolo, S. Corradetti, R. Dima, C. Gennari, G.A. Longo, S. Mancin, M. Manzolaro, G. Meneghetti, A. Pepato, E. Visconti, and M. Vedani, Pure Molybdenum Manufactured by Laser Powder Bed Fusion: Thermal and Mechanical Characterization at Room and High Temperature, Addit. Manuf., 2021, 47, p 102277.CAS Google Scholar 
  20. D. Wang, C. Yu, J. Ma, W. Liu, and Z. Shen, Densification and Crack Suppression in Selective Laser Melting of Pure Molybdenum, Mater. Des., 2017, 129, p 44–52.Article CAS Google Scholar 
  21. K.-H. Leitz, P. Singer, A. Plankensteiner, B. Tabernig, H. Kestler, and L.S. Sigl, Multi-physical Simulation of Selective Laser Melting, Met. Powder Rep., 2017, 72, p 331–338.Article Google Scholar 
  22. D.G.J. Zhang, Y. Yang, H. Zhang, H. Chen, D. Dai, and K. Lin, Influence of Particle Size on Laser Absorption and Scanning Track Formation Mechanisms of Pure Tungsten Powder During Selective Laser Melting, Engineering, 2019, 5, p 736–745.Article CAS Google Scholar 
  23. L. Caprio, A.G. Demir, and B. Previtali, Influence of Pulsed and Continuous Wave Emission on Melting Efficiency in Selective Laser Melting, J. Mater. Process. Technol., 2019, 266, p 429–441.Article CAS Google Scholar 
  24. D. Gu, M. Xia, and D. Dai, On the Role of Powder Flow Behavior in Fluid Thermodynamics and Laser Processability of Ni-based Composites by Selective Laser Melting, Int. J. Mach. Tools Manuf, 2018, 137, p 67–78.Article Google Scholar 
  25. W.-I. Cho, S.-J. Na, C. Thomy, and F. Vollertsen, Numerical Simulation of Molten Pool Dynamics in High Power Disk Laser Welding, J. Mater. Process. Technol., 2012, 212(1), p 262–275.Article CAS Google Scholar 
  26. S.W. Han, J. Ahn, and S.J. Na, A Study on Ray Tracing Method for CFD Simulations of Laser Keyhole Welding: Progressive Search Method, Weld. World, 2016, 60, p 247–258.Article CAS Google Scholar 
  27. W. Ge, S. Han, Y. Fang, J. Cheon, and S.J. Na, Mechanism of Surface Morphology in Electron Beam Melting of Ti6Al4V Based on Computational Flow Patterns, Appl. Surf. Sci., 2017, 419, p 150–158.Article CAS ADS Google Scholar 
  28. W.-I. Cho, S.-J. Na, C. Thomy, and F. Vollertsen, Numerical Simulation of Molten Pool Dynamics in High Power Disk Laser Welding, J. Mater. Process. Technol., 2012, 212, p 262–275.Article CAS Google Scholar 
  29. W. Ma, J. Ning, L.-J. Zhang, and S.-J. Na, Regulation of Microstructures and Properties of Molybdenum-Silicon-Boron Alloy Subjected to Selective Laser Melting, J. Manuf. Process., 2021, 69, p 593–601.Article Google Scholar 
  30. S. Haeri, Y. Wang, O. Ghita, and J. Sun, Discrete Element Simulation and Experimental Study of Powder Spreading Process in Additive Manufacturing, Powder Technol., 2016, 306, p 45–54.Article Google Scholar 
  31. D. Yao, X. Liu, J. Wang, W. Fan, M. Li, H. Fu, H. Zhang, X. Yang, Q. Zou, and X. An, Numerical Insights on the Spreading of Practical 316 L Stainless Steel Powder in SLM Additive Manufacturing, Powder Technol., 2021, 390, p 197–208.Article CAS Google Scholar 
  32. S. Vock, B. Klöden, A. Kirchner, T. Weißgärber, and B. Kieback, Powders for Powder Bed Fusion: A Review, Prog. Addit. Manuf., 2019, 4, p 383–397.Article Google Scholar 
  33. X. Luo, C. Yang, Z.Q. Fu, L.H. Liu, H.Z. Lu, H.W. Ma, Z. Wang, D.D. Li, L.C. Zhang, and Y.Y. Li, Achieving Ultrahigh-Strength in Beta-Type Titanium Alloy by Controlling the Melt Pool Mode in Selective Laser Melting, Mater. Sci. Eng. A, 2021, 823, p 141731.Article CAS Google Scholar 
  34. J. Braun, L. Kaserer, J. Stajkovic, K.-H. Leitz, B. Tabernig, P. Singer, P. Leibenguth, C. Gspan, H. Kestler, and G. Leichtfried, Molybdenum and Tungsten Manufactured by Selective Laser Melting: Analysis of Defect Structure and Solidification Mechanisms, Int. J. Refract. Met. Hard Mater., 2019, 84, p 104999.Article CAS Google Scholar 
  35. L. Kaserera, J. Brauna, J. Stajkovica, K.-H. Leitzb, B. Tabernigb, P. Singerb, I. Letofsky-Papstc, H. Kestlerb, and G. Leichtfried, Fully Dense and Crack Free Molybdenum Manufactured by Selective Laser Melting Through Alloying with Carbon, Int. J. Refract Metal Hard Mater., 2019, 84, p 105000.Article Google Scholar 

Lab-on-a-Chip 시스템의 혈류 역학에 대한 검토: 엔지니어링 관점

Review on Blood Flow Dynamics in Lab-on-a-Chip Systems: An Engineering Perspective

  • Bin-Jie Lai
  • Li-Tao Zhu
  • Zhe Chen*
  • Bo Ouyang*
  • , and 
  • Zheng-Hong Luo*

Abstract

다양한 수송 메커니즘 하에서, “LOC(lab-on-a-chip)” 시스템에서 유동 전단 속도 조건과 밀접한 관련이 있는 혈류 역학은 다양한 수송 현상을 초래하는 것으로 밝혀졌습니다.

본 연구는 적혈구의 동적 혈액 점도 및 탄성 거동과 같은 점탄성 특성의 역할을 통해 LOC 시스템의 혈류 패턴을 조사합니다. 모세관 및 전기삼투압의 주요 매개변수를 통해 LOC 시스템의 혈액 수송 현상에 대한 연구는 실험적, 이론적 및 수많은 수치적 접근 방식을 통해 제공됩니다.

전기 삼투압 점탄성 흐름에 의해 유발되는 교란은 특히 향후 연구 기회를 위해 혈액 및 기타 점탄성 유체를 취급하는 LOC 장치의 혼합 및 분리 기능 향상에 논의되고 적용됩니다. 또한, 본 연구는 보다 정확하고 단순화된 혈류 모델에 대한 요구와 전기역학 효과 하에서 점탄성 유체 흐름에 대한 수치 연구에 대한 강조와 같은 LOC 시스템 하에서 혈류 역학의 수치 모델링의 문제를 식별합니다.

전기역학 현상을 연구하는 동안 제타 전위 조건에 대한 보다 실용적인 가정도 강조됩니다. 본 연구는 모세관 및 전기삼투압에 의해 구동되는 미세유체 시스템의 혈류 역학에 대한 포괄적이고 학제적인 관점을 제공하는 것을 목표로 한다.

KEYWORDS: 

1. Introduction

1.1. Microfluidic Flow in Lab-on-a-Chip (LOC) Systems

Over the past several decades, the ability to control and utilize fluid flow patterns at microscales has gained considerable interest across a myriad of scientific and engineering disciplines, leading to growing interest in scientific research of microfluidics. 

(1) Microfluidics, an interdisciplinary field that straddles physics, engineering, and biotechnology, is dedicated to the behavior, precise control, and manipulation of fluids geometrically constrained to a small, typically submillimeter, scale. 

(2) The engineering community has increasingly focused on microfluidics, exploring different driving forces to enhance working fluid transport, with the aim of accurately and efficiently describing, controlling, designing, and applying microfluidic flow principles and transport phenomena, particularly for miniaturized applications. 

(3) This attention has chiefly been fueled by the potential to revolutionize diagnostic and therapeutic techniques in the biomedical and pharmaceutical sectorsUnder various driving forces in microfluidic flows, intriguing transport phenomena have bolstered confidence in sustainable and efficient applications in fields such as pharmaceutical, biochemical, and environmental science. The “lab-on-a-chip” (LOC) system harnesses microfluidic flow to enable fluid processing and the execution of laboratory tasks on a chip-sized scale. LOC systems have played a vital role in the miniaturization of laboratory operations such as mixing, chemical reaction, separation, flow control, and detection on small devices, where a wide variety of fluids is adapted. Biological fluid flow like blood and other viscoelastic fluids are notably studied among the many working fluids commonly utilized by LOC systems, owing to the optimization in small fluid sample volumed, rapid response times, precise control, and easy manipulation of flow patterns offered by the system under various driving forces. 

(4)The driving forces in blood flow can be categorized as passive or active transport mechanisms and, in some cases, both. Under various transport mechanisms, the unique design of microchannels enables different functionalities in driving, mixing, separating, and diagnosing blood and drug delivery in the blood. 

(5) Understanding and manipulating these driving forces are crucial for optimizing the performance of a LOC system. Such knowledge presents the opportunity to achieve higher efficiency and reliability in addressing cellular level challenges in medical diagnostics, forensic studies, cancer detection, and other fundamental research areas, for applications of point-of-care (POC) devices. 

(6)

1.2. Engineering Approach of Microfluidic Transport Phenomena in LOC Systems

Different transport mechanisms exhibit unique properties at submillimeter length scales in microfluidic devices, leading to significant transport phenomena that differ from those of macroscale flows. An in-depth understanding of these unique transport phenomena under microfluidic systems is often required in fluidic mechanics to fully harness the potential functionality of a LOC system to obtain systematically designed and precisely controlled transport of microfluids under their respective driving force. Fluid mechanics is considered a vital component in chemical engineering, enabling the analysis of fluid behaviors in various unit designs, ranging from large-scale reactors to separation units. Transport phenomena in fluid mechanics provide a conceptual framework for analytically and descriptively explaining why and how experimental results and physiological phenomena occur. The Navier–Stokes (N–S) equation, along with other governing equations, is often adapted to accurately describe fluid dynamics by accounting for pressure, surface properties, velocity, and temperature variations over space and time. In addition, limiting factors and nonidealities for these governing equations should be considered to impose corrections for empirical consistency before physical models are assembled for more accurate controls and efficiency. Microfluidic flow systems often deviate from ideal conditions, requiring adjustments to the standard governing equations. These deviations could arise from factors such as viscous effects, surface interactions, and non-Newtonian fluid properties from different microfluid types and geometrical layouts of microchannels. Addressing these nonidealities supports the refining of theoretical models and prediction accuracy for microfluidic flow behaviors.

The analytical calculation of coupled nonlinear governing equations, which describes the material and energy balances of systems under ideal conditions, often requires considerable computational efforts. However, advancements in computation capabilities, cost reduction, and improved accuracy have made numerical simulations using different numerical and modeling methods a powerful tool for effectively solving these complex coupled equations and modeling various transport phenomena. Computational fluid dynamics (CFD) is a numerical technique used to investigate the spatial and temporal distribution of various flow parameters. It serves as a critical approach to provide insights and reasoning for decision-making regarding the optimal designs involving fluid dynamics, even prior to complex physical model prototyping and experimental procedures. The integration of experimental data, theoretical analysis, and reliable numerical simulations from CFD enables systematic variation of analytical parameters through quantitative analysis, where adjustment to delivery of blood flow and other working fluids in LOC systems can be achieved.

Numerical methods such as the Finite-Difference Method (FDM), Finite-Element-Method (FEM), and Finite-Volume Method (FVM) are heavily employed in CFD and offer diverse approaches to achieve discretization of Eulerian flow equations through filling a mesh of the flow domain. A more in-depth review of numerical methods in CFD and its application for blood flow simulation is provided in Section 2.2.2.

1.3. Scope of the Review

In this Review, we explore and characterize the blood flow phenomena within the LOC systems, utilizing both physiological and engineering modeling approaches. Similar approaches will be taken to discuss capillary-driven flow and electric-osmotic flow (EOF) under electrokinetic phenomena as a passive and active transport scheme, respectively, for blood transport in LOC systems. Such an analysis aims to bridge the gap between physical (experimental) and engineering (analytical) perspectives in studying and manipulating blood flow delivery by different driving forces in LOC systems. Moreover, the Review hopes to benefit the interests of not only blood flow control in LOC devices but also the transport of viscoelastic fluids, which are less studied in the literature compared to that of Newtonian fluids, in LOC systems.

Section 2 examines the complex interplay between viscoelastic properties of blood and blood flow patterns under shear flow in LOC systems, while engineering numerical modeling approaches for blood flow are presented for assistance. Sections 3 and 4 look into the theoretical principles, numerical governing equations, and modeling methodologies for capillary driven flow and EOF in LOC systems as well as their impact on blood flow dynamics through the quantification of key parameters of the two driving forces. Section 5 concludes the characterized blood flow transport processes in LOC systems under these two forces. Additionally, prospective areas of research in improving the functionality of LOC devices employing blood and other viscoelastic fluids and potentially justifying mechanisms underlying microfluidic flow patterns outside of LOC systems are presented. Finally, the challenges encountered in the numerical studies of blood flow under LOC systems are acknowledged, paving the way for further research.

2. Blood Flow Phenomena

ARTICLE SECTIONS

Jump To


2.1. Physiological Blood Flow Behavior

Blood, an essential physiological fluid in the human body, serves the vital role of transporting oxygen and nutrients throughout the body. Additionally, blood is responsible for suspending various blood cells including erythrocytes (red blood cells or RBCs), leukocytes (white blood cells), and thrombocytes (blood platelets) in a plasma medium.Among the cells mentioned above, red blood cells (RBCs) comprise approximately 40–45% of the volume of healthy blood. 

(7) An RBC possesses an inherent elastic property with a biconcave shape of an average diameter of 8 μm and a thickness of 2 μm. This biconcave shape maximizes the surface-to-volume ratio, allowing RBCs to endure significant distortion while maintaining their functionality. 

(8,9) Additionally, the biconcave shape optimizes gas exchange, facilitating efficient uptake of oxygen due to the increased surface area. The inherent elasticity of RBCs allows them to undergo substantial distortion from their original biconcave shape and exhibits high flexibility, particularly in narrow channels.RBC deformability enables the cell to deform from a biconcave shape to a parachute-like configuration, despite minor differences in RBC shape dynamics under shear flow between initial cell locations. As shown in Figure 1(a), RBCs initiating with different resting shapes and orientations displaying display a similar deformation pattern 

(10) in terms of its shape. Shear flow induces an inward bending of the cell at the rear position of the rim to the final bending position, 

(11) resulting in an alignment toward the same position of the flow direction.

Figure 1. Images of varying deformation of RBCs and different dynamic blood flow behaviors. (a) The deforming shape behavior of RBCs at four different initiating positions under the same experimental conditions of a flow from left to right, (10) (b) RBC aggregation, (13) (c) CFL region. (18) Reproduced with permission from ref (10). Copyright 2011 Elsevier. Reproduced with permission from ref (13). Copyright 2022 The Authors, under the terms of the Creative Commons (CC BY 4.0) License https://creativecommons.org/licenses/by/4.0/. Reproduced with permission from ref (18). Copyright 2019 Elsevier.

The flexible property of RBCs enables them to navigate through narrow capillaries and traverse a complex network of blood vessels. The deformability of RBCs depends on various factors, including the channel geometry, RBC concentration, and the elastic properties of the RBC membrane. 

(12) Both flexibility and deformability are vital in the process of oxygen exchange among blood and tissues throughout the body, allowing cells to flow in vessels even smaller than the original cell size prior to deforming.As RBCs serve as major components in blood, their collective dynamics also hugely affect blood rheology. RBCs exhibit an aggregation phenomenon due to cell to cell interactions, such as adhesion forces, among populated cells, inducing unique blood flow patterns and rheological behaviors in microfluidic systems. For blood flow in large vessels between a diameter of 1 and 3 cm, where shear rates are not high, a constant viscosity and Newtonian behavior for blood can be assumed. However, under low shear rate conditions (0.1 s

–1) in smaller vessels such as the arteries and venules, which are within a diameter of 0.2 mm to 1 cm, blood exhibits non-Newtonian properties, such as shear-thinning viscosity and viscoelasticity due to RBC aggregation and deformability. The nonlinear viscoelastic property of blood gives rise to a complex relationship between viscosity and shear rate, primarily influenced by the highly elastic behavior of RBCs. A wide range of research on the transient behavior of the RBC shape and aggregation characteristics under varied flow circumstances has been conducted, aiming to obtain a better understanding of the interaction between blood flow shear forces from confined flows.

For a better understanding of the unique blood flow structures and rheological behaviors in microfluidic systems, some blood flow patterns are introduced in the following section.

2.1.1. RBC Aggregation

RBC aggregation is a vital phenomenon to be considered when designing LOC devices due to its impact on the viscosity of the bulk flow. Under conditions of low shear rate, such as in stagnant or low flow rate regions, RBCs tend to aggregate, forming structures known as rouleaux, resembling stacks of coins as shown in Figure 1(b). 

(13) The aggregation of RBCs increases the viscosity at the aggregated region, 

(14) hence slowing down the overall blood flow. However, when exposed to high shear rates, RBC aggregates disaggregate. As shear rates continue to increase, RBCs tend to deform, elongating and aligning themselves with the direction of the flow. 

(15) Such a dynamic shift in behavior from the cells in response to the shear rate forms the basis of the viscoelastic properties observed in whole blood. In essence, the viscosity of the blood varies according to the shear rate conditions, which are related to the velocity gradient of the system. It is significant to take the intricate relationship between shear rate conditions and the change of blood viscosity due to RBC aggregation into account since various flow driving conditions may induce varied effects on the degree of aggregation.

2.1.2. Fåhræus-Lindqvist Effect

The Fåhræus–Lindqvist (FL) effect describes the gradual decrease in the apparent viscosity of blood as the channel diameter decreases. 

(16) This effect is attributed to the migration of RBCs toward the central region in the microchannel, where the flow rate is higher, due to the presence of higher pressure and asymmetric distribution of shear forces. This migration of RBCs, typically observed at blood vessels less than 0.3 mm, toward the higher flow rate region contributes to the change in blood viscosity, which becomes dependent on the channel size. Simultaneously, the increase of the RBC concentration in the central region of the microchannel results in the formation of a less viscous region close to the microchannel wall. This region called the Cell-Free Layer (CFL), is primarily composed of plasma. 

(17) The combination of the FL effect and the following CFL formation provides a unique phenomenon that is often utilized in passive and active plasma separation mechanisms, involving branched and constriction channels for various applications in plasma separation using microfluidic systems.

2.1.3. Cell-Free Layer Formation

In microfluidic blood flow, RBCs form aggregates at the microchannel core and result in a region that is mostly devoid of RBCs near the microchannel walls, as shown in Figure 1(c). 

(18) The region is known as the cell-free layer (CFL). The CFL region is often known to possess a lower viscosity compared to other regions within the blood flow due to the lower viscosity value of plasma when compared to that of the aggregated RBCs. Therefore, a thicker CFL region composed of plasma correlates to a reduced apparent whole blood viscosity. 

(19) A thicker CFL region is often established following the RBC aggregation at the microchannel core under conditions of decreasing the tube diameter. Apart from the dependence on the RBC concentration in the microchannel core, the CFL thickness is also affected by the volume concentration of RBCs, or hematocrit, in whole blood, as well as the deformability of RBCs. Given the influence CFL thickness has on blood flow rheological parameters such as blood flow rate, which is strongly dependent on whole blood viscosity, investigating CFL thickness under shear flow is crucial for LOC systems accounting for blood flow.

2.1.4. Plasma Skimming in Bifurcation Networks

The uneven arrangement of RBCs in bifurcating microchannels, commonly termed skimming bifurcation, arises from the axial migration of RBCs within flowing streams. This uneven distribution contributes to variations in viscosity across differing sizes of bifurcating channels but offers a stabilizing effect. Notably, higher flow rates in microchannels are associated with increased hematocrit levels, resulting in higher viscosity compared with those with lower flow rates. Parametric investigations on bifurcation angle, 

(20) thickness of the CFL, 

(21) and RBC dynamics, including aggregation and deformation, 

(22) may alter the varying viscosity of blood and its flow behavior within microchannels.

2.2. Modeling on Blood Flow Dynamics

2.2.1. Blood Properties and Mathematical Models of Blood Rheology

Under different shear rate conditions in blood flow, the elastic characteristics and dynamic changes of the RBC induce a complex velocity and stress relationship, resulting in the incompatibility of blood flow characterization through standard presumptions of constant viscosity used for Newtonian fluid flow. Blood flow is categorized as a viscoelastic non-Newtonian fluid flow where constitutive equations governing this type of flow take into consideration the nonlinear viscometric properties of blood. To mathematically characterize the evolving blood viscosity and the relationship between the elasticity of RBC and the shear blood flow, respectively, across space and time of the system, a stress tensor (τ) defined by constitutive models is often coupled in the Navier–Stokes equation to account for the collective impact of the constant dynamic viscosity (η) and the elasticity from RBCs on blood flow.The dynamic viscosity of blood is heavily dependent on the shear stress applied to the cell and various parameters from the blood such as hematocrit value, plasma viscosity, mechanical properties of the RBC membrane, and red blood cell aggregation rate. The apparent blood viscosity is considered convenient for the characterization of the relationship between the evolving blood viscosity and shear rate, which can be defined by Casson’s law, as shown in eq 1.

𝜇=𝜏0𝛾˙+2𝜂𝜏0𝛾˙⎯⎯⎯⎯⎯⎯⎯√+𝜂�=�0�˙+2��0�˙+�

(1)where τ

0 is the yield stress–stress required to initiate blood flow motion, η is the Casson rheological constant, and γ̇ is the shear rate. The value of Casson’s law parameters under blood with normal hematocrit level can be defined as τ

0 = 0.0056 Pa and η = 0.0035 Pa·s. 

(23) With the known property of blood and Casson’s law parameters, an approximation can be made to the dynamic viscosity under various flow condition domains. The Power Law model is often employed to characterize the dynamic viscosity in relation to the shear rate, since precise solutions exist for specific geometries and flow circumstances, acting as a fundamental standard for definition. The Carreau and Carreau–Yasuda models can be advantageous over the Power Law model due to their ability to evaluate the dynamic viscosity at low to zero shear rate conditions. However, none of the above-mentioned models consider the memory or other elastic behavior of blood and its RBCs. Some other commonly used mathematical models and their constants for the non-Newtonian viscosity property characterization of blood are listed in Table 1 below. 

(24−26)Table 1. Comparison of Various Non-Newtonian Models for Blood Viscosity 

(24−26)

ModelNon-Newtonian ViscosityParameters
Power Law(2)n = 0.61, k = 0.42
Carreau(3)μ0 = 0.056 Pa·s, μ = 0.00345 Pa·s, λ = 3.1736 s, m = 2.406, a = 0.254
Walburn–Schneck(4)C1 = 0.000797 Pa·s, C2 = 0.0608 Pa·s, C3 = 0.00499, C4 = 14.585 g–1, TPMA = 25 g/L
Carreau–Yasuda(5)μ0 = 0.056 Pa·s, μ = 0.00345 Pa·s, λ = 1.902 s, n = 0.22, a = 1.25
Quemada(6)μp = 0.0012 Pa·s, k = 2.07, k0 = 4.33, γ̇c = 1.88 s–1

The blood rheology is commonly known to be influenced by two key physiological factors, namely, the hematocrit value (H

t) and the fibrinogen concentration (c

f), with an average value of 42% and 0.252 gd·L

–1, respectively. Particularly in low shear conditions, the presence of varying fibrinogen concentrations affects the tendency for aggregation and rouleaux formation, while the occurrence of aggregation is contingent upon specific levels of hematocrit. 

(27) The study from Apostolidis et al. 

(28) modifies the Casson model through emphasizing its reliance on hematocrit and fibrinogen concentration parameter values, owing to the extensive knowledge of the two physiological blood parameters.The viscoelastic response of blood is heavily dependent on the elasticity of the RBC, which is defined by the relationship between the deformation and stress relaxation from RBCs under a specific location of shear flow as a function of the velocity field. The stress tensor is usually characterized by constitutive equations such as the Upper-Convected Maxwell Model 

(29) and the Oldroyd-B model 

(30) to track the molecule effects under shear from different driving forces. The prominent non-Newtonian features, such as shear thinning and yield stress, have played a vital role in the characterization of blood rheology, particularly with respect to the evaluation of yield stress under low shear conditions. The nature of stress measurement in blood, typically on the order of 1 mPa, is challenging due to its low magnitude. The occurrence of the CFL complicates the measurement further due to the significant decrease in apparent viscosity near the wall over time and a consequential disparity in viscosity compared to the bulk region.In addition to shear thinning viscosity and yield stress, the formation of aggregation (rouleaux) from RBCs under low shear rates also contributes to the viscoelasticity under transient flow 

(31) and thixotropy 

(32) of whole blood. Given the difficulty in evaluating viscoelastic behavior of blood under low strain magnitudes and limitations in generalized Newtonian models, the utilization of viscoelastic models is advocated to encompass elasticity and delineate non-shear components within the stress tensor. Extending from the Oldroyd-B model, Anand et al. 

(33) developed a viscoelastic model framework for adapting elasticity within blood samples and predicting non-shear stress components. However, to also address the thixotropic effects, the model developed by Horner et al. 

(34) serves as a more comprehensive approach than the viscoelastic model from Anand et al. Thixotropy 

(32) typically occurs from the structural change of the rouleaux, where low shear rate conditions induce rouleaux formation. Correspondingly, elasticity increases, while elasticity is more representative of the isolated RBCs, under high shear rate conditions. The model of Horner et al. 

(34) considers the contribution of rouleaux to shear stress, taking into account factors such as the characteristic time for Brownian aggregation, shear-induced aggregation, and shear-induced breakage. Subsequent advancements in the model from Horner et al. often revolve around refining the three aforementioned key terms for a more substantial characterization of rouleaux dynamics. Notably, this has led to the recently developed mHAWB model 

(35) and other model iterations to enhance the accuracy of elastic and viscoelastic contributions to blood rheology, including the recently improved model suggested by Armstrong et al. 

(36)

2.2.2. Numerical Methods (FDM, FEM, FVM)

Numerical simulation has become increasingly more significant in analyzing the geometry, boundary layers of flow, and nonlinearity of hyperbolic viscoelastic flow constitutive equations. CFD is a powerful and efficient tool utilizing numerical methods to solve the governing hydrodynamic equations, such as the Navier–Stokes (N–S) equation, continuity equation, and energy conservation equation, for qualitative evaluation of fluid motion dynamics under different parameters. CFD overcomes the challenge of analytically solving nonlinear forms of differential equations by employing numerical methods such as the Finite-Difference Method (FDM), Finite-Element Method (FEM), and Finite-Volume Method (FVM) to discretize and solve the partial differential equations (PDEs), allowing for qualitative reproduction of transport phenomena and experimental observations. Different numerical methods are chosen to cope with various transport systems for optimization of the accuracy of the result and control of error during the discretization process.FDM is a straightforward approach to discretizing PDEs, replacing the continuum representation of equations with a set of finite-difference equations, which is typically applied to structured grids for efficient implementation in CFD programs. 

(37) However, FDM is often limited to simple geometries such as rectangular or block-shaped geometries and struggles with curved boundaries. In contrast, FEM divides the fluid domain into small finite grids or elements, approximating PDEs through a local description of physics. 

(38) All elements contribute to a large, sparse matrix solver. However, FEM may not always provide accurate results for systems involving significant deformation and aggregation of particles like RBCs due to large distortion of grids. 

(39) FVM evaluates PDEs following the conservation laws and discretizes the selected flow domain into small but finite size control volumes, with each grid at the center of a finite volume. 

(40) The divergence theorem allows the conversion of volume integrals of PDEs with divergence terms into surface integrals of surface fluxes across cell boundaries. Due to its conservation property, FVM offers efficient outcomes when dealing with PDEs that embody mass, momentum, and energy conservation principles. Furthermore, widely accessible software packages like the OpenFOAM toolbox 

(41) include a viscoelastic solver, making it an attractive option for viscoelastic fluid flow modeling. 

(42)

2.2.3. Modeling Methods of Blood Flow Dynamics

The complexity in the blood flow simulation arises from deformability and aggregation that RBCs exhibit during their interaction with neighboring cells under different shear rate conditions induced by blood flow. Numerical models coupled with simulation programs have been applied as a groundbreaking method to predict such unique rheological behavior exhibited by RBCs and whole blood. The conventional approach of a single-phase flow simulation is often applied to blood flow simulations within large vessels possessing a moderate shear rate. However, such a method assumes the properties of plasma, RBCs and other cellular components to be evenly distributed as average density and viscosity in blood, resulting in the inability to simulate the mechanical dynamics, such as RBC aggregation under high-shear flow field, inherent in RBCs. To accurately describe the asymmetric distribution of RBC and blood flow, multiphase flow simulation, where numerical simulations of blood flows are often modeled as two immiscible phases, RBCs and blood plasma, is proposed. A common assumption is that RBCs exhibit non-Newtonian behavior while the plasma is treated as a continuous Newtonian phase.Numerous multiphase numerical models have been proposed to simulate the influence of RBCs on blood flow dynamics by different assumptions. In large-scale simulations (above the millimeter range), continuum-based methods are wildly used due to their lower computational demands. 

(43) Eulerian multiphase flow simulations offer the solution of a set of conservation equations for each separate phase and couple the phases through common pressure and interphase exchange coefficients. Xu et al. 

(44) utilized the combined finite-discrete element method (FDEM) to replicate the dynamic behavior and distortion of RBCs subjected to fluidic forces, utilizing the Johnson–Kendall–Roberts model 

(45) to define the adhesive forces of cell-to-cell interactions. The iterative direct-forcing immersed boundary method (IBM) is commonly employed in simulations of the fluid–cell interface of blood. This method effectively captures the intricacies of the thin and flexible RBC membranes within various external flow fields. 

(46) The study by Xu et al. 

(44) also adopts this approach to bridge the fluid dynamics and RBC deformation through IBM. Yoon and You utilized the Maxwell model to define the viscosity of the RBC membrane. 

(47) It was discovered that the Maxwell model could represent the stress relaxation and unloading processes of the cell. Furthermore, the reduced flexibility of an RBC under particular situations such as infection is specified, which was unattainable by the Kelvin–Voigt model 

(48) when compared to the Maxwell model in the literature. The Yeoh hyperplastic material model was also adapted to predict the nonlinear elasticity property of RBCs with FEM employed to discretize the RBC membrane using shell-type elements. Gracka et al. 

(49) developed a numerical CFD model with a finite-volume parallel solver for multiphase blood flow simulation, where an updated Maxwell viscoelasticity model and a Discrete Phase Model are adopted. In the study, the adapted IBM, based on unstructured grids, simulates the flow behavior and shape change of the RBCs through fluid-structure coupling. It was found that the hybrid Euler–Lagrange (E–L) approach 

(50) for the development of the multiphase model offered better results in the simulated CFL region in the microchannels.To study the dynamics of individual behaviors of RBCs and the consequent non-Newtonian blood flow, cell-shape-resolved computational models are often adapted. The use of the boundary integral method has become prevalent in minimizing computational expenses, particularly in the exclusive determination of fluid velocity on the surfaces of RBCs, incorporating the option of employing IBM or particle-based techniques. The cell-shaped-resolved method has enabled an examination of cell to cell interactions within complex ambient or pulsatile flow conditions 

(51) surrounding RBC membranes. Recently, Rydquist et al. 

(52) have looked to integrate statistical information from macroscale simulations to obtain a comprehensive overview of RBC behavior within the immediate proximity of the flow through introduction of respective models characterizing membrane shape definition, tension, bending stresses of RBC membranes.At a macroscopic scale, continuum models have conventionally been adapted for assessing blood flow dynamics through the application of elasticity theory and fluid dynamics. However, particle-based methods are known for their simplicity and adaptability in modeling complex multiscale fluid structures. Meshless methods, such as the boundary element method (BEM), smoothed particle hydrodynamics (SPH), and dissipative particle dynamics (DPD), are often used in particle-based characterization of RBCs and the surrounding fluid. By representing the fluid as discrete particles, meshless methods provide insights into the status and movement of the multiphase fluid. These methods allow for the investigation of cellular structures and microscopic interactions that affect blood rheology. Non-confronting mesh methods like IBM can also be used to couple a fluid solver such as FEM, FVM, or the Lattice Boltzmann Method (LBM) through membrane representation of RBCs. In comparison to conventional CFD methods, LBM has been viewed as a favorable numerical approach for solving the N–S equations and the simulation of multiphase flows. LBM exhibits the notable advantage of being amenable to high-performance parallel computing environments due to its inherently local dynamics. In contrast to DPD and SPH where RBC membranes are modeled as physically interconnected particles, LBM employs the IBM to account for the deformation dynamics of RBCs 

(53,54) under shear flows in complex channel geometries. 

(54,55) However, it is essential to acknowledge that the utilization of LBM in simulating RBC flows often entails a significant computational overhead, being a primary challenge in this context. Krüger et al. 

(56) proposed utilizing LBM as a fluid solver, IBM to couple the fluid and FEM to compute the response of membranes to deformation under immersed fluids. This approach decouples the fluid and membranes but necessitates significant computational effort due to the requirements of both meshes and particles.Despite the accuracy of current blood flow models, simulating complex conditions remains challenging because of the high computational load and cost. Balachandran Nair et al. 

(57) suggested a reduced order model of RBC under the framework of DEM, where the RBC is represented by overlapping constituent rigid spheres. The Morse potential force is adapted to account for the RBC aggregation exhibited by cell to cell interactions among RBCs at different distances. Based upon the IBM, the reduced-order RBC model is adapted to simulate blood flow transport for validation under both single and multiple RBCs with a resolved CFD-DEM solver. 

(58) In the resolved CFD-DEM model, particle sizes are larger than the grid size for a more accurate computation of the surrounding flow field. A continuous forcing approach is taken to describe the momentum source of the governing equation prior to discretization, which is different from a Direct Forcing Method (DFM). 

(59) As no body-conforming moving mesh is required, the continuous forcing approach offers lower complexity and reduced cost when compared to the DFM. Piquet et al. 

(60) highlighted the high complexity of the DFM due to its reliance on calculating an additional immersed boundary flux for the velocity field to ensure its divergence-free condition.The fluid–structure interaction (FSI) method has been advocated to connect the dynamic interplay of RBC membranes and fluid plasma within blood flow such as the coupling of continuum–particle interactions. However, such methodology is generally adapted for anatomical configurations such as arteries 

(61,62) and capillaries, 

(63) where both the structural components and the fluid domain undergo substantial deformation due to the moving boundaries. Due to the scope of the Review being blood flow simulation within microchannels of LOC devices without deformable boundaries, the Review of the FSI method will not be further carried out.In general, three numerical methods are broadly used: mesh-based, particle-based, and hybrid mesh–particle techniques, based on the spatial scale and the fundamental numerical approach, mesh-based methods tend to neglect the effects of individual particles, assuming a continuum and being efficient in terms of time and cost. However, the particle-based approach highlights more of the microscopic and mesoscopic level, where the influence of individual RBCs is considered. A review from Freund et al. 

(64) addressed the three numerical methodologies and their respective modeling approaches of RBC dynamics. Given the complex mechanics and the diverse levels of study concerning numerical simulations of blood and cellular flow, a broad spectrum of numerical methods for blood has been subjected to extensive review. 

(64−70) Ye at al. 

(65) offered an extensive review of the application of the DPD, SPH, and LBM for numerical simulations of RBC, while Rathnayaka et al. 

(67) conducted a review of the particle-based numerical modeling for liquid marbles through drawing parallels to the transport of RBCs in microchannels. A comparative analysis between conventional CFD methods and particle-based approaches for cellular and blood flow dynamic simulation can be found under the review by Arabghahestani et al. 

(66) Literature by Li et al. 

(68) and Beris et al. 

(69) offer an overview of both continuum-based models at micro/macroscales and multiscale particle-based models encompassing various length and temporal dimensions. Furthermore, these reviews deliberate upon the potential of coupling continuum-particle methods for blood plasma and RBC modeling. Arciero et al. 

(70) investigated various modeling approaches encompassing cellular interactions, such as cell to cell or plasma interactions and the individual cellular phases. A concise overview of the reviews is provided in Table 2 for reference.

Table 2. List of Reviews for Numerical Approaches Employed in Blood Flow Simulation

ReferenceNumerical methods
Li et al. (2013) (68)Continuum-based modeling (BIM), particle-based modeling (LBM, LB-FE, SPH, DPD)
Freund (2014) (64)RBC dynamic modeling (continuum-based modeling, complementary discrete microstructure modeling), blood flow dynamic modeling (FDM, IBM, LBM, particle-mesh methods, coupled boundary integral and mesh-based methods, DPD)
Ye et al. (2016) (65)DPD, SPH, LBM, coupled IBM-Smoothed DPD
Arciero et al. (2017) (70)LBM, IBM, DPD, conventional CFD Methods (FDM, FVM, FEM)
Arabghahestani et al. (2019) (66)Particle-based methods (LBM, DPD, direct simulation Monte Carlo, molecular dynamics), SPH, conventional CFD methods (FDM, FVM, FEM)
Beris et al. (2021) (69)DPD, smoothed DPD, IBM, LBM, BIM
Rathnayaka (2022) (67)SPH, CG, LBM

3. Capillary Driven Blood Flow in LOC Systems

ARTICLE SECTIONS

Jump To


3.1. Capillary Driven Flow Phenomena

Capillary driven (CD) flow is a pivotal mechanism in passive microfluidic flow systems 

(9) such as the blood circulation system and LOC systems. 

(71) CD flow is essentially the movement of a liquid to flow against drag forces, where the capillary effect exerts a force on the liquid at the borders, causing a liquid–air meniscus to flow despite gravity or other drag forces. A capillary pressure drops across the liquid–air interface with surface tension in the capillary radius and contact angle. The capillary effect depends heavily on the interaction between the different properties of surface materials. Different values of contact angles can be manipulated and obtained under varying levels of surface wettability treatments to manipulate the surface properties, resulting in different CD blood delivery rates for medical diagnostic device microchannels. CD flow techniques are appealing for many LOC devices, because they require no external energy. However, due to the passive property of liquid propulsion by capillary forces and the long-term instability of surface treatments on channel walls, the adaptability of CD flow in geometrically complex LOC devices may be limited.

3.2. Theoretical and Numerical Modeling of Capillary Driven Blood Flow

3.2.1. Theoretical Basis and Assumptions of Microfluidic Flow

The study of transport phenomena regarding either blood flow driven by capillary forces or externally applied forces under microfluid systems all demands a comprehensive recognition of the significant differences in flow dynamics between microscale and macroscale. The fundamental assumptions and principles behind fluid transport at the microscale are discussed in this section. Such a comprehension will lay the groundwork for the following analysis of the theoretical basis of capillary forces and their role in blood transport in LOC systems.

At the macroscale, fluid dynamics are often strongly influenced by gravity due to considerable fluid mass. However, the high surface to volume ratio at the microscale shifts the balance toward surface forces (e.g., surface tension and viscous forces), much larger than the inertial force. This difference gives rise to transport phenomena unique to microscale fluid transport, such as the prevalence of laminar flow due to a very low Reynolds number (generally lower than 1). Moreover, the fluid in a microfluidic system is often assumed to be incompressible due to the small flow velocity, indicating constant fluid density in both space and time.Microfluidic flow behaviors are governed by the fundamental principles of mass and momentum conservation, which are encapsulated in the continuity equation and the Navier–Stokes (N–S) equation. The continuity equation describes the conservation of mass, while the N–S equation captures the spatial and temporal variations in velocity, pressure, and other physical parameters. Under the assumption of the negligible influence of gravity in microfluidic systems, the continuity equation and the Eulerian representation of the incompressible N–S equation can be expressed as follows:

∇·𝐮⇀=0∇·�⇀=0

(7)

−∇𝑝+𝜇∇2𝐮⇀+∇·𝝉⇀−𝐅⇀=0−∇�+�∇2�⇀+∇·�⇀−�⇀=0

(8)Here, p is the pressure, u is the fluid viscosity, 

𝝉⇀�⇀ represents the stress tensor, and F is the body force exerted by external forces if present.

3.2.2. Theoretical Basis and Modeling of Capillary Force in LOC Systems

The capillary force is often the major driving force to manipulate and transport blood without an externally applied force in LOC systems. Forces induced by the capillary effect impact the free surface of fluids and are represented not directly in the Navier–Stokes equations but through the pressure boundary conditions of the pressure term p. For hydrophilic surfaces, the liquid generally induces a contact angle between 0° and 30°, encouraging the spread and attraction of fluid under a positive cos θ condition. For this condition, the pressure drop becomes positive and generates a spontaneous flow forward. A hydrophobic solid surface repels the fluid, inducing minimal contact. Generally, hydrophobic solids exhibit a contact angle larger than 90°, inducing a negative value of cos θ. Such a value will result in a negative pressure drop and a flow in the opposite direction. The induced contact angle is often utilized to measure the wall exposure of various surface treatments on channel walls where different wettability gradients and surface tension effects for CD flows are established. Contact angles between different interfaces are obtainable through standard values or experimental methods for reference. 

(72)For the characterization of the induced force by the capillary effect, the Young–Laplace (Y–L) equation 

(73) is widely employed. In the equation, the capillary is considered a pressure boundary condition between the two interphases. Through the Y–L equation, the capillary pressure force can be determined, and subsequently, the continuity and momentum balance equations can be solved to obtain the blood filling rate. Kim et al. 

(74) studied the effects of concentration and exposure time of a nonionic surfactant, Silwet L-77, on the performance of a polydimethylsiloxane (PDMS) microchannel in terms of plasma and blood self-separation. The study characterized the capillary pressure force by incorporating the Y–L equation and further evaluated the effects of the changing contact angle due to different levels of applied channel wall surface treatments. The expression of the Y–L equation utilized by Kim et al. 

(74) is as follows:

𝑃=−𝜎(cos𝜃b+cos𝜃tℎ+cos𝜃l+cos𝜃r𝑤)�=−�(cos⁡�b+cos⁡�tℎ+cos⁡�l+cos⁡�r�)

(9)where σ is the surface tension of the liquid and θ

bθ

tθ

l, and θ

r are the contact angle values between the liquid and the bottom, top, left, and right walls, respectively. A numerical simulation through Coventor software is performed to evaluate the dynamic changes in the filling rate within the microchannel. The simulation results for the blood filling rate in the microchannel are expressed at a specific time stamp, shown in Figure 2. The results portray an increasing instantaneous filling rate of blood in the microchannel following the decrease in contact angle induced by a higher concentration of the nonionic surfactant treated to the microchannel wall.

Figure 2. Numerical simulation of filling rate of capillary driven blood flow under various contact angle conditions at a specific timestamp. (74) Reproduced with permission from ref (74). Copyright 2010 Elsevier.

When in contact with hydrophilic or hydrophobic surfaces, blood forms a meniscus with a contact angle due to surface tension. The Lucas–Washburn (L–W) equation 

(75) is one of the pioneering theoretical definitions for the position of the meniscus over time. In addition, the L–W equation provides the possibility for research to obtain the velocity of the blood formed meniscus through the derivation of the meniscus position. The L–W equation 

(75) can be shown below:

𝐿(𝑡)=𝑅𝜎cos(𝜃)𝑡2𝜇⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯√�(�)=��⁡cos(�)�2�

(10)Here L(t) represents the distance of the liquid driven by the capillary forces. However, the generalized L–W equation solely assumes the constant physical properties from a Newtonian fluid rather than considering the non-Newtonian fluid behavior of blood. Cito et al. 

(76) constructed an enhanced version of the L–W equation incorporating the power law to consider the RBC aggregation and the FL effect. The non-Newtonian fluid apparent viscosity under the Power Law model is defined as

𝜇=𝑘·(𝛾˙)𝑛−1�=�·(�˙)�−1

(11)where γ̇ is the strain rate tensor defined as 

𝛾˙=12𝛾˙𝑖𝑗𝛾˙𝑗𝑖⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯√�˙=12�˙���˙��. The stress tensor term τ is computed as τ = μγ̇

ij. The updated L–W equation by Cito 

(76) is expressed as

𝐿(𝑡)=𝑅[(𝑛+13𝑛+1)(𝜎cos(𝜃)𝑅𝑘)1/𝑛𝑡]𝑛/𝑛+1�(�)=�[(�+13�+1)(�⁡cos(�)��)1/��]�/�+1

(12)where k is the flow consistency index and n is the power law index, respectively. The power law index, from the Power Law model, characterizes the extent of the non-Newtonian behavior of blood. Both the consistency and power law index rely on blood properties such as hematocrit, the appearance of the FL effect, the formation of RBC aggregates, etc. The updated L–W equation computes the location and velocity of blood flow caused by capillary forces at specified time points within the LOC devices, taking into account the effects of blood flow characteristics such as RBC aggregation and the FL effect on dynamic blood viscosity.Apart from the blood flow behaviors triggered by inherent blood properties, unique flow conditions driven by capillary forces that are portrayed under different microchannel geometries also hold crucial implications for CD blood delivery. Berthier et al. 

(77) studied the spontaneous Concus–Finn condition, the condition to initiate the spontaneous capillary flow within a V-groove microchannel, as shown in Figure 3(a) both experimentally and numerically. Through experimental studies, the spontaneous Concus–Finn filament development of capillary driven blood flow is observed, as shown in Figure 3(b), while the dynamic development of blood flow is numerically simulated through CFD simulation.

Figure 3. (a) Sketch of the cross-section of Berthier’s V-groove microchannel, (b) experimental view of blood in the V-groove microchannel, (78) (c) illustration of the dynamic change of the extension of filament from FLOW 3D under capillary flow at three increasing time intervals. (78) Reproduced with permission from ref (78). Copyright 2014 Elsevier.

Berthier et al. 

(77) characterized the contact angle needed for the initiation of the capillary driving force at a zero-inlet pressure, through the half-angle (α) of the V-groove geometry layout, and its relation to the Concus–Finn filament as shown below:

𝜃<𝜋2−𝛼sin𝛼1+2(ℎ2/𝑤)sin𝛼<cos𝜃{�<�2−�sin⁡�1+2(ℎ2/�)⁡sin⁡�<cos⁡�

(13)Three possible regimes were concluded based on the contact angle value for the initiation of flow and development of Concus–Finn filament:

𝜃>𝜃1𝜃1>𝜃>𝜃0𝜃0no SCFSCF without a Concus−Finn filamentSCF without a Concus−Finn filament{�>�1no SCF�1>�>�0SCF without a Concus−Finn filament�0SCF without a Concus−Finn filament

(14)Under Newton’s Law, the force balance with low Reynolds and Capillary numbers results in the neglect of inertial terms. The force balance between the capillary forces and the viscous force induced by the channel wall is proposed to derive the analytical fluid velocity. This relation between the two forces offers insights into the average flow velocity and the penetration distance function dependent on time. The apparent blood viscosity is defined by Berthier et al. 

(78) through Casson’s law, 

(23) given in eq 1. The research used the FLOW-3D program from Flow Science Inc. software, which solves transient, free-surface problems using the FDM in multiple dimensions. The Volume of Fluid (VOF) method 

(79) is utilized to locate and track the dynamic extension of filament throughout the advancing interface within the channel ahead of the main flow at three progressing time stamps, as depicted in Figure 3(c).

4. Electro-osmotic Flow (EOF) in LOC Systems

ARTICLE SECTIONS

Jump To


The utilization of external forces, such as electric fields, has significantly broadened the possibility of manipulating microfluidic flow in LOC systems. 

(80) Externally applied electric field forces induce a fluid flow from the movement of ions in fluid terms as the “electro-osmotic flow” (EOF).Unique transport phenomena, such as enhanced flow velocity and flow instability, induced by non-Newtonian fluids, particularly viscoelastic fluids, under EOF, have sparked considerable interest in microfluidic devices with simple or complicated geometries within channels. 

(81) However, compared to the study of Newtonian fluids and even other electro-osmotic viscoelastic fluid flows, the literature focusing on the theoretical and numerical modeling of electro-osmotic blood flow is limited due to the complexity of blood properties. Consequently, to obtain a more comprehensive understanding of the complex blood flow behavior under EOF, theoretical and numerical studies of the transport phenomena in the EOF section will be based on the studies of different viscoelastic fluids under EOF rather than that of blood specifically. Despite this limitation, we believe these studies offer valuable insights that can help understand the complex behavior of blood flow under EOF.

4.1. EOF Phenomena

Electro-osmotic flow occurs at the interface between the microchannel wall and bulk phase solution. When in contact with the bulk phase, solution ions are absorbed or dissociated at the solid–liquid interface, resulting in the formation of a charge layer, as shown in Figure 4. This charged channel surface wall interacts with both negative and positive ions in the bulk sample, causing repulsion and attraction forces to create a thin layer of immobilized counterions, known as the Stern layer. The induced electric potential from the wall gradually decreases with an increase in the distance from the wall. The Stern layer potential, commonly termed the zeta potential, controls the intensity of the electrostatic interactions between mobile counterions and, consequently, the drag force from the applied electric field. Next to the Stern layer is the diffuse mobile layer, mainly composed of a mobile counterion. These two layers constitute the “electrical double layer” (EDL), the thickness of which is directly proportional to the ionic strength (concentration) of the bulk fluid. The relationship between the two parameters is characterized by a Debye length (λ

D), expressed as

𝜆𝐷=𝜖𝑘B𝑇2(𝑍𝑒)2𝑐0⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯√��=��B�2(��)2�0

(15)where ϵ is the permittivity of the electrolyte solution, k

B is the Boltzmann constant, T is the electron temperature, Z is the integer valence number, e is the elementary charge, and c

0 is the ionic density.

Figure 4. Schematic diagram of an electro-osmotic flow in a microchannel with negative surface charge. (82) Reproduced with permission from ref (82). Copyright 2012 Woodhead Publishing.

When an electric field is applied perpendicular to the EDL, viscous drag is generated due to the movement of excess ions in the EDL. Electro-osmotic forces can be attributed to the externally applied electric potential (ϕ) and the zeta potential, the system wall induced potential by charged walls (ψ). As illustrated in Figure 4, the majority of ions in the bulk phase have a uniform velocity profile, except for a shear rate condition confined within an extremely thin Stern layer. Therefore, EOF displays a unique characteristic of a “near flat” or plug flow velocity profile, different from the parabolic flow typically induced by pressure-driven microfluidic flow (Hagen–Poiseuille flow). The plug-shaped velocity profile of the EOF possesses a high shear rate above the Stern layer.Overall, the EOF velocity magnitude is typically proportional to the Debye Length (λ

D), zeta potential, and magnitude of the externally applied electric field, while a more viscous liquid reduces the EOF velocity.

4.2. Modeling on Electro-osmotic Viscoelastic Fluid Flow

4.2.1. Theoretical Basis of EOF Mechanisms

The EOF of an incompressible viscoelastic fluid is commonly governed by the continuity and incompressible N–S equations, as shown in eqs 7 and 8, where the stress tensor and the electrostatic force term are coupled. The electro-osmotic body force term F, representing the body force exerted by the externally applied electric force, is defined as 

𝐹⇀=𝑝𝐸𝐸⇀�⇀=���⇀, where ρ

E and 

𝐸⇀�⇀ are the net electric charge density and the applied external electric field, respectively.Numerous models are established to theoretically study the externally applied electric potential and the system wall induced potential by charged walls. The following Laplace equation, expressed as eq 16, is generally adapted and solved to calculate the externally applied potential (ϕ).

∇2𝜙=0∇2�=0

(16)Ion diffusion under applied electric fields, together with mass transport resulting from convection and diffusion, transports ionic solutions in bulk flow under electrokinetic processes. The Nernst–Planck equation can describe these transport methods, including convection, diffusion, and electro-diffusion. Therefore, the Nernst–Planck equation is used to determine the distribution of the ions within the electrolyte. The electric potential induced by the charged channel walls follows the Poisson–Nernst–Plank (PNP) equation, which can be written as eq 17.

∇·[𝐷𝑖∇𝑛𝑖−𝑢⇀𝑛𝑖+𝑛𝑖𝐷𝑖𝑧𝑖𝑒𝑘𝑏𝑇∇(𝜙+𝜓)]=0∇·[��∇��−�⇀��+����������∇(�+�)]=0

(17)where D

in

i, and z

i are the diffusion coefficient, ionic concentration, and ionic valence of the ionic species I, respectively. However, due to the high nonlinearity and numerical stiffness introduced by different lengths and time scales from the PNP equations, the Poisson–Boltzmann (PB) model is often considered the major simplified method of the PNP equation to characterize the potential distribution of the EDL region in microchannels. In the PB model, it is assumed that the ionic species in the fluid follow the Boltzmann distribution. This model is typically valid for steady-state problems where charge transport can be considered negligible, the EDLs do not overlap with each other, and the intrinsic potentials are low. It provides a simplified representation of the potential distribution in the EDL region. The PB equation governing the EDL electric potential distribution is described as

∇2𝜓=(2𝑒𝑧𝑛0𝜀𝜀0)sinh(𝑧𝑒𝜓𝑘b𝑇)∇2�=(2���0��0)⁡sinh(����b�)

(18)where n

0 is the ion bulk concentration, z is the ionic valence, and ε

0 is the electric permittivity in the vacuum. Under low electric potential conditions, an even further simplified model to illustrate the EOF phenomena is the Debye–Hückel (DH) model. The DH model is derived by obtaining a charge density term by expanding the exponential term of the Boltzmann equation in a Taylor series.

4.2.2. EOF Modeling for Viscoelastic Fluids

Many studies through numerical modeling were performed to obtain a deeper understanding of the effect exhibited by externally applied electric fields on viscoelastic flow in microchannels under various geometrical designs. Bello et al. 

(83) found that methylcellulose solution, a non-Newtonian polymer solution, resulted in stronger electro-osmotic mobility in experiments when compared to the predictions by the Helmholtz–Smoluchowski equation, which is commonly used to define the velocity of EOF of a Newtonian fluid. Being one of the pioneers to identify the discrepancies between the EOF of Newtonian and non-Newtonian fluids, Bello et al. attributed such discrepancies to the presence of a very high shear rate in the EDL, resulting in a change in the orientation of the polymer molecules. Park and Lee 

(84) utilized the FVM to solve the PB equation for the characterization of the electric field induced force. In the study, the concept of fractional calculus for the Oldroyd-B model was adapted to illustrate the elastic and memory effects of viscoelastic fluids in a straight microchannel They observed that fluid elasticity and increased ratio of viscoelastic fluid contribution to overall fluid viscosity had a significant impact on the volumetric flow rate and sensitivity of velocity to electric field strength compared to Newtonian fluids. Afonso et al. 

(85) derived an analytical expression for EOF of viscoelastic fluid between parallel plates using the DH model to account for a zeta potential condition below 25 mV. The study established the understanding of the electro-osmotic viscoelastic fluid flow under low zeta potential conditions. Apart from the electrokinetic forces, pressure forces can also be coupled with EOF to generate a unique fluid flow behavior within the microchannel. Sousa et al. 

(86) analytically studied the flow of a standard viscoelastic solution by combining the pressure gradient force with an externally applied electric force. It was found that, at a near wall skimming layer and the outer layer away from the wall, macromolecules migrating away from surface walls in viscoelastic fluids are observed. In the study, the Phan-Thien Tanner (PTT) constitutive model is utilized to characterize the viscoelastic properties of the solution. The approach is found to be valid when the EDL is much thinner than the skimming layer under an enhanced flow rate. Zhao and Yang 

(87) solved the PB equation and Carreau model for the characterization of the EOF mechanism and non-Newtonian fluid respectively through the FEM. The numerical results depict that, different from the EOF of Newtonian fluids, non-Newtonian fluids led to an increase of electro-osmotic mobility for shear thinning fluids but the opposite for shear thickening fluids.Like other fluid transport driving forces, EOF within unique geometrical layouts also portrays unique transport phenomena. Pimenta and Alves 

(88) utilized the FVM to perform numerical simulations of the EOF of viscoelastic fluids considering the PB equation and the Oldroyd-B model, in a cross-slot and flow-focusing microdevices. It was found that electroelastic instabilities are formed due to the development of large stresses inside the EDL with streamlined curvature at geometry corners. Bezerra et al. 

(89) used the FDM to numerically analyze the vortex formation and flow instability from an electro-osmotic non-Newtonian fluid flow in a microchannel with a nozzle geometry and parallel wall geometry setting. The PNP equation is utilized to characterize the charge motion in the EOF and the PTT model for non-Newtonian flow characterization. A constriction geometry is commonly utilized in blood flow adapted in LOC systems due to the change in blood flow behavior under narrow dimensions in a microchannel. Ji et al. 

(90) recently studied the EOF of viscoelastic fluid in a constriction microchannel connected by two relatively big reservoirs on both ends (as seen in Figure 5) filled with the polyacrylamide polymer solution, a viscoelastic fluid, and an incompressible monovalent binary electrolyte solution KCl.

Figure 5. Schematic diagram of a negatively charged constriction microchannel connected to two reservoirs at both ends. An electro-osmotic flow is induced in the system by the induced potential difference between the anode and cathode. (90) Reproduced with permission from ref (90). Copyright 2021 The Authors, under the terms of the Creative Commons (CC BY 4.0) License https://creativecommons.org/licenses/by/4.0/.

In studying the EOF of viscoelastic fluids, the Oldroyd-B model is often utilized to characterize the polymeric stress tensor and the deformation rate of the fluid. The Oldroyd-B model is expressed as follows:

𝜏=𝜂p𝜆(𝐜−𝐈)�=�p�(�−�)

(19)where η

p, λ, c, and I represent the polymer dynamic viscosity, polymer relaxation time, symmetric conformation tensor of the polymer molecules, and the identity matrix, respectively.A log-conformation tensor approach is taken to prevent convergence difficulty induced by the viscoelastic properties. The conformation tensor (c) in the polymeric stress tensor term is redefined by a new tensor (Θ) based on the natural logarithm of the c. The new tensor is defined as

Θ=ln(𝐜)=𝐑ln(𝚲)𝐑Θ=ln(�)=�⁡ln(�)�

(20)in which Λ is the diagonal matrix and R is the orthogonal matrix.Under the new conformation tensor, the induced EOF of a viscoelastic fluid is governed by the continuity and N–S equations adapting the Oldroyd-B model, which is expressed as

∂𝚯∂𝑡+𝐮·∇𝚯=𝛀Θ−ΘΩ+2𝐁+1𝜆(eΘ−𝐈)∂�∂�+�·∇�=�Θ−ΘΩ+2�+1�(eΘ−�)

(21)where Ω and B represent the anti-symmetric matrix and the symmetric traceless matrix of the decomposition of the velocity gradient tensor ∇u, respectively. The conformation tensor can be recovered by c = exp(Θ). The PB model and Laplace equation are utilized to characterize the charged channel wall induced potential and the externally applied potential.The governing equations are numerically solved through the FVM by RheoTool, 

(42) an open-source viscoelastic EOF solver on the OpenFOAM platform. A SIMPLEC (Semi-Implicit Method for Pressure Linked Equations-Consistent) algorithm was applied to solve the velocity-pressure coupling. The pressure field and velocity field were computed by the PCG (Preconditioned Conjugate Gradient) solver and the PBiCG (Preconditioned Biconjugate Gradient) solver, respectively.Ranging magnitudes of an applied electric field or fluid concentration induce both different streamlines and velocity magnitudes at various locations and times of the microchannel. In the study performed by Ji et al., 

(90) notable fluctuation of streamlines and vortex formation is formed at the upper stream entrance of the constriction as shown in Figure 6(a) and (b), respectively, due to the increase of electrokinetic effect, which is seen as a result of the increase in polymeric stress (τ

xx). 

(90) The contraction geometry enhances the EOF velocity within the constriction channel under high E

app condition (600 V/cm). Such phenomena can be attributed to the dependence of electro-osmotic viscoelastic fluid flow on the system wall surface and bulk fluid properties. 

(91)

Figure 6. Schematic diagram of vortex formation and streamlines of EOF depicting flow instability at (a) 1.71 s and (b) 1.75 s. Spatial distribution of the elastic normal stress at (c) high Eapp condition. Streamline of an electro-osmotic flow under Eapp of 600 V/cm (90) for (d) non-Newtonian and (e) Newtonian fluid through a constriction geometry. Reproduced with permission from ref (90). Copyright 2021 The Authors, under the terms of the Creative Commons (CC BY 4.0) License https://creativecommons.org/licenses/by/4.0/.

As elastic normal stress exceeds the local shear stress, flow instability and vortex formation occur. The induced elastic stress under EOF not only enhances the instability of the flow but often generates an irregular secondary flow leading to strong disturbance. 

(92) It is also vital to consider the effect of the constriction layout of microchannels on the alteration of the field strength within the system. The contraction geometry enhances a larger electric field strength compared with other locations of the channel outside the constriction region, resulting in a higher velocity gradient and stronger extension on the polymer within the viscoelastic solution. Following the high shear flow condition, a higher magnitude of stretch for polymer molecules in viscoelastic fluids exhibits larger elastic stresses and enhancement of vortex formation at the region. 

(93)As shown in Figure 6(c), significant elastic normal stress occurs at the inlet of the constriction microchannel. Such occurrence of a polymeric flow can be attributed to the dominating elongational flow, giving rise to high deformation of the polymers within the viscoelastic fluid flow, resulting in higher elastic stress from the polymers. Such phenomena at the entrance result in the difference in velocity streamline as circled in Figure 6(d) compared to that of the Newtonian fluid at the constriction entrance in Figure 6(e). 

(90) The difference between the Newtonian and polymer solution at the exit, as circled in Figure 6(d) and (e), can be attributed to the extrudate swell effect of polymers 

(94) within the viscoelastic fluid flow. The extrudate swell effect illustrates that, as polymers emerge from the constriction exit, they tend to contract in the flow direction and grow in the normal direction, resulting in an extrudate diameter greater than the channel size. The deformation of polymers within the polymeric flow at both the entrance and exit of the contraction channel facilitates the change in shear stress conditions of the flow, leading to the alteration in streamlines of flows for each region.

4.3. EOF Applications in LOC Systems

4.3.1. Mixing in LOC Systems

Rather than relying on the micromixing controlled by molecular diffusion under low Reynolds number conditions, active mixers actively leverage convective instability and vortex formation induced by electro-osmotic flows from alternating current (AC) or direct current (DC) electric fields. Such adaptation is recognized as significant breakthroughs for promotion of fluid mixing in chemical and biological applications such as drug delivery, medical diagnostics, chemical synthesis, and so on. 

(95)Many researchers proposed novel designs of electro-osmosis micromixers coupled with numerical simulations in conjunction with experimental findings to increase their understanding of the role of flow instability and vortex formation in the mixing process under electrokinetic phenomena. Matsubara and Narumi 

(96) numerically modeled the mixing process in a microchannel with four electrodes on each side of the microchannel wall, which generated a disruption through unstable electro-osmotic vortices. It was found that particle mixing was sensitive to both the convection effect induced by the main and secondary vortex within the micromixer and the change in oscillation frequency caused by the supplied AC voltage when the Reynolds number was varied. Qaderi et al. 

(97) adapted the PNP equation to numerically study the effect of the geometry and zeta potential configuration of the microchannel on the mixing process with a combined electro-osmotic pressure driven flow. It was reported that the application of heterogeneous zeta potential configuration enhances the mixing efficiency by around 23% while the height of the hurdles increases the mixing efficiency at most 48.1%. Cho et al. 

(98) utilized the PB model and Laplace equation to numerically simulate the electro-osmotic non-Newtonian fluid mixing process within a wavy and block layout of microchannel walls. The Power Law model is adapted to describe the fluid rheological characteristic. It was found that shear-thinning fluids possess a higher volumetric flow rate, which could result in poorer mixing efficiency compared to that of Newtonian fluids. Numerous studies have revealed that flow instability and vortex generation, in particular secondary vortices produced by barriers or greater magnitudes of heterogeneous zeta potential distribution, enhance mixing by increasing bulk flow velocity and reducing flow distance.To better understand the mechanism of disturbance formed in the system due to externally applied forces, known as electrokinetic instability, literature often utilize the Rayleigh (Ra) number, 

(1) as described below:

𝑅𝑎𝑣=𝑢ev𝑢eo=(𝛾−1𝛾+1)2𝑊𝛿2𝐸el2𝐻2𝜁𝛿Ra�=�ev�eo=(�−1�+1)2��2�el2�2��

(22)where γ is the conductivity ratio of the two streams and can be written as 

𝛾=𝜎el,H𝜎el,L�=�el,H�el,L. The Ra number characterizes the ratio between electroviscous and electro-osmotic flow. A high Ra

v value often results in good mixing. It is evident that fluid properties such as the conductivity (σ) of the two streams play a key role in the formation of disturbances to enhance mixing in microsystems. At the same time, electrokinetic parameters like the zeta potential (ζ) in the Ra number is critical in the characterization of electro-osmotic velocity and a slip boundary condition at the microchannel wall.To understand the mixing result along the channel, the concentration field can be defined and simulated under the assumption of steady state conditions and constant diffusion coefficient for each of the working fluid within the system through the convection–diffusion equation as below:

∂𝑐𝒊∂𝑡+∇⇀(𝑐𝑖𝑢⇀−𝐷𝑖∇⇀𝑐𝒊)=0∂��∂�+∇⇀(���⇀−��∇⇀��)=0

(23)where c

i is the species concentration of species i and D

i is the diffusion coefficient of the corresponding species.The standard deviation of concentration (σ

sd) can be adapted to evaluate the mixing quality of the system. 

(97) The standard deviation for concentration at a specific portion of the channel may be calculated using the equation below:

𝜎sd=∫10(𝐶∗(𝑦∗)−𝐶m)2d𝑦∗∫10d𝑦∗⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯�sd=∫01(�*(�*)−�m)2d�*∫01d�*

(24)where C*(y*) and C

m are the non-dimensional concentration profile and the mean concentration at the portion, respectively. C* is the non-dimensional concentration and can be calculated as 

𝐶∗=𝐶𝐶ref�*=��ref, where C

ref is the reference concentration defined as the bulk solution concentration. The mean concentration profile can be calculated as 

𝐶m=∫10(𝐶∗(𝑦∗)d𝑦∗∫10d𝑦∗�m=∫01(�*(�*)d�*∫01d�*. With the standard deviation of concentration, the mixing efficiency 

(97) can then be calculated as below:

𝜀𝑥=1−𝜎sd𝜎sd,0��=1−�sd�sd,0

(25)where σ

sd,0 is the standard derivation of the case of no mixing. The value of the mixing efficiency is typically utilized in conjunction with the simulated flow field and concentration field to explore the effect of geometrical and electrokinetic parameters on the optimization of the mixing results.

5. Summary

ARTICLE SECTIONS

Jump To


5.1. Conclusion

Viscoelastic fluids such as blood flow in LOC systems are an essential topic to proceed with diagnostic analysis and research through microdevices in the biomedical and pharmaceutical industries. The complex blood flow behavior is tightly controlled by the viscoelastic characteristics of blood such as the dynamic viscosity and the elastic property of RBCs under various shear rate conditions. Furthermore, the flow behaviors under varied driving forces promote an array of microfluidic transport phenomena that are critical to the management of blood flow and other adapted viscoelastic fluids in LOC systems. This review addressed the blood flow phenomena, the complicated interplay between shear rate and blood flow behaviors, and their numerical modeling under LOC systems through the lens of the viscoelasticity characteristic. Furthermore, a theoretical understanding of capillary forces and externally applied electric forces leads to an in-depth investigation of the relationship between blood flow patterns and the key parameters of the two driving forces, the latter of which is introduced through the lens of viscoelastic fluids, coupling numerical modeling to improve the knowledge of blood flow manipulation in LOC systems. The flow disturbances triggered by the EOF of viscoelastic fluids and their impact on blood flow patterns have been deeply investigated due to their important role and applications in LOC devices. Continuous advancements of various numerical modeling methods with experimental findings through more efficient and less computationally heavy methods have served as an encouraging sign of establishing more accurate illustrations of the mechanisms for multiphase blood and other viscoelastic fluid flow transport phenomena driven by various forces. Such progress is fundamental for the manipulation of unique transport phenomena, such as the generated disturbances, to optimize functionalities offered by microdevices in LOC systems.

The following section will provide further insights into the employment of studied blood transport phenomena to improve the functionality of micro devices adapting LOC technology. A discussion of the novel roles that external driving forces play in microfluidic flow behaviors is also provided. Limitations in the computational modeling of blood flow and electrokinetic phenomena in LOC systems will also be emphasized, which may provide valuable insights for future research endeavors. These discussions aim to provide guidance and opportunities for new paths in the ongoing development of LOC devices that adapt blood flow.

5.2. Future Directions

5.2.1. Electro-osmosis Mixing in LOC Systems

Despite substantial research, mixing results through flow instability and vortex formation phenomena induced by electro-osmotic mixing still deviate from the effective mixing results offered by chaotic mixing results such as those seen in turbulent flows. However, recent discoveries of a mixing phenomenon that is generally observed under turbulent flows are found within electro-osmosis micromixers under low Reynolds number conditions. Zhao 

(99) experimentally discovered a rapid mixing process in an AC applied micromixer, where the power spectrum of concentration under an applied voltage of 20 V

p-p induces a −5/3 slope within a frequency range. This value of the slope is considered as the O–C spectrum in macroflows, which is often visible under relatively high Re conditions, such as the Taylor microscale Reynolds number Re > 500 in turbulent flows. 

(100) However, the Re value in the studied system is less than 1 at the specific location and applied voltage. A secondary flow is also suggested to occur close to microchannel walls, being attributed to the increase of convective instability within the system.Despite the experimental phenomenon proposed by Zhao et al., 

(99) the range of effects induced by vital parameters of an EOF mixing system on the enhanced mixing results and mechanisms of disturbance generated by the turbulent-like flow instability is not further characterized. Such a gap in knowledge may hinder the adaptability and commercialization of the discovery of micromixers. One of the parameters for further evaluation is the conductivity gradient of the fluid flow. A relatively strong conductivity gradient (5000:1) was adopted in the system due to the conductive properties of the two fluids. The high conductivity gradients may contribute to the relatively large Rayleigh number and differences in EDL layer thickness, resulting in an unusual disturbance in laminar flow conditions and enhanced mixing results. However, high conductivity gradients are not always achievable by the working fluids due to diverse fluid properties. The reliance on turbulent-like phenomena and rapid mixing results in a large conductivity gradient should be established to prevent the limited application of fluids for the mixing system. In addition, the proposed system utilizes distinct zeta potential distributions at the top and bottom walls due to their difference in material choices, which may be attributed to the flow instability phenomena. Further studies should be made on varying zeta potential magnitude and distribution to evaluate their effect on the slip boundary conditions of the flow and the large shear rate condition close to the channel wall of EOF. Such a study can potentially offer an optimized condition in zeta potential magnitude through material choices and geometrical layout of the zeta potential for better mixing results and manipulation of mixing fluid dynamics. The two vital parameters mentioned above can be varied with the aid of numerical simulation to understand the effect of parameters on the interaction between electro-osmotic forces and electroviscous forces. At the same time, the relationship of developed streamlines of the simulated velocity and concentration field, following their relationship with the mixing results, under the impact of these key parameters can foster more insight into the range of impact that the two parameters have on the proposed phenomena and the microfluidic dynamic principles of disturbances.

In addition, many of the current investigations of electrokinetic mixers commonly emphasize the fluid dynamics of mixing for Newtonian fluids, while the utilization of biofluids, primarily viscoelastic fluids such as blood, and their distinctive response under shear forces in these novel mixing processes of LOC systems are significantly less studied. To develop more compatible microdevice designs and efficient mixing outcomes for the biomedical industry, it is necessary to fill the knowledge gaps in the literature on electro-osmotic mixing for biofluids, where properties of elasticity, dynamic viscosity, and intricate relationship with shear flow from the fluid are further considered.

5.2.2. Electro-osmosis Separation in LOC Systems

Particle separation in LOC devices, particularly in biological research and diagnostics, is another area where disturbances may play a significant role in optimization. 

(101) Plasma analysis in LOC systems under precise control of blood flow phenomena and blood/plasma separation procedures can detect vital information about infectious diseases from particular antibodies and foreign nucleic acids for medical treatments, diagnostics, and research, 

(102) offering more efficient results and simple operating procedures compared to that of the traditional centrifugation method for blood and plasma separation. However, the adaptability of LOC devices for blood and plasma separation is often hindered by microchannel clogging, where flow velocity and plasma yield from LOC devices is reduced due to occasional RBC migration and aggregation at the filtration entrance of microdevices. 

(103)It is important to note that the EOF induces flow instability close to microchannel walls, which may provide further solutions to clogging for the separation process of the LOC systems. Mohammadi et al. 

(104) offered an anti-clogging effect of RBCs at the blood and plasma separating device filtration entry, adjacent to the surface wall, through RBC disaggregation under high shear rate conditions generated by a forward and reverse EOF direction.

Further theoretical and numerical research can be conducted to characterize the effect of high shear rate conditions near microchannel walls toward the detachment of binding blood cells on surfaces and the reversibility of aggregation. Through numerical modeling with varying electrokinetic parameters to induce different degrees of disturbances or shear conditions at channel walls, it may be possible to optimize and better understand the process of disrupting the forces that bind cells to surface walls and aggregated cells at filtration pores. RBCs that migrate close to microchannel walls are often attracted by the adhesion force between the RBC and the solid surface originating from the van der Waals forces. Following RBC migration and attachment by adhesive forces adjacent to the microchannel walls as shown in Figure 7, the increase in viscosity at the region causes a lower shear condition and encourages RBC aggregation (cell–cell interaction), which clogs filtering pores or microchannels and reduces flow velocity at filtration region. Both the impact that shear forces and disturbances may induce on cell binding forces with surface walls and other cells leading to aggregation may suggest further characterization. Kinetic parameters such as activation energy and the rate-determining step for cell binding composition attachment and detachment should be considered for modeling the dynamics of RBCs and blood flows under external forces in LOC separation devices.

Figure 7. Schematic representations of clogging at a microchannel pore following the sequence of RBC migration, cell attachment to channel walls, and aggregation. (105) Reproduced with permission from ref (105). Copyright 2018 The Authors under the terms of the Creative Commons (CC BY 4.0) License https://creativecommons.org/licenses/by/4.0/.

5.2.3. Relationship between External Forces and Microfluidic Systems

In blood flow, a thicker CFL suggests a lower blood viscosity, suggesting a complex relationship between shear stress and shear rate, affecting the blood viscosity and blood flow. Despite some experimental and numerical studies on electro-osmotic non-Newtonian fluid flow, limited literature has performed an in-depth investigation of the role that applied electric forces and other external forces could play in the process of CFL formation. Additional studies on how shear rates from external forces affect CFL formation and microfluidic flow dynamics can shed light on the mechanism of the contribution induced by external driving forces to the development of a separate phase of layer, similar to CFL, close to the microchannel walls and distinct from the surrounding fluid within the system, then influencing microfluidic flow dynamics.One of the mechanisms of phenomena to be explored is the formation of the Exclusion Zone (EZ) region following a “Self-Induced Flow” (SIF) phenomenon discovered by Li and Pollack, 

(106) as shown in Figure 8(a) and (b), respectively. A spontaneous sustained axial flow is observed when hydrophilic materials are immersed in water, resulting in the buildup of a negative layer of charges, defined as the EZ, after water molecules absorb infrared radiation (IR) energy and break down into H and OH

+.

Figure 8. Schematic representations of (a) the Exclusion Zone region and (b) the Self Induced Flow through visualization of microsphere movement within a microchannel. (106) Reproduced with permission from ref (106). Copyright 2020 The Authors under the terms of the Creative Commons (CC BY 4.0) License https://creativecommons.org/licenses/by/4.0/.

Despite the finding of such a phenomenon, the specific mechanism and role of IR energy have yet to be defined for the process of EZ development. To further develop an understanding of the role of IR energy in such phenomena, a feasible study may be seen through the lens of the relationships between external forces and microfluidic flow. In the phenomena, the increase of SIF velocity under a rise of IR radiation resonant characteristics is shown in the participation of the external electric field near the microchannel walls under electro-osmotic viscoelastic fluid flow systems. The buildup of negative charges at the hydrophilic surfaces in EZ is analogous to the mechanism of electrical double layer formation. Indeed, research has initiated the exploration of the core mechanisms for EZ formation through the lens of the electrokinetic phenomena. 

(107) Such a similarity of the role of IR energy and the transport phenomena of SIF with electrokinetic phenomena paves the way for the definition of the unknown SIF phenomena and EZ formation. Furthermore, Li and Pollack 

(106) suggest whether CFL formation might contribute to a SIF of blood using solely IR radiation, a commonly available source of energy in nature, as an external driving force. The proposition may be proven feasible with the presence of the CFL region next to the negatively charged hydrophilic endothelial glycocalyx layer, coating the luminal side of blood vessels. 

(108) Further research can dive into the resonating characteristics between the formation of the CFL region next to the hydrophilic endothelial glycocalyx layer and that of the EZ formation close to hydrophilic microchannel walls. Indeed, an increase in IR energy is known to rapidly accelerate EZ formation and SIF velocity, depicting similarity to the increase in the magnitude of electric field forces and greater shear rates at microchannel walls affecting CFL formation and EOF velocity. Such correlation depicts a future direction in whether SIF blood flow can be observed and characterized theoretically further through the lens of the relationship between blood flow and shear forces exhibited by external energy.

The intricate link between the CFL and external forces, more specifically the externally applied electric field, can receive further attention to provide a more complete framework for the mechanisms between IR radiation and EZ formation. Such characterization may also contribute to a greater comprehension of the role IR can play in CFL formation next to the endothelial glycocalyx layer as well as its role as a driving force to propel blood flow, similar to the SIF, but without the commonly assumed pressure force from heart contraction as a source of driving force.

5.3. Challenges

Although there have been significant improvements in blood flow modeling under LOC systems over the past decade, there are still notable constraints that may require special attention for numerical simulation applications to benefit the adaptability of the designs and functionalities of LOC devices. Several points that require special attention are mentioned below:

1.The majority of CFD models operate under the relationship between the viscoelasticity of blood and the shear rate conditions of flow. The relative effect exhibited by the presence of highly populated RBCs in whole blood and their forces amongst the cells themselves under complex flows often remains unclearly defined. Furthermore, the full range of cell populations in whole blood requires a much more computational load for numerical modeling. Therefore, a vital goal for future research is to evaluate a reduced modeling method where the impact of cell–cell interaction on the viscoelastic property of blood is considered.
2.Current computational methods on hemodynamics rely on continuum models based upon non-Newtonian rheology at the macroscale rather than at molecular and cellular levels. Careful considerations should be made for the development of a constructive framework for the physical and temporal scales of micro/nanoscale systems to evaluate the intricate relationship between fluid driving forces, dynamic viscosity, and elasticity.
3.Viscoelastic fluids under the impact of externally applied electric forces often deviate from the assumptions of no-slip boundary conditions due to the unique flow conditions induced by externally applied forces. Furthermore, the mechanism of vortex formation and viscoelastic flow instability at laminar flow conditions should be better defined through the lens of the microfluidic flow phenomenon to optimize the prediction of viscoelastic flow across different geometrical layouts. Mathematical models and numerical methods are needed to better predict such disturbance caused by external forces and the viscoelasticity of fluids at such a small scale.
4.Under practical situations, zeta potential distribution at channel walls frequently deviates from the common assumption of a constant distribution because of manufacturing faults or inherent surface charges prior to the introduction of electrokinetic influence. These discrepancies frequently lead to inconsistent surface potential distribution, such as excess positive ions at relatively more negatively charged walls. Accordingly, unpredicted vortex formation and flow instability may occur. Therefore, careful consideration should be given to these discrepancies and how they could trigger the transport process and unexpected results of a microdevice.

Author Information

ARTICLE SECTIONS

Jump To


  • Corresponding Authors
    • Zhe Chen – Department of Chemical Engineering, School of Chemistry and Chemical Engineering, State Key Laboratory of Metal Matrix Composites, Shanghai Jiao Tong University, Shanghai 200240, P. R. China;  Email: zaccooky@sjtu.edu.cn
    • Bo Ouyang – Department of Chemical Engineering, School of Chemistry and Chemical Engineering, State Key Laboratory of Metal Matrix Composites, Shanghai Jiao Tong University, Shanghai 200240, P. R. China;  Email: bouy93@sjtu.edu.cn
    • Zheng-Hong Luo – Department of Chemical Engineering, School of Chemistry and Chemical Engineering, State Key Laboratory of Metal Matrix Composites, Shanghai Jiao Tong University, Shanghai 200240, P. R. China;  Orcidhttps://orcid.org/0000-0001-9011-6020; Email: luozh@sjtu.edu.cn
  • Authors
    • Bin-Jie Lai – Department of Chemical Engineering, School of Chemistry and Chemical Engineering, State Key Laboratory of Metal Matrix Composites, Shanghai Jiao Tong University, Shanghai 200240, P. R. China;  Orcidhttps://orcid.org/0009-0002-8133-5381
    • Li-Tao Zhu – Department of Chemical Engineering, School of Chemistry and Chemical Engineering, State Key Laboratory of Metal Matrix Composites, Shanghai Jiao Tong University, Shanghai 200240, P. R. China;  Orcidhttps://orcid.org/0000-0001-6514-8864
  • NotesThe authors declare no competing financial interest.

Acknowledgments

ARTICLE SECTIONS

Jump To


This work was supported by the National Natural Science Foundation of China (No. 22238005) and the Postdoctoral Research Foundation of China (No. GZC20231576).

Vocabulary

ARTICLE SECTIONS

Jump To


Microfluidicsthe field of technological and scientific study that investigates fluid flow in channels with dimensions between 1 and 1000 μm
Lab-on-a-Chip Technologythe field of research and technological development aimed at integrating the micro/nanofluidic characteristics to conduct laboratory processes on handheld devices
Computational Fluid Dynamics (CFD)the method utilizing computational abilities to predict physical fluid flow behaviors mathematically through solving the governing equations of corresponding fluid flows
Shear Ratethe rate of change in velocity where one layer of fluid moves past the adjacent layer
Viscoelasticitythe property holding both elasticity and viscosity characteristics relying on the magnitude of applied shear stress and time-dependent strain
Electro-osmosisthe flow of fluid under an applied electric field when charged solid surface is in contact with the bulk fluid
Vortexthe rotating motion of a fluid revolving an axis line

References

ARTICLE SECTIONS

Jump To


This article references 108 other publications.

  1. 1Neethirajan, S.; Kobayashi, I.; Nakajima, M.; Wu, D.; Nandagopal, S.; Lin, F. Microfluidics for food, agriculture and biosystems industries. Lab Chip 201111 (9), 1574– 1586,  DOI: 10.1039/c0lc00230eViewGoogle Scholar
  2. 2Whitesides, G. M. The origins and the future of microfluidics. Nature 2006442 (7101), 368– 373,  DOI: 10.1038/nature05058ViewGoogle Scholar
  3. 3Burklund, A.; Tadimety, A.; Nie, Y.; Hao, N.; Zhang, J. X. J. Chapter One – Advances in diagnostic microfluidics; Elsevier, 2020; DOI:  DOI: 10.1016/bs.acc.2019.08.001 .ViewGoogle Scholar
  4. 4Abdulbari, H. A. Chapter 12 – Lab-on-a-chip for analysis of blood. In Nanotechnology for Hematology, Blood Transfusion, and Artificial Blood; Denizli, A., Nguyen, T. A., Rajan, M., Alam, M. F., Rahman, K., Eds.; Elsevier, 2022; pp 265– 283.ViewGoogle Scholar
  5. 5Vladisavljević, G. T.; Khalid, N.; Neves, M. A.; Kuroiwa, T.; Nakajima, M.; Uemura, K.; Ichikawa, S.; Kobayashi, I. Industrial lab-on-a-chip: Design, applications and scale-up for drug discovery and delivery. Advanced Drug Delivery Reviews 201365 (11), 1626– 1663,  DOI: 10.1016/j.addr.2013.07.017ViewGoogle Scholar
  6. 6Kersaudy-Kerhoas, M.; Dhariwal, R.; Desmulliez, M. P. Y.; Jouvet, L. Hydrodynamic blood plasma separation in microfluidic channels. Microfluid. Nanofluid. 20108 (1), 105– 114,  DOI: 10.1007/s10404-009-0450-5ViewGoogle Scholar
  7. 7Popel, A. S.; Johnson, P. C. Microcirculation and Hemorheology. Annu. Rev. Fluid Mech. 200537 (1), 43– 69,  DOI: 10.1146/annurev.fluid.37.042604.133933ViewGoogle Scholar
  8. 8Fedosov, D. A.; Peltomäki, M.; Gompper, G. Deformation and dynamics of red blood cells in flow through cylindrical microchannels. Soft Matter 201410 (24), 4258– 4267,  DOI: 10.1039/C4SM00248BViewGoogle Scholar
  9. 9Chakraborty, S. Dynamics of capillary flow of blood into a microfluidic channel. Lab Chip 20055 (4), 421– 430,  DOI: 10.1039/b414566fViewGoogle Scholar
  10. 10Tomaiuolo, G.; Guido, S. Start-up shape dynamics of red blood cells in microcapillary flow. Microvascular Research 201182 (1), 35– 41,  DOI: 10.1016/j.mvr.2011.03.004ViewGoogle Scholar
  11. 11Sherwood, J. M.; Dusting, J.; Kaliviotis, E.; Balabani, S. The effect of red blood cell aggregation on velocity and cell-depleted layer characteristics of blood in a bifurcating microchannel. Biomicrofluidics 20126 (2), 24119,  DOI: 10.1063/1.4717755ViewGoogle Scholar
  12. 12Nader, E.; Skinner, S.; Romana, M.; Fort, R.; Lemonne, N.; Guillot, N.; Gauthier, A.; Antoine-Jonville, S.; Renoux, C.; Hardy-Dessources, M.-D. Blood Rheology: Key Parameters, Impact on Blood Flow, Role in Sickle Cell Disease and Effects of Exercise. Frontiers in Physiology 201910, 01329,  DOI: 10.3389/fphys.2019.01329ViewGoogle Scholar
  13. 13Trejo-Soto, C.; Lázaro, G. R.; Pagonabarraga, I.; Hernández-Machado, A. Microfluidics Approach to the Mechanical Properties of Red Blood Cell Membrane and Their Effect on Blood Rheology. Membranes 202212 (2), 217,  DOI: 10.3390/membranes12020217ViewGoogle Scholar
  14. 14Wagner, C.; Steffen, P.; Svetina, S. Aggregation of red blood cells: From rouleaux to clot formation. Comptes Rendus Physique 201314 (6), 459– 469,  DOI: 10.1016/j.crhy.2013.04.004ViewGoogle Scholar
  15. 15Kim, H.; Zhbanov, A.; Yang, S. Microfluidic Systems for Blood and Blood Cell Characterization. Biosensors 202313 (1), 13,  DOI: 10.3390/bios13010013ViewGoogle Scholar
  16. 16Fåhræus, R.; Lindqvist, T. THE VISCOSITY OF THE BLOOD IN NARROW CAPILLARY TUBES. American Journal of Physiology-Legacy Content 193196 (3), 562– 568,  DOI: 10.1152/ajplegacy.1931.96.3.562ViewGoogle Scholar
  17. 17Ascolese, M.; Farina, A.; Fasano, A. The Fåhræus-Lindqvist effect in small blood vessels: how does it help the heart?. J. Biol. Phys. 201945 (4), 379– 394,  DOI: 10.1007/s10867-019-09534-4ViewGoogle Scholar
  18. 18Bento, D.; Fernandes, C. S.; Miranda, J. M.; Lima, R. In vitro blood flow visualizations and cell-free layer (CFL) measurements in a microchannel network. Experimental Thermal and Fluid Science 2019109, 109847,  DOI: 10.1016/j.expthermflusci.2019.109847ViewGoogle Scholar
  19. 19Namgung, B.; Ong, P. K.; Wong, Y. H.; Lim, D.; Chun, K. J.; Kim, S. A comparative study of histogram-based thresholding methods for the determination of cell-free layer width in small blood vessels. Physiological Measurement 201031 (9), N61,  DOI: 10.1088/0967-3334/31/9/N01ViewGoogle Scholar
  20. 20Hymel, S. J.; Lan, H.; Fujioka, H.; Khismatullin, D. B. Cell trapping in Y-junction microchannels: A numerical study of the bifurcation angle effect in inertial microfluidics. Phys. Fluids (1994) 201931 (8), 082003,  DOI: 10.1063/1.5113516ViewGoogle Scholar
  21. 21Li, X.; Popel, A. S.; Karniadakis, G. E. Blood-plasma separation in Y-shaped bifurcating microfluidic channels: a dissipative particle dynamics simulation study. Phys. Biol. 20129 (2), 026010,  DOI: 10.1088/1478-3975/9/2/026010ViewGoogle Scholar
  22. 22Yin, X.; Thomas, T.; Zhang, J. Multiple red blood cell flows through microvascular bifurcations: Cell free layer, cell trajectory, and hematocrit separation. Microvascular Research 201389, 47– 56,  DOI: 10.1016/j.mvr.2013.05.002ViewGoogle Scholar
  23. 23Shibeshi, S. S.; Collins, W. E. The Rheology of Blood Flow in a Branched Arterial System. Appl. Rheol 200515 (6), 398– 405,  DOI: 10.1515/arh-2005-0020ViewGoogle Scholar
  24. 24Sequeira, A.; Janela, J. An Overview of Some Mathematical Models of Blood Rheology. In A Portrait of State-of-the-Art Research at the Technical University of Lisbon; Pereira, M. S., Ed.; Springer Netherlands: Dordrecht, 2007; pp 65– 87.ViewGoogle Scholar
  25. 25Walburn, F. J.; Schneck, D. J. A constitutive equation for whole human blood. Biorheology 197613, 201– 210,  DOI: 10.3233/BIR-1976-13307ViewGoogle Scholar
  26. 26Quemada, D. A rheological model for studying the hematocrit dependence of red cell-red cell and red cell-protein interactions in blood. Biorheology 198118, 501– 516,  DOI: 10.3233/BIR-1981-183-615ViewGoogle Scholar
  27. 27Varchanis, S.; Dimakopoulos, Y.; Wagner, C.; Tsamopoulos, J. How viscoelastic is human blood plasma?. Soft Matter 201814 (21), 4238– 4251,  DOI: 10.1039/C8SM00061AViewGoogle Scholar
  28. 28Apostolidis, A. J.; Moyer, A. P.; Beris, A. N. Non-Newtonian effects in simulations of coronary arterial blood flow. J. Non-Newtonian Fluid Mech. 2016233, 155– 165,  DOI: 10.1016/j.jnnfm.2016.03.008ViewGoogle Scholar
  29. 29Luo, X. Y.; Kuang, Z. B. A study on the constitutive equation of blood. J. Biomech. 199225 (8), 929– 934,  DOI: 10.1016/0021-9290(92)90233-QViewGoogle Scholar
  30. 30Oldroyd, J. G.; Wilson, A. H. On the formulation of rheological equations of state. Proceedings of the Royal Society of London. Series A. Mathematical and Physical Sciences 1950200 (1063), 523– 541,  DOI: 10.1098/rspa.1950.0035ViewGoogle Scholar
  31. 31Prado, G.; Farutin, A.; Misbah, C.; Bureau, L. Viscoelastic transient of confined red blood cells. Biophys J. 2015108 (9), 2126– 2136,  DOI: 10.1016/j.bpj.2015.03.046ViewGoogle Scholar
  32. 32Huang, C. R.; Pan, W. D.; Chen, H. Q.; Copley, A. L. Thixotropic properties of whole blood from healthy human subjects. Biorheology 198724 (6), 795– 801,  DOI: 10.3233/BIR-1987-24630ViewGoogle Scholar
  33. 33Anand, M.; Kwack, J.; Masud, A. A new generalized Oldroyd-B model for blood flow in complex geometries. International Journal of Engineering Science 201372, 78– 88,  DOI: 10.1016/j.ijengsci.2013.06.009ViewGoogle Scholar
  34. 34Horner, J. S.; Armstrong, M. J.; Wagner, N. J.; Beris, A. N. Investigation of blood rheology under steady and unidirectional large amplitude oscillatory shear. J. Rheol. 201862 (2), 577– 591,  DOI: 10.1122/1.5017623ViewGoogle Scholar
  35. 35Horner, J. S.; Armstrong, M. J.; Wagner, N. J.; Beris, A. N. Measurements of human blood viscoelasticity and thixotropy under steady and transient shear and constitutive modeling thereof. J. Rheol. 201963 (5), 799– 813,  DOI: 10.1122/1.5108737ViewGoogle Scholar
  36. 36Armstrong, M.; Tussing, J. A methodology for adding thixotropy to Oldroyd-8 family of viscoelastic models for characterization of human blood. Phys. Fluids 202032 (9), 094111,  DOI: 10.1063/5.0022501ViewGoogle Scholar
  37. 37Crank, J.; Nicolson, P. A practical method for numerical evaluation of solutions of partial differential equations of the heat-conduction type. Mathematical Proceedings of the Cambridge Philosophical Society 194743 (1), 50– 67,  DOI: 10.1017/S0305004100023197ViewGoogle Scholar
  38. 38Clough, R. W. Original formulation of the finite element method. Finite Elements in Analysis and Design 19907 (2), 89– 101,  DOI: 10.1016/0168-874X(90)90001-UViewGoogle Scholar
  39. 39Liu, W. K.; Liu, Y.; Farrell, D.; Zhang, L.; Wang, X. S.; Fukui, Y.; Patankar, N.; Zhang, Y.; Bajaj, C.; Lee, J.Immersed finite element method and its applications to biological systems. Computer Methods in Applied Mechanics and Engineering 2006195 (13), 1722– 1749,  DOI: 10.1016/j.cma.2005.05.049ViewGoogle Scholar
  40. 40Lopes, D.; Agujetas, R.; Puga, H.; Teixeira, J.; Lima, R.; Alejo, J. P.; Ferrera, C. Analysis of finite element and finite volume methods for fluid-structure interaction simulation of blood flow in a real stenosed artery. International Journal of Mechanical Sciences 2021207, 106650,  DOI: 10.1016/j.ijmecsci.2021.106650ViewGoogle Scholar
  41. 41Favero, J. L.; Secchi, A. R.; Cardozo, N. S. M.; Jasak, H. Viscoelastic flow analysis using the software OpenFOAM and differential constitutive equations. J. Non-Newtonian Fluid Mech. 2010165 (23), 1625– 1636,  DOI: 10.1016/j.jnnfm.2010.08.010ViewGoogle Scholar
  42. 42Pimenta, F.; Alves, M. A. Stabilization of an open-source finite-volume solver for viscoelastic fluid flows. J. Non-Newtonian Fluid Mech. 2017239, 85– 104,  DOI: 10.1016/j.jnnfm.2016.12.002ViewGoogle Scholar
  43. 43Chee, C. Y.; Lee, H. P.; Lu, C. Using 3D fluid-structure interaction model to analyse the biomechanical properties of erythrocyte. Phys. Lett. A 2008372 (9), 1357– 1362,  DOI: 10.1016/j.physleta.2007.09.067ViewGoogle Scholar
  44. 44Xu, D.; Kaliviotis, E.; Munjiza, A.; Avital, E.; Ji, C.; Williams, J. Large scale simulation of red blood cell aggregation in shear flows. J. Biomech. 201346 (11), 1810– 1817,  DOI: 10.1016/j.jbiomech.2013.05.010ViewGoogle Scholar
  45. 45Johnson, K. L.; Kendall, K.; Roberts, A. Surface energy and the contact of elastic solids. Proceedings of the royal society of London. A. mathematical and physical sciences 1971324 (1558), 301– 313,  DOI: 10.1098/rspa.1971.0141ViewGoogle Scholar
  46. 46Shi, L.; Pan, T.-W.; Glowinski, R. Deformation of a single red blood cell in bounded Poiseuille flows. Phys. Rev. E 201285 (1), 016307,  DOI: 10.1103/PhysRevE.85.016307ViewGoogle Scholar
  47. 47Yoon, D.; You, D. Continuum modeling of deformation and aggregation of red blood cells. J. Biomech. 201649 (11), 2267– 2279,  DOI: 10.1016/j.jbiomech.2015.11.027ViewGoogle Scholar
  48. 48Mainardi, F.; Spada, G. Creep, relaxation and viscosity properties for basic fractional models in rheology. European Physical Journal Special Topics 2011193 (1), 133– 160,  DOI: 10.1140/epjst/e2011-01387-1ViewGoogle Scholar
  49. 49Gracka, M.; Lima, R.; Miranda, J. M.; Student, S.; Melka, B.; Ostrowski, Z. Red blood cells tracking and cell-free layer formation in a microchannel with hyperbolic contraction: A CFD model validation. Computer Methods and Programs in Biomedicine 2022226, 107117,  DOI: 10.1016/j.cmpb.2022.107117ViewGoogle Scholar
  50. 50Aryan, H.; Beigzadeh, B.; Siavashi, M. Euler-Lagrange numerical simulation of improved magnetic drug delivery in a three-dimensional CT-based carotid artery bifurcation. Computer Methods and Programs in Biomedicine 2022219, 106778,  DOI: 10.1016/j.cmpb.2022.106778ViewGoogle Scholar
  51. 51Czaja, B.; Závodszky, G.; Azizi Tarksalooyeh, V.; Hoekstra, A. G. Cell-resolved blood flow simulations of saccular aneurysms: effects of pulsatility and aspect ratio. J. R Soc. Interface 201815 (146), 20180485,  DOI: 10.1098/rsif.2018.0485ViewGoogle Scholar
  52. 52Rydquist, G.; Esmaily, M. A cell-resolved, Lagrangian solver for modeling red blood cell dynamics in macroscale flows. J. Comput. Phys. 2022461, 111204,  DOI: 10.1016/j.jcp.2022.111204ViewGoogle Scholar
  53. 53Dadvand, A.; Baghalnezhad, M.; Mirzaee, I.; Khoo, B. C.; Ghoreishi, S. An immersed boundary-lattice Boltzmann approach to study the dynamics of elastic membranes in viscous shear flows. Journal of Computational Science 20145 (5), 709– 718,  DOI: 10.1016/j.jocs.2014.06.006ViewGoogle Scholar
  54. 54Krüger, T.; Holmes, D.; Coveney, P. V. Deformability-based red blood cell separation in deterministic lateral displacement devices─A simulation study. Biomicrofluidics 20148 (5), 054114,  DOI: 10.1063/1.4897913ViewGoogle Scholar
  55. 55Takeishi, N.; Ito, H.; Kaneko, M.; Wada, S. Deformation of a Red Blood Cell in a Narrow Rectangular Microchannel. Micromachines 201910 (3), 199,  DOI: 10.3390/mi10030199ViewGoogle Scholar
  56. 56Krüger, T.; Varnik, F.; Raabe, D. Efficient and accurate simulations of deformable particles immersed in a fluid using a combined immersed boundary lattice Boltzmann finite element method. Computers & Mathematics with Applications 201161 (12), 3485– 3505,  DOI: 10.1016/j.camwa.2010.03.057ViewGoogle Scholar
  57. 57Balachandran Nair, A. N.; Pirker, S.; Umundum, T.; Saeedipour, M. A reduced-order model for deformable particles with application in bio-microfluidics. Computational Particle Mechanics 20207 (3), 593– 601,  DOI: 10.1007/s40571-019-00283-8ViewGoogle Scholar
  58. 58Balachandran Nair, A. N.; Pirker, S.; Saeedipour, M. Resolved CFD-DEM simulation of blood flow with a reduced-order RBC model. Computational Particle Mechanics 20229 (4), 759– 774,  DOI: 10.1007/s40571-021-00441-xViewGoogle Scholar
  59. 59Mittal, R.; Iaccarino, G. IMMERSED BOUNDARY METHODS. Annu. Rev. Fluid Mech. 200537 (1), 239– 261,  DOI: 10.1146/annurev.fluid.37.061903.175743ViewGoogle Scholar
  60. 60Piquet, A.; Roussel, O.; Hadjadj, A. A comparative study of Brinkman penalization and direct-forcing immersed boundary methods for compressible viscous flows. Computers & Fluids 2016136, 272– 284,  DOI: 10.1016/j.compfluid.2016.06.001ViewGoogle Scholar
  61. 61Akerkouch, L.; Le, T. B. A Hybrid Continuum-Particle Approach for Fluid-Structure Interaction Simulation of Red Blood Cells in Fluid Flows. Fluids 20216 (4), 139,  DOI: 10.3390/fluids6040139ViewGoogle Scholar
  62. 62Barker, A. T.; Cai, X.-C. Scalable parallel methods for monolithic coupling in fluid-structure interaction with application to blood flow modeling. J. Comput. Phys. 2010229 (3), 642– 659,  DOI: 10.1016/j.jcp.2009.10.001ViewGoogle Scholar
  63. 63Cetin, A.; Sahin, M. A monolithic fluid-structure interaction framework applied to red blood cells. International Journal for Numerical Methods in Biomedical Engineering 201935 (2), e3171  DOI: 10.1002/cnm.3171ViewGoogle Scholar
  64. 64Freund, J. B. Numerical Simulation of Flowing Blood Cells. Annu. Rev. Fluid Mech. 201446 (1), 67– 95,  DOI: 10.1146/annurev-fluid-010313-141349ViewGoogle Scholar
  65. 65Ye, T.; Phan-Thien, N.; Lim, C. T. Particle-based simulations of red blood cells─A review. J. Biomech. 201649 (11), 2255– 2266,  DOI: 10.1016/j.jbiomech.2015.11.050ViewGoogle Scholar
  66. 66Arabghahestani, M.; Poozesh, S.; Akafuah, N. K. Advances in Computational Fluid Mechanics in Cellular Flow Manipulation: A Review. Applied Sciences 20199 (19), 4041,  DOI: 10.3390/app9194041ViewGoogle Scholar
  67. 67Rathnayaka, C. M.; From, C. S.; Geekiyanage, N. M.; Gu, Y. T.; Nguyen, N. T.; Sauret, E. Particle-Based Numerical Modelling of Liquid Marbles: Recent Advances and Future Perspectives. Archives of Computational Methods in Engineering 202229 (5), 3021– 3039,  DOI: 10.1007/s11831-021-09683-7ViewGoogle Scholar
  68. 68Li, X.; Vlahovska, P. M.; Karniadakis, G. E. Continuum- and particle-based modeling of shapes and dynamics of red blood cells in health and disease. Soft Matter 20139 (1), 28– 37,  DOI: 10.1039/C2SM26891DViewGoogle Scholar
  69. 69Beris, A. N.; Horner, J. S.; Jariwala, S.; Armstrong, M. J.; Wagner, N. J. Recent advances in blood rheology: a review. Soft Matter 202117 (47), 10591– 10613,  DOI: 10.1039/D1SM01212FViewGoogle Scholar
  70. 70Arciero, J.; Causin, P.; Malgaroli, F. Mathematical methods for modeling the microcirculation. AIMS Biophysics 20174 (3), 362– 399,  DOI: 10.3934/biophy.2017.3.362ViewGoogle Scholar
  71. 71Maria, M. S.; Chandra, T. S.; Sen, A. K. Capillary flow-driven blood plasma separation and on-chip analyte detection in microfluidic devices. Microfluid. Nanofluid. 201721 (4), 72,  DOI: 10.1007/s10404-017-1907-6ViewGoogle Scholar
  72. 72Huhtamäki, T.; Tian, X.; Korhonen, J. T.; Ras, R. H. A. Surface-wetting characterization using contact-angle measurements. Nat. Protoc. 201813 (7), 1521– 1538,  DOI: 10.1038/s41596-018-0003-zViewGoogle Scholar
  73. 73Young, T., III. An essay on the cohesion of fluids. Philosophical Transactions of the Royal Society of London 180595, 65– 87,  DOI: 10.1098/rstl.1805.0005ViewGoogle Scholar
  74. 74Kim, Y. C.; Kim, S.-H.; Kim, D.; Park, S.-J.; Park, J.-K. Plasma extraction in a capillary-driven microfluidic device using surfactant-added poly(dimethylsiloxane). Sens. Actuators, B 2010145 (2), 861– 868,  DOI: 10.1016/j.snb.2010.01.017ViewGoogle Scholar
  75. 75Washburn, E. W. The Dynamics of Capillary Flow. Physical Review 192117 (3), 273– 283,  DOI: 10.1103/PhysRev.17.273ViewGoogle Scholar
  76. 76Cito, S.; Ahn, Y. C.; Pallares, J.; Duarte, R. M.; Chen, Z.; Madou, M.; Katakis, I. Visualization and measurement of capillary-driven blood flow using spectral domain optical coherence tomography. Microfluid Nanofluidics 201213 (2), 227– 237,  DOI: 10.1007/s10404-012-0950-6ViewGoogle Scholar
  77. 77Berthier, E.; Dostie, A. M.; Lee, U. N.; Berthier, J.; Theberge, A. B. Open Microfluidic Capillary Systems. Anal Chem. 201991 (14), 8739– 8750,  DOI: 10.1021/acs.analchem.9b01429ViewGoogle Scholar
  78. 78Berthier, J.; Brakke, K. A.; Furlani, E. P.; Karampelas, I. H.; Poher, V.; Gosselin, D.; Cubizolles, M.; Pouteau, P. Whole blood spontaneous capillary flow in narrow V-groove microchannels. Sens. Actuators, B 2015206, 258– 267,  DOI: 10.1016/j.snb.2014.09.040ViewGoogle Scholar
  79. 79Hirt, C. W.; Nichols, B. D. Volume of fluid (VOF) method for the dynamics of free boundaries. J. Comput. Phys. 198139 (1), 201– 225,  DOI: 10.1016/0021-9991(81)90145-5ViewGoogle Scholar
  80. 80Chen, J.-L.; Shih, W.-H.; Hsieh, W.-H. AC electro-osmotic micromixer using a face-to-face, asymmetric pair of planar electrodes. Sens. Actuators, B 2013188, 11– 21,  DOI: 10.1016/j.snb.2013.07.012ViewGoogle Scholar
  81. 81Zhao, C.; Yang, C. Electrokinetics of non-Newtonian fluids: A review. Advances in Colloid and Interface Science 2013201-202, 94– 108,  DOI: 10.1016/j.cis.2013.09.001ViewGoogle Scholar
  82. 82Oh, K. W. 6 – Lab-on-chip (LOC) devices and microfluidics for biomedical applications. In MEMS for Biomedical Applications; Bhansali, S., Vasudev, A., Eds.; Woodhead Publishing, 2012; pp 150– 171.ViewGoogle Scholar
  83. 83Bello, M. S.; De Besi, P.; Rezzonico, R.; Righetti, P. G.; Casiraghi, E. Electroosmosis of polymer solutions in fused silica capillaries. ELECTROPHORESIS 199415 (1), 623– 626,  DOI: 10.1002/elps.1150150186ViewGoogle Scholar
  84. 84Park, H. M.; Lee, W. M. Effect of viscoelasticity on the flow pattern and the volumetric flow rate in electroosmotic flows through a microchannel. Lab Chip 20088 (7), 1163– 1170,  DOI: 10.1039/b800185eViewGoogle Scholar
  85. 85Afonso, A. M.; Alves, M. A.; Pinho, F. T. Analytical solution of mixed electro-osmotic/pressure driven flows of viscoelastic fluids in microchannels. J. Non-Newtonian Fluid Mech. 2009159 (1), 50– 63,  DOI: 10.1016/j.jnnfm.2009.01.006ViewGoogle Scholar
  86. 86Sousa, J. J.; Afonso, A. M.; Pinho, F. T.; Alves, M. A. Effect of the skimming layer on electro-osmotic─Poiseuille flows of viscoelastic fluids. Microfluid. Nanofluid. 201110 (1), 107– 122,  DOI: 10.1007/s10404-010-0651-yViewGoogle Scholar
  87. 87Zhao, C.; Yang, C. Electro-osmotic mobility of non-Newtonian fluids. Biomicrofluidics 20115 (1), 014110,  DOI: 10.1063/1.3571278ViewGoogle Scholar
  88. 88Pimenta, F.; Alves, M. A. Electro-elastic instabilities in cross-shaped microchannels. J. Non-Newtonian Fluid Mech. 2018259, 61– 77,  DOI: 10.1016/j.jnnfm.2018.04.004ViewGoogle Scholar
  89. 89Bezerra, W. S.; Castelo, A.; Afonso, A. M. Numerical Study of Electro-Osmotic Fluid Flow and Vortex Formation. Micromachines (Basel) 201910 (12), 796,  DOI: 10.3390/mi10120796ViewGoogle Scholar
  90. 90Ji, J.; Qian, S.; Liu, Z. Electroosmotic Flow of Viscoelastic Fluid through a Constriction Microchannel. Micromachines (Basel) 202112 (4), 417,  DOI: 10.3390/mi12040417ViewGoogle Scholar
  91. 91Zhao, C.; Yang, C. Exact solutions for electro-osmotic flow of viscoelastic fluids in rectangular micro-channels. Applied Mathematics and Computation 2009211 (2), 502– 509,  DOI: 10.1016/j.amc.2009.01.068ViewGoogle Scholar
  92. 92Gerum, R.; Mirzahossein, E.; Eroles, M.; Elsterer, J.; Mainka, A.; Bauer, A.; Sonntag, S.; Winterl, A.; Bartl, J.; Fischer, L. Viscoelastic properties of suspended cells measured with shear flow deformation cytometry. Elife 202211, e78823,  DOI: 10.7554/eLife.78823ViewGoogle Scholar
  93. 93Sadek, S. H.; Pinho, F. T.; Alves, M. A. Electro-elastic flow instabilities of viscoelastic fluids in contraction/expansion micro-geometries. J. Non-Newtonian Fluid Mech. 2020283, 104293,  DOI: 10.1016/j.jnnfm.2020.104293ViewGoogle Scholar
  94. 94Spanjaards, M.; Peters, G.; Hulsen, M.; Anderson, P. Numerical Study of the Effect of Thixotropy on Extrudate Swell. Polymers 202113 (24), 4383,  DOI: 10.3390/polym13244383ViewGoogle Scholar
  95. 95Rashidi, S.; Bafekr, H.; Valipour, M. S.; Esfahani, J. A. A review on the application, simulation, and experiment of the electrokinetic mixers. Chemical Engineering and Processing – Process Intensification 2018126, 108– 122,  DOI: 10.1016/j.cep.2018.02.021ViewGoogle Scholar
  96. 96Matsubara, K.; Narumi, T. Microfluidic mixing using unsteady electroosmotic vortices produced by a staggered array of electrodes. Chemical Engineering Journal 2016288, 638– 647,  DOI: 10.1016/j.cej.2015.12.013ViewGoogle Scholar
  97. 97Qaderi, A.; Jamaati, J.; Bahiraei, M. CFD simulation of combined electroosmotic-pressure driven micro-mixing in a microchannel equipped with triangular hurdle and zeta-potential heterogeneity. Chemical Engineering Science 2019199, 463– 477,  DOI: 10.1016/j.ces.2019.01.034ViewGoogle Scholar
  98. 98Cho, C.-C.; Chen, C.-L.; Chen, C. o.-K. Mixing enhancement in crisscross micromixer using aperiodic electrokinetic perturbing flows. International Journal of Heat and Mass Transfer 201255 (11), 2926– 2933,  DOI: 10.1016/j.ijheatmasstransfer.2012.02.006ViewGoogle Scholar
  99. 99Zhao, W.; Yang, F.; Wang, K.; Bai, J.; Wang, G. Rapid mixing by turbulent-like electrokinetic microflow. Chemical Engineering Science 2017165, 113– 121,  DOI: 10.1016/j.ces.2017.02.027ViewGoogle Scholar
  100. 100Tran, T.; Chakraborty, P.; Guttenberg, N.; Prescott, A.; Kellay, H.; Goldburg, W.; Goldenfeld, N.; Gioia, G. Macroscopic effects of the spectral structure in turbulent flows. Nat. Phys. 20106 (6), 438– 441,  DOI: 10.1038/nphys1674ViewGoogle Scholar
  101. 101Toner, M.; Irimia, D. Blood-on-a-chip. Annu. Rev. Biomed Eng. 20057, 77– 103,  DOI: 10.1146/annurev.bioeng.7.011205.135108ViewGoogle Scholar
  102. 102Maria, M. S.; Rakesh, P. E.; Chandra, T. S.; Sen, A. K. Capillary flow of blood in a microchannel with differential wetting for blood plasma separation and on-chip glucose detection. Biomicrofluidics 201610 (5), 054108,  DOI: 10.1063/1.4962874ViewGoogle Scholar
  103. 103Tripathi, S.; Varun Kumar, Y. V. B.; Prabhakar, A.; Joshi, S. S.; Agrawal, A. Passive blood plasma separation at the microscale: a review of design principles and microdevices. Journal of Micromechanics and Microengineering 201525 (8), 083001,  DOI: 10.1088/0960-1317/25/8/083001ViewGoogle Scholar
  104. 104Mohammadi, M.; Madadi, H.; Casals-Terré, J. Microfluidic point-of-care blood panel based on a novel technique: Reversible electroosmotic flow. Biomicrofluidics 20159 (5), 054106,  DOI: 10.1063/1.4930865ViewGoogle Scholar
  105. 105Kang, D. H.; Kim, K.; Kim, Y. J. An anti-clogging method for improving the performance and lifespan of blood plasma separation devices in real-time and continuous microfluidic systems. Sci. Rep 20188 (1), 17015,  DOI: 10.1038/s41598-018-35235-4ViewGoogle Scholar
  106. 106Li, Z.; Pollack, G. H. Surface-induced flow: A natural microscopic engine using infrared energy as fuel. Science Advances 20206 (19), eaba0941  DOI: 10.1126/sciadv.aba0941ViewGoogle Scholar
  107. 107Mercado-Uribe, H.; Guevara-Pantoja, F. J.; García-Muñoz, W.; García-Maldonado, J. S.; Méndez-Alcaraz, J. M.; Ruiz-Suárez, J. C. On the evolution of the exclusion zone produced by hydrophilic surfaces: A contracted description. J. Chem. Phys. 2021154 (19), 194902,  DOI: 10.1063/5.0043084ViewGoogle Scholar
  108. 108Yalcin, O.; Jani, V. P.; Johnson, P. C.; Cabrales, P. Implications Enzymatic Degradation of the Endothelial Glycocalyx on the Microvascular Hemodynamics and the Arteriolar Red Cell Free Layer of the Rat Cremaster Muscle. Front Physiol 20189, 168,  DOI: 10.3389/fphys.2018.00168ViewGoogle Scholar
Predicting solid-state phase transformations during metal additive manufacturing: A case study on electron-beam powder bed fusion of Inconel-738

Predicting solid-state phase transformations during metal additive manufacturing: A case study on electron-beam powder bed fusion of Inconel-738

금속 적층 제조 중 고체 상 변형 예측: Inconel-738의 전자빔 분말층 융합에 대한 사례 연구

Nana Kwabena Adomako a, Nima Haghdadi a, James F.L. Dingle bc, Ernst Kozeschnik d, Xiaozhou Liao bc, Simon P. Ringer bc, Sophie Primig a

Abstract

Metal additive manufacturing (AM) has now become the perhaps most desirable technique for producing complex shaped engineering parts. However, to truly take advantage of its capabilities, advanced control of AM microstructures and properties is required, and this is often enabled via modeling. The current work presents a computational modeling approach to studying the solid-state phase transformation kinetics and the microstructural evolution during AM. Our approach combines thermal and thermo-kinetic modelling. A semi-analytical heat transfer model is employed to simulate the thermal history throughout AM builds. Thermal profiles of individual layers are then used as input for the MatCalc thermo-kinetic software. The microstructural evolution (e.g., fractions, morphology, and composition of individual phases) for any region of interest throughout the build is predicted by MatCalc. The simulation is applied to an IN738 part produced by electron beam powder bed fusion to provide insights into how γ′ precipitates evolve during thermal cycling. Our simulations show qualitative agreement with our experimental results in predicting the size distribution of γ′ along the build height, its multimodal size character, as well as the volume fraction of MC carbides. Our findings indicate that our method is suitable for a range of AM processes and alloys, to predict and engineer their microstructures and properties.

Graphical Abstract

ga1

Keywords

Additive manufacturing, Simulation, Thermal cycles, γ′ phase, IN738

1. Introduction

Additive manufacturing (AM) is an advanced manufacturing method that enables engineering parts with intricate shapes to be fabricated with high efficiency and minimal materials waste. AM involves building up 3D components layer-by-layer from feedstocks such as powder [1]. Various alloys, including steel, Ti, Al, and Ni-based superalloys, have been produced using different AM techniques. These techniques include directed energy deposition (DED), electron- and laser powder bed fusion (E-PBF and L-PBF), and have found applications in a variety of industries such as aerospace and power generation [2][3][4]. Despite the growing interest, certain challenges limit broader applications of AM fabricated components in these industries and others. One of such limitations is obtaining a suitable and reproducible microstructure that offers the desired mechanical properties consistently. In fact, the AM as-built microstructure is highly complex and considerably distinctive from its conventionally processed counterparts owing to the complicated thermal cycles arising from the deposition of several layers upon each other [5][6].

Several studies have reported that the solid-state phases and solidification microstructure of AM processed alloys such as CMSX-4, CoCr [7][8], Ti-6Al-4V [9][10][11]IN738 [6]304L stainless steel [12], and IN718 [13][14] exhibit considerable variations along the build direction. For instance, references [9][10] have reported that there is a variation in the distribution of α and β phases along the build direction in Ti-alloys. Similarly, the microstructure of an L-PBF fabricated martensitic steel exhibits variations in the fraction of martensite [15]. Furthermore, some of the present authors and others [6][16][17][18][19][20] have recently reviewed and reported that there is a difference in the morphology and fraction of nanoscale precipitates as a function of build height in Ni-based superalloys. These non-uniformities in the as-built microstructure result in an undesired heterogeneity in mechanical and other important properties such as corrosion and oxidation [19][21][22][23]. To obtain the desired microstructure and properties, additional processing treatments are utilized, but this incurs extra costs and may lead to precipitation of detrimental phases and grain coarsening. Therefore, a through-process understanding of the microstructure evolution under repeated heating and cooling is now needed to further advance 3D printed microstructure and property control.

It is now commonly understood that the microstructure evolution during printing is complex, and most AM studies concentrate on the microstructure and mechanical properties of the final build only. Post-printing studies of microstructure characteristics at room temperature miss crucial information on how they evolve. In-situ measurements and modelling approaches are required to better understand the complex microstructural evolution under repeated heating and cooling. Most in-situ measurements in AM focus on monitoring the microstructural changes, such as phase transformations and melt pool dynamics during fabrication using X-ray scattering and high-speed X-ray imaging [24][25][26][27]. For example, Zhao et al. [25] measured the rate of solidification and described the α/β phase transformation during L-PBF of Ti-6Al-4V in-situ. Also, Wahlmann et al. [21] recently used an L-PBF machine coupled with X-ray scattering to investigate the changes in CMSX-4 phase during successive melting processes. Although these techniques provide significant understanding of the basic principles of AM, they are not widely accessible. This is due to the great cost of the instrument, competitive application process, and complexities in terms of the experimental set-up, data collection, and analysis [26][28].

Computational modeling techniques are promising and more widely accessible tools that enable advanced understanding, prediction, and engineering of microstructures and properties during AM. So far, the majority of computational studies have concentrated on physics based process models for metal AM, with the goal of predicting the temperature profile, heat transfer, powder dynamics, and defect formation (e.g., porosity) [29][30]. In recent times, there have been efforts in modeling of the AM microstructure evolution using approaches such as phase-field [31], Monte Carlo (MC) [32], and cellular automata (CA) [33], coupled with finite element simulations for temperature profiles. However, these techniques are often restricted to simulating the evolution of solidification microstructures (e.g., grain and dendrite structure) and defects (e.g., porosity). For example, Zinovieva et al. [33] predicted the grain structure of L-PBF Ti-6Al-4V using finite difference and cellular automata methods. However, studies on the computational modelling of the solid-state phase transformations, which largely determine the resulting properties, remain limited. This can be attributed to the multi-component and multi-phase nature of most engineering alloys in AM, along with the complex transformation kinetics during thermal cycling. This kind of research involves predictions of the thermal cycle in AM builds, and connecting it to essential thermodynamic and kinetic data as inputs for the model. Based on the information provided, the thermokinetic model predicts the history of solid-state phase microstructure evolution during deposition as output. For example, a multi-phase, multi-component mean-field model has been developed to simulate the intermetallic precipitation kinetics in IN718 [34] and IN625 [35] during AM. Also, Basoalto et al. [36] employed a computational framework to examine the contrasting distributions of process-induced microvoids and precipitates in two Ni-based superalloys, namely IN718 and CM247LC. Furthermore, McNamara et al. [37] established a computational model based on the Johnson-Mehl-Avrami model for non-isothermal conditions to predict solid-state phase transformation kinetics in L-PBF IN718 and DED Ti-6Al-4V. These models successfully predicted the size and volume fraction of individual phases and captured the repeated nucleation and dissolution of precipitates that occur during AM.

In the current study, we propose a modeling approach with appreciably short computational time to investigate the detailed microstructural evolution during metal AM. This may include obtaining more detailed information on the morphologies of phases, such as size distribution, phase fraction, dissolution and nucleation kinetics, as well as chemistry during thermal cycling and final cooling to room temperature. We utilize the combination of the MatCalc thermo-kinetic simulator and a semi-analytical heat conduction model. MatCalc is a software suite for simulation of phase transformations, microstructure evolution and certain mechanical properties in engineering alloys. It has successfully been employed to simulate solid-state phase transformations in Ni-based superalloys [38][39], steels [40], and Al alloys [41] during complex thermo-mechanical processes. MatCalc uses the classical nucleation theory as well as the so-called Svoboda-Fischer-Fratzl-Kozeschnik (SFFK) growth model as the basis for simulating precipitation kinetics [42]. Although MatCalc was originally developed for conventional thermo-mechanical processes, we will show that it is also applicable for AM if the detailed time-temperature profile of the AM build is known. The semi-analytical heat transfer code developed by Stump and Plotkowski [43] is used to simulate these profile throughout the AM build.

1.1. Application to IN738

Inconel-738 (IN738) is a precipitation hardening Ni-based superalloy mainly employed in high-temperature components, e.g. in gas turbines and aero-engines owing to its exceptional mechanical properties at temperatures up to 980 °C, coupled with high resistance to oxidation and corrosion [44]. Its superior high-temperature strength (∼1090 MPa tensile strength) is provided by the L12 ordered Ni3(Al,Ti) γ′ phase that precipitates in a face-centered cubic (FCC) γ matrix [45][46]. Despite offering great properties, IN738, like most superalloys with high γ′ fractions, is challenging to process owing to its propensity to hot cracking [47][48]. Further, machining of such alloys is challenging because of their high strength and work-hardening rates. It is therefore difficult to fabricate complex INC738 parts using traditional manufacturing techniques like casting, welding, and forging.

The emergence of AM has now made it possible to fabricate such parts from IN738 and other superalloys. Some of the current authors’ recent research successfully applied E-PBF to fabricate defect-free IN738 containing γ′ throughout the build [16][17]. The precipitated γ′ were heterogeneously distributed. In particular, Haghdadi et al. [16] studied the origin of the multimodal size distribution of γ′, while Lim et al. [17] investigated the gradient in γ′ character with build height and its correlation to mechanical properties. Based on these results, the present study aims to extend the understanding of the complex and site-specific microstructural evolution in E-PBF IN738 by using a computational modelling approach. New experimental evidence (e.g., micrographs not published previously) is presented here to support the computational results.

2. Materials and Methods

2.1. Materials preparation

IN738 Ni-based superalloy (59.61Ni-8.48Co-7.00Al-17.47Cr-3.96Ti-1.01Mo-0.81W-0.56Ta-0.49Nb-0.47C-0.09Zr-0.05B, at%) gas-atomized powder was used as feedstock. The powders, with average size of 60 ± 7 µm, were manufactured by Praxair and distributed by Astro Alloys Inc. An Arcam Q10 machine by GE Additive with an acceleration voltage of 60 kV was used to fabricate a 15 × 15 × 25 mm3 block (XYZ, Z: build direction) on a 316 stainless steel substrate. The block was 3D-printed using a ‘random’ spot melt pattern. The random spot melt pattern involves randomly selecting points in any given layer, with an equal chance of each point being melted. Each spot melt experienced a dwell time of 0.3 ms, and the layer thickness was 50 µm. Some of the current authors have previously characterized the microstructure of the very same and similar builds in more detail [16][17]. A preheat temperature of ∼1000 °C was set and kept during printing to reduce temperature gradients and, in turn, thermal stresses [49][50][51]. Following printing, the build was separated from the substrate through electrical discharge machining. It should be noted that this sample was simultaneously printed with the one used in [17] during the same build process and on the same build plate, under identical conditions.

2.2. Microstructural characterization

The printed sample was longitudinally cut in the direction of the build using a Struers Accutom-50, ground, and then polished to 0.25 µm suspension via standard techniques. The polished x-z surface was electropolished and etched using Struers A2 solution (perchloric acid in ethanol). Specimens for image analysis were polished using a 0.06 µm colloidal silica. Microstructure analyses were carried out across the height of the build using optical microscopy (OM) and scanning electron microscopy (SEM) with focus on the microstructure evolution (γ′ precipitates) in individual layers. The position of each layer being analyzed was determined by multiplying the layer number by the layer thickness (50 µm). It should be noted that the position of the first layer starts where the thermal profile is tracked (in this case, 2 mm from the bottom). SEM images were acquired using a JEOL 7001 field emission microscope. The brightness and contrast settings, acceleration voltage of 15 kV, working distance of 10 mm, and other SEM imaging parameters were all held constant for analysis of the entire build. The ImageJ software was used for automated image analysis to determine the phase fraction and size of γ′ precipitates and carbides. A 2-pixel radius Gaussian blur, following a greyscale thresholding and watershed segmentation was used [52]. Primary γ′ sizes (>50 nm), were measured using equivalent spherical diameters. The phase fractions were considered equal to the measured area fraction. Secondary γ′ particles (<50 nm) were not considered here. The γ′ size in the following refers to the diameter of a precipitate.

2.3. Hardness testing

A Struers DuraScan tester was utilized for Vickers hardness mapping on a polished x-z surface, from top to bottom under a maximum load of 100 mN and 10 s dwell time. 30 micro-indentations were performed per row. According to the ASTM standard [53], the indentations were sufficiently distant (∼500 µm) to assure that strain-hardened areas did not interfere with one another.

2.4. Computational simulation of E-PBF IN738 build

2.4.1. Thermal profile modeling

The thermal history was generated using the semi-analytical heat transfer code (also known as the 3DThesis code) developed by Stump and Plotkowski [43]. This code is an open-source C++ program which provides a way to quickly simulate the conductive heat transfer found in welding and AM. The key use case for the code is the simulation of larger domains than is practicable with Computational Fluid Dynamics/Finite Element Analysis programs like FLOW-3D AM. Although simulating conductive heat transfer will not be an appropriate simplification for some investigations (for example the modelling of keyholding or pore formation), the 3DThesis code does provide fast estimates of temperature, thermal gradient, and solidification rate which can be useful for elucidating microstructure formation across entire layers of an AM build. The mathematics involved in the code is as follows:

In transient thermal conduction during welding and AM, with uniform and constant thermophysical properties and without considering fluid convection and latent heat effects, energy conservation can be expressed as:(1)��∂�∂�=�∇2�+�̇where � is density, � specific heat, � temperature, � time, � thermal conductivity, and �̇ a volumetric heat source. By assuming a semi-infinite domain, Eq. 1 can be analytically solved. The solution for temperature at a given time (t) using a volumetric Gaussian heat source is presented as:(2)��,�,�,�−�0=33�����32∫0�1������exp−3�′�′2��+�′�′2��+�′�′2����′(3)and��=12��−�′+��2for�=�,�,�(4)and�′�′=�−���′Where � is the vector �,�,� and �� is the location of the heat source.

The numerical integration scheme used is an adaptive Gaussian quadrature method based on the following nondimensionalization:(5)�=��xy2�,�′=��xy2�′,�=��xy,�=��xy,�=��xy,�=���xy

A more detailed explanation of the mathematics can be found in reference [43].

The main source of the thermal cycling present within a powder-bed fusion process is the fusion of subsequent layers. Therefore, regions near the top of a build are expected to undergo fewer thermal cycles than those closer to the bottom. For this purpose, data from the single scan’s thermal influence on multiple layers was spliced to represent the thermal cycles experienced at a single location caused by multiple subsequent layers being fused.

The cross-sectional area simulated by this model was kept constant at 1 × 1 mm2, and the depth was dependent on the build location modelled with MatCalc. For a build location 2 mm from the bottom, the maximum number of layers to simulate is 460. Fig. 1a shows a stitched overview OM image of the entire build indicating the region where this thermal cycle is simulated and tracked. To increase similarity with the conditions of the physical build, each thermal history was constructed from the results of two simulations generated with different versions of a random scan path. The parameters used for these thermal simulations can be found in Table 1. It should be noted that the main purpose of the thermal profile modelling was to demonstrate how the conditions at different locations of the build change relative to each other. Accurately predicting the absolute temperature during the build would require validation via a temperature sensor measurement during the build process which is beyond the scope of the study. Nonetheless, to establish the viability of the heat source as a suitable approximation for this study, an additional sensitivity analysis was conducted. This analysis focused on the influence of energy input on γ′ precipitation behavior, the central aim of this paper. This was achieved by employing varying beam absorption energies (0.76, 0.82 – the values utilized in the simulation, and 0.9). The direct impact of beam absorption efficiency on energy input into the material was investigated. Specifically, the initial 20 layers of the build were simulated and subsequently compared to experimental data derived from SEM. While phase fractions were found to be consistent across all conditions, disparities emerged in the mean size of γ′ precipitates. An absorption efficiency of 0.76 yielded a mean size of approximately 70 nm. Conversely, absorption efficiencies of 0.82 and 0.9 exhibited remarkably similar mean sizes of around 130 nm, aligning closely with the outcomes of the experiments.

Fig. 1

Table 1. A list of parameters used in thermal simulation of E-PBF.

ParameterValue
Spatial resolution5 µm
Time step0.5 s
Beam diameter200 µm
Beam penetration depth1 µm
Beam power1200 W
Beam absorption efficiency0.82
Thermal conductivity25.37 W/(m⋅K)
Chamber temperature1000 °C
Specific heat711.756 J/(kg⋅K)
Density8110 kg/m3

2.4.2. Thermo-kinetic simulation

The numerical analyses of the evolution of precipitates was performed using MatCalc version 6.04 (rel 0.011). The thermodynamic (‘mc_ni.tdb’, version 2.034) and diffusion (‘mc_ni.ddb’, version 2.007) databases were used. MatCalc’s basic principles are elaborated as follows:

The nucleation kinetics of precipitates are computed using a computational technique based on a classical nucleation theory [54] that has been modified for systems with multiple components [42][55]. Accordingly, the transient nucleation rate (�), which expresses the rate at which nuclei are formed per unit volume and time, is calculated as:(6)�=�0��*∙�xp−�*�∙�∙exp−��where �0 denotes the number of active nucleation sites, �* the rate of atomic attachment, � the Boltzmann constant, � the temperature, �* the critical energy for nucleus formation, τ the incubation time, and t the time. � (Zeldovich factor) takes into consideration that thermal excitation destabilizes the nucleus as opposed to its inactive state [54]. Z is defined as follows:(7)�=−12�kT∂2∆�∂�2�*12where ∆� is the overall change in free energy due to the formation of a nucleus and n is the nucleus’ number of atoms. ∆�’s derivative is evaluated at n* (critical nucleus size). �* accounts for the long-range diffusion of atoms required for nucleation, provided that the matrix’ and precipitates’ composition differ. Svoboda et al. [42] developed an appropriate multi-component equation for �*, which is given by:(8)�*=4��*2�4�∑�=1��ki−�0�2�0��0�−1where �* denotes the critical radius for nucleation, � represents atomic distance, and � is the molar volume. �ki and �0� represent the concentration of elements in the precipitate and matrix, respectively. The parameter �0� denotes the rate of diffusion of the ith element within the matrix. The expression for the incubation time � is expressed as [54]:(9)�=12�*�2

and �*, which represents the critical energy for nucleation:(10)�*=16�3�3∆�vol2where � is the interfacial energy, and ∆Gvol the change in the volume free energy. The critical nucleus’ composition is similar to the γ′ phase’s equilibrium composition at the same temperature. � is computed based on the precipitate and matrix compositions, using a generalized nearest neighbor broken bond model, with the assumption of interfaces being planar, sharp, and coherent [56][57][58].

In Eq. 7, it is worth noting that �* represents the fundamental variable in the nucleation theory. It contains �3/∆�vol2 and is in the exponent of the nucleation rate. Therefore, even small variations in γ and/or ∆�vol can result in notable changes in �, especially if �* is in the order of �∙�. This is demonstrated in [38] for UDIMET 720 Li during continuous cooling, where these quantities change steadily during precipitation due to their dependence on matrix’ and precipitate’s temperature and composition. In the current work, these changes will be even more significant as the system is exposed to multiple cycles of rapid cooling and heating.

Once nucleated, the growth of a precipitate is assessed using the radius and composition evolution equations developed by Svoboda et al. [42] with a mean-field method that employs the thermodynamic extremal principle. The expression for the total Gibbs free energy of a thermodynamic system G, which consists of n components and m precipitates, is given as follows:(11)�=∑���0��0�+∑�=1�4���33��+∑�=1��ki�ki+∑�=1�4���2��.

The chemical potential of component � in the matrix is denoted as �0�(�=1,…,�), while the chemical potential of component � in the precipitate is represented by �ki(�=1,…,�,�=1,…,�). These chemical potentials are defined as functions of the concentrations �ki(�=1,…,�,�=1,…,�). The interface energy density is denoted as �, and �� incorporates the effects of elastic energy and plastic work resulting from the volume change of each precipitate.

Eq. (12) establishes that the total free energy of the system in its current state relies on the independent state variables: the sizes (radii) of the precipitates �� and the concentrations of each component �ki. The remaining variables can be determined by applying the law of mass conservation to each component �. This can be represented by the equation:(12)��=�0�+∑�=1�4���33�ki,

Furthermore, the global mass conservation can be expressed by equation:(13)�=∑�=1���When a thermodynamic system transitions to a more stable state, the energy difference between the initial and final stages is dissipated. This model considers three distinct forms of dissipation effects [42]. These include dissipations caused by the movement of interfaces, diffusion within the precipitate and diffusion within the matrix.

Consequently, �̇� (growth rate) and �̇ki (chemical composition’s rate of change) of the precipitate with index � are derived from the linear system of equation system:(14)�ij��=��where �� symbolizes the rates �̇� and �̇ki [42]. Index i contains variables for precipitate radius, chemical composition, and stoichiometric boundary conditions suggested by the precipitate’s crystal structure. Eq. (10) is computed separately for every precipitate �. For a more detailed description of the formulae for the coefficients �ij and �� employed in this work please refer to [59].

The MatCalc software was used to perform the numerical time integration of �̇� and �̇ki of precipitates based on the classical numerical method by Kampmann and Wagner [60]. Detailed information on this method can be found in [61]. Using this computational method, calculations for E-PBF thermal cycles (cyclic heating and cooling) were computed and compared to experimental data. The simulation took approximately 2–4 hrs to complete on a standard laptop.

3. Results

3.1. Microstructure

Fig. 1 displays a stitched overview image and selected SEM micrographs of various γ′ morphologies and carbides after observations of the X-Z surface of the build from the top to 2 mm above the bottom. Fig. 2 depicts a graph that charts the average size and phase fraction of the primary γ′, as it changes with distance from the top to the bottom of the build. The SEM micrographs show widespread primary γ′ precipitation throughout the entire build, with the size increasing in the top to bottom direction. Particularly, at the topmost height, representing the 460th layer (Z = 22.95 mm), as seen in Fig. 1b, the average size of γ′ is 110 ± 4 nm, exhibiting spherical shapes. This is representative of the microstructure after it solidifies and cools to room temperature, without experiencing additional thermal cycles. The γ′ size slightly increases to 147 ± 6 nm below this layer and remains constant until 0.4 mm (∼453rd layer) from the top. At this position, the microstructure still closely resembles that of the 460th layer. After the 453rd layer, the γ′ size grows rapidly to ∼503 ± 19 nm until reaching the 437th layer (1.2 mm from top). The γ′ particles here have a cuboidal shape, and a small fraction is coarser than 600 nm. γ′ continue to grow steadily from this position to the bottom (23 mm from the top). A small fraction of γ′ is > 800 nm.

Fig. 2

Besides primary γ′, secondary γ′ with sizes ranging from 5 to 50 nm were also found. These secondary γ′ precipitates, as seen in Fig. 1f, were present only in the bottom and middle regions. A detailed analysis of the multimodal size distribution of γ′ can be found in [16]. There is no significant variation in the phase fraction of the γ′ along the build. The phase fraction is ∼ 52%, as displayed in Fig. 2. It is worth mentioning that the total phase fraction of γ′ was estimated based on the primary γ′ phase fraction because of the small size of secondary γ′. Spherical MC carbides with sizes ranging from 50 to 400 nm and a phase fraction of 0.8% were also observed throughout the build. The carbides are the light grey precipitates in Fig. 1g. The light grey shade of carbides in the SEM images is due to their composition and crystal structure [52]. These carbides are not visible in Fig. 1b-e because they were dissolved during electro-etching carried out after electropolishing. In Fig. 1g, however, the sample was examined directly after electropolishing, without electro-etching.

Table 2 shows the nominal and measured composition of γ′ precipitates throughout the build by atom probe microscopy as determined in our previous study [17]. No build height-dependent composition difference was observed in either of the γ′ precipitate populations. However, there was a slight disparity between the composition of primary and secondary γ′. Among the main γ′ forming elements, the primary γ′ has a high Ti concentration while secondary γ′ has a high Al concentration. A detailed description of the atom distribution maps and the proxigrams of the constituent elements of γ′ throughout the build can be found in [17].

Table 2. Bulk IN738 composition determined using inductively coupled plasma atomic emission spectroscopy (ICP-AES). Compositions of γ, primary γ′, and secondary γ′ at various locations in the build measured by APT. This information is reproduced from data in Ref. [17] with permission.

at%NiCrCoAlMoWTiNbCBZrTaOthers
Bulk59.1217.478.487.001.010.813.960.490.470.050.090.560.46
γ matrix
Top50.4832.9111.591.941.390.820.440.80.030.030.020.24
Mid50.3732.6111.931.791.540.890.440.10.030.020.020.010.23
Bot48.1034.5712.082.141.430.880.480.080.040.030.010.12
Primary γ′
Top72.172.513.4412.710.250.397.780.560.030.020.050.08
Mid71.602.573.2813.550.420.687.040.730.010.030.040.04
Bot72.342.473.8612.500.260.447.460.500.050.020.020.030.04
Secondary γ′
Mid70.424.203.2314.190.631.035.340.790.030.040.040.05
Bot69.914.063.6814.320.811.045.220.650.050.100.020.11

3.2. Hardness

Fig. 3a shows the Vickers hardness mapping performed along the entire X-Z surface, while Fig. 3b shows the plot of average hardness at different build heights. This hardness distribution is consistent with the γ′ precipitate size gradient across the build direction in Fig. 1Fig. 2. The maximum hardness of ∼530 HV1 is found at ∼0.5 mm away from the top surface (Z = 22.5), where γ′ particles exhibit the smallest observed size in Fig. 2b. Further down the build (∼ 2 mm from the top), the hardness drops to the 440–490 HV1 range. This represents the region where γ′ begins to coarsen. The hardness drops further to 380–430 HV1 at the bottom of the build.

Fig. 3

3.3. Modeling of the microstructural evolution during E-PBF

3.3.1. Thermal profile modeling

Fig. 4 shows the simulated thermal profile of the E-PBF build at a location of 23 mm from the top of the build, using a semi-analytical heat conduction model. This profile consists of the time taken to deposit 460 layers until final cooling, as shown in Fig. 4a. Fig. 4b-d show the magnified regions of Fig. 4a and reveal the first 20 layers from the top, a single layer (first layer from the top), and the time taken for the build to cool after the last layer deposition, respectively.

Fig. 4

The peak temperatures experienced by previous layers decrease progressively as the number of layers increases but never fall below the build preheat temperature (1000 °C). Our simulated thermal cycle may not completely capture the complexity of the actual thermal cycle utilized in the E-PBF build. For instance, the top layer (Fig. 4c), also representing the first deposit’s thermal profile without additional cycles (from powder heating, melting, to solidification), recorded the highest peak temperature of 1390 °C. Although this temperature is above the melting range of the alloy (1230–1360 °C) [62], we believe a much higher temperature was produced by the electron beam to melt the powder. Nevertheless, the solidification temperature and dynamics are outside the scope of this study as our focus is on the solid-state phase transformations during deposition. It takes ∼25 s for each layer to be deposited and cooled to the build temperature. The interlayer dwell time is 125 s. The time taken for the build to cool to room temperature (RT) after final layer deposition is ∼4.7 hrs (17,000 s).

3.3.2. MatCalc simulation

During the MatCalc simulation, the matrix phase is defined as γ. γ′, and MC carbide are included as possible precipitates. The domain of these precipitates is set to be the matrix (γ), and nucleation is assumed to be homogenous. In homogeneous nucleation, all atoms of the unit volume are assumed to be potential nucleation sitesTable 3 shows the computational parameters used in the simulation. All other parameters were set at default values as recommended in the version 6.04.0011 of MatCalc. The values for the interfacial energies are automatically calculated according to the generalized nearest neighbor broken bond model and is one of the most outstanding features in MatCalc [56][57][58]. It should be noted that the elastic misfit strain was not included in the calculation. The output of MatCalc includes phase fraction, size, nucleation rate, and composition of the precipitates. The phase fraction in MatCalc is the volume fraction. Although the experimental phase fraction is the measured area fraction, it is relatively similar to the volume fraction. This is because of the generally larger precipitate size and similar morphology at the various locations along the build [63]. A reliable phase fraction comparison between experiment and simulation can therefore be made.

Table 3. Computational parameters used in the simulation.

Precipitation domainγ
Nucleation site γ′Bulk (homogenous)
Nucleation site MC carbideBulk (Homogenous)
Precipitates class size250
Regular solution critical temperature γ′2500 K[64]
Calculated interfacial energyγ′ = 0.080–0.140 J/m2 and MC carbide = 0.410–0.430 J/m2
3.3.2.1. Precipitate phase fraction

Fig. 5a shows the simulated phase fraction of γ′ and MC carbide during thermal cycling. Fig. 5b is a magnified view of 5a showing the simulated phase fraction at the center points of the top 70 layers, whereas Fig. 5c corresponds to the first two layers from the top. As mentioned earlier, the top layer (460th layer) represents the microstructure after solidification. The microstructure of the layers below is determined by the number of thermal cycles, which increases with distance to the top. For example, layers 459, 458, 457, up to layer 1 (region of interest) experience 1, 2, 3 and 459 thermal cycles, respectively. In the top layer in Fig. 5c, the volume fraction of γ′ and carbides increases with temperature. For γ′, it decreases to zero when the temperature is above the solvus temperature after a few seconds. Carbides, however, remain constant in their volume fraction reaching equilibrium (phase fraction ∼ 0.9%) in a short time. The topmost layer can be compared to the first deposit, and the peak in temperature symbolizes the stage where the electron beam heats the powder until melting. This means γ′ and carbide precipitation might have started in the powder particles during heating from the build temperature and electron beam until the onset of melting, where γ′ dissolves, but carbides remain stable [28].

Fig. 5

During cooling after deposition, γ′ reprecipitates at a temperature of 1085 °C, which is below its solvus temperature. As cooling progresses, the phase fraction increases steadily to ∼27% and remains constant at 1000 °C (elevated build temperature). The calculated equilibrium fraction of phases by MatCalc is used to show the complex precipitation characteristics in this alloy. Fig. 6 shows that MC carbides form during solidification at 1320 °C, followed by γ′, which precipitate when the solidified layer cools to 1140 °C. This indicates that all deposited layers might contain a negligible amount of these precipitates before subsequent layer deposition, while being at the 1000 °C build temperature or during cooling to RT. The phase diagram also shows that the equilibrium fraction of the γ′ increases as temperature decreases. For instance, at 1000, 900, and 800 °C, the phase fractions are ∼30%, 38%, and 42%, respectively.

Fig. 6

Deposition of subsequent layers causes previous layers to undergo phase transformations as they are exposed to several thermal cycles with different peak temperatures. In Fig. 5c, as the subsequent layer is being deposited, γ′ in the previous layer (459th layer) begins to dissolve as the temperature crosses the solvus temperature. This is witnessed by the reduction of the γ′ phase fraction. This graph also shows how this phase dissolves during heating. However, the phase fraction of MC carbide remains stable at high temperatures and no dissolution is seen during thermal cycling. Upon cooling, the γ′ that was dissolved during heating reprecipitates with a surge in the phase fraction until 1000 °C, after which it remains constant. This microstructure is similar to the solidification microstructure (layer 460), with a similar γ′ phase fraction (∼27%).

The complete dissolution and reprecipitation of γ′ continue for several cycles until the 50th layer from the top (layer 411), where the phase fraction does not reach zero during heating to the peak temperature (see Fig. 5d). This indicates the ‘partial’ dissolution of γ′, which continues progressively with additional layers. It should be noted that the peak temperatures for layers that underwent complete dissolution were much higher (1170–1300 °C) than the γ′ solvus.

The dissolution and reprecipitation of γ′ during thermal cycling are further confirmed in Fig. 7, which summarizes the nucleation rate, phase fraction, and concentration of major elements that form γ′ in the matrix. Fig. 7b magnifies a single layer (3rd layer from top) within the full dissolution region in Fig. 7a to help identify the nucleation and growth mechanisms. From Fig. 7b, γ′ nucleation begins during cooling whereby the nucleation rate increases to reach a maximum value of approximately 1 × 1020 m−3s−1. This fast kinetics implies that some rearrangement of atoms is required for γ′ precipitates to form in the matrix [65][66]. The matrix at this stage is in a non-equilibrium condition. Its composition is similar to the nominal composition and remains unchanged. The phase fraction remains insignificant at this stage although nucleation has started. The nucleation rate starts declining upon reaching the peak value. Simultaneously, diffusion-controlled growth of existing nuclei occurs, depleting the matrix of γ′ forming elements (Al and Ti). Thus, from (7)(11), ∆�vol continuously decreases until nucleation ceases. The growth of nuclei is witnessed by the increase in phase fraction until a constant level is reached at 27% upon cooling to and holding at build temperature. This nucleation event is repeated several times.

Fig. 7

At the onset of partial dissolution, the nucleation rate jumps to 1 × 1021 m−3s−1, and then reduces sharply at the middle stage of partial dissolution. The nucleation rate reaches 0 at a later stage. Supplementary Fig. S1 shows a magnified view of the nucleation rate, phase fraction, and thermal profile, underpinning this trend. The jump in nucleation rate at the onset is followed by a progressive reduction in the solute content of the matrix. The peak temperatures (∼1130–1160 °C) are lower than those in complete dissolution regions but still above or close to the γ′ solvus. The maximum phase fraction (∼27%) is similar to that of the complete dissolution regions. At the middle stage, the reduction in nucleation rate is accompanied by a sharp drop in the matrix composition. The γ′ fraction drops to ∼24%, where the peak temperatures of the layers are just below or at γ′ solvus. The phase fraction then increases progressively through the later stage of partial dissolution to ∼30% towards the end of thermal cycling. The matrix solute content continues to drop although no nucleation event is seen. The peak temperatures are then far below the γ′ solvus. It should be noted that the matrix concentration after complete dissolution remains constant. Upon cooling to RT after final layer deposition, the nucleation rate increases again, indicating new nucleation events. The phase fraction reaches ∼40%, with a further depletion of the matrix in major γ′ forming elements.

3.3.2.2. γ′ size distribution

Fig. 8 shows histograms of the γ′ precipitate size distributions (PSD) along the build height during deposition. These PSDs are predicted at the end of each layer of interest just before final cooling to room temperature, to separate the role of thermal cycles from final cooling on the evolution of γ′. The PSD for the top layer (layer 460) is shown in Fig. 8a (last solidified region with solidification microstructure). The γ′ size ranges from 120 to 230 nm and is similar to the 44 layers below (2.2 mm from the top).

Fig. 8

Further down the build, γ′ begins to coarsen after layer 417 (44th layer from top). Fig. 8c shows the PSD after the 44th layer, where the γ′ size exhibits two peaks at ∼120–230 and ∼300 nm, with most of the population being in the former range. This is the onset of partial dissolution where simultaneously with the reprecipitation and growth of fresh γ′, the undissolved γ′ grows rapidly through diffusive transport of atoms to the precipitates. This is shown in Fig. 8c, where the precipitate class sizes between 250 and 350 represent the growth of undissolved γ′. Although this continues in the 416th layer, the phase fractions plot indicates that the onset of partial dissolution begins after the 411th layer. This implies that partial dissolution started early, but the fraction of undissolved γ′ was too low to impact the phase fraction. The reprecipitated γ′ are mostly in the 100–220 nm class range and similar to those observed during full dissolution.

As the number of layers increases, coarsening intensifies with continued growth of more undissolved γ′, and reprecipitation and growth of partially dissolved ones. Fig. 8d, e, and f show this sequence. Further down the build, coarsening progresses rapidly, as shown in Figs. 8d, 8e, and 8f. The γ′ size ranges from 120 to 1100 nm, with the peaks at 160, 180, and 220 nm in Figs. 8d, 8e, and 8f, respectively. Coarsening continues until nucleation ends during dissolution, where only the already formed γ′ precipitates continue to grow during further thermal cycling. The γ′ size at this point is much larger, as observed in layers 361 and 261, and continues to increase steadily towards the bottom (layer 1). Two populations in the ranges of ∼380–700 and ∼750–1100 nm, respectively, can be seen. The steady growth of γ′ towards the bottom is confirmed by the gradual decrease in the concentration of solute elements in the matrix (Fig. 7a). It should be noted that for each layer, the γ′ class with the largest size originates from continuous growth of the earliest set of the undissolved precipitates.

Fig. 9Fig. 10 and supplementary Figs. S2 and S3 show the γ′ size evolution during heating and cooling of a single layer in the full dissolution region, and early, middle stages, and later stages of partial dissolution, respectively. In all, the size of γ′ reduces during layer heating. Depending on the peak temperature of the layer which varies with build height, γ′ are either fully or partially dissolved as mentioned earlier. Upon cooling, the dissolved γ′ reprecipitate.

Fig. 9
Fig. 10

In Fig. 9, those layers that underwent complete dissolution (top layers) were held above γ′ solvus temperature for longer. In Fig. 10, layers at the early stage of partial dissolution spend less time in the γ′ solvus temperature region during heating, leading to incomplete dissolution. In such conditions, smaller precipitates are fully dissolved while larger ones shrink [67]. Layers in the middle stages of partial dissolution have peak temperatures just below or at γ′ solvus, not sufficient to achieve significant γ′ dissolution. As seen in supplementary Fig. S2, only a few smaller γ′ are dissolved back into the matrix during heating, i.e., growth of precipitates is more significant than dissolution. This explains the sharp decrease in concentration of Al and Ti in the matrix in this layer.

The previous sections indicate various phenomena such as an increase in phase fraction, further depletion of matrix composition, and new nucleation bursts during cooling. Analysis of the PSD after the final cooling of the build to room temperature allows a direct comparison to post-printing microstructural characterization. Fig. 11 shows the γ′ size distribution of layer 1 (460th layer from the top) after final cooling to room temperature. Precipitation of secondary γ′ is observed, leading to the multimodal size distribution of secondary and primary γ′. The secondary γ′ size falls within the 10–80 nm range. As expected, a further growth of the existing primary γ′ is also observed during cooling.

Fig. 11
3.3.2.3. γ′ chemistry after deposition

Fig. 12 shows the concentration of the major elements that form γ′ (Al, Ti, and Ni) in the primary and secondary γ′ at the bottom of the build, as calculated by MatCalc. The secondary γ′ has a higher Al content (13.5–14.5 at% Al), compared to 13 at% Al in the primary γ′. Additionally, within the secondary γ′, the smallest particles (∼10 nm) have higher Al contents than larger ones (∼70 nm). In contrast, for the primary γ′, there is no significant variation in the Al content as a function of their size. The Ni concentration in secondary γ′ (71.1–72 at%) is also higher in comparison to the primary γ′ (70 at%). The smallest secondary γ′ (∼10 nm) have higher Ni contents than larger ones (∼70 nm), whereas there is no substantial change in the Ni content of primary γ′, based on their size. As expected, Ti shows an opposite size-dependent variation. It ranges from ∼ 7.7–8.7 at% Ti in secondary γ′ to ∼9.2 at% in primary γ′. Similarly, within the secondary γ′, the smallest (∼10 nm) have lower Al contents than the larger ones (∼70 nm). No significant variation is observed for Ti content in primary γ′.

Fig. 12

4. Discussion

A combined modelling method is utilized to study the microstructural evolution during E-PBF of IN738. The presented results are discussed by examining the precipitation and dissolution mechanism of γ′ during thermal cycling. This is followed by a discussion on the phase fraction and size evolution of γ′ during thermal cycling and after final cooling. A brief discussion on carbide morphology is also made. Finally, a comparison is made between the simulation and experimental results to assess their agreement.

4.1. γ′ morphology as a function of build height

4.1.1. Nucleation of γ′

The fast precipitation kinetics of the γ′ phase enables formation of γ′ upon quenching from higher temperatures (above solvus) during thermal cycling [66]. In Fig. 7b, for a single layer in the full dissolution region, during cooling, the initial increase in nucleation rate signifies the first formation of nuclei. The slight increase in nucleation rate during partial dissolution, despite a decrease in the concentration of γ′ forming elements, may be explained by the nucleation kinetics. During partial dissolution and as the precipitates shrink, it is assumed that the regions at the vicinity of partially dissolved precipitates are enriched in γ′ forming elements [68][69]. This differs from the full dissolution region, in which case the chemical composition is evenly distributed in the matrix. Several authors have attributed the solute supersaturation of the matrix around primary γ′ to partial dissolution during isothermal ageing [69][70][71][72]. The enhanced supersaturation in the regions close to the precipitates results in a much higher driving force for nucleation, leading to a higher nucleation rate upon cooling. This phenomenon can be closely related to the several nucleation bursts upon continuous cooling of Ni-based superalloys, where second nucleation bursts exhibit higher nucleation rates [38][68][73][74].

At middle stages of partial dissolution, the reduction in the nucleation rate indicates that the existing composition and low supersaturation did not trigger nucleation as the matrix was closer to the equilibrium state. The end of a nucleation burst means that the supersaturation of Al and Ti has reached a low level, incapable of providing sufficient driving force during cooling to or holding at 1000 °C for further nucleation [73]. Earlier studies on Ni-based superalloys have reported the same phenomenon during ageing or continuous cooling from the solvus temperature to RT [38][73][74].

4.1.2. Dissolution of γ′ during thermal cycling

γ′ dissolution kinetics during heating are fast when compared to nucleation due to exponential increase in phase transformation and diffusion activities with temperature [65]. As shown in Fig. 9Fig. 10, and supplementary Figs. S2 and S3, the reduction in γ′ phase fraction and size during heating indicates γ′ dissolution. This is also revealed in Fig. 5 where phase fraction decreases upon heating. The extent of γ′ dissolution mostly depends on the temperature, time spent above γ′ solvus, and precipitate size [75][76][77]. Smaller γ′ precipitates are first to be dissolved [67][77][78]. This is mainly because more solute elements need to be transported away from large γ′ precipitates than from smaller ones [79]. Also, a high temperature above γ′ solvus temperature leads to a faster dissolution rate [80]. The equilibrium solvus temperature of γ′ in IN738 in our MatCalc simulation (Fig. 6) and as reported by Ojo et al. [47] is 1140 °C and 1130–1180 °C, respectively. This means the peak temperature experienced by previous layers decreases progressively from γ′ supersolvus to subsolvus, near-solvus, and far from solvus as the number of subsequent layers increases. Based on the above, it can be inferred that the degree of dissolution of γ′ contributes to the gradient in precipitate distribution.

Although the peak temperatures during later stages of partial dissolution are much lower than the equilibrium γ′ solvus, γ′ dissolution still occurs but at a significantly lower rate (supplementary Fig. S3). Wahlmann et al. [28] also reported a similar case where they observed the rapid dissolution of γ′ in CMSX-4 during fast heating and cooling cycles at temperatures below the γ′ solvus. They attributed this to the γ′ phase transformation process taking place in conditions far from the equilibrium. While the same reasoning may be valid for our study, we further believe that the greater surface area to volume ratio of the small γ′ precipitates contributed to this. This ratio means a larger area is available for solute atoms to diffuse into the matrix even at temperatures much below the solvus [81].

4.2. γ′ phase fraction and size evolution

4.2.1. During thermal cycling

In the first layer, the steep increase in γ′ phase fraction during heating (Fig. 5), which also represents γ′ precipitation in the powder before melting, has qualitatively been validated in [28]. The maximum phase fraction of 27% during the first few layers of thermal cycling indicates that IN738 theoretically could reach the equilibrium state (∼30%), but the short interlayer time at the build temperature counteracts this. The drop in phase fraction at middle stages of partial dissolution is due to the low number of γ′ nucleation sites [73]. It has been reported that a reduction of γ′ nucleation sites leads to a delay in obtaining the final volume fraction as more time is required for γ′ precipitates to grow and reach equilibrium [82]. This explains why even upon holding for 150 s before subsequent layer deposition, the phase fraction does not increase to those values that were observed in the previous full γ′ dissolution regions. Towards the end of deposition, the increase in phase fraction to the equilibrium value of 30% is as a result of the longer holding at build temperature or close to it [83].

During thermal cycling, γ′ particles begin to grow immediately after they first precipitate upon cooling. This is reflected in the rapid increase in phase fraction and size during cooling in Fig. 5 and supplementary Fig. S2, respectively. The rapid growth is due to the fast diffusion of solute elements at high temperatures [84]. The similar size of γ′ for the first 44 layers from the top can be attributed to the fact that all layers underwent complete dissolution and hence, experienced the same nucleation event and growth during deposition. This corresponds with the findings by Balikci et al. [85], who reported that the degree of γ′ precipitation in IN738LC does not change when a solution heat treatment is conducted above a certain critical temperature.

The increase in coarsening rate (Fig. 8) during thermal cycling can first be ascribed to the high peak temperature of the layers [86]. The coarsening rate of γ′ is known to increase rapidly with temperature due to the exponential growth of diffusion activity. Also, the simultaneous dissolution with coarsening could be another reason for the high coarsening rate, as γ′ coarsening is a diffusion-driven process where large particles grow by consuming smaller ones [78][84][86][87]. The steady growth of γ′ towards the bottom of the build is due to the much lower layer peak temperature, which is almost close to the build temperature, and reduced dissolution activity, as is seen in the much lower solute concentration in γ′ compared to those in the full and partial dissolution regions.

4.2.2. During cooling

The much higher phase fraction of ∼40% upon cooling signifies the tendency of γ′ to reach equilibrium at lower temperatures (Fig. 4). This is due to the precipitation of secondary γ′ and a further increase in the size of existing primary γ′, which leads to a multimodal size distribution of γ′ after cooling [38][73][88][89][90]. The reason for secondary γ′ formation during cooling is as follows: As cooling progresses, it becomes increasingly challenging to redistribute solute elements in the matrix owing to their lower mobility [38][73]. A higher supersaturation level in regions away from or free of the existing γ′ precipitates is achieved, making them suitable sites for additional nucleation bursts. More cooling leads to the growth of these secondary γ′ precipitates, but as the temperature and in turn, the solute diffusivity is low, growth remains slow.

4.3. Carbides

MC carbides in IN738 are known to have a significant impact on the high-temperature strength. They can also act as effective hardening particles and improve the creep resistance [91]. Precipitation of MC carbides in IN738 and several other superalloys is known to occur during solidification or thermal treatments (e.g., hot isostatic pressing) [92]. In our case, this means that the MC carbides within the E-PBF build formed because of the thermal exposure from the E-PBF thermal cycle in addition to initial solidification. Our simulation confirms this as MC carbides appear during layer heating (Fig. 5). The constant and stable phase fraction of MC carbides during thermal cycling can be attributed to their high melting point (∼1360 °C) and the short holding time at peak temperatures [75][93][94]. The solvus temperature for most MC carbides exceeds most of the peak temperatures observed in our simulation, and carbide dissolution kinetics at temperatures above the solvus are known to be comparably slow [95]. The stable phase fraction and random distribution of MC carbides signifies the slight influence on the gradient in hardness.

4.4. Comparison of simulations and experiments

4.4.1. Precipitate phase fraction and morphology as a function of build height

A qualitative agreement is observed for the phase fraction of carbides, i.e. ∼0.8% in the experiment and ∼0.9% in the simulation. The phase fraction of γ′ differs, with the experiment reporting a value of ∼51% and the simulation, 40%. Despite this, the size distribution of primary γ′ along the build shows remarkable consistency between experimental and computational analyses. It is worth noting that the primary γ′ morphology in the experimental analysis is observed in the as-fabricated state, whereas the simulation (Fig. 8) captures it during deposition process. The primary γ′ size in the experiment is expected to experience additional growth during the cooling phase. Regardless, both show similar trends in primary γ′ size increments from the top to the bottom of the build. The larger primary γ’ size in the simulation versus the experiment can be attributed to the fact that experimental and simulation results are based on 2D and 3D data, respectively. The absence of stereological considerations [96] in our analysis could have led to an underestimation of the precipitate sizes from SEM measurements. The early starts of coarsening (8th layer) in the experiment compared to the simulation (45th layer) can be attributed to a higher actual γ′ solvus temperature than considered in our simulation [47]. The solvus temperature of γ′ in a Ni-based superalloy is mainly determined by the detailed composition. A high amount of Cr and Co are known to reduce the solvus temperature, whereas Ta and Mo will increase it [97][98][99]. The elemental composition from our experimental work was used for the simulation except for Ta. It should be noted that Ta is not included in the thermodynamic database in MatCalc used, and this may have reduced the solvus temperature. This could also explain the relatively higher γ′ phase fraction in the experiment than in simulation, as a higher γ′ solvus temperature will cause more γ′ to precipitate and grow early during cooling [99][100].

Another possible cause of this deviation can be attributed to the extent of γ′ dissolution, which is mainly determined by the peak temperature. It can be speculated that individual peak temperatures at different layers in the simulation may have been over-predicted. However, one needs to consider that the true thermal profile is likely more complicated in the actual E-PBF process [101]. For example, the current model assumes that the thermophysical properties of the material are temperature-independent, which is not realistic. Many materials, including IN738, exhibit temperature-dependent properties such as thermal conductivityspecific heat capacity, and density [102]. This means that heat transfer simulations may underestimate or overestimate the temperature gradients and cooling rates within the powder bed and the solidified part. Additionally, the model does not account for the reduced thermal diffusivity through unmelted powder, where gas separating the powder acts as insulation, impeding the heat flow [1]. In E-PBF, the unmelted powder regions with trapped gas have lower thermal diffusivity compared to the fully melted regions, leading to localized temperature variations, and altered solidification behavior. These limitations can impact the predictions, particularly in relation to the carbide dissolution, as the peak temperatures may be underestimated.

While acknowledging these limitations, it is worth emphasizing that achieving a detailed and accurate representation of each layer’s heat source would impose tough computational challenges. Given the substantial layer count in E-PBF, our decision to employ a semi-analytical approximation strikes a balance between computational feasibility and the capture of essential trends in thermal profiles across diverse build layers. In future work, a dual-calibration strategy is proposed to further reduce simulation-experiment disparities. By refining temperature-independent thermophysical property approximations and absorptivity in the heat source model, and by optimizing interfacial energy descriptions in the kinetic model, the predictive precision could be enhanced. Further refining the simulation controls, such as adjusting the precipitate class size may enhance quantitative comparisons between modeling outcomes and experimental data in future work.

4.4.2. Multimodal size distribution of γ′ and concentration

Another interesting feature that sees qualitative agreement between the simulation and the experiment is the multimodal size distribution of γ′. The formation of secondary γ′ particles in the experiment and most E-PBF Ni-based superalloys is suggested to occur at low temperatures, during final cooling to RT [16][73][90]. However, so far, this conclusion has been based on findings from various continuous cooling experiments, as the study of the evolution during AM would require an in-situ approach. Our simulation unambiguously confirms this in an AM context by providing evidence for secondary γ′ precipitation during slow cooling to RT. Additionally, it is possible to speculate that the chemical segregation occurring during solidification, due to the preferential partitioning of certain elements between the solid and liquid phases, can contribute to the multimodal size distribution during deposition [51]. This is because chemical segregation can result in variations in the local composition of superalloys, which subsequently affects the nucleation and growth of γ′. Regions with higher concentrations of alloying elements will encourage the formation of larger γ′ particles, while regions with lower concentrations may favor the nucleation of smaller precipitates. However, it is important to acknowledge that the elevated temperature during the E-PBF process will largely homogenize these compositional differences [103][104].

A good correlation is also shown in the composition of major γ′ forming elements (Al and Ti) in primary and secondary γ′. Both experiment and simulation show an increasing trend for Al content and a decreasing trend for Ti content from primary to secondary γ′. The slight composition differences between primary and secondary γ′ particles are due to the different diffusivity of γ′ stabilizers at different thermal conditions [105][106]. As the formation of multimodal γ′ particles with different sizes occurs over a broad temperature range, the phase chemistry of γ′ will be highly size dependent. The changes in the chemistry of various γ′ (primary, secondary, and tertiary) have received significant attention since they have a direct influence on the performance [68][105][107][108][109]. Chen et al. [108][109], reported a high Al content in the smallest γ′ precipitates compared to the largest, while Ti showed an opposite trend during continuous cooling in a RR1000 Ni-based superalloy. This was attributed to the temperature and cooling rate at which the γ′ precipitates were formed. The smallest precipitates formed last, at the lowest temperature and cooling rate. A comparable observation is evident in the present investigation, where the secondary γ′ forms at a low temperature and cooling rate in comparison to the primary. The temperature dependence of γ′ chemical composition is further evidenced in supplementary Fig. S4, which shows the equilibrium chemical composition of γ′ as a function of temperature.

5. Conclusions

A correlative modelling approach capable of predicting solid-state phase transformations kinetics in metal AM was developed. This approach involves computational simulations with a semi-analytical heat transfer model and the MatCalc thermo-kinetic software. The method was used to predict the phase transformation kinetics and detailed morphology and chemistry of γ′ and MC during E-PBF of IN738 Ni-based superalloy. The main conclusions are:

  • 1.The computational simulations are in qualitative agreement with the experimental observations. This is particularly true for the γ′ size distribution along the build height, the multimodal size distribution of particles, and the phase fraction of MC carbides.
  • 2.The deviations between simulation and experiment in terms of γ′ phase fraction and location in the build are most likely attributed to a higher γ′ solvus temperature during the experiment than in the simulation, which is argued to be related to the absence of Ta in the MatCalc database.
  • 3.The dissolution and precipitation of γ′ occur fast and under non-equilibrium conditions. The level of γ′ dissolution determines the gradient in γ′ size distribution along the build. After thermal cycling, the final cooling to room temperature has further significant impacts on the final γ′ size, morphology, and distribution.
  • 4.A negligible amount of γ′ forms in the first deposited layer before subsequent layer deposition, and a small amount of γ′ may also form in the powder induced by the 1000 °C elevated build temperature before melting.

Our findings confirm the suitability of MatCalc to predict the microstructural evolution at various positions throughout a build in a Ni-based superalloy during E-PBF. It also showcases the suitability of a tool which was originally developed for traditional thermo-mechanical processing of alloys to the new additive manufacturing context. Our simulation capabilities are likely extendable to other alloy systems that undergo solid-state phase transformations implemented in MatCalc (various steels, Ni-based superalloys, and Al-alloys amongst others) as well as other AM processes such as L-DED and L-PBF which have different thermal cycle characteristics. New tools to predict the microstructural evolution and properties during metal AM are important as they provide new insights into the complexities of AM. This will enable control and design of AM microstructures towards advanced materials properties and performances.

CRediT authorship contribution statement

Primig Sophie: Writing – review & editing, Supervision, Resources, Project administration, Funding acquisition, Conceptualization. Adomako Nana Kwabena: Writing – original draft, Writing – review & editing, Visualization, Software, Investigation, Formal analysis, Conceptualization. Haghdadi Nima: Writing – review & editing, Supervision, Project administration, Methodology, Conceptualization. Dingle James F.L.: Methodology, Conceptualization, Software, Writing – review & editing, Visualization. Kozeschnik Ernst: Writing – review & editing, Software, Methodology. Liao Xiaozhou: Writing – review & editing, Project administration, Funding acquisition. Ringer Simon P: Writing – review & editing, Project administration, Funding acquisition.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

This research was sponsored by the Department of Industry, Innovation, and Science under the auspices of the AUSMURI program – which is a part of the Commonwealth’s Next Generation Technologies Fund. The authors acknowledge the facilities and the scientific and technical assistance at the Electron Microscope Unit (EMU) within the Mark Wainwright Analytical Centre (MWAC) at UNSW Sydney and Microscopy Australia. Nana Adomako is supported by a UNSW Scientia PhD scholarship. Michael Haines’ (UNSW Sydney) contribution to the revised version of the original manuscript is thankfully acknowledged.

Appendix A. Supplementary material

Download : Download Word document (462KB)

Supplementary material.

Data Availability

Data will be made available on request.

References

Study on the critical sediment concentration determining the optimal transport capability of submarine sediment flows with different particle size composition

Study on the critical sediment concentration determining the optimal transport capability of submarine sediment flows with different particle size composition

Yupeng Ren abc, Huiguang Zhou cd, Houjie Wang ab, Xiao Wu ab, Guohui Xu cd, Qingsheng Meng cd

Abstract

해저 퇴적물 흐름은 퇴적물을 심해로 운반하는 주요 수단 중 하나이며, 종종 장거리를 이동하고 수십 또는 수백 킬로미터에 걸쳐 상당한 양의 퇴적물을 운반합니다. 그것의 강력한 파괴력은 종종 이동 과정에서 잠수함 유틸리티에 심각한 손상을 초래합니다.

퇴적물 흐름의 퇴적물 농도는 주변 해수와의 밀도차를 결정하며, 이 밀도 차이는 퇴적물 흐름의 흐름 능력을 결정하여 이송된 퇴적물의 최종 퇴적 위치에 영향을 미칩니다. 본 논문에서는 다양한 미사 및 점토 중량비(미사/점토 비율이라고 함)를 갖는 다양한 퇴적물 농도의 퇴적물 흐름을 수로 테스트를 통해 연구합니다.

우리의 테스트 결과는 특정 퇴적물 구성에 대해 퇴적물 흐름이 가장 빠르게 이동하는 임계 퇴적물 농도가 있음을 나타냅니다. 4가지 미사/점토 비율 각각에 대한 임계 퇴적물 농도와 이에 상응하는 최대 속도가 구해집니다. 결과는 점토 함량이 임계 퇴적물 농도와 선형적으로 음의 상관 관계가 있음을 나타냅니다.

퇴적물 농도가 증가함에 따라 퇴적물의 흐름 거동은 흐름 상태에서 붕괴된 상태로 변환되고 흐름 거동이 변화하는 두 탁한 현탁액의 유체 특성은 모두 Bingham 유체입니다.

또한 본 논문에서는 퇴적물 흐름 내 입자 배열을 분석하여 위에서 언급한 결과에 대한 미시적 설명도 제공합니다.

Submarine sediment flows is one of the main means for transporting sediment to the deep sea, often traveling long-distance and transporting significant volumes of sediment for tens or even hundreds of kilometers. Its strong destructive force often causes serious damage to submarine utilities on its course of movement. The sediment concentration of the sediment flow determines its density difference with the ambient seawater, and this density difference determines the flow ability of the sediment flow, and thus affects the final deposition locations of the transported sediment. In this paper, sediment flows of different sediment concentration with various silt and clay weight ratios (referred to as silt/clay ratio) are studied using flume tests. Our test results indicate that there is a critical sediment concentration at which sediment flows travel the fastest for a specific sediment composition. The critical sediment concentrations and their corresponding maximum velocities for each of the four silt/clay ratios are obtained. The results further indicate that the clay content is linearly negatively correlated with the critical sediment concentration. As the sediment concentration increases, the flow behaviors of sediment flows transform from the flow state to the collapsed state, and the fluid properties of the two turbid suspensions with changing flow behaviors are both Bingham fluids. Additionally, this paper also provides a microscopic explanation of the above-mentioned results by analyzing the arrangement of particles within the sediment flow.

Introduction

Submarine sediment flows are important carriers for sea floor sediment movement and may carry and transport significant volumes of sediment for tens or even hundreds of kilometers (Prior et al., 1987; Pirmez and Imran, 2003; Zhang et al., 2018). Earthquakes, storms, and floods may all trigger submarine sediment flow events (Hsu et al., 2008; Piper and Normark, 2009; Pope et al., 2017b; Gavey et al., 2017). Sediment flows have strong forces during the movement, which will cause great harm to submarine structures such as cables and pipelines (Pope et al., 2017a). It was first confirmed that the cable breaking event caused by the sediment flow occurred in 1929. The sediment flow triggered by the Grand Banks earthquake damaged 12 cables. According to the time sequence of the cable breaking, the maximum velocity of the sediment flow is as high as 28 m/s (Heezen and Ewing, 1952; Kuenen, 1952; Heezen et al., 1954). Subsequent research shows that the lowest turbidity velocity that can break the cable also needs to reach 19 m/s (Piper et al., 1988). Since then, there have been many damage events of submarine cables and oil and gas pipelines caused by sediment flows in the world (Hsu et al., 2008; Carter et al., 2012; Cattaneo et al., 2012; Carter et al., 2014). During its movement, the sediment flow will gradually deposit a large amount of sediment carried by it along the way, that is, the deposition process of the sediment flow. On the one hand, this process brings a large amount of terrestrial nutrients and other materials to the ocean, while on the other hand, it causes damage and burial to benthic organisms, thus forming the largest sedimentary accumulation on Earth – submarine fans, which are highly likely to become good reservoirs for oil and gas resources (Daly, 1936; Yuan et al., 2010; Wu et al., 2022). The study on sediment flows (such as, the study of flow velocity and the forces acting on seabed structures) can provide important references for the safe design of seabed structures, the protection of submarine ecosystems, and exploration of turbidity sediments related oil and gas deposits. Therefore, it is of great significance to study the movement of sediment flows.

The sediment flow, as a highly sediment-concentrated fluid flowing on the sea floor, has a dense bottom layer and a dilute turbulent cloud. Observations at the Monterey Canyon indicated that the sediment flow can maintain its movement over long distances if its bottom has a relatively high sediment concentration. This dense bottom layer can be very destructive along its movement path to any facilities on the sea floor (Paull et al., 2018; Heerema et al., 2020; Wang et al., 2020). The sediment flow mentioned in this research paper is the general term of sediment density flow.

The sediment flow, which occurs on the seafloor, has the potential to cause erosion along its path. In this process, the suspended sediment is replenished, allowing the sediment flow to maintain its continuous flow capacity (Zhao et al., 2018). The dynamic force of sediment flow movement stem from its own gravity and density difference with surrounding water. In cases that the gravity drive of the slope is absent (on a flat sea floor), the flow velocity and distance of sediment flows are essentially determined by the sediment composition and concentration of the sediment flows as previous studies have demonstrated. Ilstad et al. (2004) conducted underwater flow tests in a sloped tank and employed high speed video camera to perform particle tracking. The results indicated that the premixed sand-rich and clay-rich slurries demonstrated different flow velocity and flow behavior. Using mixed kaolinite(d50 = 6 μm) and silica flour(d50 = 9 μm) in three compositions with total volumetric concentration ranged 22% or 28%, Felix and Peakall (2006) carried out underwater flow tests in a 5° slope Perspex channel and found that the flow ability of sediment flows is different depending on sediment compositions and concentrations. Sumner et al. (2009) used annular flume experiments to investigate the depositional dynamics and deposits of waning sediment-laden flows, finding that decelerating fast flows with fixed sand content and variable mud content resulted in four different deposit types. Chowdhury and Testik (2011) used lock-exchange tank, and experimented the kaolin clay sediment flows in the concentration range of 25–350 g/L, and predicted the fluid mud sediment flows propagation characteristics, but this study focused on giving sediment flows propagate phase transition time parameters, and is limited to clay. Lv et al. (2017) found through experiments that the rheological properties and flow behavior of kaolin clay (d50 = 3.7 μm) sediment flows were correlated to clay concentrations. In the field monitoring conducted by Liu et al. (2023) at the Manila Trench in the South China Sea in 2021, significant differences in the velocity, movement distance, and flow morphology of turbidity currents were observed. These differences may be attributed to variations in the particle composition of the turbidity currents.

On low and gentle slopes, although sediment flow with sand as the main sediment composition moves faster, it is difficult to propagate over long distances because sand has greater settling velocity and subaqueous angle of repose. Whereas the sediment flows with silt and clay as main composition may maintain relatively stable currents. Although its movement speed is slow, it has the ability to propagate over long distances because of the low settling rate of the fine particles (Ilstad et al., 2004; Liu et al., 2023). In a field observation at the Gaoping submarine canyon, the sediments collected from the sediment flows exhibited grain size gradation and the sediment was mostly composed of silt and clay (Liu et al., 2012). At the largest deltas in the world, for instance, the Mississippi River Delta, the sediments are mainly composed of silt and clay, which generally distributed along the coast in a wide range and provided the sediment sources for further distribution. The sediment flows originated and transported sediment from the coast to the deep sea are therefore share the same sediment compositions as delta sediments. To study the sediment flows composed of silt and clay is of great importance.

The sediment concentration of the sediment flows determines the density difference between the sediment flows and the ambient water and plays a key role in its flow ability. For the sediment flow with sediment composed of silt and clay, low sediment concentration means low density and therefore leads to low flow ability; however, although high sediment concentration results in high density, since there is cohesion between fine particles, it changes fluid properties and leads to low flow ability as well. Therefore, there should be a critical sediment concentration with mixed composition of silt and clay, at which the sediment flow maintains its strongest flow capacity and have the highest movement speed. In other words, the two characteristics of particle diameter and concentration of the sediment flow determine its own motion ability, which, if occurs, may become the most destructive force to submarine structures.

The objectives of this work was to study how the sediment composition (measured in relative weight of silt and clay, and referred as silt/clay ratio) and sediment concentration affect flow ability and behavior of the sediment flows, and to quantify the critical sediment concentration at which the sediment flows reached the greatest flow velocity under the experiment setting. We used straight flume without slope and conducted a series of flume tests with varying sediment compositions (silt-rich or clay-rich) and concentrations (96 to 1212 g/L). Each sediment flow sample was tested and analyzed for rheological properties using a rheometer, in order to characterize the relationship between flow behavior and rheological properties. Combined with the particle diameter, density and viscosity characteristics of the sediment flows measured in the experiment, a numerical modeling study is conducted, which are mutually validated with the experimental results.

The sediment concentration determines the arrangements of the sediment particles in the turbid suspension, and the arrangement impacts the fluid properties of the turbid suspension. The microscopic mode of particle arrangement in the turbid suspension can be constructed to further analyze the relationship between the fluid properties of turbid suspension and the flow behaviors of the sediment flow, and then characterize the critical sediment concentration at which the sediment flow runs the fastest. A simplified microscopic model of particle arrangement in turbid suspension was constructed to analyze the microscopic arrangement characteristics of sediment particles in turbid suspension with the fastest velocity.

Section snippets

Equipment and materials

The sediment flows flow experiments were performed in a Perspex channel with smooth transparent walls. The layout and dimensions of the experimental set-up were shown in Fig. 1. The bottom of the channel was flat and straight, and a gate was arranged to separate the two tanks. In order to study the flow capacity of turbidity currents from the perspective of their own composition (particle size distribution and concentration), we used a straight channel instead of an inclined one, to avoid any

Relationship between sediment flow flow velocity and sediment concentration

After the sediment flow is generated, its movement in the first half (50 cm) of the channel is relatively stable, and there is obvious shock diffusion in the second half. The reason is that the excitation wave (similar to the surge) will be formed during the sediment flow movement, and its speed is much faster than the speed of the sediment flow head. When the excitation wave reaches the tail of the channel, it will be reflected, thus affecting the subsequent flow of the sediment flow.

Sediment flows motion simulation based on FLOW-3D

As a relatively mature 3D fluid simulation software, FLOW-3D can accurately predict the free surface flow, and has been used to simulate the movement process of sediment flows for many times (Heimsund, 2007). The model adopted in this paper is RNG turbulence model, which can better deal with the flow with high strain rate and is suitable for the simulation of sediment flows with variable shape during movement. The governing equations of the numerical model involved include continuity equation,

Conclusions

In this study, we conducted a series of sediment flow flume tests with mixed silt and clay sediment samples in four silt/clay ratios on a flat slope. Rheological measurements were carried out on turbid suspension samples and microstructure analysis of the sediment particle arrangements was conducted, we concluded that:

  • (1)The flow velocity of the sediment flow is controlled by the sediment concentration and its own particle diameter composition, the flow velocity increased with the increase of the

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

This work was supported by the National Natural Science Foundation of China [Grant no. 42206055]; the National Natural Science Foundation of China [Grant no. 41976049]; and the National Natural Science Foundation of China [Grant no. 42272327].

References (39)

There are more references available in the full text version of this article.

Figure 11. Sketch of scour mechanism around USAF under random waves.

Scour Characteristics and Equilibrium Scour Depth Prediction around Umbrella Suction Anchor Foundation under Random Waves

by Ruigeng Hu 1,Hongjun Liu 2,Hao Leng 1,Peng Yu 3 andXiuhai Wang 1,2,*

1College of Environmental Science and Engineering, Ocean University of China, Qingdao 266000, China

2Key Lab of Marine Environment and Ecology (Ocean University of China), Ministry of Education, Qingdao 266000, China

3Qingdao Geo-Engineering Survering Institute, Qingdao 266100, China

*Author to whom correspondence should be addressed.

J. Mar. Sci. Eng. 20219(8), 886; https://doi.org/10.3390/jmse9080886

Received: 6 July 2021 / Revised: 8 August 2021 / Accepted: 13 August 2021 / Published: 17 August 2021

(This article belongs to the Section Ocean Engineering)

Download 

Abstract

A series of numerical simulation were conducted to study the local scour around umbrella suction anchor foundation (USAF) under random waves. In this study, the validation was carried out firstly to verify the accuracy of the present model. Furthermore, the scour evolution and scour mechanism were analyzed respectively. In addition, two revised models were proposed to predict the equilibrium scour depth Seq around USAF. At last, a parametric study was carried out to study the effects of the Froude number Fr and Euler number Eu for the Seq. The results indicate that the present numerical model is accurate and reasonable for depicting the scour morphology under random waves. The revised Raaijmakers’s model shows good agreement with the simulating results of the present study when KCs,p < 8. The predicting results of the revised stochastic model are the most favorable for n = 10 when KCrms,a < 4. The higher Fr and Eu both lead to the more intensive horseshoe vortex and larger Seq.

Keywords: 

scournumerical investigationrandom wavesequilibrium scour depthKC number

1. Introduction

The rapid expansion of cities tends to cause social and economic problems, such as environmental pollution and traffic jam. As a kind of clean energy, offshore wind power has developed rapidly in recent years. The foundation of offshore wind turbine (OWT) supports the upper tower, and suffers the cyclic loading induced by waves, tides and winds, which exerts a vital influence on the OWT system. The types of OWT foundation include the fixed and floating foundation, and the fixed foundation was used usually for nearshore wind turbine. After the construction of fixed foundation, the hydrodynamic field changes in the vicinity of the foundation, leading to the horseshoe vortex formation and streamline compression at the upside and sides of foundation respectively [1,2,3,4]. As a result, the neighboring soil would be carried away by the shear stress induced by vortex, and the scour hole would emerge in the vicinity of foundation. The scour holes increase the cantilever length, and weaken the lateral bearing capacity of foundation [5,6,7,8,9]. Moreover, the natural frequency of OWT system increases with the increase of cantilever length, causing the resonance occurs when the system natural frequency equals the wave or wind frequency [10,11,12]. Given that, an innovative foundation called umbrella suction anchor foundation (USAF) has been designed for nearshore wind power. The previous studies indicated the USAF was characterized by the favorable lateral bearing capacity with the low cost [6,13,14]. The close-up of USAF is shown in Figure 1, and it includes six parts: 1-interal buckets, 2-external skirt, 3-anchor ring, 4-anchor branch, 5-supporting rod, 6-telescopic hook. The detailed description and application method of USAF can be found in reference [13].

Jmse 09 00886 g001 550

Figure 1. The close-up of umbrella suction anchor foundation (USAF).

Numerical and experimental investigations of scour around OWT foundation under steady currents and waves have been extensively studied by many researchers [1,2,15,16,17,18,19,20,21,22,23,24]. The seabed scour can be classified as two types according to Shields parameter θ, i.e., clear bed scour (θ < θcr) or live bed scour (θ > θcr). Due to the set of foundation, the adverse hydraulic pressure gradient exists at upstream foundation edges, resulting in the streamline separation between boundary layer flow and seabed. The separating boundary layer ascended at upstream anchor edges and developed into the horseshoe vortex. Then, the horseshoe vortex moved downstream gradually along the periphery of the anchor, and the vortex shed off continually at the lee-side of the anchor, i.e., wake vortex. The core of wake vortex is a negative pressure center, liking a vacuum cleaner. Hence, the soil particles were swirled into the negative pressure core and carried away by wake vortexes. At the same time, the onset of scour at rear side occurred. Finally, the wake vortex became downflow when the turbulence energy could not support the survival of wake vortex. According to Tavouktsoglou et al. [25], the scale of pile wall boundary layer is proportional to 1/ln(Rd) (Rd is pile Reynolds), which means the turbulence intensity induced by the flow-structure interaction would decrease with Rd increases, but the effects of Rd can be neglected only if the flow around the foundation is fully turbulent [26]. According to previous studies [1,15,27,28,29,30,31,32], the scour development around pile foundation under waves was significantly influenced by Shields parameter θ and KC number simultaneously (calculated by Equation (1)). Sand ripples widely existed around pile under waves in the case of live bed scour, and the scour morphology is related with θ and KC. Compared with θKC has a greater influence on the scour morphology [21,27,28]. The influence mechanism of KC on the scour around the pile is reflected in two aspects: the horseshoe vortex at upstream and wake vortex shedding at downstream.

KC=UwmTD��=�wm��(1)

where, Uwm is the maximum velocity of the undisturbed wave-induced oscillatory flow at the sea bottom above the wave boundary layer, T is wave period, and D is pile diameter.

There are two prerequisites to satisfy the formation of horseshoe vortex at upstream pile edges: (1) the incoming flow boundary layer with sufficient thickness and (2) the magnitude of upstream adverse pressure gradient making the boundary layer separating [1,15,16,18,20]. The smaller KC results the lower adverse pressure gradient, and the boundary layer cannot separate, herein, there is almost no horseshoe vortex emerging at upside of pile. Sumer et al. [1,15] carried out several sets of wave flume experiments under regular and irregular waves respectively, and the experiment results show that there is no horseshoe vortex when KC is less than 6. While the scale and lifespan of horseshoe vortex increase evidently with the increase of KC when KC is larger than 6. Moreover, the wake vortex contributes to the scour at lee-side of pile. Similar with the case of horseshoe vortex, there is no wake vortex when KC is less than 6. The wake vortex is mainly responsible for scour around pile when KC is greater than 6 and less than O(100), while horseshoe vortex controls scour nearly when KC is greater than O(100).

Sumer et al. [1] found that the equilibrium scour depth was nil around pile when KC was less than 6 under regular waves for live bed scour, while the equilibrium scour depth increased with the increase of KC. Based on that, Sumer proposed an equilibrium scour depth predicting equation (Equation (2)). Carreiras et al. [33] revised Sumer’s equation with m = 0.06 for nonlinear waves. Different with the findings of Sumer et al. [1] and Carreiras et al. [33], Corvaro et al. [21] found the scour still occurred for KC ≈ 4, and proposed the revised equilibrium scour depth predicting equation (Equation (3)) for KC > 4.

Rudolph and Bos [2] conducted a series of wave flume experiments to investigate the scour depth around monopile under waves only, waves and currents combined respectively, indicting KC was one of key parameters in influencing equilibrium scour depth, and proposed the equilibrium scour depth predicting equation (Equation (4)) for low KC (1 < KC < 10). Through analyzing the extensive data from published literatures, Raaijmakers and Rudolph [34] developed the equilibrium scour depth predicting equation (Equation (5)) for low KC, which was suitable for waves only, waves and currents combined. Khalfin [35] carried out several sets of wave flume experiments to study scour development around monopile, and proposed the equilibrium scour depth predicting equation (Equation (6)) for low KC (0.1 < KC < 3.5). Different with above equations, the Khalfin’s equation considers the Shields parameter θ and KC number simultaneously in predicting equilibrium scour depth. The flow reversal occurred under through in one wave period, so sand particles would be carried away from lee-side of pile to upside, resulting in sand particles backfilled into the upstream scour hole [20,29]. Considering the backfilling effects, Zanke et al. [36] proposed the equilibrium scour depth predicting equation (Equation (7)) around pile by theoretical analysis, and the equation is suitable for the whole range of KC number under regular waves and currents combined.

S/D=1.3(1−exp([−m(KC−6)])�/�=1.3(1−exp(−�(��−6))(2)

where, m = 0.03 for linear waves.

S/D=1.3(1−exp([−0.02(KC−4)])�/�=1.3(1−exp(−0.02(��−4))(3)

S/D=1.3γKwaveKhw�/�=1.3��wave�ℎw(4)

where, γ is safety factor, depending on design process, typically γ = 1.5, Kwave is correction factor considering wave action, Khw is correction factor considering water depth.

S/D=1.5[tanh(hwD)]KwaveKhw�/�=1.5tanh(ℎw�)�wave�ℎw(5)

where, hw is water depth.

S/D=0.0753(θθcr−−−√−0.5)0.69KC0.68�/�=0.0753(��cr−0.5)0.69��0.68(6)

where, θ is shields parameter, θcr is critical shields parameter.

S/D=2.5(1−0.5u/uc)xrelxrel=xeff/(1+xeff)xeff=0.03(1−0.35ucr/u)(KC−6)⎫⎭⎬⎪⎪�/�=2.5(1−0.5�/��)��������=����/(1+����)����=0.03(1−0.35�cr/�)(��−6)(7)

where, u is near-bed orbital velocity amplitude, uc is critical velocity corresponding the onset of sediment motion.

S/D=1.3{1−exp[−0.03(KC2lnn+36)1/2−6]}�/�=1.31−exp−0.03(��2ln�+36)1/2−6(8)

where, n is the 1/n’th highest wave for random waves

For predicting equilibrium scour depth under irregular waves, i.e., random waves, Sumer and Fredsøe [16] found it’s suitable to take Equation (2) to predict equilibrium scour depth around pile under random waves with the root-mean-square (RMS) value of near-bed orbital velocity amplitude Um and peak wave period TP to calculate KC. Khalfin [35] recommended the RMS wave height Hrms and peak wave period TP were used to calculate KC for Equation (6). References [37,38,39,40] developed a series of stochastic theoretical models to predict equilibrium scour depth around pile under random waves, nonlinear random waves plus currents respectively. The stochastic approach thought the 1/n’th highest wave were responsible for scour in vicinity of pile under random waves, and the KC was calculated in Equation (8) with Um and mean zero-crossing wave period Tz. The results calculated by Equation (8) agree well with experimental values of Sumer and Fredsøe [16] if the 1/10′th highest wave was used. To author’s knowledge, the stochastic approach proposed by Myrhaug and Rue [37] is the only theoretical model to predict equilibrium scour depth around pile under random waves for the whole range of KC number in published documents. Other methods of predicting scour depth under random waves are mainly originated from the equation for regular waves-only, waves and currents combined, which are limited to the large KC number, such as KC > 6 for Equation (2) and KC > 4 for Equation (3) respectively. However, situations with relatively low KC number (KC < 4) often occur in reality, for example, monopile or suction anchor for OWT foundations in ocean environment. Moreover, local scour around OWT foundations under random waves has not yet been investigated fully. Therefore, further study are still needed in the aspect of scour around OWT foundations with low KC number under random waves. Given that, this study presents the scour sediment model around umbrella suction anchor foundation (USAF) under random waves. In this study, a comparison of equilibrium scour depth around USAF between this present numerical models and the previous theoretical models and experimental results was presented firstly. Then, this study gave a comprehensive analysis for the scour mechanisms around USAF. After that, two revised models were proposed according to the model of Raaijmakers and Rudolph [34] and the stochastic model developed by Myrhaug and Rue [37] respectively to predict the equilibrium scour depth. Finally, a parametric study was conducted to study the effects of the Froude number (Fr) and Euler number (Eu) to equilibrium scour depth respectively.

2. Numerical Method

2.1. Governing Equations of Flow

The following equations adopted in present model are already available in Flow 3D software. The authors used these theoretical equations to simulate scour in random waves without modification. The incompressible viscous fluid motion satisfies the Reynolds-averaged Navier-Stokes (RANS) equation, so the present numerical model solves RANS equations:

∂u∂t+1VF(uAx∂u∂x+vAy∂u∂y+wAz∂u∂z)=−1ρf∂p∂x+Gx+fx∂�∂�+1��(���∂�∂�+���∂�∂�+���∂�∂�)=−1�f∂�∂�+��+��(9)

∂v∂t+1VF(uAx∂v∂x+vAy∂v∂y+wAz∂v∂z)=−1ρf∂p∂y+Gy+fy∂�∂�+1��(���∂�∂�+���∂�∂�+���∂�∂�)=−1�f∂�∂�+��+��(10)

∂w∂t+1VF(uAx∂w∂x+vAy∂w∂y+wAz∂w∂z)=−1ρf∂p∂z+Gz+fz∂�∂�+1��(���∂�∂�+���∂�∂�+���∂�∂�)=−1�f∂�∂�+��+��(11)

where, VF is the volume fraction; uv, and w are the velocity components in xyz direction respectively with Cartesian coordinates; Ai is the area fraction; ρf is the fluid density, fi is the viscous fluid acceleration, Gi is the fluid body acceleration (i = xyz).

2.2. Turbulent Model

The turbulence closure is available by the turbulent model, such as one-equation, the one-equation k-ε model, the standard k-ε model, RNG k-ε turbulent model and large eddy simulation (LES) model. The LES model requires very fine mesh grid, so the computational time is large, which hinders the LES model application in engineering. The RNG k-ε model can reduce computational time greatly with high accuracy in the near-wall region. Furthermore, the RNG k-ε model computes the maximum turbulent mixing length dynamically in simulating sediment scour model. Therefore, the RNG k-ε model was adopted to study the scour around anchor under random waves [41,42].

∂kT∂T+1VF(uAx∂kT∂x+vAy∂kT∂y+wAz∂kT∂z)=PT+GT+DiffkT−εkT∂��∂�+1��(���∂��∂�+���∂��∂�+���∂��∂�)=��+��+������−���(12)

∂εT∂T+1VF(uAx∂εT∂x+vAy∂εT∂y+wAz∂εT∂z)=CDIS1εTkT(PT+CDIS3GT)+Diffε−CDIS2ε2TkT∂��∂�+1��(���∂��∂�+���∂��∂�+���∂��∂�)=����1����(��+����3��)+�����−����2��2��(13)

where, kT is specific kinetic energy involved with turbulent velocity, GT is the turbulent energy generated by buoyancy; εT is the turbulent energy dissipating rate, PT is the turbulent energy, Diffε and DiffkT are diffusion terms associated with VFAiCDIS1CDIS2 and CDIS3 are dimensionless parameters, and CDIS1CDIS3 have default values of 1.42, 0.2 respectively. CDIS2 can be obtained from PT and kT.

2.3. Sediment Scour Model

The sand particles may suffer four processes under waves, i.e., entrainment, bed load transport, suspended load transport, and deposition, so the sediment scour model should depict the above processes efficiently. In present numerical simulation, the sediment scour model includes the following aspects:

2.3.1. Entrainment and Deposition

The combination of entrainment and deposition determines the net scour rate of seabed in present sediment scour model. The entrainment lift velocity of sand particles was calculated as [43]:

ulift,i=αinsd0.3∗(θ−θcr)1.5∥g∥di(ρi−ρf)ρf−−−−−−−−−−−−√�lift,i=�����*0.3(�−�cr)1.5���(��−�f)�f(14)

where, αi is the entrainment parameter, ns is the outward point perpendicular to the seabed, d* is the dimensionless diameter of sand particles, which was calculated by Equation (15), θcr is the critical Shields parameter, g is the gravity acceleration, di is the diameter of sand particles, ρi is the density of seabed species.

d∗=di(∥g∥ρf(ρi−ρf)μ2f)1/3�*=��(��f(��−�f)�f2)1/3(15)

where μf is the fluid dynamic viscosity.

In Equation (14), the entrainment parameter αi confirms the rate at which sediment erodes when the given shear stress is larger than the critical shear stress, and the recommended value 0.018 was adopted according to the experimental data of Mastbergen and Von den Berg [43]. ns is the outward pointing normal to the seabed interface, and ns = (0,0,1) according to the Cartesian coordinates used in present numerical model.

The shields parameter was obtained from the following equation:

θ=U2f,m(ρi/ρf−1)gd50�=�f,m2(��/�f−1)��50(16)

where, Uf,m is the maximum value of the near-bed friction velocity; d50 is the median diameter of sand particles. The detailed calculation procedure of θ was available in Soulsby [44].

The critical shields parameter θcr was obtained from the Equation (17) [44]

θcr=0.31+1.2d∗+0.055[1−exp(−0.02d∗)]�cr=0.31+1.2�*+0.0551−exp(−0.02�*)(17)

The sand particles begin to deposit on seabed when the turbulence energy weaken and cann’t support the particles suspending. The setting velocity of the particles was calculated from the following equation [44]:

usettling,i=νfdi[(10.362+1.049d3∗)0.5−10.36]�settling,�=�f��(10.362+1.049�*3)0.5−10.36(18)

where νf is the fluid kinematic viscosity.

2.3.2. Bed Load Transport

This is called bed load transport when the sand particles roll or bounce over the seabed and always have contact with seabed. The bed load transport velocity was computed by [45]:

ubedload,i=qb,iδicb,ifb�bedload,�=�b,����b,��b(19)

where, qb,i is the bed load transport rate, which was obtained from Equation (20), δi is the bed load thickness, which was calculated by Equation (21), cb,i is the volume fraction of sand i in the multiple species, fb is the critical packing fraction of the seabed.

qb,i=8[∥g∥(ρi−ρfρf)d3i]1/2�b,�=8�(��−�f�f)��31/2(20)

δi=0.3d0.7∗(θθcr−1)0.5di��=0.3�*0.7(��cr−1)0.5��(21)

2.3.3. Suspended Load Transport

Through the following transport equation, the suspended sediment concentration could be acquired.

∂Cs,i∂t+∇(us,iCs,i)=∇∇(DfCs,i)∂�s,�∂�+∇(�s,��s,�)=∇∇(�f�s,�)(22)

where, Cs,i is the suspended sand particles mass concentration of sand i in the multiple species, us,i is the sand particles velocity of sand iDf is the diffusivity.

The velocity of sand i in the multiple species could be obtained from the following equation:

us,i=u¯¯+usettling,ics,i�s,�=�¯+�settling,��s,�(23)

where, u¯�¯ is the velocity of mixed fluid-particles, which can be calculated by the RANS equation with turbulence model, cs,i is the suspended sand particles volume concentration, which was computed from Equation (24).

cs,i=Cs,iρi�s,�=�s,���(24)

3. Model Setup

The seabed-USAF-wave three-dimensional scour numerical model was built using Flow-3D software. As shown in Figure 2, the model includes sandy seabed, USAF model, sea water, two baffles and porous media. The dimensions of USAF are shown in Table 1. The sandy bed (210 m in length, 30 m in width and 11 m in height) is made up of uniform fine sand with median diameter d50 = 0.041 cm. The USAF model includes upper steel tube with the length of 20 m, which was installed in the middle of seabed. The location of USAF is positioned at 140 m from the upstream inflow boundary and 70 m from the downstream outflow boundary. Two baffles were installed at two ends of seabed. In order to eliminate the wave reflection basically, the porous media was set at the outflow side on the seabed.

Jmse 09 00886 g002 550

Figure 2. (a) The sketch of seabed-USAF-wave three-dimensional model; (b) boundary condation:Wv-wave boundary, S-symmetric boundary, O-outflow boundary; (c) USAF model.

Table 1. Numerical simulating cases.

Table

3.1. Mesh Geometric Dimensions

In the simulation of the scour under the random waves, the model includes the umbrella suction anchor foundation, seabed and fluid. As shown in Figure 3, the model mesh includes global mesh grid and nested mesh grid, and the total number of grids is 1,812,000. The basic procedure for building mesh grid consists of two steps. Step 1: Divide the global mesh using regular hexahedron with size of 0.6 × 0.6. The global mesh area is cubic box, embracing the seabed and whole fluid volume, and the dimensions are 210 m in length, 30 m in width and 32 m in height. The details of determining the grid size can see the following mesh sensitivity section. Step 2: Set nested fine mesh grid in vicinity of the USAF with size of 0.3 × 0.3 so as to shorten the computation cost and improve the calculation accuracy. The encryption range is −15 m to 15 m in x direction, −15 m to 15 m in y direction and 0 m to 32 m in z direction, respectively. In order to accurately capture the free-surface dynamics, such as the fluid-air interface, the volume of fluid (VOF) method was adopted for tracking the free water surface. One specific algorithm called FAVORTM (Fractional Area/Volume Obstacle Representation) was used to define the fractional face areas and fractional volumes of the cells which are open to fluid flow.

Jmse 09 00886 g003 550

Figure 3. The sketch of mesh grid.

3.2. Boundary Conditions

As shown in Figure 2, the initial fluid length is 210 m as long as seabed. A wave boundary was specified at the upstream offshore end. The details of determining the random wave spectrum can see the following wave parameters section. The outflow boundary was set at the downstream onshore end. The symmetry boundary was used at the top and two sides of the model. The symmetric boundaries were the better strategy to improve the computation efficiency and save the calculation cost [46]. At the seabed bottom, the wall boundary was adopted, which means the u = v = w= 0. Besides, the upper steel tube of USAF was set as no-slip condition.

3.3. Wave Parameters

The random waves with JONSWAP wave spectrum were used for all simulations as realistic representation of offshore conditions. The unidirectional JONSWAP frequency spectrum was described as [47]:

S(ω)=αg2ω5exp[−54(ωpω)4]γexp[−(ω−ωp)22σ2ω2p]�(�)=��2�5exp−54(�p�)4�exp−(�−�p)22�2�p2(25)

where, α is wave energy scale parameter, which is calculated by Equation (26), ω is frequency, ωp is wave spectrum peak frequency, which can be obtained from Equation (27). γ is wave spectrum peak enhancement factor, in this study γ = 3.3. σ is spectral width factor, σ equals 0.07 for ω ≤ ωp and 0.09 for ω > ωp respectively.

α=0.0076(gXU2)−0.22�=0.0076(���2)−0.22(26)

ωp=22(gU)(gXU2)−0.33�p=22(��)(���2)−0.33(27)

where, X is fetch length, U is average wind velocity at 10 m height from mean sea level.

In present numerical model, the input key parameters include X and U for wave boundary with JONSWAP wave spectrum. The objective wave height and period are available by different combinations of X and U. In this study, we designed 9 cases with different wave heights, periods and water depths for simulating scour around USAF under random waves (see Table 2). For random waves, the wave steepness ε and Ursell number Ur were acquired form Equations (28) and (29) respectively

ε=2πgHsT2a�=2���s�a2(28)

Ur=Hsk2h3w�r=�s�2ℎw3(29)

where, Hs is significant wave height, Ta is average wave period, k is wave number, hw is water depth. The Shield parameter θ satisfies θ > θcr for all simulations in current study, indicating the live bed scour prevails.

Table 2. Numerical simulating cases.

Table

3.4. Mesh Sensitivity

In this section, a mesh sensitivity analysis was conducted to investigate the influence of mesh grid size to results and make sure the calculation is mesh size independent and converged. Three mesh grid size were chosen: Mesh 1—global mesh grid size of 0.75 × 0.75, nested fine mesh grid size of 0.4 × 0.4, and total number of grids 1,724,000, Mesh 2—global mesh grid size of 0.6 × 0.6, nested fine mesh grid size of 0.3 × 0.3, and total number of grids 1,812,000, Mesh 3—global mesh grid size of 0.4 × 0.4, nested fine mesh grid size of 0.2 × 0.2, and total number of grids 1,932,000. The near-bed shear velocity U* is an important factor for influencing scour process [1,15], so U* at the position of (4,0,11.12) was evaluated under three mesh sizes. As the Figure 4 shown, the maximum error of shear velocity ∆U*1,2 is about 39.8% between the mesh 1 and mesh 2, and 4.8% between the mesh 2 and mesh 3. According to the mesh sensitivity criterion adopted by Pang et al. [48], it’s reasonable to think the results are mesh size independent and converged with mesh 2. Additionally, the present model was built according to prototype size, and the mesh size used in present model is larger than the mesh size adopted by Higueira et al. [49] and Corvaro et al. [50]. If we choose the smallest cell size, it will take too much time. For example, the simulation with Mesh3 required about 260 h by using a computer with Intel Xeon Scalable Gold 4214 CPU @24 Cores, 2.2 GHz and 64.00 GB RAM. Therefore, in this case, considering calculation accuracy and computation efficiency, the mesh 2 was chosen for all the simulation in this study.

Jmse 09 00886 g004 550

Figure 4. Comparison of near-bed shear velocity U* with different mesh grid size.

The nested mesh block was adopted for seabed in vicinity of the USAF, which was overlapped with the global mesh block. When two mesh blocks overlap each other, the governing equations are by default solved on the mesh block with smaller average cell size (i.e., higher grid resolution). It is should be noted that the Flow 3D software used the moving mesh captures the scour evolution and automatically adjusts the time step size to be as large as possible without exceeding any of the stability limits, affecting accuracy, or unduly increasing the effort required to enforce the continuity condition [51].

3.5. Model Validation

In order to verify the reliability of the present model, the results of present study were compared with the experimental data of Khosronejad et al. [52]. The experiment was conducted in an open channel with a slender vertical pile under unidirectional currents. The comparison of scour development between the present results and the experimental results is shown in Figure 5. The Figure 5 reveals that the present results agree well with the experimental data of Khosronejad et al. [52]. In the first stage, the scour depth increases rapidly. After that, the scour depth achieves a maximum value gradually. The equilibrium scour depth calculated by the present model is basically corresponding with the experimental results of Khosronejad et al. [52], although scour depth in the present model is slightly larger than the experimental results at initial stage.

Jmse 09 00886 g005 550

Figure 5. Comparison of time evolution of scour between the present study and Khosronejad et al. [52], Petersen et al. [17].

Secondly, another comparison was further conducted between the results of present study and the experimental data of Petersen et al. [17]. The experiment was carried out in a flume with a circular vertical pile in combined waves and current. Figure 4 shows a comparison of time evolution of scour depth between the simulating and the experimental results. As Figure 5 indicates, the scour depth in this study has good overall agreement with the experimental results proposed in Petersen et al. [17]. The equilibrium scour depth calculated by the present model is 0.399 m, which equals to the experimental value basically. Overall, the above verifications prove the present model is accurate and capable in dealing with sediment scour under waves.

In addition, in order to calibrate and validate the present model for hydrodynamic parameters, the comparison of water surface elevation was carried out with laboratory experiments conducted by Stahlmann [53] for wave gauge No. 3. The Figure 6 depicts the surface wave profiles between experiments and numerical model results. The comparison indicates that there is a good agreement between the model results and experimental values, especially the locations of wave crest and trough. Comparison of the surface elevation instructs the present model has an acceptable relative error, and the model is a calibrated in terms of the hydrodynamic parameters.

Jmse 09 00886 g006 550

Figure 6. Comparison of surface elevation between the present study and Stahlmann [53].

Finally, another comparison was conducted for equilibrium scour depth or maximum scour depth under random waves with the experimental data of Sumer and Fredsøe [16] and Schendel et al. [22]. The Figure 7 shows the comparison between the numerical results and experimental data of Run01, Run05, Run21 and Run22 in Sumer and Fredsøe [16] and test A05 and A09 in Schendel et al. [22]. As shown in Figure 7, the equilibrium scour depth or maximum scour depth distributed within the ±30 error lines basically, meaning the reliability and accuracy of present model for predicting equilibrium scour depth around foundation in random waves. However, compared with the experimental values, the present model overestimated the equilibrium scour depth generally. Given that, a calibration for scour depth was carried out by multiplying the mean reduced coefficient 0.85 in following section.

Jmse 09 00886 g007 550

Figure 7. Comparison of equilibrium (or maximum) scour depth between the present study and Sumer and Fredsøe [16], Schendel et al. [22].

Through the various examination for hydrodynamic and morphology parameters, it can be concluded that the present model is a validated and calibrated model for scour under random waves. Thus, the present numerical model would be utilized for scour simulation around foundation under random waves.

4. Numerical Results and Discussions

4.1. Scour Evolution

Figure 8 displays the scour evolution for case 1–9. As shown in Figure 8a, the scour depth increased rapidly at the initial stage, and then slowed down at the transition stage, which attributes to the backfilling occurred in scour holes under live bed scour condition, resulting in the net scour decreasing. Finally, the scour reached the equilibrium state when the amount of sediment backfilling equaled to that of scouring in the scour holes, i.e., the net scour transport rate was nil. Sumer and Fredsøe [16] proposed the following formula for the scour development under waves

St=Seq(1−exp(−t/Tc))�t=�eq(1−exp(−�/�c))(30)

where Tc is time scale of scour process.

Jmse 09 00886 g008 550

Figure 8. Time evolution of scour for case 1–9: (a) Case 1–5; (b) Case 6–9.

The computing time is 3600 s and the scour development curves in Figure 8 kept fluctuating, meaning it’s still not in equilibrium scour stage in these cases. According to Sumer and Fredsøe [16], the equilibrium scour depth can be acquired by fitting the data with Equation (30). From Figure 8, it can be seen that the scour evolution obtained from Equation (30) is consistent with the present study basically at initial stage, but the scour depth predicted by Equation (30) developed slightly faster than the simulating results and the Equation (30) overestimated the scour depth to some extent. Overall, the whole tendency of the results calculated by Equation (30) agrees well with the simulating results of the present study, which means the Equation (30) is applicable to depict the scour evolution around USAF under random waves.

4.2. Scour Mechanism under Random Waves

The scour morphology and scour evolution around USAF are similar under random waves in case 1~9. Taking case 7 as an example, the scour morphology is shown in Figure 9.

Jmse 09 00886 g009 550

Figure 9. Scour morphology under different times for case 7.

From Figure 9, at the initial stage (t < 1200 s), the scour occurred at upstream foundation edges between neighboring anchor branches. The maximum scour depth appeared at the lee-side of the USAF. Correspondingly, the sediments deposited at the periphery of the USAF, and the location of the maximum accretion depth was positioned at an angle of about 45° symmetrically with respect to the wave propagating direction in the lee-side of the USAF. After that, when t > 2400 s, the location of the maximum scour depth shifted to the upside of the USAF at an angle of about 45° with respect to the wave propagating direction.

According to previous studies [1,15,16,19,30,31], the horseshoe vortex, streamline compression and wake vortex shedding were responsible for scour around foundation. The Figure 10 displays the distribution of flow velocity in vicinity of foundation, which reflects the evolving processes of horseshoe vertex.

Jmse 09 00886 g010a 550
Jmse 09 00886 g010b 550

Figure 10. Velocity profile around USAF: (a) Flow runup and down stream at upstream anchor edges; (b) Horseshoe vortex at upstream anchor edges; (c) Flow reversal during wave through stage at lee side.

As shown in Figure 10, the inflow tripped to the upstream edges of the USAF and it was blocked by the upper tube of USAF. Then, the downflow formed the horizontal axis clockwise vortex and rolled on the seabed bypassing the tube, that is, the horseshoe vortex (Figure 11). The Figure 12 displays the turbulence intensity around the tube on the seabed. From Figure 12, it can be seen that the turbulence intensity was high-intensity with respect to the region of horseshoe vortex. This phenomenon occurred because of drastic water flow momentum exchanging in the horseshoe vortex. As a result, it created the prominent shear stress on the seabed, causing the local scour at the upstream edges of USAF. Besides, the horseshoe vortex moved downstream gradually along the periphery of the tube and the wake vortex shed off continually at the lee-side of the USAF, i.e., wake vortex.

Jmse 09 00886 g011 550

Figure 11. Sketch of scour mechanism around USAF under random waves.

Jmse 09 00886 g012 550

Figure 12. Turbulence intensity: (a) Turbulence intensity of horseshoe vortex; (b) Turbulence intensity of wake vortex; (c) Turbulence intensity of accretion area.

The core of wake vortex is a negative pressure center, liking a vacuum cleaner [11,42]. Hence, the soil particles were swirled into the negative pressure core and carried away by wake vortex. At the same time, the onset of scour at rear side occurred. Finally, the wake vortex became downflow at the downside of USAF. As is shown in Figure 12, the turbulence intensity was low where the downflow occurred at lee-side, which means the turbulence energy may not be able to support the survival of wake vortex, leading to accretion happening. As mentioned in previous section, the formation of horseshoe vortex was dependent with adverse pressure gradient at upside of foundation. As shown in Figure 13, the evaluated range of pressure distribution is −15 m to 15 m in x direction. The t = 450 s and t = 1800 s indicate that the wave crest and trough arrived at the upside and lee-side of the foundation respectively, and the t = 350 s was neither the wave crest nor trough. The adverse gradient pressure reached the maximum value at t = 450 s corresponding to the wave crest phase. In this case, it’s helpful for the wave boundary separating fully from seabed, which leads to the formation of horseshoe vortex with high turbulence intensity. Therefore, the horseshoe vortex is responsible for the local scour between neighboring anchor branches at upside of USAF. What’s more, due to the combination of the horseshoe vortex and streamline compression, the maximum scour depth occurred at the upside of the USAF with an angle of about 45° corresponding to the wave propagating direction. This is consistent with the findings of Pang et al. [48] and Sumer et al. [1,15] in case of regular waves. At the wave trough phase (t = 1800 s), the pressure gradient became positive at upstream USAF edges, which hindered the separating of wave boundary from seabed. In the meantime, the flow reversal occurred (Figure 10) and the adverse gradient pressure appeared at downstream USAF edges, but the magnitude of adverse gradient pressure at lee-side was lower than the upstream gradient pressure under wave crest. In this way, the intensity of horseshoe vortex behind the USAF under wave trough was low, which explains the difference of scour depth at upstream and downstream, i.e., the scour asymmetry. In other words, the scour asymmetry at upside and downside of USAF was attributed to wave asymmetry for random waves, and the phenomenon became more evident for nonlinear waves [21]. Briefly speaking, the vortex system at wave crest phase was mainly related to the scour process around USAF under random waves.

Jmse 09 00886 g013 550

Figure 13. Pressure distribution around USAF.

4.3. Equilibrium Scour Depth

The KC number is a key parameter for horseshoe vortex emerging and evolving under waves. According to Equation (1), when pile diameter D is fixed, the KC depends on the maximum near-bed velocity Uwm and wave period T. For random waves, the Uwm can be denoted by the root-mean-square (RMS) value of near-bed velocity amplitude Uwm,rms or the significant value of near-bed velocity amplitude Uwm,s. The Uwm,rms and Uwm,s for all simulating cases of the present study are listed in Table 3 and Table 4. The T can be denoted by the mean up zero-crossing wave period Ta, peak wave period Tp, significant wave period Ts, the maximum wave period Tm, 1/10′th highest wave period Tn = 1/10 and 1/5′th highest wave period Tn = 1/5 for random waves, so the different combinations of Uwm and T will acquire different KC. The Table 3 and Table 4 list 12 types of KC, for example, the KCrms,s was calculated by Uwm,rms and Ts. Sumer and Fredsøe [16] conducted a series of wave flume experiments to investigate the scour depth around monopile under random waves, and found the equilibrium scour depth predicting equation (Equation (2)) for regular waves was applicable for random waves with KCrms,p. It should be noted that the Equation (2) is only suitable for KC > 6 under regular waves or KCrms,p > 6 under random waves.

Table 3. Uwm,rms and KC for case 1~9.

Table

Table 4. Uwm,s and KC for case 1~9.

Table

Raaijmakers and Rudolph [34] proposed the equilibrium scour depth predicting model (Equation (5)) around pile under waves, which is suitable for low KC. The format of Equation (5) is similar with the formula proposed by Breusers [54], which can predict the equilibrium scour depth around pile at different scour stages. In order to verify the applicability of Raaijmakers’s model for predicting the equilibrium scour depth around USAF under random waves, a validation of the equilibrium scour depth Seq between the present study and Raaijmakers’s equation was conducted. The position where the scour depth Seq was evaluated is the location of the maximum scour depth, and it was depicted in Figure 14. The Figure 15 displays the comparison of Seq with different KC between the present study and Raaijmakers’s model.

Jmse 09 00886 g014 550

Figure 14. Sketch of the position where the Seq was evaluated.

Jmse 09 00886 g015a 550
Jmse 09 00886 g015b 550

Figure 15. Comparison of the equilibrium scour depth between the present model and the model of Raaijmakers and Rudolph [34]: (aKCrms,sKCrms,a; (bKCrms,pKCrms,m; (cKCrms,n = 1/10KCrms,n = 1/5; (dKCs,sKCs,a; (eKCs,pKCs,m; (fKCs,n = 1/10KCs,n = 1/5.

As shown in Figure 15, there is an error in predicting Seq between the present study and Raaijmakers’s model, and Raaijmakers’s model underestimates the results generally. Although the error exists, the varying trend of Seq with KC obtained from Raaijmakers’s model is consistent with the present study basically. What’s more, the error is minimum and the Raaijmakers’s model is of relatively high accuracy for predicting scour around USAF under random waves by using KCs,p. Based on this, a further revision was made to eliminate the error as much as possible, i.e., add the deviation value ∆S/D in the Raaijmakers’s model. The revised equilibrium scour depth predicting equation based on Raaijmakers’s model can be written as

S′eq/D=1.95[tanh(hD)](1−exp(−0.012KCs,p))+ΔS/D�eq′/�=1.95tanh(ℎ�)(1−exp(−0.012��s,p))+∆�/�(31)

As the Figure 16 shown, through trial-calculation, when ∆S/D = 0.05, the results calculated by Equation (31) show good agreement with the simulating results of the present study. The maximum error is about 18.2% and the engineering requirements have been met basically. In order to further verify the accuracy of the revised model for large KC (KCs,p > 4) under random waves, a validation between the revised model and the previous experimental results [21]. The experiment was conducted in a flume (50 m in length, 1.0 m in width and 1.3 m in height) with a slender vertical pile (D = 0.1 m) under random waves. The seabed is composed of 0.13 m deep layer of sand with d50 = 0.6 mm and the water depth is 0.5 m for all tests. The significant wave height is 0.12~0.21 m and the KCs,p is 5.52~11.38. The comparison between the predicting results by Equation (31) and the experimental results of Corvaro et al. [21] is shown in Figure 17. From Figure 17, the experimental data evenly distributes around the predicted results and the prediction accuracy is favorable when KCs,p < 8. However, the gap between the predicting results and experimental data becomes large and the Equation (31) overestimates the equilibrium scour depth to some extent when KCs,p > 8.

Jmse 09 00886 g016 550

Figure 16. Comparison of Seq between the simulating results and the predicting values by Equation (31).

Jmse 09 00886 g017 550

Figure 17. Comparison of Seq/D between the Experimental results of Corvaro et al. [21] and the predicting values by Equation (31).

In ocean environment, the waves are composed of a train of sinusoidal waves with different frequencies and amplitudes. The energy of constituent waves with very large and very small frequencies is relatively low, and the energy of waves is mainly concentrated in a certain range of moderate frequencies. Myrhaug and Rue [37] thought the 1/n’th highest wave was responsible for scour and proposed the stochastic model to predict the equilibrium scour depth around pile under random waves for full range of KC. Noteworthy is that the KC was denoted by KCrms,a in the stochastic model. To verify the application of the stochastic model for predicting scour depth around USAF, a validation between the simulating results of present study and predicting results by the stochastic model with n = 2,3,5,10,20,500 was carried out respectively.

As shown in Figure 18, compared with the simulating results, the stochastic model underestimates the equilibrium scour depth around USAF generally. Although the error exists, the varying trend of Seq with KCrms,a obtained from the stochastic model is consistent with the present study basically. What’s more, the gap between the predicting values by stochastic model and the simulating results decreases with the increase of n, but for large n, for example n = 500, the varying trend diverges between the predicting values and simulating results, meaning it’s not feasible only by increasing n in stochastic model to predict the equilibrium scour depth around USAF.

Jmse 09 00886 g018 550

Figure 18. Comparison of Seq between the simulating results and the predicting values by Equation (8).

The Figure 19 lists the deviation value ∆Seq/D′ between the predicting values and simulating results with different KCrms,a and n. Then, fitted the relationship between the ∆S′and n under different KCrms,a, and the fitting curve can be written by Equation (32). The revised stochastic model (Equation (33)) can be acquired by adding ∆Seq/D′ to Equation (8).

ΔSeq/D=0.052*exp(−n/6.566)+0.068∆�eq/�=0.052*exp(−�/6.566)+0.068(32)

S′eq¯/D=S′eq/D+0.052*exp(−n/6.566)+0.068�eq′¯/�=�eq′/�+0.052*exp(−�/6.566)+0.068(33)

Jmse 09 00886 g019 550

Figure 19. The fitting line between ∆S′and n.

The comparison between the predicting results by Equation (33) and the simulating results of present study is shown in Figure 20. According to the Figure 20, the varying trend of Seq with KCrms,a obtained from the stochastic model is consistent with the present study basically. Compared with predicting results by the stochastic model, the results calculated by Equation (33) is favorable. Moreover, comparison with simulating results indicates that the predicting results are the most favorable for n = 10, which is consistent with the findings of Myrhaug and Rue [37] for equilibrium scour depth predicting around slender pile in case of random waves.

Jmse 09 00886 g020 550

Figure 20. Comparison of Seq between the simulating results and the predicting values by Equation (33).

In order to further verify the accuracy of the Equation (33) for large KC (KCrms,a > 4) under random waves, a validation was conducted between the Equation (33) and the previous experimental results of Sumer and Fredsøe [16] and Corvaro et al. [21]. The details of experiments conducted by Corvaro et al. [21] were described in above section. Sumer and Fredsøe [16] investigated the local scour around pile under random waves. The experiments were conducted in a wave basin with a slender vertical pile (D = 0.032, 0.055 m). The seabed is composed of 0.14 m deep layer of sand with d50 = 0.2 mm and the water depth was maintained at 0.5 m. The JONSWAP wave spectrum was used and the KCrms,a was 5.29~16.95. The comparison between the predicting results by Equation (33) and the experimental results of Sumer and Fredsøe [16] and Corvaro et al. [21] are shown in Figure 21. From Figure 21, contrary to the case of low KCrms,a (KCrms,a < 4), the error between the predicting values and experimental results increases with decreasing of n for KCrms,a > 4. Therefore, the predicting results are the most favorable for n = 2 when KCrms,a > 4.

Jmse 09 00886 g021 550

Figure 21. Comparison of Seq between the experimental results of Sumer and Fredsøe [16] and Corvaro et al. [21] and the predicting values by Equation (33).

Noteworthy is that the present model was built according to prototype size, so the errors between the numerical results and experimental data of References [16,21] may be attribute to the scale effects. In laboratory experiments on scouring process, it is typically impossible to ensure a rigorous similarity of all physical parameters between the model and prototype structure, leading to the scale effects in the laboratory experiments. To avoid a cohesive behaviour, the bed material was not scaled geometrically according to model scale. As a consequence, the relatively large-scaled sediments sizes may result in the overestimation of bed load transport and underestimation of suspended load transport compared with field conditions. What’s more, the disproportional scaled sediment presumably lead to the difference of bed roughness between the model and prototype, and thus large influences for wave boundary layer on the seabed and scour process. Besides, according to Corvaro et al. [21] and Schendel et al. [55], the pile Reynolds numbers and Froude numbers both affect the scour depth for the condition of non fully developed turbulent flow in laboratory experiments.

4.4. Parametric Study

4.4.1. Influence of Froude Number

As described above, the set of foundation leads to the adverse pressure gradient appearing at upstream, leading to the wave boundary layer separating from seabed, then horseshoe vortex formatting and the horseshoe vortex are mainly responsible for scour around foundation (see Figure 22). The Froude number Fr is the key parameter to influence the scale and intensity of horseshoe vortex. The Fr under waves can be calculated by the following formula [42]

Fr=UwgD−−−√�r=�w��(34)

where Uw is the mean water particle velocity during 1/4 cycle of wave oscillation, obtained from the following formula. Noteworthy is that the root-mean-square (RMS) value of near-bed velocity amplitude Uwm,rms is used for calculating Uwm.

Uw=1T/4∫0T/4Uwmsin(t/T)dt=2πUwm�w=1�/4∫0�/4�wmsin(�/�)��=2��wm(35)

Jmse 09 00886 g022 550

Figure 22. Sketch of flow field at upstream USAF edges.

Tavouktsoglou et al. [25] proposed the following formula between Fr and the vertical location of the stagnation y

yh∝Fer�ℎ∝�r�(36)

where e is constant.

The Figure 23 displays the relationship between Seq/D and Fr of the present study. In order to compare with the simulating results, the experimental data of Corvaro et al. [21] was also depicted in Figure 23. As shown in Figure 23, the equilibrium scour depth appears a logarithmic increase as Fr increases and approaches the mathematical asymptotic value, which is also consistent with the experimental results of Corvaro et al. [21]. According to Figure 24, the adverse pressure gradient pressure at upstream USAF edges increases with the increase of Fr, which is benefit for the wave boundary layer separating from seabed, resulting in the high-intensity horseshoe vortex, hence, causing intensive scour around USAF. Based on the previous study of Tavouktsoglou et al. [25] for scour around pile under currents, the high Fr leads to the stagnation point is closer to the mean sea level for shallow water, causing the stronger downflow kinetic energy. As mentioned in previous section, the energy of downflow at upstream makes up the energy of the subsequent horseshoe vortex, so the stronger downflow kinetic energy results in the more intensive horseshoe vortex. Therefore, the higher Fr leads to the more intensive horseshoe vortex by influencing the position of stagnation point y presumably. Qi and Gao [19] carried out a series of flume tests to investigate the scour around pile under regular waves, and proposed the fitting formula between Seq/D and Fr as following

lg(Seq/D)=Aexp(B/Fr)+Clg(�eq/�)=�exp(�/�r)+�(37)

where AB and C are constant.

Jmse 09 00886 g023 550

Figure 23. The fitting curve between Seq/D and Fr.

Jmse 09 00886 g024 550

Figure 24. Sketch of adverse pressure gradient at upstream USAF edges.

Took the Equation (37) to fit the simulating results with A = −0.002, B = 0.686 and C = −0.808, and the results are shown in Figure 23. From Figure 23, the simulating results evenly distribute around the Equation (37) and the varying trend of Seq/D and Fr in present study is consistent with Equation (37) basically, meaning the Equation (37) is applicable to express the relationship of Seq/D with Fr around USAF under random waves.

4.4.2. Influence of Euler Number

The Euler number Eu is the influencing factor for the hydrodynamic field around foundation. The Eu under waves can be calculated by the following formula. The Eu can be represented by the Equation (38) for uniform cylinders [25]. The root-mean-square (RMS) value of near-bed velocity amplitude Um,rms is used for calculating Um.

Eu=U2mgD�u=�m2��(38)

where Um is depth-averaged flow velocity.

The Figure 25 displays the relationship between Seq/D and Eu of the present study. In order to compare with the simulating results, the experimental data of Sumer and Fredsøe [16] and Corvaro et al. [21] were also plotted in Figure 25. As shown in Figure 25, similar with the varying trend of Seq/D and Fr, the equilibrium scour depth appears a logarithmic increase as Eu increases and approaches the mathematical asymptotic value, which is also consistent with the experimental results of Sumer and Fredsøe [16] and Corvaro et al. [21]. According to Figure 24, the adverse pressure gradient pressure at upstream USAF edges increases with the increasing of Eu, which is benefit for the wave boundary layer separating from seabed, inducing the high-intensity horseshoe vortex, hence, causing intensive scour around USAF.

Jmse 09 00886 g025 550

Figure 25. The fitting curve between Seq/D and Eu.

Therefore, the variation of Fr and Eu reflect the magnitude of adverse pressure gradient pressure at upstream. Given that, the Equation (37) also was used to fit the simulating results with A = 8.875, B = 0.078 and C = −9.601, and the results are shown in Figure 25. From Figure 25, the simulating results evenly distribute around the Equation (37) and the varying trend of Seq/D and Eu in present study is consistent with Equation (37) basically, meaning the Equation (37) is also applicable to express the relationship of Seq/D with Eu around USAF under random waves. Additionally, according to the above description of Fr, it can be inferred that the higher Fr and Eu both lead to the more intensive horseshoe vortex by influencing the position of stagnation point y presumably.

5. Conclusions

A series of numerical models were established to investigate the local scour around umbrella suction anchor foundation (USAF) under random waves. The numerical model was validated for hydrodynamic and morphology parameters by comparing with the experimental data of Khosronejad et al. [52], Petersen et al. [17], Sumer and Fredsøe [16] and Schendel et al. [22]. Based on the simulating results, the scour evolution and scour mechanisms around USAF under random waves were analyzed respectively. Two revised models were proposed according to the model of Raaijmakers and Rudolph [34] and the stochastic model developed by Myrhaug and Rue [37] to predict the equilibrium scour depth around USAF under random waves. Finally, a parametric study was carried out with the present model to study the effects of the Froude number Fr and Euler number Eu to the equilibrium scour depth around USAF under random waves. The main conclusions can be described as follows.(1)

The packed sediment scour model and the RNG k−ε turbulence model were used to simulate the sand particles transport processes and the flow field around UASF respectively. The scour evolution obtained by the present model agrees well with the experimental results of Khosronejad et al. [52], Petersen et al. [17], Sumer and Fredsøe [16] and Schendel et al. [22], which indicates that the present model is accurate and reasonable for depicting the scour morphology around UASF under random waves.(2)

The vortex system at wave crest phase is mainly related to the scour process around USAF under random waves. The maximum scour depth appeared at the lee-side of the USAF at the initial stage (t < 1200 s). Subsequently, when t > 2400 s, the location of the maximum scour depth shifted to the upside of the USAF at an angle of about 45° with respect to the wave propagating direction.(3)

The error is negligible and the Raaijmakers’s model is of relatively high accuracy for predicting scour around USAF under random waves when KC is calculated by KCs,p. Given that, a further revision model (Equation (31)) was proposed according to Raaijmakers’s model to predict the equilibrium scour depth around USAF under random waves and it shows good agreement with the simulating results of the present study when KCs,p < 8.(4)

Another further revision model (Equation (33)) was proposed according to the stochastic model established by Myrhaug and Rue [37] to predict the equilibrium scour depth around USAF under random waves, and the predicting results are the most favorable for n = 10 when KCrms,a < 4. However, contrary to the case of low KCrms,a, the predicting results are the most favorable for n = 2 when KCrms,a > 4 by the comparison with experimental results of Sumer and Fredsøe [16] and Corvaro et al. [21].(5)

The same formula (Equation (37)) is applicable to express the relationship of Seq/D with Eu or Fr, and it can be inferred that the higher Fr and Eu both lead to the more intensive horseshoe vortex and larger Seq.

Author Contributions

Conceptualization, H.L. (Hongjun Liu); Data curation, R.H. and P.Y.; Formal analysis, X.W. and H.L. (Hao Leng); Funding acquisition, X.W.; Writing—original draft, R.H. and P.Y.; Writing—review & editing, X.W. and H.L. (Hao Leng); The final manuscript has been approved by all the authors. All authors have read and agreed to the published version of the manuscript.

Funding

This research was funded by the Fundamental Research Funds for the Central Universities (grant number 202061027) and the National Natural Science Foundation of China (grant number 41572247).

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

The data presented in this study are available on request from the corresponding author.

Conflicts of Interest

The authors declare no conflict of interest.

References

  1. Sumer, B.M.; Fredsøe, J.; Christiansen, N. Scour Around Vertical Pile in Waves. J. Waterw. Port. Coast. Ocean Eng. 1992118, 15–31. [Google Scholar] [CrossRef]
  2. Rudolph, D.; Bos, K. Scour around a monopile under combined wave-current conditions and low KC-numbers. In Proceedings of the 6th International Conference on Scour and Erosion, Amsterdam, The Netherlands, 1–3 November 2006; pp. 582–588. [Google Scholar]
  3. Nielsen, A.W.; Liu, X.; Sumer, B.M.; Fredsøe, J. Flow and bed shear stresses in scour protections around a pile in a current. Coast. Eng. 201372, 20–38. [Google Scholar] [CrossRef]
  4. Ahmad, N.; Bihs, H.; Myrhaug, D.; Kamath, A.; Arntsen, Ø.A. Three-dimensional numerical modelling of wave-induced scour around piles in a side-by-side arrangement. Coast. Eng. 2018138, 132–151. [Google Scholar] [CrossRef]
  5. Li, H.; Ong, M.C.; Leira, B.J.; Myrhaug, D. Effects of Soil Profile Variation and Scour on Structural Response of an Offshore Monopile Wind Turbine. J. Offshore Mech. Arct. Eng. 2018140, 042001. [Google Scholar] [CrossRef]
  6. Li, H.; Liu, H.; Liu, S. Dynamic analysis of umbrella suction anchor foundation embedded in seabed for offshore wind turbines. Géoméch. Energy Environ. 201710, 12–20. [Google Scholar] [CrossRef]
  7. Fazeres-Ferradosa, T.; Rosa-Santos, P.; Taveira-Pinto, F.; Vanem, E.; Carvalho, H.; Correia, J.A.F.D.O. Editorial: Advanced research on offshore structures and foundation design: Part 1. Proc. Inst. Civ. Eng. Marit. Eng. 2019172, 118–123. [Google Scholar] [CrossRef]
  8. Chavez, C.E.A.; Stratigaki, V.; Wu, M.; Troch, P.; Schendel, A.; Welzel, M.; Villanueva, R.; Schlurmann, T.; De Vos, L.; Kisacik, D.; et al. Large-Scale Experiments to Improve Monopile Scour Protection Design Adapted to Climate Change—The PROTEUS Project. Energies 201912, 1709. [Google Scholar] [CrossRef][Green Version]
  9. Wu, M.; De Vos, L.; Chavez, C.E.A.; Stratigaki, V.; Fazeres-Ferradosa, T.; Rosa-Santos, P.; Taveira-Pinto, F.; Troch, P. Large Scale Experimental Study of the Scour Protection Damage Around a Monopile Foundation Under Combined Wave and Current Conditions. J. Mar. Sci. Eng. 20208, 417. [Google Scholar] [CrossRef]
  10. Sørensen, S.P.H.; Ibsen, L.B. Assessment of foundation design for offshore monopiles unprotected against scour. Ocean Eng. 201363, 17–25. [Google Scholar] [CrossRef]
  11. Prendergast, L.; Gavin, K.; Doherty, P. An investigation into the effect of scour on the natural frequency of an offshore wind turbine. Ocean Eng. 2015101, 1–11. [Google Scholar] [CrossRef][Green Version]
  12. Fazeres-Ferradosa, T.; Chambel, J.; Taveira-Pinto, F.; Rosa-Santos, P.; Taveira-Pinto, F.; Giannini, G.; Haerens, P. Scour Protections for Offshore Foundations of Marine Energy Harvesting Technologies: A Review. J. Mar. Sci. Eng. 20219, 297. [Google Scholar] [CrossRef]
  13. Yang, Q.; Yu, P.; Liu, Y.; Liu, H.; Zhang, P.; Wang, Q. Scour characteristics of an offshore umbrella suction anchor foundation under the combined actions of waves and currents. Ocean Eng. 2020202, 106701. [Google Scholar] [CrossRef]
  14. Yu, P.; Hu, R.; Yang, J.; Liu, H. Numerical investigation of local scour around USAF with different hydraulic conditions under currents and waves. Ocean Eng. 2020213, 107696. [Google Scholar] [CrossRef]
  15. Sumer, B.M.; Christiansen, N.; Fredsøe, J. The horseshoe vortex and vortex shedding around a vertical wall-mounted cylinder exposed to waves. J. Fluid Mech. 1997332, 41–70. [Google Scholar] [CrossRef]
  16. Sumer, B.M.; Fredsøe, J. Scour around Pile in Combined Waves and Current. J. Hydraul. Eng. 2001127, 403–411. [Google Scholar] [CrossRef]
  17. Petersen, T.U.; Sumer, B.M.; Fredsøe, J. Time scale of scour around a pile in combined waves and current. In Proceedings of the 6th International Conference on Scour and Erosion, Paris, France, 27–31 August 2012. [Google Scholar]
  18. Petersen, T.U.; Sumer, B.M.; Fredsøe, J.; Raaijmakers, T.C.; Schouten, J.-J. Edge scour at scour protections around piles in the marine environment—Laboratory and field investigation. Coast. Eng. 2015106, 42–72. [Google Scholar] [CrossRef]
  19. Qi, W.; Gao, F. Equilibrium scour depth at offshore monopile foundation in combined waves and current. Sci. China Ser. E Technol. Sci. 201457, 1030–1039. [Google Scholar] [CrossRef][Green Version]
  20. Larsen, B.E.; Fuhrman, D.R.; Baykal, C.; Sumer, B.M. Tsunami-induced scour around monopile foundations. Coast. Eng. 2017129, 36–49. [Google Scholar] [CrossRef][Green Version]
  21. Corvaro, S.; Marini, F.; Mancinelli, A.; Lorenzoni, C.; Brocchini, M. Hydro- and Morpho-dynamics Induced by a Vertical Slender Pile under Regular and Random Waves. J. Waterw. Port. Coast. Ocean Eng. 2018144, 04018018. [Google Scholar] [CrossRef]
  22. Schendel, A.; Welzel, M.; Schlurmann, T.; Hsu, T.-W. Scour around a monopile induced by directionally spread irregular waves in combination with oblique currents. Coast. Eng. 2020161, 103751. [Google Scholar] [CrossRef]
  23. Fazeres-Ferradosa, T.; Taveira-Pinto, F.; Romão, X.; Reis, M.; das Neves, L. Reliability assessment of offshore dynamic scour protections using copulas. Wind. Eng. 201843, 506–538. [Google Scholar] [CrossRef]
  24. Fazeres-Ferradosa, T.; Welzel, M.; Schendel, A.; Baelus, L.; Santos, P.R.; Pinto, F.T. Extended characterization of damage in rubble mound scour protections. Coast. Eng. 2020158, 103671. [Google Scholar] [CrossRef]
  25. Tavouktsoglou, N.S.; Harris, J.M.; Simons, R.R.; Whitehouse, R.J.S. Equilibrium Scour-Depth Prediction around Cylindrical Structures. J. Waterw. Port. Coast. Ocean Eng. 2017143, 04017017. [Google Scholar] [CrossRef][Green Version]
  26. Ettema, R.; Melville, B.; Barkdoll, B. Scale Effect in Pier-Scour Experiments. J. Hydraul. Eng. 1998124, 639–642. [Google Scholar] [CrossRef]
  27. Umeda, S. Scour Regime and Scour Depth around a Pile in Waves. J. Coast. Res. Spec. Issue 201164, 845–849. [Google Scholar]
  28. Umeda, S. Scour process around monopiles during various phases of sea storms. J. Coast. Res. 2013165, 1599–1604. [Google Scholar] [CrossRef]
  29. Baykal, C.; Sumer, B.; Fuhrman, D.R.; Jacobsen, N.; Fredsøe, J. Numerical simulation of scour and backfilling processes around a circular pile in waves. Coast. Eng. 2017122, 87–107. [Google Scholar] [CrossRef][Green Version]
  30. Miles, J.; Martin, T.; Goddard, L. Current and wave effects around windfarm monopile foundations. Coast. Eng. 2017121, 167–178. [Google Scholar] [CrossRef][Green Version]
  31. Miozzi, M.; Corvaro, S.; Pereira, F.A.; Brocchini, M. Wave-induced morphodynamics and sediment transport around a slender vertical cylinder. Adv. Water Resour. 2019129, 263–280. [Google Scholar] [CrossRef]
  32. Yu, T.; Zhang, Y.; Zhang, S.; Shi, Z.; Chen, X.; Xu, Y.; Tang, Y. Experimental study on scour around a composite bucket foundation due to waves and current. Ocean Eng. 2019189, 106302. [Google Scholar] [CrossRef]
  33. Carreiras, J.; Larroudé, P.; Seabra-Santos, F.; Mory, M. Wave Scour Around Piles. In Proceedings of the Coastal Engineering 2000, American Society of Civil Engineers (ASCE), Sydney, Australia, 16–21 July 2000; pp. 1860–1870. [Google Scholar]
  34. Raaijmakers, T.; Rudolph, D. Time-dependent scour development under combined current and waves conditions—Laboratory experiments with online monitoring technique. In Proceedings of the 4th International Conference on Scour and Erosion, Tokyo, Japan, 5–7 November 2008; pp. 152–161. [Google Scholar]
  35. Khalfin, I.S. Modeling and calculation of bed score around large-diameter vertical cylinder under wave action. Water Resour. 200734, 357. [Google Scholar] [CrossRef][Green Version]
  36. Zanke, U.C.; Hsu, T.-W.; Roland, A.; Link, O.; Diab, R. Equilibrium scour depths around piles in noncohesive sediments under currents and waves. Coast. Eng. 201158, 986–991. [Google Scholar] [CrossRef]
  37. Myrhaug, D.; Rue, H. Scour below pipelines and around vertical piles in random waves. Coast. Eng. 200348, 227–242. [Google Scholar] [CrossRef]
  38. Myrhaug, D.; Ong, M.C.; Føien, H.; Gjengedal, C.; Leira, B.J. Scour below pipelines and around vertical piles due to second-order random waves plus a current. Ocean Eng. 200936, 605–616. [Google Scholar] [CrossRef]
  39. Myrhaug, D.; Ong, M.C. Random wave-induced onshore scour characteristics around submerged breakwaters using a stochastic method. Ocean Eng. 201037, 1233–1238. [Google Scholar] [CrossRef]
  40. Ong, M.C.; Myrhaug, D.; Hesten, P. Scour around vertical piles due to long-crested and short-crested nonlinear random waves plus a current. Coast. Eng. 201373, 106–114. [Google Scholar] [CrossRef]
  41. Yakhot, V.; Orszag, S.A. Renormalization group analysis of turbulence. I. Basic theory. J. Sci. Comput. 19861, 3–51. [Google Scholar] [CrossRef]
  42. Yakhot, V.; Smith, L.M. The renormalization group, the e-expansion and derivation of turbulence models. J. Sci. Comput. 19927, 35–61. [Google Scholar] [CrossRef]
  43. Mastbergen, D.R.; Berg, J.V.D. Breaching in fine sands and the generation of sustained turbidity currents in submarine canyons. Sedimentology 200350, 625–637. [Google Scholar] [CrossRef]
  44. Soulsby, R. Dynamics of Marine Sands; Thomas Telford Ltd.: London, UK, 1998. [Google Scholar] [CrossRef]
  45. Van Rijn, L.C. Sediment Transport, Part I: Bed Load Transport. J. Hydraul. Eng. 1984110, 1431–1456. [Google Scholar] [CrossRef][Green Version]
  46. Zhang, Q.; Zhou, X.-L.; Wang, J.-H. Numerical investigation of local scour around three adjacent piles with different arrangements under current. Ocean Eng. 2017142, 625–638. [Google Scholar] [CrossRef]
  47. Yu, Y.X.; Liu, S.X. Random Wave and Its Applications to Engineering, 4th ed.; Dalian University of Technology Press: Dalian, China, 2011. [Google Scholar]
  48. Pang, A.; Skote, M.; Lim, S.; Gullman-Strand, J.; Morgan, N. A numerical approach for determining equilibrium scour depth around a mono-pile due to steady currents. Appl. Ocean Res. 201657, 114–124. [Google Scholar] [CrossRef]
  49. Higuera, P.; Lara, J.L.; Losada, I.J. Three-dimensional interaction of waves and porous coastal structures using Open-FOAM®. Part I: Formulation and validation. Coast. Eng. 201483, 243–258. [Google Scholar] [CrossRef]
  50. Corvaro, S.; Crivellini, A.; Marini, F.; Cimarelli, A.; Capitanelli, L.; Mancinelli, A. Experimental and Numerical Analysis of the Hydrodynamics around a Vertical Cylinder in Waves. J. Mar. Sci. Eng. 20197, 453. [Google Scholar] [CrossRef][Green Version]
  51. Flow3D User Manual, version 11.0.3; Flow Science, Inc.: Santa Fe, NM, USA, 2013.
  52. Khosronejad, A.; Kang, S.; Sotiropoulos, F. Experimental and computational investigation of local scour around bridge piers. Adv. Water Resour. 201237, 73–85. [Google Scholar] [CrossRef]
  53. Stahlmann, A. Experimental and Numerical Modeling of Scour at Foundation Structures for Offshore Wind Turbines. Ph.D. Thesis, Franzius-Institute for Hydraulic, Estuarine and Coastal Engineering, Leibniz Universität Hannover, Hannover, Germany, 2013. [Google Scholar]
  54. Breusers, H.N.C.; Nicollet, G.; Shen, H. Local Scour Around Cylindrical Piers. J. Hydraul. Res. 197715, 211–252. [Google Scholar] [CrossRef]
  55. Schendel, A.; Hildebrandt, A.; Goseberg, N.; Schlurmann, T. Processes and evolution of scour around a monopile induced by tidal currents. Coast. Eng. 2018139, 65–84. [Google Scholar] [CrossRef]
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

© 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).

Share and Cite

      

MDPI and ACS Style

Hu, R.; Liu, H.; Leng, H.; Yu, P.; Wang, X. Scour Characteristics and Equilibrium Scour Depth Prediction around Umbrella Suction Anchor Foundation under Random Waves. J. Mar. Sci. Eng. 20219, 886. https://doi.org/10.3390/jmse9080886

AMA Style

Hu R, Liu H, Leng H, Yu P, Wang X. Scour Characteristics and Equilibrium Scour Depth Prediction around Umbrella Suction Anchor Foundation under Random Waves. Journal of Marine Science and Engineering. 2021; 9(8):886. https://doi.org/10.3390/jmse9080886Chicago/Turabian Style

Hu, Ruigeng, Hongjun Liu, Hao Leng, Peng Yu, and Xiuhai Wang. 2021. “Scour Characteristics and Equilibrium Scour Depth Prediction around Umbrella Suction Anchor Foundation under Random Waves” Journal of Marine Science and Engineering 9, no. 8: 886. https://doi.org/10.3390/jmse9080886

Find Other Styles

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

For more information on the journal statistics, click here.

Multiple requests from the same IP address are counted as one view.

Figure 5. Schematic view of flap and support structure [32]

Design Optimization of Ocean Renewable Energy Converter Using a Combined Bi-level Metaheuristic Approach

결합된 Bi-level 메타휴리스틱 접근법을 사용한 해양 재생 에너지 변환기의 설계 최적화

Erfan Amini a1, Mahdieh Nasiri b1, Navid Salami Pargoo a, Zahra Mozhgani c, Danial Golbaz d, Mehrdad Baniesmaeil e, Meysam Majidi Nezhad f, Mehdi Neshat gj, Davide Astiaso Garcia h, Georgios Sylaios i

Abstract

In recent years, there has been an increasing interest in renewable energies in view of the fact that fossil fuels are the leading cause of catastrophic environmental consequences. Ocean wave energy is a renewable energy source that is particularly prevalent in coastal areas. Since many countries have tremendous potential to extract this type of energy, a number of researchers have sought to determine certain effective factors on wave converters’ performance, with a primary emphasis on ambient factors. In this study, we used metaheuristic optimization methods to investigate the effects of geometric factors on the performance of an Oscillating Surge Wave Energy Converter (OSWEC), in addition to the effects of hydrodynamic parameters. To do so, we used CATIA software to model different geometries which were then inserted into a numerical model developed in Flow3D software. A Ribed-surface design of the converter’s flap is also introduced in this study to maximize wave-converter interaction. Besides, a Bi-level Hill Climbing Multi-Verse Optimization (HCMVO) method was also developed for this application. The results showed that the converter performs better with greater wave heights, flap freeboard heights, and shorter wave periods. Additionally, the added ribs led to more wave-converter interaction and better performance, while the distance between the flap and flume bed negatively impacted the performance. Finally, tracking the changes in the five-dimensional objective function revealed the optimum value for each parameter in all scenarios. This is achieved by the newly developed optimization algorithm, which is much faster than other existing cutting-edge metaheuristic approaches.

Keywords

Wave Energy Converter

OSWEC

Hydrodynamic Effects

Geometric Design

Metaheuristic Optimization

Multi-Verse Optimizer

1Introduction

The increase in energy demand, the limitations of fossil fuels, as well as environmental crises, such as air pollution and global warming, are the leading causes of calling more attention to harvesting renewable energy recently [1][2][3]. While still in its infancy, ocean wave energy has neither reached commercial maturity nor technological convergence. In recent decades, remarkable progress has been made in the marine energy domain, which is still in the early stage of development, to improve the technology performance level (TPL) [4][5]and technology readiness level (TRL) of wave energy converters (WECs). This has been achieved using novel modeling techniques [6][7][8][9][10][11][12][13][14] to gain the following advantages [15]: (i) As a source of sustainable energy, it contributes to the mix of energy resources that leads to greater diversity and attractiveness for coastal cities and suppliers. [16] (ii) Since wave energy can be exploited offshore and does not require any land, in-land site selection would be less expensive and undesirable visual effects would be reduced. [17] (iii) When the best layout and location of offshore site are taken into account, permanent generation of energy will be feasible (as opposed to using solar energy, for example, which is time-dependent) [18].

In general, the energy conversion process can be divided into three stages in a WEC device, including primary, secondary, and tertiary stages [19][20]. In the first stage of energy conversion, which is the subject of this study, the wave power is converted to mechanical power by wave-structure interaction (WSI) between ocean waves and structures. Moreover, the mechanical power is transferred into electricity in the second stage, in which mechanical structures are coupled with power take-off systems (PTO). At this stage, optimal control strategies are useful to tune the system dynamics to maximize power output [10][13][12]. Furthermore, the tertiary energy conversion stage revolves around transferring the non-standard AC power into direct current (DC) power for energy storage or standard AC power for grid integration [21][22]. We discuss only the first stage regardless of the secondary and tertiary stages. While Page 1 of 16 WECs include several categories and technologies such as terminators, point absorbers, and attenuators [15][23], we focus on oscillating surge wave energy converters (OSWECs) in this paper due to its high capacity for industrialization [24].

Over the past two decades, a number of studies have been conducted to understand how OSWECs’ structures and interactions between ocean waves and flaps affect converters performance. Henry et al.’s experiment on oscillating surge wave energy converters is considered as one of the most influential pieces of research [25], which demonstrated how the performance of oscillating surge wave energy converters (OSWECs) is affected by seven different factors, including wave period, wave power, flap’s relative density, water depth, free-board of the flap, the gap between the tubes, gap underneath the flap, and flap width. These parameters were assessed in their two models in order to estimate the absorbed energy from incoming waves [26][27]. In addition, Folly et al. investigated the impact of water depth on the OSWECs performance analytically, numerically, and experimentally. According to this and further similar studies, the average annual incident wave power is significantly reduced by water depth. Based on the experimental results, both the surge wave force and the power capture of OSWECs increase in shallow water [28][29]. Following this, Sarkar et al. found that under such circumstances, the device that is located near the coast performs much better than those in the open ocean [30]. On the other hand, other studies are showing that the size of the converter, including height and width, is relatively independent of the location (within similar depth) [31]. Subsequently, Schmitt et al. studied OSWECs numerically and experimentally. In fact, for the simulation of OSWEC, OpenFOAM was used to test the applicability of Reynolds-averaged Navier-Stokes (RANS) solvers. Then, the experimental model reproduced the numerical results with satisfying accuracy [32]. In another influential study, Wang et al. numerically assessed the effect of OSWEC’s width on their performance. According to their findings, as converter width increases, its efficiency decreases in short wave periods while increases in long wave periods [33]. One of the main challenges in the analysis of the OSWEC is the coupled effect of hydrodynamic and geometric variables. As a result, numerous cutting-edge geometry studies have been performed in recent years in order to find the optimal structure that maximizes power output and minimizes costs. Garcia et al. reviewed hull geometry optimization studies in the literature in [19]. In addition, Guo and Ringwood surveyed geometric optimization methods to improve the hydrodynamic performance of OSWECs at the primary stage [14]. Besides, they classified the hull geometry of OSWECs based on Figure 1. Subsequently, Whittaker et al. proposed a different design of OSWEC called Oyster2. There have been three examples of different geometries of oysters with different water depths. Based on its water depth, they determined the width and height of the converter. They also found that in the constant wave period the less the converter’s width, the less power captures the converter has [34]. Afterward, O’Boyle et al. investigated a type of OSWEC called Oyster 800. They compared the experimental and numerical models with the prototype model. In order to precisely reproduce the shape, mass distribution, and buoyancy properties of the prototype, a 40th-scale experimental model has been designed. Overall, all the models were fairly accurate according to the results [35].

Inclusive analysis of recent research avenues in the area of flap geometry has revealed that the interaction-based designs of such converters are emerging as a novel approach. An initiative workflow is designed in the current study to maximizing the wave energy extrication by such systems. To begin with, a sensitivity analysis plays its role of determining the best hydrodynamic values for installing the converter’s flap. Then, all flap dimensions and characteristics come into play to finalize the primary model. Following, interactive designs is proposed to increase the influence of incident waves on the body by adding ribs on both sides of the flap as a novel design. Finally, a new bi-level metaheuristic method is proposed to consider the effects of simultaneous changes in ribs properties and other design parameters. We hope this novel approach will be utilized to make big-scale projects less costly and justifiable. The efficiency of the method is also compared with four well known metaheuristic algorithms and out weight them for this application.

This paper is organized as follows. First, the research methodology is introduced by providing details about the numerical model implementation. To that end, we first introduced the primary model’s geometry and software details. That primary model is later verified with a benchmark study with regard to the flap angle of rotation and water surface elevation. Then, governing equations and performance criteria are presented. In the third part of the paper, we discuss the model’s sensitivity to lower and upper parts width (we proposed a two cross-sectional design for the flap), bottom elevation, and freeboard. Finally, the novel optimization approach is introduced in the final part and compared with four recent metaheuristic algorithms.

2. Numerical Methods

In this section, after a brief introduction of the numerical software, Flow3D, boundary conditions are defined. Afterwards, the numerical model implementation, along with primary model properties are described. Finally, governing equations, as part of numerical process, are discussed.

2.1Model Setup

FLOW-3D is a powerful and comprehensive CFD simulation platform for studying fluid dynamics. This software has several modules to solve many complex engineering problems. In addition, modeling complex flows is simple and effective using FLOW-3D’s robust meshing capabilities [36]. Interaction between fluid and moving objects might alter the computational range. Dynamic meshes are used in our modeling to take these changes into account. At each time step, the computational node positions change in order to adapt the meshing area to the moving object. In addition, to choose mesh dimensions, some factors are taken into account such as computational accuracy, computational time, and stability. The final grid size is selected based on the detailed procedure provided in [37]. To that end, we performed grid-independence testing on a CFD model using three different mesh grid sizes of 0.01, 0.015, and 0.02 meters. The problem geometry and boundary conditions were defined the same, and simulations were run on all three grids under the same conditions. The predicted values of the relevant variable, such as velocity, was compared between the grids. The convergence behavior of the numerical solution was analyzed by calculating the relative L2 norm error between two consecutive grids. Based on the results obtained, it was found that the grid size of 0.02 meters showed the least error, indicating that it provided the most accurate and reliable solution among the three grids. Therefore, the grid size of 0.02 meters was selected as the optimal spatial resolution for the mesh grid.

In this work, the flume dimensions are 10 meters long, 0.1 meters wide, and 2.2 meters high, which are shown in figure2. In addition, input waves with linear characteristics have a height of 0.1 meters and a period of 1.4 seconds. Among the linear wave methods included in this software, RNGk-ε and k- ε are appropriate for turbulence model. The research of Lopez et al. shows that RNGk- ε provides the most accurate simulation of turbulence in OSWECs [21]. We use CATIA software to create the flap primary model and other innovative designs for this project. The flap measures 0.1 m x 0.65 m x 0.360 m in x, y and z directions, respectively. In Figure 3, the primary model of flap and its dimensions are shown. In this simulation, five boundaries have been defined, including 1. Inlet, 2. Outlet, 3. Converter flap, 4. Bed flume, and 5. Water surface, which are shown in figure 2. Besides, to avoid wave reflection in inlet and outlet zones, Flow3D is capable of defining some areas as damping zones, the length of which has to be one to one and a half times the wavelength. Therefore, in the model, this length is considered equal to 2 meters. Furthermore, there is no slip in all the boundaries. In other words, at every single time step, the fluid velocity is zero on the bed flume, while it is equal to the flap velocity on the converter flap. According to the wave theory defined in the software, at the inlet boundary, the water velocity is called from the wave speed to be fed into the model.

2.2Verification

In the current study, we utilize the Schmitt experimental model as a benchmark for verification, which was developed at the Queen’s University of Belfast. The experiments were conducted on the flap of the converter, its rotation, and its interaction with the water surface. Thus, the details of the experiments are presented below based up on the experimental setup’s description [38]. In the experiment, the laboratory flume has a length of 20m and a width of 4.58m. Besides, in order to avoid incident wave reflection, a wave absorption source is devised at the end of the left flume. The flume bed, also, includes two parts with different slops. The flap position and dimensions of the flume can be seen in Figure4. In addition, a wave-maker with 6 paddles is installed at one end. At the opposite end, there is a beach with wire meshes. Additionally, there are 6 indicators to extract the water level elevation. In the flap model, there are three components: the fixed support structure, the hinge, and the flap. The flap measures 0.1m x 0.65m x 0.341m in x, y and z directions, respectively. In Figure5, the details are given [32]. The support structure consists of a 15 mm thick stainless steel base plate measuring 1m by 1.4m, which is screwed onto the bottom of the tank. The hinge is supported by three bearing blocks. There is a foam centerpiece on the front and back of the flap which is sandwiched between two PVC plates. Enabling changes of the flap, three metal fittings link the flap to the hinge. Moreover, in this experiment, the selected wave is generated based on sea wave data at scale 1:40. The wave height and the wave period are equal to 0.038 (m) and 2.0625 (s), respectively, which are tantamount to a wave with a period of 13 (s) and a height of 1.5 (m).

Two distinct graphs illustrate the numerical and experi-mental study results. Figure6 and Figure7 are denoting the angle of rotation of flap and surface elevation in computational and experimental models, respectively. The two figures roughly represent that the numerical and experimental models are a good match. However, for the purpose of verifying the match, we calculated the correlation coefficient (C) and root mean square error (RMSE). According to Figure6, correlation coefficient and RMSE are 0.998 and 0.003, respectively, and in Figure7 correlation coefficient and RMSE are respectively 0.999 and 0.001. Accordingly, there is a good match between the numerical and empirical models. It is worth mentioning that the small differences between the numerical and experimental outputs may be due to the error of the measuring devices and the calibration of the data collection devices.

Including continuity equation and momentum conserva- tion for incompressible fluid are given as [32][39]:(1)

where P represents the pressure, g denotes gravitational acceleration, u represents fluid velocity, and Di is damping coefficient. Likewise, the model uses the same equation. to calculate the fluid velocity in other directions as well. Considering the turbulence, we use the two-equation model of RNGK- ε. These equations are:

(3)��t(��)+����(����)=����[�eff�������]+��-��and(4)���(��)+����(����)=����[�eff�������]+�1�∗����-��2��2�Where �2� and �1� are constants. In addition, �� and �� represent the turbulent Prandtl number of � and k, respectively.

�� also denote the production of turbulent kinetic energy of k under the effect of velocity gradient, which is calculated as follows:(5)��=�eff[�����+�����]�����(6)�eff=�+��(7)�eff=�+��where � is molecular viscosity,�� represents turbulence viscosity, k denotes kinetic energy, and ∊∊ is energy dissipation rate. The values of constant coefficients in the two-equation RNGK ∊-∊ model is as shown in the Table 1 [40].Table 2.

Table 1. Constant coefficients in RNGK- model

Factors�0�1�2������
Quantity0.0124.381.421.681.391.390.084

Table 2. Flap properties

Joint height (m)0.476
Height of the center of mass (m)0.53
Weight (Kg)10.77

It is worth mentioning that the volume of fluid method is used to separate water and air phases in this software [41]. Below is the equation of this method [40].(8)����+����(���)=0where α and 1 − α are portion of water phase and air phase, respectively. As a weighting factor, each fluid phase portion is used to determine the mixture properties. Finally, using the following equations, we calculate the efficiency of converters [42][34][43]:(9)�=14|�|2�+�2+(�+�a)2(�n2-�2)2where �� represents natural frequency, I denotes the inertia of OSWEC, Ia is the added inertia, F is the complex wave force, and B denotes the hydrodynamic damping coefficient. Afterward, the capture factor of the converter is calculated by [44]:(10)��=�1/2��2����gw where �� represents the capture factor, which is the total efficiency of device per unit length of the wave crest at each time step [15], �� represent the dimensional amplitude of the incident wave, w is the flap’s width, and Cg is the group velocity of the incident wave, as below:(11)��=��0·121+2�0ℎsinh2�0ℎwhere �0 denotes the wave number, h is water depth, and H is the height of incident waves.

According to previous sections ∊,����-∊ modeling is used for all models simulated in this section. For this purpose, the empty boundary condition is used for flume walls. In order to preventing wave reflection at the inlet and outlet of the flume, the length of wave absorption is set to be at least one incident wavelength. In addition, the structured mesh is chosen, and the mesh dimensions are selected in two distinct directions. In each model, all grids have a length of 2 (cm) and a height of 1 (cm). Afterwards, as an input of the software for all of the models, we define the time step as 0.001 (s). Moreover, the run time of every simulation is 30 (s). As mentioned before, our primary model is Schmitt model, and the flap properties is given in table2. For all simulations, the flume measures 15 meters in length and 0.65 meters in width, and water depth is equal to 0.335 (m). The flap is also located 7 meters from the flume’s inlet.

Finally, in order to compare the results, the capture factor is calculated for each simulation and compared to the primary model. It is worth mentioning that capture factor refers to the ratio of absorbed wave energy to the input wave energy.

According to primary model simulation and due to the decreasing horizontal velocity with depth, the wave crest has the highest velocity. Considering the fact that the wave’s orbital velocity causes the flap to move, the contact between the upper edge of the flap and the incident wave can enhance its performance. Additionally, the numerical model shows that the dynamic pressure decreases as depth increases, and the hydrostatic pressure increases as depth increases.

To determine the OSWEC design, it is imperative to understand the correlation between the capture factor, wave period, and wave height. Therefore, as it is shown in Figure8, we plot the change in capture factor over the variations in wave period and wave height in 3D and 2D. In this diagram, the first axis features changes in wave period, the second axis displays changes in wave height, and the third axis depicts changes in capture factor. According to our wave properties in the numerical model, the wave period and wave height range from 2 to 14 seconds and 2 to 8 meters, respectively. This is due to the fact that the flap does not oscillate if the wave height is less than 2 (m), and it does not reverse if the wave height is more than 8 (m). In addition, with wave periods more than 14 (s), the wavelength would be so long that it would violate the deep-water conditions, and with wave periods less than 2 (s), the flap would not oscillate properly due to the shortness of wavelength. The results of simulation are shown in Figure 8. As it can be perceived from Figure 8, in a constant wave period, the capture factor is in direct proportion to the wave height. It is because of the fact that waves with more height have more energy to rotate the flap. Besides, in a constant wave height, the capture factor increases when the wave period increases, until a given wave period value. However, the capture factor falls after this point. These results are expected since the flap’s angular displacement is not high in lower wave periods, while the oscillating motion of that is not fast enough to activate the power take-off system in very high wave periods.

As is shown in Figure 9, we plot the change in capture factor over the variations in wave period (s) and water depth (m) in 3D. As it can be seen in this diagram, the first axis features changes in water depth (m), the second axis depicts the wave period (s), and the third axis displays OSWEC’s capture factor. The wave period ranges from 0 to 10 seconds based on our wave properties, which have been adopted from Schmitt’s model, while water depth ranges from 0 to 0.5 meters according to the flume and flap dimensions and laboratory limitations. According to Figure9, for any specific water depth, the capture factor increases in a varying rate when the wave period increases, until a given wave period value. However, the capture factor falls steadily after this point. In fact, the maximum capture factor occurs when the wave period is around 6 seconds. This trend is expected since, in a specific water depth, the flap cannot oscillate properly when the wavelength is too short. As the wave period increases, the flap can oscillate more easily, and consequently its capture factor increases. However, the capture factor drops in higher wave periods because the wavelength is too large to move the flap. Furthermore, in a constant wave period, by changing the water depth, the capture factor does not alter. In other words, the capture factor does not depend on the water depth when it is around its maximum value.

3Sensitivity Analysis

Based on previous studies, in addition to the flap design, the location of the flap relative to the water surface (freeboard) and its elevation relative to the flume bed (flap bottom elevation) play a significant role in extracting energy from the wave energy converter. This study measures the sensitivity of the model to various parameters related to the flap design including upper part width of the flap, lower part width of the flap, the freeboard, and the flap bottom elevation. Moreover, as a novel idea, we propose that the flap widths differ in the lower and upper parts. In Figure10, as an example, a flap with an upper thickness of 100 (mm) and a lower thickness of 50 (mm) and a flap with an upper thickness of 50 (mm) and a lower thickness of 100 (mm) are shown. The influence of such discrepancy between the widths of the upper and lower parts on the interaction between the wave and the flap, or in other words on the capture factor, is evaluated. To do so, other parameters are remained constant, such as the freeboard, the distance between the flap and the flume bed, and the wave properties.

In Figure11, models are simulated with distinct upper and lower widths. As it is clear in this figure, the first axis depicts the lower part width of the flap, the second axis indicates the upper part width of the flap, and the colors represent the capture factor values. Additionally, in order to consider a sufficient range of change, the flap thickness varies from half to double the value of the primary model for each part.

According to this study, the greater the discrepancy in these two parts, the lower the capture factor. It is on account of the fact that when the lower part of the flap is thicker than the upper part, and this thickness difference in these two parts is extremely conspicuous, the inertia against the motion is significant at zero degrees of rotation. Consequently, it is difficult to move the flap, which results in a low capture factor. Similarly, when the upper part of the flap is thicker than the lower part, and this thickness difference in these two parts is exceedingly noticeable, the inertia is so great that the flap can not reverse at the maximum degree of rotation. As the results indicate, the discrepancy can enhance the performance of the converter if the difference between these two parts is around 20%. As it is depicted in the Figure11, the capture factor reaches its own maximum amount, when the lower part thickness is from 5 to 6 (cm), and the upper part thickness is between 6 and 7 (cm). Consequently, as a result of this discrepancy, less material will be used, and therefore there will be less cost.

As illustrated in Figure12, this study examines the effects of freeboard (level difference between the flap top and water surface) and the flap bottom elevation (the distance between the flume bed and flap bottom) on the converter performance. In this diagram, the first axis demonstrates the freeboard and the second axis on the left side displays the flap bottom elevation, while the colors indicate the capture factor. In addition, the feasible range of freeboard is between -15 to 15 (cm) due to the limitation of the numerical model, so that we can take the wave slamming and the overtopping into consideration. Additionally, based on the Schmitt model and its scaled model of 1:40 of the base height, the flap bottom should be at least 9 (cm) high. Since the effect of surface waves is distributed over the depth of the flume, it is imperative to maintain a reasonable flap height exposed to incoming waves. Thus, the maximum flap bottom elevation is limited to 19 (cm). As the Figure12 pictures, at constant negative values of the freeboard, the capture factor is in inverse proportion with the flap bottom elevation, although slightly.

Furthermore, at constant positive values of the freeboard, the capture factor fluctuates as the flap bottom elevation decreases while it maintains an overall increasing trend. This is on account of the fact that increasing the flap bottom elevation creates turbulence flow behind the flap, which encumbers its rotation, as well as the fact that the flap surface has less interaction with the incoming waves. Furthermore, while keeping the flap bottom elevation constant, the capture factor increases by raising the freeboard. This is due to the fact that there is overtopping with adverse impacts on the converter performance when the freeboard is negative and the flap is under the water surface. Besides, increasing the freeboard makes the wave slam more vigorously, which improves the converter performance.

Adding ribs to the flap surface, as shown in Figure13, is a novel idea that is investigated in the next section. To achieve an optimized design for the proposed geometry of the flap, we determine the optimal number and dimensions of ribs based on the flap properties as our decision variables in the optimization process. As an example, Figure13 illustrates a flap with 3 ribs on each side with specific dimensions.

Figure14 shows the flow velocity field around the flap jointed to the flume bed. During the oscillation of the flap, the pressure on the upper and lower surfaces of the flap changes dynamically due to the changing angle of attack and the resulting change in the direction of fluid flow. As the flap moves upwards, the pressure on the upper surface decreases, and the pressure on the lower surface increases. Conversely, as the flap moves downwards, the pressure on the upper surface increases, and the pressure on the lower surface decreases. This results in a cyclic pressure variation around the flap. Under certain conditions, the pressure field around the flap can exhibit significant variations in magnitude and direction, forming vortices and other flow structures. These flow structures can affect the performance of the OSWEC by altering the lift and drag forces acting on the flap.

4Design Optimization

We consider optimizing the design parameters of the flap of converter using a nature-based swarm optimization method, that fall in the category of metaheuristic algorithms [45]. Accordingly, we choose four state-of-the-art algorithms to perform an optimization study. Then, based on their performances to achieve the highest capture factor, one of them will be chosen to be combined with the Hill Climb algorithm to carry out a local search. Therefore, in the remainder of this section, we discuss the search process of each algorithm and visualize their performance and convergence curve as they try to find the best values for decision variables.

4.1. Metaheuristic Approaches

As the first considered algorithm, the Gray Wolf Optimizer (GWO) algorithm simulates the natural leadership and hunting performance of gray wolves which tend to live in colonies. Hunters must obey the alpha wolf, the leader, who is responsible for hunting. Then, the beta wolf is at the second level of the gray wolf hierarchy. A subordinate of alpha wolf, beta stands under the command of the alpha. At the next level in this hierarchy, there are the delta wolves. They are subordinate to the alpha and beta wolves. This category of wolves includes scouts, sentinels, elders, hunters, and caretakers. In this ranking, omega wolves are at the bottom, having the lowest level and obeying all other wolves. They are also allowed to eat the prey just after others have eaten. Despite the fact that they seem less important than others, they are really central to the pack survival. Since, it has been shown that without omega wolves, the entire pack would experience some problems like fighting, violence, and frustration. In this simulation, there are three primary steps of hunting including searching, surrounding, and finally attacking the prey. Mathematically model of gray wolves’ hunting technique and their social hierarchy are applied in determined by optimization. this study. As mentioned before, gray wolves can locate their prey and surround them. The alpha wolf also leads the hunt. Assuming that the alpha, beta, and delta have more knowledge about prey locations, we can mathematically simulate gray wolf hunting behavior. Hence, in addition to saving the top three best solutions obtained so far, we compel the rest of the search agents (also the omegas) to adjust their positions based on the best search agent. Encircling behavior can be mathematically modeled by the following equations: [46].(12)�→=|�→·��→(�)-�→(�)|(13)�→(�+1)=��→(�)-�→·�→(14)�→=2.�2→(15)�→=2�→·�1→-�→Where �→indicates the position vector of gray wolf, ��→ defines the vector of prey, t indicates the current iteration, and �→and �→are coefficient vectors. To force the search agent to diverge from the prey, we use �→ with random values greater than 1 or less than -1. In addition, C→ contains random values in the range [0,2], and �→ 1 and �2→ are random vectors in [0,1]. The second considered technique is the Moth Flame Optimizer (MFO) algorithm. This method revolves around the moths’ navigation mechanism, which is realized by positioning themselves and maintaining a fixed angle relative to the moon while flying. This effective mechanism helps moths to fly in a straight path. However, when the source of light is artificial, maintaining an angle with the light leads to a spiral flying path towards the source that causes the moth’s death [47]. In MFO algorithm, moths and flames are both solutions. The moths are actual search agents that fly in hyper-dimensional space by changing their position vectors, and the flames are considered pins that moths drop when searching the search space [48]. The problem’s variables are the position of moths in the space. Each moth searches around a flame and updates it in case of finding a better solution. The fitness value is the return value of each moth’s fitness (objective) function. The position vector of each moth is passed to the fitness function, and the output of the fitness function is assigned to the corresponding moth. With this mechanism, a moth never loses its best solution [49]. Some attributes of this algorithm are as follows:

  • •It takes different values to converge moth in any point around the flame.
  • •Distance to the flame is lowered to be eventually minimized.
  • •When the position gets closer to the flame, the updated positions around the flame become more frequent.

As another method, the Multi-Verse Optimizer is based on a multiverse theory which proposes there are other universes besides the one in which we all live. According to this theory, there are more than one big bang in the universe, and each big bang leads to the birth of a new universe [50]. Multi-Verse Optimizer (MVO) is mainly inspired by three phenomena in cosmology: white holes, black holes, and wormholes. A white hole has never been observed in our universe, but physicists believe the big bang could be considered a white hole [51]. Black holes, which behave completely in contrast to white holes, attract everything including light beams with their extremely high gravitational force [52]. In the multiverse theory, wormholes are time and space tunnels that allow objects to move instantly between any two corners of a universe (or even simultaneously from one universe to another) [53]. Based on these three concepts, mathematical models are designed to perform exploration, exploitation, and local search, respectively. The concept of white and black holes is implied as an exploration phase, while the concept of wormholes is considered as an exploitation phase by MVO. Additionally, each solution is analogous to a universe, and each variable in the solution represents an object in that universe. Furthermore, each solution is assigned an inflation rate, and the time is used instead of iterations. Following are the universe rules in MVO:

  • •The possibility of having white hole increases with the inflation rate.
  • •The possibility of having black hole decreases with the inflation rate.
  • •Objects tend to pass through black holes more frequently in universes with lower inflation rates.
  • •Regardless of inflation rate, wormholes may cause objects in universes to move randomly towards the best universe. [54]

Modeling the white/black hole tunnels and exchanging objects of universes mathematically was accomplished by using the roulette wheel mechanism. With every iteration, the universes are sorted according to their inflation rates, then, based on the roulette wheel, the one with the white hole is selected as the local extremum solution. This is accomplished through the following steps:

Assume that

(16)���=����1<��(��)����1≥��(��)

Where ��� represents the jth parameter of the ith universe, Ui indicates the ith universe, NI(Ui) is normalized inflation rate of the ith universe, r1 is a random number in [0,1], and j xk shows the jth parameter of the kth universe selected by a roulette wheel selection mechanism [54]. It is assumed that wormhole tunnels always exist between a universe and the best universe formed so far. This mechanism is as follows:(17)���=if�2<���:��+���×((���-���)×�4+���)�3<0.5��-���×((���-���)×�4+���)�3≥0.5����:���where Xj indicates the jth parameter of the best universe formed so far, TDR and WEP are coefficients, where Xj indicates the jth parameter of the best universelbjshows the lower bound of the jth variable, ubj is the upper bound of the jth variable, and r2, r3, and r4 are random numbers in [1][54].

Finally, one of the newest optimization algorithms is WOA. The WOA algorithm simulates the movement of prey and the whale’s discipline when looking for their prey. Among several species, Humpback whales have a specific method of hunting [55]. Humpback whales can recognize the location of prey and encircle it before hunting. The optimal design position in the search space is not known a priori, and the WOA algorithm assumes that the best candidate solution is either the target prey or close to the optimum. This foraging behavior is called the bubble-net feeding method. Two maneuvers are associated with bubbles: upward spirals and double loops. A unique behavior exhibited only by humpback whales is bubble-net feeding. In fact, The WOA algorithm starts with a set of random solutions. At each iteration, search agents update their positions for either a randomly chosen search agent or the best solution obtained so far [56][55]. When the best search agent is determined, the other search agents will attempt to update their positions toward that agent. It is important to note that humpback whales swim around their prey simultaneously in a circular, shrinking circle and along a spiral-shaped path. By using a mathematical model, the spiral bubble-net feeding maneuver is optimized. The following equation represents this behavior:(18)�→(�+1)=�′→·�bl·cos(2��)+�∗→(�)

Where:(19)�′→=|�∗→(�)-�→(�)|

X→(t+ 1) indicates the distance of the it h whale to the prey (best solution obtained so far),� is a constant for defining the shape of the logarithmic spiral, l is a random number in [−1, 1], and dot (.) is an element-by-element multiplication [55].

Comparing the four above-mentioned methods, simulations are run with 10 search agents for 400 iterations. In Figure 15, there are 20 plots the optimal values of different parameters in optimization algorithms. The five parameters of this study are freeboard, bottom elevations, number of ribs on the converter, rib thickness, and rib Height. The optimal value for each was found by optimization algorithms, naming WOA, MVO, MFO, and GWO. By looking through the first row, the freeboard parameter converges to its maximum possible value in the optimization process of GWO after 300 iterations. Similarly, MFO finds the same result as GWO. In contrast, the freeboard converges to its minimum possible value in MVO optimizing process, which indicates positioning the converter under the water. Furthermore, WOA found the optimal value of freeboard as around 0.02 after almost 200 iterations. In the second row, the bottom elevation is found at almost 0.11 (m) in all algorithms; however, the curves follow different trends in each algorithm. The third row shows the number of ribs, where results immediately reveal that it should be over 4. All algorithms coincide at 5 ribs as the optimal number in this process. The fourth row displays the trends of algorithms to find optimal rib thickness. MFO finds the optimal value early and sets it to around 0.022, while others find the same value in higher iterations. Finally, regarding the rib height, MVO, MFO, and GWO state that the optimal value is 0.06 meters, but WOA did not find a higher value than 0.039.

4.2. HCMVO Bi-level Approach

Despite several strong search characteristics of MVO and its high performance in various optimization problems, it suffers from a few deficiencies in local and global search mechanisms. For instance, it is trapped in the local optimum when wormholes stochastically generate many solutions near the best universe achieved throughout iterations, especially in solving complex multimodal problems with high dimensions [57]. Furthermore, MVO needs to be modified by an escaping strategy from the local optima to enhance the global search abilities. To address these shortages, we propose a fast and effective meta-algorithm (HCMVO) to combine MVO with a Random-restart hill-climbing local search. This meta-algorithm uses MVO on the upper level to develop global tracking and provide a range of feasible and proper solutions. The hill-climbing algorithm is designed to develop a comprehensive neighborhood search around the best-found solution proposed by the upper-level (MVO) when MVO is faced with a stagnation issue or falling into a local optimum. The performance threshold is formulated as follows.(20)Δ����THD=∑�=1�����TH��-����TH��-1�where BestTHDis the best-found solution per generation, andM is related to the domain of iterations to compute the average performance of MVO. If the proposed best solution by the local search is better than the initial one, the global best of MVO will be updated. HCMVO iteratively runs hill climbing when the performance of MVO goes down, each time with an initial condition to prepare for escaping such undesirable situations. In order to get a better balance between exploration and exploitation, the search step size linearly decreases as follows:(21)��=��-����Ma�iter��+1where iter and Maxiter are the current iteration and maximum number of evaluation, respectively. �� stands for the step size of the neighborhood search. Meanwhile, this strategy can improve the convergence rate of MVO compared with other algorithms.

Algorithm 1 shows the technical details of the proposed optimization method (HCMVO). The initial solution includes freeboard (�), bottom elevation (�), number of ribs (Nr), rib thickness (�), and rib height(�).

5. Conclusion

The high trend of diminishing worldwide energy resources has entailed a great crisis upon vulnerable societies. To withstand this effect, developing renewable energy technologies can open doors to a more reliable means, among which the wave energy converters will help the coastal residents and infrastructure. This paper set out to determine the optimized design for such devices that leads to the highest possible power output. The main goal of this research was to demonstrate the best design for an oscillating surge wave energy converter using a novel metaheuristic optimization algorithm. In this regard, the methodology was devised such that it argued the effects of influential parameters, including wave characteristics, WEC design, and interaction criteria.

To begin with, a numerical model was developed in Flow 3D software to simulate the response of the flap of a wave energy converter to incoming waves, followed by a validation study based upon a well-reputed experimental study to verify the accuracy of the model. Secondly, the hydrodynamics of the flap was investigated by incorporating the turbulence. The effect of depth, wave height, and wave period are also investigated in this part. The influence of two novel ideas on increasing the wave-converter interaction was then assessed: i) designing a flap with different widths in the upper and lower part, and ii) adding ribs on the surface of the flap. Finally, four trending single-objective metaheuristic optimization methods

Empty CellAlgorithm 1: Hill Climb Multiverse Optimization
01:procedure HCMVO
02:�=30,�=5▹���������������������������������
03:�=〈F1,B1,N,R,H1〉,…〈FN,B2,N,R,HN〉⇒lb1N⩽�⩽ubN
04:Initialize parameters�ER,�DR,�EP,Best�,���ite��▹Wormhole existence probability (WEP)
05:��=����(��)
06:��=Normalize the inflation rate��
07:for iter in[1,⋯,���iter]do
08:for�in[1,⋯,�]do
09:Update�EP,�DR,Black����Index=�
10:for���[1,⋯,�]��
11:�1=����()
12:if�1≤��(��)then
13:White HoleIndex=Roulette�heelSelection(-��)
14:�(Black HoleIndex,�)=��(White HoleIndex,�)
15:end if
16:�2=����([0,�])
17:if�2≤�EPthen
18:�3=����(),�4=����()
19:if�3<0.5then
20:�1=((��(�)-��(�))�4+��(�))
21:�(�,�)=Best�(�)+�DR�
22:else
23:�(�,�)=Best�(�)-�DR�
24:end if
25:end if
26:end for
27:end for
28:�HD=����([�1,�2,⋯,�Np])
29:Bes�TH�itr=����HD
30:ΔBestTHD=∑�=1�BestTII��-BestTII��-1�
31:ifΔBestTHD<��then▹Perform hill climbing local search
32:BestTHD=����-�lim��������THD
33:end if
34:end for
35:return�,BestTHD▹Final configuration
36:end procedure

The implementation details of the hill-climbing algorithm applied in HCMPA can be seen in Algorithm 2. One of the critical parameters isg, which denotes the resolution of the neighborhood search around the proposed global best by MVO. If we set a small step size for hill-climbing, the convergence speed will be decreased. On the other hand, a large step size reinforces the exploration ability. Still, it may reduce the exploitation ability and in return increase the act of jumping from a global optimum or surfaces with high-potential solutions. Per each decision variable, the neighborhood search evaluates two different direct searches, incremental or decremental. After assessing the generated solutions, the best candidate will be selected to iterate the search algorithm. It is noted that the hill-climbing algorithm should not be applied in the initial iteration of the optimization process due to the immense tendency for converging to local optima. Meanwhile, for optimizing largescale problems, hill-climbing is not an appropriate selection. In order to improve understanding of the proposed hybrid optimization algorithm’s steps, the flowchart of HCMVO is designed and can be seen in Figure 16.

Figure 17 shows the observed capture factor (which is the absorbed energy with respect to the available energy) by each optimization algorithm from iterations 1 to 400. The algorithms use ten search agents in their modified codes to find the optimal solutions. While GWO and MFO remain roughly constant after iterations 54 and 40, the other three algorithms keep improving the capture factor. In this case, HCMVO and MVO worked very well in the optimizing process with a capture factor obtained by the former as 0.594 and by the latter as 0.593. MFO almost found its highest value before the iteration 50, which means the exploration part of the algorithm works out well. Similarly, HCMVO does the same. However, it keeps finding the better solution during the optimization process until the last iteration, indicating the strong exploitation part of the algorithm. GWO reveals a weakness in exploration and exploitation because not only does it evoke the least capture factor value, but also the curve remains almost unchanged throughout 350 iterations.

Figure 18 illustrates complex interactions between the five optimization parameters and the capture factor for HCMVO (a), MPA (b), and MFO (c) algorithms. The first interesting observation is that there is a high level of nonlinear relationships among the setting parameters that can make a multi-modal search space. The dark blue lines represent the best-found configuration throughout the optimisation process. Based on both HCMVO (a) and MVO (b), we can infer that the dark blue lines concentrate in a specific range, showing the high convergence ability of both HCMVO and MVO. However, MFO (c) could not find the exact optimal range of the decision variables, and the best-found solutions per generation distribute mostly all around the search space.

Empty CellAlgorithm 1: Hill Climb Multiverse Optimization
01:procedure HCMVO
02:Initialization
03:Initialize the constraints��1�,��1�
04:�1�=Mi�1�+���1�/�▹Compute the step size,�is search resolution
05:So�1=〈�,�,�,�,�〉▹���������������
06:�������1=����So�1▹���������ℎ���������
07:Main loop
08:for iter≤���ita=do
09:���=���±��
10:while�≤���(Sol1)do
11:���=���+�,▹����ℎ���ℎ��������ℎ
12:fitness��iter=�������
13:t = t+1
14:end while
15:〈�����,������max〉=����������
16:���itev=���Inde�max▹�������ℎ�������������������������������ℎ�������
17:��=��-����Max��+1▹�����������������
18:end for
19:return���iter,����
20:end procedure

were utilized to illuminate the optimum values of the design parameters, and the best method was chosen to develop a new algorithm that performs both local and global search methods.

The correlation between hydrodynamic parameters and the capture factor of the converter was supported by the results. For any given water depth, the capture factor increases as the wave period increases, until a certain wave period value (6 seconds) is reached, after which the capture factor gradually decreases. It is expected since the flap cannot oscillate effectively when the wavelength is too short for a certain water depth. Conversely, when the wavelength is too long, the capture factor decreases. Furthermore, under a constant wave period, increasing the water depth does not affect the capture factor. Regarding the sensitivity analysis, the study found that increasing the flap bottom elevation causes turbulence flow behind the flap and limitation of rotation, which leads to less interaction with the incoming waves. Furthermore, while keeping the flap bottom elevation constant, increasing the freeboard improves the capture factor. Overtopping happens when the freeboard is negative and the flap is below the water surface, which has a detrimental influence on converter performance. Furthermore, raising the freeboard causes the wave impact to become more violent, which increases converter performance.

In the last part, we discussed the search process of each algorithm and visualized their performance and convergence curves as they try to find the best values for decision variables. Among the four selected metaheuristic algorithms, the Multi-verse Optimizer proved to be the most effective in achieving the best answer in terms of the WEC capture factor. However, the MVO needed modifications regarding its escape approach from the local optima in order to improve its global search capabilities. To overcome these constraints, we presented a fast and efficient meta-algorithm (HCMVO) that combines MVO with a Random-restart hill-climbing local search. On a higher level, this meta-algorithm employed MVO to generate global tracking and present a range of possible and appropriate solutions. Taken together, the results demonstrated that there is a significant degree of nonlinearity among the setup parameters that might result in a multimodal search space. Since MVO was faced with a stagnation issue or fell into a local optimum, we constructed a complete neighborhood search around the best-found solution offered by the upper level. In sum, the newly-developed algorithm proved to be highly effective for the problem compared to other similar optimization methods. The strength of the current findings may encourage future investigation on design optimization of wave energy converters using developed geometry as well as the novel approach.

CRediT authorship contribution statement

Erfan Amini: Conceptualization, Methodology, Validation, Data curation, Writing – original draft, Writing – review & editing, Visualization. Mahdieh Nasiri: Conceptualization, Methodology, Validation, Data curation, Writing – original draft, Writing – review & editing, Visualization. Navid Salami Pargoo: Writing – original draft, Writing – review & editing. Zahra Mozhgani: Conceptualization, Methodology. Danial Golbaz: Writing – original draft. Mehrdad Baniesmaeil: Writing – original draft. Meysam Majidi Nezhad: . Mehdi Neshat: Supervision, Conceptualization, Writing – original draft, Writing – review & editing, Visualization. Davide Astiaso Garcia: Supervision. Georgios Sylaios: Supervision.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgement

This research has been carried out within ILIAD (Inte-grated Digital Framework for Comprehensive Maritime Data and Information Services) project that received funding from the European Union’s H2020 programme.

Data availability

Data will be made available on request.

References

Figure 1: Mold drawings

3D Flow and Temperature Analysis of Filling a Plutonium Mold

플루토늄 주형 충전의 3D 유동 및 온도 분석

Authors: Orenstein, Nicholas P. [1]

Publication Date:2013-07-24
Research Org.: Los Alamos National Lab. (LANL), Los Alamos, NM (United States)
Sponsoring Org.: DOE/LANL
OSTI Identifier: 1088904
Report Number(s): LA-UR-13-25537
DOE Contract Number: AC52-06NA25396
Resource Type: Technical Report
Country of Publication: United States
Language: English
Subject: Engineering(42); Materials Science(36); Radiation Chemistry, Radiochemistry, & Nuclear Chemistry(38)

Introduction

The plutonium foundry at Los Alamos National Laboratory casts products for various special nuclear applications. However, plutonium’s radioactivity, material properties, and security constraints complicate the ability to perform experimental analysis of mold behavior. The Manufacturing Engineering and Technologies (MET-2) group previously developed a graphite mold to vacuum cast small plutonium disks to be used by the Department of Homeland Security as point sources for radiation sensor testing.

A two-stage pouring basin consisting of a funnel and an angled cavity directs the liquid into a vertical runner. A stack of ten disk castings connect to the runner by horizontal gates. Volumetric flow rates were implemented to limit overflow into the funnel and minimize foundry returns. Models using Flow-3D computational fluid dynamics software are employed here to determine liquid Pu flow paths, optimal pour regimes, temperature changes, and pressure variations.

Setup

Hardcopy drawings provided necessary information to create 3D .stl models for import into Flow-3D (Figs. 1 and 2). The mesh was refined over several iterations to isolate the disk cavities, runner, angled cavity, funnel, and input pour. The final flow and mold-filling simulation utilizes a fine mesh with ~5.5 million total cells. For the temperature study, the mesh contained 1/8 as many cells to reduce computational time and set temperatures to 850 °C for the molten plutonium and 500 °C for the solid graphite mold components (Fig. 3).

Flow-3D solves mass continuity and Navier-Stokes momentum equations over the structured rectangular grid model using finite difference and finite volume numerical algorithms. The solver includes terms in the momentum equation for body and viscous accelerations and uses convective heat transfer.

Simulation settings enabled Flow-3D physics calculations for gravity at 980.665 cm/s 2 in the negative Z direction (top of mold to bottom); viscous, turbulent, incompressible flow using dynamically-computed Renormalized Group Model turbulence calculations and no-slip/partial slip wall shear, and; first order, full energy equation heat transfer.

Mesh boundaries were all set to symmetric boundary conditions except for the Zmin boundary set to outflow and the Zmax boundary set to a volume flow. Vacuum casting conditions and the high reactivity of remaining air molecules with Pu validate the assumption of an initially fluidless void.

Results

The flow follows a unique three-dimensional path. The mold fills upwards with two to three disks receiving fluid in a staggered sequence. Figures 5-9 show how the fluid fills the cavity, and Figure 7 includes the color scale for pressure levels in these four figures. The narrow gate causes a high pressure region which forces the fluid to flow down the cavity centerline.

It proceeds to splash against the far wall and then wrap around the circumference back to the gate (Figs. 5 and 6). Flow in the angled region of the pouring basin cascades over the bottom ledge and attaches to the far wall of the runner, as seen in Figure 7.

This channeling becomes less pronounced as fluid volume levels increase. Finally, two similar but non-uniform depressed regions form about the centerline. These regions fill from their perimeter and bottom until completion (Fig. 8). Such a pattern is counter, for example, to a steady scenario in which a circle of molten Pu encompassing the entire bottom surface rises as a growing cylinder.

Cavity pressure becomes uniform when the cavity is full. Pressure levels build in the rising well section of the runner, where impurities were found to settle in actual casting. Early test simulations optimized the flow as three pours so that the fluid would never overflow to the funnel, the cavities would all fill completely, and small amounts of fluid would remain as foundry returns in the angled cavity.

These rates and durations were translated to the single 2.7s pour at 100 cm 3 per second used here. Figure 9 shows anomalous pressure fluctuations which occurred as the cavities became completely filled. Multiple simulations exhibited a rapid change in pressure from positive to negative and back within the newly-full disk and surrounding, already-full disks.

The time required to completely fill each cavity is plotted in Figure 10. Results show negligible temperature change within the molten Pu during mold filling and, as seen in Figure 11, at fill completion.

Figure 1: Mold drawings
Figure 1: Mold drawings
Figure 2: Mold Assembly
Figure 2: Mold Assembly
Figure 4: Actual mold and cast Pu
Figure 4: Actual mold and cast Pu
Figure 5: Bottom cavity filling
from runner
Figure 5: Bottom cavity filling from runner
Figure 6: Pouring and filling
Figure 6: Pouring and filling
Figure 8: Edge detection of cavity fill geometry. Two similar depressed areas form
about the centerline. Top cavity shown; same pressure scale as other figures
Figure 8: Edge detection of cavity fill geometry. Two similar depressed areas form about the centerline. Top cavity shown; same pressure scale as other figures
Figure 10: Cavity fill times,from first fluid contact with pouring basin, Figure 11:Fluid temperature remains essentially constant
Figure 10: Cavity fill times,from first fluid contact with pouring basin, Figure 11:Fluid temperature remains essentially constant

Conclusions

Non-uniform cavity filling could cause crystal microstructure irregularities during solidification. However, the small temperature changes seen – due to large differences in specific heat between Pu and graphite – over a relatively short time make such problems unlikely in this case.

In the actual casting, cooling required approximately ten minutes. This large difference in time scales further reduces the chance for temperature effects in such a superheated scenario. Pouring basin emptying decreases pressure at the gate which extends fill time of the top two cavities.

The bottom cavity takes longer to fill because fluid must first enter the runner and fill the well. Fill times continue linearly until the top two cavities. The anomalous pressure fluctuations may be due to physical attempts by the system to reach equilibrium, but they are more likely due to numerical errors in the Flow3D solver.

Unsuccessful tests were performed to remove them by halving fluid viscosity. The fine mesh reduced, but did not eliminate, the extent of the fluctuations. Future work is planned to study induction and heat transfer in the full Pu furnace system, including quantifying temporal lag of the cavity void temperature to the mold wall temperature during pre-heat and comparing heat flux levels between furnace components during cool-down.

Thanks to Doug Kautz for the opportunity to work with MET-2 and for assigning an interesting unclassified project. Additional thanks to Mike Bange for CFD guidance, insight of the project’s history, and draft review.

Sketch of approach channel and spillway of the Kamal-Saleh dam

CFD modeling of flow pattern in spillway’s approach channel

Sustainable Water Resources Management volume 1, pages245–251 (2015)Cite this article

Abstract

Analysis of behavior and hydraulic characteristics of flow over the dam spillway is a complicated task that takes lots of money and time in water engineering projects planning. To model those hydraulic characteristics, several methods such as physical and numerical methods can be used. Nowadays, by utilizing new methods in computational fluid dynamics (CFD) and by the development of fast computers, the numerical methods have become accessible for use in the analysis of such sophisticated flows. The CFD softwares have the capability to analyze two- and three-dimensional flow fields. In this paper, the flow pattern at the guide wall of the Kamal-Saleh dam was modeled by Flow 3D. The results show that the current geometry of the left wall causes instability in the flow pattern and making secondary and vortex flow at beginning approach channel. This shape of guide wall reduced the performance of weir to remove the peak flood discharge.

댐 여수로 흐름의 거동 및 수리학적 특성 분석은 물 공학 프로젝트 계획에 많은 비용과 시간이 소요되는 복잡한 작업입니다. 이러한 수력학적 특성을 모델링하기 위해 물리적, 수치적 방법과 같은 여러 가지 방법을 사용할 수 있습니다. 요즘에는 전산유체역학(CFD)의 새로운 방법을 활용하고 빠른 컴퓨터의 개발로 이러한 정교한 흐름의 해석에 수치 방법을 사용할 수 있게 되었습니다. CFD 소프트웨어에는 2차원 및 3차원 유동장을 분석하는 기능이 있습니다. 본 논문에서는 Kamal-Saleh 댐 유도벽의 흐름 패턴을 Flow 3D로 모델링하였다. 결과는 왼쪽 벽의 현재 형상이 흐름 패턴의 불안정성을 유발하고 시작 접근 채널에서 2차 및 와류 흐름을 만드는 것을 보여줍니다. 이러한 형태의 안내벽은 첨두방류량을 제거하기 위해 둑의 성능을 저하시켰다.

Introduction

Spillways are one of the main structures used in the dam projects. Design of the spillway in all types of dams, specifically earthen dams is important because the inability of the spillway to remove probable maximum flood (PMF) discharge may cause overflow of water which ultimately leads to destruction of the dam (Das and Saikia et al. 2009; E 2013 and Novak et al. 2007). So study on the hydraulic characteristics of this structure is important. Hydraulic properties of spillway including flow pattern at the entrance of the guide walls and along the chute. Moreover, estimating the values of velocity and pressure parameters of flow along the chute is very important (Chanson 2004; Chatila and Tabbara 2004). The purpose of the study on the flow pattern is the effect of wall geometry on the creation transverse waves, flow instability, rotating and reciprocating flow through the inlet of spillway and its chute (Parsaie and Haghiabi 2015ab; Parsaie et al. 2015; Wang and Jiang 2010). The purpose of study on the values of velocity and pressure is to calculate the potential of the structure to occurrence of phenomena such as cavitation (Fattor and Bacchiega 2009; Ma et al. 2010). Sometimes, it can be seen that the spillway design parameters of pressure and velocity are very suitable, but geometry is considered not suitable for conducting walls causing unstable flow pattern over the spillway, rotating flows at the beginning of the spillway and its design reduced the flood discharge capacity (Fattor and Bacchiega 2009). Study on spillway is usually conducted using physical models (Su et al. 2009; Suprapto 2013; Wang and Chen 2009; Wang and Jiang 2010). But recently, with advances in the field of computational fluid dynamics (CFD), study on hydraulic characteristics of this structure has been done with these techniques (Chatila and Tabbara 2004; Zhenwei et al. 2012). Using the CFD as a powerful technique for modeling the hydraulic structures can reduce the time and cost of experiments (Tabbara et al. 2005). In CFD field, the Navier–Stokes equation is solved by powerful numerical methods such as finite element method and finite volumes (Kim and Park 2005; Zhenwei et al. 2012). In order to obtain closed-form Navier–Stokes equations turbulence models, such k − ε and Re-Normalisation Group (RNG) models have been presented. To use the technique of computational fluid dynamics, software packages such as Fluent and Flow 3D, etc., are provided. Recently, these two software packages have been widely used in hydraulic engineering because the performance and their accuracy are very suitable (Gessler 2005; Kim 2007; Kim et al. 2012; Milési and Causse 2014; Montagna et al. 2011). In this paper, to assess the flow pattern at Kamal-Saleh guide wall, numerical method has been used. All the stages of numerical modeling were conducted in the Flow 3D software.

Materials and methods

Firstly, a three-dimensional model was constructed according to two-dimensional map that was prepared for designing the spillway. Then a small model was prepared with scale of 1:80 and entered into the Flow 3D software; all stages of the model construction was conducted in AutoCAD 3D. Flow 3D software numerically solved the Navier–Stokes equation by finite volume method. Below is a brief reference on the equations that used in the software. Figure 1 shows the 3D sketch of Kamal-Saleh spillway and Fig. 2 shows the uploading file of the Kamal-Saleh spillway in Flow 3D software.

figure 1
Fig. 1
figure 2
Fig. 2

Review of the governing equations in software Flow 3D

Continuity equation at three-dimensional Cartesian coordinates is given as Eq (1).

vf∂ρ∂t+∂∂x(uAx)+∂∂x(vAy)+∂∂x(wAz)=PSORρ,vf∂ρ∂t+∂∂x(uAx)+∂∂x(vAy)+∂∂x(wAz)=PSORρ,

(1)

where uvz are velocity component in the x, y, z direction; A xA yA z cross-sectional area of the flow; ρ fluid density; PSOR the source term; v f is the volume fraction of the fluid and three-dimensional momentum equations given in Eq (2).

∂u∂t+1vf(uAx∂u∂x+vAy∂u∂y+wAz∂u∂z)=−1ρ∂P∂x+Gx+fx∂v∂t+1vf(uAx∂v∂x+vAy∂v∂y+wAz∂v∂z)=−1ρ∂P∂y+Gy+fy∂w∂t+1vf(uAx∂w∂x+vAy∂w∂y+wAz∂w∂z)=−1ρ∂P∂y+Gz+fz,∂u∂t+1vf(uAx∂u∂x+vAy∂u∂y+wAz∂u∂z)=−1ρ∂P∂x+Gx+fx∂v∂t+1vf(uAx∂v∂x+vAy∂v∂y+wAz∂v∂z)=−1ρ∂P∂y+Gy+fy∂w∂t+1vf(uAx∂w∂x+vAy∂w∂y+wAz∂w∂z)=−1ρ∂P∂y+Gz+fz,

(2)

where P is the fluid pressure; G xG yG z the acceleration created by body fluids; f xf yf z viscosity acceleration in three dimensions and v f is related to the volume of fluid, defined by Eq. (3). For modeling of free surface profile the VOF technique based on the volume fraction of the computational cells has been used. Since the volume fraction F represents the amount of fluid in each cell, it takes value between 0 and 1.

∂F∂t+1vf[∂∂x(FAxu)+∂∂y(FAyv)+∂∂y(FAzw)]=0∂F∂t+1vf[∂∂x(FAxu)+∂∂y(FAyv)+∂∂y(FAzw)]=0

(3)

Turbulence models

Flow 3D offers five types of turbulence models: Prantl mixing length, k − ε equation, RNG models, Large eddy simulation model. Turbulence models that have been proposed recently are based on Reynolds-averaged Navier–Stokes equations. This approach involves statistical methods to extract an averaged equation related to the turbulence quantities.

Steps of solving a problem in Flow 3D software

(1) Preparing the 3D model of spillway by AutoCAD software. (2) Uploading the file of 3D model in Flow 3D software and defining the problem in the software and checking the final mesh. (3) Choosing the basic equations that should be solved. (4) Defining the characteristics of fluid. (5) Defining the boundary conditions; it is notable that this software has a wide range of boundary conditions. (6) Initializing the flow field. (7) Adjusting the output. (8) Adjusting the control parameters, choice of the calculation method and solution formula. (9) Start of calculation. Figure 1 shows the 3D model of the Kamal-Saleh spillway; in this figure, geometry of the left and right guide wall is shown.

Figure 2 shows the uploading of the 3D spillway dam in Flow 3D software. Moreover, in this figure the considered boundary condition in software is shown. At the entrance and end of spillway, the flow rate or fluid elevation and outflow was considered as BC. The bottom of spillway was considered as wall and left and right as symmetry.

Model calibration

Calibration of the Flow 3D for modeling the effect of geometry of guide wall on the flow pattern is included for comparing the results of Flow 3D with measured water surface profile. Calibration the Flow 3D software could be conducted in two ways: first, changing the value of upstream boundary conditions is continued until the results of water surface profile of the Flow 3D along the spillway successfully covered the measurement water surface profile; second is the assessment the mesh sensitivity. Analyzing the size of mesh is a trial-and-error process where the size of mesh is evaluated form the largest to the smallest. With fining the size of mesh the accuracy of model is increased; whereas, the cost of computation is increased. In this research, the value of upstream boundary condition was adjusted with measured data during the experimental studies on the scaled model and the mesh size was equal to 1 × 1 × 1 cm3.

Results and discussion

The behavior of water in spillway is strongly affected by the flow pattern at the entrance of the spillway, the flow pattern formation at the entrance is affected by the guide wall, and choice of an optimized form for the guide wall has a great effect on rising the ability of spillway for easy passing the PMF, so any nonuniformity in flow in the approach channel can cause reduction of spillway capacity, reduction in discharge coefficient of spillway, and even probability of cavitation. Optimizing the flow guiding walls (in terms of length, angle and radius) can cause the loss of turbulence and flow disturbances on spillway. For this purpose, initially geometry proposed for model for the discharge of spillway dam, Kamal-Saleh, 80, 100, and 120 (L/s) were surveyed. These discharges of flow were considered with regard to the flood return period, 5, 100 and 1000 years. Geometric properties of the conducting guidance wall are given in Table 1.Table 1 Characteristics and dimensions of the guidance walls tested

Full size table

Results of the CFD simulation for passing the flow rate 80 (L/s) are shown in Fig. 3. Figure 3 shows the secondary flow and vortex at the left guide wall.

figure 3
Fig. 3

For giving more information about flow pattern at the left and right guide wall, Fig. 4 shows the flow pattern at the right side guide wall and Fig. 5 shows the flow pattern at the left side guide wall.

figure 4
Fig. 4
figure 5
Fig. 5

With regard to Figs. 4 and 5 and observing the streamlines, at discharge equal to 80 (L/s), the right wall has suitable performance but the left wall has no suitable performance and the left wall of the geometric design creates a secondary and circular flow, and vortex motion in the beginning of the entrance of spillway that creates cross waves at the beginning of spillway. By increasing the flow rate (Q = 100 L/s), at the inlet spillway secondary flows and vortex were removed, but the streamline is severely distorted. Results of the guide wall performances at the Q = 100 (L/s) are shown in Fig. 6.

figure 6
Fig. 6

Also more information about the performance of each guide wall can be derived from Figs. 7 and 8. These figures uphold that the secondary and vortex flows were removed, but the streamlines were fully diverted specifically near the left side guide wall.

figure 7
Fig. 7
figure 8
Fig. 8

As mentioned in the past, these secondary and vortex flows and diversion in streamline cause nonuniformity and create cross wave through the spillway. Figure 9 shows the cross waves at the crest of the spillway.

figure 9
Fig. 9

The performance of guide walls at the Q = 120 (L/s) also was assessed. The result of simulation is shown in Fig. 10. Figures 11 and 12 show a more clear view of the streamlines near to right and left side guide wall, respectively. As seen in Fig. 12, the left side wall still causes vortex flow and creation of and diversion in streamline.

figure 10
Fig. 10
figure 11
Fig. 11
figure 12
Fig. 12

The results of the affected left side guide wall shape on the cross wave creation are shown in Fig. 13. As seen from Fig. 3, the left side guide wall also causes cross wave at the spillway crest.

figure 13
Fig. 13

As can be seen clearly in Figs. 9 and 13, by moving from the left side to the right side of the spillway, the cross waves and the nonuniformity in flow is removed. By reviewing Figs. 9 and 13, it is found that the right side guide wall removes the cross waves and nonuniformity. With this point as aim, a geometry similar to the right side guide wall was considered instead of the left side guide wall. The result of simulation for Q = 120 (L/s) is shown in Fig. 14. As seen from this figure, the proposed geometry for the left side wall has suitable performance smoothly passing the flow through the approach channel and spillway.

figure 14
Fig. 14

More information about the proposed shape for the left guide wall is shown in Fig. 15. As seen from this figure, this shape has suitable performance for removing the cross waves and vortex flows.

figure 15
Fig. 15

Figure 16 shows the cross section of flow at the crest of spillway. As seen in this figure, the proposed shape for the left side guide wall is suitable for removing the cross waves and secondary flows.

figure 16
Fig. 16

Conclusion

Analysis of behavior and hydraulic properties of flow over the spillway dam is a complicated task which is cost and time intensive. Several techniques suitable to the purposes of study have been undertaken in this research. Physical modeling, usage of expert experience, usage of mathematical models on simulation flow in one-dimensional, two-dimensional and three-dimensional techniques, are some of the techniques utilized to study this phenomenon. The results of the modeling show that the CFD technique is a suitable tool for simulating the flow pattern in the guide wall. Using this tools helps the designer for developing the optimal shape for hydraulic structure which the flow pattern through them are important.

References

  • Chanson H (2004) 19—Design of weirs and spillways. In: Chanson H (ed) Hydraulics of open channel flow, 2nd edn. Butterworth-Heinemann, Oxford, pp 391–430Chapter Google Scholar 
  • Chatila J, Tabbara M (2004) Computational modeling of flow over an ogee spillway. Comput Struct 82:1805–1812Article Google Scholar 
  • Das MM, Saikia MD (2009) Irrigation and water power engineering. PHI Learning, New DelhiGoogle Scholar 
  • E, Department Of Army: U.S. Army Corps (2013) Hydraulic Design of Spillways. BiblioBazaar, CharlestonGoogle Scholar 
  • Fattor C, Bacchiega J (2009) Design conditions for morning-glory spillways: application to potrerillos dam spillway. Adv Water Res Hydraul Eng Springer, Berlin, pp 2123–2128Google Scholar 
  • Gessler D (2005) CFD modeling of spillway performance. Impacts Glob Clim Change. doi:10.1061/40792(173)398
  • Kim D-G (2007) Numerical analysis of free flow past a sluice gate. KSCE J Civ Eng 11:127–132Article Google Scholar 
  • Kim D, Park J (2005) Analysis of flow structure over ogee-spillway in consideration of scale and roughness effects by using CFD model. KSCE J Civ Eng 9:161–169Article Google Scholar 
  • Kim S, Yu K, Yoon B, Lim Y (2012) A numerical study on hydraulic characteristics in the ice Harbor-type fishway. KSCE J Civ Eng 16:265–272Article Google Scholar 
  • Ma X-D, Dai G-Q, Yang Q, Li G-J, Zhao L (2010) Analysis of influence factors of cavity length in the spillway tunnel downstream of middle gate chamber outlet with sudden lateral enlargement and vertical drop aerator. J Hydrodyn Ser B 22:680–686Article Google Scholar 
  • Milési G, Causse S (2014) 3D numerical modeling of a side-channel spillway. In: Gourbesville P, Cunge J, Caignaert G (eds) Advances in hydroinformatics. Springer, Singapore, pp 487–498Chapter Google Scholar 
  • Montagna F, Bellotti G, Di Risio M (2011) 3D numerical modeling of landslide-generated tsunamis around a conical island. Nat Hazards 58:591–608Article Google Scholar 
  • Novak P, Moffat AIB, Nalluri C, Narayanan R (2007) Hydraulic structures. Taylor & Francis, LondonGoogle Scholar 
  • Parsaie A, Haghiabi A (2015a) Computational modeling of pollution transmission in rivers. Appl Water Sci. doi:10.1007/s13201-015-0319-6
  • Parsaie A, Haghiabi A (2015b) The effect of predicting discharge coefficient by neural network on increasing the numerical modeling accuracy of flow over side weir. Water Res Manag 29:973–985Article Google Scholar 
  • Parsaie A, Yonesi H, Najafian S (2015) Predictive modeling of discharge in compound open channel by support vector machine technique. Model Earth Syst Environ 1:1–6Article Google Scholar 
  • Su P-L, Liao H-S, Qiu Y, Li CJ (2009) Experimental study on a new type of aerator in spillway with low Froude number and mild slope flow. J Hydrodyn Ser B 21:415–422Article Google Scholar 
  • Suprapto M (2013) Increase spillway capacity using Labyrinth Weir. Procedia Eng 54:440–446Article Google Scholar 
  • Tabbara M, Chatila J, Awwad R (2005) Computational simulation of flow over stepped spillways. Comput Struct 83:2215–2224Article Google Scholar 
  • Wang J, Chen H (2009) Experimental study of elimination of vortices along guide wall of bank spillway. Adv Water Res Hydraul Eng Springer, Berlin, pp 2059–2063Google Scholar 
  • Wang Y, Jiang C (2010) Investigation of the surface vortex in a spillway tunnel intake. Tsinghua Sci Technol 15:561–565Article Google Scholar 
  • Zhenwei MU, Zhiyan Z, Tao Z (2012) Numerical simulation of 3-D flow field of spillway based on VOF method. Procedia Eng 28:808–812Article Google Scholar 

Download references

Author information

Authors and Affiliations

  1. Department of Water Engineering, Lorestan University, Khorram Abad, IranAbbas Parsaie, Amir Hamzeh Haghiabi & Amir Moradinejad

Corresponding author

Correspondence to Abbas Parsaie.

Reprints and Permissions

About this article

Cite this article

Parsaie, A., Haghiabi, A.H. & Moradinejad, A. CFD modeling of flow pattern in spillway’s approach channel. Sustain. Water Resour. Manag. 1, 245–251 (2015). https://doi.org/10.1007/s40899-015-0020-9

Download citation

  • Received28 April 2015
  • Accepted28 August 2015
  • Published15 September 2015
  • Issue DateSeptember 2015
  • DOIhttps://doi.org/10.1007/s40899-015-0020-9

Share this article

Anyone you share the following link with will be able to read this content:Get shareable link

Provided by the Springer Nature SharedIt content-sharing initiative

Keywords

  • Approach channel
  • Kamal-Saleh dam
  • Guide wall
  • Flow pattern
  • Numerical modeling
  • Flow 3D software
    Effect of tailwater depth on non-cohesive earth dam failure due to overtopping

    Effect of tailwater depth on non-cohesive earth dam failure due to overtopping

    범람으로 인한 비점착성 흙댐 붕괴에 대한 테일워터 깊이의 영향

    ShaimaaAmanaMohamedAbdelrazek RezkbRabieaNasrc

    Abstract

    본 연구에서는 범람으로 인한 토사댐 붕괴에 대한 테일워터 깊이의 영향을 실험적으로 조사하였다. 테일워터 깊이의 네 가지 다른 값을 검사합니다. 각 실험에 대해 댐 수심 측량 프로파일의 진화, 고장 기간, 침식 체적 및 유출 수위곡선을 관찰하고 기록합니다.

    결과는 tailwater 깊이를 늘리면 고장 시간이 최대 57% 감소하고 상대적으로 침식된 마루 높이가 최대 77.6% 감소한다는 것을 보여줍니다. 또한 상대 배수 깊이가 3, 4, 5인 경우 누적 침식 체적의 감소는 각각 23, 36.5 및 75%인 반면 최대 유출량의 감소는 각각 7, 14 및 17.35%입니다.

    실험 결과는 침식 과정을 복제할 때 Flow 3D 소프트웨어의 성능을 평가하는 데 활용됩니다. 수치 모델은 비응집성 흙댐의 침식 과정을 성공적으로 시뮬레이션합니다.

    The influence of tailwater depth on earth dam failure due to overtopping is investigated experimentally in this work. Four different values of tailwater depths are examined. For each experiment, the evolution of the dam bathymetry profile, the duration of failure, the eroded volume, and the outflow hydrograph are observed and recorded. The results reveal that increasing the tailwater depth reduces the time of failure by up to 57% and decreases the relative eroded crest height by up to 77.6%. In addition, for relative tailwater depths equal to 3, 4, and 5, the reduction in the cumulative eroded volume is 23, 36.5, and 75%, while the reduction in peak discharge is 7, 14, and 17.35%, respectively. The experimental results are utilized to evaluate the performance of the Flow 3D software in replicating the erosion process. The numerical model successfully simulates the erosion process of non-cohesive earth dams.

    Keywords

    Earth dam, Eroded volume, Flow 3D model, Non-cohesive soil, Overtopping failure, Tailwater depth

    Notation

    d50

    Mean partical diameterWc

    Optimum water contentZo

    Dam height (cm)do

    Tailwater depth (cm)Zeroded

    Eroded height of the dam measured at distance of 0.7 m from the dam heel (cm)t

    Total time of failure (sec)t1

    Time of crest width erosion (sec)Zcrest

    The crest height (cm)Vtotal

    Total volume of the dam (m3)Veroded

    Cumulative eroded volume (m3)RMSE

    The statistical variable root- mean- square errord

    Degree of agreement indexyu.s.

    The upstream water depth (cm)yd.s

    The downstream water depth (cm)H

    Water surface elevation over sharp crested weir (cm)Q

    Outflow discharge (liter/sec)Qpeak

    Peak discharge (liter/sec)

    1. Introduction

    Earth dams are compacted structures composed of natural materials that are usually mined or quarried from local locations. The failures of the earth dams have proven to be deadly, destructive, and costly. According to People’s Daily, two earthen dams, Yong’an Dam and Xinfa Dam located in Hulun Buir City in North China’s Inner Mongolia failed on 2021, due to a surge in the water level of the Nuomin River caused by heavy rain. The dam breach affected 16,660 people, flooded 325,622 mu of farmland (21708.1 ha), and destroyed 22 bridges, 124 culverts, and 15.6 km of roadways. Also, the failure of south fork dam (earth and rock fill dam) near Johnstown on 1889 is considered the worst U.S dam disaster in terms of loss of life. The dam was overtopped and washed away due to unexpected heavy rains, releasing 20 million tons of water which destroyed Johnstown and resulted in 2209 deaths, [1][2]. Piping or shear sliding, failure due to natural factors, and failure due to overtopping are all possible causes of earth dam failure. However, overtopping failure is the most frequent cause of dam failure. According to The International Committee on Large Dams (ICOLD, 1995), and [3], more than one-third of the total known dam failures were caused by dam overtopping.

    Overtopping occurs as the result of insufficient flood design or freeboard in some cases. Extreme rainstorms can cause floods which can overtop the dam and cause it to fail. The size and geometry of the reservoir or the dam (side slopes, top width, height, etc.), the homogeneity of the material used in the construction of the dam, overtopping depth, and the presence or absence of tailwater are all elements that influence this type of failure which will be illustrated in the following literature. Overtopping failures of earth dams may be divided into several failure mechanisms based on the material composition and the inner structure of the dam. For cohesive earth dams because of low permeability, no seepage exists on the slopes. Erosion often begins at the earth dam toe during turbulent erosion and moves upstream, undercutting the slope, causing the removal of large chunks of materials. While for non-cohesive earth dams the downstream face of the dam flattens progressively and is often said to rotate around a point near the downstream toe [4][5][6] In the last few decades, the study of failures due to overtopping has gained popularity among researchers. The overtopping failure, in fact, has been widely investigated in coastal and river hydraulics and morpho dynamic. In addition, several laboratory experimental studies have been conducted in this field in order to better understand different involved factors. Also, many numerical types of research have been conducted to investigate the process of overtopping failure as well as the elements that influence this type of failure.

    Tabrizi et al. [5] conducted a series of embankment overtopping tests to find the effect of compaction on the failure of a homogenous sand embankment. A plane breach process occurred across the flume width due to the narrow flume width. They measured the downstream hydrographs and embankment surface profile for every case. They concluded that the peak discharge decreased with a high compaction level, while the time to peak increased. Kansoh et al. [6] studied experimentally the failure of compacted homogeneous non-cohesive earthen embankment due to overtopping. They investigated the influence of different shape parameters including the downstream slope, the crest width, and the height of the embankment on the erosion process. The erosion process was initiated by carving a pilot channel into the embankment crest. They evaluated the time of embankment failure for different shape parameters. They concluded that the failure time increases with increasing the downstream slope and the crest width. Zhu et al. [7] investigated experimentally the breaching of five embankments, one constructed with pure sand, and four with different sand-silt–clay mixtures. The erosion pattern was similar across the flume width. They stated that for cohesive soil mixtures the head cut erosion was the most important factor that affected the breach growth, while for non-cohesive soil the breach erosion was affected by shear erosion.

    Amaral et al. [8] studied experimentally the failure by overtopping for two embankments built from silt sand material. They studied the effect of the degree of compaction of the embankment and the geometry of the pilot channel carved at the centre of the dam crest. They studied two shapes of pilot channel a rectangular shape and triangular shape. They stated that the breach development is influenced by a higher degree of compaction, however, the pilot channel geometry did not influence the breach’s final form. Bereta et al. [9] studied experimentally the breach formation of five dam models, three of them were homogenous clay soil while two were sandy-clay mixtures. The erosion process was initiated by cutting a pilot channel at the centre of the dam crest. They observed the initiation of erosion, flow shear erosion, sidewall bottom erosion, and distinguished the soil mechanical slope mass failure from the head cut vertically and laterally during these tests. Verma et al. [10] investigated experimentally a two-dimensional erosion phenomenon due to overtopping by using a wooden fuse plug model and five different soils. They concluded that the erosion process was affected mostly by cohesiveness and degree of compaction. For cohesive soils, a head cut erosion was observed, while for non-cohesive soils surface erosion occurred gradually. Also, the dimensions of fuse plug, type of fill material, reservoir capacity, and inflow were found to affect the behaviour of the overall breaching process.

    Wu and Qin [11] studied the effect of adding coarse grains to the downstream face of a non-cohesive dam as a result of tailings deposition. The process of overtopping during tailings dam failures is analyzed and its effect on delaying the dam-break process and disaster mitigation are investigated. They found that the tested protective measures decreased the breach area, the maximum breaching flow discharge and flow velocity, and the downstream inundated area. Khankandi et al. [12] studied experimentally the effect of reservoir geometry on dam break flow in case of dry and wet bed conditions. They considered four different reservoir shapes, a long reservoir, a wide, a trapezoidal shaped and one with a 90◦ bend all with identical water volume and horizontal bed. The dam break is simulated by the sudden gate removal using a pneumatic jack. They measured the variation of water level over time with ultrasonic sensors and flow velocity component with an acoustic Doppler velocimeter. Also, the experimental results of water level variation are compared with Ritters solution (1892) [13]. They stated that for dry bed condition the long and 90 bend reservoirs results are close to the analytical solution by ritter also in these two shapes a 1D flow is noticed. However, for wide and trapezoidal reservoirs a 2D effect is significant due to flow contraction at channel entrance.

    Rifai et al. [14] conducted a series of experiments to investigate the effect of tailwater depth on the outflow discharge and breach geometry during non-cohesive homogenous fluvial dikes overtopping failure. They cut an initial notch in the crest at 0.8 m from the upstream end of the dike to initiate overtopping. They compared their results to previous experiments under different main channel inflow discharges combined with a free floodplain. They divided the dike breaching process into three stages: gradual start of overtopping flow resulting in slow initiation of dike erosion, deepening and widening breach due to large flow depth and velocity, finally the flow depth starts stabilizing at its minimal level with or without sustained breach expansion. They stated that breach discharge has lower values than in free floodplain tests. Jiang [15] studied the effect of bed slope on breach parameters and peak discharge in non-cohesive embankment failure. An initial triangular breach with a depth and width of 4 cm was pre-set on one side of the dam. He stated that peak discharge increases with the increase of bed slope and then decreases.

    Ozmen-cagatay et al. [16] studied experimentally flood wave propagation resulted from a sudden dam break event. For dam-break modelling, they used a mechanism that permitted the rapid removal of a vertical plate with a thickness of 4 mm and made of rigid plastic. They conducted three tests, one with dry bed condition and two tests with tailwater depths equal 0.025 m and 0.1 m respectively. They recorded the free surface profile during initial stages of dam break by using digital image processing. Finally, they compared the experimental results with the with a commercially available VOF-based CFD program solving the Reynolds-averaged Navier –Stokes equations (RANS) with the k– Ɛ turbulence model and the shallow water equations (SWEs). They concluded that Wave breaking was delayed with increasing the tailwater depth to initial reservoir depth ratio. They also stated that the SWE approach is sufficient more to represent dam break flows for wet bed condition. Evangelista [17] investigated experimentally and numerically using a depth-integrated two-phase model, the erosion of sand dike caused by the impact of a dam break wave. The dam break is simulated by a sudden opening of an upstream reservoir gate resulting in the overtopping of a downstream trapezoidal sand dike. The evolution of the water wave caused from the gate opening and dike erosion process are recorded by using a computer-controlled camera. The experimental results demonstrated that the progression of the wave front and dike erosion have a considerable influence on each other during the process. In addition, the dike constructed from fine sands was more resistant to erosion than the one built with coarse sand. They also stated that the numerical model can is capable of accurately predicting wave front position and dike erosion. Also, Di Cristo et al. [18] studied the effect of dam break wave propagation on a sand embankment both experimentally and numerically using a two-phase shallow-water model. The evolution of free surface and of the embankment bottom are recorded and used in numerical model assessment. They stated that the model allows reasonable simulation of the experimental trends of the free surface elevation regardeless of the geofailure operator.

    Lots of numerical models have been developed over the past few years to simulate the dam break flooding problem. A one-dimensional model, such as Hec-Ras, DAMBRK and MIKE 11, ect. A two-dimensional model such as iRIC Nay2DH is used in earth embankment breach simulation. Other researchers studied the failure process numerically using (3D) computational fluid dynamics (CFD) models, such as FLOW-3D, and FLUENT. Goharnejad et al. [19] determined the outflow hydrograph which results from the embankment dam break due to overtopping. Hu et al. [20] performed a comparison between Flow-3D and MIKE3 FM numerical models in simulating a dam break event under dry and wet bed conditions with different tailwater depths. Kaurav et al. [21] simulated a planar dam breach process due to overtopping. They conducted a sensitivity analysis to find the effect of dam material, dam height, downstream slope, crest width, and inlet discharge on the erosion process and peak discharge through breach. They concluded that downstream slope has a significant influence on breaching process. Yusof et al. [22] studied the effect of embankment sediment sizes and inflow rates on breaching geometric and hydrodynamic parameters. They stated that the peak outflow hydrograph increases with increasing sediment size and inflow rates while time of failure decreases.

    In the present work, the effect of tailwater depth on earth dam failure during overtopping is studied experimentally. The relation between the eroded volume of the dam and the tailwater depth is presented. Also, the percentage of reduction in peak discharge due to tailwater existence is calculated. An assessment of Flow 3D software performance in simulating the erosion process during earth dam failure is introduced. The statistical variable root- mean- square error, RMSE, and the agreement degree index, d, are used in model assessment.

    2. Material and methods

    The tests are conducted in a straight rectangular flume in the laboratory of Irrigation Engineering and Hydraulics Department, Faculty of Engineering, Alexandria University, Egypt. The flume dimensions are 10 m long, 0.86 m wide, and 0.5 m deep. The front part of the flume is connected to a storage basin 1 m long by 0.86 m wide. The storage basin is connected to a collecting tank for water recirculation during the experiments as shown in Fig. 1Fig. 2. A sharp-crested weir is placed at a distance of 4 m downstream the constructed dam to keep a constant tailwater depth in each experiment and to measure the outflow discharge.

    To measure the eroded volume with time a rods technique is used. This technique consists of two parallel wooden plates with 10 cm distance in between and five rows of stainless-steel rods passing vertically through the wooden plates at a spacing of 20 cm distributed across flume width. Each row consists of four rods with 15 cm spacing between them. Also, a graph board is provided to measure the drop in each rod with time as shown in Fig. 3Fig. 4. After dam construction the rods are carefully rested on the dam, with the first line of rods resting in the middle of the dam crest and then a constant distance of 15 cm between rods lines is maintained.

    A soil sample is taken and tested in the laboratory of the soil mechanics to find the soil geotechnical parameters. The soil particle size distribution is also determined by sieve analysis as shown in Fig. 5. The soil mean diameter d50,equals 0.38 mm and internal friction angle equals 32.6°.

    2.1. Experimental procedures

    To investigate the effect of the tailwater depth (do), the tailwater depth is changed four times 5, 15, 20, and 25 cm on the sand dam model. The dam profile is 35 cm height, with crest width = 15 cm, the dam base width is 155 cm, and the upstream and downstream slopes are 2:1 as shown in Fig. 6. The dam dimensions are set as the flume permitted to allow observation of the dam erosion process under the available flume dimensions and conditions. All of the conducted experiments have the same dimensions and configurations.

    The optimum water content, Wc, from the standard proctor test is found to be 8 % and the maximum dry unit weight is 19.42 kN/m3. The soil and water are mixed thoroughly to ensure consistency and then placed on three horizontal layers. Each layer is compacted according to ASTM standard with 25 blows by using a rammer (27 cm × 20.5 cm) weighing 4 kg. Special attention is paid to the compaction of the soil to guarantee the repeatability of the tests.

    After placing and compacting the three layers, the dam slopes are trimmed carefully to form the trapezoidal shape of the dam. A small triangular pilot channel with 1 cm height and 1:1 side slopes is cut into the dam crest to initiate the erosion process. The position of triangular pilot channel is presented in Fig. 1. Three digital video cameras with a resolution of 1920 × 1080 pixels and a frame rate of 60 fps are placed in three different locations. One camera on one side of the flume to record the progress of the dam profile during erosion. Another to track the water level over the sharp-crested rectangular weir placed at the downstream end of the flume. And the third camera is placed above the flume at the downstream side of the dam and in front of the rods to record the drop of the tip of the rods with time as shown previously in Fig. 1.

    Before starting the experiment, the water is pumped into the storage basin by using pump with capacity 360 m3/hr, and then into the upstream section of the flume. The upstream boundary is an inflow condition. The flow discharge provided to the storage basin is kept at a constant rate of 6 L/sec for all experiments, while the downstream boundary is an outflow boundary condition.

    Also, the required tailwater depth for each experiment is filled to the desired depth. A dye container valve is opened to color the water upstream of the dam to make it easy to distinguish the dam profile from the water profile. A wooden board is placed just upstream of the dam to prevent water from overtopping the dam until the water level rises to a certain level above the dam crest and then the wooden board is removed slowly to start the experiment.

    2.2. Repeatability

    To verify the accuracy of the results, each experiment is repeated two times under the same conditions. Fig. 7 shows the relative eroded crest height, Zeroded / Zo, with time for 5 cm tailwater depth. From the Figure, it can be noticed that results for all runs are consistent, and accuracy is achieved.

    3. Numerical model

    The commercially available numerical model, Flow 3D is used to simulate the dam failure due to overtopping for the cases of 15 cm, 20 cm and 25 cm tailwater depths. For numerical model calibration, experimental results for dam surface evolution are used. The numerical model is calibrated for selection of the optimal turbulence model (RNG, K-e, and k-w) and sediment scour equations (Van Rin, Meyer- peter and Muller, and Nielsen) that produce the best results. In this, the flow field is solved by the RNG turbulence model, and the van Rijn equation is used for the sediment scour model. A geometry file is imported before applying the mesh.

    A Mesh sensitivity is analyzed and checked for various cell sizes, and it is found that decreasing the cell size significantly increases the simulation time with insignificant differences in the result. It is noticed that the most important factor influencing cell size selection is the value of the dam’s upstream and downstream slopes. For example, the slopes in the dam model are 2:1, thus the cell size ratio in X and Z directions should be 2:1 as well. The cell size in a mesh block is set to be 0.02 m, 0.025 m, and 0.01 m in X, Y and Z directions respectively.

    In the numerical computations, the boundary conditions employed are the walls for sidewalls and the channel bottom. The pressure boundary condition is applied at the top, at the air–water interface, to account for atmospheric pressure on the free surface. The upstream boundary is volume flow rate while the downstream boundary is outflow discharge.

    The initial condition is a fluid region, which is used to define fluid areas both upstream and downstream of the dam. To assess the model accuracy, the statistical variable root- mean- square error, RMSE, and the agreement degree index, d, are calculated as(1)RMSE=1N∑i=1N(Pi-Mi)2(2)d=1-∑Mi-Pi2∑Mi-M¯+Pi-P¯2

    where N is the number of samples, Pi and Mi are the models and experimental values, P and M are the means of the model and experimental values. The best fit between the experimental and model results would have an RMSE = 0 and degree of agreement, d = 1.

    4. Results of experimental work

    The results of the total time of failure, t (defined as the time from when the water begins to overtop the dam crest until the erosion reaches a steady state, when no erosion occurs), time of crest width erosion t1, cumulative eroded volume Veroded, and peak discharge Qpeak for each experiment are listed in Table 1. The case of 5 cm tailwater depth is considered as a reference case in this work.

    Table 1. Results of experimental work.

    Tailwater depth, do (cm)Total time of failure, t (sec)Time of crest width erosion, t1 (sec)cumulative eroded volume, Veroded (m3)Peak discharge, Qpeak (liter/sec)
    5255220.2113.12
    15165300.1612.19
    20140340.1311.29
    25110390.0510.84

    5. Discussion

    5.1. Side erosion

    The evolution of the bathymetry of the erosion line recorded by the video camera1. The videos are split into frames (60 frames/sec) by the Free Video to JPG Converter v.5.063 build and then converted into an excel spreadsheet using MATLAB code as shown in Fig. 8.

    Fig. 9 shows a sample of numerical model output. Fig. 10Fig. 11Fig. 12 show a dam profile development for different time steps from both experimental and numerical model, for tailwater depths equal 15 cm, 20 cm and 25 cm. Also, the values of RMSE and d for each figure are presented. The comparison shows that the Flow 3D software can simulate the erosion process of non-cohesive earth dam during overtopping with an RMSE value equals 0.023, 0.0218, and 0.0167 and degree of agreement, d, equals 0.95, 0.968, and 0.988 for relative tailwater depths, do/(do)ref, = 3, 4 and 5, respectively. The low values of RMSE and high values of d show that the Flow 3D can effectively simulate the erosion process. From Fig. 10Fig. 11Fig. 12, it can be noticed that the model is not capable of reproducing the head cut, while it can simulate well the degradation of the crest height with a minor difference from experimental work. The reason of this could be due to inability of simulation of all physical conditions which exists in the experimental work, such as channel friction and the grain size distribution of the dam soil which is surely has a great effect on the erosion process and breach development. In the experimental work the grain size distribution is shown in Fig. 5, while the numerical model considers that the soil is uniform and exactly 50 % of the dam particles diameter are equal to the d50 value. Another reason is that the model is not considering the increased resistance of the dam due to the apparent cohesion which happens due to dam saturation [23].

    It is clear from both the experimental and numerical results that for a 5 cm tailwater depth, do/(do)ref = 1.0, erosion begins near the dam toe and continues upward on the downstream slope until it reaches the crest. After eroding the crest width, the crest is lowered, resulting in increased flow rates and the speeding up of the erosion process. While for relative tailwater depths, do/(do)ref = 3, 4, and 5 erosion starts at the point of intersection between the downstream slope and tailwater. The existence of tailwater works as an energy dissipater for the falling water which reduces the erosion process and prevents the dam from failure as shown in Fig. 13. It is found that the time of the failure decreases with increasing the tailwater depth because most of the dam height is being submerged with water which decreases the erosion process. The reduction in time of failure from the referenced case is found to be 35.3, 45, and 57 % for relative tailwater depth, do /(do)ref equals 3, 4, and 5, respectively.

    The relation between the relative eroded crest height, Zeroded /Zo, with time is drawn as shown in Fig. 14. It is found that the relative eroded crest height decreases with increasing tailwater depth by 10, 41, and 77.6 % for relative tailwater depth, do /(do)ref equals 3, 4, and 5, respectively. The time required for the erosion of the crest width, t1, is calculated for each experiment. The relation between relative tailwater depth and relative time of crest width erosion is shown in Fig. 15. It is found that the time of crest width erosion increases linearly with increasing, do /Zo. The percent of increase is 36.4, 54.5 and 77.3 % for relative tailwater depth, do /(do)ref = 3, 4 and 5, respectively.

    Crest height, Zcrest is calculated from the experimental results and the Flow 3D results for relative tailwater depths, do/(do)ref, = 3, 4, and 5. A relation between relative crest height, Zcrest/Zo with time from experimental and numerical results is presented in Fig. 16. From Fig. 16, it is seen that there is a good consistency between the results of numerical model and the experimental results in the case of tracking the erosion of the crest height with time.

    5.2. Upstream and downstream water depths

    It is noticed that at the beginning of the erosion process, both upstream and downstream water depths increase linearly with time as long as erosion of the crest height did not take place. However, when the crest height starts to lower the upstream water depth decreases with time while the downstream water depth increases. At the end of the experiment, the two depths are nearly equal. A relation between relative downstream and upstream water depths with time is drawn for each experiment as shown in Fig. 17.

    5.3. Eroded volume

    A MATLAB code is used to calculate the cumulative eroded volume every time interval for each experiment. The total volume of the dam, Vtotal is 0.256 m3. The cumulative eroded volume, Veroded is 0.21, 0.16, 0.13, and 0.05 m3 for tailwater depths, do = 5, 15, 20, and 25 cm, respectively. Fig. 18 presents the relation between cumulative eroded volume, Veroded and time. From Fig. 18, it is observed that the cumulative eroded volume decreases with increasing the tailwater depth. The reduction in cumulative eroded volume is 23, 36.5, and 75 % for relative tailwater depth, do /(do)ref = 3, 4, and 5, respectively. The relative remained volume of the dam equals 0.18, 0.375, 0.492, and 0.8 for tailwater depths = 5, 15, 20, and 25 cm, respectively. Fig. 19 shows a relation between relative tailwater depth and relative cumulative eroded volume from experimental results. From that figure, it is noticed that the eroded volume decreases exponentially with increasing relative tailwater depth.

    5.4. The outflow discharge

    The inflow discharge provided to the storage tank is maintained constant for all experiments. The water surface elevation, H, over the sharp-crested weir placed at the downstream side is recorded by the video camera 2. For each experiment, the outflow discharge is then calculated by using the sharp-crested rectangular weir equation every 10 sec.

    The outflow discharge is found to increase rapidly until it reaches its peak then it decreases until it is constant. For high values of tailwater depths, the peak discharge becomes less than that in the case of small tailwater depth as shown in Fig. 20 which agrees well with the results of Rifai et al. [14] The reduction in peak discharge is 7, 14, and 17.35 % for relative tailwater depth, do /(do)ref = 3, 4, and 5, respectively.

    The scenario presented in this article in which the tailwater depth rises due to unexpected heavy rainfall, is investigated to find the effect of rising tailwater depth on earth dam failure. The results revealed that rising tailwater depth positively affects the process of dam failure in terms of preventing the dam from complete failure and reducing the outflow discharge.

    6. Conclusions

    The effect of tailwater depth on earth dam failure due to overtopping is investigated experimentally in this work. The study focuses on the effect of tailwater depth on side erosion, upstream and downstream water depths, eroded volume, outflow hydrograph, and duration of the failure process. The Flow 3D numerical software is used to simulate the dam failure, and a comparison is made between the experimental and numerical results to find the ability of this software to simulate the erosion process. The following are the results of the investigation:

    The existence of tailwater with high depths prevents the dam from completely collapsing thereby turning it into a broad crested weir. The failure time decreases with increasing the tailwater depth and the reduction from the reference case is found to be 35.3, 45, and 57 % for relative tailwater depth, do /(do)ref = 3, 4, and 5, respectively. The difference between the upstream and downstream water depths decreases with time till it became almost negligible at the end of the experiment. The reduction in cumulative eroded volume is 23, 36.5, and 75 % for relative tailwater depth, do /(do)ref = 3, 4, and 5, respectively. The peak discharge decreases by 7, 14, and 17.35 % for relative tailwater depth, do /(do)ref = 3, 4, and 5, respectively. The relative eroded crest height decreases linearly with increasing the tailwater depth by 10, 41, and 77.6 % for relative tailwater depth, do /(do)ref = 3, 4, and 5, respectively. The numerical model can reproduce the erosion process with a minor deviation from the experimental results, particularly in terms of tracking the degradation of the crest height with time.

    Declaration of Competing Interest

    The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

    Reference

    [1]

    D. McCullough

    The Johnstown Flood

    Simon and Schuster, NY (1968)

    Google Scholar[2]Rose AT. The influence of dam failures on dam safety laws in Pennsylvania. Association of State Dam Safety Officials Annual Conference 2013, Dam Safety 2013. 2013;1:738–56.

    Google Scholar[3]

    M. Foster, R. Fell, M. Spannagle

    The statistics of embankment dam failures and accidents

    Can Geotech J, 37 (5) (2000), pp. 1000-1024, 10.1139/t00-030 View PDF

    View Record in ScopusGoogle Scholar[4]Pickert, G., Jirka, G., Bieberstein, A., Brauns, J. Soil/water interaction during the breaching process of overtopped embankments. In: Greco, M., Carravetta, A., Morte, R.D. (Eds.), Proceedings of the Conference River-Flow 2004, Balkema.

    Google Scholar[5]

    A. Asghari Tabrizi, E. Elalfy, M. Elkholy, M.H. Chaudhry, J. Imran

    Effects of compaction on embankment breach due to overtopping

    J Hydraul Res, 55 (2) (2017), pp. 236-247, 10.1080/00221686.2016.1238014 View PDF

    View Record in ScopusGoogle Scholar[6]

    R.M. Kansoh, M. Elkholy, G. Abo-Zaid

    Effect of Shape Parameters on Failure of Earthen Embankment due to Overtopping

    KSCE J Civ Eng, 24 (5) (2020), pp. 1476-1485, 10.1007/s12205-020-1107-x View PDF

    View Record in ScopusGoogle Scholar[7]

    YongHui Zhu, P.J. Visser, J.K. Vrijling, GuangQian Wang

    Experimental investigation on breaching of embankments

    Experimental investigation on breaching of embankments, 54 (1) (2011), pp. 148-155 View PDF

    CrossRefView Record in ScopusGoogle Scholar[8]Amaral S, Jónatas R, Bento AM, Palma J, Viseu T, Cardoso R, et al. Failure by overtopping of earth dams. Quantification of the discharge hydrograph. Proceedings of the 3rd IAHR Europe Congress: 14-15 April 2014, Portugal. 2014;(1):182–93.

    Google Scholar[9]

    G. Bereta, P. Hui, H. Kai, L. Guang, P. Kefan, Y.Z. Zhao

    Experimental study of cohesive embankment dam breach formation due to overtopping

    Periodica Polytechnica Civil Engineering, 64 (1) (2020), pp. 198-211, 10.3311/PPci.14565 View PDF

    View Record in ScopusGoogle Scholar[10]

    D.K. Verma, B. Setia, V.K. Arora

    Experimental study of breaching of an earthen dam using a fuse plug model

    Int J Eng Trans A, 30 (4) (2017), pp. 479-485, 10.5829/idosi.ije.2017.30.04a.04 View PDF

    View Record in ScopusGoogle Scholar[11]Wu T, Qin J. Experimental Study of a Tailings Impoundment Dam Failure Due to Overtopping. Mine Water and the Environment [Internet]. 2018;37(2):272–80. Available from: doi: 10.1007/s10230-018-0529-x.

    Google Scholar[12]

    A. Feizi Khankandi, A. Tahershamsi, S. Soares-Frazo

    Experimental investigation of reservoir geometry effect on dam-break flow

    J Hydraul Res, 50 (4) (2012), pp. 376-387 View PDF

    CrossRefView Record in ScopusGoogle Scholar[13]

    A. Ritter

    Die Fortpflanzung der Wasserwellen (The propagation of water waves)

    Zeitschrift Verein Deutscher Ingenieure, 36 (33) (1892), pp. 947-954

    [in German]

    View Record in ScopusGoogle Scholar[14]

    I. Rifai, K. El Kadi Abderrezzak, S. Erpicum, P. Archambeau, D. Violeau, M. Pirotton, et al.

    Floodplain Backwater Effect on Overtopping Induced Fluvial Dike Failure

    Water Resour Res, 54 (11) (2018), pp. 9060-9073 View PDF

    This article is free to access.

    CrossRefView Record in ScopusGoogle Scholar[15]

    X. Jiang

    Laboratory Experiments on Breaching Characteristics of Natural Dams on Sloping Beds

    Advances in Civil Engineering, 2019 (2019), pp. 1-14

    View Record in ScopusGoogle Scholar[16]

    H. Ozmen-Cagatay, S. Kocaman

    Dam-break flows during initial stage using SWE and RANS approaches

    J Hydraul Res, 48 (5) (2010), pp. 603-611 View PDF

    CrossRefView Record in ScopusGoogle Scholar[17]

    S. Evangelista

    Experiments and numerical simulations of dike erosion due to a wave impact

    Water (Switzerland), 7 (10) (2015), pp. 5831-5848 View PDF

    CrossRefView Record in ScopusGoogle Scholar[18]

    C. Di Cristo, S. Evangelista, M. Greco, M. Iervolino, A. Leopardi, A. Vacca

    Dam-break waves over an erodible embankment: experiments and simulations

    J Hydraul Res, 56 (2) (2018), pp. 196-210 View PDF

    CrossRefView Record in ScopusGoogle Scholar[19]Goharnejad H, Sm M, Zn M, Sadeghi L, Abadi K. Numerical Modeling and Evaluation of Embankment Dam Break Phenomenon (Case Study : Taleghan Dam) ISSN : 2319-9873. 2016;5(3):104–11.

    Google Scholar[20]Hu H, Zhang J, Li T. Dam-Break Flows : Comparison between Flow-3D , MIKE 3 FM , and Analytical Solutions with Experimental Data. 2018;1–24. doi: 10.3390/app8122456.

    Google Scholar[21]

    R. Kaurav, P.K. Mohapatra, D. Ph

    Studying the Peak Discharge through a Planar Dam Breach, 145 (6) (2019), pp. 1-8 View PDF

    CrossRef[22]

    Z.M. Yusof, Z.A.L. Shirling, A.K.A. Wahab, Z. Ismail, S. Amerudin

    A hydrodynamic model of an embankment breaching due to overtopping flow using FLOW-3D

    IOP Conference Series: Earth and Environmental Science, 920 (1) (2021)

    Google Scholar[23]

    G. Pickert, V. Weitbrecht, A. Bieberstein

    Breaching of overtopped river embankments controlled by apparent cohesion

    J Hydraul Res, 49 (2) (Apr. 2011), pp. 143-156, 10.1080/00221686.2011.552468 View PDF

    View Record in ScopusGoogle Scholar

    Cited by (0)

    My name is Shaimaa Ibrahim Mohamed Aman and I am a teaching assistant in Irrigation and Hydraulics department, Faculty of Engineering, Alexandria University. I graduated from the Faculty of Engineering, Alexandria University in 2013. I had my MSc in Irrigation and Hydraulic Engineering in 2017. My research interests lie in the area of earth dam Failures.

    Peer review under responsibility of Ain Shams University.

    © 2022 THE AUTHORS. Published by Elsevier BV on behalf of Faculty of Engineering, Ain Shams University.

    Extratropical cyclone damage to the seawall in Dawlish, UK: eyewitness accounts, sea level analysis and numerical modelling

    영국 Dawlish의 방파제에 대한 온대 저기압 피해: 목격자 설명, 해수면 분석 및 수치 모델링

    Extratropical cyclone damage to the seawall in Dawlish, UK: eyewitness accounts, sea level analysis and numerical modelling

    Natural Hazards (2022)Cite this article

    Abstract

    2014년 2월 영국 해협(영국)과 특히 Dawlish에 영향을 미친 온대 저기압 폭풍 사슬은 남서부 지역과 영국의 나머지 지역을 연결하는 주요 철도에 심각한 피해를 입혔습니다.

    이 사건으로 라인이 두 달 동안 폐쇄되어 5천만 파운드의 피해와 12억 파운드의 경제적 손실이 발생했습니다. 이 연구에서는 폭풍의 파괴력을 해독하기 위해 목격자 계정을 수집하고 해수면 데이터를 분석하며 수치 모델링을 수행합니다.

    우리의 분석에 따르면 이벤트의 재난 관리는 성공적이고 효율적이었으며 폭풍 전과 도중에 인명과 재산을 구하기 위해 즉각적인 조치를 취했습니다. 파도 부이 분석에 따르면 주기가 4–8, 8–12 및 20–25초인 복잡한 삼중 봉우리 바다 상태가 존재하는 반면, 조위계 기록에 따르면 최대 0.8m의 상당한 파도와 최대 1.5m의 파도 성분이 나타났습니다.

    이벤트에서 가능한 기여 요인으로 결합된 진폭. 최대 286 KN의 상당한 임펄스 파동이 손상의 시작 원인일 가능성이 가장 높았습니다. 수직 벽의 반사는 파동 진폭의 보강 간섭을 일으켜 파고가 증가하고 최대 16.1m3/s/m(벽의 미터 너비당)의 상당한 오버탑핑을 초래했습니다.

    이 정보와 우리의 공학적 판단을 통해 우리는 이 사고 동안 다중 위험 계단식 실패의 가장 가능성 있는 순서는 다음과 같다고 결론을 내립니다. 조적 파괴로 이어지는 파도 충격력, 충전물 손실 및 연속적인 조수에 따른 구조물 파괴.

    The February 2014 extratropical cyclonic storm chain, which impacted the English Channel (UK) and Dawlish in particular, caused significant damage to the main railway connecting the south-west region to the rest of the UK. The incident caused the line to be closed for two months, £50 million of damage and an estimated £1.2bn of economic loss. In this study, we collate eyewitness accounts, analyse sea level data and conduct numerical modelling in order to decipher the destructive forces of the storm. Our analysis reveals that the disaster management of the event was successful and efficient with immediate actions taken to save lives and property before and during the storm. Wave buoy analysis showed that a complex triple peak sea state with periods at 4–8, 8–12 and 20–25 s was present, while tide gauge records indicated that significant surge of up to 0.8 m and wave components of up to 1.5 m amplitude combined as likely contributing factors in the event. Significant impulsive wave force of up to 286 KN was the most likely initiating cause of the damage. Reflections off the vertical wall caused constructive interference of the wave amplitudes that led to increased wave height and significant overtopping of up to 16.1 m3/s/m (per metre width of wall). With this information and our engineering judgement, we conclude that the most probable sequence of multi-hazard cascading failure during this incident was: wave impact force leading to masonry failure, loss of infill and failure of the structure following successive tides.

    Introduction

    The progress of climate change and increasing sea levels has started to have wide ranging effects on critical engineering infrastructure (Shakou et al. 2019). The meteorological effects of increased atmospheric instability linked to warming seas mean we may be experiencing more frequent extreme storm events and more frequent series or chains of events, as well as an increase in the force of these events, a phenomenon called storminess (Mölter et al. 2016; Feser et al. 2014). Features of more extreme weather events in extratropical latitudes (30°–60°, north and south of the equator) include increased gusting winds, more frequent storm squalls, increased prolonged precipitation and rapid changes in atmospheric pressure and more frequent and significant storm surges (Dacre and Pinto 2020). A recent example of these events impacting the UK with simultaneous significant damage to coastal infrastructure was the extratropical cyclonic storm chain of winter 2013/2014 (Masselink et al. 2016; Adams and Heidarzadeh 2021). The cluster of storms had a profound effect on both coastal and inland infrastructure, bringing widespread flooding events and large insurance claims (RMS 2014).

    The extreme storms of February 2014, which had a catastrophic effect on the seawall of the south Devon stretch of the UK’s south-west mainline, caused a two-month closure of the line and significant disruption to the local and regional economy (Fig. 1b) (Network Rail 2014; Dawson et al. 2016; Adams and Heidarzadeh 2021). Restoration costs were £35 m, and economic effects to the south-west region of England were estimated up to £1.2bn (Peninsula Rail Taskforce 2016). Adams and Heidarzadeh (2021) investigated the disparate cascading failure mechanisms which played a part in the failure of the railway through Dawlish and attempted to put these in the context of the historical records of infrastructure damage on the line. Subsequent severe storms in 2016 in the region have continued to cause damage and disruption to the line in the years since 2014 (Met Office 2016). Following the events of 2014, Network Rail Footnote1 who owns the network has undertaken a resilience study. As a result, it has proposed a £400 m refurbishment of the civil engineering assets that support the railway (Fig. 1) (Network Rail 2014). The new seawall structure (Fig. 1a,c), which is constructed of pre-cast concrete sections, encases the existing Brunel seawall (named after the project lead engineer, Isambard Kingdom Brunel) and has been improved with piled reinforced concrete foundations. It is now over 2 m taller to increase the available crest freeboard and incorporates wave return features to minimise wave overtopping. The project aims to increase both the resilience of the assets to extreme weather events as well as maintain or improve amenity value of the coastline for residents and visitors.

    figure 1
    Fig. 1

    In this work, we return to the Brunel seawall and the damage it sustained during the 2014 storms which affected the assets on the evening of the 4th and daytime of the 5th of February and eventually resulted in a prolonged closure of the line. The motivation for this research is to analyse and model the damage made to the seawall and explain the damage mechanisms in order to improve the resilience of many similar coastal structures in the UK and worldwide. The innovation of this work is the multidisciplinary approach that we take comprising a combination of analysis of eyewitness accounts (social science), sea level and wave data analysis (physical science) as well as numerical modelling and engineering judgement (engineering sciences). We investigate the contemporary wave climate and sea levels by interrogating the real-time tide gauge and wave buoys installed along the south-west coast of the English Channel. We then model a typical masonry seawall (Fig. 2), applying the computational fluid dynamics package FLOW3D-Hydro,Footnote2 to quantify the magnitude of impact forces that the seawall would have experienced leading to its failure. We triangulate this information to determine the probable sequence of failures that led to the disaster in 2014.

    figure 2
    Fig. 2

    Data and methods

    Our data comprise eyewitness accounts, sea level records from coastal tide gauges and offshore wave buoys as well as structural details of the seawall. As for methodology, we analyse eyewitness data, process and investigate sea level records through Fourier transform and conduct numerical simulations using the Flow3D-Hydro package (Flow Science 2022). Details of the data and methodology are provided in the following.

    Eyewitness data

    The scale of damage to the seawall and its effects led the local community to document the first-hand accounts of those most closely affected by the storms including residents, local businesses, emergency responders, politicians and engineering contractors involved in the post-storm restoration work. These records now form a permanent exhibition in the local museum in DawlishFootnote3, and some of these accounts have been transcribed into a DVD account of the disaster (Dawlish Museum 2015). We have gathered data from the Dawlish Museum, national and international news reports, social media tweets and videos. Table 1 provides a summary of the eyewitness accounts. Overall, 26 entries have been collected around the time of the incident. Our analysis of the eyewitness data is provided in the third column of Table 1 and is expanded in Sect. 3.Table 1 Eyewitness accounts of damage to the Dawlish railway due to the February 2014 storm and our interpretations

    Full size table

    Sea level data and wave environment

    Our sea level data are a collection of three tide gauge stations (Newlyn, Devonport and Swanage Pier—Fig. 5a) owned and operated by the UK National Tide and Sea Level FacilityFootnote4 for the Environment Agency and four offshore wave buoys (Dawlish, West Bay, Torbay and Chesil Beach—Fig. 6a). The tide gauge sites are all fitted with POL-EKO (www.pol-eko.com.pl) data loggers. Newlyn has a Munro float gauge with one full tide and one mid-tide pneumatic bubbler system. Devonport has a three-channel data pneumatic bubbler system, and Swanage Pier consists of a pneumatic gauge. Each has a sampling interval of 15 min, except for Swanage Pier which has a sampling interval of 10 min. The tide gauges are located within the port areas, whereas the offshore wave buoys are situated approximately 2—3.3 km from the coast at water depths of 10–15 m. The wave buoys are all Datawell Wavemaker Mk III unitsFootnote5 and come with sampling interval of 0.78 s. The buoys have a maximum saturation amplitude of 20.5 m for recording the incident waves which implies that every wave larger than this threshold will be recorded at 20.5 m. The data are provided by the British Oceanographic Data CentreFootnote6 for tide gauges and the Channel Coastal ObservatoryFootnote7 for wave buoys.

    Sea level analysis

    The sea level data underwent quality control to remove outliers and spikes as well as gaps in data (e.g. Heidarzadeh et al. 2022; Heidarzadeh and Satake 2015). We processed the time series of the sea level data using the Matlab signal processing tool (MathWorks 2018). For calculations of the tidal signals, we applied the tidal package TIDALFIT (Grinsted 2008), which is based on fitting tidal harmonics to the observed sea level data. To calculate the surge signals, we applied a 30-min moving average filter to the de-tided data in order to remove all wind, swell and infra-gravity waves from the time series. Based on the surge analysis and the variations of the surge component before the time period of the incident, an error margin of approximately ± 10 cm is identified for our surge analysis. Spectral analysis of the wave buoy data is performed using the fast Fourier transform (FFT) of Matlab package (Mathworks 2018).

    Numerical modelling

    Numerical modelling of wave-structure interaction is conducted using the computational fluid dynamics package Flow3D-Hydro version 1.1 (Flow Science 2022). Flow3D-Hydro solves the transient Navier–Stokes equations of conservation of mass and momentum using a finite difference method and on Eulerian and Lagrangian frameworks (Flow Science 2022). The aforementioned governing equations are:

    ∇.u=0∇.u=0

    (1)

    ∂u∂t+u.∇u=−∇Pρ+υ∇2u+g∂u∂t+u.∇u=−∇Pρ+υ∇2u+g

    (2)

    where uu is the velocity vector, PP is the pressure, ρρ is the water density, υυ is the kinematic viscosity and gg is the gravitational acceleration. A Fractional Area/Volume Obstacle Representation (FAVOR) is adapted in Flow3D-Hydro, which applies solid boundaries within the Eulerian grid and calculates the fraction of areas and volume in partially blocked volume in order to compute flows on corresponding boundaries (Hirt and Nichols 1981). We validated the numerical modelling through comparing the results with Sainflou’s analytical equation for the design of vertical seawalls (Sainflou 1928; Ackhurst 2020), which is as follows:

    pd=ρgHcoshk(d+z)coshkdcosσtpd=ρgHcoshk(d+z)coshkdcosσt

    (3)

    where pdpd is the hydrodynamic pressure, ρρ is the water density, gg is the gravitational acceleration, HH is the wave height, dd is the water depth, kk is the wavenumber, zz is the difference in still water level and mean water level, σσ is the angular frequency and tt is the time. The Sainflou’s equation (Eq. 3) is used to calculate the dynamic pressure from wave action, which is combined with static pressure on the seawall.

    Using Flow3D-Hydro, a model of the Dawlish seawall was made with a computational domain which is 250.0 m in length, 15.0 m in height and 0.375 m in width (Fig. 3a). The computational domain was discretised using a single uniform grid with a mesh size of 0.125 m. The model has a wave boundary at the left side of the domain (x-min), an outflow boundary on the right side (x-max), a symmetry boundary at the bottom (z-min) and a wall boundary at the top (z-max). A wall boundary implies that water or waves are unable to pass through the boundary, whereas a symmetry boundary means that the two edges of the boundary are identical and therefore there is no flow through it. The water is considered incompressible in our model. For volume of fluid advection for the wave boundary (i.e. the left-side boundary) in our simulations, we utilised the “Split Lagrangian Method”, which guarantees the best accuracy (Flow Science, 2022).

    figure 3
    Fig. 3

    The stability of the numerical scheme is controlled and maintained through checking the Courant number (CC) as given in the following:

    C=VΔtΔxC=VΔtΔx

    (4)

    where VV is the velocity of the flow, ΔtΔt is the time step and ΔxΔx is the spatial step (i.e. grid size). For stability and convergence of the numerical simulations, the Courant number must be sufficiently below one (Courant et al. 1928). This is maintained by a careful adjustment of the ΔxΔx and ΔtΔt selections. Flow3D-Hydro applies a dynamic Courant number, meaning the program adjusts the value of time step (ΔtΔt) during the simulations to achieve a balance between accuracy of results and speed of simulation. In our simulation, the time step was in the range ΔtΔt = 0.0051—0.051 s.

    In order to achieve the most efficient mesh resolution, we varied cell size for five values of ΔxΔx = 0.1 m, 0.125 m, 0.15 m, 0.175 m and 0.20 m. Simulations were performed for all mesh sizes, and the results were compared in terms of convergence, stability and speed of simulation (Fig. 3). A linear wave with an amplitude of 1.5 m and a period of 6 s was used for these optimisation simulations. We considered wave time histories at two gauges A and B and recorded the waves from simulations using different mesh sizes (Fig. 3). Although the results are close (Fig. 3), some limited deviations are observed for larger mesh sizes of 0.20 m and 0.175 m. We therefore selected mesh size of 0.125 m as the optimum, giving an extra safety margin as a conservative solution.

    The pressure from the incident waves on the vertical wall is validated in our model by comparing them with the analytical equation of Sainflou (1928), Eq. (3), which is one of the most common set of equations for design of coastal structures (Fig. 4). The model was tested by running a linear wave of period 6 s and wave amplitude of 1.5 m against the wall, with a still water level of 4.5 m. It can be seen that the model results are very close to those from analytical equations of Sainflou (1928), indicating that our numerical model is accurately modelling the wave-structure interaction (Fig. 4).

    figure 4
    Fig. 4

    Eyewitness account analysis

    Contemporary reporting of the 4th and 5th February 2014 storms by the main national news outlets in the UK highlights the extreme nature of the events and the significant damage and disruption they were likely to have on the communities of the south-west of England. In interviews, this was reinforced by Network Rail engineers who, even at this early stage, were forecasting remedial engineering works to last for at least 6 weeks. One week later, following subsequent storms the cascading nature of the events was obvious. Multiple breaches of the seawall had taken place with up to 35 separate landslide events and significant damage to parapet walls along the coastal route also were reported. Residents of the area reported extreme effects of the storm, one likening it to an earthquake and reporting water ingress through doors windows and even through vertical chimneys (Table 1). This suggests extreme wave overtopping volumes and large wave impact forces. One resident described the structural effects as: “the house was jumping up and down on its footings”.

    Disaster management plans were quickly and effectively put into action by the local council, police service and National Rail. A major incident was declared, and decisions regarding evacuation of the residents under threat were taken around 2100 h on the night of 4th February when reports of initial damage to the seawall were received (Table 1). Local hotels were asked to provide short-term refuge to residents while local leisure facilities were prepared to accept residents later that evening. Initial repair work to the railway line was hampered by successive high spring tides and storms in the following days although significant progress was still made when weather conditions permitted (Table 1).

    Sea level observations and spectral analysis

    The results of surge and wave analyses are presented in Figs. 5 and 6. A surge height of up to 0.8 m was recorded in the examined tide gauge stations (Fig. 5b-d). Two main episodes of high surge heights are identified: the first surge started on 3rd February 2014 at 03:00 (UTC) and lasted until 4th of February 2014 at 00:00; the second event occurred in the period 4th February 2014 15:00 to 5th February 2014 at 17:00 (Fig. 5b-d). These data imply surge durations of 21 h and 26 h for the first and the second events, respectively. Based on the surge data in Fig. 5, we note that the storm event of early February 2014 and the associated surges was a relatively powerful one, which impacted at least 230 km of the south coast of England, from Land’s End to Weymouth, with large surge heights.

    figure 5
    Fig. 5
    figure 6
    Fig. 6

    Based on wave buoy records, the maximum recorded amplitudes are at least 20.5 m in Dawlish and West Bay, 1.9 m in Tor Bay and 4.9 m in Chesil (Fig. 6a-b). The buoys at Tor Bay and Chesil recorded dual peak period bands of 4–8 and 8–12 s, whereas at Dawlish and West Bay registered triple peak period bands at 4–8, 8–12 and 20–25 s (Fig. 6c, d). It is important to note that the long-period waves at 20–25 s occur with short durations (approximately 2 min) while the waves at the other two bands of 4–8 and 8–12 s appear to be present at all times during the storm event.

    The wave component at the period band of 4–8 s can be most likely attributed to normal coastal waves while the one at 8–12 s, which is longer, is most likely the swell component of the storm. Regarding the third component of the waves with long period of 20 -25 s, which occurs with short durations of 2 min, there are two hypotheses; it is either the result of a local (port and harbour) and regional (the Lyme Bay) oscillations (eg. Rabinovich 1997; Heidarzadeh and Satake 2014; Wang et al. 1992), or due to an abnormally long swell. To test the first hypothesis, we consider various water bodies such as Lyme Bay (approximate dimensions of 70 km × 20 km with an average water depth of 30 m; Fig. 6), several local bays (approximate dimensions of 3.6 km × 0.6 km with an average water depth of 6 m) and harbours (approximate dimensions of 0.5 km × 0.5 km with an average water depth of 4 m). Their water depths are based on the online Marine navigation website.Footnote8 According to Rabinovich (2010), the oscillation modes of a semi-enclosed rectangle basin are given by the following equation:

    Tmn=2gd−−√[(m2L)2+(nW)2]−1/2Tmn=2gd[(m2L)2+(nW)2]−1/2

    (5)

    where TmnTmn is the oscillation period, gg is the gravitational acceleration, dd is the water depth, LL is the length of the basin, WW is the width of the basin, m=1,2,3,…m=1,2,3,… and n=0,1,2,3,…n=0,1,2,3,…; mm and nn are the counters of the different modes. Applying Eq. (5) to the aforementioned water bodies results in oscillation modes of at least 5 min, which is far longer than the observed period of 20–25 s. Therefore, we rule out the first hypothesis and infer that the long period of 20–25 s is most likely a long swell wave coming from distant sources. As discussed by Rabinovich (1997) and Wang et al. (2022), comparison between sea level spectra before and after the incident is a useful method to distinguish the spectrum of the weather event. A visual inspection of Fig. 6 reveals that the forcing at the period band of 20–25 s is non-existent before the incident.

    Numerical simulations of wave loading and overtopping

    Based on the results of sea level data analyses in the previous section (Fig. 6), we use a dual peak wave spectrum with peak periods of 10.0 s and 25.0 s for numerical simulations because such a wave would be comprised of the most energetic signals of the storm. For variations of water depth (2.0–4.0 m), coastal wave amplitude (0.5–1.5 m) (Fig. 7) and storm surge height (0.5–0.8 m) (Fig. 5), we developed 20 scenarios (Scn) which we used in numerical simulations (Table 2). Data during the incident indicated that water depth was up to the crest level of the seawall (approximately 4 m water depth); therefore, we varied water depth from 2 to 4 m in our simulation scenarios. Regarding wave amplitudes, we referred to the variations at a nearby tide gauge station (West Bay) which showed wave amplitude up to 1.2 m (Fig. 7). Therefore, wave amplitude was varied from 0.5 m to 1.5 m by considering a factor a safety of 25% for the maximum wave amplitude. As for the storm surge component, time series of storm surges calculated at three coastal stations adjacent to Dawlish showed that it was in the range of 0.5 m to 0.8 m (Fig. 5). These 20 scenarios would help to study uncertainties associated with wave amplitudes and pressures. Figure 8 shows snapshots of wave propagation and impacts on the seawall at different times.

    figure 7
    Fig. 7

    Table 2 The 20 scenarios considered for numerical simulations in this study

    Full size table

    figure 8
    Fig. 8

    Results of wave amplitude simulations

    Large wave amplitudes can induce significant wave forcing on the structure and cause overtopping of the seawall, which could eventually cascade to other hazards such as erosion of the backfill and scour (Adams and Heidarzadeh, 2021). The first 10 scenarios of our modelling efforts are for the same incident wave amplitudes of 0.5 m, which occur at different water depths (2.0–4.0 m) and storm surge heights (0.5–0.8 m) (Table 2 and Fig. 9). This is because we aim at studying the impacts of effective water depth (deff—the sum of mean sea level and surge height) on the time histories of wave amplitudes as the storm evolves. As seen in Fig. 9a, by decreasing effective water depth, wave amplitude increases. For example, for Scn-1 with effective depth of 4.5 m, the maximum amplitude of the first wave is 1.6 m, whereas it is 2.9 m for Scn-2 with effective depth of 3.5 m. However, due to intensive reflections and interferences of the waves in front of the vertical seawall, such a relationship is barely seen for the second and the third wave peaks. It is important to note that the later peaks (second or third) produce the largest waves rather than the first wave. Extraordinary wave amplifications are seen for the Scn-2 (deff = 3.5 m) and Scn-7 (deff = 3.3 m), where the corresponding wave amplitudes are 4.5 m and 3.7 m, respectively. This may indicate that the effective water depth of deff = 3.3–3.5 m is possibly a critical water depth for this structure resulting in maximum wave amplitudes under similar storms. In the second wave impact, the combined wave height (i.e. the wave amplitude plus the effective water depth), which is ultimately an indicator of wave overtopping, shows that the largest wave heights are generated by Scn-2, 7 and 8 (Fig. 9a) with effective water depths of 3.5 m, 3.3 m and 3.8 m and combined heights of 8.0 m, 7.0 m and 6.9 m (Fig. 9b). Since the height of seawall is 5.4 m, the combined wave heights for Scn-2, 7 and 8 are greater than the crest height of the seawall by 2.6 m, 1.6 m and 1.5 m, respectively, which indicates wave overtopping.

    figure 9
    Fig. 9

    For scenarios 11–20 (Fig. 10), with incident wave amplitudes of 1.5 m (Table 2), the largest wave amplitudes are produced by Scn-17 (deff = 3.3 m), Scn-13 (deff = 2.5 m) and Scn-12 (deff = 3.5 m), which are 5.6 m, 5.1 m and 4.5 m. The maximum combined wave heights belong to Scn-11 (deff = 4.5 m) and Scn-17 (deff = 3.3 m), with combined wave heights of 9.0 m and 8.9 m (Fig. 10b), which are greater than the crest height of the seawall by 4.6 m and 3.5 m, respectively.

    figure 10
    Fig. 10

    Our simulations for all 20 scenarios reveal that the first wave is not always the largest and wave interactions, reflections and interferences play major roles in amplifying the waves in front of the seawall. This is primarily because the wall is fully vertical and therefore has a reflection coefficient of close to one (i.e. full reflection). Simulations show that the combined wave height is up to 4.6 m higher than the crest height of the wall, implying that severe overtopping would be expected.

    Results of wave loading calculations

    The pressure calculations for scenarios 1–10 are given in Fig. 11 and those of scenarios 11–20 in Fig. 12. The total pressure distribution in Figs. 1112 mostly follows a triangular shape with maximum pressure at the seafloor as expected from the Sainflou (1928) design equations. These pressure plots comprise both static (due to mean sea level in front of the wall) and dynamic (combined effects of surge and wave) pressures. For incident wave amplitudes of 0.5 m (Fig. 11), the maximum wave pressure varies in the range of 35–63 kPa. At the sea surface, it is in the range of 4–20 kPa (Fig. 11). For some scenarios (Scn-2 and 7), the pressure distribution deviates from a triangular shape and shows larger pressures at the top, which is attributed to the wave impacts and partial breaking at the sea surface. This adds an additional triangle-shaped pressure distribution at the sea surface elevation consistent with the design procedure developed by Goda (2000) for braking waves. The maximum force on the seawall due to scenarios 1–10, which is calculated by integrating the maximum pressure distribution over the wave-facing surface of the seawall, is in the range of 92–190 KN (Table 2).

    figure 11
    Fig. 11
    figure 12
    Fig. 12

    For scenarios 11–20, with incident wave amplitude of 1.5 m, wave pressures of 45–78 kPa and 7–120 kPa, for  the bottom and top of the wall, respectively, were observed (Fig. 12). Most of the plots show a triangular pressure distribution, except for Scn-11 and 15. A significant increase in wave impact pressure is seen for Scn-15 at the top of the structure, where a maximum pressure of approximately 120 kPa is produced while other scenarios give a pressure of 7–32 kPa for the sea surface. In other words, the pressure from Scn-15 is approximately four times larger than the other scenarios. Such a significant increase of the pressure at the top is most likely attributed to the breaking wave impact loads as detailed by Goda (2000) and Cuomo et al. (2010). The wave simulation snapshots in Fig. 8 show that the wave breaks before reaching the wall. The maximum force due to scenarios 11–20 is 120–286 KN.

    The breaking wave impacts peaking at 286 KN in our simulations suggest destabilisation of the upper masonry blocks, probably by grout malfunction. This significant impact force initiated the failure of the seawall which in turn caused extensive ballast erosion. Wave impact damage was proposed by Adams and Heidarzadeh (2021) as one of the primary mechanisms in the 2014 Dawlish disaster. In the multi-hazard risk model proposed by these authors, damage mechanism III (failure pathway 5 in Adams and Heidarzadeh, 2021) was characterised by wave impact force causing damage to the masonry elements, leading to failure of the upper sections of the seawall and loss of infill material. As blocks were removed, access to the track bed was increased for inbound waves allowing infill material from behind the seawall to be fluidised and subsequently removed by backwash. The loss of infill material critically compromised the stability of the seawall and directly led to structural failure. In parallel, significant wave overtopping (discussed in the next section) led to ballast washout and cascaded, in combination with masonry damage, to catastrophic failure of the wall and suspension of the rails in mid-air (Fig. 1b), leaving the railway inoperable for two months.

    Wave Overtopping

    The two most important factors contributing to the 2014 Dawlish railway catastrophe were wave impact forces and overtopping. Figure 13 gives the instantaneous overtopping rates for different scenarios, which experienced overtopping. It can be seen that the overtopping rates range from 0.5 m3/s/m to 16.1 m3/s/m (Fig. 13). Time histories of the wave overtopping rates show that the phenomenon occurs intermittently, and each time lasts 1.0–7.0 s. It is clear that the longer the overtopping time, the larger the volume of the water poured on the structure. The largest wave overtopping rates of 16.1 m3/s/m and 14.4 m3/s/m belong to Scn-20 and 11, respectively. These are the two scenarios that also give the largest combined wave heights (Fig. 10b).

    figure 13
    Fig. 13

    The cumulative overtopping curves (Figs. 1415) show the total water volume overtopped the structure during the entire simulation time. This is an important hazard factor as it determines the level of soil saturation, water pore pressure in the soil and soil erosion (Van der Meer et al. 2018). The maximum volume belongs to Scn-20, which is 65.0 m3/m (m-cubed of water per metre length of the wall). The overtopping volumes are 42.7 m3/m for Scn-11 and 28.8 m3/m for Scn-19. The overtopping volume is in the range of 0.7–65.0 m3/m for all scenarios.

    figure 14
    Fig. 14
    figure 15
    Fig. 15

    For comparison, we compare our modelling results with those estimated using empirical equations. For the case of the Dawlish seawall, we apply the equation proposed by Van Der Meer et al. (2018) to estimate wave overtopping rates, based on a set of decision criteria which are the influence of foreshore, vertical wall, possible breaking waves and low freeboard:

    qgH3m−−−−√=0.0155(Hmhs)12e(−2.2RcHm)qgHm3=0.0155(Hmhs)12e(−2.2RcHm)

    (6)

    where qq is the mean overtopping rate per metre length of the seawall (m3/s/m), gg is the acceleration due to gravity, HmHm is the incident wave height at the toe of the structure, RcRc is the wall crest height above mean sea level, hshs is the deep-water significant wave height and e(x)e(x) is the exponential function. It is noted that Eq. (6) is valid for 0.1<RcHm<1.350.1<RcHm<1.35. For the case of the Dawlish seawall and considering the scenarios with larger incident wave amplitude of 1.5 m (hshs= 1.5 m), the incident wave height at the toe of the structure is HmHm = 2.2—5.6 m, and the wall crest height above mean sea level is RcRc = 0.6–2.9 m. As a result, Eq. (6) gives mean overtopping rates up to approximately 2.9 m3/s/m. A visual inspection of simulated overtopping rates in Fig. 13 for Scn 11–20 shows that the mean value of the simulated overtopping rates (Fig. 13) is close to estimates using Eq. (6).

    Discussion and conclusions

    We applied a combination of eyewitness account analysis, sea level data analysis and numerical modelling in combination with our engineering judgement to explain the damage to the Dawlish railway seawall in February 2014. Main findings are:

    • Eyewitness data analysis showed that the extreme nature of the event was well forecasted in the hours prior to the storm impact; however, the magnitude of the risks to the structures was not well understood. Multiple hazards were activated simultaneously, and the effects cascaded to amplify the damage. Disaster management was effective, exemplified by the establishment of an emergency rendezvous point and temporary evacuation centre during the storm, indicating a high level of hazard awareness and preparedness.
    • Based on sea level data analysis, we identified triple peak period bands at 4–8, 8–12 and 20–25 s in the sea level data. Storm surge heights and wave oscillations were up to 0.8 m and 1.5 m, respectively.
    • Based on the numerical simulations of 20 scenarios with different water depths, incident wave amplitudes, surge heights and peak periods, we found that the wave oscillations at the foot of the seawall result in multiple wave interactions and interferences. Consequently, large wave amplitudes, up to 4.6 m higher than the height of the seawall, were generated and overtopped the wall. Extreme impulsive wave impact forces of up to 286 KN were generated by the waves interacting with the seawall.
    • We measured maximum wave overtopping rates of 0.5–16.1 m3/s/m for our scenarios. The cumulative overtopping water volumes per metre length of the wall were 0.7–65.0 m3/m.
    • Analysis of all the evidence combined with our engineering judgement suggests that the most likely initiating cause of the failure was impulsive wave impact forces destabilising one or more grouted joints between adjacent masonry blocks in the wall. Maximum observed pressures of 286 KN in our simulations are four times greater in magnitude than background pressures leading to block removal and initiating failure. Therefore, the sequence of cascading events was :1) impulsive wave impact force causing damage to masonry, 2) failure of the upper sections of the seawall, 3) loss of infill resulting in a reduction of structural strength in the landward direction, 4) ballast washout as wave overtopping and inbound wave activity increased and 5) progressive structural failure following successive tides.

    From a risk mitigation point of view, the stability of the seawall in the face of future energetic cyclonic storm events and sea level rise will become a critical factor in protecting the rail network. Mitigation efforts will involve significant infrastructure investment to strengthen the civil engineering assets combined with improved hazard warning systems consisting of meteorological forecasting and real-time wave observations and instrumentation. These efforts must take into account the amenity value of coastal railway infrastructure to local communities and the significant number of tourists who visit every year. In this regard, public awareness and active engagement in the planning and execution of the project will be crucial in order to secure local stakeholder support for the significant infrastructure project that will be required for future resilience.

    Notes

    1. https://www.networkrail.co.uk/..
    2. https://www.flow3d.com/products/flow-3d-hydro/.
    3. https://www.devonmuseums.net/Dawlish-Museum/Devon-Museums/.
    4. https://ntslf.org/.
    5. https://www.datawell.nl/Products/Buoys/DirectionalWaveriderMkIII.aspx.
    6. https://www.bodc.ac.uk/.
    7. https://coastalmonitoring.org/cco/.
    8. https://webapp.navionics.com/#boating@8&key=iactHlwfP.

    References

    Download references

    Acknowledgements

    We are grateful to Brunel University London for administering the scholarship awarded to KA. The Flow3D-Hydro used in this research for numerical modelling is licenced to Brunel University London through an academic programme contract. We sincerely thank Prof Harsh Gupta (Editor-in-Chief) and two anonymous reviewers for their constructive review comments.

    Funding

    This project was funded by the UK Engineering and Physical Sciences Research Council (EPSRC) through a PhD scholarship to Keith Adams.

    Author information

    Authors and Affiliations

    1. Department of Civil and Environmental Engineering, Brunel University London, Uxbridge, UB8 3PH, UKKeith Adams
    2. Department of Architecture and Civil Engineering, University of Bath, Bath, BA2 7AY, UKMohammad Heidarzadeh

    Corresponding author

    Correspondence to Keith Adams.

    Ethics declarations

    Conflict of interest

    The authors have no relevant financial or non-financial interests to disclose.

    Availability of data

    All data used in this study are provided in the body of the article.

    Additional information

    Publisher’s Note

    Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

    Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

    Reprints and Permissions

    About this article

    Cite this article

    Adams, K., Heidarzadeh, M. Extratropical cyclone damage to the seawall in Dawlish, UK: eyewitness accounts, sea level analysis and numerical modelling. Nat Hazards (2022). https://doi.org/10.1007/s11069-022-05692-2

    Download citation

    • Received17 May 2022
    • Accepted17 October 2022
    • Published14 November 2022
    • DOIhttps://doi.org/10.1007/s11069-022-05692-2

    Share this article

    Anyone you share the following link with will be able to read this content:Get shareable link

    Provided by the Springer Nature SharedIt content-sharing initiative

    Keywords

    • Storm surge
    • Cyclone
    • Railway
    • Climate change
    • Infrastructure
    • Resilience
    Fig. 1. (a) Dimensions of the casting with runners (unit: mm), (b) a melt flow simulation using Flow-3D software together with Reilly's model[44], predicted that a large amount of bifilms (denoted by the black particles) would be contained in the final casting. (c) A solidification simulation using Pro-cast software showed that no shrinkage defect was contained in the final casting.

    AZ91 합금 주물 내 연행 결함에 대한 캐리어 가스의 영향

    TianLiabJ.M.T.DaviesaXiangzhenZhuc
    aUniversity of Birmingham, Birmingham B15 2TT, United Kingdom
    bGrainger and Worrall Ltd, Bridgnorth WV15 5HP, United Kingdom
    cBrunel Centre for Advanced Solidification Technology, Brunel University London, Kingston Ln, London, Uxbridge UB8 3PH, United Kingdom

    Abstract

    An entrainment defect (also known as a double oxide film defect or bifilm) acts a void containing an entrapped gas when submerged into a light-alloy melt, thus reducing the quality and reproducibility of the final castings. Previous publications, carried out with Al-alloy castings, reported that this trapped gas could be subsequently consumed by the reaction with the surrounding melt, thus reducing the void volume and negative effect of entrainment defects. Compared with Al-alloys, the entrapped gas within Mg-alloy might be more efficiently consumed due to the relatively high reactivity of magnesium. However, research into the entrainment defects within Mg alloys has been significantly limited. In the present work, AZ91 alloy castings were produced under different carrier gas atmospheres (i.e., SF6/CO2, SF6/air). The evolution processes of the entrainment defects contained in AZ91 alloy were suggested according to the microstructure inspections and thermodynamic calculations. The defects formed in the different atmospheres have a similar sandwich-like structure, but their oxide films contained different combinations of compounds. The use of carrier gases, which were associated with different entrained-gas consumption rates, affected the reproducibility of AZ91 castings.

    연행 결함(이중 산화막 결함 또는 이중막이라고도 함)은 경합금 용융물에 잠길 때 갇힌 가스를 포함하는 공극으로 작용하여 최종 주물의 품질과 재현성을 저하시킵니다. Al-합금 주물을 사용하여 수행된 이전 간행물에서는 이 갇힌 가스가 주변 용융물과의 반응에 의해 후속적으로 소모되어 공극 부피와 연행 결함의 부정적인 영향을 줄일 수 있다고 보고했습니다. Al-합금에 비해 마그네슘의 상대적으로 높은 반응성으로 인해 Mg-합금 내에 포집된 가스가 더 효율적으로 소모될 수 있습니다. 그러나 Mg 합금 내 연행 결함에 대한 연구는 상당히 제한적이었습니다. 현재 작업에서 AZ91 합금 주물은 다양한 캐리어 가스 분위기(즉, SF6/CO2, SF6/공기)에서 생산되었습니다. AZ91 합금에 포함된 연행 결함의 진화 과정은 미세 조직 검사 및 열역학 계산에 따라 제안되었습니다. 서로 다른 분위기에서 형성된 결함은 유사한 샌드위치 구조를 갖지만 산화막에는 서로 다른 화합물 조합이 포함되어 있습니다. 다른 동반 가스 소비율과 관련된 운반 가스의 사용은 AZ91 주물의 재현성에 영향을 미쳤습니다.

    Keywords

    Magnesium alloy, Casting, Oxide film, Bifilm, Entrainment defect, Reproducibility

    1. Introduction

    As the lightest structural metal available on Earth, magnesium became one of the most attractive light metals over the last few decades. The magnesium industry has consequently experienced a rapid development in the last 20 years [1,2], indicating a large growth in demand for Mg alloys all over the world. Nowadays, the use of Mg alloys can be found in the fields of automobiles, aerospace, electronics and etc.[3,4]. It has been predicted that the global consumption of Mg metals will further increase in the future, especially in the automotive industry, as the energy efficiency requirement of both traditional and electric vehicles further push manufactures lightweight their design [3,5,6].

    The sustained growth in demand for Mg alloys motivated a wide interest in the improvement of the quality and mechanical properties of Mg-alloy castings. During a Mg-alloy casting process, surface turbulence of the melt can lead to the entrapment of a doubled-over surface film containing a small quantity of the surrounding atmosphere, thus forming an entrainment defect (also known as a double oxide film defect or bifilm) [7][8][9][10]. The random size, quantity, orientation, and placement of entrainment defects are widely accepted to be significant factors linked to the variation of casting properties [7]. In addition, Peng et al. [11] found that entrained oxides films in AZ91 alloy melt acted as filters to Al8Mn5 particles, trapping them as they settle. Mackie et al. [12] further suggested that entrained oxide films can act to trawl the intermetallic particles, causing them to cluster and form extremely large defects. The clustering of intermetallic compounds made the entrainment defects more detrimental for the casting properties.

    Most of the previous studies regarding entrainment defects were carried out on Al-alloys [7,[13][14][15][16][17][18], and a few potential methods have been suggested for diminishing their negative effect on the quality of Al-alloy castings. Nyahumwa et al.,[16] shows that the void volume within entrainment defects could be reduced by a hot isostatic pressing (HIP) process. Campbell [7] suggested the entrained gas within the defects could be consumed due to reaction with the surrounding melt, which was further verified by Raiszedeh and Griffiths [19].The effect of the entrained gas consumption on the mechanical properties of Al-alloy castings has been investigated by [8,9], suggesting that the consumption of the entrained gas promoted the improvement of the casting reproducibility.

    Compared with the investigation concerning the defects within Al-alloys, research into the entrainment defects within Mg-alloys has been significantly limited. The existence of entrainment defects has been demonstrated in Mg-alloy castings [20,21], but their behaviour, evolution, as well as entrained gas consumption are still not clear.

    In a Mg-alloy casting process, the melt is usually protected by a cover gas to avoid magnesium ignition. The cavities of sand or investment moulds are accordingly required to be flushed with the cover gas prior to the melt pouring [22]. Therefore, the entrained gas within Mg-alloy castings should contain the cover gas used in the casting process, rather than air only, which may complicate the structure and evolution of the corresponding entrainment defects.

    SF6 is a typical cover gas widely used for Mg-alloy casting processes [23][24][25]. Although this cover gas has been restricted to use in European Mg-alloy foundries, a commercial report has pointed out that this cover is still popular in global Mg-alloy industry, especially in the countries which dominated the global Mg-alloy production, such as China, Brazil, India, etc. [26]. In addition, a survey in academic publications also showed that this cover gas was widely used in recent Mg-alloy studies [27]. The protective mechanism of SF6 cover gas (i.e., the reaction between liquid Mg-alloy and SF6 cover gas) has been investigated by several previous researchers, but the formation process of the surface oxide film is still not clearly understood, and even some published results are conflicting with each other. In early 1970s, Fruehling [28] found that the surface film formed under SF6 was MgO mainly with traces of fluorides, and suggested that SF6 was absorbed in the Mg-alloy surface film. Couling [29] further noticed that the absorbed SF6 reacted with the Mg-alloy melt to form MgF2. In last 20 years, different structures of the Mg-alloy surface films have been reported, as detailed below.(1)

    Single-layered film. Cashion [30,31] used X-ray Photoelectron Spectroscopy (XPS) and Auger Spectroscopy (AES) to identify the surface film as MgO and MgF2. He also found that composition of the film was constant throughout the thickness and the whole experimental holding time. The film observed by Cashion had a single-layered structure created from a holding time from 10 min to 100 min.(2)

    Double-layered film. Aarstad et. al [32] reported a doubled-layered surface oxide film in 2003. They observed several well-distributed MgF2 particles attached to the preliminary MgO film and grew until they covered 25–50% of the total surface area. The inward diffusion of F through the outer MgO film was the driving force for the evolution process. This double-layered structure was also supported by Xiong’s group [25,33] and Shih et al. [34].(3)

    Triple-layered film. The triple-layered film and its evolution process were reported in 2002 by Pettersen [35]. Pettersen found that the initial surface film was a MgO phase and then gradually evolved to the stable MgF2 phase by the inward diffusion of F. In the final stage, the film has a triple-layered structure with a thin O-rich interlayer between the thick top and bottom MgF2 layers.(4)

    Oxide film consisted of discrete particles. Wang et al [36] stirred the Mg-alloy surface film into the melt under a SF6 cover gas, and then inspect the entrained surface film after the solidification. They found that the entrained surface films were not continues as the protective surface films reported by other researchers but composed of discrete particles. The young oxide film was composed of MgO nano-sized oxide particles, while the old oxide films consist of coarse particles (about 1  µm in average size) on one side that contained fluorides and nitrides.

    The oxide films of a Mg-alloy melt surface or an entrained gas are both formed due to the reaction between liquid Mg-alloy and the cover gas, thus the above-mentioned research regarding the Mg-alloy surface film gives valuable insights into the evolution of entrainment defects. The protective mechanism of SF6 cover gas (i.e., formation of a Mg-alloy surface film) therefore indicated a potential complicated evolution process of the corresponding entrainment defects.

    However, it should be noted that the formation of a surface film on a Mg-alloy melt is in a different situation to the consumption of an entrained gas that is submerged into the melt. For example, a sufficient amount of cover gas was supported during the surface film formation in the studies previously mentioned, which suppressed the depletion of the cover gas. In contrast, the amount of entrained gas within a Mg-alloy melt is finite, and the entrained gas may become fully depleted. Mirak [37] introduced 3.5%SF6/air bubbles into a pure Mg-alloy melt solidifying in a specially designed permanent mould. It was found that the gas bubbles were entirely consumed, and the corresponding oxide film was a mixture of MgO and MgF2. However, the nucleation sites (such as the MgF2 spots observed by Aarstad [32] and Xiong [25,33]) were not observed. Mirak also speculated that the MgF2 formed prior to MgO in the oxide film based on the composition analysis, which was opposite to the surface film formation process reported in previous literatures (i.e., MgO formed prior to MgF2). Mirak’s work indicated that the oxide-film formation of an entrained gas may be quite different from that of surface films, but he did not reveal the structure and evolution of the oxide films.

    In addition, the use of carrier gas in the cover gases also influenced the reaction between the cover gas and the liquid Mg-alloy. SF6/air required a higher content of SF6 than did a SF6/CO2 carrier gas [38], to avoid the ignition of molten magnesium, revealing different gas-consumption rates. Liang et.al [39] suggested that carbon was formed in the surface film when CO2 was used as a carrier gas, which was different from the films formed in SF6/air. An investigation into Mg combustion [40] reported a detection of Mg2C3 in the Mg-alloy sample after burning in CO2, which not only supported Liang’s results, but also indicated a potential formation of Mg carbides in double oxide film defects.

    The work reported here is an investigation into the behaviour and evolution of entrainment defects formed in AZ91 Mg-alloy castings, protected by different cover gases (i.e., SF6/air and SF6/CO2). These carrier gases have different protectability for liquid Mg alloy, which may be therefore associated with different consumption rates and evolution processes of the corresponding entrained gases. The effect of the entrained-gas consumption on the reproducibility of AZ91 castings was also studied.

    2. Experiment

    2.1. Melting and casting

    Three kilograms AZ91 alloy was melted in a mild steel crucible at 700 ± 5 °C. The composition of the AZ91 alloy has been shown in Table 1. Prior to heating, all oxide scale on the ingot surface was removed by machining. The cover gases used were 0.5%SF6/air or 0.5%SF6/CO2 (vol.%) at a flow rate of 6 L/min for different castings. The melt was degassed by argon with a flow rate of 0.3 L/min for 15 min [41,42], and then poured into sand moulds. Prior to pouring, the sand mould cavity was flushed with the cover gas for 20 min [22]. The residual melt (around 1 kg) was solidified in the crucible.

    Table 1. Composition (wt.%) of the AZ91 alloy used in this study.

    AlZnMnSiFeNiMg
    9.40.610.150.020.0050.0017Residual

    Fig. 1(a) shows the dimensions of the casting with runners. A top-filling system was deliberately used to generate entrainment defects in the final castings. Green and Campbell [7,43] suggested that a top-filling system caused more entrainment events (i.e., bifilms) during a casting process, compared with a bottom-filling system. A melt flow simulation (Flow-3D software) of this mould, using Reilly’s model [44] regarding the entrainment events, also predicted that a large amount of bifilms would be contained in the final casting (denoted by the black particles in Fig. 1b).

    Fig. 1. (a) Dimensions of the casting with runners (unit: mm), (b) a melt flow simulation using Flow-3D software together with Reilly's model[44], predicted that a large amount of bifilms (denoted by the black particles) would be contained in the final casting. (c) A solidification simulation using Pro-cast software showed that no shrinkage defect was contained in the final casting.

    Shrinkage defects also affect the mechanical properties and reproducibility of castings. Since this study focused on the effect of bifilms on the casting quality, the mould has been deliberately designed to avoid generating shrinkage defects. A solidification simulation using ProCAST software showed that no shrinkage defect would be contained in the final casting, as shown in Fig. 1c. The casting soundness has also been confirmed using a real time X-ray prior to the test bar machining.

    The sand moulds were made from resin-bonded silica sand, containing 1wt. % PEPSET 5230 resin and 1wt. % PEPSET 5112 catalyst. The sand also contained 2 wt.% Na2SiF6 to act as an inhibitor [45]. The pouring temperature was 700 ± 5 °C. After the solidification, a section of the runner bars was sent to the Sci-Lab Analytical Ltd for a H-content analysis (LECO analysis), and all the H-content measurements were carried out on the 5th day after the casting process. Each of the castings was machined into 40 test bars for a tensile strength test, using a Zwick 1484 tensile test machine with a clip extensometer. The fracture surfaces of the broken test bars were examined using Scanning Electron Microscope (SEM, Philips JEOL7000) with an accelerating voltage of 5–15 kV. The fractured test bars, residual Mg-alloy solidified in the crucible, and the casting runners were then sectioned, polished and also inspected using the same SEM. The cross-section of the oxide film found on the test-bar fracture surface was exposed by the Focused Ion Beam milling technique (FIB), using a CFEI Quanta 3D FEG FIB-SEM. The oxide film required to be analysed was coated with a platinum layer. Then, a gallium ion beam, accelerated to 30 kV, milled the material substrate surrounding the platinum coated area to expose the cross section of the oxide film. EDS analysis of the oxide film’s cross section was carried out using the FIB equipment at accelerating voltage of 30 kV.

    2.2. Oxidation cell

    As previously mentioned, several past researchers investigated the protective film formed on a Mg-alloy melt surface [38,39,[46][47][48][49][50][51][52]. During these experiments, the amount of cover gas used was sufficient, thus suppressing the depletion of fluorides in the cover gas. The experiment described in this section used a sealed oxidation cell, which limited the supply of cover gas, to study the evolution of the oxide films of entrainment defects. The cover gas contained in the oxidation cell was regarded as large-size “entrained bubble”.

    As shown in Fig. 2, the main body of the oxidation cell was a closed-end mild steel tube which had an inner length of 400 mm, and an inner diameter of 32 mm. A water-cooled copper tube was wrapped around the upper section of the cell. When the tube was heated, the cooling system created a temperature difference between the upper and lower sections, causing the interior gas to convect within the tube. The temperature was monitored by a type-K thermocouple located at the top of the crucible. Nie et al. [53] suggested that the SF6 cover gas would react with the steel wall of the holding furnace when they investigated the surface film of a Mg-alloy melt. To avoid this reaction, the interior surface of the steel oxidation cell (shown in Fig. 2) and the upper half section of the thermocouple were coated with boron nitride (the Mg-alloy was not in contact with boron nitride).

    Fig. 2. Schematic of the oxidation cell used to study the evolution of the oxide films of the entrainment defects (unit mm).

    During the experiment, a block of solid AZ91 alloy was placed in a magnesia crucible located at the bottom of the oxidation cell. The cell was heated to 100 °C in an electric resistance furnace under a gas flow rate of 1 L/min. The cell was held at this temperature for 20 min, to replace the original trapped atmosphere (i.e. air). Then, the oxidation cell was further heated to 700 °C, melting the AZ91 sample. The gas inlet and exit valves were then closed, creating a sealed environment for oxidation under a limited supply of cover gas. The oxidation cell was then held at 700 ± 10 °C for periods of time from 5 min to 30 min in 5-min intervals. At the end of each holding time, the cell was quenched in water. After cooling to room temperature, the oxidised sample was sectioned, polished, and subsequently examined by SEM.

    3. Results

    3.1. Structure and composition of the entrainment defects formed in SF6/air

    The structure and composition of the entrainment defect formed in the AZ91 castings under a cover gas of 0.5%SF6/air was observed by SEM and EDS. The results indicate that there exist two types of entrainment defects which are sketched in Fig. 3: (1) Type A defect whose oxide film has a traditional single-layered structure and (2) Type B defect, whose oxide film has two layers. The details of these defects were introduced in the following. Here it should be noticed that, as the entrainment defects are also known as biofilms or double oxide film, the oxide films of Type B defect were referred to as “multi-layered oxide film” or “multi-layered structure” in the present work to avoid a confusing description such as “the double-layered oxide film of a double oxide film defect”.

    Fig. 3. Schematic of the different types of entrainment defects found in AZ91 castings. (a) Type A defect with a single-layered oxide film and (b) Type B defect with two-layered oxide film.

    Fig. 4(a-b) shows a Type A defect having a compact single-layered oxide film with about 0.4 µm thickness. Oxygen, fluorine, magnesium and aluminium were detected in this film (Fig. 4c). It is speculated that oxide film is the mixture of fluoride and oxide of magnesium and aluminium. The detection of fluorine revealed that an entrained cover gas was contained in the formation of this defect. That is to say that the pores shown in Fig. 4(a) were not shrinkage defects or hydrogen porosity, but entrainment defects. The detection of aluminium was different with Xiong and Wang’s previous study [47,48], which showed that no aluminium was contained in their surface film of an AZ91 melt protected by a SF6 cover gas. Sulphur could not be clearly recognized in the element map, but there was a S-peak in the corresponding ESD spectrum.

    Fig. 4. (a) A Type A entrainment defect formed in SF6/air and having a single-layered oxide film, (b) the oxide film of this defect, (c) SEM-EDS element maps (using Philips JEOL7000) corresponding to the area highlighted in (b).

    Fig. 5(a-b) shows a Type B entrainment defect having a multi-layered oxide film. The compact outer layers of the oxide films were enriched with fluorine and oxygen (Fig. 5c), while their relatively porous inner layers were only enriched with oxygen (i.e., poor in fluorine) and partly grew together, thus forming a sandwich-like structure. Therefore, it is speculated that the outer layer is the mixture of fluoride and oxide, while the inner layer is mainly oxide. Sulphur could only be recognized in the EDX spectrum and could not be clearly identified in the element map, which might be due to the small S-content in the cover gas (i.e., 0.5% volume content of SF6 in the cover gas). In this oxide film, aluminium was contained in the outer layer of this oxide film but could not be clearly detected in the inner layer. Moreover, the distribution of Al seems to be uneven. It can be found that, in the right side of the defect, aluminium exists in the film but its concentration can not be identified to be higher than the matrix. However, there is a small area with much higher aluminium concentration in the left side of the defect. Such an uneven distribution of aluminium was also observed in other defects (shown in the following), and it is the result of the formation of some oxide particles in or under the film.

    Fig. 5. (a) A Type B entrainment defect formed in SF6/air and having a multi-layered oxide film, (b) the oxide films of this defect have grown together, (c) SEM-EDS element maps (using Philips JEOL7000) corresponding to the area shown in (b).

    Figs. 4 and 5 show cross sectional observations of the entrainment defects formed in the AZ91 alloy sample cast under a cover gas of SF6/air. It is not sufficient to characterize the entrainment defects only by the figures observed from the two-dimensional section. To have a further understanding, the surface of the entrainment defects (i.e. the oxide film) was further studied by observing the fracture surface of the test bars.

    Fig. 6(a) shows fracture surfaces of an AZ91 alloy tensile test bar produced in SF6/air. Symmetrical dark regions can be seen on both sides of the fracture surfaces. Fig. 6(b) shows boundaries between the dark and bright regions. The bright region consisted of jagged and broken features, while the surface of the dark region was relatively smooth and flat. In addition, the EDS results (Fig. 6c-d and Table 2) show that fluorine, oxygen, sulphur, and nitrogen were only detected in the dark regions, indicating that the dark regions were surface protective films entrained into the melt. Therefore, it could be suggested that the dark regions were an entrainment defect with consideration of their symmetrical nature. Similar defects on fracture surfaces of Al-alloy castings have been previously reported [7]Nitrides were only found in the oxide films on the test-bar fracture surfaces but never detected in the cross-sectional samples shown in Figs. 4 and 5. An underlying reason is that the nitrides contained in these samples may have hydrolysed during the sample polishing process [54].

    Fig. 6. (a) A pair of the fracture surfaces of a AZ91 alloy tensile test bar produced under a cover gas of SF6/air. The dimension of the fracture surface is 5 mm × 6 mm, (b) a section of the boundary between the dark and bright regions shown in (a), (c-d) EDS spectrum of the (c) bright regions and (d) dark regions, (e) schematic of an entrainment defect contained in a test bar.

    Table 2. EDS results (wt.%) corresponding to the regions shown in Fig. 6 (cover gas: SF6/air).

    Empty CellCOMgFAlZnSN
    Dark region in Fig. 6(b)3.481.3279.130.4713.630.570.080.73
    Bright region in Fig. 6(b)3.5884.4811.250.68

    In conjunction with the cross-sectional observation of the defects shown in Figs. 4 and 5, the structure of an entrainment defect contained in a tensile test bar was sketched as shown in Fig. 6(e). The defect contained an entrained gas enclosed by its oxide film, creating a void section inside the test bar. When the tensile force applied on the defect during the fracture process, the crack was initiated at the void section and propagated along the entrainment defect, since cracks would be propagated along the weakest path [55]. Therefore, when the test bar was finally fractured, the oxide films of entrainment defect appeared on both fracture surfaces of the test bar, as shown in Fig. 6(a).

    3.2. Structure and composition of the entrainment defects formed in SF6/CO2

    Similar to the entrainment defect formed in SF6/air, the defects formed under a cover gas of 0.5%SF6/CO2 also had two types of oxide films (i.e., single-layered and multi-layered types). Fig. 7(a) shows an example of the entrainment defects containing a multi-layered oxide film. A magnified observation to the defect (Fig. 7b) shows that the inner layers of the oxide films had grown together, presenting a sandwich-like structure, which was similar to the defects formed in an atmosphere of SF6/air (Fig. 5b). An EDS spectrum (Fig. 7c) revealed that the joint area (inner layer) of this sandwich-like structure mainly contained magnesium oxides. Peaks of fluorine, sulphur, and aluminium were recognized in this EDS spectrum, but their amount was relatively small. In contrast, the outer layers of the oxide films were compact and composed of a mixture of fluorides and oxides (Fig. 7d-e).

    Fig. 7. (a) An example of entrainment defects formed in SF6/CO2 and having a multi-layered oxide film, (b) magnified observation of the defect, showing the inner layer of the oxide films has grown together, (c) EDS spectrum of the point denoted in (b), (d) outer layer of the oxide film, (e) SEM-EDS element maps (using Philips JEOL7000) corresponding to the area shown in (d).

    Fig. 8(a) shows an entrainment defect on the fracture surfaces of an AZ91 alloy tensile test bar, which was produced in an atmosphere of 0.5%SF6/CO2. The corresponding EDS results (Table 3) showed that oxide film contained fluorides and oxides. Sulphur and nitrogen were not detected. Besides, a magnified observation (Fig. 8b) indicated spots on the oxide film surface. The diameter of the spots ranged from hundreds of nanometres to a few micron meters.

    Fig. 8. (a) A pair of the fracture surfaces of a AZ91 alloy tensile test bar, produced in an atmosphere of SF6/CO2. The dimension of the fracture surface is 5 mm × 6 mm, (b) surface appearance of the oxide films on the fracture surfaces, showing spots on the film surface.

    To further reveal the structure and composition of the oxide film clearly, the cross-section of the oxide film on a test-bar fracture surface was onsite exposed using the FIB technique (Fig. 9). As shown in Fig. 9a, a continuous oxide film was found between the platinum coating layer and the Mg-Al alloy substrate. Fig. 9 (b-c) shows a magnified observation to oxide films, indicating a multi-layered structure (denoted by the red box in Fig. 9c). The bottom layer was enriched with fluorine and oxygen and should be the mixture of fluoride and oxide, which was similar to the “outer layer” shown in Figs. 5 and 7, while the only-oxygen-enriched top layer was similar to the “inner layer” shown in Figs. 5 and 7.

    Fig. 9. (a) A cross-sectional observation of the oxide film on the fracture surface of the AZ91 casting produced in SF6/CO2, exposed by FIB, (b) a magnified observation of area highlighted in (a), and (c) SEM-EDS elements map of the area shown in (b), obtained by CFEI Quanta 3D FEG FIB-SEM.

    Except the continuous film, some individual particles were also observed in or below the continuous film, as shown in Fig. 9. An Al-enriched particle was detected in the left side of the oxide film shown in Fig. 9b and might be speculated to be spinel Mg2AlO4 because it also contains abundant magnesium and oxygen elements. The existing of such Mg2AlO4 particles is responsible for the high concentration of aluminium in small areas of the observed film and the uneven distribution of aluminium, as shown in Fig. 5(c). Here it should be emphasized that, although the other part of the bottom layer of the continuous oxide film contains less aluminium than this Al-enriched particle, the Fig. 9c indicated that the amount of aluminium in this bottom layer was still non-negligible, especially when comparing with the outer layer of the film. Below the right side of the oxide film shown in Fig. 9b, a particle was detected and speculated to be MgO because it is rich in Mg and O. According to Wang’s result [56], lots of discrete MgO particles can be formed on the surface of the Mg melt by the oxidation of Mg melt and Mg vapor. The MgO particles observed in our present work may be formed due to the same reasons. While, due to the differences in experimental conditions, less Mg melt can be vapored or react with O2, thus only a few of MgO particles formed in our work. An enrichment of carbon was also found in the film, revealing that CO2 was able to react with the melt, thus forming carbon or carbides. This carbon concentration was consistent with the relatively high carbon content of the oxide film shown in Table 3 (i.e., the dark region). In the area next to the oxide film.

    Table 3. EDS results (wt.%) corresponding to the regions shown in Fig. 8 (cover gas: SF6/ CO2).

    Empty CellCOMgFAlZnSN
    Dark region in Fig. 8(a)7.253.6469.823.827.030.86
    Bright region in Fig. 8(a)2.100.4482.8313.261.36

    This cross-sectional observation of the oxide film on a test bar fracture surface (Fig. 9) further verified the schematic of the entrainment defect shown in Fig. 6(e). The entrainment defects formed in different atmospheres of SF6/CO2 and SF6/air had similar structures, but their compositions were different.

    3.3. Evolution of the oxide films in the oxidation cell

    The results in Section 3.1 and 3.2 have shown the structures and compositions of entrainment defects formed in AZ91 castings under cover gases of SF6/air and SF6/CO2. Different stages of the oxidation reaction may lead to the different structures and compositions of entrainment defects. Although Campbell has conjectured that an entrained gas may react with the surrounding melt, it is rarely reported that the reaction occurring between the Mg-alloy melt and entrapped cover gas. Previous researchers normally focus on the reaction between a Mg-alloy melt and the cover gas in an open environment [38,39,[46][47][48][49][50][51][52], which was different from the situation of a cover gas trapped into the melt. To further understand the formation of the entrainment defect in an AZ91 alloy, the evolution process of oxide films of the entrainment defect was further studied using an oxidation cell.

    Fig. 10 (a and d) shows a surface film held for 5 min in the oxidation cell, protected by 0.5%SF6/air. There was only one single layer consisting of fluoride and oxide (MgF2 and MgO). In this surface film. Sulphur was detected in the EDS spectrum, but its amount was too small to be recognized in the element map. The structure and composition of this oxide film was similar to the single-layered films of entrainment defects shown in Fig. 4.

    Fig. 10. Oxide films formed in the oxidation cell under a cover gas of 0.5%SF6/air and held at 700 °C for (a) 5 min; (b) 10 min; (c) 30 min, and (d-f) the SEM-EDS element maps (using Philips JEOL7000) corresponding to the oxide film shown in (a-c) respectively, (d) 5 min; (e) 10 min; (f) 30 min. The red points in (c and f) are the location references, denoting the boundary of the F-enriched layer in different element maps.

    After a holding time of 10 min, a thin (O, S)-enriched top layer (around 700 nm) appeared upon the preliminary F-enriched film, forming a multi-layered structure, as shown in Fig. 10(b and e). The thickness of the (O, S)-enriched top layer increased with increased holding time. As shown in Fig. 10(c and f), the oxide film held for 30 min also had a multi-layered structure, but the thickness of its (O, S)-enriched top layer (around 2.5 µm) was higher than the that of the 10-min oxide film. The multi-layered oxide films shown in Fig. 10(b-c) presented a similar appearance to the films of the sandwich-like defect shown in Fig. 5.

    The different structures of the oxide films shown in Fig. 10 indicated that fluorides in the cover gas would be preferentially consumed due to the reaction with the AZ91 alloy melt. After the depletion of fluorides, the residual cover gas reacted further with the liquid AZ91 alloy, forming the top (O, S)-enriched layer in the oxide film. Therefore, the different structures and compositions of entrainment defects shown in Figs. 4 and 5 may be due to an ongoing oxidation reaction between melt and entrapped cover gas.

    This multi-layered structure has not been reported in previous publications concerning the protective surface film formed on a Mg-alloy melt [38,[46][47][48][49][50][51]. This may be due to the fact that previous researchers carried out their experiments with an un-limited amount of cover gas, creating a situation where the fluorides in the cover gas were not able to become depleted. Therefore, the oxide film of an entrainment defect had behaviour traits similar to the oxide films shown in Fig. 10, but different from the oxide films formed on the Mg-alloy melt surface reported in [38,[46][47][48][49][50][51].

    Similar with the oxide films held in SF6/air, the oxide films formed in SF6/CO2 also had different structures with different holding times in the oxidation cell. Fig. 11(a) shows an oxide film, held on an AZ91 melt surface under a cover gas of 0.5%SF6/CO2 for 5 min. This film had a single-layered structure consisting of MgF2. The existence of MgO could not be confirmed in this film. After the holding time of 30 min, the film had a multi-layered structure; the inner layer was of a compact and uniform appearance and composed of MgF2, while the outer layer is the mixture of MgF2 and MgO. Sulphur was not detected in this film, which was different from the surface film formed in 0.5%SF6/air. Therefore, fluorides in the cover gas of 0.5%SF6/CO2 were also preferentially consumed at an early stage of the film growth process. Compared with the film formed in SF6/air, the MgO in film formed in SF6/CO2 appeared later and sulphide did not appear within 30 min. It may mean that the formation and evolution of film in SF6/air is faster than SF6/CO2. CO2 may have subsequently reacted with the melt to form MgO, while sulphur-containing compounds accumulated in the cover gas and reacted to form sulphide in very late stage (may after 30 min in oxidation cell).

    Fig. 11. Oxide films formed in the oxidation cell under a cover gas of 0.5%SF6/CO2, and their SEM-EDS element maps (using Philips JEOL7000). They were held at 700 °C for (a) 5 min; (b) 30 min. The red points in (b) are the location references, denoting the boundary between the top and bottom layers in the oxide film.

    4. Discussion

    4.1. Evolution of entrainment defects formed in SF6/air

    HSC software from Outokumpu HSC Chemistry for Windows (http://www.hsc-chemistry.net/) was used to carry out thermodynamic calculations needed to explore the reactions which might occur between the trapped gases and liquid AZ91 alloy. The solutions to the calculations suggest which products are most likely to form in the reaction process between a small amount of cover gas (i.e., the amount within a trapped bubble) and the AZ91-alloy melt.

    In the trials, the pressure was set to 1 atm, and the temperature set to 700 °C. The amount of the cover gas was assumed to be 7 × 10−7 kg, with a volume of approximately 0.57 cm3 (3.14 × 10−8 kmol) for 0.5%SF6/air, and 0.35 cm3 (3.12 × 10−8 kmol) for 0.5%SF6/CO2. The amount of the AZ91 alloy melt in contact with the trapped gas was assumed to be sufficient to complete all reactions. The decomposition products of SF6 were SF5, SF4, SF3, SF2, F2, S(g), S2(g) and F(g) [57][58][59][60].

    Fig. 12 shows the equilibrium diagram of the thermodynamic calculation of the reaction between the AZ91 alloy and 0.5%SF6/air. In the diagram, the reactants and products with less than 10−15 kmol have not been shown, as this was 5 orders of magnitude less than the amount of SF6 present (≈ 1.57 × 10−10 kmol) and therefore would not affect the observed process in a practical way.

    Fig. 12. An equilibrium diagram for the reaction between 7e-7 kg 0.5%SF6/air and a sufficient amount of AZ91 alloy. The X axis is the amount of AZ91 alloy melt having reacted with the entrained gas, and the vertical Y-axis is the amount of the reactants and products.

    This reaction process could be divided into 3 stages.

    Stage 1: The formation of fluorides. the AZ91 melt preferentially reacted with SF6 and its decomposition products, producing MgF2, AlF3, and ZnF2. However, the amount of ZnF2 may have been too small to be detected practically (1.25 × 10−12 kmol of ZnF2 compared with 3 × 10−10 kmol of MgF2), which may be the reason why Zn was not detected in any the oxide films shown in Sections 3.13.3. Meanwhile, sulphur accumulated in the residual gas as SO2.

    Stage 2: The formation of oxides. After the liquid AZ91 alloy had depleted all the available fluorides in the entrapped gas, the amount of AlF3 and ZnF2 quickly reduced due to a reaction with Mg. O2(g) and SO2 reacted with the AZ91 melt, forming MgO, Al2O3, MgAl2O4, ZnO, ZnSO4 and MgSO4. However, the amount of ZnO and ZnSO4 would have been too small to be found practically by EDS (e.g. 9.5 × 10−12 kmol of ZnO,1.38 × 10−14 kmol of ZnSO4, in contrast to 4.68 × 10−10 kmol of MgF2, when the amount of AZ91 on the X-axis is 2.5 × 10−9 kmol). In the experimental cases, the concentration of F in the cover gas is very low, whole the concentration f O is much higher. Therefore, the stage 1 and 2, i.e, the formation of fluoride and oxide may happen simultaneously at the beginning of the reaction, resulting in the formation of a singer-layered mixture of fluoride and oxide, as shown in Figs. 4 and 10(a). While an inner layer consisted of oxides but fluorides could form after the complete depletion of F element in the cover gas.

    Stages 1- 2 theoretically verified the formation process of the multi-layered structure shown in Fig. 10.

    The amount of MgAl2O4 and Al2O3 in the oxide film was of a sufficient amount to be detected, which was consistent with the oxide films shown in Fig. 4. However, the existence of aluminium could not be recognized in the oxide films grown in the oxidation cell, as shown in Fig. 10. This absence of Al may be due to the following reactions between the surface film and AZ91 alloy melt:(1)

    Al2O3 + 3Mg + = 3MgO + 2Al, △G(700 °C) = -119.82 kJ/mol(2)

    Mg + MgAl2O4 = MgO + Al, △G(700 °C) =-106.34 kJ/molwhich could not be simulated by the HSC software since the thermodynamic calculation was carried out under an assumption that the reactants were in full contact with each other. However, in a practical process, the AZ91 melt and the cover gas would not be able to be in contact with each other completely, due to the existence of the protective surface film.

    Stage 3: The formation of Sulphide and nitride. After a holding time of 30 min, the gas-phase fluorides and oxides in the oxidation cell had become depleted, allowing the melt reaction with the residual gas, forming an additional sulphur-enriched layer upon the initial F-enriched or (F, O)-enriched surface film, thus resulting in the observed multi-layered structure shown in Fig. 10 (b and c). Besides, nitrogen reacted with the AZ91 melt until all reactions were completed. The oxide film shown in Fig. 6 may correspond to this reaction stage due to its nitride content. However, the results shows that the nitrides were not detected in the polished samples shown in Figs. 4 and 5, but only found on the test bar fracture surfaces. The nitrides may have hydrolysed during the sample preparation process, as follows [54]:(3)

    Mg3N2 + 6H2O =3Mg(OH)2 + 2NH3↑(4)

    AlN+ 3H2O =Al(OH)3 + NH3

    In addition, Schmidt et al. [61] found that Mg3N2 and AlN could react to form ternary nitrides (Mg3AlnNn+2, n= 1, 2, 3…). HSC software did not contain the database of ternary nitrides, and it could not be added into the calculation. The oxide films in this stage may also contain ternary nitrides.

    4.2. Evolution of entrainment defects formed in SF6/CO2

    Fig. 13 shows the results of the thermodynamic calculation between AZ91 alloy and 0.5%SF6/CO2. This reaction processes can also be divided into three stages.

    Fig. 13. An equilibrium diagram for the reaction between 7e-7 kg 0.5%SF6/CO2 and a sufficient amount of AZ91 alloy. The X axis denotes the amount of Mg alloy melt having reacted with the entrained gas, and the vertical Y-axis denotes the amounts of the reactants and products.

    Stage 1: The formation of fluorides. SF6 and its decomposition products were consumed by the AZ91 melt, forming MgF2, AlF3, and ZnF2. As in the reaction of AZ91 in 0.5%SF6/air, the amount of ZnF2 was too small to be detected practically (1.51 × 10−13 kmol of ZnF2 compared with 2.67 × 10−10 kmol of MgF2). Sulphur accumulated in the residual trapped gas as S2(g) and a portion of the S2(g) reacted with CO2, to form SO2 and CO. The products in this reaction stage were consistent with the film shown in Fig. 11(a), which had a single layer structure that contained fluorides only.

    Stage 2: The formation of oxides. AlF3 and ZnF2 reacted with the Mg in the AZ91 melt, forming MgF2, Al and Zn. The SO2 began to be consumed, producing oxides in the surface film and S2(g) in the cover gas. Meanwhile, the CO2 directly reacted with the AZ91 melt, forming CO, MgO, ZnO, and Al2O3. The oxide films shown in Figs. 9 and 11(b) may correspond to this reaction stage due to their oxygen-enriched layer and multi-layered structure.

    The CO in the cover gas could further react with the AZ91 melt, producing C. This carbon may further react with Mg to form Mg carbides, when the temperature reduced (during solidification period) [62]. This may be the reason for the high carbon content in the oxide film shown in Figs. 89. Liang et al. [39] also reported carbon-detection in an AZ91 alloy surface film protected by SO2/CO2. The produced Al2O3 may be further combined with MgO, forming MgAl2O4 [63]. As discussed in Section 4.1, the alumina and spinel can react with Mg, causing an absence of aluminium in the surface films, as shown in Fig. 11.

    Stage 3: The formation of Sulphide. the AZ91 melt began to consume S2(g) in the residual entrapped gas, forming ZnS and MgS. These reactions did not occur until the last stage of the reaction process, which could be the reason why the S-content in the defect shown Fig. 7(c) was small.

    In summary, thermodynamic calculations indicate that the AZ91 melt will react with the cover gas to form fluorides firstly, then oxides and sulphides in the last. The oxide film in the different reaction stages would have different structures and compositions.

    4.3. Effect of the carrier gases on consumption of the entrained gas and the reproducibility of AZ91 castings

    The evolution processes of entrainment defects, formed in SF6/air and SF6/CO2, have been suggested in Sections 4.1 and 4.2. The theoretical calculations were verified with respect to the corresponding oxide films found in practical samples. The atmosphere within an entrainment defect could be efficiently consumed due to the reaction with liquid Mg-alloy, in a scenario dissimilar to the Al-alloy system (i.e., nitrogen in an entrained air bubble would not efficiently react with Al-alloy melt [64,65], however, nitrogen would be more readily consumed in liquid Mg alloys, commonly referred to as “nitrogen burning” [66]).

    The reaction between the entrained gas and the surrounding liquid Mg-alloy converted the entrained gas into solid compounds (e.g. MgO) within the oxide film, thus reducing the void volume of the entrainment defect and hence probably causing a collapse of the defect (e.g., if an entrained gas of air was depleted by the surrounding liquid Mg-alloy, under an assumption that the melt temperature is 700 °C and the depth of liquid Mg-alloy is 10 cm, the total volume of the final solid products would be 0.044% of the initial volume taken by the entrapped air).

    The relationship between the void volume reduction of entrainment defects and the corresponding casting properties has been widely studied in Al-alloy castings. Nyahumwa and Campbell [16] reported that the Hot Isostatic Pressing (HIP) process caused the entrainment defects in Al-alloy castings to collapse and their oxide surfaces forced into contact. The fatigue lives of their castings were improved after HIP. Nyahumwa and Campbell [16] also suggested a potential bonding of the double oxide films that were in contact with each other, but there was no direct evidence to support this. This binding phenomenon was further investigated by Aryafar et.al.[8], who re-melted two Al-alloy bars with oxide skins in a steel tube and then carried out a tensile strength test on the solidified sample. They found that the oxide skins of the Al-alloy bars strongly bonded with each other and became even stronger with an extension of the melt holding time, indicating a potential “healing” phenomenon due to the consumption of the entrained gas within the double oxide film structure. In addition, Raidszadeh and Griffiths [9,19] successfully reduced the negative effect of entrainment defects on the reproducibility of Al-alloy castings, by extending the melt holding time before solidification, which allowed the entrained gas to have a longer time to react with the surrounding melt.

    With consideration of the previous work mentioned, the consumption of the entrained gas in Mg-alloy castings may diminish the negative effect of entrainment defects in the following two ways.

    (1) Bonding phenomenon of the double oxide films. The sandwich-like structure shown in Fig. 5 and 7 indicated a potential bonding of the double oxide film structure. However, more evidence is required to quantify the increase in strength due to the bonding of the oxide films.

    (2) Void volume reduction of entrainment defects. The positive effect of void-volume reduction on the quality of castings has been widely demonstrated by the HIP process [67]. As the evolution processes discussed in Section 4.14.2, the oxide films of entrainment defects can grow together due to an ongoing reaction between the entrained gas and surrounding AZ91 alloy melt. The volume of the final solid products was significant small compared with the entrained gas (i.e., 0.044% as previously mentioned).

    Therefore, the consumption rate of the entrained gas (i.e., the growth rate of oxide films) may be a critical parameter for improving the quality of AZ91 alloy castings. The oxide film growth rate in the oxidization cell was accordingly further investigated.

    Fig. 14 shows a comparison of the surface film growth rates in different cover gases (i.e., 0.5%SF6/air and 0.5%SF6/CO2). 15 random points on each sample were selected for film thickness measurements. The 95% confidence interval (95%CI) was computed under an assumption that the variation of the film thickness followed a Gaussian distribution. It can be seen that all the surface films formed in 0.5%SF6/air grew faster than those formed in 0.5%SF6/CO2. The different growth rates suggested that the entrained-gas consumption rate of 0.5%SF6/air was higher than that of 0.5%SF6/CO2, which was more beneficial for the consumption of the entrained gas.

    Fig. 14. A comparison of the AZ91 alloy oxide film growth rates in 0.5%SF6/air and 0.5%SF6/CO2

    It should be noted that, in the oxidation cell, the contact area of liquid AZ91 alloy and cover gas (i.e. the size of the crucible) was relatively small with consideration of the large volume of melt and gas. Consequently, the holding time for the oxide film growth within the oxidation cell was comparatively long (i.e., 5–30 min). However, the entrainment defects contained in a real casting are comparatively very small (i.e., a few microns size as shown in Figs. 36, and [7]), and the entrained gas is fully enclosed by the surrounding melt, creating a relatively large contact area. Hence the reaction time for cover gas and the AZ91 alloy melt may be comparatively short. In addition, the solidification time of real Mg-alloy sand castings can be a few minutes (e.g. Guo [68] reported that a Mg-alloy sand casting with 60 mm diameter required 4 min to be solidified). Therefore, it can be expected that an entrained gas trapped during an Mg-alloy melt pouring process will be readily consumed by the surrounding melt, especially for sand castings and large-size castings, where solidification times are long.

    Therefore, the different cover gases (0.5%SF6/air and 0.5%SF6/CO2) associated with different consumption rates of the entrained gases may affect the reproducibility of the final castings. To verify this assumption, the AZ91 castings produced in 0.5%SF6/air and 0.5%SF6/CO2 were machined into test bars for mechanical evaluation. A Weibull analysis was carried out using both linear least square (LLS) method and non-linear least square (non-LLS) method [69].

    Fig. 15(a-b) shows a traditional 2-p linearized Weibull plot of the UTS and elongation of the AZ91 alloy castings, obtained by the LLS method. The estimator used is P= (i-0.5)/N, which was suggested to cause the lowest bias among all the popular estimators [69,70]. The casting produced in SF6/air has an UTS Weibull moduli of 16.9, and an elongation Weibull moduli of 5.0. In contrast, the UTS and elongation Weibull modulus of the casting produced in SF6/CO2 are 7.7 and 2.7 respectively, suggesting that the reproducibility of the casting protected by SF6/CO2 were much lower than that produced in SF6/air.

    Fig. 15. The Weibull modulus of AZ91 castings produced in different atmospheres, estimated by (a-b) the linear least square method, (c-d) the non-linear least square method, where SSR is the sum of residual squares.

    In addition, the author’s previous publication [69] demonstrated a shortcoming of the linearized Weibull plots, which may cause a higher bias and incorrect R2 interruption of the Weibull estimation. A Non-LLS Weibull estimation was therefore carried out, as shown in Fig. 15 (c-d). The UTS Weibull modulus of the SF6/air casting was 20.8, while the casting produced under SF6/CO2 had a lower UTS Weibull modulus of 11.4, showing a clear difference in their reproducibility. In addition, the SF6/air elongation (El%) dataset also had a Weibull modulus (shape = 5.8) higher than the elongation dataset of SF6/CO2 (shape = 3.1). Therefore, both the LLS and Non-LLS estimations suggested that the SF6/air casting has a higher reproducibility than the SF6/CO2 casting. It supports the method that the use of air instead of CO2 contributes to a quicker consumption of the entrained gas, which may reduce the void volume within the defects. Therefore, the use of 0.5%SF6/air instead of 0.5%SF6/CO2 (which increased the consumption rate of the entrained gas) improved the reproducibility of the AZ91 castings.

    However, it should be noted that not all the Mg-alloy foundries followed the casting process used in present work. The Mg-alloy melt in present work was degassed, thus reducing the effect of hydrogen on the consumption of the entrained gas (i.e., hydrogen could diffuse into the entrained gas, potentially suppressing the depletion of the entrained gas [7,71,72]). In contrast, in Mg-alloy foundries, the Mg-alloy melt is not normally degassed, since it was widely believed that there is not a ‘gas problem’ when casting magnesium and hence no significant change in tensile properties [73]. Although studies have shown the negative effect of hydrogen on the mechanical properties of Mg-alloy castings [41,42,73], a degassing process is still not very popular in Mg-alloy foundries.

    Moreover, in present work, the sand mould cavity was flushed with the SF6 cover gas prior to pouring [22]. However, not all the Mg-alloy foundries flushed the mould cavity in this way. For example, the Stone Foundry Ltd (UK) used sulphur powder instead of the cover-gas flushing. The entrained gas within their castings may be SO2/air, rather than the protective gas.

    Therefore, although the results in present work have shown that using air instead of CO2 improved the reproducibility of the final casting, it still requires further investigations to confirm the effect of carrier gases with respect to different industrial Mg-alloy casting processes.

    7. Conclusion

    Entrainment defects formed in an AZ91 alloy were observed. Their oxide films had two types of structure: single-layered and multi-layered. The multi-layered oxide film can grow together forming a sandwich-like structure in the final casting.2.

    Both the experimental results and the theoretical thermodynamic calculations demonstrated that fluorides in the trapped gas were depleted prior to the consumption of sulphur. A three-stage evolution process of the double oxide film defects has been suggested. The oxide films contained different combinations of compounds, depending on the evolution stage. The defects formed in SF6/air had a similar structure to those formed in SF6/CO2, but the compositions of their oxide films were different. The oxide-film formation and evolution process of the entrainment defects were different from that of the Mg-alloy surface films previous reported (i.e., MgO formed prior to MgF2).3.

    The growth rate of the oxide film was demonstrated to be greater under SF6/air than SF6/CO2, contributing to a quicker consumption of the damaging entrapped gas. The reproducibility of an AZ91 alloy casting improved when using SF6/air instead of SF6/CO2.

    Acknowledgements

    The authors acknowledge funding from the EPSRC LiME grant EP/H026177/1, and the help from Dr W.D. Griffiths and Mr. Adrian Carden (University of Birmingham). The casting work was carried out in University of Birmingham.

    Reference

    [1]

    M.K. McNutt, SALAZAR K.

    Magnesium, Compounds & Metal, U.S. Geological Survey and U.S. Department of the Interior

    Reston, Virginia (2013)

    Google Scholar[2]

    Magnesium

    Compounds & Metal, U.S. Geological Survey and U.S. Department of the Interior

    (1996)

    Google Scholar[3]

    I. Ostrovsky, Y. Henn

    ASTEC’07 International Conference-New Challenges in Aeronautics, Moscow (2007), pp. 1-5

    Aug 19-22

    View Record in ScopusGoogle Scholar[4]

    Y. Wan, B. Tang, Y. Gao, L. Tang, G. Sha, B. Zhang, N. Liang, C. Liu, S. Jiang, Z. Chen, X. Guo, Y. Zhao

    Acta Mater., 200 (2020), pp. 274-286

    ArticleDownload PDFView Record in Scopus[5]

    J.T.J. Burd, E.A. Moore, H. Ezzat, R. Kirchain, R. Roth

    Appl. Energy, 283 (2021), Article 116269

    ArticleDownload PDFView Record in Scopus[6]

    A.M. Lewis, J.C. Kelly, G.A. Keoleian

    Appl. Energy, 126 (2014), pp. 13-20

    ArticleDownload PDFView Record in Scopus[7]

    J. Campbell

    Castings

    Butterworth-Heinemann, Oxford (2004)

    Google Scholar[8]

    M. Aryafar, R. Raiszadeh, A. Shalbafzadeh

    J. Mater. Sci., 45 (2010), pp. 3041-3051 View PDF

    CrossRefView Record in Scopus[9]

    R. Raiszadeh, W.D. Griffiths

    Metall. Mater. Trans. B-Process Metall. Mater. Process. Sci., 42 (2011), pp. 133-143 View PDF

    CrossRefView Record in Scopus[10]

    R. Raiszadeh, W.D. Griffiths

    J. Alloy. Compd., 491 (2010), pp. 575-580

    ArticleDownload PDFView Record in Scopus[11]

    L. Peng, G. Zeng, T.C. Su, H. Yasuda, K. Nogita, C.M. Gourlay

    JOM, 71 (2019), pp. 2235-2244 View PDF

    CrossRefView Record in Scopus[12]

    S. Ganguly, A.K. Mondal, S. Sarkar, A. Basu, S. Kumar, C. Blawert

    Corros. Sci., 166 (2020)[13]

    G.E. Bozchaloei, N. Varahram, P. Davami, S.K. Kim

    Mater. Sci. Eng. A-Struct. Mater. Prop. Microstruct. Process., 548 (2012), pp. 99-105

    View Record in Scopus[14]

    S. Fox, J. Campbell

    Scr. Mater., 43 (2000), pp. 881-886

    ArticleDownload PDFView Record in Scopus[15]

    M. Cox, R.A. Harding, J. Campbell

    Mater. Sci. Technol., 19 (2003), pp. 613-625

    View Record in Scopus[16]

    C. Nyahumwa, N.R. Green, J. Campbell

    Metall. Mater. Trans. A-Phys. Metall. Mater. Sci., 32 (2001), pp. 349-358

    View Record in Scopus[17]

    A. Ardekhani, R. Raiszadeh

    J. Mater. Eng. Perform., 21 (2012), pp. 1352-1362 View PDF

    CrossRefView Record in Scopus[18]

    X. Dai, X. Yang, J. Campbell, J. Wood

    Mater. Sci. Technol., 20 (2004), pp. 505-513

    View Record in Scopus[19]

    E.M. Elgallad, M.F. Ibrahim, H.W. Doty, F.H. Samuel

    Philos. Mag., 98 (2018), pp. 1337-1359 View PDF

    CrossRefView Record in Scopus[20]

    W.D. Griffiths, N.W. Lai

    Metall. Mater. Trans. A-Phys. Metall. Mater. Sci., 38A (2007), pp. 190-196 View PDF

    CrossRefView Record in Scopus[21]

    A.R. Mirak, M. Divandari, S.M.A. Boutorabi, J. Campbell

    Int. J. Cast Met. Res., 20 (2007), pp. 215-220 View PDF

    CrossRefView Record in Scopus[22]

    C. Cingi

    Laboratory of Foundry Engineering

    Helsinki University of Technology, Espoo, Finland (2006)

    Google Scholar[23]

    Y. Jia, J. Hou, H. Wang, Q. Le, Q. Lan, X. Chen, L. Bao

    J. Mater. Process. Technol., 278 (2020), Article 116542

    ArticleDownload PDFView Record in Scopus[24]

    S. Ouyang, G. Yang, H. Qin, S. Luo, L. Xiao, W. Jie

    Mater. Sci. Eng. A, 780 (2020), Article 139138

    ArticleDownload PDFView Record in Scopus[25]

    S.-m. Xiong, X.-F. Wang

    Trans. Nonferrous Met. Soc. China, 20 (2010), pp. 1228-1234

    ArticleDownload PDFView Record in Scopus[26]

    G.V. Research

    Grand View Research

    (2018)

    USA

    Google Scholar[27]

    T. Li, J. Davies

    Metall. Mater. Trans. A, 51 (2020), pp. 5389-5400 View PDF

    CrossRefView Record in Scopus[28]J.F. Fruehling, The University of Michigan, 1970.

    Google Scholar[29]

    S. Couling

    36th Annual World Conference on Magnesium, Norway (1979), pp. 54-57

    View Record in ScopusGoogle Scholar[30]

    S. Cashion, N. Ricketts, P. Hayes

    J. Light Met., 2 (2002), pp. 43-47

    ArticleDownload PDFView Record in Scopus[31]

    S. Cashion, N. Ricketts, P. Hayes

    J. Light Met., 2 (2002), pp. 37-42

    ArticleDownload PDFView Record in Scopus[32]

    K. Aarstad, G. Tranell, G. Pettersen, T.A. Engh

    Various Techniques to Study the Surface of Magnesium Protected by SF6

    TMS (2003)

    Google Scholar[33]

    S.-M. Xiong, X.-L. Liu

    Metall. Mater. Trans. A, 38 (2007), pp. 428-434 View PDF

    CrossRefView Record in Scopus[34]

    T.-S. Shih, J.-B. Liu, P.-S. Wei

    Mater. Chem. Phys., 104 (2007), pp. 497-504

    ArticleDownload PDFView Record in Scopus[35]

    G. Pettersen, E. Øvrelid, G. Tranell, J. Fenstad, H. Gjestland

    Mater. Sci. Eng. A, 332 (2002), pp. 285-294

    ArticleDownload PDFView Record in Scopus[36]

    H. Bo, L.B. Liu, Z.P. Jin

    J. Alloy. Compd., 490 (2010), pp. 318-325

    ArticleDownload PDFView Record in Scopus[37]

    A. Mirak, C. Davidson, J. Taylor

    Corros. Sci., 52 (2010), pp. 1992-2000

    ArticleDownload PDFView Record in Scopus[38]

    B.D. Lee, U.H. Beak, K.W. Lee, G.S. Han, J.W. Han

    Mater. Trans., 54 (2013), pp. 66-73 View PDF

    View Record in Scopus[39]

    W.Z. Liang, Q. Gao, F. Chen, H.H. Liu, Z.H. Zhao

    China Foundry, 9 (2012), pp. 226-230 View PDF

    CrossRef[40]

    U.I. Gol’dshleger, E.Y. Shafirovich

    Combust. Explos. Shock Waves, 35 (1999), pp. 637-644[41]

    A. Elsayed, S.L. Sin, E. Vandersluis, J. Hill, S. Ahmad, C. Ravindran, S. Amer Foundry

    Trans. Am. Foundry Soc., 120 (2012), pp. 423-429[42]

    E. Zhang, G.J. Wang, Z.C. Hu

    Mater. Sci. Technol., 26 (2010), pp. 1253-1258

    View Record in Scopus[43]

    N.R. Green, J. Campbell

    Mater. Sci. Eng. A-Struct. Mater. Prop. Microstruct. Process., 173 (1993), pp. 261-266

    ArticleDownload PDFView Record in Scopus[44]

    C Reilly, MR Jolly, NR Green

    Proceedings of MCWASP XII – 12th Modelling of Casting, Welding and Advanced Solidifcation Processes, Vancouver, Canada (2009)

    Google Scholar[45]H.E. Friedrich, B.L. Mordike, Springer, Germany, 2006.

    Google Scholar[46]

    C. Zheng, B.R. Qin, X.B. Lou

    Proceedings of the 2010 International Conference on Mechanical, Industrial, and Manufacturing Technologies, ASME (2010), pp. 383-388

    Mimt 2010 View PDF

    CrossRefView Record in ScopusGoogle Scholar[47]

    S.M. Xiong, X.F. Wang

    Trans. Nonferrous Met. Soc. China, 20 (2010), pp. 1228-1234

    ArticleDownload PDFView Record in Scopus[48]

    S.M. Xiong, X.L. Liu

    Metall. Mater. Trans. A-Phys. Metall. Mater. Sci., 38A (2007), pp. 428-434 View PDF

    CrossRefView Record in Scopus[49]

    T.S. Shih, J.B. Liu, P.S. Wei

    Mater. Chem. Phys., 104 (2007), pp. 497-504

    ArticleDownload PDFView Record in Scopus[50]

    K. Aarstad, G. Tranell, G. Pettersen, T.A. Engh

    Magn. Technol. (2003), pp. 5-10[51]

    G. Pettersen, E. Ovrelid, G. Tranell, J. Fenstad, H. Gjestland

    Mater. Sci. Eng. A-Struct. Mater. Prop. Microstruct. Process., 332 (2002), pp. 285-294

    ArticleDownload PDFView Record in Scopus[52]

    X.F. Wang, S.M. Xiong

    Corros. Sci., 66 (2013), pp. 300-307

    ArticleDownload PDFView Record in Scopus[53]

    S.H. Nie, S.M. Xiong, B.C. Liu

    Mater. Sci. Eng. A-Struct. Mater. Prop. Microstruct. Process., 422 (2006), pp. 346-351

    ArticleDownload PDFView Record in Scopus[54]

    C. Bauer, A. Mogessie, U. Galovsky

    Zeitschrift Fur Metallkunde, 97 (2006), pp. 164-168 View PDF

    CrossRef[55]

    Q.G. Wang, D. Apelian, D.A. Lados

    J. Light Met., 1 (2001), pp. 73-84

    ArticleDownload PDFView Record in Scopus[56]

    S. Wang, Y. Wang, Q. Ramasse, Z. Fan

    Metall. Mater. Trans. A, 51 (2020), pp. 2957-2974[57]

    S. Hayashi, W. Minami, T. Oguchi, H.J. Kim

    Kag. Kog. Ronbunshu, 35 (2009), pp. 411-415 View PDF

    CrossRefView Record in Scopus[58]

    K. Aarstad

    Norwegian University of Science and Technology

    (2004)

    Google Scholar[59]

    R.L. Wilkins

    J. Chem. Phys., 51 (1969), p. 853

    -&

    View Record in Scopus[60]

    O. Kubaschewski, K. Hesselemam

    Thermo-Chemical Properties of Inorganic Substances

    Springer-Verlag, Belin (1991)

    Google Scholar[61]

    R. Schmidt, M. Strobele, K. Eichele, H.J. Meyer

    Eur. J. Inorg. Chem. (2017), pp. 2727-2735 View PDF

    CrossRefView Record in Scopus[62]

    B. Hu, Y. Du, H. Xu, W. Sun, W.W. Zhang, D. Zhao

    J. Min. Metall. Sect. B-Metall., 46 (2010), pp. 97-103

    View Record in Scopus[63]

    O. Salas, H. Ni, V. Jayaram, K.C. Vlach, C.G. Levi, R. Mehrabian

    J. Mater. Res., 6 (1991), pp. 1964-1981

    View Record in Scopus[64]

    S.S.S. Kumari, U.T.S. Pillai, B.C. Pai

    J. Alloy. Compd., 509 (2011), pp. 2503-2509

    ArticleDownload PDFView Record in Scopus[65]

    H. Scholz, P. Greil

    J. Mater. Sci., 26 (1991), pp. 669-677

    View Record in Scopus[66]

    P. Biedenkopf, A. Karger, M. Laukotter, W. Schneider

    Magn. Technol., 2005 (2005), pp. 39-42

    View Record in Scopus[67]

    H.V. Atkinson, S. Davies

    Metall. Mater. Trans. A, 31 (2000), pp. 2981-3000 View PDF

    CrossRefView Record in Scopus[68]

    E.J. Guo, L. Wang, Y.C. Feng, L.P. Wang, Y.H. Chen

    J. Therm. Anal. Calorim., 135 (2019), pp. 2001-2008 View PDF

    CrossRefView Record in Scopus[69]

    T. Li, W.D. Griffiths, J. Chen

    Metall. Mater. Trans. A-Phys. Metall. Mater. Sci., 48A (2017), pp. 5516-5528 View PDF

    CrossRefView Record in Scopus[70]

    M. Tiryakioglu, D. Hudak

    J. Mater. Sci., 42 (2007), pp. 10173-10179 View PDF

    CrossRefView Record in Scopus[71]

    Y. Yue, W.D. Griffiths, J.L. Fife, N.R. Green

    Proceedings of the 1st International Conference on 3d Materials Science (2012), pp. 131-136 View PDF

    CrossRefView Record in ScopusGoogle Scholar[72]

    R. Raiszadeh, W.D. Griffiths

    Metall. Mater. Trans. B-Process Metall. Mater. Process. Sci., 37 (2006), pp. 865-871

    View Record in Scopus[73]

    Z.C. Hu, E.L. Zhang, S.Y. Zeng

    Mater. Sci. Technol., 24 (2008), pp. 1304-1308 View PDF

    CrossRefView Record in Scopus

    Fig. 1. Schematic figure showing the PREP with additional gas flowing on the end face of electrode.

    플라즈마 회전 전극 공정 중 분말 형성에 대한 공정 매개변수 및 냉각 가스의 영향

    Effects of process parameters and cooling gas on powder formation during the plasma rotating electrode process

    Yujie Cuia Yufan Zhaoa1 Haruko Numatab Kenta Yamanakaa Huakang Biana Kenta Aoyagia AkihikoChibaa
    aInstitute for Materials Research, Tohoku University, Sendai 980-8577, JapanbDepartment of Materials Processing, Graduate School of Engineering, Tohoku University, Sendai 980-8577, Japan

    Highlights

    •The limitation of increasing the rotational speed in decreasing powder size was clarified.

    •Cooling and disturbance effects varied with the gas flowing rate.

    •Inclined angle of the residual electrode end face affected powder formation.

    •Additional cooling gas flowing could be applied to control powder size.

    Abstract

    The plasma rotating electrode process (PREP) is rapidly becoming an important powder fabrication method in additive manufacturing. However, the low production rate of fine PREP powder limits the development of PREP. Herein, we investigated different factors affecting powder formation during PREP by combining experimental methods and numerical simulations. The limitation of increasing the rotation electrode speed in decreasing powder size is attributed to the increased probability of adjacent droplets recombining and the decreased tendency of granulation. The effects of additional Ar/He gas flowing on the rotational electrode on powder formation is determined through the cooling effect, the disturbance effect, and the inclined effect of the residual electrode end face simultaneously. A smaller-sized powder was obtained in the He atmosphere owing to the larger inclined angle of the residual electrode end face compared to the Ar atmosphere. Our research highlights the route for the fabrication of smaller-sized powders using PREP.

    플라즈마 회전 전극 공정(PREP)은 적층 제조 에서 중요한 분말 제조 방법으로 빠르게 자리잡고 있습니다. 그러나 미세한 PREP 분말의 낮은 생산율은 PREP의 개발을 제한합니다. 여기에서 우리는 실험 방법과 수치 시뮬레이션을 결합하여 PREP 동안 분말 형성에 영향을 미치는 다양한 요인을 조사했습니다. 분말 크기 감소에서 회전 전극 속도 증가의 한계는 인접한 액적 재결합 확률 증가 및 과립화 경향 감소에 기인합니다.. 회전 전극에 흐르는 추가 Ar/He 가스가 분말 형성에 미치는 영향은 냉각 효과, 외란 효과 및 잔류 전극 단면의 경사 효과를 통해 동시에 결정됩니다. He 분위기에서는 Ar 분위기에 비해 잔류 전극 단면의 경사각이 크기 때문에 더 작은 크기의 분말이 얻어졌다. 우리의 연구는 PREP를 사용하여 더 작은 크기의 분말을 제조하는 경로를 강조합니다.

    Keywords

    Plasma rotating electrode process

    Ti-6Al-4 V alloy, Rotating speed, Numerical simulation, Gas flowing, Powder size

    Introduction

    With the development of additive manufacturing, there has been a significant increase in high-quality powder production demand [1,2]. The initial powder characteristics are closely related to the uniform powder spreading [3,4], packing density [5], and layer thickness observed during additive manufacturing [6], thus determining the mechanical properties of the additive manufactured parts [7,8]. Gas atomization (GA) [9–11], centrifugal atomization (CA) [12–15], and the plasma rotating electrode process (PREP) are three important powder fabrication methods.

    Currently, GA is the dominant powder fabrication method used in additive manufacturing [16] for the fabrication of a wide range of alloys [11]. GA produces powders by impinging a liquid metal stream to droplets through a high-speed gas flow of nitrogen, argon, or helium. With relatively low energy consumption and a high fraction of fine powders, GA has become the most popular powder manufacturing technology for AM.

    The entrapped gas pores are generally formed in the powder after solidification during GA, in which the molten metal is impacted by a high-speed atomization gas jet. In addition, satellites are formed in GA powder when fine particles adhere to partially molten particles.

    The gas pores of GA powder result in porosity generation in the additive manufactured parts, which in turn deteriorates its mechanical properties because pores can become crack initiation sites [17]. In CA, a molten metal stream is poured directly onto an atomizer disc spinning at a high rotational speed. A thin film is formed on the surface of the disc, which breaks into small droplets due to the centrifugal force. Metal powder is obtained when these droplets solidify.

    Compared with GA powder, CA powder exhibits higher sphericity, lower impurity content, fewer satellites, and narrower particle size distribution [12]. However, very high speed is required to obtain fine powder by CA. In PREP, the molten metal, melted using the plasma arc, is ejected from the rotating rod through centrifugal force. Compared with GA powder, PREP-produced powders also have higher sphericity and fewer pores and satellites [18].

    For instance, PREP-fabricated Ti6Al-4 V alloy powder with a powder size below 150 μm exhibits lower porosity than gas-atomized powder [19], which decreases the porosity of additive manufactured parts. Furthermore, the process window during electron beam melting was broadened using PREP powder compared to GA powder in Inconel 718 alloy [20] owing to the higher sphericity of the PREP powder.

    In summary, PREP powder exhibits many advantages and is highly recommended for powder-based additive manufacturing and direct energy deposition-type additive manufacturing. However, the low production rate of fine PREP powder limits the widespread application of PREP powder in additive manufacturing.

    Although increasing the rotating speed is an effective method to decrease the powder size [21,22], the reduction in powder size becomes smaller with the increased rotating speed [23]. The occurrence of limiting effects has not been fully clarified yet.

    Moreover, the powder size can be decreased by increasing the rotating electrode diameter [24]. However, these methods are quite demanding for the PREP equipment. For instance, it is costly to revise the PREP equipment to meet the demand of further increasing the rotating speed or electrode diameter.

    Accordingly, more feasible methods should be developed to further decrease the PREP powder size. Another factor that influences powder formation is the melting rate [25]. It has been reported that increasing the melting rate decreases the powder size of Inconel 718 alloy [26].

    In contrast, the powder size of SUS316 alloy was decreased by decreasing the plasma current within certain ranges. This was ascribed to the formation of larger-sized droplets from fluid strips with increased thickness and spatial density at higher plasma currents [27]. The powder size of NiTi alloy also decreases at lower melting rates [28]. Consequently, altering the melting rate, varied with the plasma current, is expected to regulate the PREP powder size.

    Furthermore, gas flowing has a significant influence on powder formation [27,29–31]. On one hand, the disturbance effect of gas flowing promotes fluid granulation, which in turn contributes to the formation of smaller-sized powder [27]. On the other hand, the cooling effect of gas flowing facilitates the formation of large-sized powder due to increased viscosity and surface tension. However, there is a lack of systematic research on the effect of different gas flowing on powder formation during PREP.

    Herein, the authors systematically studied the effects of rotating speed, electrode diameter, plasma current, and gas flowing on the formation of Ti-6Al-4 V alloy powder during PREP as additive manufactured Ti-6Al-4 V alloy exhibits great application potential [32]. Numerical simulations were conducted to explain why increasing the rotating speed is not effective in decreasing powder size when the rotation speed reaches a certain level. In addition, the different factors incited by the Ar/He gas flowing on powder formation were clarified.

    Fig. 1. Schematic figure showing the PREP with additional gas flowing on the end face of electrode.
    Fig. 1. Schematic figure showing the PREP with additional gas flowing on the end face of electrode.

    References

    [1] W. Ding, G. Chen, M. Qin, Y. He, X. Qu, Low-cost Ti powders for additive manufacturing treated by fluidized bed, Powder Technol. 350 (2019) 117–122, https://doi.org/
    10.1016/j.powtec.2019.03.042.
    [2] W.S.W. Harun, M.S.I.N. Kamariah, N. Muhamad, S.A.C. Ghani, F. Ahmad, Z. Mohamed,
    A review of powder additive manufacturing processes for metallic biomaterials,
    Powder Technol. 327 (2018) 128–151, https://doi.org/10.1016/j.powtec.2017.12.
    058.
    [3] M. Ahmed, M. Pasha, W. Nan, M. Ghadiri, A simple method for assessing powder
    spreadability for additive manufacturing, Powder Technol. 367 (2020) 671–679,
    https://doi.org/10.1016/j.powtec.2020.04.033.
    [4] W. Nan, M. Pasha, M. Ghadiri, Numerical simulation of particle flow and segregation
    during roller spreading process in additive manufacturing, Powder Technol. 364
    (2020) 811–821, https://doi.org/10.1016/j.powtec.2019.12.023.
    [5] A. Averardi, C. Cola, S.E. Zeltmann, N. Gupta, Effect of particle size distribution on the
    packing of powder beds : a critical discussion relevant to additive manufacturing,
    Mater. Today Commun. 24 (2020) 100964, https://doi.org/10.1016/j.mtcomm.
    2020.100964.
    [6] K. Riener, N. Albrecht, S. Ziegelmeier, R. Ramakrishnan, L. Haferkamp, A.B. Spierings,
    G.J. Leichtfried, Influence of particle size distribution and morphology on the properties of the powder feedstock as well as of AlSi10Mg parts produced by laser powder bed fusion (LPBF), Addit. Manuf. 34 (2020) 101286, https://doi.org/10.1016/j.
    addma.2020.101286.
    [7] W.S.W. Harun, N.S. Manam, M.S.I.N. Kamariah, S. Sharif, A.H. Zulkifly, I. Ahmad, H.
    Miura, A review of powdered additive manufacturing techniques for Ti-6Al-4V biomedical applications, Powder Technol. 331 (2018) 74–97, https://doi.org/10.1016/j.
    powtec.2018.03.010.
    [8] A.T. Sutton, C.S. Kriewall, M.C. Leu, J.W. Newkirk, Powder characterisation techniques and effects of powder characteristics on part properties in powder-bed fusion processes, Virtual Phys. Prototyp. 12 (2017) (2017) 3–29, https://doi.org/10.
    1080/17452759.2016.1250605.
    [9] G. Chen, Q. Zhou, S.Y. Zhao, J.O. Yin, P. Tan, Z.F. Li, Y. Ge, J. Wang, H.P. Tang, A pore
    morphological study of gas-atomized Ti-6Al-4V powders by scanning electron microscopy and synchrotron X-ray computed tomography, Powder Technol. 330
    (2018) 425–430, https://doi.org/10.1016/j.powtec.2018.02.053.
    [10] Y. Feng, T. Qiu, Preparation, characterization and microwave absorbing properties of
    FeNi alloy prepared by gas atomization method, J. Alloys Compd. 513 (2012)
    455–459, https://doi.org/10.1016/j.jallcom.2011.10.079.

    [11] I.E. Anderson, R.L. Terpstra, Progress toward gas atomization processing with increased uniformity and control, Mater. Sci. Eng. A 326 (2002) 101–109, https://
    doi.org/10.1016/S0921-5093(01)01427-7.
    [12] P. Phairote, T. Plookphol, S. Wisutmethangoon, Design and development of a centrifugal atomizer for producing zinc metal powder, Int. J. Appl. Phys. Math. 2 (2012)
    77–82, https://doi.org/10.7763/IJAPM.2012.V2.58.
    [13] L. Tian, I. Anderson, T. Riedemann, A. Russell, Production of fine calcium powders by
    centrifugal atomization with rotating quench bath, Powder Technol. 308 (2017)
    84–93, https://doi.org/10.1016/j.powtec.2016.12.011.
    [14] M. Eslamian, J. Rak, N. Ashgriz, Preparation of aluminum/silicon carbide metal matrix composites using centrifugal atomization, Powder Technol. 184 (2008) 11–20,
    https://doi.org/10.1016/j.powtec.2007.07.045.
    [15] T. Plookphol, S. Wisutmethangoon, S. Gonsrang, Influence of process parameters on
    SAC305 lead-free solder powder produced by centrifugal atomization, Powder
    Technol. 214 (2011) 506–512, https://doi.org/10.1016/j.powtec.2011.09.015.
    [16] M.Z. Gao, B. Ludwig, T.A. Palmer, Impact of atomization gas on characteristics of austenitic stainless steel powder feedstocks for additive manufacturing, Powder
    Technol. 383 (2021) 30–42, https://doi.org/10.1016/j.powtec.2020.12.005.
    [17] X. Shui, K. Yamanaka, M. Mori, Y. Nagata, A. Chiba, Effects of post-processing on cyclic fatigue response of a titanium alloy additively manufactured by electron beam
    melting, Mater. Sci. Eng. A 680 (2017) 239–248, https://doi.org/10.1016/j.msea.
    2016.10.059.
    [18] C. Wang, X.H. Zhao, Y.C. Ma, Q.X. Wang, Y.J. Lai, S.J. Liang, Study of the spherical
    HoCu powders prepared by supreme-speed plasma rotating electrode process,
    Powder Metallurgy Technology 38 (3) (2020) 227–233, https://doi.org/10.19591/
    j.cnki.cn11-1974/tf.2020.03.011 (in Chinese).
    [19] G. Chen, S.Y. Zhao, P. Tan, J. Wang, C.S. Xiang, H.P. Tang, A comparative study of Ti6Al-4V powders for additive manufacturing by gas atomization, plasma rotating
    electrode process and plasma atomization, Powder Technol. 333 (2018) 38–46,
    https://doi.org/10.1016/j.powtec.2018.04.013.
    [20] Y. Zhao, K. Aoyagi, Y. Daino, K. Yamanaka, A. Chiba, Significance of powder feedstock
    characteristics in defect suppression of additively manufactured Inconel 718, Addit.
    Manuf. 34 (2020) 101277, https://doi.org/10.1016/j.addma.2020.101277.
    [21] Y. Nie, J. Tang, B. Yang, Q. Lei, S. Yu, Y. Li, Comparison in characteristic and atomization behavior of metallic powders produced by plasma rotating electrode process,
    Adv. Powder Technol. 31 (2020) 2152–2160, https://doi.org/10.1016/j.apt.2020.03.
    006.
    [22] Y. Cui, Y. Zhao, H. Numata, H. Bian, K. Wako, K. Yamanaka, K. Aoyagi, C. Zhang, A.
    Chiba, Effects of plasma rotating electrode process parameters on the particle size
    distribution and microstructure of Ti-6Al-4 V alloy powder, Powder Technol 376
    (2020) 363–372, https://doi.org/10.1016/j.powtec.2020.08.027.
    [23] J. Tang, Y. Nie, Q. Lei, Y. Li, Characteristics and atomization behavior of Ti-6Al-4V
    powder produced by plasma rotating electrode process Adv, Powder Technol. 10
    (2019) 2330–2337, https://doi.org/10.1016/j.apt.2019.07.015.
    [24] M. Zdujić, D. Uskoković, Production of atomized metal and alloy powders by the rotating electrode process, Sov. Powder Metall. Met. Ceram. 29 (1990) 673–683,
    https://doi.org/10.1007/BF00795571.
    [25] L. Zhang, Y. Zhao, Particle size distribution of tin powder produced by centrifugal
    atomisation using rotating cups, Powder Technol. 318 (2017) 62–67, https://doi.
    org/10.1016/j.powtec.2017.05.038.
    [26] Y. Liu, S. Liang, Z. Han, J. Song, Q. Wang, A novel model of calculating particle sizes in
    plasma rotating electrode process for superalloys, Powder Technol. 336 (2018)
    406–414, https://doi.org/10.1016/j.powtec.2018.06.002.
    [27] Y. Zhao, Y. Cui, H. Numata, H. Bian, K. Wako, K. Yamanaka, Centrifugal granulation
    behavior in metallic powder fabrication by plasma rotating electrode process, Sci.
    Rep. (2020) 1–15, https://doi.org/10.1038/s41598-020-75503-w.
    [28] T. Hsu, C. Wei, L. Wu, Y. Li, A. Chiba, M. Tsai, Nitinol powders generate from plasma
    rotation electrode process provide clean powder for biomedical devices used with
    suitable size, spheroid surface and pure composition, Sci. Rep. 8 (2018) 1–8,
    https://doi.org/10.1038/s41598-018-32101-1.
    [29] M. Wei, S. Chen, M. Sun, J. Liang, C. Liu, M. Wang, Atomization simulation and preparation of 24CrNiMoY alloy steel powder using VIGA technology at high gas pressure, Powder Technol. 367 (2020) 724–739, https://doi.org/10.1016/j.powtec.
    2020.04.030.
    [30] Y. Tan, X. Zhu, X.Y. He, B. Ding, H. Wang, Q. Liao, H. Li, Granulation characteristics of
    molten blast furnace slag by hybrid centrifugal-air blast technique, Powder Technol.
    323 (2018) 176–185, https://doi.org/10.1016/j.powtec.2017.09.040.
    [31] P. Xu, D.H. Liu, J. Hu, G.Y. Lin, Synthesis of Ni-Ti composite powder by radio frequency plasma spheroidization process, Nonferrous Metals Science and Engineering
    39 (1) (2020) 67–71 , (in Chinese) 10.13264/j.cnki.ysjskx.2020.01.011.
    [32] H. Mehboob, F. Tarlochan, A. Mehboob, S.H. Chang, S. Ramesh, W.S.W. Harun, K.
    Kadirgama, A novel design, analysis and 3D printing of Ti-6Al-4V alloy bioinspired porous femoral stem, J. Mater. Sci. Mater. Med. 31 (2020) 78, https://doi.
    org/10.1007/s10856-020-06420-7.
    [33] FLOW-3D® Version 11.2 [Computer software]. , Flow Science, Inc., Santa Fe, NM,
    2017https://www.flow3d.com.
    [34] M. Boivineau, C. Cagran, D. Doytier, V. Eyraud, M.H. Nadal, B. Wilthan, G. Pottlacher,
    Thermophysical properties of solid and liquid Ti-6Al-4V (TA6V) alloy, Int. J.
    Thermophys. 27 (2006) 507–529, https://doi.org/10.1007/PL00021868.
    [35] J. Liu, Q. Qin, Q. Yu, The effect of size distribution of slag particles obtained in dry
    granulation on blast furnace slag cement strength, Powder Technol. 362 (2020)
    32–36, https://doi.org/10.1016/j.powtec.2019.11.115.
    [36] M. Tanaka, S. Tashiro, A study of thermal pinch effect of welding arcs, J. Japan Weld.
    Soc. 25 (2007) 336–342, https://doi.org/10.2207/qjjws.25.336 (in Japanese).
    [37] T. Kamiya, A. Kayano, Disintegration of viscous fluid in the ligament state purged
    from a rotating disk, J. Chem. Eng. JAPAN. 4 (1971) 364–369, https://doi.org/10.
    1252/jcej.4.364.
    [38] T. Kamiya, An analysis of the ligament-type disintegration of thin liquid film at the
    edge of a rotating disk, J. Chem. Eng. Japan. 5 (1972) 391–396, https://doi.org/10.
    1252/jcej.5.391.
    [39] J. Burns, C. Ramshaw, R. Jachuck, Measurement of liquid film thickness and the determination of spin-up radius on a rotating disc using an electrical resistance technique, Chem. Eng. Sci. 58 (2003) 2245–2253, https://doi.org/10.1016/S0009-2509
    (03)00091-5.
    [40] J. Rauscher, R. Kelly, J. Cole, An asymptotic solution for the laminar flow of a thin film
    on a rotating disk, J. Appl. Mech. Trans. ASME 40 (1973) 43–47, https://doi.org/10.
    1115/1.3422970

    Figure 2. Schematic diagram for pilot-scale cooling-water circulation system (a) along with a real picture of the system (b).

    Application of Computational Fluid Dynamics in Chlorine-Dynamics Modeling of In-Situ Chlorination Systems for Cooling Systems

    Jongchan Yi 1, Jonghun Lee 1, Mohd Amiruddin Fikri 2,3, Byoung-In Sang 4 and Hyunook Kim 1,*

    Abstract

    염소화는 상대적인 효율성과 저렴한 비용으로 인해 발전소 냉각 시스템에서 생물학적 오염을 제어하는​​데 선호되는 방법입니다. 해안 지역에 발전소가 있는 경우 바닷물을 사용하여 현장에서 염소를 전기화학적으로 생성할 수 있습니다. 이를 현장 전기염소화라고 합니다. 이 접근 방식은 유해한 염소화 부산물이 적고 염소를 저장할 필요가 없다는 점을 포함하여 몇 가지 장점이 있습니다. 그럼에도 불구하고, 이 전기화학적 공정은 실제로는 아직 초기 단계에 있습니다. 이 연구에서는 파일럿 규모 냉각 시스템에서 염소 붕괴를 시뮬레이션하기 위해 병렬 1차 동역학을 적용했습니다. 붕괴가 취수관을 따라 발생하기 때문에 동역학은 전산유체역학(CFD) 코드에 통합되었으며, 이후에 파이프의 염소 거동을 시뮬레이션하는데 적용되었습니다. 실험과 시뮬레이션 데이터는 강한 난류가 형성되는 조건하에서도 파이프 벽을 따라 염소 농도가 점진적인 것으로 나타났습니다. 염소가 중간보다 파이프 표면을 따라 훨씬 더 집중적으로 남아 있다는 사실은 전기 염소화를 기반으로 하는 시스템의 전체 염소 요구량을 감소시킬 수 있었습니다. 현장 전기 염소화 방식의 냉각 시스템은 직접 주입 방식에 필요한 염소 사용량의 1/3만 소비했습니다. 따라서 현장 전기염소화는 해안 지역의 발전소에서 바이오파울링 제어를 위한 비용 효율적이고 환경 친화적인 접근 방식으로 사용될 수 있다고 결론지었습니다.

    Chlorination is the preferred method to control biofouling in a power plant cooling system due to its comparative effectiveness and low cost. If a power plant is located in a coastal area, chlorine can be electrochemically generated in-situ using seawater, which is called in-situ electrochlorination; this approach has several advantages including fewer harmful chlorination byproducts and no need for chlorine storage. Nonetheless, this electrochemical process is still in its infancy in practice. In this study, a parallel first-order kinetics was applied to simulate chlorine decay in a pilot-scale cooling system. Since the decay occurs along the water-intake pipe, the kinetics was incorporated into computational fluid dynamics (CFD) codes, which were subsequently applied to simulate chlorine behavior in the pipe. The experiment and the simulation data indicated that chlorine concentrations along the pipe wall were incremental, even under the condition where a strong turbulent flow was formed. The fact that chlorine remained much more concentrated along the pipe surface than in the middle allowed for the reduction of the overall chlorine demand of the system based on the electro-chlorination. The cooling system, with an in-situ electro-chlorination, consumed only 1/3 of the chlorine dose demanded by the direct injection method. Therefore, it was concluded that in-situ electro-chlorination could serve as a cost-effective and environmentally friendly approach for biofouling control at power plants on coastal areas.

    Keywords

    computational fluid dynamics; power plant; cooling system; electro-chlorination; insitu chlorination

    Figure 1. Electrodes and batch experiment set-up. (a) Two cylindrical electrodes used in this study. (b) Batch experiment set-up for kinetic tests.
    Figure 1. Electrodes and batch experiment set-up. (a) Two cylindrical electrodes used in this study. (b) Batch experiment set-up for kinetic tests.
    Figure 2. Schematic diagram for pilot-scale cooling-water circulation system (a) along with a real picture of the system (b).
    Figure 2. Schematic diagram for pilot-scale cooling-water circulation system (a) along with a real picture of the system (b).
    Figure 3. Free chlorine decay curves in seawater with different TOC and initial chlorine concentration. Each line represents the predicted concentration of chlorine under a given condition. (a) Artificial seawater solution with 1 mg L−1 of TOC; (b) artificial seawater solution with 2 mg L−1 of TOC; (c) artificial seawater solution with 3 mg L−1 of TOC; (d) West Sea water (1.3 mg L−1 of TOC).
    Figure 3. Free chlorine decay curves in seawater with different TOC and initial chlorine concentration. Each line represents the predicted concentration of chlorine under a given condition. (a) Artificial seawater solution with 1 mg L−1 of TOC; (b) artificial seawater solution with 2 mg L−1 of TOC; (c) artificial seawater solution with 3 mg L−1 of TOC; (d) West Sea water (1.3 mg L−1 of TOC).
    Figure 4. Correlation between model and experimental data in the chlorine kinetics using seawater.
    Figure 4. Correlation between model and experimental data in the chlorine kinetics using seawater.
    Figure 5. Free chlorine concentrations in West Sea water under different current conditions in an insitu electro-chlorination system.
    Figure 5. Free chlorine concentrations in West Sea water under different current conditions in an insitu electro-chlorination system.
    Figure 6. Free chlorine distribution along the sampling ports under different flow rates. Each dot represents experimental data, and each point on the black line is the expected chlorine concentration obtained from computational fluid dynamics (CFD) simulation with a parallel first-order decay model. The red-dotted line is the desirable concentration at the given flow rate: (a) 600 L min−1 of flow rate, (b) 700 L min−1 of flow rate, (c) 800 L min−1 of flow rate, (d) 900 L min−1 of flow rate.
    Figure 6. Free chlorine distribution along the sampling ports under different flow rates. Each dot represents experimental data, and each point on the black line is the expected chlorine concentration obtained from computational fluid dynamics (CFD) simulation with a parallel first-order decay model. The red-dotted line is the desirable concentration at the given flow rate: (a) 600 L min−1 of flow rate, (b) 700 L min−1 of flow rate, (c) 800 L min−1 of flow rate, (d) 900 L min−1 of flow rate.
    Figure 7. Fluid contour images from CFD simulation of the electro-chlorination experiment. Inlet flow rate is 800 L min−1. Outlet pressure was set to 10.8 kPa. (a) Chlorine concentration; (b) expanded view of electrode side in image (a); (c) velocity magnitude; (d) pressure.
    Figure 7. Fluid contour images from CFD simulation of the electro-chlorination experiment. Inlet flow rate is 800 L min−1. Outlet pressure was set to 10.8 kPa. (a) Chlorine concentration; (b) expanded view of electrode side in image (a); (c) velocity magnitude; (d) pressure.
    Figure 8. Chlorine concentration contour in the simulation of full-scale in-situ electro-chlorination with different cathode positions. The pipe diameter is 2 m and the flow rate is 14 m3 s−1. The figure shows 10 m of the pipeline. (a) The simulation result when the cathode is placed on the surface of the pipe wall. (b) The simulation result when the cathode is placed on the inside of the pipe with 100 mm of distance from the pipe wall.
    Figure 8. Chlorine concentration contour in the simulation of full-scale in-situ electro-chlorination with different cathode positions. The pipe diameter is 2 m and the flow rate is 14 m3 s−1. The figure shows 10 m of the pipeline. (a) The simulation result when the cathode is placed on the surface of the pipe wall. (b) The simulation result when the cathode is placed on the inside of the pipe with 100 mm of distance from the pipe wall.
    Figure 9. Comparison of in-situ electro-chlorination and direct chlorine injection in full-scale applications. (a) Estimated chlorine concentrations along the pipe surface. (b) Relative chlorine demands.
    Figure 9. Comparison of in-situ electro-chlorination and direct chlorine injection in full-scale applications. (a) Estimated chlorine concentrations along the pipe surface. (b) Relative chlorine demands.

    References

    1. Macknick, J.; Newmark, R.; Heath, G.; Hallett, K.C. Operational water consumption and withdrawal factors for electricity generating technologies: A review of existing literature. Environ. Res. Lett. 2012, 7, 045802.
    2. Pan, S.-Y.; Snyder, S.W.; Packman, A.I.; Lin, Y.J.; Chiang, P.-C. Cooling water use in thermoelectric power generation and its associated challenges for addressing water-energy nexus. Water-Energy Nexus 2018, 1, 26–41.
    3. Feeley, T.J., III; Skone, T.J.; Stiegel, G.J., Jr.; McNemar, A.; Nemeth, M.; Schimmoller, B.; Murphy, J.T.;
      Manfredo, L. Water: A critical resource in the thermoelectric power industry. Energy 2008, 33, 1–11.
    4. World Nuclear Association. World Nuclear Performance Report 2016; World Nuclear Association: London, UK, 2016.
    5. Pugh, S.; Hewitt, G.; Müller-Steinhagen, H. Fouling during the use of seawater as coolant—The development of a user guide. Heat Transf. Eng. 2005, 26, 35–43.
    6. Satpathy, K.K.; Mohanty, A.K.; Sahu, G.; Biswas, S.; Prasad, M.; Slvanayagam, M. Biofouling and its control in seawater cooled power plant cooling water system—A review. Nucl. Power 2010, 17, 191–242.
    7. Cristiani, P.; Perboni, G. Antifouling strategies and corrosion control in cooling circuits. Bioelectrochemistry 2014, 97, 120–126.
    8. Walker, M.E.; Safari, I.; Theregowda, R.B.; Hsieh, M.-K.; Abbasian, J.; Arastoopour, H.; Dzombak, D.A.; Miller, D.C. Economic impact of condenser fouling in existing thermoelectric power plants. Energy 2012,44, 429–437.
    9. Yi, J.; Ahn, Y.; Hong, M.; Kim, G.-H.; Shabnam, N.; Jeon, B.; Sang, B.-I.; Kim, H. Comparison between OCl−-Injection and In Situ Electrochlorination in the Formation of Chlorate and Perchlorate in Seawater. Appl.Sci. 2019, 9, 229.
    10. Xue, Y.; Zhao, J.; Qiu, R.; Zheng, J.; Lin, C.; Ma, B.; Wang, P. In Situ glass antifouling using Pt nanoparticle coating for periodic electrolysis of seawater. Appl. Surf. Sci. 2015, 357, 60–68.
    11. Mahfouz, A.B.; Atilhan, S.; Batchelor, B.; Linke, P.; Abdel-Wahab, A.; El-Halwagi, M.M. Optimal scheduling of biocide dosing for seawater-cooled power and desalination plants. Clean Technol. Environ. Policy 2011, 13, 783–796.
    12. Rubio, D.; López-Galindo, C.; Casanueva, J.F.; Nebot, E. Monitoring and assessment of an industrial antifouling treatment. Seasonal effects and influence of water velocity in an open once-through seawater cooling system. Appl. Therm. Eng. 2014, 67, 378–387.
    13. European Integrated Pollution Prevention and Control (IPPC) Bureau, European Commission. Reference Document on the Application of Best Available Techniques to Industrial Cooling Systems December 2001; European Commission, Tech. Rep: Brussels, Belgium, 2001.
    14. Venkatesan R.; Murthy P. S. Macrofouling Control in Power Plants. In Springer Series on Biofilms; Springer: Berlin/Heidelberg, Germany, 2008.
    15. Kastl, G.; Fisher, I.; Jegatheesan, V. Evaluation of chlorine decay kinetics expressions for drinking water distribution systems modelling. J. Water Supply Res. Technol. AQUA 1999, 48, 219–226.
    16. Fisher, I.; Kastl, G.; Sathasivan, A.; Cook, D.; Seneverathne, L. General model of chlorine decay in blends of surface waters, desalinated water, and groundwaters. J. Environ. Eng. 2015, 141, 04015039.
    17. Fisher, I.; Kastl, G.; Sathasivan, A.; Jegatheesan, V. Suitability of chlorine bulk decay models for planning and management of water distribution systems. Crit. Rev. Environ. Sci. Technol. 2011, 41, 1843–1882.
    18. Fisher, I.; Kastl, G.; Sathasivan, A. Evaluation of suitable chlorine bulk-decay models for water distribution systems. Water Res. 2011, 45, 4896–4908.
    19. Haas, C.N.; Karra, S. Kinetics of wastewater chlorine demand exertion. J. (Water Pollut. Control Fed.) 1984, 56, 170–173.
    20. Zeng, J.; Jiang, Z.; Chen, Q.; Zheng, P.; Huang, Y. The decay kinetics of residual chlorine in cooling seawater simulation experiments. Acta Oceanol. Sin. 2009, 28, 54–59.
    21. Saeed, S.; Prakash, S.; Deb, N.; Campbell, R.; Kolluru, V.; Febbo, E.; Dupont, J. Development of a sitespecific kinetic model for chlorine decay and the formation of chlorination by-products in seawater. J. Mar. Sci. Eng. 2015, 3, 772–792.
    22. Al Heboos, S.; Licskó, I. Application and comparison of two chlorine decay models for predicting bulk chlorine residuals. Period. Polytech. Civ. Eng. 2017, 61, 7–13.
    23. Shadloo, M.S.; Oger, G.; Le Touzé, D. Smoothed particle hydrodynamics method for fluid flows, towards industrial applications: Motivations, current state, and challenges. Comput. Fluids 2016, 136, 11–34.
    24. Wols, B.; Hofman, J.; Uijttewaal, W.; Rietveld, L.; Van Dijk, J. Evaluation of different disinfection calculation methods using CFD. Environ. Model. Softw. 2010, 25, 573–582.
    25. Angeloudis, A.; Stoesser, T.; Falconer, R.A. Predicting the disinfection efficiency range in chlorine contact tanks through a CFD-based approach. Water Res. 2014, 60, 118–129.
    26. Zhang, J.; Tejada-Martínez, A.E.; Zhang, Q. Developments in computational fluid dynamics-based modeling for disinfection technologies over the last two decades: A review. Environ. Model. Softw. 2014, 58,71–85.
    27. Lim, Y.H.; Deering, D.D. In Modeling Chlorine Residual in a Ground Water Supply Tank for a Small Community in Cold Conditions, World Environmental and Water Resources Congress 2017; American Society of Civil Engineers: Reston, Virginia, USA, 2017; pp. 124–138.
    28. Hernández-Cervantes, D.; Delgado-Galván, X.; Nava, J.L.; López-Jiménez, P.A.; Rosales, M.; Mora Rodríguez, J. Validation of a computational fluid dynamics model for a novel residence time distribution analysis in mixing at cross-junctions. Water 2018, 10, 733.
    29. Hua, F.; West, J.; Barker, R.; Forster, C. Modelling of chlorine decay in municipal water supplies. Water Res. 1999, 33, 2735–2746.
    30. Jonkergouw, P.M.; Khu, S.-T.; Savic, D.A.; Zhong, D.; Hou, X.Q.; Zhao, H.-B. A variable rate coefficient chlorine decay model. Environ. Sci. Technol. 2009, 43, 408–414.
    31. Nejjari, F.; Puig, V.; Pérez, R.; Quevedo, J.; Cugueró, M.; Sanz, G.; Mirats, J. Chlorine decay model calibration and comparison: Application to a real water network. Procedia Eng. 2014, 70, 1221–1230.
    32. Kohpaei, A.J.; Sathasivan, A.; Aboutalebi, H. Effectiveness of parallel second order model over second and first order models. Desalin. Water Treat. 2011, 32, 107–114.
    33. Powell, J.C.; Hallam, N.B.; West, J.R.; Forster, C.F.; Simms, J. Factors which control bulk chlorine decay rates. Water Res. 2000, 34, 117–126.
    34. Clark, R.M.; Sivaganesan, M. Predicting chlorine residuals in drinking water: Second order model. J. Water Resour. Plan. Manag. 2002, 128, 152–161.
    35. Li, X.; Li, C.; Bayier, M.; Zhao, T.; Zhang, T.; Chen, X.; Mao, X. Desalinated seawater into pilot-scale drinking water distribution system: Chlorine decay and trihalomethanes formation. Desalin. Water Treat. 2016, 57,19149–19159.
    36. United States Environmental Protection Agency (EPA). Chlorine, Total Residual (Spectrophotometric, DPD); EPA-NERL: 330.5; EPA: Cincinnati, OH, USA, 1978.
    37. Polman, H.; Verhaart, F.; Bruijs, M. Impact of biofouling in intake pipes on the hydraulics and efficiency of pumping capacity. Desalin. Water Treat. 2013, 51, 997–1003.
    38. Rajagopal, S.; Van der Velde, G.; Van der Gaag, M.; Jenner, H.A. How effective is intermittent chlorination to control adult mussel fouling in cooling water systems? Water Res. 2003, 37, 329–338.
    39. Bruijs, M.C.; Venhuis, L.P.; Daal, L. Global Experiences in Optimizing Biofouling Control through PulseChlorination®. 2017. Available online: https://www.researchgate.net/publication/318561645_Global_Experiences_in_Optimizing_Biofouling_Co ntrol_through_Pulse-ChlorinationR (accessed on 1 May 2020).
    40. Kim, H.; Hao, O.J.; McAvoy, T.J. Comparison between model-and pH/ORP-based process control for an AAA system. Tamkang J. Sci. Eng. 2000, 3, 165–172.
    41. Brdys, M.; Chang, T.; Duzinkiewicz, K. Intelligent Model Predictive Control of Chlorine Residuals in Water Distribution Systems, Bridging the Gap: Meeting the World’s Water and Environmental Resources Challenges. In Proceedings of the ASCE Water Resource Engineering and Water Resources Planning and Management, July 30–August 2, 2000; pp. 1–11
    Fig. 5. The predicted shapes of initial breach (a) Rectangular (b) V-notch. Fig. 6. Dam breaching stages.

    Investigating the peak outflow through a spatial embankment dam breach

    공간적 제방댐 붕괴를 통한 최대 유출량 조사

    Mahmoud T.GhonimMagdy H.MowafyMohamed N.SalemAshrafJatwaryFaculty of Engineering, Zagazig University, Zagazig 44519, Egypt

    Abstract

    Investigating the breach outflow hydrograph is an essential task to conduct mitigation plans and flood warnings. In the present study, the spatial dam breach is simulated by using a three-dimensional computational fluid dynamics model, FLOW-3D. The model parameters were adjusted by making a comparison with a previous experimental model. The different parameters (initial breach shape, dimensions, location, and dam slopes) are studied to investigate their effects on dam breaching. The results indicate that these parameters have a significant impact. The maximum erosion rate and peak outflow for the rectangular shape are higher than those for the V-notch by 8.85% and 5%, respectively. Increasing breach width or decreasing depth by 5% leads to increasing maximum erosion rate by 11% and 15%, respectively. Increasing the downstream slope angle by 4° leads to an increase in both peak outflow and maximum erosion rate by 2.0% and 6.0%, respectively.

    유출 유출 수문곡선을 조사하는 것은 완화 계획 및 홍수 경보를 수행하는 데 필수적인 작업입니다. 본 연구에서는 3차원 전산유체역학 모델인 FLOW-3D를 사용하여 공간 댐 붕괴를 시뮬레이션합니다. 이전 실험 모델과 비교하여 모델 매개변수를 조정했습니다.

    다양한 매개변수(초기 붕괴 형태, 치수, 위치 및 댐 경사)가 댐 붕괴에 미치는 영향을 조사하기 위해 연구됩니다. 결과는 이러한 매개변수가 상당한 영향을 미친다는 것을 나타냅니다. 직사각형 형태의 최대 침식율과 최대 유출량은 V-notch보다 각각 8.85%, 5% 높게 나타났습니다.

    위반 폭을 늘리거나 깊이를 5% 줄이면 최대 침식률이 각각 11% 및 15% 증가합니다. 하류 경사각을 4° 증가시키면 최대 유출량과 최대 침식률이 각각 2.0% 및 6.0% 증가합니다.

    Keywords

    Spatial dam breach; FLOW-3D; Overtopping erosion; Computational fluid dynamics (CFD)

    1. Introduction

    There are many purposes for dam construction, such as protection from flood disasters, water storage, and power generationEmbankment failures may have a catastrophic impact on lives and infrastructure in the downstream regions. One of the most common causes of embankment dam failure is overtopping. Once the overtopping of the dam begins, the breach formation will start in the dam body then end with the dam failure. This failure occurs within a very short time, which threatens to be very dangerous. Therefore, understanding and modeling the embankment breaching processes is essential for conducting mitigation plans, flood warnings, and forecasting flood damage.

    The analysis of the dam breaching process is implemented by different techniques: comparative methods, empirical models with dimensional and dimensionless solutions, physical-based models, and parametric models. These models were described in detail [1]Parametric modeling is commonly used to simulate breach growth as a time-dependent linear process and calculate outflow discharge from the breach using hydraulics principles [2]. Alhasan et al. [3] presented a simple one-dimensional mathematical model and a computer code to simulate the dam breaching process. These models were validated by small dams breaching during the floods in 2002 in the Czech Republic. Fread [4] developed an erosion model (BREACH) based on hydraulics principles, sediment transport, and soil mechanics to estimate breach size, time of formation, and outflow discharge. Říha et al. [5] investigated the dam break process for a cascade of small dams using a simple parametric model for piping and overtopping erosion, as well as a 2D shallow-water flow model for the flood in downstream areas. Goodarzi et al. [6] implemented mathematical and statistical methods to assess the effect of inflows and wind speeds on the dam’s overtopping failure.

    Dam breaching studies can be divided into two main modes of erosion. The first mode is called “planar dam breach” where the flow overtops the whole dam width. While the second mode is called “spatial dam breach” where the flow overtops through the initial pilot channel (i.e., a channel created in the dam body). Therefore, the erosion will be in both vertical and horizontal directions [7].

    The erosion process through the embankment dams occurs due to the shear stress applied by water flows. The dam breaching evolution can be divided into three stages [8][9], but Y. Yang et al. [10] divided the breach development into five stages: Stage I, the seepage erosion; Stage II, the initial breach formation; Stage III, the head erosion; Stage IV, the breach expansion; and Stage V, the re-equilibrium of the river channel through the breach. Many experimental tests have been carried out on non-cohesive embankment dams with an initial breach to examine the effect of upstream inflow discharges on the longitudinal profile evolution and the time to inflection point [11].

    Zhang et al. [12] studied the effect of changing downstream slope angle, sediment grain size, and dam crest length on erosion rates. They noticed that increasing dam crest length and decreasing downstream slope angle lead to decreasing sediment transport rate. While the increase in sediment grain size leads to an increased sediment transport rate at the initial stages. Höeg et al. [13] presented a series of field tests to investigate the stability of embankment dams made of various materials. Overtopping and piping were among the failure tests carried out for the dams composed of homogeneous rock-fill, clay, or gravel with a height of up to 6.0 m. Hakimzadeh et al. [14] constructed 40 homogeneous cohesive and non-cohesive embankment dams to study the effect of changing sediment diameter and dam height on the breaching process. They also used genetic programming (GP) to estimate the breach outflow. Refaiy et al. [15] studied different scenarios for the downstream drain geometry, such as length, height, and angle, to minimize the effect of piping phenomena and therefore increase dam safety.

    Zhu et al. [16] examined the effect of headcut erosion on dam breach growth, especially in the case of cohesive dams. They found that the breach growth in non-cohesive embankments is slower than cohesive embankments due to the little effect of headcut. Schmocker and Hager [7] proposed a relationship for estimating peak outflow from the dam breach process.(1)QpQin-1=1.7exp-20hc23d5013H0

    where: Qp = peak outflow discharge.

    Qin = inflow discharge.

    hc = critical flow depth.

    d50 = mean sediment diameter.

    Ho = initial dam height.

    Yu et al. [17] carried out an experimental study for homogeneous non-cohesive embankment dams in a 180° bending rectangular flume to determine the effect of overtopping flows on breaching formation. They found that the main factors influencing breach formation are water level, river discharge, and embankment material diameter.

    Wu et al. [18] carried out a series of experiments to investigate the effect of breaching geometry on both non-cohesive and cohesive embankment dams in a U-bend flume due to overtopping flows. In the case of non-cohesive embankments, the non-symmetrical lateral expansion was noticed during the breach formation. This expansion was described by a coefficient ranging from 2.7 to 3.3.

    The numerical models of the dam breach can be categorized according to different parameters, such as flow dimensions (1D, 2D, or 3D), flow governing equations, and solution methods. The 1D models are mainly used to predict the outflow hydrograph from the dam breach. Saberi et al. [19] applied the 1D Saint-Venant equation, which is solved by the finite difference method to investigate the outflow hydrograph during dam overtopping failure. Because of the ability to study dam profile evolution and breach formation, 2D models are more applicable than 1D models. Guan et al. [20] and Wu et al. [21] employed both 2D shallow water equations (SWEs) and sediment erosion equations, which are solved by the finite volume method to study the effect of the dam’s geometry parameters on outflow hydrograph and dam profile evolution. Wang et al. [22] also proposed a second-order hybrid-type of total variation diminishing (TVD) finite-difference to estimate the breach outflow by solving the 2D (SWEs). The accuracy of (SWEs) for both vertical flow contraction and surface roughness has been assessed [23]. They noted that the accuracy of (SWEs) is acceptable for milder slopes, but in the case of steeper slopes, modelers should be more careful. Generally, the accuracy of 2D models is still low, especially with velocity distribution over the flow depth, lateral momentum exchange, density-driven flows, and bottom friction [24]. Therefore, 3D models are preferred. Larocque et al. [25] and Yang et al. [26] started to use three-dimensional (3D) models that depend on the Reynolds-averaged Navier-Stokes (RANS) equations.

    Previous experimental studies concluded that there is no clear relationship between the peak outflow from the dam breach and the initial breach characteristics. Some of these studies depend on the sharp-crested weir fixed at the end of the flume to determine the peak outflow from the breach, which leads to a decrease in the accuracy of outflow calculations at the microscale. The main goals of this study are to carry out a numerical simulation for a spatial dam breach due to overtopping flows by using (FLOW-3D) software to find an empirical equation for the peak outflow discharge from the breach and determine the worst-case that leads to accelerating the dam breaching process.

    2. Numerical simulation

    The current study for spatial dam breach is simulated by using (FLOW-3D) software [27], which is a powerful computational fluid dynamics (CFD) program.

    2.1. Geometric presentations

    A stereolithographic (STL) file is prepared for each change in the initial breach geometry and dimensions. The CAD program is useful for creating solid objects and converting them to STL format, as shown in Fig. 1.

    2.2. Governing equations

    The governing equations for water flow are three-dimensional Reynolds Averaged Navier-Stokes equations (RANS).

    The continuity equation:(2)∂ui∂xi=0

    The momentum equation:(3)∂ui∂t+1VFuj∂ui∂xj=1ρ∂∂xj-pδij+ν∂ui∂xj+∂uj∂xi-ρu`iu`j¯

    where u is time-averaged velocity,ν is kinematic viscosity, VF is fractional volume open to flow, p is averaged pressure and -u`iu`j¯ are components of Reynold’s stress. The Volume of Fluid (VOF) technique is used to simulate the free surface profile. Hirt et al. [28] presented the VOF algorithm, which employs the function (F) to express the occupancy of each grid cell with fluid. The value of (F) varies from zero to unity. Zero value refers to no fluid in the grid cell, while the unity value refers to the grid cell being fully occupied with fluid. The free surface is formed in the grid cells having (F) values between zero and unity.(4)∂F∂t+1VF∂∂xFAxu+∂∂yFAyv+∂∂zFAzw=0

    where (u, v, w) are the velocity components in (x, y, z) coordinates, respectively, and (AxAyAz) are the area fractions.

    2.3. Boundary and initial conditions

    To improve the accuracy of the results, the boundary conditions should be carefully determined. In this study, two mesh blocks are used to minimize the time consumed in the simulation. The boundary conditions for mesh block 1 are as follows: The inlet and sides boundaries are defined as a wall boundary condition (wall boundary condition is usually used for bound fluid by solid regions. In the case of viscous flows, no-slip means that the tangential velocity is equal to the wall velocity and the normal velocity is zero), the outlet is defined as a symmetry boundary condition (symmetry boundary condition is usually used to reduce computational effort during CFD simulation. This condition allows the flow to be transferred from one mesh block to another. No inputs are required for this boundary condition except that its location should be defined accurately), the bottom boundary is defined as a uniform flow rate boundary condition, and the top boundary is defined as a specific pressure boundary condition with assigned atmospheric pressure. The boundary conditions for mesh block 2 are as follows: The inlet is defined as a symmetry boundary condition, the outlet is defined as a free flow boundary condition, the bottom and sides boundaries are defined as a wall boundary condition, and the top boundary is defined as a specific pressure boundary condition with assigned atmospheric pressure as shown in Fig. 2. The initial conditions required to be set for the fluid (i.e., water) inside of the domain include configuration, temperature, velocities, and pressure distribution. The configuration of water depends on the dimensions and shape of the dam reservoir. While the other conditions have been assigned as follows: temperature is normal water temperature (25 °c) and pressure distribution is hydrostatic with no initial velocity.

    2.4. Numerical method

    FLOW-3D uses the finite volume method (FVM) to solve the governing equation (Reynolds-averaged Navier-Stokes) over the computational domain. A finite-volume method is an Eulerian approach for representing and evaluating partial differential equations in algebraic equations form [29]. At discrete points on the mesh geometry, values are determined. Finite volume expresses a small volume surrounding each node point on a mesh. In this method, the divergence theorem is used to convert volume integrals with a divergence term to surface integrals. After that, these terms are evaluated as fluxes at each finite volume’s surfaces.

    2.5. Turbulent models

    Turbulence is the chaotic, unstable motion of fluids that occurs when there are insufficient stabilizing viscous forces. In FLOW-3D, there are six turbulence models available: the Prandtl mixing length model, the one-equation turbulent energy model, the two-equation (k – ε) model, the Renormalization-Group (RNG) model, the two-equation (k – ω) models, and a large eddy simulation (LES) model. For simulating flow motion, the RNG model is adopted to simulate the motion behavior better than the k – ε and k – ω.

    models [30]. The RNG model consists of two main equations for the turbulent kinetic energy KT and its dissipation.εT(5)∂kT∂t+1VFuAx∂kT∂x+vAy∂kT∂y+wAz∂kT∂z=PT+GT+DiffKT-εT(6)∂εT∂t+1VFuAx∂εT∂x+vAy∂εT∂y+wAz∂εT∂z=C1.εTKTPT+c3.GT+Diffε-c2εT2kT

    where KT is the turbulent kinetic energy, PT is the turbulent kinetic energy production, GT is the buoyancy turbulence energy, εT is the turbulent energy dissipation rate, DiffKT and Diffε are terms of diffusion, c1, c2 and c3 are dimensionless parameters, in which c1 and c3 have a constant value of 1.42 and 0.2, respectively, c2 is computed from the turbulent kinetic energy (KT) and turbulent production (PT) terms.

    2.6. Sediment scour model

    The sediment scour model available in FLOW-3D can calculate all the sediment transport processes including Entrainment transport, Bedload transport, Suspended transport, and Deposition. The erosion process starts once the water flows remove the grains from the packed bed and carry them into suspension. It happens when the applied shear stress by water flows exceeds critical shear stress. This process is represented by entrainment transport in the numerical model. After entrained, the grains carried by water flow are represented by suspended load transport. After that, some suspended grains resort to settling because of the combined effect of gravity, buoyancy, and friction. This process is described through a deposition. Finally, the grains sliding motions are represented by bedload transport in the model. For the entrainment process, the shear stress applied by the fluid motion on the packed bed surface is calculated using the standard wall function as shown in Eq.7.(7)ks,i=Cs,i∗d50

    where ks,i is the Nikuradse roughness and Cs,i is a user-defined coefficient. The critical bed shear stress is defined by a dimensionless parameter called the critical shields number as expressed in Eq.8.(8)θcr,i=τcr,i‖g‖diρi-ρf

    where θcr,i is the critical shields number, τcr,i is the critical bed shear stress, g is the absolute value of gravity acceleration, di is the diameter of the sediment grain, ρi is the density of the sediment species (i) and ρf is the density of the fluid. The value of the critical shields number is determined according to the Soulsby-Whitehouse equation.(9)θcr,i=0.31+1.2d∗,i+0.0551-exp-0.02d∗,i

    where d∗,i is the dimensionless diameter of the sediment, given by Eq.10.(10)d∗,i=diρfρi-ρf‖g‖μf213

    where μf is the fluid dynamic viscosity. For the sloping bed interface, the value of the critical shields number is modified according to Eq.11.(11)θ`cr,i=θcr,icosψsinβ+cos2βtan2φi-sin2ψsin2βtanφi

    where θ`cr,i is the modified critical shields number, φi is the angle of repose for the sediment, β is the angle of bed slope and ψ is the angle between the flow and the upslope direction. The effects of the rolling, hopping, and sliding motions of grains along the packed bed surface are taken by the bedload transport process. The volumetric bedload transport rate (qb,i) per width of the bed is expressed in Eq.12.(12)qb,i=Φi‖g‖ρi-ρfρfdi312

    where Φi is the dimensionless bedload transport rate is calculated by using Meyer Peter and Müller equation.(13)Φi=βMPM,iθi-θ`cr,i1.5cb,i

    where βMPM,i is the Meyer Peter and Müller user-defined coefficient and cb,i is the volume fraction of species i in the bed material. The suspended load transport is calculated as shown in Eq.14.(14)∂Cs,i∂t+∇∙Cs,ius,i=∇∙∇DCs,i

    where Cs,i is the suspended sediment mass concentration, D is the diffusivity, and us,i is the grain velocity of species i. Entrainment and deposition are two opposing processes that take place at the same time. The lifting and settling velocities for both entrainment and deposition processes are calculated according to Eq.15 and Eq.16, respectively.(15)ulifting,i=αid∗,i0.3θi-θ`cr,igdiρiρf-1(16)usettling,i=υfdi10.362+1.049d∗,i3-10.36

    where αi is the entrainment coefficient of species i and υf is the kinematic viscosity of the fluid.

    2.7. Grid type

    Using simple rectangular orthogonal elements in planes and hexahedral in volumes in the (FLOW-3D) program makes the mesh generation process easier, decreases the required memory, and improves numerical accuracy. Two mesh blocks were used in a joined form with a size ratio of 2:1. The first mesh block is coarser, which contains the reservoir water, and the second mesh block is finer, which contains the dam. For achieving accuracy and efficiency in results, the mesh size is determined by using a grid convergence test. The optimum uniform cell size for the first mesh block is 0.012 m and for the second mesh block is 0.006 m.

    2.8. Time step

    The maximum time step size is determined by using a Courant number, which controls the distance that the flow will travel during the simulation time step. In this study, the Courant number was taken equal to 0.25 to prevent the flow from traveling through more than one cell in the time step. Based on the Courant number, a maximum time step value of 0.00075 s was determined.

    2.9. Numerical model validation

    The numerical model accuracy was achieved by comparing the numerical model results with previous experimental results. The experimental study of Schmocker and Hager [7] was based on 31 tests with changes in six parameters (d50, Ho, Bo, Lk, XD, and Qin). All experimental tests were conducted in a straight open glass-sided flume. The horizontal flume has a rectangular cross-section with a width of 0.4 m and a height of 0.7 m. The flume was provided with a flow straightener and an intake with a length of 0.66 m. All tested dams were inserted at various distances (XD) from the intake. Test No.1 from this experimental program was chosen to validate the numerical model. The different parameters used in test No.1 are as follows:

    (1) uniform sediment with a mean diameter (d50 = 0.31 mm), (2) Ho = 0.2 m, (3) Bo = 0.2 m, (4) Lk = 0.1 m,

    (5) XD = 1.0 m, (6) Qin = 6.0 lit/s, (7) Su and Sd = 2:1, (8) mass density (ρs = 2650 kg/m3(9) Homogenous and non-cohesive embankment dam. As shown in Fig. 2, the simulation is contained within a rectangular grid with dimensions: 3.56 m in the x-direction (where 0.66 m is used as inlet, 0.9 m as dam base width, and 1.0 m as outlet), in y-direction 0.2 m (dam length), and in the z-direction 0.3 m, which represents the dam height (0.2 m) with a free distance (0.1 m) above the dam. There are two main reasons that this experimental program is preferred for the validation process. The first reason is that this program deals with homogenous, non-cohesive soil, which is available in FLOW-3D. The second reason is that this program deals with small-scale models which saves time for numerical simulation. Finally, some important assumptions were considered during the validation process. The flow is assumed to be incompressible, viscous, turbulent, and three-dimensional.

    By comparing dam profiles at different time instants for the experimental test with the current numerical model, it appears that the numerical model gives good agreement as shown in Fig. 3 and Fig. 4, with an average error percentage of 9% between the experimental results and the numerical model.

    3. Analysis and discussions

    The current model is used to study the effects of different parameters such as (initial breach shapes, dimensions, locations, upstream and downstream dam slopes) on the peak outflow discharge, QP, time of peak outflow, tP, and rate of erosion, E.

    This study consists of a group of scenarios. The first scenario is changing the shapes of the initial breach according to Singh [1], the most predicted shapes are rectangular and V-notch as shown in Fig. 5. The second scenario is changing the initial breach dimensions (i.e., width and depth). While the third scenario is changing the location of the initial breach. Eventually, the last scenario is changing the upstream and downstream dam slopes.

    All scenarios of this study were carried out under the same conditions such as inflow discharge value (Qin=1.0lit/s), dimensions of the tested dam, where dam height (Ho=0.20m), crest width.

    (Lk=0.1m), dam length (Bo=0.20m), and homogenous & non-cohesive soil with a mean diameter (d50=0.31mm).

    3.1. Dam breaching process evolution

    The dam breaching process is a very complex process due to the quick changes in hydrodynamic conditions during dam failure. The dam breaching process starts once water flows reach the downstream face of the dam. During the initial stage of dam breaching, the erosion process is relatively quiet due to low velocities of flow. As water flows continuously, erosion rates increase, especially in two main zones: the crest and the downstream face. As soon as the dam crest is totally eroded, the water levels in the dam reservoir decrease rapidly, accompanied by excessive erosion in the dam body. The erosion process continues until the water levels in the dam reservoir equal the remaining height of the dam.

    According to Zhou et al. [11], the breaching process consists of three main stages. The first stage starts with beginning overtopping flow, then ends when the erosion point directed upstream and reached the inflection point at the inflection time (ti). The second stage starts from the end of the stage1 until the occurrence of peak outflow discharge at the peak outflow time (tP). The third stage starts from the end of the stage2 until the value of outflow discharge becomes the same as the value of inflow discharge at the final time (tf). The outflow discharge from the dam breach increases rapidly during stage1 and stage2 because of the large dam storage capacity (i.e., the dam reservoir is totally full of water) and excessive erosion. While at stage3, the outflow values start to decrease slowly because most of the dam’s storage capacity was run out. The end of stage3 indicates that the dam storage capacity was totally run out, so the outflow equalized with the inflow discharge as shown in Fig. 6 and Fig. 7.

    3.2. The effect of initial breach shape

    To identify the effect of the initial breach shape on the evolution of the dam breaching process. Three tests were carried out with different cross-section areas for each shape. The initial breach is created at the center of the dam crest. Each test had an ID to make the process of arranging data easier. The rectangular shape had an ID (Rec5h & 5b), which means that its depth and width are equal to 5% of the dam height, and the V-notch shape had an ID (V-noch5h & 1:1) which means that its depth is equal to 5% of the dam height and its side slope is equal to 1:1. The comparison between rectangular and V-notch shapes is done by calculating the ratio between maximum dam height at different times (ZMax) to the initial dam height (Ho), rate of erosion, and hydrograph of outflow discharge for each test. The rectangular shape achieves maximum erosion rate and minimum inflection time, in addition to a rapid decrease in the dam reservoir levels. Therefore, the dam breaching is faster in the case of a rectangular shape than in a V-notch shape, which has the same cross-section area as shown in Fig. 8.

    Also, by comparing the hydrograph for each test, the peak outflow discharge value in the case of a rectangular shape is higher than the V-notch shape by 5% and the time of peak outflow for the rectangular shape is shorter than the V-notch shape by 9% as shown in Fig. 9.

    3.3. The effect of initial breach dimensions

    The results of the comparison between the different initial breach shapes indicate that the worst initial breach shape is rectangular, so the second scenario from this study concentrated on studying the effect of a change in the initial rectangular breach dimensions. Groups of tests were carried out with different depths and widths for the rectangular initial breach. The first group had a depth of 5% from the dam height and with three different widths of 5,10, and 15% from the dam height, the second group had a depth of 10% with three different widths of 5,10, and 15%, the third group had a depth of 15% with three different widths of 5,10, and 15% and the final group had a width of 15% with three different heights of 5, 10, and 15% for a rectangular breach shape. The comparison was made as in the previous section to determine the worst case that leads to the quick dam failure as shown in Fig. 10.

    The results show that the (Rec 5 h&15b) test achieves a maximum erosion rate for a shorter period of time and a minimum ratio for (Zmax / Ho) as shown in Fig. 10, which leads to accelerating the dam failure process. The dam breaching process is faster with the minimum initial breach depth and maximum initial breach width. In the case of a minimum initial breach depth, the retained head of water in the dam reservoir is high and the crest width at the bottom of the initial breach (L`K) is small, so the erosion point reaches the inflection point rapidly. While in the case of the maximum initial breach width, the erosion perimeter is large.

    3.4. The effect of initial breach location

    The results of the comparison between the different initial rectangular breach dimensions indicate that the worst initial breach dimension is (Rec 5 h&15b), so the third scenario from this study concentrated on studying the effect of a change in the initial breach location. Three locations were checked to determine the worst case for the dam failure process. The first location is at the center of the dam crest, which was named “Center”, the second location is at mid-distance between the dam center and dam edge, which was named “Mid”, and the third location is at the dam edge, which was named “Edge” as shown in Fig. 11. According to this scenario, the results indicate that the time of peak outflow discharge (tP) is the same in the three cases, but the maximum value of the peak outflow discharge occurs at the center location. The difference in the peak outflow values between the three cases is relatively small as shown in Fig. 12.

    The rates of erosion were also studied for the three cases. The results show that the maximum erosion rate occurs at the center location as shown in Fig. 13. By making a comparison between the three cases for the dam storage volume. The results show that the center location had the minimum values for the dam storage volume, which means that a large amount of water has passed to the downstream area as shown in Fig. 14. According to these results, the center location leads to increased erosion rate and accelerated dam failure process compared with the two other cases. Because the erosion occurs on both sides, but in the case of edge location, the erosion occurs on one side.

    3.5. The effect of upstream and downstream dam slopes

    The results of the comparison between the different initial rectangular breach locations indicate that the worst initial breach location is the center location, so the fourth scenario from this study concentrated on studying the effect of a change in the upstream (Su) and downstream (Sd) dam slopes. Three slopes were checked individually for both upstream and downstream slopes to determine the worst case for the dam failure process. The first slope value is (2H:1V), the second slope value is (2.5H:1V), and the third slope value is (3H:1V). According to this scenario, the results show that the decreasing downstream slope angle leads to increasing time of peak outflow discharge (tP) and decreasing value of peak outflow discharge. The difference in the peak outflow values between the three cases for the downstream slope is 2%, as shown in Fig. 15, but changing the upstream slope has a negligible impact on the peak outflow discharge and its time as shown in Fig. 16.

    The rates of erosion were also studied in the three cases for both upstream and downstream slopes. The results show that the maximum erosion rate increases by 6.0% with an increasing downstream slope angle by 4°, as shown in Fig. 17. The results also indicate that the erosion rates aren’t affected by increasing or decreasing the upstream slope angle, as shown in Fig. 18. According to these results, increasing the downstream slope angle leads to increased erosion rate and accelerated dam failure process compared with the upstream slope angle. Because of increasing shear stress applied by water flows in case of increasing downstream slope.

    According to all previous scenarios, the dimensionless peak outflow discharge QPQin is presented for a fixed dam height (Ho) and inflow discharge (Qin). Fig. 19 illustrates the relationship between QP∗=QPQin and.

    Lr=ho2/3∗bo2/3Ho. The deduced relationship achieves R2=0.96.(17)QP∗=2.2807exp-2.804∗Lr

    4. Conclusions

    A spatial dam breaching process was simulated by using FLOW-3D Software. The validation process was performed by making a comparison between the simulated results of dam profiles and the dam profiles obtained by Schmocker and Hager [7] in their experimental study. And also, the peak outflow value recorded an error percentage of 12% between the numerical model and the experimental study. This model was used to study the effect of initial breach shape, dimensions, location, and dam slopes on peak outflow discharge, time of peak outflow, and the erosion process. By using the parameters obtained from the validation process, the results of this study can be summarized in eight points as follows.1.

    The rectangular initial breach shape leads to an accelerating dam failure process compared with the V-notch.2.

    The value of peak outflow discharge in the case of a rectangular initial breach is higher than the V-notch shape by 5%.3.

    The time of peak outflow discharge for a rectangular initial breach is shorter than the V-notch shape by 9%.4.

    The minimum depth and maximum width for the initial breach achieve maximum erosion rates (increasing breach width, b0, or decreasing breach depth, h0, by 5% from the dam height leads to an increase in the maximum rate of erosion by 11% and 15%, respectively), so the dam failure is rapid.5.

    The center location of the initial breach leads to an accelerating dam failure compared with the edge location.6.

    The initial breach location has a negligible effect on the peak outflow discharge value and its time.7.

    Increasing the downstream slope angle by 4° leads to an increase in both peak outflow discharge and maximum rate of erosion by 2.0% and 6.0%, respectively.8.

    The upstream slope has a negligible effect on the dam breaching process.

    References

    [1]V. SinghDam breach modeling technologySpringer Science & Business Media (1996)Google Scholar[2]Wahl TL. Prediction of embankment dam breach parameters: a literature review and needs assessment. 1998.Google Scholar[3]Z. Alhasan, J. Jandora, J. ŘíhaStudy of dam-break due to overtopping of four small dams in the Czech RepublicActa Universitatis Agriculturae et Silviculturae Mendelianae Brunensis, 63 (3) (2015), pp. 717-729 View PDFCrossRefView Record in ScopusGoogle Scholar[4]D. FreadBREACH, an erosion model for earthen dam failures: Hydrologic Research LaboratoryNOAA, National Weather Service (1988)Google Scholar[5]J. Říha, S. Kotaška, L. PetrulaDam Break Modeling in a Cascade of Small Earthen Dams: Case Study of the Čižina River in the Czech RepublicWater, 12 (8) (2020), p. 2309, 10.3390/w12082309 View PDFView Record in ScopusGoogle Scholar[6]E. Goodarzi, L. Teang Shui, M. ZiaeiDam overtopping risk using probabilistic concepts–Case study: The Meijaran DamIran Ain Shams Eng J, 4 (2) (2013), pp. 185-197ArticleDownload PDFView Record in ScopusGoogle Scholar[7]L. Schmocker, W.H. HagerPlane dike-breach due to overtopping: effects of sediment, dike height and dischargeJ Hydraul Res, 50 (6) (2012), pp. 576-586 View PDFCrossRefView Record in ScopusGoogle Scholar[8]J.S. Walder, R.M. Iverson, J.W. Godt, M. Logan, S.A. SolovitzControls on the breach geometry and flood hydrograph during overtopping of noncohesive earthen damsWater Resour Res, 51 (8) (2015), pp. 6701-6724View Record in ScopusGoogle Scholar[9]H. Wei, M. Yu, D. Wang, Y. LiOvertopping breaching of river levees constructed with cohesive sedimentsNat Hazards Earth Syst Sci, 16 (7) (2016), pp. 1541-1551 View PDFCrossRefView Record in ScopusGoogle Scholar[10]Y. Yang, S.-Y. Cao, K.-J. Yang, W.-P. LiYang K-j, Li W-p. Experimental study of breach process of landslide dams by overtopping and its initiation mechanismsJ Hydrodynamics, 27 (6) (2015), pp. 872-883ArticleDownload PDFCrossRefView Record in ScopusGoogle Scholar[11]G.G.D. Zhou, M. Zhou, M.S. Shrestha, D. Song, C.E. Choi, K.F.E. Cui, et al.Experimental investigation on the longitudinal evolution of landslide dam breaching and outburst floodsGeomorphology, 334 (2019), pp. 29-43ArticleDownload PDFView Record in ScopusGoogle Scholar[12]J. Zhang, Z.-x. Guo, S.-y. CaoYang F-g. Experimental study on scour and erosion of blocked damWater Sci Eng, 5 (2012), pp. 219-229ArticleDownload PDFView Record in ScopusGoogle Scholar[13]K. Höeg, A. Løvoll, K. VaskinnStability and breaching of embankment dams: Field tests on 6 m high damsInt J Hydropower Dams, 11 (2004), pp. 88-92View Record in ScopusGoogle Scholar[14]H. Hakimzadeh, V. Nourani, A.B. AminiGenetic programming simulation of dam breach hydrograph and peak outflow dischargeJ Hydrol Eng, 19 (4) (2014), pp. 757-768View Record in ScopusGoogle Scholar[15]A.R. Refaiy, N.M. AboulAtta, N.Y. Saad, D.A. El-MollaModeling the effect of downstream drain geometry on seepage through earth damsAin Shams Eng J, 12 (3) (2021), pp. 2511-2531ArticleDownload PDFView Record in ScopusGoogle Scholar[16]Y. Zhu, P.J. Visser, J.K. Vrijling, G. WangExperimental investigation on breaching of embankmentsScience China Technological Sci, 54 (1) (2011), pp. 148-155 View PDFCrossRefView Record in ScopusGoogle Scholar[17]M.-H. Yu, H.-Y. Wei, Y.-J. Liang, Y. ZhaoInvestigation of non-cohesive levee breach by overtopping flowJ Hydrodyn, 25 (4) (2013), pp. 572-579ArticleDownload PDFCrossRefView Record in ScopusGoogle Scholar[18]S. Wu, M. Yu, H. Wei, Y. Liang, J. ZengNon-symmetrical levee breaching processes in a channel bend due to overtoppingInt J Sedim Res, 33 (2) (2018), pp. 208-215ArticleDownload PDFView Record in ScopusGoogle Scholar[19]O. Saberi, G. ZenzNumerical investigation on 1D and 2D embankment dams failure due to overtopping flowInt J Hydraulic Engineering, 5 (2016), pp. 9-18View Record in ScopusGoogle Scholar[20]M. Guan, N.G. Wright, P.A. Sleigh2D Process-Based Morphodynamic Model for Flooding by Noncohesive Dyke BreachJ Hydraul Eng, 140 (7) (2014), p. 04014022, 10.1061/(ASCE)HY.1943-7900.0000861 View PDFView Record in ScopusGoogle Scholar[21]W. Wu, R. Marsooli, Z. HeDepth-Averaged Two-Dimensional Model of Unsteady Flow and Sediment Transport due to Noncohesive Embankment Break/BreachingJ Hydraul Eng, 138 (6) (2012), pp. 503-516View Record in ScopusGoogle Scholar[22]Z. Wang, D.S. BowlesThree-dimensional non-cohesive earthen dam breach model. Part 1: Theory and methodologyAdv Water Resour, 29 (10) (2006), pp. 1528-1545ArticleDownload PDFView Record in ScopusGoogle Scholar[23]Říha J, Duchan D, Zachoval Z, Erpicum S, Archambeau P, Pirotton M, et al. Performance of a shallow-water model for simulating flow over trapezoidal broad-crested weirs. J Hydrology Hydromechanics. 2019;67:322-8.Google Scholar[24]C.B. VreugdenhilNumerical methods for shallow-water flowSpringer Science & Business Media (1994)Google Scholar[25]L.A. Larocque, J. Imran, M.H. Chaudhry3D numerical simulation of partial breach dam-break flow using the LES and k–∊ turbulence modelsJ Hydraul Res, 51 (2) (2013), pp. 145-157 View PDFCrossRefView Record in ScopusGoogle Scholar[26]C. Yang, B. Lin, C. Jiang, Y. LiuPredicting near-field dam-break flow and impact force using a 3D modelJ Hydraul Res, 48 (6) (2010), pp. 784-792 View PDFCrossRefView Record in ScopusGoogle Scholar[27]FLOW-3D. Version 11.1.1 Flow Science, Inc., Santa Fe, NM. https://wwwflow3dcom.Google Scholar[28]C.W. Hirt, B.D. NicholsVolume of fluid (VOF) method for the dynamics of free boundariesJ Comput Phys, 39 (1) (1981), pp. 201-225ArticleDownload PDFGoogle Scholar[29]S.V. PatankarNumerical heat transfer and fluid flow, Hemisphere PublCorp, New York, 58 (1980), p. 288View Record in ScopusGoogle Scholar[30]M. Alemi, R. MaiaNumerical simulation of the flow and local scour process around single and complex bridge piersInt J Civil Eng, 16 (5) (2018), pp. 475-487 View PDFCrossRefView Record in ScopusGoogle Scholar

    Figure 3 Simulation PTC pipes enhanced with copper foam and nanoparticles in FLOW-3D software.

    다공성 미디어 및 나노유체에 의해 강화된 수집기로 태양광 CCHP 시스템의 최적화

    Optimization of Solar CCHP Systems with Collector Enhanced by Porous Media and Nanofluid


    Navid Tonekaboni,1Mahdi Feizbahr,2 Nima Tonekaboni,1Guang-Jun Jiang,3,4 and Hong-Xia Chen3,4

    Abstract

    태양열 집열기의 낮은 효율은 CCHP(Solar Combined Cooling, Heating, and Power) 사이클의 문제점 중 하나로 언급될 수 있습니다. 태양계를 개선하기 위해 나노유체와 다공성 매체가 태양열 집열기에 사용됩니다.

    다공성 매질과 나노입자를 사용하는 장점 중 하나는 동일한 조건에서 더 많은 에너지를 흡수할 수 있다는 것입니다. 이 연구에서는 평균 일사량이 1b인 따뜻하고 건조한 지역의 600 m2 건물의 전기, 냉방 및 난방을 생성하기 위해 다공성 매질과 나노유체를 사용하여 태양열 냉난방 복합 발전(SCCHP) 시스템을 최적화했습니다.

    본 논문에서는 침전물이 형성되지 않는 lb = 820 w/m2(이란) 정도까지 다공성 물질에서 나노유체의 최적량을 계산하였다. 이 연구에서 태양열 집열기는 구리 다공성 매체(95% 다공성)와 CuO 및 Al2O3 나노 유체로 향상되었습니다.

    나노유체의 0.1%-0.6%가 작동 유체로 물에 추가되었습니다. 나노유체의 0.5%가 태양열 집열기 및 SCCHP 시스템에서 가장 높은 에너지 및 엑서지 효율 향상으로 이어지는 것으로 밝혀졌습니다.

    본 연구에서 포물선형 집열기(PTC)의 최대 에너지 및 엑서지 효율은 각각 74.19% 및 32.6%입니다. 그림 1은 태양 CCHP의 주기를 정확하게 설명하기 위한 그래픽 초록으로 언급될 수 있습니다.

    The low efficiency of solar collectors can be mentioned as one of the problems in solar combined cooling, heating, and power (CCHP) cycles. For improving solar systems, nanofluid and porous media are used in solar collectors. One of the advantages of using porous media and nanoparticles is to absorb more energy under the same conditions. In this research, a solar combined cooling, heating, and power (SCCHP) system has been optimized by porous media and nanofluid for generating electricity, cooling, and heating of a 600 m2 building in a warm and dry region with average solar radiation of Ib = 820 w/m2 in Iran. In this paper, the optimal amount of nanofluid in porous materials has been calculated to the extent that no sediment is formed. In this study, solar collectors were enhanced with copper porous media (95% porosity) and CuO and Al2O3 nanofluids. 0.1%–0.6% of the nanofluids were added to water as working fluids; it is found that 0.5% of the nanofluids lead to the highest energy and exergy efficiency enhancement in solar collectors and SCCHP systems. Maximum energy and exergy efficiency of parabolic thermal collector (PTC) riches in this study are 74.19% and 32.6%, respectively. Figure 1 can be mentioned as a graphical abstract for accurately describing the cycle of solar CCHP.

    1. Introduction

    Due to the increase in energy consumption, the use of clean energy is one of the important goals of human societies. In the last four decades, the use of cogeneration cycles has increased significantly due to high efficiency. Among clean energy, the use of solar energy has become more popular due to its greater availability [1]. Low efficiency of energy production, transmission, and distribution system makes a new system to generate simultaneously electricity, heating, and cooling as an essential solution to be widely used. The low efficiency of the electricity generation, transmission, and distribution system makes the CCHP system a basic solution to eliminate waste of energy. CCHP system consists of a prime mover (PM), a power generator, a heat recovery system (produce extra heating/cooling/power), and thermal energy storage (TES) [2]. Solar combined cooling, heating, and power (SCCHP) has been started three decades ago. SCCHP is a system that receives its propulsive force from solar energy; in this cycle, solar collectors play the role of propulsive for generating power in this system [3].

    Increasing the rate of energy consumption in the whole world because of the low efficiency of energy production, transmission, and distribution system causes a new cogeneration system to generate electricity, heating, and cooling energy as an essential solution to be widely used. Building energy utilization fundamentally includes power required for lighting, home electrical appliances, warming and cooling of building inside, and boiling water. Domestic usage contributes to an average of 35% of the world’s total energy consumption [4].

    Due to the availability of solar energy in all areas, solar collectors can be used to obtain the propulsive power required for the CCHP cycle. Solar energy is the main source of energy in renewable applications. For selecting a suitable area to use solar collectors, annual sunshine hours, the number of sunny days, minus temperature and frosty days, and the windy status of the region are essentially considered [5]. Iran, with an average of more than 300 sunny days, is one of the suitable countries to use solar energy. Due to the fact that most of the solar radiation is in the southern regions of Iran, also the concentration of cities is low in these areas, and transmission lines are far apart, one of the best options is to use CCHP cycles based on solar collectors [6]. One of the major problems of solar collectors is their low efficiency [7]. Low efficiency increases the area of collectors, which increases the initial cost of solar systems and of course increases the initial payback period. To increase the efficiency of solar collectors and improve their performance, porous materials and nanofluids are used to increase their workability.

    There are two ways to increase the efficiency of solar collectors and mechanical and fluid improvement. In the first method, using porous materials or helical filaments inside the collector pipes causes turbulence of the flow and increases heat transfer. In the second method, using nanofluids or salt and other materials increases the heat transfer of water. The use of porous materials has grown up immensely over the past twenty years. Porous materials, especially copper porous foam, are widely used in solar collectors. Due to the high contact surface area, porous media are appropriate candidates for solar collectors [8]. A number of researchers investigated Solar System performance in accordance with energy and exergy analyses. Zhai et al. [9] reviewed the performance of a small solar-powered system in which the energy efficiency was 44.7% and the electrical efficiency was 16.9%.

    Abbasi et al. [10] proposed an innovative multiobjective optimization to optimize the design of a cogeneration system. Results showed the CCHP system based on an internal diesel combustion engine was the applicable alternative at all regions with different climates. The diesel engine can supply the electrical requirement of 31.0% and heating demand of 3.8% for building.

    Jiang et al. [11] combined the experiment and simulation together to analyze the performance of a cogeneration system. Moreover, some research focused on CCHP systems using solar energy. It integrated sustainable and renewable technologies in the CCHP, like PV, Stirling engine, and parabolic trough collector (PTC) [21215].

    Wang et al. [16] optimized a cogeneration solar cooling system with a Rankine cycle and ejector to reach the maximum total system efficiency of 55.9%. Jing et al. analyzed a big-scale building with the SCCHP system and auxiliary heaters to produced electrical, cooling, and heating power. The maximum energy efficiency reported in their work is 46.6% [17]. Various optimization methods have been used to improve the cogeneration system, minimum system size, and performance, such as genetic algorithm [1819].

    Hirasawa et al. [20] investigated the effect of using porous media to reduce thermal waste in solar systems. They used the high-porosity metal foam on top of the flat plate solar collector and observed that thermal waste decreased by 7% due to natural heat transfer. Many researchers study the efficiency improvement of the solar collector by changing the collector’s shapes or working fluids. However, the most effective method is the use of nanofluids in the solar collector as working fluid [21]. In the experimental study done by Jouybari et al. [22], the efficiency enhancement up to 8.1% was achieved by adding nanofluid in a flat plate collector. In this research, by adding porous materials to the solar collector, collector efficiency increased up to 92% in a low flow regime. Subramani et al. [23] analyzed the thermal performance of the parabolic solar collector with Al2O3 nanofluid. They conducted their experiments with Reynolds number range 2401 to 7202 and mass flow rate 0.0083 to 0.05 kg/s. The maximum efficiency improvement in this experiment was 56% at 0.05 kg/s mass flow rate.

    Shojaeizadeh et al. [24] investigated the analysis of the second law of thermodynamic on the flat plate solar collector using Al2O3/water nanofluid. Their research showed that energy efficiency rose up to 1.9% and the exergy efficiency increased by a maximum of 0.72% compared to pure water. Tiwari et al. [25] researched on the thermal performance of solar flat plate collectors for working fluid water with different nanofluids. The result showed that using 1.5% (optimum) particle volume fraction of Al2O3 nanofluid as an absorbing medium causes the thermal efficiency to enhance up to 31.64%.

    The effect of porous media and nanofluids on solar collectors has already been investigated in the literature but the SCCHP system with a collector embedded by both porous media and nanofluid for enhancing the ratio of nanoparticle in nanofluid for preventing sedimentation was not discussed. In this research, the amount of energy and exergy of the solar CCHP cycles with parabolic solar collectors in both base and improved modes with a porous material (copper foam with 95% porosity) and nanofluid with different ratios of nanoparticles was calculated. In the first step, it is planned to design a CCHP system based on the required load, and, in the next step, it will analyze the energy and exergy of the system in a basic and optimize mode. In the optimize mode, enhanced solar collectors with porous material and nanofluid in different ratios (0.1%–0.7%) were used to optimize the ratio of nanofluids to prevent sedimentation.

    2. Cycle Description

    CCHP is one of the methods to enhance energy efficiency and reduce energy loss and costs. The SCCHP system used a solar collector as a prime mover of the cogeneration system and assisted the boiler to generate vapor for the turbine. Hot water flows from the expander to the absorption chiller in summer or to the radiator or fan coil in winter. Finally, before the hot water wants to flow back to the storage tank, it flows inside a heat exchanger for generating domestic hot water [26].

    For designing of solar cogeneration system and its analysis, it is necessary to calculate the electrical, heating (heating load is the load required for the production of warm water and space heating), and cooling load required for the case study considered in a residential building with an area of 600 m2 in the warm region of Iran (Zahedan). In Table 1, the average of the required loads is shown for the different months of a year (average of electrical, heating, and cooling load calculated with CARRIER software).Table 1 The average amount of electric charges, heating load, and cooling load used in the different months of the year in the city of Zahedan for a residential building with 600 m2.

    According to Table 1, the maximum magnitude of heating, cooling, and electrical loads is used to calculate the cogeneration system. The maximum electric load is 96 kW, the maximum amount of heating load is 62 kW, and the maximum cooling load is 118 kW. Since the calculated loads are average, all loads increased up to 10% for the confidence coefficient. With the obtained values, the solar collector area and other cogeneration system components are calculated. The cogeneration cycle is capable of producing 105 kW electric power, 140 kW cooling capacity, and 100 kW heating power.

    2.1. System Analysis Equations

    An analysis is done by considering the following assumptions:(1)The system operates under steady-state conditions(2)The system is designed for the warm region of Iran (Zahedan) with average solar radiation Ib = 820 w/m2(3)The pressure drops in heat exchangers, separators, storage tanks, and pipes are ignored(4)The pressure drop is negligible in all processes and no expectable chemical reactions occurred in the processes(5)Potential, kinetic, and chemical exergy are not considered due to their insignificance(6)Pumps have been discontinued due to insignificance throughout the process(7)All components are assumed adiabatic

    Schematic shape of the cogeneration cycle is shown in Figure 1 and all data are given in Table 2.

    Figure 1 Schematic shape of the cogeneration cycle.Table 2 Temperature and humidity of different points of system.

    Based on the first law of thermodynamic, energy analysis is based on the following steps.

    First of all, the estimated solar radiation energy on collector has been calculated:where α is the heat transfer enhancement coefficient based on porous materials added to the collector’s pipes. The coefficient α is increased by the porosity percentage, the type of porous material (in this case, copper with a porosity percentage of 95), and the flow of fluid to the collector equation.

    Collector efficiency is going to be calculated by the following equation [9]:

    Total energy received by the collector is given by [9]

    Also, the auxiliary boiler heat load is [2]

    Energy consumed from vapor to expander is calculated by [2]

    The power output form by the screw expander [9]:

    The efficiency of the expander is 80% in this case [11].

    In this step, cooling and heating loads were calculated and then, the required heating load to reach sanitary hot water will be calculated as follows:

    First step: calculating the cooling load with the following equation [9]:

    Second step: calculating heating loads [9]:

    Then, calculating the required loud for sanitary hot water will be [9]

    According to the above-mentioned equations, efficiency is [9]

    In the third step, calculated exergy analysis as follows.

    First, the received exergy collector from the sun is calculated [9]:

    In the previous equation, f is the constant of air dilution.

    The received exergy from the collector is [9]

    In the case of using natural gas in an auxiliary heater, the gas exergy is calculated from the following equation [12]:

    Delivering exergy from vapor to expander is calculated with the following equation [9]:

    In the fourth step, the exergy in cooling and heating is calculated by the following equation:

    Cooling exergy in summer is calculated [9]:

    Heating exergy in winter is calculated [9]:

    In the last step based on thermodynamic second law, exergy efficiency has been calculated from the following equation and the above-mentioned calculated loads [9]:

    3. Porous Media

    The porous medium that filled the test section is copper foam with a porosity of 95%. The foams are determined in Figure 2 and also detailed thermophysical parameters and dimensions are shown in Table 3.

    Figure 2 Copper foam with a porosity of 95%.Table 3 Thermophysical parameters and dimensions of copper foam.

    In solar collectors, copper porous materials are suitable for use at low temperatures and have an easier and faster manufacturing process than ceramic porous materials. Due to the high coefficient conductivity of copper, the use of copper metallic foam to increase heat transfer is certainly more efficient in solar collectors.

    Porous media and nanofluid in solar collector’s pipes were simulated in FLOW-3D software using the finite-difference method [27]. Nanoparticles Al2O3 and CUO are mostly used in solar collector enhancement. In this research, different concentrations of nanofluid are added to the parabolic solar collectors with porous materials (copper foam with porosity of 95%) to achieve maximum heat transfer in the porous materials before sedimentation. After analyzing PTC pipes with the nanofluid flow in FLOW-3D software, for energy and exergy efficiency analysis, Carrier software results were used as EES software input. Simulation PTC with porous media inside collector pipe and nanofluids sedimentation is shown in Figure 3.

    Figure 3 Simulation PTC pipes enhanced with copper foam and nanoparticles in FLOW-3D software.

    3.1. Nano Fluid

    In this research, copper and silver nanofluids (Al2O3, CuO) have been added with percentages of 0.1%–0.7% as the working fluids. The nanoparticle properties are given in Table 4. Also, system constant parameters are presented in Table 4, which are available as default input in the EES software.Table 4 Properties of the nanoparticles [9].

    System constant parameters for input in the software are shown in Table 5.Table 5 System constant parameters.

    The thermal properties of the nanofluid can be obtained from equations (18)–(21). The basic fluid properties are indicated by the index (bf) and the properties of the nanoparticle silver with the index (np).

    The density of the mixture is shown in the following equation [28]:where ρ is density and ϕ is the nanoparticles volume fraction.

    The specific heat capacity is calculated from the following equation [29]:

    The thermal conductivity of the nanofluid is calculated from the following equation [29]:

    The parameter β is the ratio of the nanolayer thickness to the original particle radius and, usually, this parameter is taken equal to 0.1 for the calculated thermal conductivity of the nanofluids.

    The mixture viscosity is calculated as follows [30]:

    In all equations, instead of water properties, working fluids with nanofluid are used. All of the above equations and parameters are entered in the EES software for calculating the energy and exergy of solar collectors and the SCCHP cycle. All calculation repeats for both nanofluids with different concentrations of nanofluid in the solar collector’s pipe.

    4. Results and Discussion

    In the present study, relations were written according to Wang et al. [16] and the system analysis was performed to ensure the correctness of the code. The energy and exergy charts are plotted based on the main values of the paper and are shown in Figures 4 and 5. The error rate in this simulation is 1.07%.

    Figure 4 Verification charts of energy analysis results.

    Figure 5 Verification charts of exergy analysis results.

    We may also investigate the application of machine learning paradigms [3141] and various hybrid, advanced optimization approaches that are enhanced in terms of exploration and intensification [4255], and intelligent model studies [5661] as well, for example, methods such as particle swarm optimizer (PSO) [6062], differential search (DS) [63], ant colony optimizer (ACO) [616465], Harris hawks optimizer (HHO) [66], grey wolf optimizer (GWO) [5367], differential evolution (DE) [6869], and other fusion and boosted systems [4146485054557071].

    At the first step, the collector is modified with porous copper foam material. 14 cases have been considered for the analysis of the SCCHP system (Table 6). It should be noted that the adding of porous media causes an additional pressure drop inside the collector [922263072]. All fourteen cases use copper foam with a porosity of 95 percent. To simulate the effect of porous materials and nanofluids, the first solar PTC pipes have been simulated in the FLOW-3D software and then porous media (copper foam with porosity of 95%) and fluid flow with nanoparticles (AL2O3 and CUO) are generated in the software. After analyzing PTC pipes in FLOW-3D software, for analyzing energy and exergy efficiency, software outputs were used as EES software input for optimization ratio of sedimentation and calculating energy and exergy analyses.Table 6 Collectors with different percentages of nanofluids and porous media.

    In this research, an enhanced solar collector with both porous media and Nanofluid is investigated. In the present study, 0.1–0.5% CuO and Al2O3 concentration were added to the collector fully filled by porous media to achieve maximum energy and exergy efficiencies of solar CCHP systems. All steps of the investigation are shown in Table 6.

    Energy and exergy analyses of parabolic solar collectors and SCCHP systems are shown in Figures 6 and 7.

    Figure 6 Energy and exergy efficiencies of the PTC with porous media and nanofluid.

    Figure 7 Energy and exergy efficiency of the SCCHP.

    Results show that the highest energy and exergy efficiencies are 74.19% and 32.6%, respectively, that is achieved in Step 12 (parabolic collectors with filled porous media and 0.5% Al2O3). In the second step, the maximum energy efficiency of SCCHP systems with fourteen steps of simulation are shown in Figure 7.

    In the second step, where 0.1, −0.6% of the nanofluids were added, it is found that 0.5% leads to the highest energy and exergy efficiency enhancement in solar collectors and SCCHP systems. Using concentrations more than 0.5% leads to sediment in the solar collector’s pipe and a decrease of porosity in the pipe [73]. According to Figure 7, maximum energy and exergy efficiencies of SCCHP are achieved in Step 12. In this step energy efficiency is 54.49% and exergy efficiency is 18.29%. In steps 13 and 14, with increasing concentration of CUO and Al2O3 nanofluid solution in porous materials, decreasing of energy and exergy efficiency of PTC and SCCHP system at the same time happened. This decrease in efficiency is due to the formation of sediment in the porous material. Calculations and simulations have shown that porous materials more than 0.5% nanofluids inside the collector pipe cause sediment and disturb the porosity of porous materials and pressure drop and reduce the coefficient of performance of the cogeneration system. Most experience showed that CUO and AL2O3 nanofluids with less than 0.6% percent solution are used in the investigation on the solar collectors at low temperatures and discharges [74]. One of the important points of this research is that the best ratio of nanofluids in the solar collector with a low temperature is 0.5% (AL2O3 and CUO); with this replacement, the cost of solar collectors and SCCHP cycle is reduced.

    5. Conclusion and Future Directions

    In the present study, ways for increasing the efficiency of solar collectors in order to enhance the efficiency of the SCCHP cycle are examined. The research is aimed at adding both porous materials and nanofluids for estimating the best ratio of nanofluid for enhanced solar collector and protecting sedimentation in porous media. By adding porous materials (copper foam with porosity of 95%) and 0.5% nanofluids together, high efficiency in solar parabolic collectors can be achieved. The novelty in this research is the addition of both nanofluids and porous materials and calculating the best ratio for preventing sedimentation and pressure drop in solar collector’s pipe. In this study, it was observed that, by adding 0.5% of AL2O3 nanofluid in working fluids, the energy efficiency of PTC rises to 74.19% and exergy efficiency is grown up to 32.6%. In SCCHP cycle, energy efficiency is 54.49% and exergy efficiency is 18.29%.

    In this research, parabolic solar collectors fully filled by porous media (copper foam with a porosity of 95) are investigated. In the next step, parabolic solar collectors in the SCCHP cycle were simultaneously filled by porous media and different percentages of Al2O3 and CuO nanofluid. At this step, values of 0.1% to 0.6% of each nanofluid were added to the working fluid, and the efficiency of the energy and exergy of the collectors and the SCCHP cycle were determined. In this case, nanofluid and the porous media were used together in the solar collector and maximum efficiency achieved. 0.5% of both nanofluids were used to achieve the biggest efficiency enhancement.

    In the present study, as expected, the highest efficiency is for the parabolic solar collector fully filled by porous material (copper foam with a porosity of 95%) and 0.5% Al2O3. Results of the present study are as follows:(1)The average enhancement of collectors’ efficiency using porous media and nanofluids is 28%.(2)Solutions with 0.1 to 0.5% of nanofluids (CuO and Al2O3) are used to prevent collectors from sediment occurrence in porous media.(3)Collector of solar cogeneration cycles that is enhanced by both porous media and nanofluid has higher efficiency, and the stability of output temperature is more as well.(4)By using 0.6% of the nanofluids in the enhanced parabolic solar collectors with copper porous materials, sedimentation occurs and makes a high-pressure drop in the solar collector’s pipe which causes decrease in energy efficiency.(5)Average enhancement of SCCHP cycle efficiency is enhanced by both porous media and nanofluid 13%.

    Nomenclature

    :Solar radiation
    a:Heat transfer augmentation coefficient
    A:Solar collector area
    Bf:Basic fluid
    :Specific heat capacity of the nanofluid
    F:Constant of air dilution
    :Thermal conductivity of the nanofluid
    :Thermal conductivity of the basic fluid
    :Viscosity of the nanofluid
    :Viscosity of the basic fluid
    :Collector efficiency
    :Collector energy receives
    :Auxiliary boiler heat
    :Expander energy
    :Gas energy
    :Screw expander work
    :Cooling load, in kilowatts
    :Heating load, in kilowatts
    :Solar radiation energy on collector, in Joule
    :Sanitary hot water load
    Np:Nanoparticle
    :Energy efficiency
    :Heat exchanger efficiency
    :Sun exergy
    :Collector exergy
    :Natural gas exergy
    :Expander exergy
    :Cooling exergy
    :Heating exergy
    :Exergy efficiency
    :Steam mass flow rate
    :Hot water mass flow rate
    :Specific heat capacity of water
    :Power output form by the screw expander
    Tam:Average ambient temperature
    :Density of the mixture.

    Greek symbols

    ρ:Density
    ϕ:Nanoparticles volume fraction
    β:Ratio of the nanolayer thickness.

    Abbreviations

    CCHP:Combined cooling, heating, and power
    EES:Engineering equation solver.

    Data Availability

    For this study, data were generated by CARRIER software for the average electrical, heating, and cooling load of a residential building with 600 m2 in the city of Zahedan, Iran.

    Conflicts of Interest

    The authors declare that they have no conflicts of interest.

    Acknowledgments

    This work was partially supported by the National Natural Science Foundation of China under Contract no. 71761030 and Natural Science Foundation of Inner Mongolia under Contract no. 2019LH07003.

    References

    1. A. Fudholi and K. Sopian, “Review on solar collector for agricultural produce,” International Journal of Power Electronics and Drive Systems (IJPEDS), vol. 9, no. 1, p. 414, 2018.View at: Publisher Site | Google Scholar
    2. G. Yang and X. Zhai, “Optimization and performance analysis of solar hybrid CCHP systems under different operation strategies,” Applied Thermal Engineering, vol. 133, pp. 327–340, 2018.View at: Publisher Site | Google Scholar
    3. J. Wang, Z. Han, and Z. Guan, “Hybrid solar-assisted combined cooling, heating, and power systems: a review,” Renewable and Sustainable Energy Reviews, vol. 133, p. 110256, 2020.View at: Publisher Site | Google Scholar
    4. Y. Tian and C. Y. Zhao, “A review of solar collectors and thermal energy storage in solar thermal applications,” Applied Energy, vol. 104, pp. 538–553, 2013.View at: Publisher Site | Google Scholar
    5. J. M. Hassan, Q. J. Abdul-Ghafour, and M. F. Mohammed, “CFD simulation of enhancement techniques in flat plate solar water collectors,” Al-Nahrain Journal for Engineering Sciences, vol. 20, no. 3, pp. 751–761, 2017.View at: Google Scholar
    6. M. Jahangiri, O. Nematollahi, A. Haghani, H. A. Raiesi, and A. Alidadi Shamsabadi, “An optimization of energy cost of clean hybrid solar-wind power plants in Iran,” International Journal of Green Energy, vol. 16, no. 15, pp. 1422–1435, 2019.View at: Publisher Site | Google Scholar
    7. I. H. Yılmaz and A. Mwesigye, “Modeling, simulation and performance analysis of parabolic trough solar collectors: a comprehensive review,” Applied Energy, vol. 225, pp. 135–174, 2018.View at: Google Scholar
    8. F. Wang, J. Tan, and Z. Wang, “Heat transfer analysis of porous media receiver with different transport and thermophysical models using mixture as feeding gas,” Energy Conversion and Management, vol. 83, pp. 159–166, 2014.View at: Publisher Site | Google Scholar
    9. H. Zhai, Y. J. Dai, J. Y. Wu, and R. Z. Wang, “Energy and exergy analyses on a novel hybrid solar heating, cooling and power generation system for remote areas,” Applied Energy, vol. 86, no. 9, pp. 1395–1404, 2009.View at: Publisher Site | Google Scholar
    10. M. H. Abbasi, H. Sayyaadi, and M. Tahmasbzadebaie, “A methodology to obtain the foremost type and optimal size of the prime mover of a CCHP system for a large-scale residential application,” Applied Thermal Engineering, vol. 135, pp. 389–405, 2018.View at: Google Scholar
    11. R. Jiang, F. G. F. Qin, X. Yang, S. Huang, and B. Chen, “Performance analysis of a liquid absorption dehumidifier driven by jacket-cooling water of a diesel engine in a CCHP system,” Energy and Buildings, vol. 163, pp. 70–78, 2018.View at: Publisher Site | Google Scholar
    12. F. A. Boyaghchi and M. Chavoshi, “Monthly assessments of exergetic, economic and environmental criteria and optimization of a solar micro-CCHP based on DORC,” Solar Energy, vol. 166, pp. 351–370, 2018.View at: Publisher Site | Google Scholar
    13. F. A. Boyaghchi and M. Chavoshi, “Multi-criteria optimization of a micro solar-geothermal CCHP system applying water/CuO nanofluid based on exergy, exergoeconomic and exergoenvironmental concepts,” Applied Thermal Engineering, vol. 112, pp. 660–675, 2017.View at: Publisher Site | Google Scholar
    14. B. Su, W. Han, Y. Chen, Z. Wang, W. Qu, and H. Jin, “Performance optimization of a solar assisted CCHP based on biogas reforming,” Energy Conversion and Management, vol. 171, pp. 604–617, 2018.View at: Publisher Site | Google Scholar
    15. F. A. Al-Sulaiman, F. Hamdullahpur, and I. Dincer, “Performance assessment of a novel system using parabolic trough solar collectors for combined cooling, heating, and power production,” Renewable Energy, vol. 48, pp. 161–172, 2012.View at: Publisher Site | Google Scholar
    16. J. Wang, Y. Dai, L. Gao, and S. Ma, “A new combined cooling, heating and power system driven by solar energy,” Renewable Energy, vol. 34, no. 12, pp. 2780–2788, 2009.View at: Publisher Site | Google Scholar
    17. Y.-Y. Jing, H. Bai, J.-J. Wang, and L. Liu, “Life cycle assessment of a solar combined cooling heating and power system in different operation strategies,” Applied Energy, vol. 92, pp. 843–853, 2012.View at: Publisher Site | Google Scholar
    18. J.-J. Wang, Y.-Y. Jing, and C.-F. Zhang, “Optimization of capacity and operation for CCHP system by genetic algorithm,” Applied Energy, vol. 87, no. 4, pp. 1325–1335, 2010.View at: Publisher Site | Google Scholar
    19. L. Ali, “LDA–GA–SVM: improved hepatocellular carcinoma prediction through dimensionality reduction and genetically optimized support vector machine,” Neural Computing and Applications, vol. 87, pp. 1–10, 2020.View at: Google Scholar
    20. S. Hirasawa, R. Tsubota, T. Kawanami, and K. Shirai, “Reduction of heat loss from solar thermal collector by diminishing natural convection with high-porosity porous medium,” Solar Energy, vol. 97, pp. 305–313, 2013.View at: Publisher Site | Google Scholar
    21. E. Bellos, C. Tzivanidis, and Z. Said, “A systematic parametric thermal analysis of nanofluid-based parabolic trough solar collectors,” Sustainable Energy Technologies and Assessments, vol. 39, p. 100714, 2020.View at: Publisher Site | Google Scholar
    22. H. J. Jouybari, S. Saedodin, A. Zamzamian, M. E. Nimvari, and S. Wongwises, “Effects of porous material and nanoparticles on the thermal performance of a flat plate solar collector: an experimental study,” Renewable Energy, vol. 114, pp. 1407–1418, 2017.View at: Publisher Site | Google Scholar
    23. J. Subramani, P. K. Nagarajan, S. Wongwises, S. A. El-Agouz, and R. Sathyamurthy, “Experimental study on the thermal performance and heat transfer characteristics of solar parabolic trough collector using Al2O3 nanofluids,” Environmental Progress & Sustainable Energy, vol. 37, no. 3, pp. 1149–1159, 2018.View at: Publisher Site | Google Scholar
    24. E. Shojaeizadeh, F. Veysi, and A. Kamandi, “Exergy efficiency investigation and optimization of an Al2O3-water nanofluid based Flat-plate solar collector,” Energy and Buildings, vol. 101, pp. 12–23, 2015.View at: Publisher Site | Google Scholar
    25. A. K. Tiwari, P. Ghosh, and J. Sarkar, “Solar water heating using nanofluids–a comprehensive overview and environmental impact analysis,” International Journal of Emerging Technology and Advanced Engineering, vol. 3, no. 3, pp. 221–224, 2013.View at: Google Scholar
    26. D. R. Rajendran, E. Ganapathy Sundaram, P. Jawahar, V. Sivakumar, O. Mahian, and E. Bellos, “Review on influencing parameters in the performance of concentrated solar power collector based on materials, heat transfer fluids and design,” Journal of Thermal Analysis and Calorimetry, vol. 140, no. 1, pp. 33–51, 2020.View at: Publisher Site | Google Scholar
    27. M. Feizbahr, C. Kok Keong, F. Rostami, and M. Shahrokhi, “Wave energy dissipation using perforated and non perforated piles,” International Journal of Engineering, vol. 31, no. 2, pp. 212–219, 2018.View at: Google Scholar
    28. K. Khanafer and K. Vafai, “A critical synthesis of thermophysical characteristics of nanofluids,” International Journal of Heat and Mass Transfer, vol. 54, no. 19-20, pp. 4410–4428, 2011.View at: Publisher Site | Google Scholar
    29. K. Farhana, K. Kadirgama, M. M. Rahman et al., “Improvement in the performance of solar collectors with nanofluids – a state-of-the-art review,” Nano-Structures & Nano-Objects, vol. 18, p. 100276, 2019.View at: Publisher Site | Google Scholar
    30. M. Turkyilmazoglu, “Condensation of laminar film over curved vertical walls using single and two-phase nanofluid models,” European Journal of Mechanics-B/Fluids, vol. 65, pp. 184–191, 2017.View at: Publisher Site | Google Scholar
    31. X. Zhang, J. Wang, T. Wang, R. Jiang, J. Xu, and L. Zhao, “Robust feature learning for adversarial defense via hierarchical feature alignment,” Information Sciences, vol. 2020, 2020.View at: Google Scholar
    32. X. Zhang, T. Wang, W. Luo, and P. Huang, “Multi-level fusion and attention-guided CNN for image dehazing,” IEEE Transactions on Circuits and Systems for Video Technology, vol. 1, 2020.View at: Publisher Site | Google Scholar
    33. X. Zhang, M. Fan, D. Wang, P. Zhou, and D. Tao, “Top-k feature selection framework using robust 0-1 integer programming,” IEEE Transactions on Neural Networks and Learning Systems, vol. 1, pp. 1–15, 2020.View at: Publisher Site | Google Scholar
    34. X. Zhang, D. Wang, Z. Zhou, and Y. Ma, “Robust low-rank tensor recovery with rectification and alignment,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 43, no. 1, pp. 238–255, 2019.View at: Google Scholar
    35. X. Zhang, R. Jiang, T. Wang, and J. Wang, “Recursive neural network for video deblurring,” IEEE Transactions on Circuits and Systems for Video Technology, vol. 1, 2020.View at: Publisher Site | Google Scholar
    36. X. Zhang, T. Wang, J. Wang, G. Tang, and L. Zhao, “Pyramid channel-based feature attention network for image dehazing,” Computer Vision and Image Understanding, vol. 1, 2020.View at: Google Scholar
    37. M. Mirmozaffari, “Machine learning algorithms based on an optimization model,” 2020.View at: Google Scholar
    38. M. Mirmozaffari, M. Yazdani, A. Boskabadi, H. Ahady Dolatsara, K. Kabirifar, and N. Amiri Golilarz, “A novel machine learning approach combined with optimization models for eco-efficiency evaluation,” Applied Sciences, vol. 10, no. 15, p. 5210, 2020.View at: Publisher Site | Google Scholar
    39. M. Vosoogha and A. Addeh, “An intelligent power prediction method for wind energy generation based on optimized fuzzy system,” Computational Research Progress in Applied Science & Engineering (CRPASE), vol. 5, pp. 34–43, 2019.View at: Google Scholar
    40. A. Javadi, N. Mikaeilvand, and H. Hosseinzdeh, “Presenting a new method to solve partial differential equations using a group search optimizer method (GSO),” Computational Research Progress in Applied Science and Engineering, vol. 4, no. 1, pp. 22–26, 2018.View at: Google Scholar
    41. F. J. Golrokh, Gohar Azeem, and A. Hasan, “Eco-efficiency evaluation in cement industries: DEA malmquist productivity index using optimization models,” ENG Transactions, vol. 1, pp. 1–8, 2020.View at: Google Scholar
    42. H. Yu, “Dynamic Gaussian bare-bones fruit fly optimizers with abandonment mechanism: method and analysis,” Engineering with Computers, vol. 1, pp. 1–29, 2020.View at: Google Scholar
    43. C. Yu, “SGOA: annealing-behaved grasshopper optimizer for global tasks,” Engineering with Computers, vol. 1, pp. 1–28, 2021.View at: Google Scholar
    44. W. Shan, Z. Qiao, A. A. Heidari, H. Chen, H. Turabieh, and Y. Teng, “Double adaptive weights for stabilization of moth flame optimizer: balance analysis, engineering cases, and medical diagnosis,” Knowledge-Based Systems, vol. 1, p. 106728, 2020.View at: Google Scholar
    45. J. Tu, H. Chen, J. Liu et al., “Evolutionary biogeography-based whale optimization methods with communication structure: towards measuring the balance,” Knowledge-Based Systems, vol. 212, p. 106642, 2021.View at: Publisher Site | Google Scholar
    46. Y. Zhang, “Towards augmented kernel extreme learning models for bankruptcy prediction: algorithmic behavior and comprehensive analysis,” Neurocomputing, vol. 1, 2020.View at: Google Scholar
    47. Y. Zhang, R. Liu, X. Wang, H. Chen, and C. Li, “Boosted binary Harris hawks optimizer and feature selection,” Engineering with Computers, vol. 1, pp. 1–30, 2020.View at: Google Scholar
    48. H.-L. Chen, G. Wang, C. Ma, Z.-N. Cai, W.-B. Liu, and S.-J. Wang, “An efficient hybrid kernel extreme learning machine approach for early diagnosis of Parkinson’s disease,” Neurocomputing, vol. 184, pp. 131–144, 2016.View at: Publisher Site | Google Scholar
    49. L. Hu, G. Hong, J. Ma, X. Wang, and H. Chen, “An efficient machine learning approach for diagnosis of paraquat-poisoned patients,” Computers in Biology and Medicine, vol. 59, pp. 116–124, 2015.View at: Publisher Site | Google Scholar
    50. L. Shen, H. Chen, Z. Yu et al., “Evolving support vector machines using fruit fly optimization for medical data classification,” Knowledge-Based Systems, vol. 96, pp. 61–75, 2016.View at: Publisher Site | Google Scholar
    51. J. Xia, H. Chen, Q. Li et al., “Ultrasound-based differentiation of malignant and benign thyroid Nodules: an extreme learning machine approach,” Computer Methods and Programs in Biomedicine, vol. 147, pp. 37–49, 2017.View at: Publisher Site | Google Scholar
    52. C. Li, L. Hou, B. Y. Sharma et al., “Developing a new intelligent system for the diagnosis of tuberculous pleural effusion,” Computer Methods and Programs in Biomedicine, vol. 153, pp. 211–225, 2018.View at: Publisher Site | Google Scholar
    53. X. Zhao, X. Zhang, Z. Cai et al., “Chaos enhanced grey wolf optimization wrapped ELM for diagnosis of paraquat-poisoned patients,” Computational Biology and Chemistry, vol. 78, pp. 481–490, 2019.View at: Publisher Site | Google Scholar
    54. M. Wang and H. Chen, “Chaotic multi-swarm whale optimizer boosted support vector machine for medical diagnosis,” Applied Soft Computing Journal, vol. 88, 2020.View at: Publisher Site | Google Scholar
    55. X. Xu and H.-L. Chen, “Adaptive computational chemotaxis based on field in bacterial foraging optimization,” Soft Computing, vol. 18, no. 4, pp. 797–807, 2014.View at: Publisher Site | Google Scholar
    56. R. U. Khan, X. Zhang, R. Kumar, A. Sharif, N. A. Golilarz, and M. Alazab, “An adaptive multi-layer botnet detection technique using machine learning classifiers,” Applied Sciences, vol. 9, no. 11, p. 2375, 2019.View at: Publisher Site | Google Scholar
    57. A. Addeh, A. Khormali, and N. A. Golilarz, “Control chart pattern recognition using RBF neural network with new training algorithm and practical features,” ISA Transactions, vol. 79, pp. 202–216, 2018.View at: Publisher Site | Google Scholar
    58. N. Amiri Golilarz, H. Gao, R. Kumar, L. Ali, Y. Fu, and C. Li, “Adaptive wavelet based MRI brain image de-noising,” Frontiers in Neuroscience, vol. 14, p. 728, 2020.View at: Publisher Site | Google Scholar
    59. N. A. Golilarz, H. Gao, and H. Demirel, “Satellite image de-noising with Harris hawks meta heuristic optimization algorithm and improved adaptive generalized Gaussian distribution threshold function,” IEEE Access, vol. 7, pp. 57459–57468, 2019.View at: Publisher Site | Google Scholar
    60. M. Eisazadeh and J. Rezapour, “Multi-objective optimization of the composite sheets using PSO algorithm,” 2017.View at: Google Scholar
    61. I. Bargegol, M. Nikookar, R. V. Nezafat, E. J. Lashkami, and A. M. Roshandeh, “Timing optimization of signalized intersections using shockwave theory by genetic algorithm,” Computational Research Progress in Applied Science & Engineering, vol. 1, pp. 160–167, 2015.View at: Google Scholar
    62. B. Bai, Z. Guo, C. Zhou, W. Zhang, and J. Zhang, “Application of adaptive reliability importance sampling-based extended domain PSO on single mode failure in reliability engineering,” Information Sciences, vol. 546, pp. 42–59, 2021.View at: Publisher Site | Google Scholar
    63. J. Liu, C. Wu, G. Wu, and X. Wang, “A novel differential search algorithm and applications for structure design,” Applied Mathematics and Computation, vol. 268, pp. 246–269, 2015.View at: Publisher Site | Google Scholar
    64. X. Zhao, D. Li, B. Yang, C. Ma, Y. Zhu, and H. Chen, “Feature selection based on improved ant colony optimization for online detection of foreign fiber in cotton,” Applied Soft Computing, vol. 24, pp. 585–596, 2014.View at: Publisher Site | Google Scholar
    65. D. Zhao, “Chaotic random spare ant colony optimization for multi-threshold image segmentation of 2D Kapur entropy,” Knowledge-Based Systems, vol. 24, p. 106510, 2020.View at: Google Scholar
    66. H. Chen, A. A. Heidari, H. Chen, M. Wang, Z. Pan, and A. H. Gandomi, “Multi-population differential evolution-assisted Harris hawks optimization: framework and case studies,” Future Generation Computer Systems, vol. 111, pp. 175–198, 2020.View at: Publisher Site | Google Scholar
    67. J. Hu, H. Chen, A. A. Heidari et al., “Orthogonal learning covariance matrix for defects of grey wolf optimizer: insights, balance, diversity, and feature selection,” Knowledge-Based Systems, vol. 213, p. 106684, 2021.View at: Publisher Site | Google Scholar
    68. G. Sun, B. Yang, Z. Yang, and G. Xu, “An adaptive differential evolution with combined strategy for global numerical optimization,” Soft Computing, vol. 24, pp. 1–20, 2019.View at: Google Scholar
    69. G. Sun, C. Li, and L. Deng, “An adaptive regeneration framework based on search space adjustment for differential evolution,” Neural Computing and Applications, vol. 24, pp. 1–17, 2021.View at: Google Scholar
    70. A. Addeh and M. Iri, “Brain tumor type classification using deep features of MRI images and optimized RBFNN,” ENG Transactions, vol. 2, pp. 1–7, 2021.View at: Google Scholar
    71. F. J. Golrokh and A. Hasan, “A comparison of machine learning clustering algorithms based on the DEA optimization approach for pharmaceutical companies in developing countries,” Soft Computing, vol. 1, pp. 1–8, 2020.View at: Google Scholar
    72. H. Tyagi, P. Phelan, and R. Prasher, “Predicted efficiency of a low-temperature nanofluid-based direct absorption solar collector,” Journal of Solar Energy Engineering, vol. 131, no. 4, 2009.View at: Publisher Site | Google Scholar
    73. S. Rashidi, M. Bovand, and J. A. Esfahani, “Heat transfer enhancement and pressure drop penalty in porous solar heat exchangers: a sensitivity analysis,” Energy Conversion and Management, vol. 103, pp. 726–738, 2015.View at: Publisher Site | Google Scholar
    74. N. Akram, R. Sadri, S. N. Kazi et al., “A comprehensive review on nanofluid operated solar flat plate collectors,” Journal of Thermal Analysis and Calorimetry, vol. 139, no. 2, pp. 1309–1343, 2020.View at: Publisher Site | Google Scholar
    Fig. 11. Velocity vectors along x-direction through the center of the box culvert for B0, B30, B50, and B70 respectively.

    Numerical investigation of scour characteristics downstream of blocked culverts

    막힌 암거 하류의 세굴 특성 수치 조사

    NesreenTahabMaged M.El-FekyaAtef A.El-SaiadaIsmailFathya
    aDepartment of Water and Water Structures Engineering, Faculty of Engineering, Zagazig University, Zagazig 44519, Egypt
    bLab Manager, Faculty of Engineering, Zagazig University, Zagazig 44519, Egypt

    Abstract

    횡단 구조물을 통한 막힘은 안정성을 위협하는 위험한 문제 중 하나입니다. 암거의 막힘 형상 및 하류 세굴 특성에 미치는 영향에 관한 연구는 거의 없습니다.

    이 연구의 목적은 수면과 세굴 모두에서 상자 암거를 통한 막힘의 작용을 수치적으로 논의하는 것입니다. 이를 위해 FLOW 3D v11.1.0을 사용하여 퇴적물 수송 모델을 조사했습니다.

    상자 암거를 통한 다양한 차단 비율이 연구되었습니다. FLOW 3D 모델은 실험 데이터로 보정되었습니다. 결과는 FLOW 3D 프로그램이 세굴 다운스트림 상자 암거를 정확하게 시뮬레이션할 수 있음을 나타냅니다.

    막힌 경우에 대한 속도 분포, 최대 세굴 깊이 및 수심을 플롯하고 비차단된 사례(기본 사례)와 비교했습니다.

    그 결과 암거 높이의 70% 차단율은 상류의 수심을 암거 높이의 2.3배 증가시키고 평균 유속은 기본 경우보다 3배 더 증가시키는 것으로 입증되었다. 막힘 비율의 함수로 상대 최대 세굴 깊이를 추정하는 방정식이 만들어졌습니다.

    Blockage through crossing structures is one of the dangerous problems that threaten its stability. There are few researches concerned with blockage shape in culverts and its effect on characteristics of scour downstream it.

    The study’s purpose is to discuss the action of blockage through box culvert on both water surface and scour numerically. A sediment transport model has been investigated for this purpose using FLOW 3D v11.1.0. Different ratios of blockage through box culvert have been studied. The FLOW 3D model was calibrated with experimental data.

    The results present that the FLOW 3D program was capable to simulate accurately the scour downstream box culvert. The velocity distribution, maximum scour depth and water depths for blocked cases have been plotted and compared with the non-blocked case (base case).

    The results proved that the blockage ratio 70% of culvert height makes the water depth upstream increases by 2.3 times of culvert height and mean velocity increases by 3 times more than in the base case. An equation has been created to estimate the relative maximum scour depth as a function of blockage ratio.

    1. Introduction

    Local scour is the removal of granular bed material by the action of hydrodynamic forces. As the depth of scour hole increases, the stability of the foundation of the structure may be endangered, with a consequent risk of damage and failure [1]. So the prediction and control of scour is considered to be very important for protecting the water structures from failure. Most previous studies were designed to study the different factors that impact on scour and their relationship with scour hole dimensions like fluid characteristics, flow conditions, bed properties, and culvert geometry. Many previous researches studied the effect of flow rate on scour hole by information Froude number or modified Froude number [2][3][4][5][6]. Cesar Mendoza [6] found a good correlation between the scour depth and the discharge Intensity (Qg−.5D−2.5). Breusers and Raudkiv [7] used shear velocity in the outlet-scour prediction procedure. Ali and Lim [8] used the densimetric Froude number in estimation of the scour depth [1][8][9][10][11][12][13][14]. “The densimetric Froude number presents the ratio of the tractive force on sediment particle to the submerged specific weight of the sediment” [15](1)Fd=uρsρ-1gD50

    Ali and Lim [8] pointed to the consequence of tailwater depth on scour behavior [1][2][8][13]. Abida and Townsend [2] indicated that the maximum depth of local scour downstream culvert was varying with the tailwater depth in three ways: first, for very shallow tailwater depths, local scouring decreases with a decrease in tailwater depth; second, when the ratio of tailwater depth to culvert height ranged between 0.2 and 0.7, the scour depth increases with decreasing tailwater depth; and third for a submerged outlet condition. The tailwater depth has only a marginal effect on the maximum depth of scour [2]. Ruff et al. [16] observed that for materials having similar mean grain sizes (d50) but different standard deviations (σ). As (σ) increased, the maximum scour hole depth decreased. Abt et al. [4] mentioned to role of soil type of maximum scour depth. It was noticed that local scour was more dangerous for uniform sands than for well-graded mixtures [1][2][4][9][17][18]. Abt et al [3][19] studied the culvert shape effect on scour hole. The results evidenced that the culvert shape has a limited effect on outlet scour. Under equivalent discharge conditions, it was noted that a square culvert with height equal to the diameter of a circular culvert would reduce scour [16][20]. The scour hole dimension was also effected by the culvert slope. Abt et al. [3][21] showed that the culvert slope is a key element in estimating the culvert flow velocity, the discharge capacity, and sediment transport capability. Abt et al. [21][22] tested experimentally culvert drop height effect on maximum scour depth. It was observed that as the drop height was increasing, the depth of scour was also increasing. From the previous studies, it could have noticed that the most scour prediction formula downstream unblocked culvert was the function of densimetric Froude number, soil properties (d50, σ), tailwater depth and culvert opening size. Blockage is the phenomenon of plugging water structures due to the movement of water flow loaded with sediment and debris. Water structures blockage has a bad effect on water flow where it causes increasing of upstream water level that may cause flooding around the structure and increase of scour rate downstream structures [23][24]. The blockage phenomenon through was studied experimentally and numerical [15][25][26][27][28][29][30][31][32][33]. Jaeger and Lucke [33] studied the debris transport behavior in a natural channel in Australia. Froude number scale model of an existing culvert was used. It was noticed that through rainfall event, the mobility of debris was impressed by stream shape (depth and width). The condition of the vegetation (size and quantities) through the catchment area was the main factor in debris transport. Rigby et al. [26] reported that steep slope was increasing the ability to mobilize debris that form field data of blocked culverts and bridges during a storm in Wollongong city.

    Streftaris et al. [32] studied the probability of screen blockage by debris at trash screens through a numerical model to relate between the blockage probability and nature of the area around. Recently, many commercial computational fluid programs (CFD) such as SSIIM, Fluent, and FLOW 3D are used in the analysis of the scour process. Scour and sediment transport numerical model need to validate by using experimental data or field data [34][35][36][37][38]. Epely-Chauvin et al. [36] investigated numerically the effect of a series of parallel spur diked. The experimental data were compared by SSIIM and FLOW 3D program. It was found that the accuracy of calibrated FLOW 3D model was better than SSIIM model. Nielsen et al. [35] used the physical model and FLOW 3D model to analyze the scour process around the pile. The soil around the pile was uniform coarse stones in the physical models that were simulated by regular spheres, porous media, and a mixture of them. The calibrated porous media model can be used to determine the bed shear stress. In partially blocked culverts, there aren’t many studies that explain the blockage impact on scour dimensions. Sorourian et al. [14][15] studied the effect of inlet partial blockage on scour characteristics downstream box culvert. It resulted that the partial blockage at the culvert inlet could be the main factor in estimating the depth of scour. So, this study is aiming to investigate the effects of blockage through a box culvert on flow and scour characteristics by different blockage ratios and compares the results with a non-blocked case. Create a dimensionless equation relates the blockage ratio of the culvert with scour characteristics downstream culvert.

    2. Experimental data

    The experimental work of the study was conducted in the Hydraulics and Water Engineering Laboratory, Faculty of Engineering, Zagazig University, Egypt. The flume had a rectangular cross-section of 66 cm width, 65.5 cm depth, and 16.2 m long. A rectangular culvert was built with 0.2 m width, 0.2 m height and 3.00 m long with θ = 25° gradually outlet and 0.8 m fixed apron. The model was located on the mid-point of the channel. The sediment part was extended for a distance 2.20 m with 0.66 m width and 0.20 m depth of coarse sand with specific weight 1.60 kg/cm3, d50 = 2.75 mm and σ (d90/d50) = 1.50. The particle size distribution was as shown in Fig. 1. The experimental model was tested for different inlet flow (Q) of 25, 30, 34, 40 l/s for different submerged ratio (S) of 1.25, 1.50, 1.75.

    3. Dimensional analysis

    A dimensional analysis has been used to reduce the number of variables which affecting on the scour pattern downstream partial blocked culvert. The main factors affecting the maximum scour depth are:(2)ds=f(b.h.L.hb.lb.Q.ud.hu.hd.D50.ρ.ρs.g.ls.dd.ld)

    Fig. 2 shows a definition sketch of the experimental model. The maximum scour depth can be written in a dimensionless form as:(3)dsh=f(B.Fd.S)where the ds/h is the relative maximum scour depth.

    4. Numerical work

    The FLOW 3D is (CFD) program used by many researchers and appeared high accuracy in solving hydrodynamic and sediment transport models in the three dimensions. Numerical simulation with FLOW 3D was performed to study the impacts of blockage ratio through box culvert on shear stress, velocity distribution and the sediment transport in terms of the hydrodynamic features (water surface, velocity and shear stress) and morphological parameters (scour depth and sizes) conditions in accurately and efficiently. The renormalization group (RNG) turbulence model was selected due to its high ability to predict the velocity profiles and turbulent kinetic energy for the flow through culvert [39]. The one-fluid incompressible mode was used to simulate the water surface. Volume of fluid (VOF) method was employed in FLOW 3D to tracks a liquid interface through arbitrary deformations and apply the correct boundary conditions at the interface [40].1.

    Governing equations

    Three-dimensional Reynolds-averaged Navier Stokes (RANS) equation was applied for incompressible viscous fluid motion. The continuity equation is as following:(4)VF∂ρ∂t+∂∂xρuAx+∂∂yρvAy+∂∂zρwAz=RDIF(5)∂u∂t+1VFuAx∂u∂x+vAy∂u∂y+ωAz∂u∂z=-1ρ∂P∂x+Gx+fx(6)∂v∂t+1VFuAx∂v∂x+vAy∂v∂y+ωAz∂v∂z=-1ρ∂P∂y+Gy+fy(7)∂ω∂t+1VFuAx∂ω∂x+vAy∂ω∂y+ωAz∂ω∂z=-1ρ∂P∂z+Gz+fz

    ρ is the fluid density,

    VF is the volume fraction,

    (x,y,z) is the Cartesian coordinates,

    (u,v,w) are the velocity components,

    (Ax,Ay,Az) are the area fractions and

    RDIF is the turbulent diffusion.

    P is the average hydrodynamic pressure,

    (Gx, Gy, Gz) are the body accelerations and

    (fx, fy, fz) are the viscous accelerations.

    The motion of sediment transport (suspended, settling, entrainment, bed load) is estimated by predicting the erosion, advection and deposition process as presented in [41].

    The critical shields parameter is (θcr) is defined as the critical shear stress τcr at which sediments begin to move on a flat and horizontal bed [41]:(8)θcr=τcrgd50(ρs-ρ)

    The Soulsby–Whitehouse [42] is used to predict the critical shields parameter as:(9)θcr=0.31+1.2d∗+0.0551-e(-0.02d∗)(10)d∗=d50g(Gs-1ν3where:

    d* is the dimensionless grain size

    Gs is specific weight (Gs = ρs/ρ)

    The entrainment coefficient (0.005) was used to scale the scour rates and fit the experimental data. The settling velocity controls the Soulsby deposition equation. The volumetric sediment transport rate per width of the bed is calculated using Van Rijn [43].2.

    Meshing and geometry of model

    After many trials, it was found that the uniform cell size with 0.03 m cell size is the closest to the experimental results and takes less time. As shown in Fig. 3. In x-direction, the total model length in this direction is 700 cm with mesh planes at −100, 0, 300, 380 and 600 cm respectively from the origin point, in y-direction, the total model length in this direction is 66 cm at distances 0, 23, 43 and 66 cm respectively from the origin point. In z-direction, the total model length in this direction is 120 cm. with mesh planes at −20, 0, 20 and 100 cm respectively.3.

    Boundary condition

    As shown in Fig. 4, the boundary conditions of the model have been defined to simulate the experimental flow conditions accurately. The upstream boundary was defined as the volume flow rate with a different flow rate. The downstream boundary was defined as specific pressure with different fluid elevation. Both of the right side, the left side, and the bottom boundary were defined as a wall. The top boundary defined as specified pressure with pressure value equals zero.

    5. Validation of experimental results and numerical results

    The experimental results investigated the flow and scour characteristics downstream culvert due to different flow conditions. The measured value of maximum scour depth is compared with the simulated depth from FLOW 3D model as shown in Fig. 5. The scour results show that the simulated results from the numerical model is quite close to the experimental results with an average error of 3.6%. The water depths in numerical model results is so close to the experimental results as shown in Fig. 6 where the experiment and numerical results are compared at different submerged ratios and flow rates. The results appear maximum error percentage in water depths upstream and downstream the culvert is about 2.37%. This indicated that the FLOW 3D is efficient for the prediction of maximum scour depth and the flow depths downstream box culvert.

    6. Computation time

    The run time was chosen according to reaching to the stability limit. Hydraulic stability was achieved after 50 s, where the scour development may still go on. For run 1, the numerical simulation was run for 1000 s as shown in Fig. 7 where it mostly reached to scour stability at 800 s. The simulation time was taken 500 s at about 95% of scour stability.

    7. Analysis and discussions

    Fig. 8 shows the study sections where sec 1 represents to upstream section, sec2 represents to inside section and sec3 represents to downstream stream section. Table 1 indicates the scour hole dimensions at different blockage case. The symbol (B) represents to blockage and the number points to blockage ratio. B0 case signifies to the non-blocked case, B30 is that blockage height is 30% to the culvert height and so on.

    Table 1. The scour results of different blockage ratio.

    Casehb cmB = hb/hQ lit/sSFdd50 mmds/h measuredls/hdd/hld/hds/h estimated
    B000351.261.692.50.581.500.275.000.46
    B3060.30351.261.682.50.481.250.274.250.40
    B50100.50351.221.742.50.451.100.244.000.37
    B70140.70351.231.732.50.431.500.165.500.33

    7.1. Scour hole geometry

    The scour hole geometry mainly depends on the properties of soil of the bed downstream the fixed apron. From Table 1, the results show that the maximum scour depth in B0 case is about 0.58 of culvert height while the maximum deposition in B0 is 0.27 culvert height. There is a symmetric scour hole as shown in Fig. 9 in B0 case. An asymmetric scour hole is created in B50 and B70 due to turbulences that causes the deviation of the jet direction from the center of the flume where appear in Fig. 11 and Fig. 19.

    7.2. Flow water surface

    Fig. 10 presents the relative free surface water (hw/h) along the x-direction at center of the box culvert. From the mention Figure, it is easy to release the effect of different blockage ratios. The upstream water level rises by increasing the blockage ratio. Increasing upstream water level may cause flooding over the banks of the waterway. In the 70% blockage case, the upstream water level rises to 2.3 times of culvert height more than the non-blocked case at the same discharge and submerged ratio. The water surface profile shows an increase in water level upstream the culvert due to a decrease in transverse velocity. Because of decreasing velocity downstream culvert, there is an increase in water level before it reaches its uniform depth.

    7.3. Velocity vectors

    Scour downstream hydraulic structures mainly affects by velocities distribution and bed shear stress. Fig. 11 shows the velocity vectors and their magnitude in xz plane at the same flow conditions. The difference in the upstream water level due to the different blockage ratios is so clear. The maximum water level is in B70 and the minimum level is in B0. The inlet mean velocity value is about 0.88 m/s in B0 increases to 2.86 m/s in B70. As the blockage ratio increases, the inlet velocity increases. The outlet velocity in B0 case makes downward jet causes scour hole just after the fixed apron in the middle of the bed while the blockage causes upward water flow that appears clearly in B70. The upward jet decreases the scour depth to 0.13 culvert height less than B0 case. After the scour hole, the velocity decreases and the flow becomes uniform.

    7.4. Velocity distribution

    Fig. 12 represents flow velocity (Vx) distribution along the vertical depth (z/hu) upstream the inlet for the different blockage ratios at the same flow conditions. From the Figure, the maximum velocity creates closed to bed in B0 while in blocked case, the maximum horizontal velocity creates at 0.30 of relative vertical depth (z/hu). Fig. 13 shows the (Vz) distribution along the vertical depth (z/hu) upstream culvert at sec 1. From the mentioned Figure, it is easy to note that the maximum vertical is in B70 which appears that as the blockage ratio increases the vertical ratio also increases. In the non-blocked case. The vertical velocity (Vz) is maximum at (z/hu) equals 0.64. At the end of the fixed apron (sec 3), the horizontal velocity (Vx) is slowly increasing to reach the maximum value closed to bed in B0 and B30 while the maximum horizontal velocity occurs near to the top surface in B50 and B70 as shown in Fig. 14. The vertical velocity component along the vertical depth (z/hd) is presented in Fig. 15. The vertical velocity (Vz) is maximum in B0 at vertical depth (z/hd) 0.3 with value 0.45 m/s downward. Figs. 16 and 17 observe velocity components (Vx, Vz) along the vertical depth just after the end of blockage length at the centerline of the culvert barrel. It could be noticed the uniform velocity distribution in B0 case with horizontal velocity (Vx) closed to 1.0 m/s and vertical velocity closed to zero. In the blocked case, the maximum horizontal velocity occurs in depth more than the blockage height.

    7.5. Bed velocity distribution

    Fig. 18 presents the x-velocity vectors at 1.5 cm above the bed for different blockage ratios from the velocity vectors distribution and magnitude, it is easy to realize the position of the scour hole and deposition region. In B0 and B30, the flow is symmetric so that the scour hole is created around the centerline of flow while in B50 and B70 cases, the flow is asymmetric and the scour hole creates in the right of flow direction in B50. The maximum scour depth is found in the left of flow direction in B70 case where the high velocity region is found.

    8. Maximum scour depth prediction

    Regression analysis is used to estimate maximum scour depth downstream box culvert for different ratios of blockage by correlating the maximum relative scour by other variables that affect on it in one formula. An equation is developed to predict maximum scour depth for blocked and non-blocked. As shown in the equation below, the relative maximum scour depth(ds/hd) is a function of densimetric Froude number (Fd), blockage ratio (B) and submerged ratio (S)(11)dsh=0.56Fd-0.20B+0.45S-1.05

    In this equation the coefficient of correlation (R2) is 0.82 with standard error equals 0·08. The developed equation is valid for Fd = [0.9 to 2.10] and submerged ratio (S) ≥ 1.00. Fig. 19 shows the comparison between relative maximum scour depths (ds/h) measured and estimated for different blockage ratios. Fig. 20 clears the comparison between residuals and ds/h estimated for the present study. From these figures, it could be noticed that there is a good agreement between the measured and estimated relative scour depth.

    9. Comparison with previous scour equations

    Many previous scour formulae have been produced for calculation the maximum scour depth downstream non-blockage culvert. These equations have been included the effect of flow regime, culvert shape, soil properties and the flow rate on maximum scour depth. Two of previous experimental studies data have been chosen to be compared with the present study results in non-blocked study data. Table 2 shows comparison of culvert shape, densmetric Froude number, median particle size and scour equations for these previous studies. By applying the present study data in these studies scour formula as shown in Fig. 21, it could be noticed that there are a good agreement between present formula results and others empirical equations results. Where that Lim [44] and Abt [4] are so closed to the present study data.

    Table 2. Comparison of some previous scour formula.

    ResearchersFdCulvert shaped50(mm)Proposed equationSubmerged ratio
    Present study0.9–2.11square2.75dsh=0.56Fd-0.20B+0.45S-1.051.25–1.75
    Lim [44]1–10Circular1.65dsh=0.45Fd0.47
    Abt [4]Fd ≥ 1Circular0.22–7.34-dsh=3.67Fd0.57∗D500.4∗σ-0.4

    10. Conclusions

    The present study has shown that the FLOW 3D model can accurately simulate water surface and the scour hole characteristics downstream the box culvert with error percentage in water depths does not exceed 2.37%. Velocities distribution through and outlets culvert barrel helped on understanding the scour hole shape.

    The blockage through culvert had caused of increasing of water surface upstream structure where the upstream water level in B70 was 2.3 of culvert height more than non-blocked case at the same discharge that could be dangerous on the stability of roads above. The depth averaged velocity through culvert barrel increased by 3 times its value in non-blocked case.

    On the other hand, blockage through culvert had a limited effect on the maximum scour depth. The little effect of blockage on maximum scour depth could be noticed in Fig. 11. From this Figure, it could be noted that the residual part of culvert barrel after the blockage part had made turbulences. These turbulences caused the deviation of the flow resulting in the formation of asymmetric scour hole on the side of channel. This not only but in B70 the blockage height caused upward jet which made a wide far scour hole as cleared from the results in Table 1.

    An empirical equation was developed from the results to estimate the maximum scour depth relative to culvert height function of blockage ratio (B), submerged ratio (S), and densimetric Froude number (Fd). The equation results was compared with some scour formulas at the same densimetric Froude number rang where the present study results was in between the other equations results as shown in Fig. 21.

    Declaration of Competing Interest

    The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

    References

    [1]P. Sarathi, M. Faruque, R. BalachandarInfluence of tailwater depth, sediment size and densimetric Froude number on scour by submerged square wall jetsJ. Hydraul. Res., 46 (2) (2008), pp. 158-175CrossRefView Record in ScopusGoogle Scholar[2]H. Abida, R. TownsendLocal scour downstream of box-culvert outletsJ. Irrig. Drain. Eng., 117 (3) (1991), pp. 425-440CrossRefView Record in ScopusGoogle Scholar[3]S.R. Abt, C.A. Donnell, J.F. Ruff, F.K. DoehringCulvert Slope and Shape Effects on Outlet ScourTransp. Res. Rec., 1017 (1985), pp. 24-30View Record in ScopusGoogle Scholar[4]S.R. Abt, R.L. Kloberdanz, C. MendozaUnified culvert scour determinationJ. Hydraul. Eng., 110 (10) (1984), pp. 1475-1479CrossRefView Record in ScopusGoogle Scholar[5]J.P. Bohan, Erosion And Riprap Requirements At Culvert And Storm-Drain Outlets, ARMY ENGINEER WATERWAYS EXPERIMENT STATION VICKSBURG MISS1970.Google Scholar[6]C. Mendoza, S.R. Abt, J.F. RuffHeadwall influence on scour at culvert outletsJ. Hydraul. Eng., 109 (7) (1983), pp. 1056-1060CrossRefView Record in ScopusGoogle Scholar[7]H. Breusers, A. Raudkivi, Scouring, hydraulic structures design manual, vol. 143, IAHR, AA Balkema, Rotterdam, 1991.Google Scholar[8]K. Ali, S. LimLocal scour caused by submerged wall jetsProc. Inst. Civ. Eng., 81 (4) (1986), pp. 607-645CrossRefView Record in ScopusGoogle Scholar[9]O. Aderibigbe, N. RajaratnamEffect of sediment gradation on erosion by plane turbulent wall jetsJ. Hydraul. Eng., 124 (10) (1998), pp. 1034-1042View Record in ScopusGoogle Scholar[10]F.W. Blaisdell, C.L. AndersonA comprehensive generalized study of scour at cantilevered pipe outletsJ. Hydraul. Res., 26 (4) (1988), pp. 357-376CrossRefView Record in ScopusGoogle Scholar[11]Y.-M. Chiew, S.-Y. LimLocal scour by a deeply submerged horizontal circular jetJ. Hydraul. Eng., 122 (9) (1996), pp. 529-532View Record in ScopusGoogle Scholar[12]R.A. Day, S.L. Liriano, W.R. WhiteEffect of tailwater depth and model scale on scour at culvert outletsProc. Instit. Civil Eng. – Water Marit. Eng., 148 (3) (2001), pp. 189-198http://www.icevirtuallibrary.com/doi/10.1680/wame.2001.148.3.18910.1680/wame.2001.148.3.189View Record in ScopusGoogle Scholar[13]S. Emami, A.J. SchleissPrediction of localized scour hole on natural mobile bed at culvert outletsScour and Erosion (2010), pp. 844-853CrossRefView Record in ScopusGoogle Scholar[14]S. Sorourian, A. Keshavarzi, J. Ball, B. SamaliStudy of Blockage Effect on Scouring Pattern Downstream of a Box Culvert under Unsteady FlowAustr. J Water Resor. (2013)Google Scholar[15]S. Sorourian, Turbulent Flow Characteristics At The Outlet Of Partially Blocked Box Culverts, in: 36th IAHR World Congress, The Hague, the Netherlands, 2015.Google Scholar[16]J. Ruff, S. Abt, C. Mendoza, A. Shaikh, R. KloberdanzScour at culvert outlets in mixed bed materialsUnited States. Federal Highway Administration. Office of Research and Development (1982)Google Scholar[17]S.A. Ansari, U.C. Kothyari, K.G.R. RajuInfluence of cohesion on scour under submerged circular vertical jetsJ. Hydraul. Eng., 129 (12) (2003), pp. 1014-1019View Record in ScopusGoogle Scholar[18]B. Crookston B. Tullis, Scour and Riprap Protection in a Bottomless Arch Culvert, in: World Environmental and Water Resources Congress 2008: Ahupua’A, 2008, pp. 1–10.Google Scholar[19]S.R. Abt, J. Ruff, F. Doehring, C. DonnellInfluence of culvert shape on outlet scourJ. Hydraul. Eng., 113 (3) (1987), pp. 393-400View Record in ScopusGoogle Scholar[20]Y.H. Chen, Scour at outlets of box culverts, Colorado State University, 1970.Google Scholar[21]S. Abt, P. Thompson, T. LewisEnhancement of the culvert outlet scour estimation equationsTransp. Res. Rec. J. Transp. Res. Board, 1523 (1996), pp. 178-185View Record in ScopusGoogle Scholar[22]F.K. Doehring, S.R. AbtDrop height influence on outlet scourJ. Hydraul. Eng., 120 (12) (1994), pp. 1470-1476CrossRefView Record in ScopusGoogle Scholar[23]W. Weeks, A. Barthelmess, E. Rigby, G. Witheridge, R. Adamson, Australian rainfall and runoff revison project 11: blockage of hydraulic structures, 2009.Google Scholar[24]W. Weeks, G. Witheridge, E. Rigby, A. BarthelmessProject 11: blockage of hydraulic structuresEngineers Australia (2013)Google Scholar[25]S.R. Abt, T.E. Brisbane, D.M. Frick, C.A. McKnightTrash rack blockage in supercritical flowJ. Hydraul. Eng., 118 (12) (1992), pp. 1692-1696View Record in ScopusGoogle Scholar[26]E. Rigby, M. Boyd, S. Roso, P. Silveri, A. Davis, Causes and effects of culvert blockage during large storms, in: Global solutions for urban drainage, 2002, pp. 1–16.Google Scholar[27]S. Roso, M. Boyd, E. Rigby, R. VanDrie“Prediction of increased flooding in urban catchments due to debris blockage and flow diversionsProceedings Novatech (2004), pp. 8-13View Record in ScopusGoogle Scholar[28]C.-D. Jan, C.-L. ChenDebris flows caused by Typhoon Herb in Taiwanin Debris-Flow Hazards and Related Phenomena, Springer (2005), pp. 539-563CrossRefGoogle Scholar[29]L.W. Zevenbergen, P.F. Lagasse, P.E. Clopper, Effects of debris on bridge pier scour, in: World Environmental and Water Resources Congress 2007: Restoring Our Natural Habitat, 2007, pp. 1–10.Google Scholar[30]A. Barthelmess, E. Rigby, Estimating Culvert and Bridge Blockages-a Simplified Procedure, in: Proceedings of the 34th World Congress of the International Association for Hydro-Environment Research and Engineering: 33rd Hydrology and Water Resources Symposium and 10th Conference on Hydraulics in Water Engineering, Engineers Australia, 2011, pp. 39.Google Scholar[31]E. Rigby, A. Barthelmess, Culvert Blockage Mechanisms and their Impact on Flood Behaviour, in: Proceedings of the 34th World Congress of the International Association for Hydro-Environment Research and Engineering: 33rd Hydrology and Water Resources Symposium and 10th Conference on Hydraulics in Water Engineering, Engineers Australia, 2011, pp. 380.Google Scholar[32]G. Streftaris, N. Wallerstein, G. Gibson, S. ArthurModeling probability of blockage at culvert trash screens using Bayesian approachJ. Hydraul. Eng., 139 (7) (2012), pp. 716-726Google Scholar[33]R. Jaeger, T. LuckeInvestigating the relationship between rainfall intensity, catchment vegetation and debris mobilityInt. J. GEOMATE, 12 (33) (2017), pp. 22-29 Download PDFView Record in ScopusGoogle Scholar[34]S. Amiraslani, J. Fahimi, H. Mehdinezhad, The Numerical Investigation of Free Falling Jet’s Effect On the Scour of Plunge Pool, in: XVIII International conference on water resources, Tehran University, Iran, 2008.Google Scholar[35]A.W. Nielsen, X. Liu, B.M. Sumer, J. FredsøeFlow and bed shear stresses in scour protections around a pile in a currentCoast. Eng., 72 (2013), pp. 20-38ArticleDownload PDFView Record in ScopusGoogle Scholar[36]G. Epely-Chauvin, G. De Cesare, S. SchwindtNumerical modelling of plunge pool scour evolution in non-cohesive sedimentsEng. Appl. Comput. Fluid Mech., 8 (4) (2014), pp. 477-487 Download PDFCrossRefView Record in ScopusGoogle Scholar[37]H. Karami, H. Basser, A. Ardeshir, S.H. HosseiniVerification of numerical study of scour around spur dikes using experimental dataWater Environ. J., 28 (1) (2014), pp. 124-134CrossRefView Record in ScopusGoogle Scholar[38]S.-H. Oh, K.S. Lee, W.-M. JeongThree-dimensional experiment and numerical simulation of the discharge performance of sluice passageway for tidal power plantRenew. Energy, 92 (2016), pp. 462-473ArticleDownload PDFView Record in ScopusGoogle Scholar[39]M.A. Khodier, B.P. TullisExperimental and computational comparison of baffled-culvert hydrodynamics for fish passageJ. Appl. Water Eng. Res. (2017), pp. 1-9CrossRefView Record in ScopusGoogle Scholar[40]F.S. Inc., FLOW-3D user’s manual, Flow Science, Inc., 2009.Google Scholar[41]G. Wei, J. Brethour, M. Grünzner, J. BurnhamSedimentation scour modelFlow Science Report, 7 (2014), pp. 1-29View Record in ScopusGoogle Scholar[42]R. Soulsby, R. Whitehouse, Threshold of sediment motion in coastal environments, in: Pacific Coasts and Ports’ 97: Proceedings of the 13th Australasian Coastal and Ocean Engineering Conference and the 6th Australasian Port and Harbour Conference, vol. 1, Centre for Advanced Engineering, University of Canterbury, 1997, pp. 145.Google Scholar[43]L.C.v. RijnSediment transport, part II: suspended load transportJ. Hydraul. Eng., 110 (11) (1984), pp. 1613-1641View Record in ScopusGoogle Scholar[44]S Y LIMScour below unsubmerged full-flowing culvert outletsProc. Instit. Civil Eng. – Water Marit. Energy, 112 (2) (1995), pp. 136-149http://www.icevirtuallibrary.com/doi/10.1680/iwtme.1995.2765910.1680/iwtme.1995.27659View Record in ScopusGoogle Scholar

    Peer review under responsibility of Faculty of Engineering, Alexandria University.

    Figure 1- Schematic diagram of pooled stepped spillway conducted by Felder et al. (2012A): Notes: h step height (10 cm): w pool height (3.1 cm): l horizontal step length (20 cm): lw pool weir length (1.5 cm): d' is the water depth above the crest; y' is the distance normal to the crest invert

    Study of inception point, void fraction and pressure over pooled stepped spillways using Flow-3D

    Khosro Morovati , Afshin Eghbalzadeh 
    International Journal of Numerical Methods for Heat & Fluid Flow

    ISSN: 0961-5539

    Article publication date: 3 April 2018

    Abstract

    많은 계단식 배수로 지오메트리 설계 지침이 평평한 단계를 위해 개발되었지만 통합 단계를 설계하는 것이 더 효율적으로 작동하는 배수로에 대한 적절한 대안이 될 수 있습니다.

    이 논문은 POOL의 다른 높이에서 공기 연행과 보이드 비율의 시작점을 다루는 것을 목표로 합니다. 그 후, FLOW-3D 소프트웨어를 사용하여 POOL과 경사면의 높이를 다르게 하여 폭기된 지역과 폭기되지 않은 지역에서 압력 분포를 평가했습니다.

    얻어진 수치 결과와 실험 결과의 비교는 본 연구에 사용된 모든 방류에 대해 잘 일치했습니다. POOL 높이는 시작 지점 위치에 미미한 영향을 미쳤습니다. 공극률의 값은 높은 방류에 비해 낮은 방전에서 더 많은 영향을 받았습니다.

    여수로의 마루(통기되지 않은 지역)에서는 음압이 나타나지 않았으며 각 방류에서 마루를 따라 높이가 15cm인 수영장에서 최대 압력 값이 얻어졌습니다.

    모든 사면에서 웅덩이 및 평평한 계단형 여수로의 계단층 부근에서는 음압이 형성되지 않았습니다. 그러나 평단식 여수로에 비해 평단식 여수로의 수직면 부근에서 음압이 더 많이 형성되어 평단식 슈트에서 캐비테이션 현상이 발생할 확률이 증가하였습니다.

    Study of inception point, void fraction and pressure over pooled
    stWhile many stepped spillways geometry design guidelines were developed for flat steps, designing pooled steps might be an appropriate alternative to spillways working more efficiency. This paper aims to deal with the inception point of air-entrainment and void fraction in the different height of the pools. Following that, pressure distribution was evaluated in aerated and non-aerated regions under the effect of different heights of the pools and slopes through the use of the FLOW-3D software. Comparison of obtained numerical results with experimental ones was in good agreement for all discharges used in this study. Pools height had the insignificant effect on the inception point location. The value of void fraction was more affected in lower discharges in comparison with higher ones. Negative pressure was not seen over the crest of spillway (non-aerated region), and the maximum pressure values were obtained for pools with 15 cm height along the crest in each discharge. In all slopes, negative pressure was not formed near the step bed in the pooled and flat stepped spillways. However, negative pressure was formed in more area near the vertical face in the flat stepped spillway compared with the pooled stepped spillway which increases the probability of cavitation phenomenon in the flat stepped chute.

    Design/methodology/approach

    압력, 공극률 및 시작점을 평가하기 위해 POOL된 계단식 여수로가 사용되었습니다. 또한 POOL의 다른 높이가 사용되었습니다. 이 연구의 수치 시뮬레이션은 Flow-3D 소프트웨어를 통해 수행되었습니다. 얻어진 결과는 풀이 압력, 공극률 및 시작점을 포함한 2상 유동 특성에 영향을 미칠 수 있음을 나타냅니다.

    Findings

    마루 위에는 음압이 보이지 않았습니다. 압력 값은 사용된 모든 높이와 15cm 높이에서 얻은 최대 값에 대해 다릅니다. 또한, 풀링 스텝은 플랫 케이스에 비해 음압점 감소에 더 효과적인 역할을 하였습니다. 시작 지점 위치는 특히 9 및 15cm 높이에 대해 스키밍 흐름 영역과 비교하여 낮잠 및 전환 흐름 영역에서 더 많은 영향을 받았습니다.

    Keywords

    Citation

    Morovati, K. and Eghbalzadeh, A. (2018), “Study of inception point, void fraction and pressure over pooled stepped spillways using Flow-3D”, International Journal of Numerical Methods for Heat & Fluid Flow, Vol. 28 No. 4, pp. 982-998. https://doi.org/10.1108/HFF-03-2017-0112

    Figure 1- Schematic diagram of pooled stepped spillway conducted by Felder et al. (2012A): Notes: h  step height (10 cm): w pool height (3.1 cm): l horizontal step length (20 cm): lw pool weir length (1.5 cm):  d' is the water depth above the crest; y' is the distance normal to the crest invert
    Figure 1- Schematic diagram of pooled stepped spillway conducted by Felder et al. (2012A): Notes: h step height (10 cm): w pool height (3.1 cm): l horizontal step length (20 cm): lw pool weir length (1.5 cm): d’ is the water depth above the crest; y’ is the distance normal to the crest invert
    Figure 2- meshing domain and distribution of blocks
    Figure 2- meshing domain and distribution of blocks
    Figure 3- Comparison of numerical simulation with experimental data by Felder et al. (2012A);  mesh convergence analysis; pooled stepped spillway (slope: 26.6 0 )
    Figure 3- Comparison of numerical simulation with experimental data by Felder et al. (2012A); mesh convergence analysis; pooled stepped spillway (slope: 26.6 0 )
    Figure 4- Comparison of numerical simulation with experimental data by Felder et al. (2012A);  Flat stepped spillway (slope: 0 26 6. )
    Figure 4- Comparison of numerical simulation with experimental data by Felder et al. (2012A); Flat stepped spillway (slope: 0 26 6. )
    Figure 5-Comparison of numerical simulation with experimental data by Felder et al. (2012B); pooled  and flat stepped spillways (slope: 0 9.8 )
    Figure 5-Comparison of numerical simulation with experimental data by Felder et al. (2012B); pooled and flat stepped spillways (slope: 0 9.8 )
    Figure 6- TKE distribution on steps 8, 9 and 10 for four different mesh numbers: 261252 (model 1),  288941 (model 2), 323578 (model 3) and 343154 (model 4)
    Figure 6- TKE distribution on steps 8, 9 and 10 for four different mesh numbers: 261252 (model 1), 288941 (model 2), 323578 (model 3) and 343154 (model 4)
    Figure 7- Comparison of obtained Void fraction distribution on step 10 in numerical simulation with  experimental work conducted by Felder et al. (2012A); (slope 26.60 )
    Figure 7- Comparison of obtained Void fraction distribution on step 10 in numerical simulation with experimental work conducted by Felder et al. (2012A); (slope 26.60 )
    Figure 8- Results of inception point of air entrainment in different height of the pools: comparison with  empirical correlations (Eqs 8-9), experimental (Felder et al. (2012A)) and numerical data
    Figure 8- Results of inception point of air entrainment in different height of the pools: comparison with empirical correlations (Eqs 8-9), experimental (Felder et al. (2012A)) and numerical data
    Figure 9- Void fraction distribution for different pool heights on steps 10; slope 26.6 0
    Figure 9- Void fraction distribution for different pool heights on steps 10; slope 26.6 0
    Figure 10- Comparison of pressure distribution between numerical simulation and experimental work  conducted by Zhang and Chanson (2016); flat stepped spillway (slope: 0 45 )
    Figure 10- Comparison of pressure distribution between numerical simulation and experimental work conducted by Zhang and Chanson (2016); flat stepped spillway (slope: 0 45 )
    Figure 11- A comparison of the pressure distribution above the crest of the spillway; B comparison of the  free surface profile along the crest of the spillway.  Note: x' indicates the longitudinal distance from the starting point of the crest.
    Figure 11- A comparison of the pressure distribution above the crest of the spillway; B comparison of the free surface profile along the crest of the spillway. Note: x’ indicates the longitudinal distance from the starting point of the crest.
    Figure 12- pressure distribution along crest of spillway in different discharges; slope 26.6
    Figure 12- pressure distribution along crest of spillway in different discharges; slope 26.6
    Figure 13- Pressure distribution near the last step bed for different slopes and discharges: x'' indicatesthe  longitudinal distance from the intersection of the horizontal and vertical faces of step 10; y" is the distance from the intersection of the horizontal and vertical faces in the vertical direction
    Figure 13- Pressure distribution near the last step bed for different slopes and discharges: x” indicatesthe longitudinal distance from the intersection of the horizontal and vertical faces of step 10; y” is the distance from the intersection of the horizontal and vertical faces in the vertical direction
    Figure 14- Pressure distribution adjacent the vertical face of step 9 for different discharges and slopes
    Figure 14- Pressure distribution adjacent the vertical face of step 9 for different discharges and slopes
    Table1- Used discharges for assessments of mesh convergence analysis and hydraulic  characteristics
    Table1- Used discharges for assessments of mesh convergence analysis and hydraulic characteristics

    Conclusion

    본 연구에서는 자유표면을 모사하기 위해 VOF 방법과 k -ε (RNG) 난류 모델을 활용하여 FLOW-3D 소프트웨어를 사용하였고, 계단식 배수로의 유동을 모사하기 위한 목적으로 난류 특성을 모사하였다. 얻은 결과는 수치 모델이 시작점 위치, 보이드 비율 및 압력을 적절하게 시뮬레이션했음을 나타냅니다. 풀의 높이는 공기 유입 위치에 미미한 영향을 미치므로 얻은 결과는 이 문서에서 제시된 상관 관계와 잘 일치했습니다. 즉, 사용 가능한 상관 관계를 서로 다른 풀 높이에 사용할 수 있습니다. 공극률의 결과는 스텝 풀 근처의 나프 유동 영역에서 공극율 값이 다른 배출보다 더 큰 것으로 나타났다. 더욱이 고방출량 .0 113m3/s에서 수영장 높이를 변경해도 수영장 표면 근처의 공극률 값에는 영향을 미치지 않았습니다.

    낮잠 및 전환 체제의 압력 분포에 대한 0 및 3cm 높이의 수영장 효과는 많은 지점에서 대부분 유사했습니다. 더욱이 조사된 모든 높이에서 여수로의 마루를 따라 부압이 없었습니다. 여수로 끝단의 바닥 부근의 압력 결과는 평평하고 고인 경우 부압이 발생하지 않았음을 나타냅니다. 수직면 부근의 음압은 웅덩이에 비해 평평한 계단형 여수로의 깊이(w=0 cm)의 대부분에서 발생하였다. 또한 더 큰 사면에 대한 풀링 케이스에서 음압이 제거되었습니다. 평단식 여수로에서는 계단의 수직면에 인접한 더 넓은 지역에서 음압이 발생하였기 때문에 이 여수로에서는 고형단식여수로보다 캐비테이션 현상이 발생할 가능성이 더 큽니다.

    In this study, the FLOW-3D software was used through utilizing the VOF method and k −ε (RNG) turbulence model in order to simulate free surface, and turbulence characteristics for the purpose of simulating flow over pooled stepped spillway. The results obtained indicated that the numerical model properly simulated the inception point location, void fraction, and pressure. The height of the pools has the insignificant effect on the location of air entrainment, so that obtained results were in good agreement with the correlations presented in this paper. In other words, available correlations can be used for different pool heights. The results of void fraction showed that the void fraction values in nappe flow regime near the step pool were more than the other discharges. Furthermore in high discharge, 0.113m3/s, altering pool height had no effect on the value of void fraction near the pool surface.

    The effect of the pools with 0 and 3 cm heights over the pressure distribution in nappe and transition regimes was mostly similar in many points. Furthermore, in all examined heights there was no negative pressure along the crest of the spillway. The pressure results near the bed of the step at the end of the spillway indicated that negative pressure did not occur in the flat and pooled cases. Negative pressure near the vertical face occurred in the most part of the depth in the flat stepped spillway (w=0 cm) in comparison with the pooled case. Also, the negative pressure was eliminated in the pooled case for the larger slopes. Since negative pressure occurred in a larger area adjacent the vertical face of the steps in the flat stepped spillways, it is more likely that cavitation phenomenon occurs in this spillway rather than the pooled stepped spillways.

    References

    1. André, S. (2004), “High velocity aerated flows on stepped chutes with macro-roughness elements.” Ph.D. thesis,
      Laboratoire de Constructions Hydraulics (LCH), EPFL, Lausanne, Switzerland, 272 pages.
    2. Attarian, A. Hosseini, Kh. Abdi, H and Hosseini, M. (2014), “The Effect of the Step Height on Energy
      Dissipation in Stepped Spillways Using Numerical Simulation”. Arabian Journal for Science and
      Engineering, 39(4), 2587-2594.
    3. Bombardelli, F.A. Meireles. I. Matos, J. (2011), “Laboratory measurements and multi-block numerical
      simulations of the mean flow and turbulence in the non-aerated skimming flow region of steep stepped
      spillways”. Environmental fluid mechanics, 11(3) 263-288.
    4. Chakib, B. (2013), “Numerical Computation of Inception Point Location for Flat-sloped Stepped Spillway”.
      International Journal of Hydraulic Engineering; 2(3): 47-52.
    5. Chakib, B. Mohammed, H. (2015), “Numerical Simulation of Air Entrainment for Flat-Sloped Stepped Spillway.
      Journal of computational multiphase flows”, Volume 7. Number 1.
    6. Chanson, H. Toombes, L. (2002), “Air–water flows down stepped chutes: turbulence and flow structure
      observations”. International Journal of Multiphase Flow, 28(11) 1737-1761
    7. Chen, Q. Dai, G. Liu, H. (2002), “Volume of Fluid Model for Turbulence Numerical Simulation
      of Stepped Spillway Overflow”. DOI: 10.1061/(ASCE)0733-9429128:7(683).
    8. Cheng, X. Chen, Y. Luo, L. (2006), “Numerical simulation of air-water two-phase flow over stepped spillways”.
      Science in China Series E: Technological Sciences, 49(6), 674-684.
    9. Cheng, X. Luo, L. Zhao, W. (2004), “Study of aeration in the water flow over stepped spillway”. In: Proceedings
      of the world water congress.
    10. Chinnarasri, Ch. Kositgittiwong, D. Julien, Y. (2013), “Model of flow over spillways by computational fluid
      dynamics”. Proceedings of the ICE – Water Management, Volume 167(3) 164 –175.
    11. Dastgheib, A. Niksokhan, M.H. and Nowroozpour, A.R. (2012), “Comparing of Flow Pattern and Energy
      Dissipation over different forms of Stepped Spillway”. World Environmental and Water Resources
      Congress ASCE.
    12. Eghbalzadeh, A. Javan, M. (2012), “Comparison of mixture and VOF models for numerical simulation of air
      entrainment in skimming flow over stepped spillway”. Procedia Engineering, 28. 657-660.
    13. Felder, S, Chanson, H. (2012), “Free-surface Profiles, Velocity and Pressure Distributions on a
      Broad-Crested Weir: a Physical study “Free-surface Profiles, Velocity and Pressure Distributions on a
      Broad-Crested Weir: a Physical study
    14. Felder, S. Fromm, Ch. Chanson, H. (2012B), “Air entrainment and energy dissipation on a 8.9 slope stepped
      spillway with flat and pooled steps”, School of Civil Engineering, The University of Queensland,.
      Brisbane, Australia.
    15. Felder, S. Chanson, H. (2014A), Triple decomposition technique in air–water flows: application to instationary
      flows on a stepped spillway. International Journal of Multiphase Flow, 58, 139-153.
    16. Felder, S. Chanson, H. (2014B), Effects of step pool porosity upon flow aeration and energy dissipation on
      pooled stepped spillways. Journal of Hydraulic Engineering, 140(4), 04014002.
    17. Felder, S. Chanson, H. (2013A), “Air entrainment and energy dissipation on porous pooled stepped spillways”.
      Paper presented at the International Workshop on Hydraulic Design of Low-Head Structures.
    18. Felder, S. Chanson, H. (2013B), “Aeration, flow instabilities, and residual energy on pooled stepped spillways of
      embankment dams”. Journal of irrigation and drainage engineering, 139(10) 880-887.
    19. Felder, S. Guenther, Ph. Chanson, H. (2012A). “Air-water flow properties and energy dissipation on stepped
      spillways: a physical study of several pooled stepped configurations”, School of Civil Engineering, The
      University of Queensland,. Brisbane, Australia.
    20. Flow Science, (2013). “FLOW-3D user’s manual”, version 10.1. Flow Science, Inc, Los Alamos.
    21. Frizell, K.W. Renna, F.M. Matos, J. (2012), “Cavitation potential of flow on stepped spillways”. Journal of
      Hydraulic Engineering, 139(6), 630-636.
    22. Gonzalez, C. (2005), “An experimental study of free-surface aeration on embankment stepped chutes”,
      department of civil engineering, Brisbane, Australia, Phd thesis.
    23. Gonzalez, C.A. Chanson, H. (2008), “Turbulence manipulation in air–water flows on a stepped chute: An
      experimental study”. European Journal of Mechanics-B/Fluids, 27(4), 388-408.
    24. Guenther, Ph.. Felder, S. Chanson, H. (2013), “Flow aeration, cavity processes and energy dissipation on flat and
      pooled stepped spillways for embankments”. Environmental fluid mechanics, 13(5) 503-525.
    25. Hamedi, A. Mansoori, A. Malekmohamadi, I. Roshanaei, H. (2011), “Estimating Energy Dissipation in Stepped
      Spillways with Reverse Inclined Steps and End Sill”. World Environmental and Water Resources
      Congress, ASCE.
    26. Hirt, C.W. (2003), “Modeling Turbulent Entrainment of Air at a Free Surface”. Flow Science Inc.
    27. Hunt, S.L. Kadavy, K.C. (2013), “Inception point for enbankment dam stepped spillway”. J. Hydraul. Eng.,
      139(1), 60–64.
    28. Hunt, S.L. Kadavy, K.C. (2010), “Inception Point Relationship for Flat-Sloped Stepped
      Spillways”. DOI: 10.1061/ASCEHY.1943-7900.0000297.
    29. Matos, J. Quintela, A. (2000), “Air entrainment and safety against cavitation damage in stepped spillways over
      RCC dams. In: Proceeding Intl. Workshop on Hydraulics of Stepped Spillways”, VAW, ETH-Zurich, H.E.
      Minor and W.H. Hager. Balkema. 69–76.
    30. Meireles, I. Matos, J. (2009), “Skimming flow in the nonaerated region of stepped spillways over embankment
      dams”. J. Hydraul. Eng., 135(8), 685–689.
    31. Miang-liang, ZH. Yong-ming, SH. (2008), “Three dimentional simulation of meandering river basin on 3-D
      RNG k − ε turbulence model”. Journal of hydrodynamics, 20(4): 448-455.
    32. Morovati, Kh. Eghbalzadeh, A. Javan, M. (2015), “Numerical investigation of the configuration of the pools on
      the flowPattern passing over pooled stepped spillway in skimming flow regime. Acta Mech, DOI
      10.1007/s00707-015-1444-x
    33. Morovati, Kh. Eghbalzadeh, A. Soori, S. (2016), “Numerical Study of Energy Dissipation of Pooled Stepped
      spillway”. Civil Engineering Journal. Vol. 2, No. 5.
    34. Nikseresht, A.H. Talebbeydokhti, N. and Rezaei, M.J. (2013), “Numerical simulation of two-phase flow on steppool spillways”. Scientia Iranica, A 20 (2), 222–230.
    35. Peyras, L. Royet, P. Degoutte, G. (1990), “Flow and energy dissipation over stepped gabion weirs”. ASCE
      Convention.
    36. Qun, Ch. Guang-qing, D. Feu-qing, Zh. Qing, Y. (2004). “Three-dimensional turbulence numerical simulation of
      a stepped spillway overflow”. Journal of hydrodynamics, Ser. B, 1, 74-79.
    37. Relvas, A. T. Pinheiro, A. N. (2008), Inception point and air concentration in flows on stepped chutes lined with
      wedge-shaped concrete blocks. Journal of Hydraulic Engineering, 134(8), 1042-1051
    38. Sanchez, M. (2000), “Pressure field in skimming flow over a stepped spillways”. In: Proceeding Intl. Workshop
      on Hydraulics of Stepped Spillways, VAW, ETH-Zurich, H.E. Minor and W.H. Hager. Balkema,
      137–146.
    39. Sarfaraz, M. Attari, J. Pfister, M. (2012), “Numerical Computation of Inception Point Location for Steeply
      Sloping Stepped Spillways”. 9th International Congress on Civil Engineering, May 8-10. Isfahan
      University of Technology (IUT), Isfahan, Iran.
    40. Savage, Bruce M. Michael C. Johnson. (2001), “Flow over ogee spillway: Physical and numerical model case
      study.” Journal of Hydraulic Engineering 127.8:640-649.
    41. Shahhedari, H. Jafari Nodoshan, E. Barati, R. Azhdary moghadam, M. (2014). “Discharge coeficient and energy
      dissipation over stepped spillway under skimming flow regime”. KSCE Journal of Civil Engineering, DOI
      10.1007/s12205-013-0749-3.
    42. Tabbara, M. Chatila, J. Awwad, R. (2005), “Computational simulation of flow over stepped spillways”.
      Computers & structures, 83(27) 2215-2224.
    43. Thorwarth, J. (2008), “Hydraulisches Verhalten der Treppengerinne mit eingetieften Stufen—Selbstinduzierte
      Abflussinstationaritäten und Energiedissipation” [Hydraulics of pooled stepped spillways— Self-induced
      unsteady flow and energy dissipation]. Ph.D. thesis, Univ. of Aachen, Aachen, Germany (in German).
    44. WeiLin, XU. ShuJing, LUO, QiuWen, ZH. Jing, LUO. (2015), “Experimental study on pressure and aeration
      characteristics in stepped chute flows. SCIENCE CHINA. Vol.58 No.4: 720–726. doi: 10.1007/s11431-015-
      5783-6.
    45. Xiangju, Ch. Yongcan, C. Lin, L. (2006), “Numerical simulation of air-water two-phase flow over stepped
      spillways”. Science in China Series E: Technological Sciences, 49(6), 674-684.
    46. Zare, K.H. Doering, J.C. (2012), “Inception Point of Air Entrainment and Training Wall
      Characteristics of Baffles and Sills on Stepped Spillways”. DOI: 10.1061/(ASCE)HY
      .1943-7900.0000630.
    47. Zhan, J. Zhang, J. Gong, Y. (2016), “Numerical investigation of air-entrainment in skimming flow over stepped
      spillways”. Theoretical and Applied Mechanics Letters. Volume 6. Pages 139–142.
    48. Zhang, G. Chanson, H. (2016), Hydraulics of the developing flow region of stepped spillways. II: Pressure and
      velocity fields. Journal of Hydraulic Engineering, 142(7).
    49. Zhenwei, M. Zhiyan, Zh. Tao, Zh. (2012), “Numerical Simulation of 3-D Flow Field of Spillway based on VOF
      Method”. Procedia Engineering, 28, 808-812.
    50. Zhi-yong, D. Hun-wei, L.J. (2006), “Numerical simulation of skimming flow over mild stepped channel”.
      Journal of Hydrodynamics, Ser. B, 18(3) 367-371.
    51. ZhongDong, Q. XiaoQing, H. WenXin, H. António, A. (2009), “Numerical simulation and analysis of water
      flow over stepped spillways”. Science in China Series E: Technological Sciences, 52(7) 1958-1965.
    Fig. 1. (a) Dimensions of the casting with runners (unit: mm), (b) a melt flow simulation using Flow-3D software together with Reilly's model[44], predicted that a large amount of bifilms (denoted by the black particles) would be contained in the final casting. (c) A solidification simulation using Pro-cast software showed that no shrinkage defect was contained in the final casting.

    AZ91 합금 주물 내 연행 결함에 대한 캐리어 가스의 영향

    Effect of carrier gases on the entrainment defects within AZ91 alloy castings

    Tian Liab J.M.T.Daviesa Xiangzhen Zhuc
    aUniversity of Birmingham, Birmingham B15 2TT, United Kingdom
    bGrainger and Worrall Ltd, Bridgnorth WV15 5HP, United Kingdom
    cBrunel Centre for Advanced Solidification Technology, Brunel University London, Kingston Ln, London, Uxbridge UB8 3PH, United Kingdom

    Abstract

    An entrainment defect (also known as a double oxide film defect or bifilm) acts a void containing an entrapped gas when submerged into a light-alloy melt, thus reducing the quality and reproducibility of the final castings. Previous publications, carried out with Al-alloy castings, reported that this trapped gas could be subsequently consumed by the reaction with the surrounding melt, thus reducing the void volume and negative effect of entrainment defects. Compared with Al-alloys, the entrapped gas within Mg-alloy might be more efficiently consumed due to the relatively high reactivity of magnesium. However, research into the entrainment defects within Mg alloys has been significantly limited. In the present work, AZ91 alloy castings were produced under different carrier gas atmospheres (i.e., SF6/CO2, SF6/air). The evolution processes of the entrainment defects contained in AZ91 alloy were suggested according to the microstructure inspections and thermodynamic calculations. The defects formed in the different atmospheres have a similar sandwich-like structure, but their oxide films contained different combinations of compounds. The use of carrier gases, which were associated with different entrained-gas consumption rates, affected the reproducibility of AZ91 castings.

    Keywords

    Magnesium alloyCastingOxide film, Bifilm, Entrainment defect, Reproducibility

    연행 결함(이중 산화막 결함 또는 이중막 결함이라고도 함)은 경합금 용융물에 잠길 때 갇힌 가스를 포함하는 공극으로 작용하여 최종 주물의 품질과 재현성을 저하시킵니다. Al-합금 주조로 수행된 이전 간행물에서는 이 갇힌 가스가 주변 용융물과의 반응에 의해 후속적으로 소모되어 공극 부피와 연행 결함의 부정적인 영향을 줄일 수 있다고 보고했습니다. Al-합금에 비해 마그네슘의 상대적으로 높은 반응성으로 인해 Mg-합금 내에 포집된 가스가 더 효율적으로 소모될 수 있습니다. 그러나 Mg 합금 내 연행 결함에 대한 연구는 상당히 제한적이었습니다. 현재 작업에서 AZ91 합금 주물은 다양한 캐리어 가스 분위기(즉, SF 6 /CO2 , SF 6 / 공기). AZ91 합금에 포함된 엔트레인먼트 결함의 진화 과정은 미세조직 검사 및 열역학적 계산에 따라 제안되었습니다. 서로 다른 분위기에서 형성된 결함은 유사한 샌드위치 구조를 갖지만 산화막에는 서로 다른 화합물 조합이 포함되어 있습니다. 다른 동반 가스 소비율과 관련된 운반 가스의 사용은 AZ91 주물의 재현성에 영향을 미쳤습니다.

    키워드

    마그네슘 합금주조Oxide film, Bifilm, Entrainment 불량, 재현성

    1 . 소개

    지구상에서 가장 가벼운 구조용 금속인 마그네슘은 지난 수십 년 동안 가장 매력적인 경금속 중 하나가 되었습니다. 결과적으로 마그네슘 산업은 지난 20년 동안 급속한 발전을 경험했으며 [1 , 2] , 이는 전 세계적으로 Mg 합금에 대한 수요가 크게 증가했음을 나타냅니다. 오늘날 Mg 합금의 사용은 자동차, 항공 우주, 전자 등의 분야에서 볼 수 있습니다. [3 , 4] . Mg 금속의 전 세계 소비는 특히 자동차 산업에서 앞으로 더욱 증가할 것으로 예측되었습니다. 기존 자동차와 전기 자동차 모두의 에너지 효율성 요구 사항이 설계를 경량화하도록 더욱 밀어붙이기 때문입니다 [3 , 56] .

    Mg 합금에 대한 수요의 지속적인 성장은 Mg 합금 주조의 품질 및 기계적 특성 개선에 대한 광범위한 관심을 불러일으켰습니다. Mg 합금 주조 공정 동안 용융물의 표면 난류는 소량의 주변 대기를 포함하는 이중 표면 필름의 포획으로 이어질 수 있으므로 동반 결함(이중 산화막 결함 또는 이중막 결함이라고도 함)을 형성합니다. ) [7] , [8] , [9] , [10] . 무작위 크기, 수량, 방향 및 연행 결함의 배치는 주조 특성의 변화와 관련된 중요한 요인으로 널리 받아들여지고 있습니다 [7] . 또한 Peng et al. [11]AZ91 합금 용융물에 동반된 산화물 필름이 Al 8 Mn 5 입자에 대한 필터 역할을 하여 침전될 때 가두는 것을 발견했습니다 . Mackie et al. [12]는 또한 동반된 산화막이 금속간 입자를 트롤(trawl)하는 작용을 하여 입자가 클러스터링되어 매우 큰 결함을 형성할 수 있다고 제안했습니다. 금속간 화합물의 클러스터링은 비말동반 결함을 주조 특성에 더 해롭게 만들었습니다.

    연행 결함에 관한 이전 연구의 대부분은 Al-합금에 대해 수행되었으며 [7 , [13] , [14] , [15] , [16] , [17] , [18] 몇 가지 잠재적인 방법이 제안되었습니다. 알루미늄 합금 주물의 품질에 대한 부정적인 영향을 줄이기 위해. Nyahumwa et al., [16] 은 연행 결함 내의 공극 체적이 열간 등방압 압축(HIP) 공정에 의해 감소될 수 있음을 보여줍니다. Campbell [7] 은 결함 내부의 동반된 가스가 주변 용융물과의 반응으로 인해 소모될 수 있다고 제안했으며, 이는 Raiszedeh와 Griffiths [19]에 의해 추가로 확인되었습니다 ..혼입 가스 소비가 Al-합금 주물의 기계적 특성에 미치는 영향은 [8 , 9]에 의해 조사되었으며 , 이는 혼입 가스의 소비가 주조 재현성의 개선을 촉진함을 시사합니다.

    Al-합금 내 결함에 대한 조사와 비교하여 Mg-합금 내 연행 결함에 대한 연구는 상당히 제한적입니다. 연행 결함의 존재는 Mg 합금 주물 [20 , 21] 에서 입증 되었지만 그 거동, 진화 및 연행 가스 소비는 여전히 명확하지 않습니다.

    Mg 합금 주조 공정에서 용융물은 일반적으로 마그네슘 점화를 피하기 위해 커버 가스로 보호됩니다. 따라서 모래 또는 매몰 몰드의 공동은 용융물을 붓기 전에 커버 가스로 세척해야 합니다 [22] . 따라서, Mg 합금 주물 내의 연행 가스는 공기만이 아니라 주조 공정에 사용되는 커버 가스를 포함해야 하며, 이는 구조 및 해당 연행 결함의 전개를 복잡하게 만들 수 있습니다.

    SF 6 은 Mg 합금 주조 공정에 널리 사용되는 대표적인 커버 가스입니다 [23] , [24] , [25] . 이 커버 가스는 유럽의 마그네슘 합금 주조 공장에서 사용하도록 제한되었지만 상업 보고서에 따르면 이 커버는 전 세계 마그네슘 합금 산업, 특히 다음과 같은 글로벌 마그네슘 합금 생산을 지배한 국가에서 여전히 인기가 있습니다. 중국, 브라질, 인도 등 [26] . 또한, 최근 학술지 조사에서도 이 커버가스가 최근 마그네슘 합금 연구에서 널리 사용된 것으로 나타났다 [27] . SF 6 커버 가스 의 보호 메커니즘 (즉, 액체 Mg 합금과 SF 6 사이의 반응Cover gas)에 대한 연구는 여러 선행연구자들에 의해 이루어졌으나 표면 산화막의 형성과정이 아직 명확하게 밝혀지지 않았으며, 일부 발표된 결과들도 상충되고 있다. 1970년대 초 Fruehling [28] 은 SF 6 아래에 형성된 표면 피막이 주로 미량의 불화물과 함께 MgO 임을 발견 하고 SF 6 이 Mg 합금 표면 피막에 흡수 된다고 제안했습니다 . Couling [29] 은 흡수된 SF 6 이 Mg 합금 용융물과 반응하여 MgF 2 를 형성함을 추가로 확인했습니다 . 지난 20년 동안 아래에 자세히 설명된 것처럼 Mg 합금 표면 필름의 다양한 구조가 보고되었습니다.(1)

    단층 필름 . Cashion [30 , 31] 은 X선 광전자 분광법(XPS)과 오제 분광법(AES)을 사용하여 표면 필름을 MgO 및 MgF 2 로 식별했습니다 . 그는 또한 필름의 구성이 두께와 전체 실험 유지 시간에 걸쳐 일정하다는 것을 발견했습니다. Cashion이 관찰한 필름은 10분에서 100분의 유지 시간으로 생성된 단층 구조를 가졌다.(2)

    이중층 필름 . Aarstad et. al [32] 은 2003년에 이중층 표면 산화막을 보고했습니다. 그들은 예비 MgO 막에 부착된 잘 분포된 여러 MgF 2 입자를 관찰 하고 전체 표면적의 25-50%를 덮을 때까지 성장했습니다. 외부 MgO 필름을 통한 F의 내부 확산은 진화 과정의 원동력이었습니다. 이 이중층 구조는 Xiong의 그룹 [25 , 33] 과 Shih et al. 도 지지했습니다 . [34] .(삼)

    트리플 레이어 필름 . 3층 필름과 그 진화 과정은 Pettersen [35]에 의해 2002년에 보고되었습니다 . Pettersen은 초기 표면 필름이 MgO 상이었고 F의 내부 확산에 의해 점차적으로 안정적인 MgF 2 상 으로 진화한다는 것을 발견했습니다 . 두꺼운 상부 및 하부 MgF 2 층.(4)

    산화물 필름은 개별 입자로 구성 됩니다. Wang et al [36] 은 Mg-alloy 표면 필름을 SF 6 커버 가스 하에서 용융물에 교반 한 다음 응고 후 동반된 표면 필름을 검사했습니다. 그들은 동반된 표면 필름이 다른 연구자들이 보고한 보호 표면 필름처럼 계속되지 않고 개별 입자로 구성된다는 것을 발견했습니다. 젊은 산화막은 MgO 나노 크기의 산화물 입자로 구성되어 있는 반면, 오래된 산화막은 한쪽 면에 불화물과 질화물이 포함된 거친 입자(평균 크기 약 1μm)로 구성되어 있습니다.

    Mg 합금 용융 표면의 산화막 또는 동반 가스는 모두 액체 Mg 합금과 커버 가스 사이의 반응으로 인해 형성되므로 Mg 합금 표면막에 대한 위에서 언급한 연구는 진화에 대한 귀중한 통찰력을 제공합니다. 연행 결함. 따라서 SF 6 커버 가스 의 보호 메커니즘 (즉, Mg-합금 표면 필름의 형성)은 해당 동반 결함의 잠재적인 복잡한 진화 과정을 나타냅니다.

    그러나 Mg 합금 용융물에 표면 필름을 형성하는 것은 용융물에 잠긴 동반된 가스의 소비와 다른 상황에 있다는 점에 유의해야 합니다. 예를 들어, 앞서 언급한 연구에서 표면 성막 동안 충분한 양의 커버 가스가 담지되어 커버 가스의 고갈을 억제했습니다. 대조적으로, Mg 합금 용융물 내의 동반된 가스의 양은 유한하며, 동반된 가스는 완전히 고갈될 수 있습니다. Mirak [37] 은 3.5% SF 6 /기포를 특별히 설계된 영구 금형에서 응고되는 순수한 Mg 합금 용융물에 도입했습니다. 기포가 완전히 소모되었으며, 해당 산화막은 MgO와 MgF 2 의 혼합물임을 알 수 있었다.. 그러나 Aarstad [32] 및 Xiong [25 , 33]에 의해 관찰된 MgF 2 스팟 과 같은 핵 생성 사이트 는 관찰되지 않았습니다. Mirak은 또한 조성 분석을 기반으로 산화막에서 MgO 이전에 MgF 2 가 형성 되었다고 추측했는데 , 이는 이전 문헌에서 보고된 표면 필름 형성 과정(즉, MgF 2 이전에 형성된 MgO)과 반대 입니다. Mirak의 연구는 동반된 가스의 산화막 형성이 표면막의 산화막 형성과 상당히 다를 수 있음을 나타내었지만 산화막의 구조와 진화에 대해서는 밝히지 않았습니다.

    또한 커버 가스에 캐리어 가스를 사용하는 것도 커버 가스와 액체 Mg 합금 사이의 반응에 영향을 미쳤습니다. SF 6 /air 는 용융 마그네슘의 점화를 피하기 위해 SF 6 /CO 2 운반 가스 [38] 보다 더 높은 함량의 SF 6을 필요로 하여 다른 가스 소비율을 나타냅니다. Liang et.al [39] 은 CO 2 가 캐리어 가스로 사용될 때 표면 필름에 탄소가 형성된다고 제안했는데 , 이는 SF 6 /air 에서 형성된 필름과 다릅니다 . Mg 연소 [40]에 대한 조사 에서 Mg 2 C 3 검출이 보고되었습니다.CO 2 연소 후 Mg 합금 샘플 에서 이는 Liang의 결과를 뒷받침할 뿐만 아니라 이중 산화막 결함에서 Mg 탄화물의 잠재적 형성을 나타냅니다.

    여기에 보고된 작업은 다양한 커버 가스(즉, SF 6 /air 및 SF 6 /CO 2 )로 보호되는 AZ91 Mg 합금 주물에서 형성된 연행 결함의 거동과 진화에 대한 조사 입니다. 이러한 캐리어 가스는 액체 Mg 합금에 대해 다른 보호성을 가지며, 따라서 상응하는 동반 가스의 다른 소비율 및 발생 프로세스와 관련될 수 있습니다. AZ91 주물의 재현성에 대한 동반 가스 소비의 영향도 연구되었습니다.

    2 . 실험

    2.1 . 용융 및 주조

    3kg의 AZ91 합금을 700 ± 5 °C의 연강 도가니에서 녹였습니다. AZ91 합금의 조성은 표 1 에 나타내었다 . 가열하기 전에 잉곳 표면의 모든 산화물 스케일을 기계가공으로 제거했습니다. 사용 된 커버 가스는 0.5 %이었다 SF 6 / 공기 또는 0.5 % SF 6 / CO 2 (부피. %) 다른 주물 6L / 분의 유량. 용융물은 15분 동안 0.3L/min의 유속으로 아르곤으로 가스를 제거한 다음 [41 , 42] , 모래 주형에 부었습니다. 붓기 전에 샌드 몰드 캐비티를 20분 동안 커버 가스로 플러싱했습니다 [22] . 잔류 용융물(약 1kg)이 도가니에서 응고되었습니다.

    표 1 . 본 연구에 사용된 AZ91 합금의 조성(wt%).

    아연미네소타마그네슘
    9.40.610.150.020.0050.0017잔여

    그림 1 (a)는 러너가 있는 주물의 치수를 보여줍니다. 탑 필링 시스템은 최종 주물에서 연행 결함을 생성하기 위해 의도적으로 사용되었습니다. Green과 Campbell [7 , 43] 은 탑 필링 시스템이 바텀 필링 시스템에 비해 주조 과정에서 더 많은 연행 현상(즉, 이중 필름)을 유발한다고 제안했습니다. 이 금형의 용융 흐름 시뮬레이션(Flow-3D 소프트웨어)은 연행 현상에 관한 Reilly의 모델 [44] 을 사용하여 최종 주조에 많은 양의 이중막이 포함될 것이라고 예측했습니다( 그림 1 에서 검은색 입자로 표시됨) . NS).

    그림 1

    수축 결함은 또한 주물의 기계적 특성과 재현성에 영향을 미칩니다. 이 연구는 주조 품질에 대한 이중 필름의 영향에 초점을 맞추었기 때문에 수축 결함이 발생하지 않도록 금형을 의도적으로 설계했습니다. ProCAST 소프트웨어를 사용한 응고 시뮬레이션은 그림 1c 와 같이 최종 주조에 수축 결함이 포함되지 않음을 보여주었습니다 . 캐스팅 건전함도 테스트바 가공 전 실시간 X-ray를 통해 확인했다.

    모래 주형은 1wt를 함유한 수지 결합된 규사로 만들어졌습니다. % PEPSET 5230 수지 및 1wt. % PEPSET 5112 촉매. 모래는 또한 억제제로 작용하기 위해 2중량%의 Na 2 SiF 6 을 함유했습니다 .. 주입 온도는 700 ± 5 °C였습니다. 응고 후 러너바의 단면을 Sci-Lab Analytical Ltd로 보내 H 함량 분석(LECO 분석)을 하였고, 모든 H 함량 측정은 주조 공정 후 5일째에 실시하였다. 각각의 주물은 인장 강도 시험을 위해 클립 신장계가 있는 Zwick 1484 인장 시험기를 사용하여 40개의 시험 막대로 가공되었습니다. 파손된 시험봉의 파단면을 주사전자현미경(SEM, Philips JEOL7000)을 이용하여 가속전압 5~15kV로 조사하였다. 파손된 시험 막대, 도가니에서 응고된 잔류 Mg 합금 및 주조 러너를 동일한 SEM을 사용하여 단면화하고 연마하고 검사했습니다. CFEI Quanta 3D FEG FIB-SEM을 사용하여 FIB(집속 이온 빔 밀링 기술)에 의해 테스트 막대 파괴 표면에서 발견된 산화막의 단면을 노출했습니다. 분석에 필요한 산화막은 백금층으로 코팅하였다. 그런 다음 30kV로 가속된 갈륨 이온 빔이 산화막의 단면을 노출시키기 위해 백금 코팅 영역을 둘러싼 재료 기판을 밀링했습니다. 산화막 단면의 EDS 분석은 30kV의 가속 전압에서 FIB 장비를 사용하여 수행되었습니다.

    2.2 . 산화 세포

    전술 한 바와 같이, 몇몇 최근 연구자들은 마그네슘 합금의 용탕 표면에 형성된 보호막 조사 [38 , 39 , [46] , [47] , [48] , [49] , [50] , [51] , [52 ] . 이 실험 동안 사용된 커버 가스의 양이 충분하여 커버 가스에서 불화물의 고갈을 억제했습니다. 이 섹션에서 설명하는 실험은 엔트레인먼트 결함의 산화막의 진화를 연구하기 위해 커버 가스의 공급을 제한하는 밀봉된 산화 셀을 사용했습니다. 산화 셀에 포함된 커버 가스는 큰 크기의 “동반된 기포”로 간주되었습니다.

    도 2에 도시된 바와 같이 , 산화셀의 본체는 내부 길이가 400mm, 내경이 32mm인 폐쇄형 연강관이었다. 수냉식 동관을 전지의 상부에 감았습니다. 튜브가 가열될 때 냉각 시스템은 상부와 하부 사이에 온도 차이를 만들어 내부 가스가 튜브 내에서 대류하도록 했습니다. 온도는 도가니 상단에 위치한 K형 열전대로 모니터링했습니다. Nieet al. [53] 은 Mg 합금 용융물의 표면 피막을 조사할 때 SF 6 커버 가스가 유지로의 강철 벽과 반응할 것이라고 제안했습니다 . 이 반응을 피하기 위해 강철 산화 전지의 내부 표면(그림 2 참조)) 및 열전대의 상반부는 질화붕소로 코팅되었습니다(Mg 합금은 질화붕소와 ​​접촉하지 않았습니다).

    그림 2

    실험 중에 고체 AZ91 합금 블록을 산화 셀 바닥에 위치한 마그네시아 도가니에 넣었습니다. 전지는 1L/min의 가스 유속으로 전기 저항로에서 100℃로 가열되었다. 원래의 갇힌 대기(즉, 공기)를 대체하기 위해 셀을 이 온도에서 20분 동안 유지했습니다. 그런 다음, 산화 셀을 700°C로 더 가열하여 AZ91 샘플을 녹였습니다. 그런 다음 가스 입구 및 출구 밸브가 닫혀 제한된 커버 가스 공급 하에서 산화를 위한 밀폐된 환경이 생성되었습니다. 그런 다음 산화 전지를 5분 간격으로 5분에서 30분 동안 700 ± 10°C에서 유지했습니다. 각 유지 시간이 끝날 때 세포를 물로 켄칭했습니다. 실온으로 냉각한 후 산화된 샘플을 절단하고 연마한 다음 SEM으로 검사했습니다.

    3 . 결과

    3.1 . SF 6 /air 에서 형성된 엔트레인먼트 결함의 구조 및 구성

    0.5 % SF의 커버 가스 하에서 AZ91 주물에 형성된 유입 결함의 구조 및 조성 6 / 공기는 SEM 및 EDS에 의해 관찰되었다. 결과는 그림 3에 스케치된 엔트레인먼트 결함의 두 가지 유형이 있음을 나타냅니다 . (1) 산화막이 전통적인 단층 구조를 갖는 유형 A 결함 및 (2) 산화막이 2개 층을 갖는 유형 B 결함. 이러한 결함의 세부 사항은 다음에 소개되었습니다. 여기에서 비말동반 결함은 생물막 또는 이중 산화막으로도 알려져 있기 때문에 B형 결함의 산화막은 본 연구에서 “다층 산화막” 또는 “다층 구조”로 언급되었습니다. “이중 산화막 결함의 이중층 산화막”과 같은 혼란스러운 설명을 피하기 위해.

    그림 3

    그림 4 (ab)는 약 0.4μm 두께의 조밀한 단일층 산화막을 갖는 Type A 결함을 보여줍니다. 이 필름에서 산소, 불소, 마그네슘 및 알루미늄이 검출되었습니다( 그림 4c). 산화막은 마그네슘과 알루미늄의 산화물과 불화물의 혼합물로 추측됩니다. 불소의 검출은 동반된 커버 가스가 이 결함의 형성에 포함되어 있음을 보여주었습니다. 즉, Fig. 4 (a)에 나타난 기공 은 수축결함이나 수소기공도가 아니라 연행결함이었다. 알루미늄의 검출은 Xiong과 Wang의 이전 연구 [47 , 48] 와 다르며 , SF 6으로 보호된 AZ91 용융물의 표면 필름에 알루미늄이 포함되어 있지 않음을 보여주었습니다.커버 가스. 유황은 원소 맵에서 명확하게 인식할 수 없었지만 해당 ESD 스펙트럼에서 S-피크가 있었습니다.

    그림 4

    도 5 (ab)는 다층 산화막을 갖는 Type B 엔트레인먼트 결함을 나타낸다. 산화막의 조밀한 외부 층은 불소와 산소가 풍부하지만( 그림 5c) 상대적으로 다공성인 내부 층은 산소만 풍부하고(즉, 불소가 부족) 부분적으로 함께 성장하여 샌드위치 모양을 형성합니다. 구조. 따라서 외층은 불화물과 산화물의 혼합물이며 내층은 주로 산화물로 추정된다. 황은 EDX 스펙트럼에서만 인식될 수 있었고 요소 맵에서 명확하게 식별할 수 없었습니다. 이는 커버 가스의 작은 S 함량(즉, SF 6 의 0.5% 부피 함량 때문일 수 있음)커버 가스). 이 산화막에서는 이 산화막의 외층에 알루미늄이 포함되어 있지만 내층에서는 명확하게 검출할 수 없었다. 또한 Al의 분포가 고르지 않은 것으로 보입니다. 결함의 우측에는 필름에 알루미늄이 존재하지만 그 농도는 매트릭스보다 높은 것으로 식별할 수 없음을 알 수 있다. 그러나 결함의 왼쪽에는 알루미늄 농도가 훨씬 높은 작은 영역이 있습니다. 이러한 알루미늄의 불균일한 분포는 다른 결함(아래 참조)에서도 관찰되었으며, 이는 필름 내부 또는 아래에 일부 산화물 입자가 형성된 결과입니다.

    그림 5

    무화과 도 4 및 5 는 SF 6 /air 의 커버 가스 하에 주조된 AZ91 합금 샘플에서 형성된 연행 결함의 횡단면 관찰을 나타낸다 . 2차원 단면에서 관찰된 수치만으로 연행 결함을 특성화하는 것만으로는 충분하지 않습니다. 더 많은 이해를 돕기 위해 테스트 바의 파단면을 관찰하여 엔트레인먼트 결함(즉, 산화막)의 표면을 더 연구했습니다.

    Fig. 6 (a)는 SF 6 /air 에서 생산된 AZ91 합금 인장시험봉의 파단면을 보여준다 . 파단면의 양쪽에서 대칭적인 어두운 영역을 볼 수 있습니다. 그림 6 (b)는 어두운 영역과 밝은 영역 사이의 경계를 보여줍니다. 밝은 영역은 들쭉날쭉하고 부서진 특징으로 구성되어 있는 반면, 어두운 영역의 표면은 비교적 매끄럽고 평평했습니다. 또한 EDS 결과( Fig. 6 c-d 및 Table 2) 불소, 산소, 황 및 질소는 어두운 영역에서만 검출되었으며, 이는 어두운 영역이 용융물에 동반된 표면 보호 필름임을 나타냅니다. 따라서 어두운 영역은 대칭적인 특성을 고려할 때 연행 결함이라고 제안할 수 있습니다. Al-합금 주조물의 파단면에서 유사한 결함이 이전에 보고되었습니다 [7] . 질화물은 테스트 바 파단면의 산화막에서만 발견되었지만 그림 1과 그림 4에 표시된 단면 샘플에서는 검출되지 않았습니다 4 및 5 . 근본적인 이유는 이러한 샘플에 포함된 질화물이 샘플 연마 과정에서 가수분해되었을 수 있기 때문입니다 [54] .

    그림 6

    표 2 . EDS 결과(wt.%)는 그림 6에 표시된 영역에 해당합니다 (커버 가스: SF 6 /공기).

    영형마그네슘NS아연NSNS
    그림 6 (b)의 어두운 영역3.481.3279.130.4713.630.570.080.73
    그림 6 (b)의 밝은 영역3.5884.4811.250.68

    도 1 및 도 2에 도시된 결함의 단면 관찰과 함께 도 4 및 도 5 를 참조하면, 인장 시험봉에 포함된 연행 결함의 구조를 도 6 (e) 와 같이 스케치하였다 . 결함에는 산화막으로 둘러싸인 동반된 가스가 포함되어 있어 테스트 바 내부에 보이드 섹션이 생성되었습니다. 파괴 과정에서 결함에 인장력이 가해지면 균열이 가장 약한 경로를 따라 전파되기 때문에 보이드 섹션에서 균열이 시작되어 연행 결함을 따라 전파됩니다 [55] . 따라서 최종적으로 시험봉이 파단되었을 때 Fig. 6 (a) 와 같이 시험봉의 양 파단면에 연행결함의 산화피막이 나타났다 .

    3.2 . SF 6 /CO 2 에 형성된 연행 결함의 구조 및 조성

    SF 6 /air 에서 형성된 엔트레인먼트 결함과 유사하게, 0.5% SF 6 /CO 2 의 커버 가스 아래에서 형성된 결함 도 두 가지 유형의 산화막(즉, 단층 및 다층 유형)을 가졌다. 도 7 (a)는 다층 산화막을 포함하는 엔트레인먼트 결함의 예를 도시한다. 결함에 대한 확대 관찰( 그림 7b )은 산화막의 내부 층이 함께 성장하여 SF 6 /air 의 분위기에서 형성된 결함과 유사한 샌드위치 같은 구조를 나타냄을 보여줍니다 ( 그림 7b). 5 나 ). EDS 스펙트럼( 그림 7c) 이 샌드위치형 구조의 접합부(내층)는 주로 산화마그네슘을 함유하고 있음을 보여주었다. 이 EDS 스펙트럼에서는 불소, 황, 알루미늄의 피크가 확인되었으나 그 양은 상대적으로 적었다. 대조적으로, 산화막의 외부 층은 조밀하고 불화물과 산화물의 혼합물로 구성되어 있습니다( 그림 7d-e).

    그림 7

    Fig. 8 (a)는 0.5%SF 6 /CO 2 분위기에서 제작된 AZ91 합금 인장시험봉의 파단면의 연행결함을 보여준다 . 상응하는 EDS 결과(표 3)는 산화막이 불화물과 산화물을 함유함을 보여주었다. 황과 질소는 검출되지 않았습니다. 게다가, 확대 관찰(  8b)은 산화막 표면에 반점을 나타내었다. 반점의 직경은 수백 나노미터에서 수 마이크론 미터까지 다양했습니다.

    그림 8

    산화막의 구조와 조성을 보다 명확하게 나타내기 위해 테스트 바 파단면의 산화막 단면을 FIB 기법을 사용하여 현장에서 노출시켰다( 그림 9 ). 도 9a에 도시된 바와 같이 , 백금 코팅층과 Mg-Al 합금 기재 사이에 연속적인 산화피막이 발견되었다. 그림 9 (bc)는 다층 구조( 그림 9c 에서 빨간색 상자로 표시)를 나타내는 산화막에 대한 확대 관찰을 보여줍니다 . 바닥층은 불소와 산소가 풍부하고 불소와 산화물의 혼합물이어야 합니다 . 5 와 7, 유일한 산소가 풍부한 최상층은 도 1 및 도 2에 도시 된 “내층”과 유사하였다 5 및 7 .

    그림 9

    연속 필름을 제외하고 도 9 에 도시된 바와 같이 연속 필름 내부 또는 하부에서도 일부 개별 입자가 관찰되었다 . 그림 9( b) 의 산화막 좌측에서 Al이 풍부한 입자가 검출되었으며, 마그네슘과 산소 원소도 풍부하게 함유하고 있어 스피넬 Mg 2 AlO 4 로 추측할 수 있다 . 이러한 Mg 2 AlO 4 입자의 존재는 Fig. 5 와 같이 관찰된 필름의 작은 영역에 높은 알루미늄 농도와 알루미늄의 불균일한 분포의 원인이 된다 .(씨). 여기서 강조되어야 할 것은 연속 산화막의 바닥층의 다른 부분이 이 Al이 풍부한 입자보다 적은 양의 알루미늄을 함유하고 있지만, 그림 9c는 이 바닥층의 알루미늄 양이 여전히 무시할 수 없는 수준임을 나타냅니다 . , 특히 필름의 외층과 비교할 때. 도 9b에 도시된 산화막의 우측 아래에서 입자가 검출되어 Mg와 O가 풍부하여 MgO인 것으로 추측되었다. Wang의 결과에 따르면 [56], Mg 용융물과 Mg 증기의 산화에 의해 Mg 용융물의 표면에 많은 이산 MgO 입자가 형성될 수 있다. 우리의 현재 연구에서 관찰된 MgO 입자는 같은 이유로 인해 형성될 수 있습니다. 실험 조건의 차이로 인해 더 적은 Mg 용융물이 기화되거나 O2와 반응할 수 있으므로 우리 작업에서 형성되는 MgO 입자는 소수에 불과합니다. 또한 필름에서 풍부한 탄소가 발견되어 CO 2 가 용융물과 반응하여 탄소 또는 탄화물을 형성할 수 있음을 보여줍니다 . 이 탄소 농도는 표 3에 나타낸 산화막의 상대적으로 높은 탄소 함량 (즉, 어두운 영역) 과 일치하였다 . 산화막 옆 영역.

    표 3 . 도 8에 도시된 영역에 상응하는 EDS 결과(wt.%) (커버 가스: SF 6 / CO 2 ).

    영형마그네슘NS아연NSNS
    그림 8 (a)의 어두운 영역7.253.6469.823.827.030.86
    그림 8 (a)의 밝은 영역2.100.4482.8313.261.36

    테스트 바 파단면( 도 9 ) 에서 산화막의 이 단면 관찰은 도 6 (e)에 도시된 엔트레인먼트 결함의 개략도를 추가로 확인했다 . SF 6 /CO 2 와 SF 6 /air 의 서로 다른 분위기에서 형성된 엔트레인먼트 결함 은 유사한 구조를 가졌지만 그 조성은 달랐다.

    3.3 . 산화 전지에서 산화막의 진화

    섹션 3.1 및 3.2 의 결과 는 SF 6 /air 및 SF 6 /CO 2 의 커버 가스 아래에서 AZ91 주조에서 형성된 연행 결함의 구조 및 구성을 보여줍니다 . 산화 반응의 다른 단계는 연행 결함의 다른 구조와 조성으로 이어질 수 있습니다. Campbell은 동반된 가스가 주변 용융물과 반응할 수 있다고 추측했지만 Mg 합금 용융물과 포획된 커버 가스 사이에 반응이 발생했다는 보고는 거의 없습니다. 이전 연구자들은 일반적으로 개방된 환경에서 Mg 합금 용융물과 커버 가스 사이의 반응에 초점을 맞췄습니다 [38 , 39 , [46] , [47][48] , [49] , [50] , [51] , [52] , 이는 용융물에 갇힌 커버 가스의 상황과 다릅니다. AZ91 합금에서 엔트레인먼트 결함의 형성을 더 이해하기 위해 엔트레인먼트 결함의 산화막의 진화 과정을 산화 셀을 사용하여 추가로 연구했습니다.

    .도 10 (a 및 d) 0.5 % 방송 SF 보호 산화 셀에서 5 분 동안 유지 된 표면 막 (6) / 공기. 불화물과 산화물(MgF 2 와 MgO) 로 이루어진 단 하나의 층이 있었습니다 . 이 표면 필름에서. 황은 EDS 스펙트럼에서 검출되었지만 그 양이 너무 적어 원소 맵에서 인식되지 않았습니다. 이 산화막의 구조 및 조성은 도 4 에 나타낸 엔트레인먼트 결함의 단층막과 유사하였다 .

    그림 10

    10분의 유지 시간 후, 얇은 (O,S)가 풍부한 상부층(약 700nm)이 예비 F-농축 필름에 나타나 그림 10 (b 및 e) 에서와 같이 다층 구조를 형성했습니다 . ). (O, S)가 풍부한 최상층의 두께는 유지 시간이 증가함에 따라 증가했습니다. Fig. 10 (c, f) 에서 보는 바와 같이 30분간 유지한 산화막도 다층구조를 가지고 있으나 (O,S)가 풍부한 최상층(약 2.5μm)의 두께가 10분 산화막의 그것. 도 10 (bc) 에 도시 된 다층 산화막 은 도 5에 도시된 샌드위치형 결함의 막과 유사한 외관을 나타냈다 .

    도 10에 도시된 산화막의 상이한 구조는 커버 가스의 불화물이 AZ91 합금 용융물과의 반응으로 인해 우선적으로 소모될 것임을 나타내었다. 불화물이 고갈된 후, 잔류 커버 가스는 액체 AZ91 합금과 추가로 반응하여 산화막에 상부 (O, S)가 풍부한 층을 형성했습니다. 따라서 도 1 및 도 3에 도시된 연행 결함의 상이한 구조 및 조성 4 와 5 는 용융물과 갇힌 커버 가스 사이의 진행 중인 산화 반응 때문일 수 있습니다.

    이 다층 구조는 Mg 합금 용융물에 형성된 보호 표면 필름에 관한 이전 간행물 [38 , [46] , [47] , [48] , [49] , [50] , [51] 에서 보고되지 않았습니다 . . 이는 이전 연구원들이 무제한의 커버 가스로 실험을 수행했기 때문에 커버 가스의 불화물이 고갈되지 않는 상황을 만들었기 때문일 수 있습니다. 따라서 엔트레인먼트 결함의 산화피막은 도 10에 도시된 산화피막과 유사한 거동특성을 가지나 [38 ,[46] , [47] , [48] , [49] , [50] , [51] .

    SF 유지 산화막와 마찬가지로 6 / 공기, SF에 형성된 산화물 막 (6) / CO 2는 또한 세포 산화 다른 유지 시간과 다른 구조를 가지고 있었다. .도 11 (a)는 AZ91 개최 산화막, 0.5 %의 커버 가스 하에서 SF 표면 용융 도시 6 / CO 2, 5 분. 이 필름은 MgF 2 로 이루어진 단층 구조를 가졌다 . 이 영화에서는 MgO의 존재를 확인할 수 없었다. 30분의 유지 시간 후, 필름은 다층 구조를 가졌다; 내부 층은 조밀하고 균일한 외관을 가지며 MgF 2 로 구성 되고 외부 층은 MgF 2 혼합물및 MgO. 0.5%SF 6 /air 에서 형성된 표면막과 다른 이 막에서는 황이 검출되지 않았다 . 따라서, 0.5%SF 6 /CO 2 의 커버 가스 내의 불화물 도 막 성장 과정의 초기 단계에서 우선적으로 소모되었다. SF 6 /air 에서 형성된 막과 비교하여 SF 6 /CO 2 에서 형성된 막에서 MgO 는 나중에 나타났고 황화물은 30분 이내에 나타나지 않았다. 이는 SF 6 /air 에서 필름의 형성과 진화 가 SF 6 /CO 2 보다 빠르다 는 것을 의미할 수 있습니다 . CO 2 후속적으로 용융물과 반응하여 MgO를 형성하는 반면, 황 함유 화합물은 커버 가스에 축적되어 반응하여 매우 늦은 단계에서 황화물을 형성할 수 있습니다(산화 셀에서 30분 후).

    그림 11

    4 . 논의

    4.1 . SF 6 /air 에서 형성된 연행 결함의 진화

    Outokumpu HSC Chemistry for Windows( http://www.hsc-chemistry.net/ )의 HSC 소프트웨어를 사용하여 갇힌 기체와 액체 AZ91 합금 사이에서 발생할 수 있는 반응을 탐색하는 데 필요한 열역학 계산을 수행했습니다. 계산에 대한 솔루션은 소량의 커버 가스(즉, 갇힌 기포 내의 양)와 AZ91 합금 용융물 사이의 반응 과정에서 어떤 생성물이 가장 형성될 가능성이 있는지 제안합니다.

    실험에서 압력은 1기압으로, 온도는 700°C로 설정했습니다. 커버 가스의 사용량은 7 × 10으로 가정 하였다 -7  약 0.57 cm의 양으로 kg 3 (3.14 × 10 -6  0.5 % SF위한 kmol) 6 / 공기, 0.35 cm (3) (3.12 × 10 – 8  kmol) 0.5%SF 6 /CO 2 . 포획된 가스와 접촉하는 AZ91 합금 용융물의 양은 모든 반응을 완료하기에 충분한 것으로 가정되었습니다. SF 6 의 분해 생성물 은 SF 5 , SF 4 , SF 3 , SF 2 , F 2 , S(g), S 2(g) 및 F(g) [57] , [58] , [59] , [60] .

    그림 12 는 AZ91 합금과 0.5%SF 6 /air 사이의 반응에 대한 열역학적 계산의 평형 다이어그램을 보여줍니다 . 다이어그램에서 10 -15  kmol 미만의 반응물 및 생성물은 표시되지 않았습니다. 이는 존재 하는 SF 6 의 양 (≈ 1.57 × 10 -10  kmol) 보다 5배 적 으므로 영향을 미치지 않습니다. 실제적인 방법으로 과정을 관찰했습니다.

    그림 12

    이 반응 과정은 3단계로 나눌 수 있다.

    1단계 : 불화물의 형성. AZ91 용융물은 SF 6 및 그 분해 생성물과 우선적으로 반응하여 MgF 2 , AlF 3 및 ZnF 2 를 생성 합니다. 그러나 ZnF 2 의 양 이 너무 적어서 실제적으로 검출되지  않았을 수 있습니다(  MgF 2 의 3 × 10 -10 kmol에 비해 ZnF 2 1.25 × 10 -12 kmol ). 섹션 3.1 – 3.3에 표시된 모든 산화막 . 한편, 잔류 가스에 황이 SO 2 로 축적되었다 .

    2단계 : 산화물의 형성. 액체 AZ91 합금이 포획된 가스에서 사용 가능한 모든 불화물을 고갈시킨 후, Mg와의 반응으로 인해 AlF 3 및 ZnF 2 의 양이 빠르게 감소했습니다. O 2 (g) 및 SO 2 는 AZ91 용융물과 반응하여 MgO, Al 2 O 3 , MgAl 2 O 4 , ZnO, ZnSO 4 및 MgSO 4 를 형성 합니다. 그러나 ZnO 및 ZnSO 4 의 양은 EDS에 의해 실제로 발견되기에는 너무 적었을 것입니다(예: 9.5 × 10 -12  kmol의 ZnO, 1.38 × 10 -14  kmol의 ZnSO 4 , 대조적으로 4.68 × 10−10  kmol의 MgF 2 , X 축의 AZ91 양 이 2.5 × 10 -9  kmol일 때). 실험 사례에서 커버 가스의 F 농도는 매우 낮고 전체 농도 f O는 훨씬 높습니다. 따라서 1단계와 2단계, 즉 불화물과 산화물의 형성은 반응 초기에 동시에 일어나 그림 1과 2와 같이 불화물과 산화물의 가수층 혼합물이 형성될 수 있다 . 4 및 10 (a). 내부 층은 산화물로 구성되어 있지만 불화물은 커버 가스에서 F 원소가 완전히 고갈된 후에 형성될 수 있습니다.

    단계 1-2는 도 10 에 도시 된 다층 구조의 형성 과정을 이론적으로 검증하였다 .

    산화막 내의 MgAl 2 O 4 및 Al 2 O 3 의 양은 도 4에 도시된 산화막과 일치하는 검출하기에 충분한 양이었다 . 그러나, 도 10 에 도시된 바와 같이, 산화셀에서 성장된 산화막에서는 알루미늄의 존재를 인식할 수 없었다 . 이러한 Al의 부재는 표면 필름과 AZ91 합금 용융물 사이의 다음 반응으로 인한 것일 수 있습니다.(1)

    Al 2 O 3  + 3Mg + = 3MgO + 2Al, △G(700°C) = -119.82 kJ/mol(2)

    Mg + MgAl 2 O 4  = MgO + Al, △G(700°C) = -106.34 kJ/mol이는 반응물이 서로 완전히 접촉한다는 가정 하에 열역학적 계산이 수행되었기 때문에 HSC 소프트웨어로 시뮬레이션할 수 없었습니다. 그러나 실제 공정에서 AZ91 용융물과 커버 가스는 보호 표면 필름의 존재로 인해 서로 완전히 접촉할 수 없습니다.

    3단계 : 황화물과 질화물의 형성. 30분의 유지 시간 후, 산화 셀의 기상 불화물 및 산화물이 고갈되어 잔류 가스와 용융 반응을 허용하여 초기 F-농축 또는 (F, O )이 풍부한 표면 필름, 따라서 그림 10 (b 및 c)에 표시된 관찰된 다층 구조를 생성합니다 . 게다가, 질소는 모든 반응이 완료될 때까지 AZ91 용융물과 반응했습니다. 도 6 에 도시 된 산화막 은 질화물 함량으로 인해 이 반응 단계에 해당할 수 있다. 그러나, 그 결과는 도 1 및 도 5에 도시 된 연마된 샘플에서 질화물이 검출되지 않음을 보여준다. 4 와 5, 그러나 테스트 바 파단면에서만 발견됩니다. 질화물은 다음과 같이 샘플 준비 과정에서 가수분해될 수 있습니다 [54] .(삼)

    Mg 3 N 2  + 6H 2 O = 3Mg(OH) 2  + 2NH 3 ↑(4)

    AlN+ 3H 2 O = Al(OH) 3  + NH 3 ↑

    또한 Schmidt et al. [61] 은 Mg 3 N 2 와 AlN이 반응하여 3원 질화물(Mg 3 Al n N n+2, n=1, 2, 3…) 을 형성할 수 있음을 발견했습니다 . HSC 소프트웨어에는 삼원 질화물 데이터베이스가 포함되어 있지 않아 계산에 추가할 수 없습니다. 이 단계의 산화막은 또한 삼원 질화물을 포함할 수 있습니다.

    4.2 . SF 6 /CO 2 에서 형성된 연행 결함의 진화

    도 13 은 AZ91 합금과 0.5%SF 6 /CO 2 사이의 열역학적 계산 결과를 보여준다 . 이 반응 과정도 세 단계로 나눌 수 있습니다.

    그림 13

    1단계 : 불화물의 형성. SF 6 및 그 분해 생성물은 AZ91 용융물에 의해 소비되어 MgF 2 , AlF 3 및 ZnF 2 를 형성했습니다 . 0.5% SF 6 /air 에서 AZ91의 반응에서와 같이 ZnF 2 의 양 이 너무 작아서 실제적으로 감지되지  않았습니다( 2.67 x 10 -10  kmol의 MgF 2 에 비해 ZnF 2 1.51 x 10 -13 kmol ). S와 같은 잔류 가스 트랩에 축적 유황 2 (g) 및 (S)의 일부분 (2) (g)가 CO와 반응하여 2 SO 형성하는 2및 CO. 이 반응 단계의 생성물은 도 11 (a)에 도시된 필름과 일치하며 , 이는 불화물만을 함유하는 단일 층 구조를 갖는다.

    2단계 : 산화물의 형성. ALF 3 및 ZnF 2 MgF로 형성 용융 AZ91 마그네슘의 반응 2 , Al 및 Zn으로한다. SO 2 는 소모되기 시작하여 표면 필름에 산화물을 생성 하고 커버 가스에 S 2 (g)를 생성했습니다. 한편, CO 2 는 AZ91 용융물과 직접 반응하여 CO, MgO, ZnO 및 Al 2 O 3 를 형성 합니다. 도 1에 도시 된 산화막 9 및 11 (b)는 산소가 풍부한 층과 다층 구조로 인해 이 반응 단계에 해당할 수 있습니다.

    커버 가스의 CO는 AZ91 용융물과 추가로 반응하여 C를 생성할 수 있습니다. 이 탄소는 온도가 감소할 때(응고 기간 동안) Mg와 추가로 반응하여 Mg 탄화물을 형성할 수 있습니다 [62] . 이것은 도 4에 도시된 산화막의 탄소 함량이 높은 이유일 수 있다 8 – 9 . Liang et al. [39] 또한 SO 2 /CO 2 로 보호된 AZ91 합금 표면 필름에서 탄소 검출을 보고했습니다 . 생성된 Al 2 O 3 는 MgO와 더 결합하여 MgAl 2 O [63]를 형성할 수 있습니다 . 섹션 4.1 에서 논의된 바와 같이, 알루미나 및 스피넬은 도 11 에 도시된 바와 같이 표면 필름에 알루미늄 부재를 야기하는 Mg와 반응할 수 있다 .

    3단계 : 황화물의 형성. AZ91은 용융물 S 소비하기 시작 2 인 ZnS와 MGS 형성 갇힌 잔류 가스 (g)를. 이러한 반응은 반응 과정의 마지막 단계까지 일어나지 않았으며, 이는 Fig. 7 (c)에 나타난 결함의 S-함량 이 적은 이유일 수 있다 .

    요약하면, 열역학적 계산은 AZ91 용융물이 커버 가스와 반응하여 먼저 불화물을 형성한 다음 마지막에 산화물과 황화물을 형성할 것임을 나타냅니다. 다른 반응 단계에서 산화막은 다른 구조와 조성을 가질 것입니다.

    4.3 . 운반 가스가 동반 가스 소비 및 AZ91 주물의 재현성에 미치는 영향

    SF 6 /air 및 SF 6 /CO 2 에서 형성된 연행 결함의 진화 과정은 4.1절 과 4.2  에서 제안되었습니다 . 이론적인 계산은 실제 샘플에서 발견되는 해당 산화막과 관련하여 검증되었습니다. 연행 결함 내의 대기는 Al-합금 시스템과 다른 시나리오에서 액체 Mg-합금과의 반응으로 인해 효율적으로 소모될 수 있습니다(즉, 연행된 기포의 질소가 Al-합금 용융물과 효율적으로 반응하지 않을 것입니다 [64 , 65] 그러나 일반적으로 “질소 연소”라고 하는 액체 Mg 합금에서 질소가 더 쉽게 소모될 것입니다 [66] ).

    동반된 가스와 주변 액체 Mg-합금 사이의 반응은 동반된 가스를 산화막 내에서 고체 화합물(예: MgO)로 전환하여 동반 결함의 공극 부피를 감소시켜 결함(예: 공기의 동반된 가스가 주변의 액체 Mg 합금에 의해 고갈되면 용융 온도가 700 °C이고 액체 Mg 합금의 깊이가 10 cm라고 가정할 때 최종 고체 제품의 총 부피는 0.044가 됩니다. 갇힌 공기가 취한 초기 부피의 %).

    연행 결함의 보이드 부피 감소와 해당 주조 특성 사이의 관계는 알루미늄 합금 주조에서 널리 연구되었습니다. Nyahumwa와 Campbell [16] 은 HIP(Hot Isostatic Pressing) 공정이 Al-합금 주물의 연행 결함이 붕괴되고 산화물 표면이 접촉하게 되었다고 보고했습니다. 주물의 피로 수명은 HIP 이후 개선되었습니다. Nyahumwa와 Campbell [16] 도 서로 접촉하고 있는 이중 산화막의 잠재적인 결합을 제안했지만 이를 뒷받침하는 직접적인 증거는 없었습니다. 이 결합 현상은 Aryafar et.al에 의해 추가로 조사되었습니다. [8], 그는 강철 튜브에서 산화물 스킨이 있는 두 개의 Al-합금 막대를 다시 녹인 다음 응고된 샘플에 대해 인장 강도 테스트를 수행했습니다. 그들은 Al-합금 봉의 산화물 스킨이 서로 강하게 결합되어 용융 유지 시간이 연장됨에 따라 더욱 강해짐을 발견했으며, 이는 이중 산화막 내 동반된 가스의 소비로 인한 잠재적인 “치유” 현상을 나타냅니다. 구조. 또한 Raidszadeh와 Griffiths [9 , 19] 는 연행 가스가 반응하는 데 더 긴 시간을 갖도록 함으로써 응고 전 용융 유지 시간을 연장함으로써 Al-합금 주물의 재현성에 대한 연행 결함의 부정적인 영향을 성공적으로 줄였습니다. 주변이 녹습니다.

    앞서 언급한 연구를 고려할 때, Mg 합금 주물에서 혼입 가스의 소비는 다음 두 가지 방식으로 혼입 결함의 부정적인 영향을 감소시킬 수 있습니다.

    (1) 이중 산화막의 결합 현상 . 도 5 및 도 7 에 도시 된 샌드위치형 구조 는 이중 산화막 구조의 잠재적인 결합을 나타내었다. 그러나 산화막의 결합으로 인한 강도 증가를 정량화하기 위해서는 더 많은 증거가 필요합니다.

    (2) 연행 결함의 보이드 체적 감소 . 주조품의 품질에 대한 보이드 부피 감소의 긍정적인 효과는 HIP 프로세스 [67]에 의해 널리 입증되었습니다 . 섹션 4.1 – 4.2 에서 논의된 진화 과정과 같이 , 동반된 가스와 주변 AZ91 합금 용융물 사이의 지속적인 반응으로 인해 동반 결함의 산화막이 함께 성장할 수 있습니다. 최종 고체 생성물의 부피는 동반된 기체에 비해 상당히 작았다(즉, 이전에 언급된 바와 같이 0.044%).

    따라서, 혼입 가스의 소모율(즉, 산화막의 성장 속도)은 AZ91 합금 주물의 품질을 향상시키는 중요한 매개변수가 될 수 있습니다. 이에 따라 산화 셀의 산화막 성장 속도를 추가로 조사했습니다.

    도 14 는 상이한 커버 가스(즉, 0.5%SF 6 /air 및 0.5%SF 6 /CO 2 ) 에서의 표면 필름 성장 속도의 비교를 보여준다 . 필름 두께 측정을 위해 각 샘플의 15개의 임의 지점을 선택했습니다. 95% 신뢰구간(95%CI)은 막두께의 변화가 가우시안 분포를 따른다는 가정하에 계산하였다. 0.5%SF 6 /air 에서 형성된 모든 표면막이 0.5%SF 6 /CO 2 에서 형성된 것보다 빠르게 성장함을 알 수 있다 . 다른 성장률은 0.5%SF 6 /air 의 연행 가스 소비율 이 0.5%SF 6 /CO 2 보다 더 높음 을 시사했습니다., 이는 동반된 가스의 소비에 더 유리했습니다.

    그림 14

    산화 셀에서 액체 AZ91 합금과 커버 가스의 접촉 면적(즉, 도가니의 크기)은 많은 양의 용융물과 가스를 고려할 때 상대적으로 작았다는 점에 유의해야 합니다. 결과적으로, 산화 셀 내에서 산화막 성장을 위한 유지 시간은 비교적 길었다(즉, 5-30분). 하지만, 실제 주조에 함유 된 혼입 결함은 (상대적으로 매우 적은, 즉, 수 미크론의 크기에 도시 된 바와 같이 ,도 3. – 6 및 [7]), 동반된 가스는 주변 용융물로 완전히 둘러싸여 상대적으로 큰 접촉 영역을 생성합니다. 따라서 커버 가스와 AZ91 합금 용융물의 반응 시간은 비교적 짧을 수 있습니다. 또한 실제 Mg 합금 모래 주조의 응고 시간은 몇 분일 수 있습니다(예: Guo [68] 은 직경 60mm의 Mg 합금 모래 주조가 응고되는 데 4분이 필요하다고 보고했습니다). 따라서 Mg-합금 용융주조 과정에서 포획된 동반된 가스는 특히 응고 시간이 긴 모래 주물 및 대형 주물의 경우 주변 용융물에 의해 쉽게 소모될 것으로 예상할 수 있습니다.

    따라서, 동반 가스의 다른 소비율과 관련된 다른 커버 가스(0.5%SF 6 /air 및 0.5%SF 6 /CO 2 )가 최종 주물의 재현성에 영향을 미칠 수 있습니다. 이 가정을 검증하기 위해 0.5%SF 6 /air 및 0.5%SF 6 /CO 2 에서 생산된 AZ91 주물 을 기계적 평가를 위해 테스트 막대로 가공했습니다. Weibull 분석은 선형 최소 자승(LLS) 방법과 비선형 최소 자승(비 LLS) 방법을 모두 사용하여 수행되었습니다 [69] .

    그림 15 (ab)는 LLS 방법으로 얻은 UTS 및 AZ91 합금 주물의 연신율의 전통적인 2-p 선형 Weibull 플롯을 보여줍니다. 사용된 추정기는 P= (i-0.5)/N이며, 이는 모든 인기 있는 추정기 중 가장 낮은 편향을 유발하는 것으로 제안되었습니다 [69 , 70] . SF 6 /air 에서 생산된 주물 은 UTS Weibull 계수가 16.9이고 연신율 Weibull 계수가 5.0입니다. 대조적으로, SF 6 /CO 2 에서 생산된 주물의 UTS 및 연신 Weibull 계수는 각각 7.7과 2.7로, SF 6 /CO 2 에 의해 보호된 주물의 재현성이 SF 6 /air 에서 생산된 것보다 훨씬 낮음을 시사합니다. .

    그림 15

    또한 저자의 이전 출판물 [69] 은 선형화된 Weibull 플롯의 단점을 보여주었으며, 이는 Weibull 추정 의 더 높은 편향과 잘못된 2 중단을 유발할 수 있습니다 . 따라서 그림 15 (cd) 와 같이 Non-LLS Weibull 추정이 수행되었습니다 . SF 6 /공기주조물 의 UTS Weibull 계수 는 20.8인 반면, SF 6 /CO 2 하에서 생산된 주조물의 UTS Weibull 계수는 11.4로 낮아 재현성에서 분명한 차이를 보였다. 또한 SF 6 /air elongation(El%) 데이터 세트는 SF 6 /CO 2 의 elongation 데이터 세트보다 더 높은 Weibull 계수(모양 = 5.8)를 가졌습니다.(모양 = 3.1). 따라서 LLS 및 Non-LLS 추정 모두 SF 6 /공기 주조가 SF 6 /CO 2 주조 보다 더 높은 재현성을 갖는다고 제안했습니다 . CO 2 대신 공기를 사용 하면 혼입된 가스의 더 빠른 소비에 기여하여 결함 내의 공극 부피를 줄일 수 있다는 방법을 지원합니다 . 따라서 0.5%SF 6 /CO 2 대신 0.5%SF 6 /air를 사용 하면(동반된 가스의 소비율이 증가함) AZ91 주물의 재현성이 향상되었습니다.

    그러나 모든 Mg 합금 주조 공장이 현재 작업에서 사용되는 주조 공정을 따랐던 것은 아니라는 점에 유의해야 합니다. Mg의 합금 용탕 본 작업은 탈기에 따라서, 동반 가스의 소비에 수소의 영향을 감소 (즉, 수소 잠재적 동반 가스의 고갈 억제, 동반 된 기체로 확산 될 수있다 [7 , 71 , 72] ). 대조적으로, 마그네슘 합금 주조 공장에서는 마그네슘을 주조할 때 ‘가스 문제’가 없고 따라서 인장 특성에 큰 변화가 없다고 널리 믿어지기 때문에 마그네슘 합금 용융물은 일반적으로 탈기되지 않습니다 [73] . 연구에 따르면 Mg 합금 주물의 기계적 특성에 대한 수소의 부정적인 영향 [41 ,42 , 73] , 탈기 공정은 마그네슘 합금 주조 공장에서 여전히 인기가 없습니다.

    또한 현재 작업에서 모래 주형 공동은 붓기 전에 SF 6 커버 가스 로 플러싱되었습니다 [22] . 그러나 모든 Mg 합금 주조 공장이 이러한 방식으로 금형 캐비티를 플러싱한 것은 아닙니다. 예를 들어, Stone Foundry Ltd(영국)는 커버 가스 플러싱 대신 유황 분말을 사용했습니다. 그들의 주물 내의 동반된 가스 는 보호 가스라기 보다는 SO 2 /공기일 수 있습니다 .

    따라서 본 연구의 결과는 CO 2 대신 공기를 사용 하는 것이 최종 주조의 재현성을 향상시키는 것으로 나타났지만 다른 산업용 Mg 합금 주조 공정과 관련하여 캐리어 가스의 영향을 확인하기 위해서는 여전히 추가 조사가 필요합니다.

    7 . 결론

    1.

    AZ91 합금에 형성된 연행 결함이 관찰되었습니다. 그들의 산화막은 단층과 다층의 두 가지 유형의 구조를 가지고 있습니다. 다층 산화막은 함께 성장하여 최종 주조에서 샌드위치 같은 구조를 형성할 수 있습니다.2.

    실험 결과와 이론적인 열역학적 계산은 모두 갇힌 가스의 불화물이 황을 소비하기 전에 고갈되었음을 보여주었습니다. 이중 산화막 결함의 3단계 진화 과정이 제안되었습니다. 산화막은 진화 단계에 따라 다양한 화합물 조합을 포함했습니다. SF 6 /air 에서 형성된 결함 은 SF 6 /CO 2 에서 형성된 것과 유사한 구조를 갖지만 산화막의 조성은 달랐다. 엔트레인먼트 결함의 산화막 형성 및 진화 과정은 이전에 보고된 Mg 합금 표면막(즉, MgF 2 이전에 형성된 MgO)의 것과 달랐다 .삼.

    산화막의 성장 속도는 SF하에 큰 것으로 입증되었다 (6) / SF보다 공기 6 / CO 2 손상 봉입 가스의 빠른 소비에 기여한다. AZ91 합금 주물의 재현성은 SF 6 /CO 2 대신 SF 6 /air를 사용할 때 향상되었습니다 .

    감사의 말

    저자는 EPSRC LiME 보조금 EP/H026177/1의 자금 지원 과 WD Griffiths 박사와 Adrian Carden(버밍엄 대학교)의 도움을 인정합니다. 주조 작업은 University of Birmingham에서 수행되었습니다.

    참조
    [1]
    MK McNutt , SALAZAR K.
    마그네슘, 화합물 및 금속, 미국 지질 조사국 및 미국 내무부
    레 스톤 , 버지니아 ( 2013 )
    Google 학술검색
    [2]
    마그네슘
    화합물 및 금속, 미국 지질 조사국 및 미국 내무부
    ( 1996 )
    Google 학술검색
    [삼]
    I. Ostrovsky , Y. Henn
    ASTEC’07 International Conference-New Challenges in Aeronautics , Moscow ( 2007 ) , pp. 1 – 5
    8월 19-22일
    Scopus에서 레코드 보기Google 학술검색
    [4]
    Y. Wan , B. Tang , Y. Gao , L. Tang , G. Sha , B. Zhang , N. Liang , C. Liu , S. Jiang , Z. Chen , X. Guo , Y. Zhao
    액타 메이터. , 200 ( 2020 ) , 274 – 286 페이지
    기사PDF 다운로드Scopus에서 레코드 보기
    [5]
    JTJ Burd , EA Moore , H. Ezzat , R. Kirchain , R. Roth
    적용 에너지 , 283 ( 2021 ) , 제 116269 조
    기사PDF 다운로드Scopus에서 레코드 보기
    [6]
    AM 루이스 , JC 켈리 , 조지아주 Keoleian
    적용 에너지 , 126 ( 2014 ) , pp. 13 – 20
    기사PDF 다운로드Scopus에서 레코드 보기
    [7]
    J. 캠벨
    주물
    버터워스-하이네만 , 옥스퍼드 ( 2004 )
    Google 학술검색
    [8]
    M. Aryafar , R. Raiszadeh , A. Shalbafzadeh
    J. 메이터. 과학. , 45 ( 2010 년 ) , PP. (3041) – 3051
    교차 참조Scopus에서 레코드 보기
    [9]
    R. 라이자데 , WD 그리피스
    메탈. 메이터. 트랜스. B-프로세스 메탈. 메이터. 프로세스. 과학. , 42 ( 2011 ) , 133 ~ 143페이지
    교차 참조Scopus에서 레코드 보기
    [10]
    R. 라이자데 , WD 그리피스
    J. 합금. Compd. , 491 ( 2010 ) , 575 ~ 580 쪽
    기사PDF 다운로드Scopus에서 레코드 보기
    [11]
    L. Peng , G. Zeng , TC Su , H. Yasuda , K. Nogita , CM Gourlay
    JOM , 71 ( 2019 ) , pp. 2235 – 2244
    교차 참조Scopus에서 레코드 보기
    [12]
    S. Ganguly , AK Mondal , S. Sarkar , A. Basu , S. Kumar , C. Blawert
    코로스. 과학. , 166 ( 2020 )
    [13]
    GE Bozchaloei , N. Varahram , P. Davami , SK 김
    메이터. 과학. 영어 A-구조체. 메이터. 소품 Microstruct. 프로세스. , 548 ( 2012 ) , 99 ~ 105페이지
    Scopus에서 레코드 보기
    [14]
    S. 폭스 , J. 캠벨
    Scr. 메이터. , 43 ( 2000 ) , PP. 881 – 886
    기사PDF 다운로드Scopus에서 레코드 보기
    [15]
    M. 콕스 , RA 하딩 , J. 캠벨
    메이터. 과학. 기술. , 19 ( 2003 ) , 613 ~ 625페이지
    Scopus에서 레코드 보기
    [16]
    C. Nyahumwa , NR Green , J. Campbell
    메탈. 메이터. 트랜스. A-Phys. 메탈. 메이터. 과학. , 32 ( 2001 ) , 349 ~ 358 쪽
    Scopus에서 레코드 보기
    [17]
    A. Ardekhani , R. Raiszadeh
    J. 메이터. 영어 공연하다. , 21 ( 2012 ) , pp. 1352 – 1362
    교차 참조Scopus에서 레코드 보기
    [18]
    X. Dai , X. Yang , J. Campbell , J. Wood
    메이터. 과학. 기술. , 20 ( 2004 ) , 505 ~ 513 쪽
    Scopus에서 레코드 보기
    [19]
    EM 엘갈라드 , MF 이브라힘 , HW 도티 , FH 사무엘
    필로스. 잡지. , 98 ( 2018 ) , PP. 1337 – 1359
    교차 참조Scopus에서 레코드 보기
    [20]
    WD 그리피스 , NW 라이
    메탈. 메이터. 트랜스. A-Phys. 메탈. 메이터. 과학. , 38A ( 2007 ) , PP. 190 – 196
    교차 참조Scopus에서 레코드 보기
    [21]
    AR Mirak , M. Divandari , SMA Boutorabi , J. 캠벨
    국제 J. 캐스트 만났습니다. 해상도 , 20 ( 2007 ) , PP. 215 – 220
    교차 참조Scopus에서 레코드 보기
    [22]
    C. 칭기
    주조공학 연구실
    Helsinki University of Technology , Espoo, Finland ( 2006 )
    Google 학술검색
    [23]
    Y. Jia , J. Hou , H. Wang , Q. Le , Q. Lan , X. Chen , L. Bao
    J. 메이터. 프로세스. 기술. , 278 ( 2020 ) , 제 116542 조
    기사PDF 다운로드Scopus에서 레코드 보기
    [24]
    S. Ouyang , G. Yang , H. Qin , S. Luo , L. Xiao , W. Jie
    메이터. 과학. 영어 A , 780 ( 2020 ) , 제 139138 조
    기사PDF 다운로드Scopus에서 레코드 보기
    [25]
    에스엠. Xiong , X.-F. 왕
    트랜스. 비철금속 사회 중국 , 20 ( 2010 ) , pp. 1228 – 1234
    기사PDF 다운로드Scopus에서 레코드 보기
    [26]
    지브이리서치
    그랜드뷰 리서치
    ( 2018 )
    미국
    Google 학술검색
    [27]
    T. 리 , J. 데이비스
    메탈. 메이터. 트랜스. , 51 ( 2020 ) , PP. 5,389 – (5400)
    교차 참조Scopus에서 레코드 보기
    [28]
    JF Fruehling, 미시간 대학, 1970.
    Google 학술검색
    [29]
    S. 쿨링
    제36회 세계 마그네슘 연례 회의 , 노르웨이 ( 1979 ) , pp. 54 – 57
    Scopus에서 레코드 보기Google 학술검색
    [30]
    S. Cashion , N. Ricketts , P. Hayes
    J. 가벼운 만남. , 2 ( 2002 ) , 43 ~ 47페이지
    기사PDF 다운로드Scopus에서 레코드 보기
    [31]
    S. Cashion , N. Ricketts , P. Hayes
    J. 가벼운 만남. , 2 ( 2002 ) , PP. 37 – 42
    기사PDF 다운로드Scopus에서 레코드 보기
    [32]
    K. Aarstad , G. Tranell , G. Pettersen , TA Engh
    SF6에 의해 보호되는 마그네슘의 표면을 연구하는 다양한 기술
    TMS ( 2003년 )
    Google 학술검색
    [33]
    에스엠 Xiong , X.-L. 리우
    메탈. 메이터. 트랜스. , 38 ( 2007 년 ) , PP. (428) – (434)
    교차 참조Scopus에서 레코드 보기
    [34]
    T.-S. 시 , J.-B. Liu , P.-S. 웨이
    메이터. 화학 물리. , 104 ( 2007 ) , 497 ~ 504페이지
    기사PDF 다운로드Scopus에서 레코드 보기
    [35]
    G. Pettersen , E. Øvrelid , G. Tranell , J. Fenstad , H. Gjestland
    메이터. 과학. 영어 , 332 ( 2002 ) , PP. (285) – (294)
    기사PDF 다운로드Scopus에서 레코드 보기
    [36]
    H. Bo , LB Liu , ZP Jin
    J. 합금. Compd. , 490 ( 2010 ) , 318 ~ 325 쪽
    기사PDF 다운로드Scopus에서 레코드 보기
    [37]
    A. 미락 , C. 데이비슨 , J. 테일러
    코로스. 과학. , 52 ( 2010 ) , PP. 1992 년 – 2000
    기사PDF 다운로드Scopus에서 레코드 보기
    [38]
    BD 리 , UH 부리 , KW 리 , GS 한강 , JW 한
    메이터. 트랜스. , 54 ( 2013 ) , 66 ~ 73페이지
    Scopus에서 레코드 보기
    [39]
    WZ Liang , Q. Gao , F. Chen , HH Liu , ZH Zhao
    China Foundry , 9 ( 2012 ) , pp. 226 – 230
    교차 참조Scopus에서 레코드 보기
    [40]
    UI 골드슐레거 , EY 샤피로비치
    연소. 폭발 충격파 , 35 ( 1999 ) , 637 ~ 644페이지
    Scopus에서 레코드 보기
    [41]
    A. Elsayed , SL Sin , E. Vandersluis , J. Hill , S. Ahmad , C. Ravindran , S. Amer Foundry
    트랜스. 오전. 파운드리 Soc. , 120 ( 2012 ) , 423 ~ 429페이지
    Scopus에서 레코드 보기
    [42]
    E. Zhang , GJ Wang , ZC Hu
    메이터. 과학. 기술. , 26 ( 2010 ) , 1253 ~ 1258페이지
    Scopus에서 레코드 보기
    [43]
    NR 그린 , J. 캠벨
    메이터. 과학. 영어 A-구조체. 메이터. 소품 Microstruct. 프로세스. , 173 ( 1993 ) , 261 ~ 266 쪽
    기사PDF 다운로드Scopus에서 레코드 보기
    [44]
    C 라일리 , MR 졸리 , NR 그린
    MCWASP XII 논문집 – 주조, 용접 및 고급 Solidifcation 프로세스의 12 모델링 , 밴쿠버, 캐나다 ( 2009 )
    Google 학술검색
    [45]
    HE Friedrich, BL Mordike, Springer, 독일, 2006.
    Google 학술검색
    [46]
    C. Zheng , BR Qin , XB Lou
    기계, 산업 및 제조 기술에 관한 2010 국제 회의 , ASME ( 2010 ) , pp. 383 – 388
    2010년 미트
    교차 참조Scopus에서 레코드 보기Google 학술검색
    [47]
    SM Xiong , XF 왕
    트랜스. 비철금속 사회 중국 , 20 ( 2010 ) , pp. 1228 – 1234
    기사PDF 다운로드Scopus에서 레코드 보기
    [48]
    SM Xiong , XL Liu
    메탈. 메이터. 트랜스. A-Phys. 메탈. 메이터. 과학. , 38A ( 2007 ) , PP. (428) – (434)
    교차 참조Scopus에서 레코드 보기
    [49]
    TS Shih , JB Liu , PS Wei
    메이터. 화학 물리. , 104 ( 2007 ) , 497 ~ 504페이지
    기사PDF 다운로드Scopus에서 레코드 보기
    [50]
    K. Aarstad , G. Tranell , G. Pettersen , TA Engh
    매그. 기술. ( 2003 ) , PP. (5) – (10)
    Scopus에서 레코드 보기
    [51]
    G. Pettersen , E. Ovrelid , G. Tranell , J. Fenstad , H. Gjestland
    메이터. 과학. 영어 A-구조체. 메이터. 소품 Microstruct. 프로세스. , 332 ( 2002 ) , 285 ~ 294페이지
    기사PDF 다운로드Scopus에서 레코드 보기
    [52]
    XF 왕 , SM Xiong
    코로스. 과학. , 66 ( 2013 ) , PP. 300 – 307
    기사PDF 다운로드Scopus에서 레코드 보기
    [53]
    SH Nie , SM Xiong , BC Liu
    메이터. 과학. 영어 A-구조체. 메이터. 소품 Microstruct. 프로세스. , 422 ( 2006 ) , 346 ~ 351페이지
    기사PDF 다운로드Scopus에서 레코드 보기
    [54]
    C. Bauer , A. Mogessie , U. Galovsky
    Zeitschrift 모피 Metallkunde , 97 ( 2006 ) , PP. (164) – (168)
    교차 참조Scopus에서 레코드 보기
    [55]
    QG 왕 , D. Apelian , DA Lados
    J. 가벼운 만남. , 1 ( 2001 ) , PP. (73) – 84
    기사PDF 다운로드Scopus에서 레코드 보기
    [56]
    S. Wang , Y. Wang , Q. Ramasse , Z. Fan
    메탈. 메이터. 트랜스. , 51 ( 2020 ) , PP. 2957 – 2974
    교차 참조Scopus에서 레코드 보기
    [57]
    S. Hayashi , W. Minami , T. Oguchi , HJ Kim
    카그. 코그. 론분슈 , 35 ( 2009 ) , 411 ~ 415페이지
    교차 참조Scopus에서 레코드 보기
    [58]
    K. 아르스타드
    노르웨이 과학 기술 대학교
    ( 2004년 )
    Google 학술검색
    [59]
    RL 윌킨스
    J. Chem. 물리. , 51 ( 1969 ) , p. 853
    -&
    Scopus에서 레코드 보기
    [60]
    O. Kubaschewski , K. Hesselemam
    무기물의 열화학적 성질
    Springer-Verlag , 벨린 ( 1991 )
    Google 학술검색
    [61]
    R. Schmidt , M. Strobele , K. Eichele , HJ Meyer
    유로 J. Inorg. 화학 ( 2017 ) , PP. 2727 – 2735
    교차 참조Scopus에서 레코드 보기
    [62]
    B. Hu , Y. Du , H. Xu , W. Sun , WW Zhang , D. Zhao
    제이민 메탈. 분파. B-금속. , 46 ( 2010 ) , 97 ~ 103페이지
    Scopus에서 레코드 보기
    [63]
    O. Salas , H. Ni , V. Jayaram , KC Vlach , CG Levi , R. Mehrabian
    J. 메이터. 해상도 , 6 ( 1991 ) , 1964 ~ 1981페이지
    Scopus에서 레코드 보기
    [64]
    SSS Kumari , UTS Pillai , BC 빠이
    J. 합금. Compd. , 509 ( 2011 ) , pp. 2503 – 2509
    기사PDF 다운로드Scopus에서 레코드 보기
    [65]
    H. Scholz , P. Greil
    J. 메이터. 과학. , 26 ( 1991 ) , 669 ~ 677 쪽
    Scopus에서 레코드 보기
    [66]
    P. Biedenkopf , A. Karger , M. Laukotter , W. Schneider
    매그. 기술. , 2005년 ( 2005년 ) , 39 ~ 42 쪽
    Scopus에서 레코드 보기
    [67]
    HV 앳킨슨 , S. 데이비스
    메탈. 메이터. 트랜스. , 31 ( 2000 ) , PP. 2981 – 3000
    교차 참조Scopus에서 레코드 보기
    [68]
    EJ Guo , L. Wang , YC Feng , LP Wang , YH Chen
    J. 썸. 항문. 칼로리. , 135 ( 2019 ) , PP. 2001 년 – 2008 년
    교차 참조Scopus에서 레코드 보기
    [69]
    T. Li , WD Griffiths , J. Chen
    메탈. 메이터. 트랜스. A-Phys. 메탈. 메이터. 과학. , 48A ( 2017 ) , PP. 5516 – 5528
    교차 참조Scopus에서 레코드 보기
    [70]
    M. Tiryakioglu , D. Hudak는
    J. 메이터. 과학. , 42 ( 2007 ) , pp. 10173 – 10179
    교차 참조Scopus에서 레코드 보기
    [71]
    Y. Yue , WD Griffiths , JL Fife , NR Green
    제1회 3d 재료과학 국제학술대회 논문집 ( 2012 ) , pp. 131 – 136
    교차 참조Scopus에서 레코드 보기Google 학술검색
    [72]
    R. 라이자데 , WD 그리피스
    메탈. 메이터. 트랜스. B-프로세스 메탈. 메이터. 프로세스. 과학. , 37 ( 2006 ) , PP. (865) – (871)
    Scopus에서 레코드 보기
    [73]
    ZC Hu , EL Zhang , SY Zeng
    메이터. 과학. 기술. , 24 ( 2008 ) , 1304 ~ 1308페이지
    교차 참조Scopus에서 레코드 보기

    Conflict resolution in the multi-stakeholder stepped spillway design under uncertainty by machine learning techniques

    기계 학습 기술에 의한 불확실성 하에서 다중 이해 관계자 계단형 배수로 설계의 충돌 해결

    Conflict resolution in the multi-stakeholder stepped spillway design under uncertainty by machine learning techniques

    Mehrdad GhorbaniMooseluaMohammad RezaNikoobParnian HashempourBakhtiaribNooshin BakhtiariRayanicAzizallahIzadyd
    aDepartment of Engineering Sciences, University of Agder, Norway
    bDepartment of Civil and Environmental Engineering, Shiraz University, Shiraz, Iran
    cSchool of Engineering, Department of Civil and Environmental Engineering, Shiraz University, Shiraz, IrandWater Research Center, Sultan Qaboos University, Muscat, Oman

    Abstract

    The optimal spillway design is of great significance since these structures can reduce erosion downstream of the dams. This study proposes a risk-based optimization framework for a stepped spillway to achieve an economical design scenario with the minimum loss in hydraulic performance. Accordingly, the stepped spillway was simulated in the FLOW-3D® model, and the validated model was repeatedly performed for various geometric states.

    The results were used to form a Multilayer Perceptron artificial neural network (MLP-ANN) surrogate model. Then, a risk-based optimization model was formed by coupling the MLP-ANN and NSGA-II. The concept of conditional value at risk (CVaR) was utilized to reduce the risk of the designed spillway malfunctions in high flood flow rates, while minimizing the construction cost and the loss in hydraulic performance.

    Lastly, given the conflicting objectives of stakeholders, the non-cooperative graph model for conflict resolution (GMCR) was applied to achieve a compromise on the Pareto optimal solutions. Applicability of the suggested approach in the Jarreh Dam, Iran, resulted in a practical design scenario, which simultaneously minimizes the loss in hydraulic performance and the project cost and satisfies the priorities of decision-makers.

    Keywords

    Stepped spillway, FLOW-3D® ,CVaR-based optimization model, GMCR-plus, NSGA-II

    최적의 배수로 설계는 이러한 구조가 댐 하류의 침식을 줄일 수 있기 때문에 매우 중요합니다. 본 연구에서는 유압 성능 손실을 최소화하면서 경제적인 설계 시나리오를 달성하기 위해 계단형 여수로에 대한 위험 기반 최적화 프레임워크를 제안합니다. 따라서 FLOW-3D® 모델에서 계단식 배수로를 시뮬레이션하고 다양한 기하학적 상태에 대해 검증된 모델을 반복적으로 수행했습니다.

    결과는 다층 퍼셉트론 인공 신경망(MLP-ANN) 대리 모델을 형성하는 데 사용되었습니다. 그런 다음 MLP-ANN과 NSGA-II를 결합하여 위험 기반 최적화 모델을 구성했습니다. 위험 조건부 값(CVaR)의 개념은 높은 홍수 유량에서 설계된 방수로 오작동의 위험을 줄이는 동시에 건설 비용과 수리 성능 손실을 최소화하기 위해 활용되었습니다.

    마지막으로 이해관계자의 상충되는 목표를 고려하여 파레토 최적해에 대한 절충안을 달성하기 위해 갈등 해결을 위한 비협조적 그래프 모델(GMCR)을 적용하였다. 이란 Jarreh 댐에서 제안된 접근 방식의 적용 가능성은 수력 성능 손실과 프로젝트 비용을 동시에 최소화하고 의사 결정자의 우선 순위를 만족시키는 실용적인 설계 시나리오로 귀결되었습니다.

    FLOW-3D AM

    flow3d AM-product
    FLOW-3D AM-product

    와이어 파우더 기반 DED | Wire Powder Based DED

    일부 연구자들은 부품을 만들기 위해 더 넓은 범위의 처리 조건을 사용하여 하이브리드 와이어 분말 기반 DED 시스템을 찾고 있습니다. 예를 들어, 이 시뮬레이션은 다양한 분말 및 와이어 이송 속도를 가진 하이브리드 시스템을 살펴봅니다.

    와이어 기반 DED | Wire Based DED

    와이어 기반 DED는 분말 기반 DED보다 처리량이 높고 낭비가 적지만 재료 구성 및 증착 방향 측면에서 유연성이 떨어집니다. FLOW-3D AM 은 와이어 기반 DED의 처리 결과를 이해하는데 유용하며 최적화 연구를 통해 빌드에 대한 와이어 이송 속도 및 직경과 같은 최상의 처리 매개 변수를 찾을 수 있습니다.

    FLOW-3D AM은 레이저 파우더 베드 융합 (L-PBF), 바인더 제트 및 DED (Directed Energy Deposition)와 같은 적층 제조 공정 ( additive manufacturing )을 시뮬레이션하고 분석하는 CFD 소프트웨어입니다. FLOW-3D AM 의 다중 물리 기능은 공정 매개 변수의 분석 및 최적화를 위해 분말 확산 및 압축, 용융 풀 역학, L-PBF 및 DED에 대한 다공성 형성, 바인더 분사 공정을 위한 수지 침투 및 확산에 대해 매우 정확한 시뮬레이션을 제공합니다.

    3D 프린팅이라고도하는 적층 제조(additive manufacturing)는 일반적으로 층별 접근 방식을 사용하여, 분말 또는 와이어로 부품을 제조하는 방법입니다. 금속 기반 적층 제조 공정에 대한 관심은 지난 몇 년 동안 시작되었습니다. 오늘날 사용되는 3 대 금속 적층 제조 공정은 PBF (Powder Bed Fusion), DED (Directed Energy Deposition) 및 바인더 제트 ( Binder jetting ) 공정입니다.  FLOW-3D  AM  은 이러한 각 프로세스에 대한 고유 한 시뮬레이션 통찰력을 제공합니다.

    파우더 베드 융합 및 직접 에너지 증착 공정에서 레이저 또는 전자 빔을 열원으로 사용할 수 있습니다. 두 경우 모두 PBF용 분말 형태와 DED 공정용 분말 또는 와이어 형태의 금속을 완전히 녹여 융합하여 층별로 부품을 형성합니다. 그러나 바인더 젯팅(Binder jetting)에서는 결합제 역할을 하는 수지가 금속 분말에 선택적으로 증착되어 층별로 부품을 형성합니다. 이러한 부품은 더 나은 치밀화를 달성하기 위해 소결됩니다.

    FLOW-3D AM 의 자유 표면 추적 알고리즘과 다중 물리 모델은 이러한 각 프로세스를 높은 정확도로 시뮬레이션 할 수 있습니다. 레이저 파우더 베드 융합 (L-PBF) 공정 모델링 단계는 여기에서 자세히 설명합니다. DED 및 바인더 분사 공정에 대한 몇 가지 개념 증명 시뮬레이션도 표시됩니다.

    레이저 파우더 베드 퓨전 (L-PBF)

    LPBF 공정에는 유체 흐름, 열 전달, 표면 장력, 상 변화 및 응고와 같은 복잡한 다중 물리학 현상이 포함되어 공정 및 궁극적으로 빌드 품질에 상당한 영향을 미칩니다. FLOW-3D AM 의 물리적 모델은 질량, 운동량 및 에너지 보존 방정식을 동시에 해결하는 동시에 입자 크기 분포 및 패킹 비율을 고려하여 중규모에서 용융 풀 현상을 시뮬레이션합니다.

    FLOW-3D DEM FLOW-3D WELD 는 전체 파우더 베드 융합 공정을 시뮬레이션하는 데 사용됩니다. L-PBF 공정의 다양한 단계는 분말 베드 놓기, 분말 용융 및 응고,이어서 이전에 응고 된 층에 신선한 분말을 놓는 것, 그리고 다시 한번 새 층을 이전 층에 녹이고 융합시키는 것입니다. FLOW-3D AM  은 이러한 각 단계를 시뮬레이션하는 데 사용할 수 있습니다.

    파우더 베드 부설 공정

    FLOW-3D DEM을 통해 분말 크기 분포, 재료 특성, 응집 효과는 물론 롤러 또는 블레이드 움직임 및 상호 작용과 같은 기하학적 효과와 관련된 분말 확산 및 압축을 이해할 수 있습니다. 이러한 시뮬레이션은 공정 매개 변수가 후속 인쇄 공정에서 용융 풀 역학에 직접적인 영향을 미치는 패킹 밀도와 같은 분말 베드 특성에 어떻게 영향을 미치는지에 대한 정확한 이해를 제공합니다.

    다양한 파우더 베드 압축을 달성하는 한 가지 방법은 베드를 놓는 동안 다양한 입자 크기 분포를 선택하는 것입니다. 아래에서 볼 수 있듯이 세 가지 크기의 입자 크기 분포가 있으며, 이는 가장 높은 압축을 제공하는 Case 2와 함께 다양한 분말 베드 압축을 초래합니다.

    파우더 베드 분포 다양한 입자 크기 분포
    세 가지 다른 입자 크기 분포를 사용하여 파우더 베드 배치
    파우더 베드 압축 결과
    세 가지 다른 입자 크기 분포를 사용한 분말 베드 압축

    입자-입자 상호 작용, 유체-입자 결합 및 입자 이동 물체 상호 작용은 FLOW-3D DEM을 사용하여 자세히 분석 할 수도 있습니다 . 또한 입자간 힘을 지정하여 분말 살포 응용 분야를 보다 정확하게 연구 할 수도 있습니다.

    FLOW-3D AM  시뮬레이션은 이산 요소 방법 (DEM)을 사용하여 역 회전하는 원통형 롤러로 인한 분말 확산을 연구합니다. 비디오 시작 부분에서 빌드 플랫폼이 위로 이동하는 동안 분말 저장소가 아래로 이동합니다. 그 직후, 롤러는 분말 입자 (초기 위치에 따라 색상이 지정됨)를 다음 층이 녹고 구축 될 준비를 위해 구축 플랫폼으로 펼칩니다. 이러한 시뮬레이션은 저장소에서 빌드 플랫폼으로 전송되는 분말 입자의 선호 크기에 대한 추가 통찰력을 제공 할 수 있습니다.

    Melting | 파우더 베드 용해

    DEM 시뮬레이션에서 파우더 베드가 생성되면 STL 파일로 추출됩니다. 다음 단계는 CFD를 사용하여 레이저 용융 공정을 시뮬레이션하는 것입니다. 여기서는 레이저 빔과 파우더 베드의 상호 작용을 모델링 합니다. 이 프로세스를 정확하게 포착하기 위해 물리학에는 점성 흐름, 용융 풀 내의 레이저 반사 (광선 추적을 통해), 열 전달, 응고, 상 변화 및 기화, 반동 압력, 차폐 가스 압력 및 표면 장력이 포함됩니다. 이 모든 물리학은 이 복잡한 프로세스를 정확하게 시뮬레이션하기 위해 TruVOF 방법을 기반으로 개발되었습니다.

    레이저 출력 200W, 스캔 속도 3.0m / s, 스폿 반경 100μm에서 파우더 베드의 용융 풀 분석.

    용융 풀이 응고되면 FLOW-3D AM  압력 및 온도 데이터를 Abaqus 또는 MSC Nastran과 같은 FEA 도구로 가져와 응력 윤곽 및 변위 프로파일을 분석 할 수도 있습니다.

    Multilayer | 다층 적층 제조

    용융 풀 트랙이 응고되면 DEM을 사용하여 이전에 응고된 층에 새로운 분말 층의 확산을 시뮬레이션 할 수 있습니다. 유사하게, 레이저 용융은 새로운 분말 층에서 수행되어 후속 층 간의 융합 조건을 분석 할 수 있습니다.

    해석 진행 절차는 첫 번째 용융층이 응고되면 입자의 두 번째 층이 응고 층에 증착됩니다. 새로운 분말 입자 층에 레이저 공정 매개 변수를 지정하여 용융 풀 시뮬레이션을 다시 수행합니다. 이 프로세스를 여러 번 반복하여 연속적으로 응고된 층 간의 융합, 빌드 내 온도 구배를 평가하는 동시에 다공성 또는 기타 결함의 형성을 모니터링 할 수 있습니다.

    다층 적층 적층 제조 시뮬레이션

    LPBF의 키홀 링 | Keyholing in LPBF

    키홀링 중 다공성은 어떻게 형성됩니까? 이것은 TU Denmark의 연구원들이 FLOW-3D AM을 사용하여 답변한 질문이었습니다. 레이저 빔의 적용으로 기판이 녹으면 기화 및 상 변화로 인한 반동 압력이 용융 풀을 압박합니다. 반동 압력으로 인한 하향 흐름과 레이저 반사로 인한 추가 레이저 에너지 흡수가 공존하면 폭주 효과가 발생하여 용융 풀이 Keyholing으로 전환됩니다. 결국, 키홀 벽을 따라 온도가 변하기 때문에 표면 장력으로 인해 벽이 뭉쳐져서 진행되는 응고 전선에 의해 갇힐 수 있는 공극이 생겨 다공성이 발생합니다. FLOW-3D AM 레이저 파우더 베드 융합 공정 모듈은 키홀링 및 다공성 형성을 시뮬레이션 하는데 필요한 모든 물리 모델을 보유하고 있습니다.

    바인더 분사 (Binder jetting)

    Binder jetting 시뮬레이션은 모세관 힘의 영향을받는 파우더 베드에서 바인더의 확산 및 침투에 대한 통찰력을 제공합니다. 공정 매개 변수와 재료 특성은 증착 및 확산 공정에 직접적인 영향을 미칩니다.

    Scan Strategy | 스캔 전략

    스캔 전략은 온도 구배 및 냉각 속도에 영향을 미치기 때문에 미세 구조에 직접적인 영향을 미칩니다. 연구원들은 FLOW-3D AM 을 사용하여 결함 형성과 응고된 금속의 미세 구조에 영향을 줄 수 있는 트랙 사이에서 발생하는 재 용융을 이해하기 위한 최적의 스캔 전략을 탐색하고 있습니다. FLOW-3D AM 은 하나 또는 여러 레이저에 대해 시간에 따른 방향 속도를 구현할 때 완전한 유연성을 제공합니다.

    Beam Shaping | 빔 형성

    레이저 출력 및 스캔 전략 외에도 레이저 빔 모양과 열유속 분포는 LPBF 공정에서 용융 풀 역학에 큰 영향을 미칩니다. AM 기계 제조업체는 공정 안정성 및 처리량에 대해 다중 코어 및 임의 모양의 레이저 빔 사용을 모색하고 있습니다. FLOW-3D AM을 사용하면 멀티 코어 및 임의 모양의 빔 프로파일을 구현할 수 있으므로 생산량을 늘리고 부품 품질을 개선하기 위한 최상의 구성에 대한 통찰력을 제공 할 수 있습니다.

    이 영역에서 수행 된 일부 작업에 대해 자세히 알아 보려면 “The Next Frontier of Metal AM”웨비나를 시청하십시오.

    Multi-material Powder Bed Fusion | 다중 재료 분말 베드 융합

    이 시뮬레이션에서 스테인리스 강 및 알루미늄 분말은 FLOW-3D AM 이 용융 풀 역학을 정확하게 포착하기 위해 추적하는 독립적으로 정의 된 온도 의존 재료 특성을 가지고 있습니다. 시뮬레이션은 용융 풀에서 재료 혼합을 이해하는 데 도움이됩니다.

    다중 재료 용접 사례 연구

    이종 금속의 레이저 키홀 용접에서 금속 혼합 조사

    GM과 University of Utah의 연구원들은 FLOW-3D WELD 를 사용 하여 레이저 키홀 용접을 통한 이종 금속의 혼합을 이해했습니다. 그들은 반동 압력 및 Marangoni 대류와 관련하여 구리와 알루미늄의 혼합 농도에 대한 레이저 출력 및 스캔 속도의 영향을 조사했습니다. 그들은 시뮬레이션을 실험 결과와 비교했으며 샘플 내의 절단 단면에서 재료 농도 사이에 좋은 일치를 발견했습니다.

    이종 금속의 레이저 키홀 용접에서 금속 혼합 조사
    이종 금속의 레이저 키홀 용접에서 금속 혼합 조사
    참조 : Wenkang Huang, Hongliang Wang, Teresa Rinker, Wenda Tan, 이종 금속의 레이저 키홀 용접에서 금속 혼합 조사 , Materials & Design, Volume 195, (2020). https://doi.org/10.1016/j.matdes.2020.109056
    참조 : Wenkang Huang, Hongliang Wang, Teresa Rinker, Wenda Tan, 이종 금속의 레이저 키홀 용접에서 금속 혼합 조사 , Materials & Design, Volume 195, (2020). https://doi.org/10.1016/j.matdes.2020.109056

    방향성 에너지 증착

    FLOW-3D AM 의 내장 입자 모델 을 사용하여 직접 에너지 증착 프로세스를 시뮬레이션 할 수 있습니다. 분말 주입 속도와 고체 기질에 입사되는 열유속을 지정함으로써 고체 입자는 용융 풀에 질량, 운동량 및 에너지를 추가 할 수 있습니다. 다음 비디오에서 고체 금속 입자가 용융 풀에 주입되고 기판에서 용융 풀의 후속 응고가 관찰됩니다.

    Figure 6. Evolution of melt pool in the overhang region (θ = 45°, P = 100 W, v = 1000 mm/s, the streamlines are shown by arrows).

    Experimental and numerical investigation of the origin of surface roughness in laser powder bed fused overhang regions

    레이저 파우더 베드 융합 오버행 영역에서 표면 거칠기의 원인에 대한 실험 및 수치 조사

    Shaochuan Feng,Amar M. Kamat,Soheil Sabooni &Yutao PeiPages S66-S84 | Received 18 Jan 2021, Accepted 25 Feb 2021, Published online: 10 Mar 2021

    ABSTRACT

    Surface roughness of laser powder bed fusion (L-PBF) printed overhang regions is a major contributor to deteriorated shape accuracy/surface quality. This study investigates the mechanisms behind the evolution of surface roughness (Ra) in overhang regions. The evolution of surface morphology is the result of a combination of border track contour, powder adhesion, warp deformation, and dross formation, which is strongly related to the overhang angle (θ). When 0° ≤ θ ≤ 15°, the overhang angle does not affect Ra significantly since only a small area of the melt pool boundaries contacts the powder bed resulting in slight powder adhesion. When 15° < θ ≤ 50°, powder adhesion is enhanced by the melt pool sinking and the increased contact area between the melt pool boundary and powder bed. When θ > 50°, large waviness of the overhang contour, adhesion of powder clusters, severe warp deformation and dross formation increase Ra sharply.

    레이저 파우더 베드 퓨전 (L-PBF) 프린팅 오버행 영역의 표면 거칠기는 형상 정확도 / 표면 품질 저하의 주요 원인입니다. 이 연구 는 오버행 영역에서 표면 거칠기 (Ra ) 의 진화 뒤에 있는 메커니즘을 조사합니다 . 표면 형태의 진화는 오버행 각도 ( θ ) 와 밀접한 관련이있는 경계 트랙 윤곽, 분말 접착, 뒤틀림 변형 및 드로스 형성의 조합의 결과입니다 . 0° ≤  θ  ≤ 15° 인 경우 , 용융풀 경계의 작은 영역 만 분말 베드와 접촉하여 약간의 분말 접착이 발생하기 때문에 오버행 각도가 R a에 큰 영향을 주지 않습니다 . 15° < θ 일 때  ≤ 50°, 용융 풀 싱킹 및 용융 풀 경계와 분말 베드 사이의 증가된 접촉 면적으로 분말 접착력이 향상됩니다. θ  > 50° 일 때 오버행 윤곽의 큰 파형, 분말 클러스터의 접착, 심한 휨 변형 및 드 로스 형성이 Ra 급격히 증가 합니다.

    KEYWORDS: Laser powder bed fusion (L-PBF), melt pool dynamics, overhang region, shape deviation, surface roughness

    1. Introduction

    레이저 분말 베드 융합 (L-PBF)은 첨단 적층 제조 (AM) 기술로, 집중된 레이저 빔을 사용하여 금속 분말을 선택적으로 융합하여 슬라이스 된 3D 컴퓨터 지원에 따라 층별로 3 차원 (3D) 금속 부품을 구축합니다. 설계 (CAD) 모델 (Chatham, Long 및 Williams 2019 ; Tan, Zhu 및 Zhou 2020 ). 재료가 인쇄 층 아래에 ​​존재하는지 여부에 따라 인쇄 영역은 각각 솔리드 영역 또는 돌출 영역으로 분류 될 수 있습니다. 따라서 오버행 영역은 고체 기판이 아니라 분말 베드 바로 위에 건설되는 특수 구조입니다 (Patterson, Messimer 및 Farrington 2017). 오버행 영역은지지 구조를 포함하거나 포함하지 않고 구축 할 수 있으며, 지지대가있는 돌출 영역의 L-PBF는 지지체가 더 낮은 밀도로 구축된다는 점을 제외 하고 (Wang and Chou 2018 ) 고체 기판의 공정과 유사합니다 (따라서 기계적 강도가 낮기 때문에 L-PBF 공정 후 기계적으로 쉽게 제거 할 수 있습니다. 따라서지지 구조로 인쇄 된 오버행 영역은 L-PBF 공정 후 지지물 제거, 연삭 및 연마와 같은 추가 후 처리 단계가 필요합니다.

    수평 내부 채널의 제작과 같은 일부 특정 경우에는 공정 후 지지대를 제거하기가 어려우므로 채널 상단 절반의 돌출부 영역을 지지대없이 건설해야합니다 (Hopkinson and Dickens 2000 ). 수평 내부 채널에 사용할 수없는지지 구조 외에도 내부 표면, 특히 등각 냉각 채널 (Feng, Kamat 및 Pei 2021 ) 에서 발생하는 복잡한 3D 채널 네트워크의 경우 표면 마감 프로세스를 구현하는 것도 어렵습니다 . 결과적으로 오버행 영역은 (i) 잔류 응력에 의한 변형, (ii) 계단 효과 (Kuo et al. 2020 ; Li et al. 2020 )로 인해 설계된 모양에서 벗어날 수 있습니다 .) 및 (iii) 원하지 않는 분말 소결로 인한 향상된 표면 거칠기; 여기서, 앞의 두 요소는 일반적으로 mm 길이 스케일에서 ‘매크로’편차로 분류되고 후자는 일반적으로 µm 길이 스케일에서 ‘마이크로’편차로 인식됩니다.

    열 응력에 의한 변형은 오버행 영역에서 발생하는 중요한 문제입니다 (Patterson, Messimer 및 Farrington 2017 ). 국부적 인 용융 / 냉각은 용융 풀 내부 및 주변에서 큰 온도 구배를 유도하여 응고 된 층에 집중적 인 열 응력을 유발합니다. 열 응력에 의한 뒤틀림은 고체 영역을 현저하게 변형하지 않습니다. 이러한 영역은 아래의 여러 레이어에 의해 제한되기 때문입니다. 반면에 오버행 영역은 구속되지 않고 공정 중 응력 완화로 인해 상당한 변형이 발생합니다 (Kamat 및 Pei 2019 ). 더욱이 용융 깊이는 레이어 두께보다 큽니다 (이전 레이어도 재용 해되어 빌드 된 레이어간에 충분한 결합을 보장하기 때문입니다 [Yadroitsev et al. 2013 ; Kamath et al.2014 ]),응고 된 두께가 설계된 두께보다 크기 때문에형태 편차 (예 : 드 로스 [Charles et al. 2020 ; Feng et al. 2020 ])가 발생합니다. 마이크로 스케일에서 인쇄 된 표면 (R a 및 S a ∼ 10 μm)은 기계적으로 가공 된 표면보다 거칠다 (Duval-Chaneac et al. 2018 ; Wen et al. 2018 ). 이 문제는고형화 된 용융 풀의 가장자리에 부착 된 용융되지 않은 분말의 결과로 표면 거칠기 (R a )가 일반적으로 약 20 μm인 오버행 영역에서 특히 심각합니다 (Mazur et al. 2016 ; Pakkanen et al. 2016 ).

    오버행 각도 ( θ , 빌드 방향과 관련하여 측정)는 오버행 영역의 뒤틀림 편향과 표면 거칠기에 영향을 미치는 중요한 매개 변수입니다 (Kamat and Pei 2019 ; Mingear et al. 2019 ). θ ∼ 45 ° 의 오버행 각도 는 일반적으로지지 구조없이 오버행 영역을 인쇄 할 수있는 임계 값으로 합의됩니다 (Pakkanen et al. 2016 ; Kadirgama et al. 2018 ). θ 일 때이 임계 값보다 크면 오버행 영역을 허용 가능한 표면 품질로 인쇄 할 수 없습니다. 오버행 각도 외에도 레이저 매개 변수 (레이저 에너지 밀도와 관련된)는 용융 풀의 모양 / 크기 및 용융 풀 역학에 영향을줌으로써 오버행 영역의 표면 거칠기에 영향을줍니다 (Wang et al. 2013 ; Mingear et al . 2019 ).

    용융 풀 역학은 고체 (Shrestha 및 Chou 2018 ) 및 오버행 (Le et al. 2020 ) 영역 모두에서 수행되는 L-PBF 공정을 포함한 레이저 재료 가공의 일반적인 물리적 현상입니다 . 용융 풀 모양, 크기 및 냉각 속도는 잔류 응력으로 인한 변형과 ​​표면 거칠기에 모두 영향을 미치므로 처리 매개 변수와 표면 형태 / 품질 사이의 다리 역할을하며 용융 풀을 이해하기 위해 수치 시뮬레이션을 사용하여 추가 조사를 수행 할 수 있습니다. 거동과 표면 거칠기에 미치는 영향. 현재까지 고체 영역의 L-PBF 동안 용융 풀 동작을 시뮬레이션하기 위해 여러 연구가 수행되었습니다. 유한 요소 방법 (FEM)과 같은 시뮬레이션 기술 (Roberts et al. 2009 ; Du et al.2019 ), 유한 차분 법 (FDM) (Wu et al. 2018 ), 전산 유체 역학 (CFD) (Lee and Zhang 2016 ), 임의의 Lagrangian-Eulerian 방법 (ALE) (Khairallah and Anderson 2014 )을 사용하여 증발 반동 압력 (Hu et al. 2018 ) 및 Marangoni 대류 (Zhang et al. 2018 ) 현상을포함하는 열 전달 (온도 장) 및 물질 전달 (용융 흐름) 프로세스. 또한 이산 요소법 (DEM)을 사용하여 무작위 분산 분말 베드를 생성했습니다 (Lee and Zhang 2016 ; Wu et al. 2018 ). 이 모델은 분말 규모의 L-PBF 공정을 시뮬레이션했습니다 (Khairallah et al. 2016) 메조 스케일 (Khairallah 및 Anderson 2014 ), 단일 트랙 (Leitz et al. 2017 )에서 다중 트랙 (Foroozmehr et al. 2016 ) 및 다중 레이어 (Huang, Khamesee 및 Toyserkani 2019 )로.

    그러나 결과적인 표면 거칠기를 결정하는 오버행 영역의 용융 풀 역학은 문헌에서 거의 관심을받지 못했습니다. 솔리드 영역의 L-PBF에 대한 기존 시뮬레이션 모델이 어느 정도 참조가 될 수 있지만 오버행 영역과 솔리드 영역 간의 용융 풀 역학에는 상당한 차이가 있습니다. 오버행 영역에서 용융 금속은 분말 입자 사이의 틈새로 아래로 흘러 용융 풀이 다공성 분말 베드가 제공하는 약한 지지체 아래로 가라 앉습니다. 이것은 중력과 표면 장력의 영향이 용융 풀의 결과적인 모양 / 크기를 결정하는 데 중요하며, 결과적으로 오버행 영역의 마이크로 스케일 형태의 진화에 중요합니다. 또한 분말 입자 사이의 공극, 열 조건 (예 : 에너지 흡수,2019 ; Karimi et al. 2020 ; 노래와 영 2020 ). 표면 거칠기는 (마이크로) 형상 편차를 증가시킬뿐만 아니라 주기적 하중 동안 미세 균열의 시작 지점 역할을함으로써 기계적 강도를 저하시킵니다 (Günther et al. 2018 ). 오버행 영역의 높은 표면 거칠기는 (마이크로) 정확도 / 품질에 대한 엄격한 요구 사항이있는 부품 제조에서 L-PBF의 적용을 제한합니다.

    본 연구는 실험 및 시뮬레이션 연구를 사용하여 오버행 영역 (지지물없이 제작)의 미세 형상 편차 형성 메커니즘과 표면 거칠기의 기원을 체계적이고 포괄적으로 조사합니다. 결합 된 DEM-CFD 시뮬레이션 모델은 경계 트랙 윤곽, 분말 접착 및 뒤틀림 변형의 효과를 고려하여 오버행 영역의 용융 풀 역학과 표면 형태의 형성 메커니즘을 나타 내기 위해 개발되었습니다. 표면 거칠기 R의 시뮬레이션 및 단일 요인 L-PBF 인쇄 실험을 사용하여 오버행 각도의 함수로 연구됩니다. 용융 풀의 침몰과 관련된 오버행 영역에서 분말 접착의 세 가지 메커니즘이 식별되고 자세히 설명됩니다. 마지막으로, 인쇄 된 오버행 영역에서 높은 표면 거칠기 문제를 완화 할 수 있는 잠재적 솔루션에 대해 간략하게 설명합니다.

    The shape and size of the L-PBF printed samples are illustrated in Figure 1
    The shape and size of the L-PBF printed samples are illustrated in Figure 1
    Figure 2. Borders in the overhang region depending on the overhang angle θ
    Figure 2. Borders in the overhang region depending on the overhang angle θ
    Figure 3. (a) Profile of the volumetric heat source, (b) the model geometry of single-track printing on a solid substrate (unit: µm), and (c) the comparison of melt pool dimensions obtained from the experiment (right half) and simulation (left half) for a calibrated optical penetration depth of 110 µm (laser power 200 W and scan speed 800 mm/s, solidified layer thickness 30 µm, powder size 10–45 µm).
    Figure 3. (a) Profile of the volumetric heat source, (b) the model geometry of single-track printing on a solid substrate (unit: µm), and (c) the comparison of melt pool dimensions obtained from the experiment (right half) and simulation (left half) for a calibrated optical penetration depth of 110 µm (laser power 200 W and scan speed 800 mm/s, solidified layer thickness 30 µm, powder size 10–45 µm).
    Figure 4. The model geometry of an overhang being L-PBF processed: (a) 3D view and (b) right view.
    Figure 4. The model geometry of an overhang being L-PBF processed: (a) 3D view and (b) right view.
    Figure 5. The cross-sectional contour of border tracks in a 45° overhang region.
    Figure 5. The cross-sectional contour of border tracks in a 45° overhang region.
    Figure 6. Evolution of melt pool in the overhang region (θ = 45°, P = 100 W, v = 1000 mm/s, the streamlines are shown by arrows).
    Figure 6. Evolution of melt pool in the overhang region (θ = 45°, P = 100 W, v = 1000 mm/s, the streamlines are shown by arrows).
    Figure 7. The overhang contour is contributed by (a) only outer borders when θ ≤ 60° (b) both inner borders and outer borders when θ > 60°.
    Figure 7. The overhang contour is contributed by (a) only outer borders when θ ≤ 60° (b) both inner borders and outer borders when θ > 60°.
    Figure 8. Schematic of powder adhesion on a 45° overhang region.
    Figure 8. Schematic of powder adhesion on a 45° overhang region.
    Figure 9. The L-PBF printed samples with various overhang angle (a) θ = 0° (cube), (b) θ = 30°, (c) θ = 45°, (d) θ = 55° and (e) θ = 60°.
    Figure 9. The L-PBF printed samples with various overhang angle (a) θ = 0° (cube), (b) θ = 30°, (c) θ = 45°, (d) θ = 55° and (e) θ = 60°.
    Figure 10. Two mechanisms of powder adhesion related to the overhang angle: (a) simulation-predicted, θ = 45°; (b) simulation-predicted, θ = 60°; (c, e) optical micrographs, θ = 45°; (d, f) optical micrographs, θ = 60°. (e) and (f) are partial enlargement of (c) and (d), respectively.
    Figure 10. Two mechanisms of powder adhesion related to the overhang angle: (a) simulation-predicted, θ = 45°; (b) simulation-predicted, θ = 60°; (c, e) optical micrographs, θ = 45°; (d, f) optical micrographs, θ = 60°. (e) and (f) are partial enlargement of (c) and (d), respectively.
    Figure 11. Simulation-predicted surface morphology in the overhang region at different overhang angle: (a) θ = 15°, (b) θ = 30°, (c) θ = 45°, (d) θ = 60° and (e) θ = 80° (Blue solid lines: simulation-predicted contour; red dashed lines: the planar profile of designed overhang region specified by the overhang angles).
    Figure 11. Simulation-predicted surface morphology in the overhang region at different overhang angle: (a) θ = 15°, (b) θ = 30°, (c) θ = 45°, (d) θ = 60° and (e) θ = 80° (Blue solid lines: simulation-predicted contour; red dashed lines: the planar profile of designed overhang region specified by the overhang angles).
    Figure 12. Effect of overhang angle on surface roughness Ra in overhang regions
    Figure 12. Effect of overhang angle on surface roughness Ra in overhang regions
    Figure 13. Surface morphology of L-PBF printed overhang regions with different overhang angle: (a) θ = 15°, (b) θ = 30°, (c) θ = 45° and (d) θ = 60° (overhang border parameters: P = 100 W, v = 1000 mm/s).
    Figure 13. Surface morphology of L-PBF printed overhang regions with different overhang angle: (a) θ = 15°, (b) θ = 30°, (c) θ = 45° and (d) θ = 60° (overhang border parameters: P = 100 W, v = 1000 mm/s).
    Figure 14. Effect of (a) laser power (scan speed = 1000 mm/s) and (b) scan speed (lase power = 100 W) on surface roughness Ra in overhang regions (θ = 45°, laser power and scan speed referred to overhang border parameters, and the other process parameters are listed in Table 2).
    Figure 14. Effect of (a) laser power (scan speed = 1000 mm/s) and (b) scan speed (lase power = 100 W) on surface roughness Ra in overhang regions (θ = 45°, laser power and scan speed referred to overhang border parameters, and the other process parameters are listed in Table 2).

    References

    • Cai, Chao, Chrupcala Radoslaw, Jinliang Zhang, Qian Yan, Shifeng Wen, Bo Song, and Yusheng Shi. 2019. “In-Situ Preparation and Formation of TiB/Ti-6Al-4V Nanocomposite via Laser Additive Manufacturing: Microstructure Evolution and Tribological Behavior.” Powder Technology 342: 73–84. doi:10.1016/j.powtec.2018.09.088. [Crossref], [Web of Science ®], [Google Scholar]
    • Cai, Chao, Wei Shian Tey, Jiayao Chen, Wei Zhu, Xingjian Liu, Tong Liu, Lihua Zhao, and Kun Zhou. 2021. “Comparative Study on 3D Printing of Polyamide 12 by Selective Laser Sintering and Multi Jet Fusion.” Journal of Materials Processing Technology 288 (August 2020): 116882. doi:10.1016/j.jmatprotec.2020.116882. [Crossref], [Web of Science ®], [Google Scholar]
    • Cai, Chao, Xu Wu, Wan Liu, Wei Zhu, Hui Chen, Jasper Dong Qiu Chua, Chen Nan Sun, Jie Liu, Qingsong Wei, and Yusheng Shi. 2020. “Selective Laser Melting of Near-α Titanium Alloy Ti-6Al-2Zr-1Mo-1V: Parameter Optimization, Heat Treatment and Mechanical Performance.” Journal of Materials Science and Technology 57: 51–64. doi:10.1016/j.jmst.2020.05.004. [Crossref], [Web of Science ®], [Google Scholar]
    • Charles, Amal, Ahmed Elkaseer, Lore Thijs, and Steffen G. Scholz. 2020. “Dimensional Errors Due to Overhanging Features in Laser Powder Bed Fusion Parts Made of Ti-6Al-4V.” Applied Sciences 10 (7): 2416. doi:10.3390/app10072416. [Crossref], [Google Scholar]
    • Chatham, Camden A., Timothy E. Long, and Christopher B. Williams. 2019. “A Review of the Process Physics and Material Screening Methods for Polymer Powder Bed Fusion Additive Manufacturing.” Progress in Polymer Science 93: 68–95. doi:10.1016/j.progpolymsci.2019.03.003. [Crossref], [Web of Science ®], [Google Scholar]
    • Du, Yang, Xinyu You, Fengbin Qiao, Lijie Guo, and Zhengwu Liu. 2019. “A Model for Predicting the Temperature Field during Selective Laser Melting.” Results in Physics 12 (November 2018): 52–60. doi:10.1016/j.rinp.2018.11.031. [Crossref], [Web of Science ®], [Google Scholar]
    • Duval-Chaneac, M. S., S. Han, C. Claudin, F. Salvatore, J. Bajolet, and J. Rech. 2018. “Experimental Study on Finishing of Internal Laser Melting (SLM) Surface with Abrasive Flow Machining (AFM).” Precision Engineering 54 (July 2017): 1–6. doi:10.1016/j.precisioneng.2018.03.006. [Crossref], [Web of Science ®], [Google Scholar]
    • Feng, Shaochuan, Shijie Chen, Amar M. Kamat, Ru Zhang, Mingji Huang, and Liangcai Hu. 2020. “Investigation on Shape Deviation of Horizontal Interior Circular Channels Fabricated by Laser Powder Bed Fusion.” Additive Manufacturing 36 (December): 101585. doi:10.1016/j.addma.2020.101585. [Crossref], [Web of Science ®], [Google Scholar]
    • Feng, Shaochuan, Chuanzhen Huang, Jun Wang, Hongtao Zhu, Peng Yao, and Zhanqiang Liu. 2017. “An Analytical Model for the Prediction of Temperature Distribution and Evolution in Hybrid Laser-Waterjet Micro-Machining.” Precision Engineering 47: 33–45. doi:10.1016/j.precisioneng.2016.07.002. [Crossref], [Web of Science ®], [Google Scholar]
    • Feng, Shaochuan, Amar M. Kamat, and Yutao Pei. 2021. “Design and Fabrication of Conformal Cooling Channels in Molds: Review and Progress Updates.” International Journal of Heat and Mass Transfer. doi:10.1016/j.ijheatmasstransfer.2021.121082. [Crossref], [PubMed], [Web of Science ®], [Google Scholar]
    • Flow-3D V11.2 Documentation. 2016. Flow Science, Inc. [Crossref], [Google Scholar]
    • Foroozmehr, Ali, Mohsen Badrossamay, Ehsan Foroozmehr, and Sa’id Golabi. 2016. “Finite Element Simulation of Selective Laser Melting Process Considering Optical Penetration Depth of Laser in Powder Bed.” Materials and Design 89: 255–263. doi:10.1016/j.matdes.2015.10.002. [Crossref], [Web of Science ®], [Google Scholar]
    • “Geometrical Product Specifications (GPS) — Surface Texture: Profile Method — Rules and Procedures for the Assessment of Surface Texture (ISO 4288).” 1996. International Organization for Standardization. https://www.iso.org/standard/2096.html. [Google Scholar]
    • Günther, Johannes, Stefan Leuders, Peter Koppa, Thomas Tröster, Sebastian Henkel, Horst Biermann, and Thomas Niendorf. 2018. “On the Effect of Internal Channels and Surface Roughness on the High-Cycle Fatigue Performance of Ti-6Al-4V Processed by SLM.” Materials & Design 143: 1–11. doi:10.1016/j.matdes.2018.01.042. [Crossref], [Web of Science ®], [Google Scholar]
    • Hopkinson, Neil, and Phill Dickens. 2000. “Conformal Cooling and Heating Channels Using Laser Sintered Tools.” In Solid Freeform Fabrication Conference, 490–497. Texas. doi:10.26153/tsw/3075. [Crossref], [Google Scholar]
    • Hu, Zhiheng, Haihong Zhu, Changchun Zhang, Hu Zhang, Ting Qi, and Xiaoyan Zeng. 2018. “Contact Angle Evolution during Selective Laser Melting.” Materials and Design 139: 304–313. doi:10.1016/j.matdes.2017.11.002. [Crossref], [Web of Science ®], [Google Scholar]
    • Hu, Cheng, Kejia Zhuang, Jian Weng, and Donglin Pu. 2019. “Three-Dimensional Analytical Modeling of Cutting Temperature for Round Insert Considering Semi-Infinite Boundary and Non-Uniform Heat Partition.” International Journal of Mechanical Sciences 155 (October 2018): 536–553. doi:10.1016/j.ijmecsci.2019.03.019. [Crossref], [Web of Science ®], [Google Scholar]
    • Huang, Yuze, Mir Behrad Khamesee, and Ehsan Toyserkani. 2019. “A New Physics-Based Model for Laser Directed Energy Deposition (Powder-Fed Additive Manufacturing): From Single-Track to Multi-Track and Multi-Layer.” Optics & Laser Technology 109 (August 2018): 584–599. doi:10.1016/j.optlastec.2018.08.015. [Crossref], [Web of Science ®], [Google Scholar]
    • Kadirgama, K., W. S. W. Harun, F. Tarlochan, M. Samykano, D. Ramasamy, Mohd Zaidi Azir, and H. Mehboob. 2018. “Statistical and Optimize of Lattice Structures with Selective Laser Melting (SLM) of Ti6AL4V Material.” International Journal of Advanced Manufacturing Technology 97 (1–4): 495–510. doi:10.1007/s00170-018-1913-1. [Crossref], [Web of Science ®], [Google Scholar]
    • Kamat, Amar M, and Yutao Pei. 2019. “An Analytical Method to Predict and Compensate for Residual Stress-Induced Deformation in Overhanging Regions of Internal Channels Fabricated Using Powder Bed Fusion.” Additive Manufacturing 29 (March): 100796. doi:10.1016/j.addma.2019.100796. [Crossref], [Web of Science ®], [Google Scholar]
    • Kamath, Chandrika, Bassem El-Dasher, Gilbert F. Gallegos, Wayne E. King, and Aaron Sisto. 2014. “Density of Additively-Manufactured, 316L SS Parts Using Laser Powder-Bed Fusion at Powers up to 400 W.” International Journal of Advanced Manufacturing Technology 74 (1–4): 65–78. doi:10.1007/s00170-014-5954-9. [Crossref], [Web of Science ®], [Google Scholar]
    • Karimi, J., C. Suryanarayana, I. Okulov, and K. G. Prashanth. 2020. “Selective Laser Melting of Ti6Al4V: Effect of Laser Re-Melting.” Materials Science and Engineering A (July): 140558. doi:10.1016/j.msea.2020.140558. [Crossref], [Web of Science ®], [Google Scholar]
    • Khairallah, Saad A., and Andy Anderson. 2014. “Mesoscopic Simulation Model of Selective Laser Melting of Stainless Steel Powder.” Journal of Materials Processing Technology 214 (11): 2627–2636. doi:10.1016/j.jmatprotec.2014.06.001. [Crossref], [Web of Science ®], [Google Scholar]
    • Khairallah, Saad A., Andrew T. Anderson, Alexander Rubenchik, and Wayne E. King. 2016. “Laser Powder-Bed Fusion Additive Manufacturing: Physics of Complex Melt Flow and Formation Mechanisms of Pores, Spatter, and Denudation Zones.” Edited by Adedeji B. Badiru, Vhance V. Valencia, and David Liu. Acta Materialia 108 (April): 36–45. doi:10.1016/j.actamat.2016.02.014. [Crossref], [Web of Science ®], [Google Scholar]
    • Kuo, C. N., C. K. Chua, P. C. Peng, Y. W. Chen, S. L. Sing, S. Huang, and Y. L. Su. 2020. “Microstructure Evolution and Mechanical Property Response via 3D Printing Parameter Development of Al–Sc Alloy.” Virtual and Physical Prototyping 15 (1): 120–129. doi:10.1080/17452759.2019.1698967. [Taylor & Francis Online], [Web of Science ®], [Google Scholar]
    • Le, K. Q., C. H. Wong, K. H. G. Chua, C. Tang, and H. Du. 2020. “Discontinuity of Overhanging Melt Track in Selective Laser Melting Process.” International Journal of Heat and Mass Transfer 162 (December): 120284. doi:10.1016/j.ijheatmasstransfer.2020.120284. [Crossref], [Web of Science ®], [Google Scholar]
    • Lee, Y. S., and W. Zhang. 2016. “Modeling of Heat Transfer, Fluid Flow and Solidification Microstructure of Nickel-Base Superalloy Fabricated by Laser Powder Bed Fusion.” Additive Manufacturing 12: 178–188. doi:10.1016/j.addma.2016.05.003. [Crossref], [Web of Science ®], [Google Scholar]
    • Leitz, K. H., P. Singer, A. Plankensteiner, B. Tabernig, H. Kestler, and L. S. Sigl. 2017. “Multi-Physical Simulation of Selective Laser Melting.” Metal Powder Report 72 (5): 331–338. doi:10.1016/j.mprp.2016.04.004. [Crossref], [Google Scholar]
    • Li, Jian, Jing Hu, Yi Zhu, Xiaowen Yu, Mengfei Yu, and Huayong Yang. 2020. “Surface Roughness Control of Root Analogue Dental Implants Fabricated Using Selective Laser Melting.” Additive Manufacturing 34 (September 2019): 101283. doi:10.1016/j.addma.2020.101283. [Crossref], [Web of Science ®], [Google Scholar]
    • Li, Yingli, Kun Zhou, Pengfei Tan, Shu Beng Tor, Chee Kai Chua, and Kah Fai Leong. 2018. “Modeling Temperature and Residual Stress Fields in Selective Laser Melting.” International Journal of Mechanical Sciences 136 (February): 24–35. doi:10.1016/j.ijmecsci.2017.12.001. [Crossref], [Web of Science ®], [Google Scholar]
    • Mazur, MacIej, Martin Leary, Matthew McMillan, Joe Elambasseril, and Milan Brandt. 2016. “SLM Additive Manufacture of H13 Tool Steel with Conformal Cooling and Structural Lattices.” Rapid Prototyping Journal 22 (3): 504–518. doi:10.1108/RPJ-06-2014-0075. [Crossref], [Web of Science ®], [Google Scholar]
    • Mingear, Jacob, Bing Zhang, Darren Hartl, and Alaa Elwany. 2019. “Effect of Process Parameters and Electropolishing on the Surface Roughness of Interior Channels in Additively Manufactured Nickel-Titanium Shape Memory Alloy Actuators.” Additive Manufacturing 27 (October 2018): 565–575. doi:10.1016/j.addma.2019.03.027. [Crossref], [Web of Science ®], [Google Scholar]
    • Pakkanen, Jukka, Flaviana Calignano, Francesco Trevisan, Massimo Lorusso, Elisa Paola Ambrosio, Diego Manfredi, and Paolo Fino. 2016. “Study of Internal Channel Surface Roughnesses Manufactured by Selective Laser Melting in Aluminum and Titanium Alloys.” Metallurgical and Materials Transactions A 47 (8): 3837–3844. doi:10.1007/s11661-016-3478-7. [Crossref], [Web of Science ®], [Google Scholar]
    • Patterson, Albert E., Sherri L. Messimer, and Phillip A. Farrington. 2017. “Overhanging Features and the SLM/DMLS Residual Stresses Problem: Review and Future Research Need.” Technologies 5 (4): 15. doi:10.3390/technologies5020015. [Crossref], [Web of Science ®], [Google Scholar]
    • Roberts, I. A., C. J. Wang, R. Esterlein, M. Stanford, and D. J. Mynors. 2009. “A Three-Dimensional Finite Element Analysis of the Temperature Field during Laser Melting of Metal Powders in Additive Layer Manufacturing.” International Journal of Machine Tools and Manufacture 49 (12–13): 916–923. doi:10.1016/j.ijmachtools.2009.07.004. [Crossref], [Web of Science ®], [Google Scholar]
    • Shrestha, Subin, and Kevin Chou. 2018. “Computational Analysis of Thermo-Fluid Dynamics with Metallic Powder in SLM.” In CFD Modeling and Simulation in Materials Processing 2018, edited by Laurentiu Nastac, Koulis Pericleous, Adrian S. Sabau, Lifeng Zhang, and Brian G. Thomas, 85–95. Cham, Switzerland: Springer Nature. doi:10.1007/978-3-319-72059-3_9. [Crossref], [Google Scholar]
    • Sing, S. L., and W. Y. Yeong. 2020. “Laser Powder Bed Fusion for Metal Additive Manufacturing: Perspectives on Recent Developments.” Virtual and Physical Prototyping 15 (3): 359–370. doi:10.1080/17452759.2020.1779999. [Taylor & Francis Online], [Web of Science ®], [Google Scholar]
    • Šmilauer, Václav, Emanuele Catalano, Bruno Chareyre, Sergei Dorofeenko, Jérôme Duriez, Nolan Dyck, Jan Eliáš, et al. 2015. Yade Documentation. 2nd ed. The Yade Project. doi:10.5281/zenodo.34073. [Crossref], [Google Scholar]
    • Tan, Pengfei, Fei Shen, Biao Li, and Kun Zhou. 2019. “A Thermo-Metallurgical-Mechanical Model for Selective Laser Melting of Ti6Al4V.” Materials & Design 168 (April): 107642. doi:10.1016/j.matdes.2019.107642. [Crossref], [Web of Science ®], [Google Scholar]
    • Tan, Lisa Jiaying, Wei Zhu, and Kun Zhou. 2020. “Recent Progress on Polymer Materials for Additive Manufacturing.” Advanced Functional Materials 30 (43): 1–54. doi:10.1002/adfm.202003062. [Crossref], [Web of Science ®], [Google Scholar]
    • Wang, Xiaoqing, and Kevin Chou. 2018. “Effect of Support Structures on Ti-6Al-4V Overhang Parts Fabricated by Powder Bed Fusion Electron Beam Additive Manufacturing.” Journal of Materials Processing Technology 257 (February): 65–78. doi:10.1016/j.jmatprotec.2018.02.038. [Crossref], [Web of Science ®], [Google Scholar]
    • Wang, Di, Yongqiang Yang, Ziheng Yi, and Xubin Su. 2013. “Research on the Fabricating Quality Optimization of the Overhanging Surface in SLM Process.” International Journal of Advanced Manufacturing Technology 65 (9–12): 1471–1484. doi:10.1007/s00170-012-4271-4. [Crossref], [Web of Science ®], [Google Scholar]
    • Wen, Peng, Maximilian Voshage, Lucas Jauer, Yanzhe Chen, Yu Qin, Reinhart Poprawe, and Johannes Henrich Schleifenbaum. 2018. “Laser Additive Manufacturing of Zn Metal Parts for Biodegradable Applications: Processing, Formation Quality and Mechanical Properties.” Materials and Design 155: 36–45. doi:10.1016/j.matdes.2018.05.057. [Crossref], [Web of Science ®], [Google Scholar]
    • Wu, Yu-che, Cheng-hung San, Chih-hsiang Chang, Huey-jiuan Lin, Raed Marwan, Shuhei Baba, and Weng-Sing Hwang. 2018. “Numerical Modeling of Melt-Pool Behavior in Selective Laser Melting with Random Powder Distribution and Experimental Validation.” Journal of Materials Processing Technology 254 (November 2017): 72–78. doi:10.1016/j.jmatprotec.2017.11.032. [Crossref], [Web of Science ®], [Google Scholar]
    • Yadroitsev, I., P. Krakhmalev, I. Yadroitsava, S. Johansson, and I. Smurov. 2013. “Energy Input Effect on Morphology and Microstructure of Selective Laser Melting Single Track from Metallic Powder.” Journal of Materials Processing Technology 213 (4): 606–613. doi:10.1016/j.jmatprotec.2012.11.014. [Crossref], [Web of Science ®], [Google Scholar]
    • Yu, Wenhui, Swee Leong Sing, Chee Kai Chua, and Xuelei Tian. 2019. “Influence of Re-Melting on Surface Roughness and Porosity of AlSi10Mg Parts Fabricated by Selective Laser Melting.” Journal of Alloys and Compounds 792: 574–581. doi:10.1016/j.jallcom.2019.04.017. [Crossref], [Web of Science ®], [Google Scholar]
    • Zhang, Dongyun, Pudan Zhang, Zhen Liu, Zhe Feng, Chengjie Wang, and Yanwu Guo. 2018. “Thermofluid Field of Molten Pool and Its Effects during Selective Laser Melting (SLM) of Inconel 718 Alloy.” Additive Manufacturing 21 (100): 567–578. doi:10.1016/j.addma.2018.03.031. [Crossref], [Web of Science ®], [Google Scholar]
    Dam-Break Flows: Comparison between Flow-3D, MIKE 3 FM, and Analytical Solutions with Experimental Data

    Dam-Break Flows: Comparison between Flow-3D, MIKE 3 FM, and Analytical Solutions with Experimental Data

    by Hui Hu,Jianfeng Zhang andTao Li *
    State Key Laboratory Base of Eco-Hydraulic Engineering in Arid Area, School of Water Resources and Hydropower, Xi’an University of Technology, Xi’an 710048, China
    *Author to whom correspondence should be addressed.
    Appl. Sci.20188(12), 2456; https://doi.org/10.3390/app8122456Received: 14 October 2018 /
    Revised: 20 November 2018 / Accepted: 29 November 2018 / Published: 2 December 2018

    Abstract

    The objective of this study was to evaluate the applicability of a flow model with different numbers of spatial dimensions in a hydraulic features solution, with parameters such a free surface profile, water depth variations, and averaged velocity evolution in a dam-break under dry and wet bed conditions with different tailwater depths. Two similar three-dimensional (3D) hydrodynamic models (Flow-3D and MIKE 3 FM) were studied in a dam-break simulation by performing a comparison with published experimental data and the one-dimensional (1D) analytical solution. The results indicate that the Flow-3D model better captures the free surface profile of wavefronts for dry and wet beds than other methods. The MIKE 3 FM model also replicated the free surface profiles well, but it underestimated them during the initial stage under wet-bed conditions. However, it provided a better approach to the measurements over time. Measured and simulated water depth variations and velocity variations demonstrate that both of the 3D models predict the dam-break flow with a reasonable estimation and a root mean square error (RMSE) lower than 0.04, while the MIKE 3 FM had a small memory footprint and the computational time of this model was 24 times faster than that of the Flow-3D. Therefore, the MIKE 3 FM model is recommended for computations involving real-life dam-break problems in large domains, leaving the Flow-3D model for fine calculations in which knowledge of the 3D flow structure is required. The 1D analytical solution was only effective for the dam-break wave propagations along the initially dry bed, and its applicability was fairly limited. 

    Keywords: dam breakFlow-3DMIKE 3 FM1D Ritter’s analytical solution

    이 연구의 목적은 자유 표면 프로파일, 수심 변화 및 건식 및 댐 파괴에서 평균 속도 변화와 같은 매개 변수를 사용하여 유압 기능 솔루션에서 서로 다른 수의 공간 치수를 가진 유동 모델의 적용 가능성을 평가하는 것이었습니다.

    테일 워터 깊이가 다른 습식베드 조건. 2 개의 유사한 3 차원 (3D) 유체 역학 모델 (Flow-3D 및 MIKE 3 FM)이 게시된 실험 데이터와 1 차원 (1D) 분석 솔루션과의 비교를 수행하여 댐 브레이크 시뮬레이션에서 연구되었습니다.

    결과는 FLOW-3D 모델이 다른 방법보다 건식 및 습식 베드에 대한 파면의 자유 표면 프로파일을 더 잘 포착함을 나타냅니다. MIKE 3 FM 모델도 자유 표면 프로파일을 잘 복제했지만, 습식 조건에서 초기 단계에서 과소 평가했습니다. 그러나 시간이 지남에 따라 측정에 더 나은 접근 방식을 제공했습니다.

    측정 및 시뮬레이션 된 수심 변화와 속도 변화는 두 3D 모델 모두 합리적인 추정치와 0.04보다 낮은 RMSE (root mean square error)로 댐 브레이크 흐름을 예측하는 반면 MIKE 3 FM은 메모리 공간이 적고 이 모델의 계산 시간은 Flow-3D보다 24 배 더 빠릅니다.

    따라서 MIKE 3 FM 모델은 대규모 도메인의 실제 댐 브레이크 문제와 관련된 계산에 권장되며 3D 흐름 구조에 대한 지식이 필요한 미세 계산을 위해 Flow-3D 모델을 남겨 둡니다. 1D 분석 솔루션은 초기 건조 층을 따라 전파되는 댐 파괴에만 효과적이었으며 그 적용 가능성은 상당히 제한적이었습니다.

    1. Introduction

    저수지에 저장된 물의 통제되지 않은 방류[1]로 인해 댐 붕괴와 그로 인해 하류에서 발생할 수 있는 잠재적 홍수로 인해 큰 자연 위험이 발생한다. 이러한 영향을 최대한 완화하기 위해서는 홍수[2]로 인한 위험을 관리하고 감소시키기 위해 홍수의 시간적 및 공간적 진화를 모두 포착하여 댐 붕괴 파동의 움직임을 예측하고 댐 붕괴 파동의 전파 과정 효과를 다운스트림[3]으로 예측하는 것이 중요하다. 

    그러나 이러한 수량을 예측하는 것은 어려운 일이며, 댐 붕괴 홍수의 움직임을 정확하게 시뮬레이션하고 유동장에 대한 유용한 정보를 제공하기 위한 적절한 모델을 선택하는 것은 그러므로 필수적인 단계[4]이다.

    적절한 수학적 및 수치적 모델의 선택은 댐 붕괴 홍수 분석에서 매우 중요한 것으로 나타났다.분석적 해결책에서 행해진 댐 붕괴 흐름에 대한 연구는 100여 년 전에 시작되었다. 

    리터[5]는 먼저 건조한 침대 위에 1D de 생베넌트 방정식의 초기 분석 솔루션을 도출했고, 드레슬러[6,7]와 휘담[8]은 마찰저항의 영향을 받은 파동학을 연구했으며, 스토커[9]는 젖은 침대를 위한 1D 댐 붕괴 문제에 리터의 솔루션을 확장했다. 

    마샬과 멩데즈[10]는 고두노프가 가스 역학의 오일러 방정식을 위해 개발한 방법론[11]을 적용하여 젖은 침대 조건에서 리만 문제를 해결하기 위한 일반적인 절차를 고안했다. Toro [12]는 습식 및 건식 침대 조건을 모두 해결하기 위해 완전한 1D 정밀 리만 용해제를 실시했다. 

    Chanson [13]은 특성 방법을 사용하여 갑작스러운 댐 붕괴로 인한 홍수에 대한 간단한 분석 솔루션을 연구했다. 그러나 이러한 분석 솔루션은 특히 댐 붕괴 초기 단계에서 젖은 침대의 정확한 결과를 도출하지 못했다[14,15].과거 연구의 발전은 이른바 댐 붕괴 홍수 문제 해결을 위한 여러 수치 모델[16]을 제공했으며, 헥-라스, DAMBRK, MIK 11 등과 같은 1차원 모델을 댐 붕괴 홍수를 모델링하는 데 사용하였다.

    [17 2차원(2D) 깊이 평균 방정식도 댐 붕괴 흐름 문제를 시뮬레이션하는 데 널리 사용되어 왔으며[18,19,20,21,22] 그 결과 천수(shallow water) 방정식(SWE)이 유체 흐름을 나타내는 데 적합하다는 것을 알 수 있다. 그러나, 경우에 따라 2D 수치해결기가 제공하는 해결책이 특히 근거리 분야에서 실험과 일관되지 않을 수 있다[23,24]. 더욱이, 1차원 및 2차원 모델은 3차원 현상에 대한 일부 세부사항을 포착하는 데 한계가 있다.

    [25]. RANS(Reynolds-averageed Navier-Stok크스 방정식)에 기초한 여러 3차원(3D) 모델이 천수(shallow water) 모델의 일부 단점을 극복하기 위해 적용되었으며, 댐 붕괴 초기 단계에서의 복잡한 흐름의 실제 동작을 이해하기 위해 사용되었다 [26,27,28]장애물이나 바닥 실에 대한 파장의 충격으로 인한 튜디 댐 붕괴 흐름 [19,29] 및 근거리 영역의 난류 댐 붕괴 흐름 거동 [4] 최근 상용화된 수치 모델 중 잘 알려진 유체 방식(VOF) 기반 CFD 모델링 소프트웨어 FLOW-3D는 컴퓨터 기술의 진보에 따른 계산력 증가로 인해 불안정한 자유 표면 흐름을 분석하는 데 널리 사용되고 있다. 

    이 소프트웨어는 유한 차이 근사치를 사용하여 RANS 방정식에 대한 수치 해결책을 계산하며, 자유 표면을 추적하기 위해 VOF를 사용한다 [30,31]; 댐 붕괴 흐름을 모델링하는 데 성공적으로 사용되었다 [32,33].그러나, 2D 천수(shallow water) 모델을 사용하여 포착할 수 없는 공간과 시간에 걸친 댐 붕괴 흐름의 특정한 유압적 특성이 있다. 

    실생활 현장 척도 시뮬레이션을 위한 완전한 3D Navier-Stokes 방정식의 적용은 더 높은 계산 비용[34]을 가지고 있으며, 원하는 결과는 천수(shallow water) 모델[35]보다 더 정확한 결과를 산출하지 못할 수 있다. 따라서, 본 논문은 3D 모델의 기능과 그 계산 효율을 평가하기 위해 댐 붕괴 흐름 시뮬레이션을 위한 단순화된 3D 모델-MIKE 3 FM을 시도한다. 

    MIK 3 모델은 자연 용수 분지의 여러 유체 역학 시뮬레이션 조사에 적용되었다. 보치 외 연구진이 사용해 왔다. [36], 니콜라오스 및 게오르기오스 [37], 고얄과 라토드[38] 등 현장 연구에서 유체역학 시뮬레이션을 위한 것이다. 이러한 저자들의 상당한 연구에도 불구하고, MIK 3 FM을 이용한 댐 붕괴의 모델링에 관한 연구는 거의 없었다. 

    또한 댐 붕괴 홍수 전파 문제를 해결하기 위한 3D 천수(shallow water)과 완전한 3D RANS 모델의 성능을 비교한 연구도 아직 보고되지 않았다. 이 공백을 메우기 위해 현재 연구의 주요 목표는 댐 붕괴 흐름을 시뮬레이션하기 위한 단순화된 3D SWE, 상세 RANS 모델 및 분석 솔루션을 평가하여 댐 붕괴 문제에 대한 정확도와 적용 가능성을 평가하는 것이다.실제 댐 붕괴 문제를 해결하기 위해 유체역학 시뮬레이션을 시도하기 전에 수치 모델을 검증할 필요가 있다. 

    일련의 실험 벤치마크를 사용하여 수치 모델을 확인하는 것은 용인된 관행이다. 현장 데이터 확보가 어려워 최근 몇 년 동안 제한된 측정 데이터를 취득했다. 

    본 논문은 Ozmen-Cagatay와 Kocaman[30] 및 Khankandi 외 연구진이 제안한 두 가지 테스트 사례에 의해 제안된 검증에서 인용한 것이다. [39] 오즈멘-카가테이와 코카만[30]이 수행한 첫 번째 실험에서, 다른 미숫물 수위에 걸쳐 초기 단계 동안 댐 붕괴 홍수파가 발생했으며, 자유 지표면 프로파일의 측정치를 제공했다. Ozmen-Cagatay와 Kocaman[30]은 초기 단계에서 Flow-3D 소프트웨어가 포함된 2D SWE와 3D RANS의 숫자 솔루션에 의해 계산된 자유 표면 프로필만 비교했다. 

    Khankandi 등이 고안한 두 번째 실험 동안. [39], 이 실험의 측정은 홍수 전파를 시뮬레이션하고 측정된 데이터를 제공하는 것을 목적으로 하는 수치 모델을 검증하기 위해 사용되었으며, 말기 동안의 자유 표면 프로필, 수위의 시간 진화 및 속도 변화를 포함한다. Khankandi 등의 연구. [39] 주로 실험 조사에 초점을 맞추었으며, 초기 단계에서는 리터의 솔루션과의 수위만을 언급하고 있다.

    경계 조건(상류 및 하류 모두 무한 채널 길이를 갖는 1D 분석 솔루션에서는 실험 결과를 리터와 비교하는 것이 타당하지 않기 때문이다(건조 be)d) 또는 스토커(웨트 베드) 솔루션은 벽의 반사가 깊이 프로파일에 영향을 미쳤을 때, 그리고 참조 [39]의 실험에 대한 수치 시뮬레이션과의 추가 비교가 불량할 때. 이 논문은 이러한 문제를 직접 겨냥하여 전체 댐 붕괴 과정에서의 자유 표면 프로필, 수심 변화 및 속도 변화에 대한 완전한 비교 연구를 제시한다. 

    여기서 댐 붕괴파의 수치 시뮬레이션은 초기에 건조하고 습한 직사각형 채널을 가진 유한 저장소의 순간 댐 붕괴에 대해 두 개의 3D 모델을 사용하여 개발된다.본 논문은 다음과 같이 정리되어 있다. 두 모델에 대한 통치 방정식은 숫자 체계를 설명하기 전에 먼저 도입된다. 

    일반적인 단순화된 시험 사례는 3D 수치 모델과 1D 분석 솔루션을 사용하여 시뮬레이션했다. 모델 결과와 이들이 실험실 실험과 비교하는 방법이 논의되고, 서로 다른 수심비에서 시간에 따른 유압 요소의 변동에 대한 시뮬레이션 결과가 결론을 도출하기 전에 제시된다.

    2. Materials and Methods

    2.1. Data

    첫째, 수평 건조 및 습식 침상에 대한 초기 댐 붕괴 단계 동안의 자유 표면 프로필 측정은 Ozmen-Cagatay와 Kocaman에 의해 수행되었다[30]. 이 시험 동안, 매끄럽고 직사각형의 수평 채널은 그림 1에서 표시한 대로 너비 0.30m, 높이 0.30m, 길이 8.9m이었다. 

    채널은 채널 입구에서 4.65m 떨어진 수직 플레이트(담) 즉, 저장소의 길이 L0=4.65mL0에 의해 분리되었다., 및 다운스트림 채널 L1=4.25 mL1. m저수지는 댐의 좌측에 위치하고 처음에는 침수된 것으로 간주되었다; 저수지의 초기 상류 수심 h0 0.25m로 일정했다.

    오른쪽의 초기 수심 h1h1 건식침대의 경우 0m, 습식침대의 경우 0.025m, 0.1m이므로 수심비 α=h1/h0α으로 세 가지 상황이 있었다. 0, 0.1, 0.4의 습식침대 조건은 플룸 끝에 낮은 보를 사용함으로써 만들어졌다. 물 표면 프로필은 3개의 고속 디지털 카메라(50프레임/s)를 사용하여 초기에 관찰되었으며, 계측 측정의 정확도는 참고문헌 [30]에서 입증되었다. In the following section, the corresponding numerical results refer to positions x = −1 m (P1), −0.5 m (P2), −0.2 m (P3), +0.2 m (P4), +0.5 m (P5), +1 m (P6), +2 m (P7), and +2.85 m (P8), where the origin of the coordinate system x = 0 is at the dam site. 3수심비 ααα 0, 0.1, 0.4의 경우 x,yx의 경우 좌표는 h0.으로 정규화된다.

    <중략> ……

    Figure 1. Schematic view of the experimental conditions by Ozmen-Cagatay and Kocaman [30]: (a) α = 0; (b) α = 0.1; and (c) α = 0.4.
    Figure 1. Schematic view of the experimental conditions by Ozmen-Cagatay and Kocaman [30]: (a) α = 0; (b) α = 0.1; and (c) α = 0.4.

    Figure 2. Schematic view of the experimental conditions by Khankandi et al. [39]: (a) α = 0 and (b) α = 0.2.
    Figure 2. Schematic view of the experimental conditions by Khankandi et al. [39]: (a) α = 0 and (b) α = 0.2.
    Figure 3. Typical profiles of the dam-break flow regimes for Stoker’s analytical solution [9]: Wet-bed downstream
    Figure 3. Typical profiles of the dam-break flow regimes for Stoker’s analytical solution [9]: Wet-bed downstream
    Figure 4. Sensitivity analysis of the numerical simulation using Flow-3D for the different mesh sizes of the experiments in Reference [30].
    Figure 4. Sensitivity analysis of the numerical simulation using Flow-3D for the different mesh sizes of the experiments in Reference [30].
    Figure 5. Sensitivity analysis of the numerical simulation using MIKE 3 FM for the different mesh sizes of the experiments in Reference [30].
    Figure 5. Sensitivity analysis of the numerical simulation using MIKE 3 FM for the different mesh sizes of the experiments in Reference [30].
    Figure 6. Comparison between observed and simulated free surface profiles at dimensionless times T = t(g/h0)1/2 and for dry-bed (α=0). The experimental data are from Reference [30].
    Figure 6. Comparison between observed and simulated free surface profiles at dimensionless times T = t(g/h0)1/2 and for dry-bed (α=0). The experimental data are from Reference [30].
    Figure 7. Comparison between observed and simulated free surface profiles at dimensionless times T = t(g/h0)1/2 and for a wet-bed (α = 0.1). The experimental data are from Reference [30].
    Figure 7. Comparison between observed and simulated free surface profiles at dimensionless times T = t(g/h0)1/2 and for a wet-bed (α = 0.1). The experimental data are from Reference [30].
    Figure 8. Comparison between observed and simulated free surface profiles at dimensionless times T = t(g/h0)1/2 and for the wet-bed (α = 0.4). The experimental data are from Reference [30].
    Figure 8. Comparison between observed and simulated free surface profiles at dimensionless times T = t(g/h0)1/2 and for the wet-bed (α = 0.4). The experimental data are from Reference [30].
    Figure 9. Experimental and numerical comparison of free surface profiles h/h0(x/h0) during late stages at various dimensionless times T after the failure in the dry-bed by Khankandi et al. [39].
    Figure 9. Experimental and numerical comparison of free surface profiles h/h0(x/h0) during late stages at various dimensionless times T after the failure in the dry-bed by Khankandi et al. [39].

    Table 2. RMSE values for the free surface profiles observed by Khankandi et al. [39].

    Table 2. RMSE values for the free surface profiles observed by Khankandi et al. [39].
    Table 2. RMSE values for the free surface profiles observed by Khankandi et al. [39].
    Figure 10. Measured and computed water level hydrograph at various positions for dry-bed by Khankandi et al. [39]: (a) G1 (−0.5 m); (b) G2 (−0.1 m); (c) G3 (0.1 m); (d) G4 (0.8 m); (e) G6 (1.2 m); (f) G8 (5.5 m).
    Figure 10. Measured and computed water level hydrograph at various positions for dry-bed by Khankandi et al. [39]: (a) G1 (−0.5 m); (b) G2 (−0.1 m); (c) G3 (0.1 m); (d) G4 (0.8 m); (e) G6 (1.2 m); (f) G8 (5.5 m).
    Figure 11. Measured and computed water level hydrographs at various positions for the wet-bed by Khankandi et al. [39]: (a) G1 (−0.5 m); (b) G2 (−0.1 m); (c) G4 (0.8 m); and (d) G5 (1.0 m).
    Figure 11. Measured and computed water level hydrographs at various positions for the wet-bed by Khankandi et al. [39]: (a) G1 (−0.5 m); (b) G2 (−0.1 m); (c) G4 (0.8 m); and (d) G5 (1.0 m).

    Table 3. RMSE values for the water depth variations observed by Khankandi et al. [39] at the late stage.

    Table 3. RMSE values for the water depth variations observed by Khankandi et al. [39] at the late stage.
    Table 3. RMSE values for the water depth variations observed by Khankandi et al. [39] at the late stage.
    Figure 13. Comparison of simulated velocity profiles at various locations upstream and downstream of the dam at t = 0.8 s, 2 s, and 5 s for water depth ratios α = 0.1 by Ozmen-Cagatay and Kocaman [30]: (a) P1(−1 m); (b) P3 (+0.2 m); (c) P5 (+1 m); and (d) P6 (+2 m).
    Figure 13. Comparison of simulated velocity profiles at various locations upstream and downstream of the dam at t = 0.8 s, 2 s, and 5 s for water depth ratios α = 0.1 by Ozmen-Cagatay and Kocaman [30]: (a) P1(−1 m); (b) P3 (+0.2 m); (c) P5 (+1 m); and (d) P6 (+2 m).
    Table 5. The required computational time for the two models to address dam break flows in all cases
    Table 5. The required computational time for the two models to address dam break flows in all cases

    References

    1. Gallegos, H.A.; Schubert, J.E.; Sanders, B.F. Two-dimensional high-resolution modeling of urban dam-break flooding: A case study of Baldwin Hills, California. Adv. Water Resour. 200932, 1323–1335. [Google Scholar] [CrossRef]
    2. Kim, K.S. A Mesh-Free Particle Method for Simulation of Mobile-Bed Behavior Induced by Dam Break. Appl. Sci. 20188, 1070. [Google Scholar] [CrossRef]
    3. Robb, D.M.; Vasquez, J.A. Numerical simulation of dam-break flows using depth-averaged hydrodynamic and three-dimensional CFD models. In Proceedings of the Canadian Society for Civil Engineering Hydrotechnical Conference, Québec, QC, Canada, 21–24 July 2015. [Google Scholar]
    4. LaRocque, L.A.; Imran, J.; Chaudhry, M.H. 3D numerical simulation of partial breach dam-break flow using the LES and k-ε. J. Hydraul. Res. 201351, 145–157. [Google Scholar] [CrossRef]
    5. Ritter, A. Die Fortpflanzung der Wasserwellen (The propagation of water waves). Z. Ver. Dtsch. Ing. 189236, 947–954. [Google Scholar]
    6. Dressler, R.F. Hydraulic resistance effect upon the dam-break functions. J. Res. Nat. Bur. Stand. 195249, 217–225. [Google Scholar] [CrossRef]
    7. Dressler, R.F. Comparison of theories and experiments for the hydraulic dam-break wave. Int. Assoc. Sci. Hydrol. 195438, 319–328. [Google Scholar]
    8. Whitham, G.B. The effects of hydraulic resistance in the dam-break problem. Proc. R. Soc. Lond. 1955227A, 399–407. [Google Scholar] [CrossRef]
    9. Stoker, J.J. Water Waves: The Mathematical Theory with Applications; Wiley and Sons: New York, NY, USA, 1957; ISBN 0-471-57034-6. [Google Scholar]
    10. Marshall, G.; Méndez, R. Computational Aspects of the Random Choice Method for Shallow Water Equations. J. Comput. Phys. 198139, 1–21. [Google Scholar] [CrossRef]
    11. Godunov, S.K. Finite Difference Methods for the Computation of Discontinuous Solutions of the Equations of Fluid Dynamics. Math. Sb. 195947, 271–306. [Google Scholar]
    12. Toro, E.F. Shock-Capturing Methods for Free-Surface Shallow Flows; Wiley and Sons Ltd.: New York, NY, USA, 2001. [Google Scholar]
    13. Chanson, H. Application of the method of characteristics to the dam break wave problem. J. Hydraul. Res. 200947, 41–49. [Google Scholar] [CrossRef][Green Version]
    14. Cagatay, H.; Kocaman, S. Experimental Study of Tail Water Level Effects on Dam-Break Flood Wave Propagation; 2008 Kubaba Congress Department and Travel Services: Ankara, Turkey, 2008; pp. 635–644. [Google Scholar]
    15. Stansby, P.K.; Chegini, A.; Barnes, T.C.D. The initial stages of dam-break flow. J. Fluid Mech. 1998374, 407–424. [Google Scholar] [CrossRef]
    16. Soares-Frazao, S.; Zech, Y. Dam Break in Channels with 90° Bend. J. Hydraul. Eng. 2002128, 956–968. [Google Scholar] [CrossRef]
    17. Zolghadr, M.; Hashemi, M.R.; Zomorodian, S.M.A. Assessment of MIKE21 model in dam and dike-break simulation. IJST-Trans. Mech. Eng. 201135, 247–262. [Google Scholar]
    18. Bukreev, V.I.; Gusev, A.V. Initial stage of the generation of dam-break waves. Dokl. Phys. 200550, 200–203. [Google Scholar] [CrossRef]
    19. Soares-Frazao, S.; Noel, B.; Zech, Y. Experiments of dam-break flow in the presence of obstacles. Proc. River Flow 20042, 911–918. [Google Scholar]
    20. Aureli, F.; Maranzoni, A.; Mignosa, P.; Ziveri, C. Dambreak flows: Acquisition of experimental data through an imaging technique and 2D numerical modelling. J. Hydraul. Eng. 2008134, 1089–1101. [Google Scholar] [CrossRef]
    21. Rehman, K.; Cho, Y.S. Bed Evolution under Rapidly Varying Flows by a New Method for Wave Speed Estimation. Water 20168, 212. [Google Scholar] [CrossRef]
    22. Wu, G.F.; Yang, Z.H.; Zhang, K.F.; Dong, P.; Lin, Y.T. A Non-Equilibrium Sediment Transport Model for Dam Break Flow over Moveable Bed Based on Non-Uniform Rectangular Mesh. Water 201810, 616. [Google Scholar] [CrossRef]
    23. Ferrari, A.; Fraccarollo, L.; Dumbser, M.; Toro, E.F.; Armanini, A. Three-dimensional flow evolution after a dam break. J. Fluid Mech. 2010663, 456–477. [Google Scholar] [CrossRef]
    24. Liang, D. Evaluating shallow water assumptions in dam-break flows. Proc. Inst. Civ. Eng. Water Manag. 2010163, 227–237. [Google Scholar] [CrossRef]
    25. Biscarini, C.; Francesco, S.D.; Manciola, P. CFD modelling approach for dam break flow studies. Hydrol. Earth Syst. Sci. 201014, 705–718. [Google Scholar] [CrossRef][Green Version]
    26. Oertel, M.; Bung, D.B. Initial stage of two-dimensional dam-break waves: Laboratory versus VOF. J. Hydraul. Res. 201250, 89–97. [Google Scholar] [CrossRef]
    27. Quecedo, M.; Pastor, M.; Herreros, M.I.; Merodo, J.A.F.; Zhang, Q. Comparison of two mathematical models for solving the dam break problem using the FEM method. Comput. Method Appl. Mech. Eng. 2005194, 3984–4005. [Google Scholar] [CrossRef]
    28. Shigematsu, T.; Liu, P.L.F.; Oda, K. Numerical modeling of the initial stages of dam-break waves. J. Hydraul. Res. 200442, 183–195. [Google Scholar] [CrossRef]
    29. Soares-Frazao, S. Experiments of dam-break wave over a triangular bottom sill. J. Hydraul. Res. 200745, 19–26. [Google Scholar] [CrossRef]
    30. Ozmen-Cagatay, H.; Kocaman, S. Dam-break flows during initial stage using SWE and RANS approaches. J. Hydraul. Res. 201048, 603–611. [Google Scholar] [CrossRef]
    31. Vasquez, J.; Roncal, J. Testing River2D and FLOW-3D for Sudden Dam-Break Flow Simulations. In Proceedings of the Canadian Dam Association’s 2009 Annual Conference: Protecting People, Property and the Environment, Whistler, BC, Canada, 3–8 October 2009. [Google Scholar]
    32. Ozmen-Cagatay, H.; Kocaman, S. Dam-break flow in the presence of obstacle: Experiment and CFD simulation. Eng. Appl. Comput. Fluid 20115, 541–552. [Google Scholar] [CrossRef]
    33. Ozmen-Cagatay, H.; Kocaman, S.; Guzel, H. Investigation of dam-break flood waves in a dry channel with a hump. J. Hydro-Environ. Res. 20148, 304–315. [Google Scholar] [CrossRef]
    34. Gu, S.L.; Zheng, S.P.; Ren, L.Q.; Xie, H.W.; Huang, Y.F.; Wei, J.H.; Shao, S.D. SWE-SPHysics Simulation of Dam Break Flows at South-Gate Gorges Reservoir. Water 20179, 387. [Google Scholar] [CrossRef]
    35. Evangelista, S. Experiments and Numerical Simulations of Dike Erosion due to a Wave Impact. Water 20157, 5831–5848. [Google Scholar] [CrossRef][Green Version]
    36. Bocci, M.; Chiarlo, R.; De Nat, L.; Fanelli, A.; Petersen, O.; Sorensen, J.T.; Friss-Christensen, A. Modelling of impacts from a long sea outfall outside of the Venice Lagoon (Italy). In Proceedings of the MWWD—IEMES 2006 Conference, Antalya, Turkey, 6–10 November 2006; MWWD Organization: Antalya, Turkey, 2006. [Google Scholar]
    37. Nikolaos, T.F.; Georgios, M.H. Three-dimensional numerical simulation of wind-induced barotropic circulation in the Gulf of Patras. Ocean Eng. 201037, 355–364. [Google Scholar]
    38. Goyal, R.; Rathod, P. Hydrodynamic Modelling for Salinity of Singapore Strait and Johor Strait using MIKE 3FM. In Proceedings of the 2011 2nd International Conference on Environmental Science and Development, Singapore, 26–28 February 2011. [Google Scholar]
    39. Khankandi, A.F.; Tahershamsi, A.; Soares-Frazão, S. Experimental investigation of reservoir geometry effect on dam-break flow. J. Hydraul. Res. 201250, 376–387. [Google Scholar] [CrossRef]
    40. Flow Science Inc. FLOW-3D User’s Manuals; Flow Science Inc.: Santa Fe, NM, USA, 2007. [Google Scholar]
    41. Danish Hydraulic Institute (DHI). MIKE 3 Flow Model FM. Hydrodynamic Module-User Guide; DHI: Horsholm, Denmark, 2014. [Google Scholar]
    42. Pilotti, M.; Tomirotti, M.; Valerio, G. Simplified Method for the Characterization of the Hydrograph following a Sudden Partial Dam Break. J. Hydraul. Eng. 2010136, 693–704. [Google Scholar] [CrossRef]
    43. Hooshyaripor, F.; Tahershamsi, A.; Razi, S. Dam break flood wave under different reservoir’s capacities and lengths. Sādhanā 201742, 1557–1569. [Google Scholar] [CrossRef]
    44. Kocaman, S.; Ozmen-Cagatay, H. Investigation of dam-break induced shock waves impact on a vertical Wall. J. Hydrol. 2015525, 1–12. [Google Scholar] [CrossRef]
    45. Liu, H.; Liu, H.J.; Guo, L.H.; Lu, S.X. Experimental Study on the Dam-Break Hydrographs at the Gate Location. J. Ocean Univ. China 201716, 697–702. [Google Scholar] [CrossRef]
    46. Marra, D.; Earl, T.; Ancey, C. Experimental Investigations of Dam Break Flows down an Inclined Channel. In Proceedings of the 34th World Congress of the International Association for Hydro- Environment Research and Engineering: 33rd Hydrology and Water Resources Symposium and 10th Conference on Hydraulics in Water Engineering, Brisbane, Australia, 26 June–1 July 2011. [Google Scholar]
    47. Wang, J.; Liang, D.F.; Zhang, J.X.; Xiao, Y. Comparison between shallow water and Boussinesq models for predicting cascading dam-break flows. Nat. Hazards 201683, 327–343. [Google Scholar] [CrossRef]
    48. Yang, C.; Lin, B.L.; Jiang, C.B.; Liu, Y. Predicting near-field dam-break flow and impact force using a 3D model. J. Hydraul. Res. 201048, 784–792. [Google Scholar] [CrossRef]
    Figure 5 - Modeling a simple lotus overflow symmetrically in FLOW-3D software

    Flow-3D를 이용한 나팔형 여수로의 방류계수에 대한 와류방지 블레이드 45 도의 효과

    Effect of Vortex Breaker Blades 45 Degree on Discharge Coefficient of Morning Glory Spillway Using Flow-3D

    Authors

    S. Noruzi1
    and J. Ahadiyan2*
    1– M.Sc. Student, Faculty of Water Sciences Engineering, Shahid Chamran University of Ahvaz, Iran.
    2*-Corresponding Author, Associate Professor, Faculty of Water Sciences Engineering, Shahid Chamran
    University of Ahvaz, Iran.

    Abstract

    The discharge coefficient of morning glory spillway is decreased with eddies created by vortex at the inlet part of weir. However, a series of specific blades can reduce vortices which result in the spillway efficiency is increased. Hence, in this research numerical modeling of installed breaker blade on morning glory spillway was evaluated using Flow-3D model. To achieve these purposes, morning glory spillway was modeled without and with blades 3, 4 and 6 blades at 45 degree angle. To simulate the turbulence fluctuations, the modified k-e model (RNG k-e) was used and its results were compared to the experimental data. Results showed that by installing blades, the discharge coefficient increases up to 42 percent with 25 percent decreasing in the upstream water level. Moreover, among the three different arrangements of blades, the six-blade model was found to have more satisfactory results than other models. In comparison to control model, for H/D between 0 to 0.1 and 0.1 to 0.2 the discharge coefficient has been increased 40 and 57 percent for six-blade arrangement, respectively. 

    모닝 글로리의 방류계수는 위어 입구 부분의 와류에 의해 생성된 소용돌이로 감소합니다. 그러나 일련의 특정 블레이드는 와류를 줄여 여수로 효율성을 높일 수 있습니다. 따라서 본 연구에서는 모닝 글로리 여수로에 설치된 브레이커 블레이드의 수치 모델링을 FLOW-3D 모델을 사용하여 평가했습니다. 이러한 목적을 달성하기 위해 모닝 글로리 여수로는 45도 각도로 블레이드 3, 4 및 6 블레이드 있는 모델과 없는 모델로 모델링되었습니다. 난류 변동을 시뮬레이션하기 위해 수정된 k-e 모델 (RNG k-e)을 사용하고 그 결과를 실험 데이터와 비교했습니다. 결과에 따르면 블레이드를 설치하면 상류 수위가 25 % 감소하면서 배출 계수가 42 %까지 증가합니다. 또한 3 개의 서로다른 블레이드 배열 중 6 개 블레이드 모델이 다른 모델보다 더 만족스러운 결과를 나타냈다. 기본 모델과 비교하여 H / D가 0 ~ 0.1 및 0.1 ~ 0.2 인 경우, 6개 블레이드 배열에서 방류계수가 각각 40 % 및 57 % 증가했습니다.

    Keywords

    Figure 1 - Dimensions of the vortex blade
    Figure 1 – Dimensions of the vortex blade
    Figure 3 - A (Physical model of lotus overflow without blade, b) Physical model of lotus overflow with eddy blades.
    Figure 3 – A (Physical model of lotus overflow without blade, b) Physical model of lotus overflow with eddy blades.
    Figure 5 - Modeling a simple lotus overflow symmetrically in FLOW-3D software
    Figure 5 – Modeling a simple lotus overflow symmetrically in FLOW-3D software
    Figure 7 - Comparison of Ashley flow chart with numerical model and laboratory
    Figure 7 – Comparison of Ashley flow chart with numerical model and laboratory
    Figure 8 - Comparison of flow coefficient diagram - immersion ratio of numerical model with laboratory: a (overflow without blade, b) overflow with three blades, c (overflow with four blades, d) overflow with six blades
    Figure 8 – Comparison of flow coefficient diagram – immersion ratio of numerical model with laboratory: a (overflow without blade, b) overflow with three blades, c (overflow with four blades, d) overflow with six blades

    Reference

    1 -حیدری ارجلو، س.، موسوی جهرمی، س. ح. و ادیب، ا. 1386 .بررسی تاثیر شیب بر تعداد بهینه پلکانها در سرریزهای پلکانی، مجله علوم و مهندسی
    .)123-136 :)2(33 ،كشاورزی علمی )آبیاری
    2 -حاجیپور، گ. 1363 .بررسی آزمایشگاهی تأثیر تیغههای گردابشکن بر هیدرولیک جریان سرریز نیلوفری. پایاننامه كارشناسی ارشد رشته سازههای آبی،
    دانشکده مهندسی علوم آب، دانشگاه شهید چمران اهواز.
    3 -رنجبر ملکشاه، م.، 1363 .بررسی رفتار سرریز نیلوفری با پایین دست تاج پلکانی بوسیله مدلسازی رایانهای، پایاننامه كارشناسی ارشد مهندسی عمران،
    دانشکده مهندسی عمران، دانشگاه خواجه نصیر طوسی.
    4 -رمضانی، س. كاویانپور، م ر. و ع. حسنی نژاد. 1362 .بررسی پارامترهای مؤثر بر آبگذری سرریزهای نیلوفری. هفتمین كنگره ملی مهندسی عمران،
    دانشکده مهندسی شهید نیکبخت، زاهدان.
    1 -سامانی، م. 1331 .طراحی سازههای هیدرولیکی. انتشارات شركت مهندسی مشاور دز آب اهواز
    1 -قاسمزاده، ف. 1362 .شبیه سازی مسائل هیدرولیکی در 3D-FLOW .تهران، نوآور.
    3 -كمانبدست، 1 ،.موسوی، س.ر. 1361 .مطالعه آزمایشگاهی تأثیر تعداد و زاویه گرداب شکن بر مشخصات جریان در سرریز نیلوفری مربعی، نشریه علوم
    آب و خاک )غعلوم و فنون كشاورزی و منابع طبیعی(، سال بیستم، شماره 38 ،صفحه 182-131 .
    8 -نظری پوركیانی، ع ا. 1363 .بررسی فشار و سرعت جریان در سرریز نیلوفری سد البرز با استفاده از نرمافزار 3D-FLOW .اولین كنفرانس سراسری
    توسعه محوری مهندسی عمران، معماری، برق و مکانیک ایران.
    6 -نوحانی، ا.، جمالی امام قیس، ر. 1364 .بررسی آزمایشگاهی تأثیرشکل تیغه های ضد گرداب برراندمان تخلیهی سرریزهای نیلوفری، نشریه آبیاری و
    زهکشی ایران، جلد 6 ،شماره 1 ،صفحه 346-341 .
    10-Akbari, A A., Nohani, E and A. Afrous. 2015. Numerical study of the effect of anti-vortex plates on the
    inflow pattern in shaft spillways. Indian Journal of Fundamental and Applied Life Sciences, 5(S1):
    3819-3826.
    11-Anonymous, 1965. Design of Small Dams. Water Resources Technical publication, U.S Department of
    the interior Bureau of Reclamation.
    12-Bagheri, A., Shafai Bajestan, M., Mousavi Jahromi, H., Kashkuli, H. and H. Sedghi. 2010. Hydraulic
    evaluation of the flow over polyhedral morning glory spillways. Word Applied Sciences Journal, 9(7):
    712-717.
    13- Fattor, C. A. and J. D. Bacchiega. 2003. Analysis of instabilities in the charge of regime in morning
    glory spillways. Journal of Hydraulic Research, 40(4): 114-123.
    14- Khatsuria, R. M. 2005. Hydraulics of spillways and energy dissipaters. Marcel Dekker. Department of
    Civil and Environmental Engineering Georgia, Institute of Technology Atlanta, Newyork, USA.
    15-Mousavi. S. R., Kamanbedast, A.A., and H. Fathian. 2013. Experimental investigation of the effect of
    number of anti-vortex piers on submergence threshold in morning glory spillway with square inlet.
    Technical Journal of Engineering and Applied Sciences, 3(24): 3534-3540.
    16- Novak, P. 2007. Hydraulic Structures, Fourth edition published by Taylor and Francis. University of
    New Castle upon, Tyne, UK, Landon and Network.
    17-Tavana, M H., Mousavi Jahromi, H., Shafai Bajestan, M., Masjedi, A. R. and H. Sedghi. 2011.
    Optimization of number and direction of vortex breakers in the morning glory spillway using physical
    model. Economy, Environmental and Conservation Journal, 17(2): 435-440.
    18-Vresteeg. H. K and W. Malalasekera. 1995. An introduction to computational fluid dynamics. Longman
    Scientific and Technical. New York.
    19-Yakhot. V and L. M. Smith. 1992. The renormalization group. The e-expansion and of turbulence
    models. Journal of Computing, 7(1): 35-61.

    The 3D computational domain model (50–18.6) slope change, and boundary condition for (50–30 slope change) model.

    Numerical investigation of flow characteristics over stepped spillways

    Güven, Aytaç
    Mahmood, Ahmed Hussein
    Water Supply (2021) 21 (3): 1344–1355.
    https://doi.org/10.2166/ws.2020.283Article history

    Abstract

    Spillways are constructed to evacuate flood discharge safely so that a flood wave does not overtop the dam body. There are different types of spillways, with the ogee type being the conventional one. A stepped spillway is an example of a nonconventional spillway. The turbulent flow over a stepped spillway was studied numerically by using the Flow-3D package. Different fluid flow characteristics such as longitudinal flow velocity, temperature distribution, density and chemical concentration can be well simulated by Flow-3D. In this study, the influence of slope changes on flow characteristics such as air entrainment, velocity distribution and dynamic pressures distribution over a stepped spillway was modelled by Flow-3D. The results from the numerical model were compared with an experimental study done by others in the literature. Two models of a stepped spillway with different discharge for each model were simulated. The turbulent flow in the experimental model was simulated by the Renormalized Group (RNG) turbulence scheme in the numerical model. A good agreement was achieved between the numerical results and the observed ones, which are exhibited in terms of graphics and statistical tables.

    배수로는 홍수가 댐 몸체 위로 넘치지 않도록 안전하게 홍수를 피할 수 있도록 건설되었습니다. 다른 유형의 배수로가 있으며, ogee 유형이 기존 유형입니다. 계단식 배수로는 비 전통적인 배수로의 예입니다. 계단식 배수로 위의 난류는 Flow-3D 패키지를 사용하여 수치적으로 연구되었습니다.

    세로 유속, 온도 분포, 밀도 및 화학 농도와 같은 다양한 유체 흐름 특성은 Flow-3D로 잘 시뮬레이션 할 수 있습니다. 이 연구에서는 계단식 배수로에 대한 공기 혼입, 속도 분포 및 동적 압력 분포와 같은 유동 특성에 대한 경사 변화의 영향을 Flow-3D로 모델링 했습니다.

    수치 모델의 결과는 문헌에서 다른 사람들이 수행한 실험 연구와 비교되었습니다. 각 모델에 대해 서로 다른 배출이 있는 계단식 배수로의 두 모델이 시뮬레이션되었습니다. 실험 모델의 난류 흐름은 수치 모델의 Renormalized Group (RNG) 난류 계획에 의해 시뮬레이션되었습니다. 수치 결과와 관찰 된 결과 사이에 좋은 일치가 이루어졌으며, 이는 그래픽 및 통계 테이블로 표시됩니다.

    HIGHLIGHTS

    ListenReadSpeaker webReader: Listen

    • A numerical model was developed for stepped spillways.
    • The turbulent flow was simulated by the Renormalized Group (RNG) model.
    • Both numerical and experimental results showed that flow characteristics are greatly affected by abrupt slope change on the steps.

    Keyword

    CFDnumerical modellingslope changestepped spillwayturbulent flow

    INTRODUCTION

    댐 구조는 물 보호가 생활의 핵심이기 때문에 물을 저장하거나 물을 운반하는 전 세계에서 가장 중요한 프로젝트입니다. 그리고 여수로는 댐의 가장 중요한 부분 중 하나로 분류됩니다. 홍수로 인한 파괴 나 피해로부터 댐을 보호하기 위해 여수로가 건설됩니다.

    수력 발전, 항해, 레크리에이션 및 어업의 중요성을 감안할 때 댐 건설 및 홍수 통제는 전 세계적으로 매우 중요한 문제로 간주 될 수 있습니다. 많은 유형의 배수로가 있지만 가장 일반적인 유형은 다음과 같습니다 : ogee 배수로, 자유 낙하 배수로, 사이펀 배수로, 슈트 배수로, 측면 채널 배수로, 터널 배수로, 샤프트 배수로 및 계단식 배수로.

    그리고 모든 여수로는 입구 채널, 제어 구조, 배출 캐리어 및 출구 채널의 네 가지 필수 구성 요소로 구성됩니다. 특히 롤러 압축 콘크리트 (RCC) 댐 건설 기술과 더 쉽고 빠르며 저렴한 건설 기술로 분류 된 계단식 배수로 건설과 관련하여 최근 수십 년 동안 많은 계단식 배수로가 건설되었습니다 (Chanson 2002; Felder & Chanson 2011).

    계단식 배수로 구조는 캐비테이션 위험을 감소시키는 에너지 소산 속도를 증가시킵니다 (Boes & Hager 2003b). 계단식 배수로는 다양한 조건에서 더 매력적으로 만드는 장점이 있습니다.

    계단식 배수로의 흐름 거동은 일반적으로 낮잠, 천이 및 스키밍 흐름 체제의 세 가지 다른 영역으로 분류됩니다 (Chanson 2002). 유속이 낮을 때 nappe 흐름 체제가 발생하고 자유 낙하하는 낮잠의 시퀀스로 특징 지워지는 반면, 스키밍 흐름 체제에서는 물이 외부 계단 가장자리 위의 유사 바닥에서 일관된 흐름으로 계단 위로 흐릅니다.

    또한 주요 흐름에서 3 차원 재순환 소용돌이가 발생한다는 것도 분명합니다 (예 : Chanson 2002; Gonzalez & Chanson 2008). 계단 가장자리 근처의 의사 바닥에서 흐름의 방향은 가상 바닥과 가상으로 정렬됩니다. Takahashi & Ohtsu (2012)에 따르면, 스키밍 흐름 체제에서 주어진 유속에 대해 흐름은 계단 가장자리 근처의 수평 계단면에 영향을 미치고 슈트 경사가 감소하면 충돌 영역의 면적이 증가합니다. 전이 흐름 체제는 나페 흐름과 스키밍 흐름 체제 사이에서 발생합니다. 계단식 배수로를 설계 할 때 스키밍 흐름 체계를 고려해야합니다 (예 : Chanson 1994, Matos 2000, Chanson 2002, Boes & Hager 2003a).

    CFD (Computational Fluid Dynamics), 즉 수력 공학의 수치 모델은 일반적으로 물리적 모델에 소요되는 총 비용과 시간을 줄여줍니다. 따라서 수치 모델은 실험 모델보다 빠르고 저렴한 것으로 분류되며 동시에 하나 이상의 목적으로 사용될 수도 있습니다. 사용 가능한 많은 CFD 소프트웨어 패키지가 있지만 가장 널리 사용되는 것은 FLOW-3D입니다. 이 연구에서는 Flow 3D 소프트웨어를 사용하여 유량이 서로 다른 두 모델에 대해 계단식 배수로에서 공기 농도, 속도 분포 및 동적 압력 분포를 시뮬레이션합니다.

    Roshan et al. (2010)은 서로 다른 수의 계단 및 배출을 가진 계단식 배수로의 두 가지 물리적 모델에 대한 흐름 체제 및 에너지 소산 조사를 연구했습니다. 실험 모델의 기울기는 각각 19.2 %, 12 단계와 23 단계의 수입니다. 결과는 23 단계 물리적 모델에서 관찰 된 흐름 영역이 12 단계 모델보다 더 수용 가능한 것으로 간주되었음을 보여줍니다. 그러나 12 단계 모델의 에너지 손실은 23 단계 모델보다 더 많았습니다. 그리고 실험은 스키밍 흐름 체제에서 23 단계 모델의 에너지 소산이 12 단계 모델보다 약 12 ​​% 더 적다는 것을 관찰했습니다.

    Ghaderi et al. (2020a)는 계단 크기와 유속이 다른 정련 매개 변수의 영향을 조사하기 위해 계단식 배수로에 대한 실험 연구를 수행했습니다. 그 결과, 흐름 체계가 냅페 흐름 체계에서 발생하는 최소 scouring 깊이와 같은 scouring 구멍 치수에 영향을 미친다는 것을 보여주었습니다. 또한 테일 워터 깊이와 계단 크기는 최대 scouring깊이에 대한 실제 매개 변수입니다. 테일 워터의 깊이를 6.31cm에서 8.54 및 11.82cm로 늘림으로써 수세 깊이가 각각 18.56 % 및 11.42 % 증가했습니다. 또한 이 증가하는 테일 워터 깊이는 scouring 길이를 각각 31.43 % 및 16.55 % 감소 시킵니다. 또한 유속을 높이면 Froude 수가 증가하고 흐름의 운동량이 증가하면 scouring이 촉진됩니다. 또한 결과는 중간의 scouring이 횡단면의 측벽보다 적다는 것을 나타냅니다. 계단식 배수로 하류의 최대 scouring 깊이를 예측 한 후 실험 결과와 비교하기 위한 실험식이 제안 되었습니다. 그리고 비교 결과 제안 된 공식은 각각 3.86 %와 9.31 %의 상대 오차와 최대 오차 내에서 scouring 깊이를 예측할 수 있음을 보여주었습니다.

    Ghaderi et al. (2020b)는 사다리꼴 미로 모양 (TLS) 단계의 수치 조사를 했습니다. 결과는 이러한 유형의 배수로가 확대 비율 LT / Wt (LT는 총 가장자리 길이, Wt는 배수로의 폭)를 증가시키기 때문에 더 나은 성능을 갖는 것으로 관찰되었습니다. 또한 사다리꼴 미로 모양의 계단식 배수로는 더 큰 마찰 계수와 더 낮은 잔류 수두를 가지고 있습니다. 마찰 계수는 다양한 배율에 대해 0.79에서 1.33까지 다르며 평평한 계단식 배수로의 경우 대략 0.66과 같습니다. 또한 TLS 계단식 배수로에서 잔류 수두의 비율 (Hres / dc)은 약 2.89이고 평평한 계단식 배수로의 경우 약 4.32와 같습니다.

    Shahheydari et al. (2015)는 Flow-3D 소프트웨어, RNG k-ε 모델 및 VOF (Volume of Fluid) 방법을 사용하여 배출 계수 및 에너지 소산과 같은 자유 표면 흐름의 프로파일을 연구하여 스키밍 흐름 체제에서 계단식 배수로에 대한 흐름을 조사했습니다. 실험 결과와 비교했습니다. 결과는 에너지 소산 율과 방전 계수율의 관계가 역으로 실험 모델의 결과와 잘 일치 함을 보여 주었다.

    Mohammad Rezapour Tabari & Tavakoli (2016)는 계단 높이 (h), 계단 길이 (L), 계단 수 (Ns) 및 단위 폭의 방전 (q)과 같은 다양한 매개 변수가 계단식 에너지 ​​소산에 미치는 영향을 조사했습니다. 방수로. 그들은 해석에 FLOW-3D 소프트웨어를 사용하여 계단식 배수로에서 에너지 손실과 임계 흐름 깊이 사이의 관계를 평가했습니다. 또한 유동 난류에 사용되는 방정식과 표준 k-ɛ 모델을 풀기 위해 유한 체적 방법을 적용했습니다. 결과에 따르면 스텝 수가 증가하고 유량 배출량이 증가하면 에너지 손실이 감소합니다. 얻은 결과를 다른 연구와 비교하고 경험적, 수학적 조사를 수행하여 결국 합격 가능한 결과를 얻었습니다.

    METHODOLOGY

    ListenReadSpeaker webReader: ListenFor all numerical models the basic principle is very similar: a set of partial differential equations (PDE) present the physical problems. The flow of fluids (gas and liquid) are governed by the conservation laws of mass, momentum and energy. For Computational Fluid Dynamics (CFD), the PDE system is substituted by a set of algebraic equations which can be worked out by using numerical methods (Versteeg & Malalasekera 2007). Flow-3D uses the finite volume approach to solve the Reynolds Averaged Navier-Stokes (RANS) equation, by applying the technique of Fractional Area/Volume Obstacle Representation (FAVOR) to define an obstacle (Flow Science Inc. 2012). Equations (1) and (2) are RANS and continuity equations with FAVOR variables that are applied for incompressible flows.

    formula

    (1)

    formula

    (2)where  is the velocity in xi direction, t is the time,  is the fractional area open to flow in the subscript directions,  is the volume fraction of fluid in each cell, p is the hydrostatic pressure,  is the density, is the gravitational force in subscript directions and  is the Reynolds stresses.

    Turbulence modelling is one of three key elements in CFD (Gunal 1996). There are many types of turbulence models, but the most common are Zero-equation models, One-equation models, Two-equation models, Reynolds Stress/Flux models and Algebraic Stress/Flux models. In FLOW-3D software, five turbulence models are available. The formulation used in the FLOW-3D software differs slightly from other formulations that includes the influence of the fractional areas/volumes of the FAVORTM method and generalizes the turbulence production (or decay) associated with buoyancy forces. The latter generalization, for example, includes buoyancy effects associated with non-inertial accelerations.

    The available turbulence models in Flow-3D software are the Prandtl Mixing Length Model, the One-Equation Turbulent Energy Model, the Two-Equation Standard  Model, the Two-Equation Renormalization-Group (RNG) Model and large Eddy Simulation Model (Flow Science Inc. 2012).In this research the RNG model was selected because this model is more commonly used than other models in dealing with particles; moreover, it is more accurate to work with air entrainment and other particles. In general, the RNG model is classified as a more widely-used application than the standard k-ɛ model. And in particular, the RNG model is more accurate in flows that have strong shear regions than the standard k-ɛ model and it is defined to describe low intensity turbulent flows. For the turbulent dissipation  it solves an additional transport equation:

    formula

    (3)where CDIS1, CDIS2, and CDIS3 are dimensionless parameters and the user can modify them. The diffusion of dissipation, Diff ɛ, is

    formula

    (4)where uv and w are the x, y and z coordinates of the fluid velocity; ⁠, ⁠,  and ⁠, are FLOW-3D’s FAVORTM defined terms;  and  are turbulence due to shearing and buoyancy effects, respectively. R and  are related to the cylindrical coordinate system. The default values of RMTKE, CDIS1 and CNU differ, being 1.39, 1.42 and 0.085 respectively. And CDIS2 is calculated from turbulent production (⁠⁠) and turbulent kinetic energy (⁠⁠).The kinematic turbulent viscosity is the same in all turbulence transport models and is calculated from

    formula

    (5)where ⁠: is the turbulent kinematic viscosity.  is defined as the numerical challenge between the RNG and the two-equation k-ɛ models, found in the equation below. To avoid an unphysically large result for  in Equation (3), since this equation could produce a value for  very close to zero and also because the physical value of  may approach to zero in such cases, the value of  is calculated from the following equation:

    formula

    (6)where ⁠: the turbulent length scale.

    VOF and FAVOR are classifications of volume-fraction methods. In these two methods, firstly the area should be subdivided into a control volume grid or a small element. Each flow parameter like velocity, temperature and pressure values within the element are computed for each element containing liquids. Generally, these values represent the volumetric average of values in the elements.Numerous methods have been used recently to solve free infinite boundaries in the various numerical simulations. VOF is an easy and powerful method created based on the concept of a fractional intensity of fluid. A significant number of studies have confirmed that this method is more flexible and efficient than others dealing with the configurations of a complex free boundary. By using VOF technology the Flow-3D free surface was modelled and first declared in Hirt & Nichols (1981). In the VOF method there are three ingredients: a planner to define the surface, an algorithm for tracking the surface as a net mediator moving over a computational grid, and application of the boundary conditions to the surface. Configurations of the fluids are defined in terms of VOF function, F (x, y, z, t) (Hirt & Nichols 1981). And this VOF function shows the volume of flow per unit volume

    formula

    (7)

    formula

    (8)

    formula

    (9)where  is the density of the fluid, is a turbulent diffusion term,  is a mass source,  is the fractional volume open to flow. The components of velocity (u, v, w) are in the direction of coordinates (x, y, z) or (r, ⁠).  in the x-direction is the fractional area open to flow,  and  are identical area fractions for flow in the y and z directions. The R coefficient is based on the selection of the coordinate system.

    The FAVOR method is a different method and uses another volume fraction technique, which is only used to define the geometry, such as the volume of liquid in each cell used to determine the position of fluid surfaces. Another fractional volume can be used to define the solid surface. Then, this information is used to determine the boundary conditions of the wall that the flow should be adapted for.

    Case study

    ListenReadSpeaker webReader: Listen

    In this study, the experimental results of Ostad Mirza (2016) was simulated. In a channel composed of two 4 m long modules, with a transparent sidewall of height 0.6 m and 0.5 m width. The upstream chute slope (i.e. pseudo-bottom angle) Ɵ1 = 50°, the downstream chute slope Ɵ2 = 30° or 18.6°, the step heights h = 0.06 m, the total number of steps along the 50° chute 41 steps, the total number of steps along the 30° chute 34 steps and the total number of steps along the 18.6° chute 20 steps.

    The flume inflow tool contained a jetbox with a maximum opening set to 0.12 meters, designed for passing the maximum unit discharge of 0.48 m2/s. The measurements of the flow properties (i.e. air concentration and velocity) were computed perpendicular to the pseudo-bottom as shown in Figure 1 at the centre of twenty stream-wise cross-sections, along the stepped chute, (i.e. in five steps up on the slope change and fifteen steps down on the slope change, namely from step number −09 to +23 on 50°–30° slope change, or from −09 to +15 on 50°–18.6° slope change, respectively).

    Sketch of the air concentration C and velocity V measured perpendicular to the pseudo-bottom used by Mirza (Ostad Mirza 2016).
    Sketch of the air concentration C and velocity V measured perpendicular to the pseudo-bottom used by Mirza (Ostad Mirza 2016).

    Sketch of the air concentration C and velocity V measured perpendicular to the pseudo-bottom used by Mirza (Ostad Mirza 2016).

    Pressure sensors were arranged with the x/l values for different slope change as shown in Table 1, where x is the distance from the step edge, along the horizontal step face, and l is the length of the horizontal step face. The location of pressure sensors is shown in Table 1.Table 1

    Location of pressure sensors on horizontal step faces

    Θ(°)L(m)x/l (–)
    50.0 0.050 0.35 0.64 – – – 
    30.0 0.104 0.17 0.50 0.84 – – 
    18.6 0.178 0.10 0.30 0.50 0.7 0.88 
    Location of pressure sensors on horizontal step faces
    Inlet boundary condition for Q = 0.235 m3/s and fluid elevation 4.21834 m.
    Inlet boundary condition for Q = 0.235 m3/s and fluid elevation 4.21834 m.

    Inlet boundary condition for Q = 0.235 m3/s and fluid elevation 4.21834 m.

    Numerical model set-up

    ListenReadSpeaker webReader: Listen

    A 3D numerical model of hydraulic phenomena was simulated based on an experimental study by Ostad Mirza (2016). The water surcharge and flow pressure over the stepped spillway was computed for two models of a stepped spillway with different discharge for each model. In this study, the package was used to simulate the flow parameters such as air entrainment, velocity distribution and dynamic pressures. The solver uses the finite volume technique to discretize the computational domain. In every test run, one incompressible fluid flow with a free surface flow selected at 20̊ was used for this simulation model. Table 2 shows the variables used in test runs.Table 2

    Variables used in test runs

    Test no.Θ1 (°)Θ2 (°)h(m)d0q (m3s1)dc/h (–)
    50 18.6 0.06 0.045 0.1 2.6 
    50 18.6 0.06 0.082 0.235 4.6 
    50 30.0 0.06 0.045 0.1 2.6 
    50 30.0 0.06 0.082 0.235 4.6 
    Table 2 Variables used in test runs

    For stepped spillway simulation, several parameters should be specified to get accurate simulations, which is the scope of this research. Viscosity and turbulent, gravity and non-inertial reference frame, air entrainment, density evaluation and drift-flux should be activated for these simulations. There are five different choices in the ‘viscosity and turbulent’ option, in the viscosity flow and Renormalized Group (RNG) model. Then a dynamical model is selected as the second option, the ‘gravity and non-inertial reference frame’. Only the z-component was inputted as a negative 9.81 m/s2 and this value represents gravitational acceleration but in the same option the x and y components will be zero. Air entrainment is selected. Finally, in the drift-flux model, the density of phase one is input as (water) 1,000 kg/m3 and the density of phase two (air) as 1.225 kg/m3. Minimum volume fraction of phase one is input equal to 0.1 and maximum volume fraction of phase two to 1 to allow air concentration to reach 90%, then the option allowing gas to escape at free surface is selected, to obtain closer simulation.

    The flow domain is divided into small regions relatively by the mesh in Flow-3D numerical model. Cells are the smallest part of the mesh, in which flow characteristics such as air concentration, velocity and dynamic pressure are calculated. The accuracy of the results and simulation time depends directly on the mesh block size so the cell size is very important. Orthogonal mesh was used in cartesian coordinate systems. A smaller cell size provides more accuracy for results, so we reduced the number of cells whilst including enough accuracy. In this study, the size of cells in x, y and z directions was selected as 0.015 m after several trials.

    Figure 3 shows the 3D computational domain model 50–18.6 slope change, that is 6.0 m length, 0.50 m width and 4.23 m height. The 3D model of the computational domain model 50–30 slope changes this to 6.0 m length, 0.50 m width and 5.068 m height and the size of meshes in x, y, and z directions are 0.015 m. For the 50–18.6 slope change model: both total number of active and passive cells = 4,009,952, total number of active cells = 3,352,307, include real cells (used for solving the flow equations) = 3,316,269, open real cells = 3,316,269, fully blocked real cells equal to zero, external boundary cells were 36,038, inter-block boundary cells = 0 (Flow-3D report). For 50–30 slope change model: both total number of active and passive cells = 4,760,002, total number of active cells equal to 4,272,109, including real cells (used for solving the flow equations) were 3,990,878, open real cells = 3,990,878 fully blocked real cells = zero, external boundary cells were 281,231, inter-block boundary cells = 0 (Flow-3D report).

    The 3D computational domain model (50–18.6) slope change, and boundary condition for (50–30 slope change) model.
    Figure3 The 3D computational domain model (50–18.6) slope change, and boundary condition for (50–30 slope change) model.

    Figure 3VIEW LARGEDOWNLOAD SLIDE

    The 3D computational domain model (50–18.6) slope change, and boundary condition for (50–30 slope change) model.

    When solving the Navier-Stokes equation and continuous equations, boundary conditions should be applied. The most important work of boundary conditions is to create flow conditions similar to physical status. The Flow-3D software has many types of boundary condition; each type can be used for the specific condition of the models. The boundary conditions in Flow-3D are symmetry, continuative, specific pressure, grid overlay, wave, wall, periodic, specific velocity, outflow, and volume flow rate.

    There are two options to input finite flow rate in the Flow-3D software either for inlet discharge of the system or for the outlet discharge of the domain: specified velocity and volume flow rate. In this research, the X-minimum boundary condition, volume flow rate, has been chosen. For X-maximum boundary condition, outflow was selected because there is nothing to be calculated at the end of the flume. The volume flow rate and the elevation of surface water was set for Q = 0.1 and 0.235 m3/s respectively (Figure 2).

    The bottom (Z-min) is prepared as a wall boundary condition and the top (Z-max) is computed as a pressure boundary condition, and for both (Y-min) and (Y-max) as symmetry.

    RESULTS AND DISCUSSION

    ListenReadSpeaker webReader: Listen

    The air concentration distribution profiles in two models of stepped spillway were obtained at an acquisition time equal to 25 seconds in skimming flow for both upstream and downstream of a slope change 50°–18.6° and 50°–30° for different discharge as in Table 2, and as shown in Figure 4 for 50°–18.6° slope change and Figure 5 for 50°–30° slope change configuration for dc/h = 4.6. The simulation results of the air concentration are very close to the experimental results in all curves and fairly close to that predicted by the advection-diffusion model for the air bubbles suggested by Chanson (1997) on a constant sloping chute.

    Figure 4 Experimental and simulated air concentration distribution for steps number −5, +1, +5, +8, +11 and +15 along the 50°–18.6° slope change for dc/h = 4.6. VIEW LARGEDOWNLOAD SLIDE Experimental and simulated air concentration distribution for steps number −5, +1, +5, +8, +11 and +15 along the 50°–18.6° slope change for dc/h = 4.6.
    Figure 4 Experimental and simulated air concentration distribution for steps number −5, +1, +5, +8, +11 and +15 along the 50°–18.6° slope change for dc/h = 4.6. VIEW LARGEDOWNLOAD SLIDE Experimental and simulated air concentration distribution for steps number −5, +1, +5, +8, +11 and +15 along the 50°–18.6° slope change for dc/h = 4.6.

    Figure 4VIEW LARGEDOWNLOAD SLIDE

    Experimental and simulated air concentration distribution for steps number −5, +1, +5, +8, +11 and +15 along the 50°–18.6° slope change for dc/h = 4.6.

    Figure5 Experimental and simulated air concentration distribution for steps number −5, +1, +5, +11, +19 and +22 along the 50°–30° slope change, for dc/h = 4.6.
    Figure5 Experimental and simulated air concentration distribution for steps number −5, +1, +5, +11, +19 and +22 along the 50°–30° slope change, for dc/h = 4.6.

    Figure 5VIEW LARGEDOWNLOAD SLIDE

    Experimental and simulated air concentration distribution for steps number −5, +1, +5, +11, +19 and +22 along the 50°–30° slope change, for dc/h = 4.6.

    Figure 6VIEW LARGEDOWNLOAD SLIDE

    Figure 6 Experimental and simulated dimensionless velocity distribution for steps number −5, −1, +1, +5, +8, +11 and +15 along the 50°–18.6° slope change for dc/h = 2.6.
    Figure 6 Experimental and simulated dimensionless velocity distribution for steps number −5, −1, +1, +5, +8, +11 and +15 along the 50°–18.6° slope change for dc/h = 2.6.

    Experimental and simulated dimensionless velocity distribution for steps number −5, −1, +1, +5, +8, +11 and +15 along the 50°–18.6° slope change for dc/h = 2.6.

    Figure 7 Experimental and simulated dimensionless velocity distribution for steps number −5, −1, +1, +5. +11, +15 and +22 along the 50°–30° slope change for dc/h = 2.6.
    Figure 7 Experimental and simulated dimensionless velocity distribution for steps number −5, −1, +1, +5. +11, +15 and +22 along the 50°–30° slope change for dc/h = 2.6.

    Figure 7VIEW LARGEDOWNLOAD SLIDE

    Experimental and simulated dimensionless velocity distribution for steps number −5, −1, +1, +5. +11, +15 and +22 along the 50°–30° slope change for dc/h = 2.6.

    But as is shown in all above mentioned figures it is clear that at the pseudo-bottom the CFD results of air concentration are less than experimental ones until the depth of water reaches a quarter of the total depth of water. Also the direction of the curves are parallel to each other when going up towards the surface water and are incorporated approximately near the surface water. For all curves, the cross-section is separate between upstream and downstream steps. Therefore the (-) sign for steps represents a step upstream of the slope change cross-section and the (+) sign represents a step downstream of the slope change cross-section.

    The dimensionless velocity distribution (V/V90) profile was acquired at an acquisition time equal to 25 seconds in skimming flow of the upstream and downstream slope change for both 50°–18.6° and 50°–30° slope change. The simulation results are compared with the experimental ones showing that for all curves there is close similarity for each point between the observed and experimental results. The curves increase parallel to each other and they merge near at the surface water as shown in Figure 6 for slope change 50°–18.6° configuration and Figure 7 for slope change 50°–30° configuration. However, at step numbers +1 and +5 in Figure 7 there are few differences between the simulated and observed results, namely the simulation curves ascend regularly meaning the velocity increases regularly from the pseudo-bottom up to the surface water.

    Figure 8 (50°–18.6° slope change) and Figure 9 (50°–30° slope change) compare the simulation results and the experimental results for the presented dimensionless dynamic pressure distribution for different points on the stepped spillway. The results show a good agreement with the experimental and numerical simulations in all curves. For some points, few discrepancies can be noted in pressure magnitudes between the simulated and the observed ones, but they are in the acceptable range. Although the experimental data do not completely agree with the simulated results, there is an overall agreement.

    Figure 8 Comparison between simulated and experimental results for the dimensionless pressure for steps number  −1, −2, −3 and +1, +2 +3 and +20 on the horizontal step faces of 50°–18.6° slope change configuration, for dc/h = 4.6, x is the distance from the step edge.
    Figure 8 Comparison between simulated and experimental results for the dimensionless pressure for steps number −1, −2, −3 and +1, +2 +3 and +20 on the horizontal step faces of 50°–18.6° slope change configuration, for dc/h = 4.6, x is the distance from the step edge.

    Figure 8VIEW LARGEDOWNLOAD SLIDE

    Comparison between simulated and experimental results for the dimensionless pressure for steps number −1, −2, −3 and +1, +2 +3 and +20 on the horizontal step faces of 50°–18.6° slope change configuration, for dc/h = 4.6, x is the distance from the step edge.

    Figure 9 Comparison between simulated and experimental results for the dimensionless pressure for steps number  −1, −2, −3 and +1, +2 and +30, +31 on the horizontal step face of 50°–30° slope change configuration, for dc/h = 4.6, x is the distance from the step edge.
    Figure 9 Comparison between simulated and experimental results for the dimensionless pressure for steps number −1, −2, −3 and +1, +2 and +30, +31 on the horizontal step face of 50°–30° slope change configuration, for dc/h = 4.6, x is the distance from the step edge.

    Figure 9VIEW LARGEDOWNLOAD SLIDE

    Comparison between simulated and experimental results for the dimensionless pressure for steps number −1, −2, −3 and +1, +2 and +30, +31 on the horizontal step face of 50°–30° slope change configuration, for dc/h = 4.6, x is the distance from the step edge.

    The pressure profiles were acquired at an acquisition time equal to 70 seconds in skimming flow on 50°–18.6°, where p is the measured dynamic pressure, h is step height and ϒ is water specific weight. A negative sign for steps represents a step upstream of the slope change cross-section and a positive sign represents a step downstream of the slope change cross-section.

    Figure 10 shows the experimental streamwise development of dimensionless pressure on the 50°–18.6° slope change for dc/h = 4.6, x/l = 0.35 on 50° sloping chute and x/l = 0.3 on 18.6° sloping chute compared with the numerical simulation. It is obvious from Figure 10 that the streamwise development of dimensionless pressure before slope change (steps number −1, −2 and −3) both of the experimental and simulated results are close to each other. However, it is clear that there is a little difference between the results of the streamwise development of dimensionless pressure at step numbers +1, +2 and +3. Moreover, from step number +3 to the end, the curves get close to each other.

    Figure 10 Comparison between experimental and simulated results for the streamwise development of the dimensionless pressure on the 50°–18.6° slope change, for dc/h = 4.6, and x/l = 0.35 on 50° sloping chute and x/l = 0.3 on 18.6° sloping chute.
    Figure 10 Comparison between experimental and simulated results for the streamwise development of the dimensionless pressure on the 50°–18.6° slope change, for dc/h = 4.6, and x/l = 0.35 on 50° sloping chute and x/l = 0.3 on 18.6° sloping chute.

    Figure 10VIEW LARGEDOWNLOAD SLIDE

    Comparison between experimental and simulated results for the streamwise development of the dimensionless pressure on the 50°–18.6° slope change, for dc/h = 4.6, and x/l = 0.35 on 50° sloping chute and x/l = 0.3 on 18.6° sloping chute.

    Figure 11 compares the experimental and the numerical results for the streamwise development of the dimensionless pressure on the 50°–30° slope change, for dc/h = 4.6, and x/l = 0.35 on 50° sloping chute and x/l = 0.17 on 30° sloping chute. It is apparent that the outcomes of the experimental work are close to the numerical results, however, the results of the simulation are above the experimental ones before the slope change, but the results of the simulation descend below the experimental ones after the slope change till the end.

    Figure 11 Comparison between experimental and simulated results for the streamwise development of the dimensionless pressure on the 50°–30° slope change, for dc/h = 4.6, and x/l = 0.35 on 50° sloping chute and x/l = 0.17 on 30° sloping chute.
    Figure 11 Comparison between experimental and simulated results for the streamwise development of the dimensionless pressure on the 50°–30° slope change, for dc/h = 4.6, and x/l = 0.35 on 50° sloping chute and x/l = 0.17 on 30° sloping chute.

    Figure 11VIEW LARGEDOWNLOAD SLIDE

    Comparison between experimental and simulated results for the streamwise development of the dimensionless pressure on the 50°–30° slope change, for dc/h = 4.6, and x/l = 0.35 on 50° sloping chute and x/l = 0.17 on 30° sloping chute.

    CONCLUSION

    ListenReadSpeaker webReader: Listen

    In this research, numerical modelling was attempted to investigate the effect of abrupt slope change on the flow properties (air entrainment, velocity distribution and dynamic pressure) over a stepped spillway with two different models and various flow rates in a skimming flow regime by using the CFD technique. The numerical model was verified and compared with the experimental results of Ostad Mirza (2016). The same domain of the numerical model was inputted as in experimental models to reduce errors as much as possible.

    Flow-3D is a well modelled tool that deals with particles. In this research, the model deals well with air entrainment particles by observing their results with experimental results. And the reason for the small difference between the numerical and the experimental results is that the program deals with particles more accurately than the laboratory. In general, both numerical and experimental results showed that near to the slope change the flow bulking, air entrainment, velocity distribution and dynamic pressure are greatly affected by abrupt slope change on the steps. Although the extent of the slope change was relatively small, the influence of the slope change was major on flow characteristics.

    The Renormalized Group (RNG) model was selected as a turbulence solver. For 3D modelling, orthogonal mesh was used as a computational domain and the mesh grid size used for X, Y, and Z direction was equal to 0.015 m. In CFD modelling, air concentration and velocity distribution were recorded for a period of 25 seconds, but dynamic pressure was recorded for a period of 70 seconds. The results showed that there is a good agreement between the numerical and the physical models. So, it can be concluded that the proposed CFD model is very suitable for use in simulating and analysing the design of hydraulic structures.

    이 연구에서 수치 모델링은 두 가지 다른 모델과 다양한 유속을 사용하여 스키밍 흐름 영역에서 계단식 배수로에 대한 유동 특성 (공기 혼입, 속도 분포 및 동적 압력)에 대한 급격한 경사 변화의 영향을 조사하기 위해 시도되었습니다. CFD 기술. 수치 모델을 검증하여 Ostad Mirza (2016)의 실험 결과와 비교 하였다. 오차를 최대한 줄이기 위해 실험 모형과 동일한 수치 모형을 입력 하였다.

    Flow-3D는 파티클을 다루는 잘 모델링 된 도구입니다. 이 연구에서 모델은 실험 결과를 통해 결과를 관찰하여 공기 혼입 입자를 잘 처리합니다. 그리고 수치와 실험 결과의 차이가 작은 이유는 프로그램이 실험실보다 입자를 더 정확하게 다루기 때문입니다. 일반적으로 수치 및 실험 결과는 경사에 가까워지면 유동 벌킹, 공기 혼입, 속도 분포 및 동적 압력이 계단의 급격한 경사 변화에 크게 영향을받는 것으로 나타났습니다. 사면 변화의 정도는 상대적으로 작았지만 사면 변화의 영향은 유동 특성에 큰 영향을 미쳤다.

    Renormalized Group (RNG) 모델이 난류 솔버로 선택되었습니다. 3D 모델링의 경우 계산 영역으로 직교 메쉬가 사용되었으며 X, Y, Z 방향에 사용 된 메쉬 그리드 크기는 0.015m입니다. CFD 모델링에서 공기 농도와 속도 분포는 25 초 동안 기록되었지만 동적 압력은 70 초 동안 기록되었습니다. 결과는 수치 모델과 물리적 모델간에 좋은 일치가 있음을 보여줍니다. 따라서 제안 된 CFD 모델은 수력 구조물의 설계 시뮬레이션 및 해석에 매우 적합하다는 결론을 내릴 수 있습니다.

    DATA AVAILABILITY STATEMENT

    ListenReadSpeaker webReader: Listen

    All relevant data are included in the paper or its Supplementary Information.

    REFERENCES

    Boes R. M. Hager W. H. 2003a Hydraulic design of stepped spillways. Journal of Hydraulic Engineering 129 (9), 671–679.
    Google Scholar
    Boes R. M. Hager W. H. 2003b Two-Phase flow characteristics of stepped spillways. Journal of Hydraulic Engineering 129 (9), 661–670.
    Google Scholar
    Chanson H. 1994 Hydraulics of skimming flows over stepped channels and spillways. Journal of Hydraulic Research 32 (3), 445–460.
    Google Scholar
    Chanson H. 1997 Air Bubble Entrainment in Free Surface Turbulent Shear Flows. Academic Press, London.
    Google Scholar
    Chanson H. 2002 The Hydraulics of Stepped Chutes and Spillways. Balkema, Lisse, The Netherlands.
    Google Scholar
    Felder S. Chanson H. 2011 Energy dissipation down a stepped spillway with nonuniform step heights. Journal of Hydraulic Engineering 137 (11), 1543–1548.
    Google Scholar
    Flow Science, Inc. 2012 FLOW-3D v10-1 User Manual. Flow Science, Inc., Santa Fe, CA.
    Ghaderi A. Daneshfaraz R. Torabi M. Abraham J. Azamathulla H. M. 2020a Experimental investigation on effective scouring parameters downstream from stepped spillways. Water Supply 20 (5), 1988–1998.
    Google Scholar
    Ghaderi A. Abbasi S. Abraham J. Azamathulla H. M. 2020b Efficiency of trapezoidal labyrinth shaped stepped spillways. Flow Measurement and Instrumentation 72, 101711.
    Google Scholar
    Gonzalez C. A. Chanson H. 2008 Turbulence and cavity recirculation in air-water skimming flows on a stepped spillway. Journal of Hydraulic Research 46 (1), 65–72.
    Google Scholar
    Gunal M. 1996 Numerical and Experimental Investigation of Hydraulic Jumps. PhD Thesis, University of Manchester, Institute of Science and Technology, Manchester, UK.
    Hirt C. W. Nichols B. D. 1981 Volume of fluid (VOF) method for the dynamics of free boundaries. Journal of Computational Physics 39 (1), 201–225.
    Google Scholar
    Matos J. 2000 Hydraulic design of stepped spillways over RCC dams. In: Intl Workshop on Hydraulics of Stepped Spillways (H.-E. Minor & W. Hager, eds). Balkema Publ, Zurich, pp. 187–194.
    Google Scholar
    Mohammad Rezapour Tabari M. Tavakoli S. 2016 Effects of stepped spillway geometry on flow pattern and energy dissipation. Arabian Journal for Science & Engineering (Springer Science & Business Media BV) 41 (4), 1215–1224.
    Google Scholar
    Ostad Mirza M. J. 2016 Experimental Study on the Influence of Abrupt Slope Changes on Flow Characteristics Over Stepped Spillways. Communications du Laboratoire de Constructions Hydrauliques, No. 64 (A. J. Schleiss, ed.). Swiss Federal Institute of Technology Lausanne (EPFL), Lausanne, Switzerland.
    Roshan R. Azamathulla H. M. Marosi M. Sarkardeh H. Pahlavan H. Ab Ghani A. 2010 Hydraulics of stepped spillways with different numbers of steps. Dams and Reservoirs 20 (3), 131–136.
    Google Scholar
    Shahheydari H. Nodoshan E. J. Barati R. Moghadam M. A. 2015 Discharge coefficient and energy dissipation over stepped spillway under skimming flow regime. KSCE Journal of Civil Engineering 19 (4), 1174–1182.
    Google Scholar
    Takahashi M. Ohtsu I. 2012 Aerated flow characteristics of skimming flow over stepped chutes. Journal of Hydraulic Research 50 (4), 427–434.
    Google Scholar
    Versteeg H. K. Malalasekera W. 2007 An Introduction to Computational Fluid Dynamics: The Finite Volume Method. Pearson Education, Harlow.
    Google Scholar
    © 2021 The Authors
    This is an Open Access article distributed under the terms of the Creative Commons Attribution Licence (CC BY 4.0), which permits copying, adaptation and redistribution, provided the original work is properly cited (http://creativecommons.org/licenses/by/4.0/).

    Figure 1. Sketch map of the port Laozi on Lake Hongze

    FLOW-3D software for substantiation the layout of the port water area

    항구 수역의 레이아웃을 입증하기 위한 FLOW-3D 소프트웨어

    B Pan1

    • and N Belyaev2
      1 Peter the Great St. Petersburg Polytechnic University, St. Petersburg, Russia
      pan3.b@edu.spbstu.ru

    Abstract

    방파제 설계시 항만 내 수역 및 접안 시설에서 중대한 파도 발생을 배제 할 필요가 있다. 이 기사는 항구 지역의 물 이동 계산 결과를 제시합니다. 항구 입구의 위치와 주어진 물체의 크기에 대한 가장 안전한 옵션이 확인되었습니다. FLOW – 3D 프로그램을 사용하여 항구 수역의 예비 배치 단계에서 계획되고 건설적인 솔루션을 선택할 가능성이 입증되었습니다.

    Introduction

    항구를 설계 할 때 계류선의 가장 합리적인 구성과 항구 수역의 레이아웃을 선택하여 항구의 영토를 배치하는 것이 필요합니다. 러시아 연방의 항구 수역 배치는 항구 수역 요소에 대한 사양을 포함하는 해로, 페어웨이 및 기동 구역에 대한 설계 표준의 요구 사항에 따라 수행됩니다 [1, 2].

    항구 물은 파도, 퇴적물 축적 및 얼음으로부터 보호되어야 합니다. 항구 계획을 작성할 때, 선박의 기동 및 연안 계류 중뿐만 아니라 선적 및 하역 및 기타 항구 운영 중에 선박 계류에 대한 정상적인 조건이 생성되도록 파도로부터 수역을 보호하는 정도를 제공해야 합니다.

    설계 결정은 새로운 포트를 설계하거나 기존 포트를 개발할 때 물리적 또는 수학적 모델링을 기반으로 합니다 [2]. 항구 수역에서 계산 된 물 흐름의 매개 변수는 수문 기상 조사, 장기 현장 관찰 및 실험실 연구의 결과를 기반으로 하도록 권장됩니다.

    공학 수문 기상 측량 데이터가 불충분하면 계산 방법을 기반으로 설계 폭풍의 파도 매개 변수를 결정할 수 있습니다. 사용된 계산 방법이 국제 실무에서 동일한 목적으로 채택된 방법 (모델)에 부합하는지 표시하는 것이 좋습니다 [3].

    Figure 1. Sketch map of the port Laozi on Lake Hongze
    Figure 1. Sketch map of the port Laozi on Lake Hongze
    Figure 2. The location of the port entrance on Lake Hongze: a – variant 1; b – variant 2; c – variants 3-5
    Figure 2. The location of the port entrance on Lake Hongze: a – variant 1; b – variant 2; c – variants 3-5
    Figure 3. Port water area plan
    Figure 3. Port water area plan
    Figure 4. Modeling of variant 1 with the movement of waves in the port water area
    Figure 4. Modeling of variant 1 with the movement of waves in the port water area
    Figure 5. Modeling of variant 2: a is prevailing movement of water towards the enclosed water area; b is prevailing reverse movement of water
    Figure 5. Modeling of variant 2: a is prevailing movement of water towards the enclosed water area; b is prevailing reverse movement of water
    Figure 6. Modeling of variant 3
    Figure 6. Modeling of variant 3
    Figure 7. Modeling of variant 4
    Figure 7. Modeling of variant 4
    Figure 8. Modeling of variant 5
    Figure 8. Modeling of variant 5
    Figure 9. Plan of the port water area with design points
    Figure 9. Plan of the port water area with design points
    Figure 10. Change in water depth at point A: a – variant 1; b – variant 2
    Figure 10. Change in water depth at point A: a – variant 1; b – variant 2
    Figure 11. Change in water depth at point A: a – variant 3; b – variant 4; c – variant 5
    Figure 11. Change in water depth at point A: a – variant 3; b – variant 4; c – variant 5
    Figure 12. Change in water depth at points A (a) and C (b) for variant 3
    Figure 12. Change in water depth at points A (a) and C (b) for variant 3
    Figure 13. Change in water depth at points A (a) and B (b) for variant 3
    Figure 13. Change in water depth at points A (a) and B (b) for variant 3
    Figure 14. Scheme of vessel traffic: a – variant 3; b – variant 4
    Figure 14. Scheme of vessel traffic: a – variant 3; b – variant 4

    References
    [1] SP 350.1326000.2018. 2018 Norms for technological design of sea ports (Moscow:
    Standartinform) p 226
    [2] SP 444.1326000.2019. 2019 Standards for the design of sea channels, fairways and
    maneuvering areas (Moscow: Standartinform) p 62
    [3] SP 38.13330.2012. 2014 Loads and impacts on Hydraulic structures (from wave, ice and ships)
    (Moscow: Ministry of Regional Development of the Russian Federation) p 112
    [4] Rijnsdorp D P Smit P B and Zijlema M 2012 Non-hydrostatic modelling of infragravity waves
    using SWASH. Proceedings of 33rd Conference on Coastal Engineering. pp 1287–1299
    [5] Kantardgi I G Zheleznyak M J 2016 Laboratory and numerical study of waves in the port area.
    Magazine of Civil Engineering No 6 pp 49–59 DOI: 10.5862/MCE.66.5
    [6] Zheleznyak M J Kantardgi I G Sorokin M S and Polyakov A I 2015 Resonance properties of
    seaport water areas Magazine of Civil Engineering № 5(57) pp 3-19 DOI:10.5862/MCE.57.1
    [7] Kantarzhi I Zuev N Shunko N 2014 Numerical and physical modelling of the waves inside the
    new marina in Gelendjik (Black Sea) Application of physical modelling to port and coastal
    protection. Proceedings of 5th international conference Coastlab (Varna) Vol 2 pp 253–262
    [8] Makarov K N and Chebotarev A G 2015 Breakwater placement at the root of a seawall
    Magazine of Civil Engineering № 3(55) pp 67-78 DOI: 10.5862/MCE.55.8
    [9] Belyaev N D Lebedev V V and Alexeeva A V 2017 Investigation of the soil structure changes
    under the tsunami waves impact on the marine hydrotechnical structures V 10 № 4 pp 44-52
    DOI: 10.7868/S2073667317040049
    [10] Lebedev V V Nudner I S and Belyaev N D 2018 The formation of the seabed surface relief near
    the gravitational object Magazine of Civil Engineering No 79(3) pp 120–131 DOI:
    10.18720/MCE.79.13
    [11] Kofoed-Hansen H Sloth P Sørensen O R Fuchs J 2000 Combined numerical and physical
    modelling of seiching in exposed new marina Proceedings of 27th international conference of
    coastal engineering pp 3600–3614
    [12] Smit P Stelling G and Zijlema M 2011 Assessment of nonhydrostatic wave-flow model
    SWASH for directionally spread waves propagating through a barred basin Proceedings of
    ACOMEN 2011 pp 1–10
    [13] Zijlema M Stelling G Smit P 2011 SWASH: An operational public domain code for simulating
    wave fields and rapidly varied flows in coastal waters. Coastal Engineering. № 10(58). pp 992–
    1012
    [14] FLOW-3D® 2008 User’s Manual Version 9.3 Flow Science Inc p 821
    [15] Pan Bayan and Belyaev N D 2019 Week of Science SPbPU: Proceedings of an international
    scientific conference The best reports. pp 3-7
    [16] Girgidov A A 2011 Hybrid simulation in hydrotechnical facilities design and FLOW-3D as a
    tool its realization Magazine of Civil Engineering №3 pp 21-27
    [17] Girgidov A A 2010 Proceeding of the VNIIG vol 260. pp 12-19

    [18] Vasquez J A Walsh B W 2009 CFD simulation of local scour in complex piers under tidal flow,
    33rd IAHR Congress: Water Engineering for a Sustainable Environment, © 2009 by
    International Association of Hydraulic Engineering & Research (IAHR) ISBN: 978-94-90365-
    01-1.
    [19] Shan-Hwei Ou Tai-Wen Hsu and Jian-Feng Lin 2010 Experimental and Numerical Studies on
    Wave Transformation over Artificial Reefs Proceedings of the International Conference on
    Coastal Engineering (Shanghai, China) No 32
    [20] Hirt C and Nichols B 1980 Volume of Fluid Method for the Dynamics of Free Boundaries
    Journal Comp. Phys 39 p 201.

    Capsule-type Vane Tank

    Numerical Simulation Analysis of Liquid Transportation in Capsule-type Vane Tank under Microgravity

    Microgravity 하에서 캡슐형 베인 탱크의 액체 수송에 대한 수치 시뮬레이션 분석

    Abstract

    Li Yong-Qiang1,2 & Dong Jun-Yan1 & Rui Wei
    1Received: 17 June 2019 /Accepted: 4 December 2019
    #Springer Nature B.V. 2020

    In order to research the influence of the guide vane on liquid transmission performance in a tank under microgravity, simulation analysis was carried out with FLOW-3D software. Firstly, it compared the working condition under the charging rate of 10% with the corresponding experiment results of the drop tower and validated the correctness of the simulation process. And then it changed the structure parameters of the guide vane, researched the influence of different quantity, gap and thickness on climbing rate of liquid, and analyzed the causing reasons of these effects in-depth. This paper provided a reference for the design of internal guiding vane of microgravity tank.

    본 논문은 가이드 베인이 미세 중력 상태의 탱크에서 액체 전달 성능에 미치는 영향을 연구하기 위해 FLOW-3D 소프트웨어를 사용하여 시뮬레이션 분석을 수행했습니다. 첫째, 10 % 충전율 하에서 작업 조건을 드롭 타워의 해당 실험 결과와 비교하여 시뮬레이션 프로세스의 정확성을 검증했다. 그리고 가이드 베인의 구조 매개 변수를 변경하고, 액체의 상승 속도에 대한 양, 간격 및 두께의 영향을 연구하고 이러한 영향의 원인을 심도있게 분석했습니다. 이 논문은 미세 중력 탱크의 내부 안내 날개 설계에 대한 참고 자료를 제공했습니다.

    Capsule-type Vane Tank
    Capsule-type Vane Tank
    The relationship curve between the square of climbing height and time with a = 6 mm
    The relationship curve between the square of climbing height and time with a = 6 mm
    The relationship curve between the vane’s liquid transportation and time under different width a
    The relationship curve between the vane’s liquid transportation and time under different width a
    Figure 7. Formation of incident and reflected waves.

    Investigate Impact Force of Dam-Break Flow against Structures by Both 2D and 3D Numerical Simulations

    2D 및 3D 수치 시뮬레이션에 의한 댐 붕괴유동의 구조물 충격력 조사

    1 Faculty of Water Resources Engineering, Thuyloi University, 175 Tay Son, Dong Da, Ha Noi 116705, Vietnam
    2 Hydraulic Construction Institute, 3/95 Chua Boc, Dong Da, Ha Noi 116705, Vietnam
    * Author to whom correspondence should be addressed.
    Academic Editor: Costanza Aricò
    Water 2021, 13(3), 344;

    Abstract

    본 논문의 목적은 일부 2D 및 3D 수치 모델이 침수 지역에 고립된 건물 또는 건물 배열이 있는 곳에서 홍수 파동을 시뮬레이션하는 능력을 조사하는 것이었습니다.

    먼저, 제안된 2D 수치 모델은 구조화된 메시에서 2D 천수(shallow water) 방정식(2D-SWEs)을 해결하기 위한 유한 볼륨 방법(FVM)을 기반으로 했습니다.

    FDS (flux-difference splitting)은 정확한 질량 균형을 얻기 위해 사용되었고 Roe 체계는 Riemann 문제를 근사하기 위해 호출되었습니다.

    둘째, 상업적으로 이용 가능한 3D CFD 소프트웨어 패키지가 선택되었으며, 여기에는 두 가지 난류 모델이 포함된 Flow 3D 모델이 포함되어 있습니다.

    RNG(Renormalized Group) 및 LES(Large-eddy Simulation)를 사용하는 레이놀즈 평균 Navier-Stokes(RAN)입니다. 댐 붕괴 흐름으로 인한 장애물에 대한 충격력의 수치 결과는 3D 솔루션이 2D 솔루션보다 훨씬 낫다는 것을 보여주었습니다.

    건물 배열에 작용하는 충격력의 3D 수치 힘 결과를 보유하고 있는 실험 데이터와 비교함으로써, 속도 유도력이 동적 힘에 미치는 영향은 Froude 숫자의 함수와 사고 파동의 수심 함수에 의해 정량화 되었습니다. 또한, 우리는 힘의 강도의 피크 값의 3D 계산 결과에 대한 초기 물 단계와 댐 붕괴 폭의 영향을 조사했습니다.

    The aim of this paper was to investigate the ability of some 2D and 3D numerical models to simulate flood waves in the presence of an isolated building or building array in an inundated area. Firstly, the proposed 2D numerical model was based on the finite-volume method (FVM) to solve 2D shallow-water equations (2D-SWEs) on structured mesh. The flux-difference splitting method (FDS) was utilized to obtain an exact mass balance while the Roe scheme was invoked to approximate Riemann problems. Secondly, the 3D commercially available CFD software package was selected, which contained a Flow 3D model with two turbulent models: Reynolds-averaged Navier-Stokes (RANs) with a renormalized group (RNG) and a large-eddy simulation (LES). The numerical results of an impact force on an obstruction due to a dam-break flow showed that a 3D solution was much better than a 2D one. By comparing the 3D numerical force results of an impact force acting on building arrays with the existence experimental data, the influence of velocity-induced force on a dynamic force was quantified by a function of the Froude number and the water depth of the incident wave. Furthermore, we investigated the effect of the initial water stage and dam-break width on the 3D-computed results of the peak value of force intensity.

    Keywords: dam-break wave2D numerical modelFlow 3D modelstructuresimpact force

    Introduction

    홍수 위험 분석에 따른 도시 계획은 최근에 큰 연구 과제였습니다.

    건물 또는 건물 그룹에 대한 홍수 파동의 영향에 대한 연구는 하류 지역에 대한 조기 경고 또는 안전 의식 향상에 중요한 역할을 했습니다. 기본적으로 댐 파괴 흐름에 대한 연구는 실험 측정이나 수치 시뮬레이션을 통해 추정 할 수 있습니다 [1,2,3,4,5,6].

    컴퓨터 처리 능력의 증가로 인해 불연속 흐름에 대한 수치 연구가 비용 효율적이되었습니다. 지난 10 년 동안 천수(shallow water) 솔버는 정확성과 계산 능력면에서 크게 향상되었습니다. 침수 가능 지역의 수심 및 속도 프로파일과 같은 유체 역학적 매개 변수에 많은주의를 기울였습니다 [1,2,3,4,5,6,7,8].

    Migot et al. [9]는 도시 홍수의 실험적 모델링에 관한 많은 기사를 검토했습니다. 그 논문에 언급 된 45 개의 작품 중 단 4 개의 프로젝트 만이 장애물에 가해지는 일정한 또는 비정상적인 흐름의 힘 또는 압력을 측정했습니다.

    또한 물리적 및 2D 수치 모델에서 건물 또는 건물 그룹에 돌발 홍수가 미치는 영향에 대한 연구는 거의 없었습니다. 천수(shallow water) 모델은 [10,11]에서 고립된 장애물에 대한 충격의 힘을 예측하는데 사용되었습니다.

    한편 Shige-eda [12]는 액체와 건물 배열 간의 상호 작용을 결정하기 위해 물리적 모델과 2D 수치 체계를 선택했습니다. Aureli와 Shige-eda는 수직 속도와 가속도를 무시하기 때문에 댐 파괴 흐름의 힘을 추정하기 위한 2D 천수(shallow water) 방정식 (SWE)의 단점을 보여주었습니다 [10,12].

    Migot [9]은 또한 장애물 주변의 시뮬레이션된 홍수 흐름에 대한 2D SWE에 대한 여러 출판물이 있었지만 이 주제에 대한 3D 수치 모델에 대한 연구는 거의 없다고 지적했습니다. 최근 전산 유체 역학 (CFD) 3D 시뮬레이션은 유체 흐름과 관련된 문제를 해결하기위한 광범위한 도구가되었습니다.

    댐 파괴 파의 특성은 [13,14,15,16]에 의해 주목되었고 Issakhov [17]는 다양한 종류의 장애물이 압력 분포에 미치는 영향을 조사하기 위해 CFD 방법을 사용했습니다. 그들은 분포가 댐 표면에서 3 배 더 낮다는 것을 밝혔다.

    Aureli [10]는 댐 파괴 파가 구조물에 미치는 영향의 정적 힘을 평가하기 위해 실험 테스트와 2D 및 3D 수치 모델을 사용했습니다. Mokarani [18]는 댐 브레이크 흐름 영향의 VOF 시뮬레이션에서 피크 압력 안정성 조건을 연구했습니다.

    앞서 언급한 작품에서 구조물이나 구조물 군에 작용하는 힘은 압력에 의한 정 수력 또는 정력이었다. 한편, 급류에서 속도로 인한 힘은 압력 력보다 크거나 같았습니다 [19]. Armanini [20]는 정상 흐름에 대해이 항을 추정하기 위한 분석적 표현 만을 제시했습니다. 우리가 아는 한, 건물 그룹에 작용하는 비정상 흐름의 동적 힘을 생성하기 위해 2D 및 3D 수학적 모델을 모두 사용하는 작업은 없습니다.

    따라서 본 연구에서는 제안된 2D 수치 모델과 3D 수학적 모델 모두에 의해 고립 된 장애물 또는 장애물 그룹에 대한 급격한 비정상 흐름의 테스트 사례를 재현했습니다. 수심 및 유속 수문 그래프와 같은 몇 가지 수력 학적 특성이 추정되었으며 측정 된 데이터와 매우 잘 일치했습니다.

    특히 댐 브레이크 흐름이 서로 다른 건물에 가하는 동적인 힘도 시뮬레이션했습니다. 속도 유도 힘이 동적 힘에 미치는 영향 수준을 나타내는 매개 변수는 Froude 수와 입사 파동의 수심의 함수인 것으로 밝혀졌습니다. 또한 붕괴된 댐 사이트 폭 (b)과 초기 수위 (h0)는 충격력의 최대 값에 영향을 미치는 변수로 고려되었습니다.

    Figure 1. (a) Configuration of experiment test (dimension in meters); (b) Gauges on the vertical front face of building.
    Figure 1. (a) Configuration of experiment test (dimension in meters); (b) Gauges on the vertical front face of building.
    Figure 2. (a) Distributed pressure profiles at centerline of front face of column; (b) Comparison of load-time histories simulated by different numerical models
    Figure 2. (a) Distributed pressure profiles at centerline of front face of column; (b) Comparison of load-time histories simulated by different numerical models
    Figure 3. Group of buildings in flooded area.
    Figure 3. Group of buildings in flooded area.
    Figure 4. Water depth and u-velocity profiles at gauge b.
    Figure 4. Water depth and u-velocity profiles at gauge b.
    Figure 5. Water hydrographs at gauges a and c.
    Figure 5. Water hydrographs at gauges a and c.
    Figure 6. Velocity component profiles at gauges a and c.
    Figure 6. Velocity component profiles at gauges a and c.
    Figure 7. Formation of incident and reflected waves.
    Figure 7. Formation of incident and reflected waves.
    Figure 8. Snapshots of streamlines of Froude number at different times: 1.0 s, 2.0 s, 5.0 s and 10 s.
    Figure 8. Snapshots of streamlines of Froude number at different times: 1.0 s, 2.0 s, 5.0 s and 10 s.
    Figure 9. Force in the flow direction exerted on 6 buildings.
    Figure 9. Force in the flow direction exerted on 6 buildings.
    Figure 10. The linear regression between forces per unit width (F) and q2b/h0.
    Figure 10. The linear regression between forces per unit width (F) and q2b/h0.

    Conclusions

    댐 붕괴 흐름으로 인한 홍수 파도는 높은 속도 또는 큰 깊이가 관련되었을 때 건물에 큰 영향을 미칩니다. 본 논문에서는 2D 및 3D 수치 모델의 건물 및 건물 그룹에 대한 빠른 흐름에 의해 발생하는 유압 특성과 충격 부하를 추정할 수 있는 능력을 조사했습니다. 천수(shallow water) 방정식에 기초한 2D 수학 모델은 FDS 방법으로 해결되었으며, FDS 방법은 최신 버전의 Flow 3D 유체 역학 모델과 함께 사용되었습니다. 연구의 주요 발견은 다음과 같습니다.
    (1) 수심 또는 속도 프로파일을 공식화하기 위해 2D 및 3D 수치 솔루션은 모두 매우 유사합니다. 제안된 2D 수치 모델은 정적 힘의 최대 값 뿐만 아니라 수심 및 속도 구성 요소를 포함하는 유압 특성을 예측하는 데 적합합니다. 그러나 LES 및 RAN 난류 모듈이 포함된 3D 유체역학 모델은 2D 얕은 흐름 모델이 1개만 제공하는 동안 두 개의 최고 충격 부하를 잘 포착할 수 있습니다. 일반적으로 3D 결과는 실험 결과와 더 가깝습니다.
    (2) 여러 건물에 대한 정적 및 동적 힘은 모두 LES 모듈을 사용하여 Flow 3D에 의해 계산되었습니다. 건물에서 속도에 의한 힘과 압력의 역할은 위치에 따라 다릅니다. 댐 현장 근처에서, 속도 유도 힘은 댐 파괴 파동의 주 방향에서 멀리 떨어져 있거나 두 번째 배열에서 압력 힘이 더 중요합니다. 속도 유도 힘의 영향은 매개 변수 α에 의해 정량화되며, 이는 사고파의 Froude 숫자와 수심 함수로 수행됩니다. q2b/h0과 정적 힘과 동적 힘의 피크 강도 사이의 선형 회귀 관계는 합리적인 R-제곱 양으로 해결됩니다.

    추가 연구에서, 제시된 2D 수치 모델의 견고성과 효과는 더 명확하게 드러날 것입니다. 대규모 도메인에 대한 홍수 흐름을 시뮬레이션하는 데 쉽게 적용할 수 있습니다. 또한, α 매개변수의 제안된 방정식(21)은 실제 사례 연구에서 다운스트림 영역의 건물에 대한 속도 유도 힘의 영향을 정확하게 평가하기 위한 매우 의미가 있습니다. 이 매개 변수의 정확도 수준을 높이려면 서로 다른 조건에서 장애물에 작용하는 여러 가지 힘 실험이 구현되어야 합니다.

    References

    1. Testa, G.; Zuccala, D.; Alcrudo, F.; Mulet, J.; Frazao, S.S. Flash flood flow experiment in a simplifed urban district. J. Hydraul. Res. 200745, 37–44. [Google Scholar] [CrossRef]
    2. Soares-Frazao, S.; Zech, Y. Dam-break flow through an idealized city. J. Hydraul. Res. 200846, 648–665. [Google Scholar] [CrossRef]
    3. Soares-Frazão, S.; Zech, Y. Experimental study of dam-break flow against an isolated obstacle. J. Hydraul. Res. 200745, 27–36. [Google Scholar] [CrossRef]
    4. Soares-Frazão, S. Experiments of dam-break wave over a triangular bottom sill. J. Hydraul. Res. 200745, 19–26. [Google Scholar] [CrossRef]
    5. di Cristo, C.; Evangelista, S.; Greco, M.; Iervolino, M.; Leopardi, A.; Vacca, A. Dam-break waves over an erodible embankment: Experiments and simulations. J. Hydraul. Res. 201856, 196–210. [Google Scholar] [CrossRef]
    6. Evangelista, S. Experiments and numerical simulations of dike erosion due to a wave impact. Water 20157, 5831–5848. [Google Scholar] [CrossRef]
    7. Li, Y.L.; Yu, C.H. Research on dam break flow induced front wave impacting a vertical wall based on the CLSVOF and level set methods. Ocean Eng. 2019178, 442–462. [Google Scholar] [CrossRef]
    8. Özgen, I.; Zhao, J.; Liang, D.; Hinkelmann, R. Urban flood modeling using shallow water equations with depth-dependent anisotropic porosity. J. Hydrol. 2016541, 1165–1184. [Google Scholar] [CrossRef]
    9. Mignot, E.; Li, X.; Dewals, B. Experimental modelling of urban flooding: A review. J. Hydrol. 2019568, 334–342. [Google Scholar] [CrossRef]
    10. Aureli, F.; Dazzi, A.; Maranzoni, A.; Mignosa, P.; Vacondio, R. Experimental and numerical evaluation of the force due to the impact of a dam break wave on a structure. Adv. Water Resour. 201576, 29–42. [Google Scholar] [CrossRef]
    11. Milanesi, L.; Pilotti, M.; Belleri, A.; Marini, A.; Fuchs, S. Vulnerability to flash floods: A simplified structural model for masonry buldings. Water Resour. Res. 201854, 7177–7197. [Google Scholar] [CrossRef]
    12. Shige-eda, M.; Akiyama, J. Numerical and experimental study on two dimensional flood flows with and without structures. J. Hydraul. Eng. 2003129, 817–821. [Google Scholar] [CrossRef]
    13. Cagatay, H.O.; Kocaman, S. Dam break flows during initial stage using SWE and RANs approaches. J. Hydraul. Res. 201048, 603–611. [Google Scholar] [CrossRef]
    14. Yang, S.; Yang, W.; Qin, S.; Li, Q.; Yang, B. Numerical study on characteristics of dam break wave. Ocean Eng. 2018159, 358–371. [Google Scholar] [CrossRef]
    15. Robb., D.M.; Vasquez., J.A. Numerical simulation of dam break flows using depth averaged hydrodynamic and three dimensional CFD models. In Proceedings of the 22nd Canadian Hydrotechnical Conference, Ottawa, ON, Canada, 28–30 April 2015. [Google Scholar]
    16. Kocaman, S.; Evangelista, S.; Viccione, G.; Guzel, H. Experimental and Numerical analysis of 3D dam break waves in an enclosed domain with a single oriented obstacles. Environ. Sci. Proc. 20202, 35. [Google Scholar] [CrossRef]
    17. Issakhov, A.; Zhandaulet, Y.; Nogaeva, A. Numerical simulation of dambreak flow for various forms of the obstacle by VOF method. Int. J. Multiph. Flow 2018109, 191–206. [Google Scholar] [CrossRef]
    18. Mokarani, C.; Abadie, S. Conditions for peak pressure stability in VOF simulations of dam break flow impact. J. Fluids Struct. 201662, 86–103. [Google Scholar] [CrossRef]
    19. Liu, L.; Sun, J.; Lin, B.; Lu, L. Building performance in dam break flow—an experimental sudy. Urban Water J. 201815, 251–258. [Google Scholar] [CrossRef]
    20. Armanini, A.; Larcher, M.; Odorizzi, M. Dynamic impact of a debris flow front against a vertical wall. In Proceedings of the 5th international conference on debris-flow hazards mitigation: Mechanics, prediction and assessment, Padua, Italy, 14–17 June 2011. [Google Scholar] [CrossRef]
    21. Hubbard, M.E.; Garcia Navarro, P. Flux difference splitting and the balancing of source terms and flux gradients. J. Comput. Phys. 2000165, 89–125. [Google Scholar] [CrossRef]
    22. Roe, P.L. A basis for upwind differencing of the two-dimensional unsteady Euler equations. In Numerical Methods in Fluids Dynamics II; Oxford University Press: Oxford, UK, 1986. [Google Scholar]
    23. Bradford, S.F.; Sander, B. Finite volume model for shallow water flooding of arbitrary topography. J. Hydraul. Eng. (ASCE) 2002128, 289–298. [Google Scholar] [CrossRef]
    24. Brufau, P.; Garica-Navarro, P. Two dimensional dam break flow simulation. Int. J. Numer. Meth. Fluids 200033, 35–57. [Google Scholar] [CrossRef]
    25. Hien, L.T.T. 2D Numerical Modeling of Dam-Break Flows with Application to Case Studies in Vietnam. Ph.D. Thesis, Brescia University, Brescia, Italy, 2014. [Google Scholar]
    26. Hien, L.T.T.; Tomirotti, M. Numerical modeling of dam break flows over complex topography. Case studies in Vietnam. In Proceedings of the 19th IAHR-APD Congress 2014, Hanoi, Vietnam, 21–24 September 2014; ISBN 978-604-82-1383-1. [Google Scholar]
    27. Flow-3D, Version 12.0; User Mannual; Flow Science Inc.: Santa Fe, NM, USA, 2020.
    28. Guney, M.S.; Tayfur, G.; Bombar, G.; Elci, S. Distorted physical model to study sudden partial dam break flow in an urban area. J. Hydraul. Eng. 2014140, 05014006. [Google Scholar] [CrossRef]
    29. Shige-eda, M.; Akiyama, J. Discussion and Closure to “Numerical and experimental study on two dimensional flood flows with and without structures” by Mirei Shige-eda and Juichiro Akiyama. J. Hydraul. Eng. 2005131, 336–337. [Google Scholar] [CrossRef]
    30. Ritter, A. Die Fortpflanzung der Wasserwelle (Generation of the water wave). Z. Ver. Dtsch. Ing. 189236, 947–954. [Google Scholar]
    Fig. 9 (a) Velocity field, keyhole profile, and breakage of the keyhole to form bubble and (b) 2D temperature and velocity field along the longitudinal section

    A Numerical Study on the Keyhole Formation During Laser Powder Bed Fusion Process

    Keyhole에 대한 수치적 연구 : 레이저 분말 중 형성 베드 퓨전 공정

    Subin Shrestha1
    J.B. Speed School of Engineering,University of Louisville,Louisville, KY 40292
    e-mail: subin.shrestha@louisville.edu

    Y. Kevin Chou
    J.B. Speed School of Engineering,University of Louisville,Louisville, KY 40292
    e-mail: kevin.chou@louisville.edu

    LPBF (Laser Powder Bed fusion) 공정 중 용융 풀의 동적 현상은 복잡하고 공정 매개 변수에 민감합니다. 에너지 밀도 입력이 특정 임계 값을 초과하면 키홀이라고 하는 거대한 증기 함몰이 형성 될 수 있습니다.

    이 연구는 수치 분석을 통해 LPBF 과정에서 키홀 거동 및 관련 기공 형성을 이해하는 데 중점을 둡니다. 이를 위해 이산 분말 입자가 있는 열 유동 모델이 개발되었습니다.

    이산 요소 방법 (DEM)에서 얻은 분말 분포는 계산 영역에 통합되어 FLOW-3D를 사용하는 3D 프로세스 물리학 모델을 개발합니다.

    전도 모드 중 용융 풀 형성과 용융의 키홀 모드가 식별되고 설명되었습니다. 높은 에너지 밀도는 증기 기둥의 형성으로 이어지고 결과적으로 레이저 스캔 트랙 아래에 구멍이 생깁니다.

    또한 다양한 레이저 출력과 스캔 속도로 인한 Keyhole 모양을 조사합니다. 수치 결과는 동일한 에너지 밀도에서도 레이저 출력이 증가함에 따라 Keyhole크기가 증가 함을 나타냅니다. Keyhole은 더 높은 출력에서 ​​안정되어 레이저 스캔 중 Keyhole 발생을 줄일 수 있습니다.

    The dynamic phenomenon of a melt pool during the laser powder bed fusion (LPBF) process is complex and sensitive to process parameters. As the energy density input exceeds a certain threshold, a huge vapor depression may form, known as the keyhole. This study focuses on understanding the keyhole behavior and related pore formation during the LPBF process through numerical analysis. For this purpose, a thermo-fluid model with discrete powder particles is developed. The powder distribution, obtained from a discrete element method (DEM), is incorporated into the computational domain to develop a 3D process physics model using flow-3d. The melt pool formation during the conduction mode and the keyhole mode of melting has been discerned and explained. The high energy density leads to the formation of a vapor column and consequently pores under the laser scan track. Further, the keyhole shape resulted from different laser powers and scan speeds is investigated. The numerical results indicated that the keyhole size increases with the increase in the laser power even with the same energy density. The keyhole becomes stable at a higher power, which may reduce the occurrence of pores during laser scanning.

    Keywords: additive manufacturing, keyhole, laser powder bed fusion, porosity

    Fig. 1 (a) Powder added to the dispenser platform and (b) powder particles settled over build plate after the recoating process
    Fig. 1 (a) Powder added to the dispenser platform and (b) powder particles settled over build plate after the recoating process
    Fig. 2 3D computational domain used for single-track simulation
    Fig. 2 3D computational domain used for single-track simulation
    Fig. 3 Temperature-dependent material properties of Ti-6Al-4V
    Fig. 3 Temperature-dependent material properties of Ti-6Al-4V
    Fig. 4 Powder and substrate melting during laser application
    Fig. 4 Powder and substrate melting during laser application
    Fig. 5 Melt region formed after complete melting and solidification
    Fig. 5 Melt region formed after complete melting and solidification
    Fig. 6 Melt pool boundary comparison between the experiment [25] and the simulation
    Fig. 6 Melt pool boundary comparison between the experiment [25] and the simulation
    Fig. 7 Equilibrium points during the formation of vapor column [27]
    Fig. 7 Equilibrium points during the formation of vapor column [27]
    Fig. 8 Multiple reflection vectors from the keyhole wall
    Fig. 8 Multiple reflection vectors from the keyhole wall
    Fig. 9 (a) Velocity field, keyhole profile, and breakage of the keyhole to form bubble and (b) 2D temperature and velocity field along the longitudinal section
    Fig. 9 (a) Velocity field, keyhole profile, and breakage of the keyhole to form bubble and (b) 2D temperature and velocity field along the longitudinal section
    Fig. 10 Fluid flow in the transverse direction during keyhole melting
    Fig. 10 Fluid flow in the transverse direction during keyhole melting
    Fig. 11 Melt pool boundary compared with the experiment [21] for 195 W laser power and 400 mm/s scan speed
    Fig. 11 Melt pool boundary compared with the experiment [21] for 195 W laser power and 400 mm/s scan speed
    Fig. 12 Melt region formed after complete melting and solidification
    Fig. 12 Melt region formed after complete melting and solidification
    Fig. 13 2D images of the pores formed at the beginning of the single track and their 3D-rendered morphology
    Fig. 13 2D images of the pores formed at the beginning of the single track and their 3D-rendered morphology
    Fig. 14 Pore number and volume from a different level of power with LED = 0.4 J/mm [29]
    Fig. 14 Pore number and volume from a different level of power with LED = 0.4 J/mm [29]
    Fig. 15 Keyhole shape at different time steps from different parameters: (a) P = 100 W, v = 250 mm/s, (b) P = 200 W, v = 500 mm/s, (c) P = 300 W, v = 750 mm/s, and (d) P = 400 W, v = 1000 mm/s
    Fig. 15 Keyhole shape at different time steps from different parameters: (a) P = 100 W, v = 250 mm/s, (b) P = 200 W, v = 500 mm/s, (c) P = 300 W, v = 750 mm/s, and (d) P = 400 W, v = 1000 mm/s
    Fig. 16 Intensity dependence in the relationship between vapor column and evaporation pressure [27]
    Fig. 16 Intensity dependence in the relationship between vapor column and evaporation pressure [27]
    Fig. 17 Temperature distribution when laser has moved 0.8 mm with P = 300 W, v = 750 mm/s and P = 400 W, v = 1000 mm/s
    Fig. 17 Temperature distribution when laser has moved 0.8 mm with P = 300 W, v = 750 mm/s and P = 400 W, v = 1000 mm/s
    Fig. 18 Melt region with different level of power with LED of 0.4 J/mm
    Fig. 18 Melt region with different level of power with LED of 0.4 J/mm

    References

    [1] Bauereiß, A., Scharowsky, T., and Körner, C., 2014, “Defect Generation and
    Propagation Mechanism During Additive Manufacturing by Selective Beam
    Melting,” J. Mater. Process. Technol., 214(11), pp. 2522–2528.
    [2] Gong, H., Rafi, K., Gu, H., Starr, T., and Stucker, B., 2014, “Analysis of Defect
    Generation in Ti–6Al–4V Parts Made Using Powder Bed Fusion Additive
    Manufacturing Processes,” Add. Manuf., 1(2014), pp. 87–98.
    [3] Wang, Y., Kamath, C., Voisin, T., and Li, Z., 2018, “A Processing Diagram for
    High-Density Ti-6Al-4V by Selective Laser Melting,” Rapid Prototyping J., 24
    (9), pp. 1469–1478.
    [4] Khairallah, S. A., and Anderson, A., 2014, “Mesoscopic Simulation Model of
    Selective Laser Melting of Stainless Steel Powder,” J. Mater. Process. Technol.,
    214(11), pp. 2627–2636.
    [5] Yadroitsev, I., Gusarov, A., Yadroitsava, I., and Smurov, I., 2010, “Single Track
    Formation in Selective Laser Melting of Metal Powders,” J. Mater. Process.
    Technol., 210(12), pp. 1624–1631.
    [6] Xia, M., Gu, D., Yu, G., Dai, D., Chen, H., and Shi, Q., 2016, “Influence of Hatch
    Spacing on Heat and Mass Transfer, Thermodynamics and Laser Processability
    During Additive Manufacturing of Inconel 718 Alloy,” Int. J. Mach. Tools
    Manuf., 109(2016), pp. 147–157.
    [7] Lee, Y., and Zhang, W., 2016, “Modeling of Heat Transfer, Fluid Flow and
    Solidification Microstructure of Nickel-Base Superalloy Fabricated by Laser
    Powder bed Fusion,” Add. Manuf., 12(2016), pp. 178–188.
    [8] Wu, Y.-C., San, C.-H., Chang, C.-H., Lin, H.-J., Marwan, R., Baba, S., and
    Hwang, W.-S., 2018, “Numerical Modeling of Melt-Pool Behavior in Selective
    Laser Melting with Random Powder Distribution and Experimental
    Validation,” J. Mater. Process. Technol., 254(2018), pp. 72–78.
    [9] Khairallah, S. A., Anderson, A. T., Rubenchik, A., and King, W. E., 2016, “Laser
    Powder-bed Fusion Additive Manufacturing: Physics of Complex Melt Flow and
    Formation Mechanisms of Pores, Spatter, and Denudation Zones,” Acta
    Materialia, 108(2016), pp. 36–45.
    [10] Tan, J., Tang, C., and Wong, C., 2018, “A Computational Study on Porosity
    Evolution in Parts Produced by Selective Laser Melting,” Metall. Mater. Trans.
    A, 49A(8), pp. 3663–3673.
    [11] Leitz, K.-H., Singer, P., Plankensteiner, A., Tabernig, B., Kestler, H., and Sigl,
    L. J. M. P. R., 2017, “Multi-Physical Simulation of Selective Laser Melting,”
    Metal Powder Report, 72(5), pp. 331–338.
    [12] Zhao, C., Fezzaa, K., Cunningham, R. W., Wen, H., Carlo, F., Chen, L., Rollett,
    A. D., and Sun, T., 2017, “Real-time Monitoring of Laser Powder Bed Fusion
    Process Using High-Speed X-ray Imaging and Diffraction,” Sci. Rep., 7(1),
    p. 3602.
    [13] Parab, N. D., Zhao, C., Cunningham, R., Escano, L. I., Fezzaa, K., Everhart, W.,
    Rollett, A. D., Chen, L., and Sun, T., 2018, “Ultrafast X-ray Imaging of Laser–
    Metal Additive Manufacturing Processes,” J. Synchrotron Radiat., 25(5),
    pp. 1467–1477.
    [14] Cunningham, R., Zhao, C., Parab, N., Kantzos, C., Pauza, J., Fezzaa, K., Sun, T.,
    and Rollett, A. D., 2019, “Keyhole Threshold and Morphology in Laser Melting
    Revealed by Ultrahigh-Speed X-Ray Imaging,” Science, 363(6429), pp. 849–852.
    [15] Shrestha, S., Starr, T., and Chou, K., 2019, “A Study of Keyhole Porosity in
    Selective Laser Melting: Single Track Scanning With Micro-CT Analysis,”
    ASME J. Manuf. Sci. Eng., 141(7), pp. 1–23.
    [16] Ye, J., Rubenchik, A. M., Crumb, M. F., Guss, G., and Matthews, M. J., 2018,
    “Laser Absorption and Scaling Behavior in Powder Bed Fusion Additive
    Manufacturing of Metals,” Proceedings of the CLEO: Science and Innovations,
    Optical Society of America, San Jose, CA, May 13–18, Optical Society of
    America, p. JW2A.117.
    [17] Mishra, B., and Rajamani, R. K., 1992, “The Discrete Element Method for the
    Simulation of Ball Mills,” Appl. Math. Modell., 16(11), pp. 598–604.
    [18] Yan, W., Qian, Y., Ge, W., Lin, S., Liu, W. K., Lin, F., and Wagner, G. J., 2018,
    “Meso-Scale Modeling of Multiple-Layer Fabrication Process in Selective
    Electron Beam Melting: Inter-Layer/Track Voids Formation,” Materials and
    Design, 141(2018), pp. 210–219.
    [19] Kloss, C., Goniva, C., Hager, A., Amberger, S., and Pirker, S., 2012, “Models,
    Algorithms and Validation for Opensource DEM and CFD–DEM,” Prog.
    Comput. Fluid Dynam. Int. J., 12(2–3), pp. 140–152.
    [20] Escano, L. I., Parab, N. D., Xiong, L., Guo, Q., Zhao, C., Fezzaa, K., Everhart,
    W., Sun, T., and Chen, L., 2018, “Revealing Particle-Scale Powder Spreading
    Dynamics in Powder-Bed-Based Additive Manufacturing Process by
    High-Speed X-Ray Imaging,” Sci. Rep., 8(1), p. 15079.
    [21] Gong, H., Gu, H., Zeng, K., Dilip, J., Pal, D., Stucker, B., Christiansen, D., Beuth,
    J., and Lewandowski, J. J., 2014, “Melt Pool Characterization for Selective Laser
    Melting of Ti-6Al-4V Pre-Alloyed Powder,” Proceedings of the Solid Freeform
    Fabrication Symposium, Austin, TX, Aug. 4–6, pp. 256–267.
    [22] Mills, K. C., 2002, Recommended Values of Thermophysical Properties for
    Selected Commercial Alloys, Woodhead Publishing, Cambridge, UK.
    [23] Shrestha, S., and Chou, K., 2017, “A Build Surface Study of Powder-Bed
    Electron Beam Additive Manufacturing by 3D Thermo-Fluid Simulation and
    White-Light Interferometry,” Int. J. Mach. Tools Manuf., 121(2017), pp. 37–49.
    [24] Cho, J.-H., and Na, S.-J., 2006, “Implementation of Real-Time Multiple
    Reflection and Fresnel Absorption of Laser Beam in Keyhole,” J. Phys. D:
    Appl. Phys., 39(24), p. 5372.
    [25] Dilip, J., Zhang, S., Teng, C., Zeng, K., Robinson, C., Pal, D., and Stucker, B.,
    2017, “Influence of Processing Parameters on the Evolution of Melt Pool,
    Porosity, and Microstructures in Ti-6Al-4V Alloy Parts Fabricated by Selective
    Laser Melting,” Prog. Add. Manuf., 2(3), pp. 157–167.
    [26] Bertoli, U. S., Wolfer, A. J., Matthews, M. J., Delplanque, J.-P. R., and Schoenung,
    J. M., 2017, “On the Limitations of Volumetric Energy Density as a Design
    Parameter for Selective Laser Melting,” Mater. Des., 113(2017), pp. 331–340.
    [27] Kroos, J., Gratzke, U., and Simon, G., 1993, “Towards a Self-Consistent Model of
    the Keyhole in Penetration Laser Beam Welding,” J. Phys. D: Appl. Phys., 26(3),
    p. 474.
    [28] Martin, A., Calta, N., Hammons, J., Khairallah, S., Nielsen, M., Shuttlesworth, R.,
    Sinclair, N., Matthews, M., Jeffries, J., and Willey, T., 2019, “Ultrafast Dynamics
    of Laser-Metal Interactions in Additive Manufacturing Alloys Captured by In Situ
    X-Ray Imaging,” Mater. Today Adv., 1(2019), p. 100002.
    [29] Shrestha, S., Starr, T., and Chou, K., 2018, “Individual and coupled contributions
    of laser power and scanning speed towards process-induced porosity in selective
    laser melting,” Proceedings of the Solid Freeform Fabrication Symposium,
    Austin, TX, Aug. 13–15, pp. 1400–1409.
    [30] Hann, D., Iammi, J., and Folkes, J., 2011, “A Simple Methodology for Predicting
    Laser-Weld Properties From Material and Laser Parameters,” J. Phys. D: Appl.
    Phys., 44(44), p. 445401.
    [31] Trapp, J., Rubenchik, A. M., Guss, G., and Matthews, M. J., 2017, “In Situ
    Absorptivity Measurements of Metallic Powders During Laser Powder-bed
    Fusion Additive Manufacturing,” Appl. Mat. Today, 9(2017), pp. 341–349.

    Cad2Stl

    FLOW-3D 유틸리티 프로그램 안내

    이 문서에서는 FLOW-3D에서 사용할 수 있는 일부 Utility Program에 대해 설명합니다. 유틸리티 프로그램의 목적은 시뮬레이션을 수행할 때 반드시 필요한 것은 아니지만 특정 작업을 쉽게 수행할 수 있도록 돕는 것입니다. 각 개별 유틸리티의 사용법은 다음과 같습니다.

    1. 파일 변환 및 STL 품질 검사 도구

    FLOW-3D는 중립 형식인 STL파일 형식만 지원하며 대부분의 CAD 패키지에서 STL형식을 지원하지만 형상을 STL형식으로 만들 수 없는 이유가 있을 수 있습니다. 이로 인해 FLOW-3D 사용자는 여러 파일 변환 유틸리티를 사용할 필요가 있을 수 있습니다. 또한 STL 파일 품질을 확인하는데 사용할 수 있는 여러 유틸리티도 사용할 수 있습니다. 아래 나열된 이러한 유틸리티는 다음 섹션에서 자세히 설명합니다.

    • Cad2Stl : 다양한 CAD 형식에서 변환 파일을 사용하는.STL파일
    • Topo2STL : 파일을topo형식에서.STL파일로 변환하는 데 사용
    • MiniMagics :.STL파일의 오류를 확인하는 데 사용
    • qAdmesh :.STL파일의 오류를 확인하고 사소한 문제를 해결하는데 사용

    Cad2Stl

    Cad2Stl 은 다른 CAD 파일 형식을 FLOW-3D에서 사용되는 STL 파일 형식으로 변환하기 위한 파일 변환 도구입니다. Cad2Stl 은 다음 파일 형식을 STL 형식으로 변환합니다.

    • Autodesk 3D Max :.3ds
    • Autodesk 별명 :.obj
    • IGES: .igs,.iges
    • BREP :.brep
    • 단계 : .stp,.step
    • 아바쿠스 6.2+ :.inp
    • NASTRAN :.blk
    • Marc Mentat : 고정 형식과 쉼표로 구분.dat

    Cad2Stl 은 파일에서 역 법선 벡터를 보정하는 기능도 있습니다. 이 유틸리티는 유지 보수 계약이 유효한 모든 FLOW-3D 고객에게 무료로 제공되며 FLOW-3D Usre Site의 유틸리티 페이지에서 다운로드 할 수 있습니다.

    Cad2Stl 은 Flow Science Japan에서 FLOW-3D 사용자를 위해 개발되었습니다 .

    Cad2Stl Program
    1. 변환 목록에 변환할 파일 추가
      • 추가 -변환 목록에 파일을 추가합니다.
      • 제거 -변환 목록에서 파일을 제거합니다. 제거하려면 변환 목록에서 파일을 강조 표시하고 제거를 선택하십시오.
      • 기본적으로 파일 이름은 import file 이름과 일치하는 CAD파일을 STL파일 이름으로 지정하는데 변경이 필요하면 더블 클릭하고 이름을 바꾸면 변경할 수 있습니다.
    2. 구체화 옵션을 사용하여 STL 파일의 품질을 선택하십시오. 선택하고 볼 수 있는 네 가지 수준의 정확도가 있습니다. 파일이 변환될 때마다 STL로 작성된 파일이 표시되므로 사용자가 만족스럽거나 더 높은 수준의 세분화가 필요한지 여부를 결정할 수 있습니다. 정확성이 향상되면 파일 크기는 증가하지만 처리 시간은 크게 증가하지 않습니다. 다른 파일 형식을 한 번에 로드하고 변환할 수 있습니다. 또한 변환 프로세스가 완료되면 파일을 로드하고 표시하기 위한 대화 상자가 열립니다. 이것은 BREP, IGES및 STEP 파일 형식에만 적용됩니다.
    3. 원하는 작업을 선택하십시오. 다른 파일 형식을 한 번에 로드하고 변환할 수 있습니다. 또한 변환 프로세스가 완료되면 파일을 로드하고 표시하기 위한 대화 상자가 열립니다.
      • 변환 -파일을 변환합니다. 한 파일을 변환하려면 로드할 파일 목록에서 해당 파일을 강조 표시하여 변환하십시오.
      • 모두 변환 -모든 파일을 변환
      • 표시 -변환된 파일을 강조 표시합니다
      • 면 방향 수정 -일반 수정 루틴
      • 변환 목록 숨기기 -더 나은 부품 표시를 위해 보기 화면을 증가 시킵니다.
      • 와이어 프레임 오버레이 -각 STL 패싯의 패싯 모서리를 오버레이 합니다. 이것은 오른쪽 하단의 확인란입니다.
      • 로그 지우기 – 변환 로그 텍스트 상자에 대한 모든 데이터 출력을 지웁니다.
    4. 종료 -프로그램을 닫습니다

    qAdmesh

    qAdmesh는 .STL파일에 오류 가 있는지 확인하는 도구이며 연결이 끊어진 패싯, 반전된 법선, 연결이 끊어진 패싯 및 누락된 패싯과 같은 사소한 문제를 해결하는 데 사용할 수 있습니다. qAdmesh를 시작하려면:

    • GUI에서: Model Setup 탭의 Tools ‣ qAdmesh로 이동하십시오.
    • Windows: 바탕 화면 아이콘을 클릭하거나 시작 메뉴에서 FLOW-3D v12.0 폴더의 형상 도구 하위 디렉토리에 있는 Admesh 항목으로 이동하십시오.
    • Linux의 경우: $F3D_HOME/utilities/qAdmesh을 실행하십시오.

    명령: qAdmesh를 열고 찾아보기 버튼을 사용하여 지오메트리 파일을 로드 하십시오. 문제를 해결하고 수정 사항으로 새 형상 파일을 생성하려면 기본 옵션을 그대로 두고 출력 유형을 선택하고 새 형상 파일의 경로를 지정하십시오. 이진 STL 은 ASCII STL 옵션 보다 작은 파일을 생성하므로 권장됩니다 (이진 및 ASCII 형식 만 FLOW-3D 로 인식됨). 그런 다음 적용을 클릭하여 파일을 확인하고 수정하십시오.

    qAdmesh program
    qAdmesh program

    qAdmesh의 출력은 인터페이스의 메시지 섹션에 표시됩니다. 출력에는 감지된 오류와 출력 옵션이 선택된 경우 이러한 문제점을 해결하기 위해 수행할 조치가 표시됩니다.

    사용자 정의 검사 옵션은 파일을 고정할 때 프로그램이 어떤 작업을 수행하는지에 대한 자세한 제어를 제공할 수 있습니다. 또한 변형 및 공차 탭에는 .STL 파일의 회전, 미러링, 크기 조정, 변환 및 병합 기능을 제공하는 옵션이 있습니다.

    qAdmesh는 무료 유틸리티입니다만 FSI에서 지원하지 않습니다. qAdmesh가 문제를 해결하는 능력은 심각도에 따라 다릅니다. 문제의 수가 증가함에 따라 qAdmesh 가 문제를 해결할 수 있는 가능성이 줄어 듭니다. 문제를 해결할 수 없는 경우 CAD 패키지를 사용하여  .STL 파일을 재생성 하는 것이 좋습니다.

    MiniMagics 

    MiniMagics 는 무료 STL파일 시각화 및 복구 유틸리티입니다. 설치는 FLOW-3D 홈 디렉토리 의 Utilites 폴더에서 찾을 수 있으며 파일 분석 및 복구를 위한 유용한 도구로 qAdmesh에서 수행된 수정 사항을 시각화하거나 qAdmesh의 대안으로 사용할 수 있습니다.

    $F3D_HOME/UtilitiesSTL

    • Topo2STL

    FLOW-3D가 지원하는 유일한 CAD 파일 형식은 .STL이지만 형식을 포함하여 다른 형식의 지형 데이터를 갖는 것은 드문 일이 아닙니다. Topo2STL의 유틸리티로 변환할 수 있습니다. Topo2STL 은 Windows 시스템에서만 사용 가능하며 유틸리티 드롭 다운 메뉴에서 액세스 할 수 있습니다.

    명령

    1. 지형 파일은 다음 형식의 ASCII 파일입니다. 각 선은 점을 나타내며 동일한 단위 시스템에서 3 개의 좌표 (일반적으로 피트 또는 미터)를 포함합니다. 좌표는 공백으로 구분됩니다. 선의 좌표 순서는 XYZ 여야 합니다. 여기서 Z는 표고입니다. 두 좌표는 동일한 XY 점을 공유할 수 없습니다. 포인트의 순서 (파일의 줄)는 중요하지 않습니다. 좌표를 포함하지 않는 머리글 줄이나 꼬리 줄이 없어야 합니다.
    2. Topo2stl.exe유틸리티가 추출된 위치에 있는 파일을 실행하여 Topo2STL에 액세스 할 수 있습니다.
    3. 유틸리티를 시작하면 변환할 파일을 선택하라는 topo 파일 찾아보기 창이 나타납니다. 파일 찾아보기 창을 이용하여 파일을 선택합니다.
    4. topo파일이 선택되면, Topo2STL의 창이 나타나고, X, Y의 범위와 Z 계산할 topo데이터 익스텐트가 계산되면 Topo 데이터 익스텐트 및 데이터의 총 포인트 수에 대한 정보가 Information: Topo data extents 아래에 표시됩니다.
    Topo2STL
    Topo2STL
    Topo2STL
    Topo2STL
    1. 변환에 필요한 사용자 입력은 공간 분해능 및 STL 최소 Z 좌표입니다. 기본적으로 공간 해상도는 0.002 * min (X 범위, Y 범위)이고 STL 최소 Z 좌표는 ZMIN-(ZMAX-ZMIN)입니다. 여기서 ZMIN 및 ZMAX는 Topo 데이터의 범위입니다.
      • 공간 해상도는 STL 파일을 생성하는 동안 Topo 데이터가 얼마나 정밀하게 분석되는지 제어합니다.
      • STL 최소 Z 좌표는 Topo 데이터의 ZMAX보다 작은 값이어야 합니다. 이것은 STL파일의 최소 ​​Z 두께를 효과적으로 설정합니다.
    2. Browse 버튼은 파일 출력 위치를 설정하는 데 사용할 수 있습니다.
    3. 변환을 클릭하면 변환 프로세스가 시작됩니다. 이 시점에서 변환 취소를 사용하여 변환이 완료되거나 종료될 때까지 Topo2STL 창을 닫을 수 없습니다.
    Topo2STL
    Topo2STL
    1. 변환이 완료 (또는 종료)되면 변환 단추가 변환 추가로 변경되어 사용자가 변환할 다른 Topo 파일을 선택할 수 있습니다.
    Topo2STL
    1. FSAI를 사용한 유한 요소 메쉬 파일 형식 변환

    FSAI의 도구에서 유한 요소 메시를 변환하는 유틸리티입니다 Abaqus6.2 이후 형식과 NASTRAN 벌크 형식에 사용되는 형식을 변환하는 FSAI는 유틸리티 드롭 다운 메뉴에서 액세스 할 수 있습니다. FSAI를 사용하려면 다음을 수행하십시오. EXODUS II

    • 적절한 모드에서 유틸리티를 엽니다 (초기 메쉬의 Abaqus 형식인지 NASTRAN 형식인지 여부에 따라 다름 )
    • 파일에서 생성 필드에서 입력 유한 요소 메쉬를 찾습니다.
    • 생성된 파일 위치 필드에서 원하는 출력 위치를 찾으십시오.
    • 생성된 파일 이름 필드에서 원하는 출력 파일 이름을 설정하십시오.
    • 생성을 누릅니다.

     노트

    이 FSAI 프로그램을 사용하려면 FLOW-3D 와 별개의 라이센스가 필요합니다. 자세한 내용은 FLOW-3D 영업 담당자에게 문의하십시오.

    1. 계산기

    유틸리티 드롭 다운 메뉴에 여러 계산기가 추가되어 알려진 매개 변수 (예: 유체 속성 등)를 기반으로 입력 수량을 추정할 수 있습니다. 사용 가능한 계산기는 다음을 계산합니다.

    • 냉각 채널의 열전달 계수
    • 재료 특성 및 시뮬레이션 시간에 따른 열 침투 깊이
    • 샷 슬리브의 유체 높이
    • 고압 다이캐스팅을 위한 피스톤 속도
    • 밸브 압력 계수
    1. MPDB (Material Properties Database) 확장

    MPDB (Material Properties Database)는 FLOW-3D 와 별도로 Flow Science, Inc 에서 구입할 수 있는 타사 데이터베이스입니다. 여기에는 문헌의 다양한 온도 의존성 고체 재료 특성이 포함되어 있습니다. FLOW-3D 용 MPDB는 사용자가 FLOW-3D의 기본 데이터베이스와 호환되는 파일 형식을 내보낼 수 있도록 하여 데이터를 FLOW-3D 로 편리하게 가져올 수 있는 MPDB 독점 버전입니다. MPDB의 재료 특성은 대부분 고체상입니다. 따라서 FLOW-3D 모든 모델 고체 특성을 요구하는 데이터, 특히 유체 구조 상호 작용, 응고 및 열 응력 진화 모델을 활용할 수 있습니다.

    MPDB는 다양한 형식으로 데이터를 내보낼 수 있는 독립형 데이터베이스로 사용될 수 있습니다. MPDB에 대한 일반적인 지침은 JAHM Software, Inc.를 방문하십시오. 여기에서는 FLOW-3D 와 함께 MPDB를 사용하는 방법에 대한 지침을 제공합니다. FLOW-3D 와 제대로 통합하려면 MPDB 용 실행 파일이 Windows와 Linux에 있어야 합니다. 실행 파일은 FLOW-3D GUI에 의해 감지되며 재료 메뉴 아래 MPDB에서 재료 가져오기 메뉴 항목 이 활성화됩니다. 이러한 조건 중 하나라도 충족되지 않으면 FLOW-3D GUI를 통해 액세스 할 수 없습니다. MPDB%F3D_HOME%\Utilities$F3D_HOME/UtilitiesMPDB_for_FLOW-3D

    FLOW-3D MPDB
    FLOW-3D MPDB

    material를 클릭 MPDB에서 가져오기 및 사용자 인터페이스 MPDB는 별도의 창에서 열립니다. 재료는 주요 요소로 분류되었습니다. Materials 탭, 테이블에서 요소를 마우스 오른쪽 버튼으로 클릭하여, 사용자는 해당 요소를 포함하는 물질의 목록을 볼 수 있습니다.

    (Material Properties Database)
    (Material Properties Database)

    예를 들어 다음 그림은 철 (Fe)이 포함된 데이터베이스의 재료 목록을 보여줍니다.

    FLOW-3D MPDB(Fe)
    FLOW-3D MPDB(Fe)

    사용자는 다른 합금, 세라믹, 유리 또는 기타 분류되지 않은 재료를 분류하는 다른 탭으로 전환할 수도 있습니다. 다음 그림은 Al & Cu 합금 목록을 보여줍니다.

    FLOW-3D MPDB(Al & Cu)
    FLOW-3D MPDB(Al & Cu)
    FLOW-3D MPDB(Fe,Ni - 1006 (UNS G10060))
    FLOW-3D MPDB(Fe,Ni – 1006 (UNS G10060))

    재료가 식별되면 재료를 두 번 클릭하면 해당 재료에 사용할 수 있는 속성 목록이 있는 별도의 창이 나타납니다. 예를 들어 Fe 및 Ni 합금에서 1006 (UNS G10060)을 엽니다. 이러한 속성이 모두 FLOW-3D에 사용되는 것은 아닙니다.

    FLOW-3D MPDB(1006(UNS G10060))
    FLOW-3D MPDB(1006(UNS G10060))

    각 속성은 이 창의 오른쪽에서 선택할 수 있는 다른 형식으로 파일에 표시, 플로팅 또는 저장할 수 있습니다. 그러나 이러한 속성 중 일부가 FLOW-3D 로 인식되는 것은 아닙니다. 

    FLOW-3D 와 호환되는 파일 형식을 생성하려면 재료 창을 닫고 FLOW-3D/SolidWorks/ANSYS 메뉴에서 시작하십시오. 재료의 특성으로 FLOW-3D로 가져올 수 있는 세 가지 파일 형식이 있습니다.  유체 데이터베이스 형식(.f3d_dbf 확장), 고체 데이터베이스 형식 (.f3d_dbs 확장), 일반 쉼표로 구분된 값(CSV형식)으로 부터 시뮬레이션에 적합한 FLOW-3D 호환 형식을 선택하십시오. MPDB의 재료는 대부분 고체이지만 사용자가 응고된 유체의 특성을 가져오려면 FLOW-3D에서 응고된 유체 특성이 유체 특성의 일부이므로 Fluids 데이터베이스 형식을 선택해야 합니다. 솔리드 및 유체 데이터베이스 파일 형식과 파일은 현재 사용자의 문서 폴더와 Windows 및 Linux에 저장됩니다.

    CSV<My Documents>\FLOW-3D\gui\MaterialsDatabase/home/<user>/FLOW-3D/gui/MaterialsDatabase

    이러한 위치는 FLOW-3D의 데이터베이스가 사용자 정의 재료를 찾는 곳입니다. MPDB에서 이러한 위치로 내보낸 모든 자료는 FLOW-3D의 기본 데이터베이스에 의해 선택됩니다.

    1006 (UNS G10060) 철 합금을 선택하십시오.

    FLOW-3D MPDB(UNS G10060)
    FLOW-3D MPDB(UNS G10060)

    이전에 사용 가능했던 일부 특성은 FLOW-3D 와 관련이 없기 때문에 사용 불가능 합니다. 각 속성이 처리되자 마자 플롯 되거나 해당 데이터가 표시되면 참조 및 메모 섹션이 활성화됩니다. 참조 탭 속성에서 찍은 위치를 나타내는 참고 섹션은 일반적으로 데이터의 구성과 정확성에 관한 사항이 포함되어 있습니다. 

    온도에 따른 특성의 동작을 이해하는 데 도움이 되도록 각 특성을 플롯 할 수 있습니다. 또한 데이터의 유효성에 대한 경고가 있을 수 있습니다. 

    예를 들어 열전도도를 먼저 플로팅하면 저온 경고가 표시됩니다. 온도의 함수로 플롯을 표시하기 전에 .f3d_dbs파일을 쓰려면 데이터베이스에 추가 버튼을 클릭하고 다음 창에서 파일에 쓸 속성을 ​​선택하십시오. 사용 가능한 단계에 대한 속성을 선택할 수 있습니다. 속성이 선택되면 데이터 쓰기 및 닫기를 클릭하십시오. 

    재료 창을 닫습니다. FLOW-3D/SolidWorks/ANSYS 메뉴에서 데이터베이스를 닫습니다.

    FLOW-3D MPDB(Low temperature warning)
    FLOW-3D MPDB(Low temperature warning)
    FLOW-3D MPDB(Temperature Plot)
    FLOW-3D MPDB(Temperature Plot)

    .f3d_dbs파일을 쓰려면 데이터베이스에 추가 버튼을 클릭하고 다음 창에서 파일에 쓸 속성을 ​​선택하십시오. 사용 가능한 단계에 대한 속성을 선택할 수 있습니다. 속성이 선택되면 데이터 쓰기 및 닫기를 클릭하십시오. 재료 창을 닫습니다. FLOW-3D/SolidWorks/ANSYS 메뉴에서 데이터베이스를 닫습니다.

    경우에 따라 재료에 사용자에게 필요한 속성이 없습니다. 데이터베이스에 사용 가능한 속성을 추가한 후 이러한 상황에서 누락된 속성은 유사한 속성을 가진 합금 (사용자의 위험 부담)에서 얻을 수 있습니다. 데이터베이스가 열려있는 동안 FLOW-3D에서 사용될 하나의 재료에 대해 속성을 혼합하고 일치시킬 수 있습니다.

    FLOW-3D MPDB(Select properties to write to file)
    FLOW-3D MPDB(Select properties to write to file)

    데이터베이스를 닫은 후 파일 이름을 묻는 메시지가 사용자에게 표시됩니다. 기본값은 MPDB 가 재료에 지정하는 것입니다. FLOW-3D 가 재료를 사용자 정의 재료로 인식하도록 파일의 위치와 확장자가 미리 설정되어 있습니다.

    FLOW-3D MPDB(File locate position)
    FLOW-3D MPDB(File locate position)

    CSV파일을 선택한 경우에도 동일한 프로세스가 적용됩니다. 데이터가 파일에 기록되면 각 테이블 형식 속성 창의 값 가져오기 버튼에서 데이터를 검색할 수 있습니다.

    첫 번째 열은 항상 온도입니다.

    FLOW-3D MPDB(csv file)
    FLOW-3D MPDB(csv file)
    1. grfedit를 사용하여 flsgrf 파일 편집

    명령 줄 유틸리티이므로 runscript와 같은 적절한 환경에서 실행해야 합니다 ( Runscripts 사용 참조 ).


    Runscripts 사용

    실행 스크립트는 작업 문제 디렉토리에서 실행되도록 설계되었습니다. 스크립트는 $F3D_HOME/local디렉토리에 있습니다. 스크립트를 사용하려면 다음 환경 변수를 설정해야합니다.

    • F3D_HOMEFLOW-3D 설치 디렉터리 의 경로를 지정합니다 .
    • F3DTKNUX_LICENSE_FILEFLOW-3D 라이선스 서버 의 위치를 ​​지정 합니다.
    • PATHPATH포함하도록 환경 변수를 수정해야합니다. $F3D_HOME/local그렇지 않으면 실행 스크립트를 찾을 수 없습니다.
    • F3D_VERSION: 사용할 솔버 버전을 지정합니다. 유효한 옵션은 double배정 밀도 버전 및 prehyd사용자 지정 배정 밀도 솔버입니다.

    명령 줄에서 실행하려면 :

    1. 명령 프롬프트 또는 터미널을 엽니 다.
    2. 필요한 환경 변수를 설정하십시오.
      • Windows : FLOW-3D 를 시작하는 데 사용되는 배치 파일에서 환경을 복사하여 수행 할 수 있습니다 . 배치 파일의 내용은 FLOW-3D 아이콘 을 마우스 오른쪽 버튼으로 클릭 하고 편집을 선택 하여 액세스 할 수 있습니다 .
      • Linux : 설치 디렉토리 에서 파일을 flow3dvars.sh가져옵니다 local.
    3. 솔버가 실행중인 디렉토리로 변경하십시오.
    4. 원하는 runscript 명령을 입력하십시오. runhyd <ext2>

    • grfedit를 연 후 사용자에게 소스 파일 (flsgrf.*데이터가 복사될 파일)의 경로를 묻는 메시지가 표시됩니다. 파일의 전체 경로 (예 c:\users\username\FLOW-3D\simulation\flsgrf.simulation:)를 입력하고 <enter>를 누르십시오.
    • 이제, 파일 입력 확장의 목표 예를 들어, (데이터를 기록할 위치로 파일) 파일을 new_output. 데이터가 파일에 기록됩니다 c:\users\username\FLOW-3D\simulation\flsgrf.new_output. 대상 파일이 존재하면 파일을 덮어쓰거나 대상 파일에 데이터를 추가하라는 메시지가 표시됩니다. 대상 파일의 시간보다 늦게 시뮬레이션 시간을 가진 소스 파일 편집 만 추가됩니다.
    • 이 시점에서 프로그램은 어떤 히스토리 데이터 편집, 데이터 편집 재시작 및 대상 파일에 쓰기 위해 선택된 데이터 편집을 묻습니다. 프롬프트에 따라 작성할 데이터 편집을 선택하십시오.
    • 대상 파일을 작성한 후 프로그램이 닫히고 다른 flsgrf.*파일처럼 사용할 수 있습니다.

     노트

    • grfedit는 FLOW-3D v11.1 이상에서 작성된 결과 파일에서만 작동합니다.
    • 소스 flsgrf.*파일은 grfedit에 의해 수정되지 않습니다
    • FLOW-3D/MP의 출력 파일로 작업할 때는 flsgrf1의 위로 flsgrf 교체 하십시오 .
    • 소스 및 대상 파일 모두에 허용되는 유일한 이름은 flsgrf및 flsgrf1입니다.

    FLOW-3D 및TruVOF는 미국 및 기타 국가에서 등록 상표입니다.

    FLOW-3D 용어 사전 테이블

    FLOW-3D Glossary | FLOW-3D 용어 사전

    FLOW-3D 용어 사전 / 용어 설명

    FLOW-3D 용어 사전 테이블
    FLOW-3D 용어 사전 테이블

    FLOW-3D 용어 사전 / 용어 설명

    Drift Flux

    드리프트 모델은 밀도가 서로 다른 두 혼합 유체 구성 요소의 상대적 흐름을 설명합니다. 구성 요소는 상이 다를 수도 있고, 상이 같지만(불가침) 유체가 다를 수도 있습니다. 분산된 위상 입자 크기가 클 경우 드리프트 모델의 적용성에 대한 제한이 존재할 수 있습니다. 이러한 제한은 일반적으로 메쉬 셀 크기의 10% 미만으로 분산된 위상 입자 크기를 유지함으로써 방지할 수 있습니다.

    배플

    얇은 형상 조각을 나타내는데 사용되는 2 차원 개체입니다. 이들은 전처리기에 의해 셀면으로 이동되고 유체의 흐름을 부분적으로 또는 완전히 차단하는 역할을 합니다. 배플은 지정된 열 전달 계수를 가질 수 있으며 배플을 통과하는 양(플럭스 표면)을 측정하는 데 사용할 수 있습니다.

    Two-dimensional objects that are used to represent thin pieces of geometry. They are moved by the preprocessor to cell faces and act to partially, or completely block the flow of fluid. Baffles can have heat transfer coefficients specified and can be used to measure quantities that pass through them (a flux surface).

    경계 조건

    도메인의 범위에서 솔루션을 정의합니다. 경계 위치에서 흐름의 실제 상태를 나타내는 경계 조건을 선택하는 것이 중요합니다.

    Defines the solution at the extents of the domain. It is important to choose boundary conditions that represent the true condition of the flow at the boundary location.

    CFD

    CFD (Computational Fluid Dynamics)는 수치 솔루션을 통해 컴퓨터의 유체 흐름을 시뮬레이션 하는 유체 역학의 한 분야입니다.

    Computational Fluid Dynamics (CFD), the branch of fluid mechanics dedicated to simulating the flow of fluid on a computer via numerical solutions.

    Complements

    Complements를 정의합니다. 예를 들어, 솔리드 구의 complements는 솔리드 재료로 둘러싸인 구형 구멍입니다.

    The inverse of a shape defines the complement. For example, the complement of a solid sphere is a spherical hole surrounded by solid material.

    Client

    클라이언트 컴퓨터는 자신이 FLOW-3D를 실행하고 있지만, FLOW-3D 소프트웨어 라이선스는 다른 컴퓨터 (서버 컴퓨터)에서 획득하는 컴퓨터를 의미합니다.

    A client machine is a computer that runs FLOW-3D  but acquires the software license from a different machine (the server machine)

    Components

    Components는 공간의 개체를 정의하며 하위 구성 요소로 구성됩니다. 구성 요소는 열 전도율, 비열 및 표면 거칠기와 같은 재료 특성을 가질 수 있습니다.

    Components define objects in space and are comprised of subcomponents. A component can have material properties such as thermal conductivity, specific heat and surface roughness.

    Custom result

    시뮬레이션 중 또는 완료 후 사용자가 생성한 데이터를 그래픽으로 표시합니다. 생성하려면 사용자가 flsgrf결과 파일을 연 다음 플로팅 매개 변수(예 : 플로팅 할 도메인 부분, 플로팅 할 수량 등)를 선택해야 합니다.

    Graphical displays of data generated by the user during the simulation or after it has completed. To generate, the user must open an flsgrf results file and then select the plotting parameter (e.g., portion of domain to plot, quantity to plot, etc.).

    Domain

    지배 방정식을 풀 영역입니다. 이것은 메쉬의 범위에 의해 정의됩니다.

    The region in which the governing equations are to be solved. This is defined by the extents of the mesh.

    Diagnostics

    전 처리기 및 솔버의 진행 상황과 오류 및 경고에 대한 정보가 포함된 파일 세트입니다.

    A suite of files that contain information on the progress of the preprocessor and solver as well as errors and warnings.

    EPSI

    압력/연속 반복이 어느 지점에서 수렴되는지를 결정하는데 사용된 수렴 기준입니다. 기본 숫자 설정을 사용하면 이 값은 FLOW-3D에 의해 자동으로 계산되며 시간 단계가 증가함에 따라 작아집니다.

    The convergence criterion that was used to determine at what point the pressure/continuity iterations have converged. With the default numerical settings, this value is automatically computed by FLOW-3D  and becomes smaller as the time step increases.

    Existing result

    prpplt.* 또는 flsplt.* 파일은 전처리 종료 솔버 실행 종료시 또는 자동으로 생성되는 플롯 파일입니다.

    A plot file that is automatically created, either at the end of preprocessing or the end of the solver run- prpplt.* or flsplt.*.

    F3D_HOME

    FLOW-3D 프로그램 파일이 있는 디렉토리를 정의하는 환경 변수.

    Environment variable that defines the directory where the FLOW-3D  program files are located.

    Floating license

    FLOW-3D는 서버 시스템에 라이센스를 액세스하는 각 클라이언트 컴퓨터와 컴퓨터 네트워크에서 실행합니다. 허용하는 라이센스 최대 동시 시뮬레이션 수는 구매한 솔버 토큰 수에 의해 제한됩니다.

    A license that allows FLOW-3D  to be run on a network of computers with each client machine accessing the license on a server machine. The maximum number of concurrent simulations is limited by the number of solver tokens purchased.

    Flsgrf file

    솔버가 생성한 결과 파일. 이 파일은 사전에 정의된 시간 간격으로 생성된 정보를 포함하며 그래픽 디스플레이를 생성하는 데 사용됩니다. 사용자 정의 플로팅 중에 포스트 프로세서에서 사용합니다.

    Results file produced by the solver. This file contains information produced at predefined time intervals and is used to produce graphical displays. Used by the postprocessor during custom plotting.

    Flsplt file

    솔버가 자동으로 생성한 플롯 파일입니다. 이 파일에는 시뮬레이션의 히스토리 데이터, 메시 등에 대한 기본 정보와의 $GRAFIC 이름 목록에 사전 정의된 그래픽 요청이 포함되어 prepin.* 파일 안에 있습니다.

    Plot file produced automatically by the solver. This file contains basic information on history data, mesh, etc. from the simulation as well as any pre-defined graphics requests in the $GRAFIC namelist in prepin.*.

    Fluid #1 surface area

    선택한 길이 단위의 자유 표면 영역을 제곱 됩니다. 인터페이스가 예리한 문제에만 해당됩니다.

    The free-surface area in the chosen length units squared. This is only relevant for problems with a sharp interface.

    Fluid thermal energy

    영역에 존재하는 모든 유체에 포함된 총 열 에너지 (에너지 전송이 켜져 있는 시뮬레이션에만 해당).

    The total thermal energy contained by all the fluid present in the domain (relevant only for simulations with energy transport turned on).

    Free surface

    유체와 유체 사이의 인터페이스. FLOW-3D에서 이 인터페이스는 전단이 없는 것으로 가정되며, 이는 빈 공간에 있는 가스가 유체에 무시할 수 있는 트랙션을 발휘함을 의미한다.

    The interface between fluid and void. In FLOW-3D , this interface is assumed to be shear-free, meaning that any gas in the void space exerted negligible traction on the fluid.

    GUI

    ” Graphical User Interface”.  GUI는 사용자가 FLOW-3D를 제어할 수 있는 그래픽 패널, 대화 상자 및 창을 제공합니다.

    “Graphical User Interface”. The GUI presents the graphical panels, dialog boxes and windows that allow the user to control FLOW-3D .

    Iteration count

    각 시간 단계에서 필요한 압력/연속 반복 횟수입니다. 압력/연속성 반복은 유체 볼륨이 유지되도록 하고 유체 전체에서 올바른 압력을 계산하는 데 필요합니다.

    The number of pressure/continuity iterations required at each time step. The pressure/continuity iterations are necessary to ensure that the fluid volume is maintained and to compute the correct pressure throughout the fluid.

    License file

    사용자가 FLOW-3D 를 실행할 수 있도록 암호화된 정보가 포함된 Flow Science에서 제공하는 전자 파일 입니다.

    Electronic file provided by Flow Science that contains encrypted information enabling the user to run FLOW-3D .

    License server

    플로팅 라이센스 시스템의 작동을 활성화하기 위해 FLEXlm 라이센스 소프트웨어가 설치된 시스템. FLOW-3D는 License Server에 설치할 필요가 없습니다.

    Computer on which the FLEXlm licensing software is installed to enable the operation of a floating license system. FLOW-3D  does not need to be installed on the license server.

    Licensing

    FLOW-3D 실행을 제어하는 ​​FLEXlm 소프트웨어.

    FLEXlm software that controls the running of FLOW-3D .

    Max. residual

    압력/연속성 반복의 최종 반복에서 연속성 방정식의 실제 발산. 이 값은 메시지가 나타나지 않는 한 일반적으로 epsi보다 작습니다 .

    The actual divergence of the continuity equation on the final iteration of the pressure/continuity iterations. This value is usually smaller than epsi unless the message, pressure iteration did not converge in xxxx iterations appears.

    Mean kinetic energy

    모든 계산 셀의 운동 에너지의 합을 도메인에 존재하는 총 유체 질량으로 나눈 값입니다. 이 양이 시간이 지남에 따라 변하지 않으면 정상 상태에 도달했음을 나타내는 좋은 지표입니다.

    The sum of kinetic energy of all the computational cells, divided by the total mass of fluid present in the domain. When this quantity ceases to change over time, it is a good indicator that steady-state has been reached.

    Node-locked license

    특정 컴퓨터에 고정된 라이센스. 노드 잠금 라이센스는 네트워크를 통해 액세스 할 수 없으므로 일반적으로 모든 작업을 한 컴퓨터에서 수행해야하는 경우에만 사용됩니다.

    A license that is locked to a particular computer. A node-locked license cannot be accessed across a network, and so is typically only used when all work is to be done on one computer.

    Non-inertial reference frame

    가속화되는 참조 프레임. 비 관성 참조 프레임은 움직이는 컨테이너를 모방하는 데 사용할 수 있습니다.

    A frame of reference that is accelerating. A non-inertial reference frame can be used to mimic a moving container.

    Pltfsi

    1D 및 2D 플롯을 생성하는 FLOW-3D에 포함된 그래픽 디스플레이 프로그램.

    Graphics display program included with FLOW-3D  that produces 1D and 2D plots.

    Postprocessor

    FLOW-3D 내의 Postprocessor 프로그램은 FLOW-3D 또는 타사 시각화 프로그램에서 읽을 수 있는 데이터 파일을 생성하거나 타사 소프트웨어 프로그램에서 읽을 텍스트 데이터를 생성하는 솔버 출력 데이터를 처리하는 프로그램입니다.

    The program within FLOW-3D  that processes the solver output data to produce data files that can be read by FLOW-3D ’s or third-party’s visualization programs, or produce text data to be read by third party software programs.

    Prepin file

    FLOW-3D 시뮬레이션을 실행하는데 필요한 모든 정보가 포함된 텍스트 파일 입니다. GUI를 사용하거나 텍스트 편집기를 사용하여 수동으로 작성할 수 있습니다.

    Text file that contains all of the information necessary to create a FLOW-3D  simulation. It can be created using the GUI or manually with a text editor.

    Preprocessor

    솔버의 실행을 준비하기 위해 입력 파일을 기반으로 메쉬 및 초기 조건을 생성하는 FLOW-3D 내의 프로그램 입니다.

    The program within FLOW-3D  that generates the mesh and initial conditions based on the input file in preparation for the running of the solver.

    Prpgrf file

    전처리기에 의해 생성된 결과 파일로 전 처리기의 정보를 포함하며 후 처리기에서 사용자 플롯을 생성하는 데 사용할 수 있습니다. 이 파일은 미리보기 버튼을 선택하거나 시뮬레이션에서 사전 프로세서(runpre 사용)를 실행하는 경우에만 실행됩니다.

    Results file produced by the preprocessor. Contains information from the preprocessor and can be used by the postprocessor to create custom plots. This file is produced only when the Preview button is selected or if only the pre-processor is run on the simulation (using runpre).

    Prpplt file

    전처리기에 의해 자동으로 생성된 파일을 플롯 합니다. 메시, 구성 요소, 초기 조건 및 재료 특성에 대한 정보가 포함되어 있습니다.

    Plot file produced automatically by the preprocessor. Contains information on meshing, components, initial conditions and material properties.

    Restart simulation

    이전 시뮬레이션에서 계속되는 시뮬레이션입니다. 이전 시뮬레이션의 결과는 다시 시작 시뮬레이션을 위한 초기 조건 및 (선택적으로) 경계 조건을 생성하는 데 사용됩니다.

    A simulation which continues from a previous simulation. The results from the previous simulation are used to generate the initial conditions and (optionally) boundary conditions for the restart simulation.

    Server

    라이센스 서버를 호스팅하는 시스템

    The machine that hosts the license server.

    Stability limit

    각 시간 단계에서 사용할 수 있는 최대 시간 단계. 더 큰 시간 단계는 수치적 불안정성과 비물리적 결과로 이어질 것이다.

    The maximum time step that can be used during each time step. A larger time step will lead to numerical instabilities and nonphysical results.

    STL (Stereolithography) File

    .STL 파일 형식은 일련의 삼각형이 있는 솔리드 모델의 표면에 근접한 표준 데이터 전송 형식이다. 삼각형은 가장자리에서 결합해야 하며 일관된 방향을 가리키는 정규식이 있어야 한다.

    The .STL file format is a standard data transmission format that approximates the surfaces of a solid model with a series of triangles. The triangles must join at the edges and must have normals that point in a consistent direction.

    Solid fraction

    응고된 영역의 유체 분율 (응고 모델이 켜져 있는 시뮬레이션에만 해당).

    The fraction of fluid in the domain that has become solidified (relevant only for simulations where the solidification model has been turned on).

    Solver

    입력 파일에 정의된 흐름 문제를 시뮬레이션하는 방정식을 계산하는 FLOW-3D 내의 솔버 프로그램 입니다.

    The program within FLOW-3D  that solves the system of equations that simulate the flow problem defined in the input file.

    STL Viewer

    스테레오리소그래피(STL) 파일을 표시하는 특수 유틸리티입니다. STL 파일은 CAD 소프트웨어로 제작되며 3 차원 객체의 표면을 형성하는 많은 삼각형으로 구성됩니다. 의 STL 뷰어 FLOW-3D는 메인 메뉴에서 유틸리티/STL 뷰어를 클릭하여 GUI를 통해 액세스 할 수 있습니다. 그러면 뷰어가 별도의 창에서 열립니다. 메쉬 및 형상 탭에서 STL 파일을 열고 볼 수도 있습니다.

    A special utility that displays stereolithography (STL) files. STL files are produced by CAD software and are composed of many triangles that form the surface of a three-dimensional object. The STL Viewer in FLOW-3D  is accessible via the GUI by clicking Utilities/STL Viewer in the main menu. This causes the viewer to open in a separate window. STL files can also be opened and viewed in the Meshing and Geometry tab.

    Subcomponents

    하위 구성 요소는 구성 요소라고하는 더 큰 모양을 형성하기 위해 결합할 수 있는 기하학적 모양입니다. 하위 구성 요소는 재료를 추가하거나 (고체로) 다른 하위 구성 요소에서 재료를 제거하거나 (구멍으로) 또는 모양 외부에 재료를 추가하도록 정의할 수 있습니다.

    Subcomponents are geometric shapes that can be combined to form larger shapes, called components. A subcomponent can be defined to add material (as solids), remove material from other subcomponents (as holes), or add material outside of the shape (as a complement).

    Time-step size

    계산에 사용된 실제 시간 단계. 이 값은 안정성 한계와 같거나 작을 수 있습니다.

    The actual time step used in the computation. This value can be equal to or less than the stability limit.

    Units

    Units are based upon the values set for the physical properties. Items such as mesh block extents and cell lengths automatically conform to the units used for setting these physical properties.

    단위는 물리적 특성에 설정된 값을 기반으로 합니다. 메쉬 블록 범위 및 셀 길이와 같은 항목은 이러한 물리적 속성을 설정하는 데 사용되는 단위를 자동으로 따릅니다.

    Volume error (%)

    주어진 시간에 도메인에 존재하는 총 유체의 백분율로 설명되지 않은 유체 부피의 백분율을 의미합니다. 따라서 단순히 총 부피가 작기 때문에 유체가 시스템 밖으로 배출되는 시뮬레이션에서 큰 비율의 부피 오류가 발생할 수 있습니다.

    The percentage of fluid volume not accounted for as a percentage of the total fluid present in the domain at a given time. Therefore, a large percentage volume error can occur for simulations where fluid is draining out of the system simply because the total volume present is small.

    Volume of fluid #1

    선택한 길이 단위로 입방체에 존재하는 유체 #1의 총 부피입니다. 2 유체 문제의 경우, 유체 #2의 부피는 항상 도메인 부피에서 유체 #1의 부피를 뺀 값입니다.

    The total volume of fluid #1 present in the system, in the chosen length units cubed. For two-fluid problems, the volume of fluid #2 is always the domain volume minus the volume of fluid #1.

    Wall shear stress

    FLOW-3D 옵션은 벽면 및 객체 인터페이스에서 전단 응력 계산을 켜거나 끌 수 있도록 해줍니다. “no-slip” 인터페이스의 효과를 모델링 하려면 벽면 전단 응력을 켜야 합니다.

    The FLOW-3D  option that allows the user to turn on or off the computation of shear stress at wall and object interfaces. Wall shear stress must be turned on to model the effect of “no-slip” interfaces.

    Workspace

    작업 공간은 시뮬레이션 프로젝트를 위한 파일 컨테이너입니다. 작업 공간은 사용자가 FLOW-3D 뿐만 아니라 하드 드라이브에서도 작업을 구성하는 데 도움이 됩니다.

    A workspace is a file container for simulation projects. Workspaces help the user organize their work, not only within FLOW-3D , but also on their hard drive.

    FLOW-3D 및TruVOF는 미국 및 기타 국가에서 등록 상표입니다.

    FLOW-3D 기술자료로 이동

    education_banner

    FLOW-3D 온라인 교육

    FLOW-3D Training Modules

    FLOW-3D GUI PART 1 OF THE FLOW-3D V12.0 TRAINING SERIES

    FLOW-3D GUI

    • Introduction to FLOW-3D graphical user interface
    • Simulation Manager Tab
    • Portfolio
    • Running Simulations and the Queue
    • Runtime Diagnostics: Text Output
    • Runtime Diagnostics: Plots
    • Runtime Controls
    • FLOW-3D File Structure
      Review the important files that are created when running simulations in FLOW-3D. Access the simulation files through a link on the Simulation Manager Tab. Identify the important setup and solver outputs files

    모델 설정 탭

    • Introduction to the Model Setup TabIntroduction to the Model Setup Tab including an orientation to its layout and how to access model inputs though the dock widgets on the process toolbar. Options for customizing the layout of the process toolbar are also reviewed.
    • Navigating the 3D ViewportLearn the basic controls for navigating the 3D viewport. This includes mouse controls, toolbar shortcuts, saving views, and moving the pivot point.
    • Other Menu/Toolbar Navigation Options
    • Working with Dock Widget Inputs
    • Model DependenciesRecognize and understand dock widget input dependencies.
    Model Setup Tab PART 2 OF THE FLOW-3D V12.0 TRAINING SERIES
    Global Settings PART 3 OF THE FLOW-3D V12.0 TRAINING SERIES

    전역 설정

    • Global Dock Widget Overview
    • Pressure Type
    • Finish Time
    • Finish Options: Additional Finish Condition
    • Finish Options: Active Simulation ControlDefine a logical condition to stop the simulation using active simulation control.
    • Restart OptionsHow to manually define the Restart options to continue running a previously completed simulation.
    • Version OptionsDefine the Version options to specify the solver version and the number of processors used when starting a new simulation run.

    물리 모델

    • Physics Dock Widget OverviewDescription of the available options in the Physics dock widget
    • Interface Tracking, Number of Fluids and Flow ModeBackground information on interface tracking methods and defining the number of fluids. Description of the Volume of Fluid (VOF) method for simulation of complex free surfaces, and how this affects the selection of the number of fluids. Examples are presented for one fluid and two fluid simulations.
    • Activating Physics ModelsDemonstration for how to activate physics models and how to limit the display of inactive physics models using the physics model filter.
    Physics Models PART 4 OF THE FLOW-3D V12.0 TRAINING SERIES
    Fluid Properties PART 5 OF THE FLOW-3D V12.0 TRAINING SERIES

    유체 속성

    • Fluids Dock Widget OverviewIntroduction to the Fluids dock widget and how to define properties for fluids in the simulation.
    • Defining Fluid Properties ManuallyExample for how to manually define fluid properties.
    • Defining Fluid Properties from the Materials DatabaseExample for how to load fluid properties from the fluids database.
    • Managing the Materials Database
      How to add and edit entries in the materials database.

    지오메트리

    • Introduction
    • Component and Subcomponent Overview
    • Creating Subcomponents: Overview
    • Creating Subcomponents: STL
    • Creating Subcomponents: Primitives Manually
    • Creating Subcomponents: Primitives Interactively
    • Creating Subcomponents: Raster
    • Subcomponent Types
    • Subcomponent Order
    • Component Order
    • Component and Subcomponent Properties
    • Transformations
    Geometry PART 6 OF THE FLOW-3D V12.0 TRAINING SERIES
    Meshing PART 7 OF THE FLOW-3D V12.0 TRAINING SERIES

    Meshing

    • Meshing Introduction
    • Coordinate Systems
    • FAVOR™
    • Meshing Basics: Meshing Overview
    • Meshing Basics: Creating Mesh Blocks
    • Meshing Basics: Domain Extents
    • Meshing Basics: Global Controls
    • Meshing Basics: Local Controls
    • Reviewing Mesh Quality: FAVORize
    • Reviewing Mesh Quality: Preprocessing
    • Multi-block Meshing
    • Conforming Mesh Blocks
    • Meshing Best Practices

    Boundary Conditions

    • Introduction
      Introductory comments regarding how boundary conditions are applied and other considerations when defining BCs.
    • Boundaries Dock Widget Overview
    • Velocity
    • Volume Flow Rate
    • Wall
    • Symmetry
    • Grid Overlay
    • Pressure
    • Continuative
    • Outflow
      Description and example setup of the Outflow BC type.
    Boundary Conditions PART 8 OF THE FLOW-3D V12.0 TRAINING SERIES
    Initial Conditions PART 9 OF THE FLOW-3D V12.0 TRAINING SERIES

    Initial Conditions

    • Introduction
      Discussion of how the initial conditions and can affect simulation results and run times.
    • Options for Defining ICs
      Example: Global Settings
      Example: Fluid Regions
    • Example: Function Coefficients
      Description and example for defining spatially varying fluid properties with user defined functions.
    • Example: Pointers
      Description and example for defining an initial condition by filling contiguous cells with the Pointer object.

    Output Options

    • Output Dock Widget Overview
    • Spatial Data
    • Spatial Data: Restart Data
    • Spatial Data: Selected Data
    • History Data
    • Diagnostics: Short Print Data
    • Diagnostics: Long Print Data
    • Example Setup
    • Batch Post-processing
    • Batch Mode: Context File
    • Batch Mode: Manual
    • Batch Mode: Generate Reports
    Output Options PART 10 OF THE FLOW-3D V12.0 TRAINING SERIES
    Baffles PART 11 OF THE FLOW-3D V12.0 TRAINING SERIES

    Baffles

    Introduction
    An introduction to the available options for creating and defining baffle objects.
    Creating Baffle Objects
    Limitations
    Force Outputs
    Porosity
    Scalar Reset Options
    Summary
    A summary of the important options for creating baffles and defining properties.

    Measurement Devices

    • History Probes 
      History probes are point measurement devices and are used to record solver output at a specific location. Examples are provided for how to create these objects interactively and by defining a coordinate value.
    • Flux Surfaces 
      Flux surfaces are a special type of baffle object with a fixed porosity of 1, and are used to calculate flux quantities. Examples are provided for how to create flux surfaces and the types of data available from their output.
    • Sampling volumes 
      Sampling volumes are are three-dimensional data collection regions. Examples are provided for how to create sampling volumes and the types of data available from their output.
    Measurement Devices PART 12 OF THE FLOW-3D V12.0 TRAINING SERIES
    W&E Exercise: Ogee Weir

    W&E Exercise: Ogee Weir

    • This exercise demonstrates the steps to setup a basic free surface or open channel flow simulation in FLOW-3D. It is intended to be a simple and fast running simulation that demonstrates the key setup steps that can be applied to a wide range of other common open channel flow applications. In this exercise, we will simulate flow over an ogee weir to predict the discharge capacity. Simulation results can be validated against discharge rating curves obtained from physical model measurements (USBR, 1996).  Special attention is given to the common types of boundary conditions used in open channel flow simulations and how to select them during the model setup. We also provide examples for common post-processing tasks using both FLOW-3D and FlowSight.
    FLOW-3D POST Optimal presentation

    FLOW-3D POST

    FLOW-3D POST 2025R1 의 새로운 기능

    이 강력한 신제품은 FLOW-3D POST의 기능을 FLOW-3D 제품군 전반으로 확장합니다.

    경로 추적 개선 사항
    사용자는 이제 즉석에서 경로 추적 재료 속성을 추가, 편집 및 조정할 수 있으므로 복잡한 기술적 결과를 더 많은 청중에게 사실적인 렌더링으로 더욱 쉽게 전달할 수 있습니다.

    Simulation of a factory chute
    Simulation of a factory chute

    History data 계산기

    이제 FLOW-3D POST 내에서 History data에 대한 수학적 연산이 직접 가능합니다. 사용자는 업데이트된 파이썬 계산기를 이력 데이터에 적용하여 프로브 및 플럭스 표면과 같은 측정 장치의 시계열 데이터에 대한 연산을 간소화할 수 있습니다.

    EXODUS 파일 포맷 성능 향상

    이 릴리스는 EXODUS 객체의 렌더링을 더 부드럽고 현실적으로 개선합니다.

    • JSON / EXODUS 파일 다시 로드: 이제 사용자는 시뮬레이션이 실행되는 동안 JSON / EXODUS 파일을 다시 로드하여 시뮬레이션 워크플로우를 중단하지 않고도 진화하는 데이터를 시각화하고 분석할 수 있습니다.
    • 수정된 다공성 시각화: EXODUS 파일 형식의 다공성 출력은 수축 다공성을 해소하고 주조물 내부의 누출 경로를 더 잘 시각화할 수 있도록 수정되었습니다.
    River bank simulation before and after going through FLOW-3D POST
    EXODUS 출력의 측면 표면은 2025R1 이후 FLOW-3D POST에서 평활화할 수 있습니다.

    FLOW-3D WELD 및 FLOW-3D AM 지원

    유체, 용융 영역, 열원, 반사 및 입자를 위한 새로운 사전 구성 객체는 FLOW-3D WELD 및 FLOW-3D AM 시뮬레이션의 시각화를 용이하게 합니다. 일반적으로 사용되는 출력의 주석은 FLOW-3D POST에서 결과 파일을 열면 자동으로 제공되므로 후처리 워크플로우가 가속화됩니다.

    Simulation image with annotations from FLOW-3D POST

    일반적으로 사용되는 출력을 쉽게 볼 수 있어 사용자가 데이터 해석과 분석에 집중할 수 있습니다.

    FLOW-3D POST 2024R1 의 새로운 기능

    FLOW-3D POST 2024R1은 EXODUS II 기반의 결과를 확장하여 유체-구조 상호작용열 응력을 시각화할 수 있는 기능을 제공합니다.

    또한, 사용자는 이제 삼각형 격자 래스터 및 LandXML 파일을 시각화할 수 있어 모델링 영역을 둘러싼 지형을 더 쉽게 확인할 수 있습니다. 이를 통해 시뮬레이션에 대한 더 나은 컨텍스트를 제공하고, 결과에 집중할 수 있도록 돕습니다.

    Land XML support
    모델링 영역 내 지형(왼쪽)과 삼각형 지형(오른쪽)의 비교. 모델링 영역에는 산과 하류 계곡이 포함되지 않은 반면, 삼각형 지형은 이를 포함하여 더 우수한 컨텍스트와 명확성을 제공합니다.

    마지막으로, 주조 사용자들은 기포 발생에 영향을 받는 지역과 냉각이 필요한 영역을 식별하는 데 도움이 되는 새로운 출력을 보게 될 것입니다.

    FLOW-3D POST 2023R2 의 새로운 기능

    새로운 결과 파일 형식

    FLOW-3D POST 2023R2는 EXODUS II 형식을 기반으로 하는 완전히 새로운 결과 파일 형식을 도입하여 더 빠른 후처리를 가능하게 합니다. 이 새로운 파일 형식은 크고 복잡한 시뮬레이션의 후처리 작업에 소요되는 시간을 크게 줄이는 동시에(평균 최대 5배!) 다른 시각화 도구와의 연결성을 향상시킵니다.

    FLOW-3D POST 2023R2 에서 사용자는 이제 선택한 데이터를 flsgrf , EXODUS II 또는 flsgrf 및 EXODUS II 파일 형식 으로 쓸 수 있습니다 . 새로운 EXODUS II 파일 형식은 각 객체에 대해 유한 요소 메쉬를 활용하므로 사용자는 다른 호환 가능한 포스트 프로세서 및 FEA 코드를 사용 하여 FLOW-3D 결과를 열 수도 있습니다. 새로운 워크플로우를 통해 사용자는 크고 복잡한 사례를 신속하게 시각화하고 임의 슬라이싱, 볼륨 렌더링 및 통계를 사용하여 보조 정보를 추출할 수 있습니다. 

    새로운 결과 파일 형식은 hydr3d 솔버의 성능을 저하시키지 않으면서 flsgrf 에 비해 시각화 작업 흐름에서 놀라운 속도 향상을 자랑합니다.

    레이 트레이싱을 이용한 화장품 크림 충전

    혼입 공기 시뮬레이션

    FLOW-3D POST의 표면 LIC

    레이 트레이싱을 이용한 화장품 크림 충전

    혼입 공기 시뮬레이션

    이 흥미로운 새로운 개발은 결과 분석의 속도와 유연성이 향상되어 원활한 시뮬레이션 경험을 제공합니다. 

    또한 FLOW-3D POST 2023R2 는 최신 버전의 ParaView로 업그레이드되었으며 ParaView 5.11.1 과 관련된 개선 사항을 제공합니다 .

    새로운 시각화 기능

    임의의 클립 및 슬라이스를 매끄럽게 만듭니다.

    EXODUS II 파일 형식을 사용하면 사용자는 모든 방향에서 부드러운 슬라이스를 생성할 수 있으므로 보고 싶은 대로 정확히 흐름을 시각화하는 것이 더 쉬워집니다.

    아크형 웨어 시뮬레이션
    호형 위어 위의 흐름 방향에 맞춰 정렬된 슬라이스입니다. Surface LIC 표현에서 매끄러운 표면과 유선형을 확인하세요.

    모델 출력의 더 나은 정량화

    EXODUS II 파일은 체적 개체이므로 흐름의 특성을 더 쉽게 정량화할 수 있습니다. 예를 들어, 아래 표시된 주조 응고 시뮬레이션에서 오른쪽 패널은 히스토그램을 사용하여 주조의 다공성 분포를 설명할 수 있는 방법을 보여줍니다. 마찬가지로 접촉 탱크의 예는 시간이 지남에 따라 소독제 및 병원체 농도 분포가 어떻게 변화하는지 보여주므로 설계 요구 사항이 충족되었는지 여부를 보여주는 데 도움이 됩니다. 

    주조 응고 결과

    접촉식 탱크 시뮬레이션의 진화

    향상된 광선 추적

    광선 추적은 기술적인 청중과 비기술적인 청중 모두에게 결과를 전달하는 데 유용한 도구이며 EXODUS II 파일 형식에서 사용할 수 있는 체적 데이터는 이 시각화 방법과 잘 작동합니다.

    광선 추적을 사용한 병 채우기 시뮬레이션
    FLOW-3D POST 의 뛰어난 광선 추적 기능을 보여주는 병 채우기 시뮬레이션

    Surface LIC로 유동장 표현

    새로운 Surface LIC 시각화 도구는 흐름 선단이 함께 모이는 재순환 및 불감대뿐만 아니라 온도, 오염 물질 등의 일반적인 이동을 강조하여 흐름장을 시각화하는 데 도움이 됩니다.

    FLOW-3D POST의 표면 LIC
    FLOW-3D POST 의 새로운 EXODUS II 파일 형식 및 Surface LIC 표현의 예

    애니메이션 유선형

    애니메이션 유선형은 표준 보기에서 보기 어려울 수 있는 흐름의 내부 구조에 대한 세부 정보를 시각화하는 데 도움이 됩니다.

    FLOW-3D POST 2023R1 의 새로운 기능

    FLOW-3D POST 2023R1은 기본 MP4 지원을 갖춘 업데이트된 ParaView 엔진, 쉬운 설치를 위한 자동 종속성 테스트 기능을 갖춘 간소화된 Linux 설치 프로그램, Windows 11 및 RHEL 8 지원을 특징으로 합니다.

    단위 표시

    단위는 엔지니어링 분석 결과를 해석하고 전달하는 핵심 부분입니다. FLOW-3D POST 2023R1 에서는 단위가 결과 파일에서 자동으로 판독되고 공간 및 히스토리 플롯의 범례에 설정되므로 시뮬레이션 결과를 쉽게 해석하고 전달할 수 있습니다.

    FLOW-3D POST 장치 디스플레이

    자동 PQ 2 플롯

    FLOW-3D CAST는 수년 동안 PQ 2 분석을 통해 HPDC 기계 성능에 대한 정보를 제공해 왔으며 이제 FLOW-3D POST 에서 시각화를 지원하도록 이 기능을 확장했습니다. PQ 2 정보는 사전 정의된 플롯에 자동으로 요약되므로 플롯의 가시성을 전환하여 기계가 주조 작업을 수행하는 방식을 확인하기만 하면 됩니다 . 추가적인 이점은 데이터와 시간을 비교하여 압력이 기계 성능을 초과하는 시기를 확인할 수도 있다는 것입니다.

    자동-pq2-플롯-flow3d-post-2023r1

    입자 시각화

    우리는 상호 작용을 보다 직관적으로 만들고 다른 응용 프로그램에서 사용하기 위해 입자를 STL 파일로 쉽게 내보내거나 FLOW-3D AM 의 경우 분말 용융 시뮬레이션의 초기 조건으로 내보낼 수 있도록 입자를 표시하는 방법을 다시 검토했습니다. FLOW-3D POST 2023R1 에서는 배율 1을 사용하여 입자의 물리적 크기를 신속하게 표시하고 파일 > 데이터 저장 옵션을 사용하여 입자를 STL로 저장할 수 있습니다.

    FLOW-3D POST 2023R1의 입자 시각화

    FLOW-3D POST 2022R1 의 새로운 기능

    FLOW-3D POST 2022R1은 FLOW-3D 의 포스트 프로세서 에 세 가지 중요한 개발을 제공합니다. 즉, 간소화된 2D 슬라이싱, ParaView의 Python 도구를 사용한 고급 자동화, 향상된 포스트 프로세싱 렌더링 속도입니다.

    2D 슬라이싱 기능

    2D 슬라이싱 기능이 확장되고 간소화되어 작업이 더욱 간단해지고 강력해졌습니다. FLOW-3D POST 사용자는 이제 슬라이스 표면의 벡터 표현과 여러 색상 변수를 사용하여 2D 슬라이스를 빠르게 생성할 수 있습니다. 이 2분짜리 비디오는 새로운 2D 슬라이스 기능의 예를 제공합니다.

    파이썬 도구

    2022R1에 ParaView의 Python 도구가 추가되면 FLOW-3D POST 의 자동화 기능이 확장 되어 반복 작업을 자동화하는 매크로는 물론 클릭 한 번으로 전체 결과 세트를 생성하는 일괄 후처리도 포함됩니다. 특정하거나 정교한 유형의 후처리, 시뮬레이션 후 시뮬레이션을 표시하려는 경우 출력을 표준화하고 후처리 작업을 자동화할 수 있는 이러한 새로운 기능을 통해 엄청난 이점을 얻을 수 있습니다.

    일괄 후처리를 사용하면 후처리 작업을 사전 정의하는 스크립트 또는 상태 파일을 사용하여 명령줄에서 후처리할 수 있으므로 DOE, 매개변수 스윕 또는 자동화된 워크플로우로 인한 여러 결과 파일에 대한 이미지 및 애니메이션 생성이 용이해집니다. 배치 스크립트 또는 상태 파일을 다양한 결과 파일이나 시뮬레이션 결과 파일의 전체 작업 공간에 적용하여 각 사례에 대해 원하는 출력을 빠르고 일관되게 생성할 수 있습니다. 또한 단일 결과 파일에 대한 일련의 다양한 시각화 출력을 생성하는 데 활용할 수도 있습니다.

    PvBatch와 매크로를 통합하여 사용자 사이트 에서 후처리 워크플로를 쉽게 자동화하고 가속화하는 방법에 대한 30분짜리 비디오 튜토리얼에 액세스하십시오 .

    성능 향상

    우리는 또한 후처리 속도에 대해 연구해 왔으며 FLOW-3D POST 2022R1은 일반적으로 FLOW-3D POST v1.1 보다 10%-30% 더 빠르지 만 정확한 속도 향상은 시뮬레이션 및 출력 세부 사항에 따라 다릅니다. 오른쪽의 몇 가지 예는 성능 향상을 보여줍니다.

    샘플 시뮬레이션속도를 올리다
    미로 위어1.3배
    벨하우징 주조1.14배
    유체-구조 상호작용1.2배

    Figure 1. Alaska requires minimum water depth for fish passage to be 2.5 times the height of the caudal fin (D) (Hotchkiss and Frei 2007).

    EFECTS OF HYDRAULIC STRUCTURES ON FISH PASSAGE: AN EVALUATION OF 2D VS 3D HYDRAULIC ANALYSIS METHODS

    물고기 통로 유압 구조효과 :2D VS 3D 유압 분석 방법의 평가

    ABSTRACT

    채널 스패닝 유압 구조물은 상류 물고기 이동에 대한 장벽 역할을 할 수 있습니다. 이러한 종단 적 서식지 연결의 중단과 관련된 부정적인 결과는 정확하고 실행 가능한 평가 기술의 필요성을 강조합니다.

    3 차원 평가 방법은 인스트림 구조에서 복잡한 흐름을 해결하고 물고기 움직임을 정확하게 예측하는 것으로 나타났습니다. 그러나 3 차원 모델링은 시간과 리소스 요구 사항으로 인해 비실용적 일 수 있습니다.

    이 연구는 2 차원 전산 유체 역학 모델과 통계 분석을 사용하여 콜로라도 주 리옹에있는 화이트 워터 공원 구조의 수력 조건을 설명하는 것을 조사합니다. 물고기의 움직임 관찰은 잠재적 인 수영 경로를 나타내는 공간적으로 명시적이고 연속적인 경로를 따라 결과 수력 변수와 쌍을 이룹니다.

    로지스틱 회귀 분석은 흐름 깊이와 속도가 어류 통과와 밀접한 관련이 있음을 나타냅니다. 결합 된 깊이 및 속도 변수무지개 송어 (92 %를 정확하게 예측Oncorhynchus mykiss) 및 갈색 송어 (Salmo trutta)는 이 유압 구조에서) 움직임 관찰을 합니다.

    이 연구의 결과는 2 차원 분석 방법이 3 차원 분석이 불가능한 경우 유사한 수력 학적 구조가 어류 통과에 미치는 영향을 평가하는 비용 효율적인 접근 방식을 제공할 수 있음을 시사합니다. 또한,이 연구의 결론은 비교적 낮은 수영 성능을 가진 송어와 물고기 모두에 대한 관리 및 설계 결정을 안내하는 데 사용할 수 있습니다.

    서문

    수력 구조물은 수생 생물의 종 방향 서식지 연결을 의도적으로든 우연히든 효과적으로 차단할 수 있습니다. 의도적 장벽은 일반적으로 침입성 종의 도입 또는 교잡을 방지하기 위해 관리자에 의해 배치됩니다 (Holthe et al. 2005; Fausch et al. 2006). 그러나 구조물을 설계하고 설치할 때 물고기 통행 촉진을 고려하지 않았기 때문에 장벽이 더 자주 생성됩니다. 따라서 인위적 장애로 인해 전 세계 수로가 분열되었습니다 (Williams et al. 2012). 철새 어종의 성공적인 수명주기를 위해서는 종단 서식지 연결이 필수적입니다 (Schlosser and Angermeier 1995). 상류 이동에 대한 지연 또는 종료는 인구에 부정적인 영향을 미치고 생태계 기능을 방해 할 수 있습니다 (Beechie et al. 2010). 

    수로를 가로 지르는 수력 구조물은 어류 통행에 미치는 영향을 철저히 평가하지 않고 하천과 강에 계속 배치됩니다 (Cada 1998; Noonan et al. 2012). 그러나 강 조각화와 관련된 문제에 대한 인식이 높아짐에 따라 설계 프로세스 전반에 걸쳐 물고기 통과 문제가 해결되는 방식에서 패러다임 전환이 일어나고 있습니다 (Katopodis and Williams 2012). 비 연어 종은 경제적 가치가 높은 종을 선호하는 경우가 많지만, 칼륨 종의 상류 이동 요구가 점점 더 중요하게 고려되고 있습니다 (Santos et al. 2012; Silva et al 2012) (Katopodis 2005; Roscoe and Hinch 2010). . 천연 자원 관리자는 제안 된 수력 구조물에 대해 의견을 제시하고 허용하도록 자주 요청받으며 (Kondratieff 2015),이 검토 과정에서 엔지니어와 과학자는 설계에 대한 예상 어류 통과 성능에 대한 모델 기반 증거를 제공하도록 요청받을 수 있습니다. 어류 통행과 관련하여 기존의 수력 구조물을 평가하고 우선 순위를 정하는 여러 방법이 현재 사용 가능하지만 (Kemp et al. 2010), 이전에 이 중요한 지점에서 제안된 구조물의 통행 효율성을 평가할 수있는 정확하고 실행 가능한 승인 및 설치 도구가 필요합니다.  

    이러한 요구를 해결하는 데 초점을 맞춘 이전 작업은 3D 수력 모델링 기술이 상류 어류 이동을 평가할 목적으로 채널 스패닝 구조의 복잡한 유체 역학을 적절하게 해결할 수 있음을 보여주었습니다 (Stephens 2014).

    이러한 새로운 3D 분석 방법은 전체 예측 정확도가 80 % 이상 (Stephens 2014)으로 매우 효과적 일 수 있지만 3D CFD (전산 유체 역학) 모델을 개발하는 데는 시간과 리소스가 많이 사용됩니다.

    추가 데이터 수집, 소프트웨어 라이선스, 모델링 전문 지식 등에 대한 필요성은 많은 하천 관리 결정에 3D 분석을 비실용적으로 만들 수 있습니다. 다양한 2D 모델 플랫폼이 홍수 배출을 추정하고 (Horritt and Bates 2002; Merwade et al. 2008) 인스 트림 평가에 광범위하게 사용 되었기 때문에 실무 엔지니어와 과학자는 대부분의 수력 구조물 프로젝트에서 2D 수력 모델링을 수행 할 가능성이 더 높습니다.

    물고기 서식지 (Clark et al. 2008; Katopodis 2012). 2D 및 3D 유압 모델의 실제 비교가보고되었지만 (Lane et al. 1999; Shen and Diplas 2008; Kolden 2013), 어류 통과에 대한 2D 및 3D 모델 기반 평가의 효능을 조사한 연구는 현재에서 발견되지 않았습니다.

    목표

    천연 자원 관리자와 설계 엔지니어가 Stephens (2014)의 매우 효과적인 3D 방법에 더 쉽게 접근 할 수 있도록하기 위해이 연구는 자유롭게 사용할 수있는 산업 표준 2D CFD 모델 인 River2D (Steffler and Blackburn 2002)를 사용하여 타당성을 조사합니다. 수력 구조가 어류 통로에 미치는 영향을 평가합니다.

    유사한 접근 방식을 기반으로하고 이전의 수력 학 및 어류 이동 데이터 세트 (Fox 2013, Kolden 2013, Stephens 2014)를 사용하여 이 2 개의 연구는 2D 분석 방법을 사용하여 St. Vrain River의 WWP (화이트 워터 파크) 구조를 평가합니다. Lyons, CO.이 연구의 구체적인 목표는 다음과 같습니다. 

    1. WWP 구조에서 복잡한 유압 환경을 설명하는 2D CFD 모델을 개발합니다. 

    2.이 2D CFD 모델의 결과를 사용하여 WWP 구조를 통해 잠재적 인 물고기 이동 경로를 따라 연속적이고 공간적으로 명시적인 수력 학적 설명을 생성합니다. 

    3. 무지개 송어 (대해 사용 가능한 어류 이동 데이터와 가장 밀접하게 관련된 수리적 변수를 결정Oncorhynchus mykiss) 및 갈색 송어 (Salmo trutta)에합니다. 

    4. 이전에 개발 된 3D 접근 방식 (Stephens 2014)의 PIT (Passive Integrated Transponder) 태그 연구의 움직임 데이터를 기반으로 한 예측 평가 능력을이 연구의 2D 접근 방식과 비교합니다. 

    5. 어류 통행의 관점에서 수력 구조물에 대한 비용 효율적인 평가를 통해 천연 자원 관리자 및 설계자를 지원하기위한 권장 사항을 제공합니다.

    배경

    상류 어류 이동에 대한 장벽은 유속 깊이, 유속 또는 유속과 거리의 조합을 포함한 다양한 물리적 조건에 의해 생성 될 수 있습니다 (Coffman 2005; Cahoon et al. 2005). 깊이 장벽은 일반적으로 흐름 깊이가 너무 얕아 통과 시도를 허용하지 않을 때 생성됩니다.

    깊이 장벽은 또한 자리 잡은 구조물의 낙하 높이 및 플런지 풀 깊이가 도약 제약으로 인해 통과를 허용하지 않을 때 존재할 수 있습니다. 유속이 구조물을 통과하려는 물고기의 수영 능력을 초과 할 때 속도 장벽이 생성되어 상류 진행을 방해합니다. 수력 구조물에 의해 생성 된 난류는 물고기의 통과에도 역할을 할 수 있습니다. 조건에 따라 난류는 물고기 수영에 긍정적 인 영향과 부정적인 영향을 모두 미칠 수 있습니다 (Liao 2007; Cotel and Webb 2012; Lacey et al. 2012).  

    수영 성능 지표는 종종 기존의 수력 학적 구조가 물고기 통행의 장벽으로 작용하는지 여부를 평가하는 데 사용됩니다. 이러한 메트릭 중 가장 일반적인 것은 달리기 속도라고도 하는 버스트 수영 속도와 지구력 곡선입니다 (Castro-Santos et al. 2013).

    물고기는 지속, 연장, 파열의 세 가지 수영 모드를 나타냅니다 (Peake et al. 1997). 지속적인 수영은 이론적으로 무한정 유지 될 수 있지만 장시간 및 버스트 수영 속도는 제한된 시간 동안만 유지 될 수 있습니다.

    지구력 곡선은 세 가지 수영 모드 (Videler and Wardle 1991)에 걸쳐 연속적으로 수영 속도와 피로 시간 사이의 역 관계를 설명하여 생성됩니다. 버스트 수영 속도는 속도 장벽을 식별 할 때 유용하며 (Haro et al. 2004) 지구력 곡선은 잠재적 인 완전 장벽을 식별하는 데 도움이됩니다 (Castro-Santos et al. 2013). 현재 물고기 수영 성능과 난류 임계 값 또는 분포 사이의 물리적 관계는 잘 알려져 있지 않습니다 (Liao 2007).

    그러나 총 운동 에너지 (TKE), 총 수력 변형, 레이놀즈 전단 응력 및 와도와 같은 일부 프록시 변수는 난류가 어류에 미치는 영향을 정량화 할 때 유용한 것으로 나타났습니다 (Nestler et al. 2008; Cotel and Webb 2012; Lacey et al. 2012; Silva et al. 2012). 

    장벽은 완전 할 수 있으며, 물고기 통행을 허용하지 않거나 선택적 통행 성공이 생리적 또는 수리적 특성에 따라 결정되는 경우 부분적 일 수 있습니다. 이 연구의 목적을 위해 총 시도 횟수에 대한 성공적인 통과 횟수를 기반으로 한 인구 수준의 통과 효율을 사용하여 유압 구조로 인한 상류 이동 억제 정도를 정량화합니다 (Haro et al. 2004). 다양한 방법 개발되었습니다. 

    장벽이 물고기 통로 (켐프와 O’Hanley 2010)에 영향을 미치는 방법을 정량화하기 위해  한 가지 접근 방식은 통계 모델을 사용하여 통과 효율 추정치를 0 ~ 100 %의 연속 척도로 표현할 수 있습니다. 과거에는 규칙 기반 또는 회귀 기법을 사용하여 암거 (Coffman 2005; Burford et al. 2009), 도로 횡단 (Warren and Pardew 1998) 또는 수로 실험 설정 (Haro et al. 2004)을 다양한 성공으로 평가했습니다.

    통계적 방법은 다양한 척도에서 수리적 변수에 대한 정보를 결합하여 통과에 큰 영향을 미치는 변수를 식별 할 수 있습니다 (Kemp and O’Hanley 2010). 이러한 모델은 현장 기반 어류 이동 관찰을 사용하여 검증 할 수도 있습니다 (Coffman 2005; Burford 2009).

    2014 년에 Stephens는 3D CFD 모델 출력 (Kolden 2013)을 활용하여 수력 구조물에서 물고기 통과를 평가하기위한 연속적이고 공간적으로 명시적인 분석 방법을 만드는 새로운 통계 방법을 개발했습니다. 이 방법은 콜로라도에있는 3 개의 파도 생성, 인공 화이트 워터 파크 (WWP) 구조물에서 수집 한 수력 측정 및 PIT 태그 통과 관찰 (Fox 2013)을 통해 검증되었습니다. 통계 결과에 따르면 Stephens (2014) 방법은 전체 정확도가 80 % 이상인 통과 효율을 예측할 수 있습니다. 

    Stephens는 3D CFD 모델의 결과를 사용했지만 다른 연구에서는 2D CFD 모델을 사용하여 물고기와 관련된 규모의 복잡한 흐름을 설명하는 데 초점을 맞추 었습니다 (Lane et al. 1999; Crowder and Diplas 2000; Shen and Diplas 2008). 2D CFD 모델링의 주요 관심사는 물고기 서식지 및 수영 성능에 중요한 중간 규모 기능과 관련된 복잡성을 포착 할 수 있는지 여부였습니다 (Crowder and Diplas 2000).

    혼합된 결과는 서식지 평가를 위해 모델링되는 도달 범위의 특성에 따라 2D CFD 모델이 수력 조건에 대한 적절한 설명을 제공하거나 제공하지 않을 수 있음을 보여줍니다 (Clark et al. 2008; Shen and Diplas 2008; Kozarek et al. 2010) . 서식지 또는 지형 모델링에 중점을 두는 경우 깊이 평균 2D 모델과 직접 비교할 때 3D 모델 사용이 선호되었습니다 (Lane et al. 1999; Shen and Diplas 2008). 그러나 수력 구조물에서 상류 어류의 움직임을 평가할 때 2D 및 3D 모델의 성능을 비교 평가 한 연구는 거의 없습니다. 

    이 연구에서 CFD 모델의 비교는 2D 소프트웨어 River2D와 3D 소프트웨어 FLOW-3D에 중점을 둡니다 (Flow Science, 2009). 2D 모델과 3D 모델의 가장 큰 차이점은 2D 모델은 각 계산 노드에서 유압 변수의 값을 깊이 평균한다는 것입니다. 이 깊이 평균은 구조물의 물고기 친화성에 큰 영향을 미칠 수있는 중요한 흐름 특징과 경계층 효과를 배제 할 수있는 잠재력을 가지고 있습니다.

    예를 들어, 수심 평균 속도 값은 WWP 구조 하류의 수력 조건이 동일한 도달 범위 내의 자연 풀에있는 것과 유사하다고 잘못 제안 할 수 있습니다. 실제로 두 유동장은 어류 개체군에 다르게 영향을 미칠 수있는 고유 한 특성을 가지고 있습니다 (Kolden 2013). River2D는 또한 정수압과 일정한 수평 속도 분포를 가정하는 반면 FLOW-3D는 이러한 가정을 피할 수 있습니다.

    대부분의 2D CFD 모델링 프로그램 (Toombes and Chanson 2011)에서 요구하는 정수압 가정은 가파른 경사 (> 10 %)와 급변하는 경사 (Steffler and Blackburn 2002)에서 계산 정확도를 제한합니다. 속도 분포가 일정하다는 가정은 수직 속도 구성 요소가 무시할 수 있음을 의미하며 본질적으로 2D CFD 모델을 사용하여 2 차 흐름 및 강한 순환을 분석하는 기능을 제거합니다 (Steffler and Blackburn 2002; Toombes and Chanson 2011).

    이러한 가정과 2D 물리적 표현의 단순화 된 특성을 고려할 때 2D CFD 모델이 물고기 통과 예측 평가를 위해 채널 스패닝 구조의 복잡한 유체 역학을 적절하게 해결할 수 있는지 여부는 불분명합니다.

    Figure 1. Alaska requires minimum water depth for fish passage to be 2.5 times the height of the caudal fin (D) (Hotchkiss and Frei 2007).
    Figure 1. Alaska requires minimum water depth for fish passage to be 2.5 times the height of the caudal fin (D) (Hotchkiss and Frei 2007).
    Figure 2. Depth (m) and velocity magnitude (m/s) River2D contours for 0.42 cms.
    Figure 2. Depth (m) and velocity magnitude (m/s) River2D contours for 0.42 cms.

    구체적인 내용은 아래 원문을 참고하시기 바랍니다.

    바람이 개방형 철광석 골재 저장소에 미치는 영향 분석 (비산먼지 배출 방지 연구)

    다양한 구성에 대한 비산 먼지 배출

    이 기사는 Dhananjay Sharma, EI, CFM, 유압 모델링 엔지니어, AECOM 에 의해 기고되었습니다  .

    바람이 개방형 골재 저장소에 미치는 영향은 전 세계적으로 환경 문제가 되고 있습니다. 2.7km2 철골 저장소 부지에서 이런 문제가 관찰되었습니다. 이 시설은 철도 운송차량를 통해 광석을 공급받는데, 이 운송차량은 자동 덤프에 의해 비워집니다. 그런 다음 이 광석은 일련의 컨베이어와 이송 지점을 통과하여 저장 장소중 하나로 운송됩니다. 비산먼지 배출은 풍력이 비축된량에 미치는 영향의 결과로 관찰된 결과입니다.

    두 가지 다른 구성(옵션 A와 B)을 FLOW-3D로 모델링하여 비산먼지 배출의 영향을 연구했습니다. 옵션 A에는 4줄에 9개 더미가 있는 36개의 비축량이 있고 옵션 B에는 1줄에 총 16개의 비축량이 있습니다.

    또한 장벽이 있는 공기와 장벽이 없는 공기의 속도를 비교하기 위해 비축물 주변을 따라 30미터 높이의 장벽을 모델링할 수도 있습니다. 10m 높이에서 기준 초속 7.5m(m/s)의 풍속이 두 구성을 모두 모델링하는데 사용되었습니다. 비축 옵션 A와 B에 대해 네 가지 풍향 방향이 분석되었습니다.

    물리적 및 수치 적 모델링

    초기 모델 설정

    FLOW-3D 에서 비산 먼지 배출을 모델링하기  위해, 공기 온도는 15 ° C로 가정되었습니다. 단일의 균일한 비압축성 유체 옵션이 선택되었습니다. z 방향에서 -9.81 m / s의 중력이 사용되었습니다. 유체는 점성과 난류로 간주되었습니다. 2- 방정식 (ke) 모델은 옵션 A 및 B 구성 모두에 대해 표면 마찰없이 난류를 계산하는데 사용되었습니다.

    초기 조건

    1/7 power 법칙 (pproximately a logrithmic law-of-the-wall distribution)에 기반한 속도 프로파일이 각 시뮬레이션에 대한 초기 조건으로 지정되었습니다. 비축 분석에서 가장 관심있는 기준 속도는 12 및 7.5m / s입니다. 풍속을 증가시키고 파일에 인접한 속도에 미치는 영향을 측정하여 분석을 수행했으며, 레이놀즈 스케일링이 이러한 속도에 대해 유지된다는 것을 확인했습니다 (즉, 들어오는 풍속 스케일링과 파일에 인접한 속도 스케일링 간의 선형 관계).   그런 다음 7.5m / s의 속도 만 사용하여 FLOW-3D 시뮬레이션을 구성했습니다. 이러한 시뮬레이션의 결과는 12m / s 조건을 충족하도록 확장 할 수 있습니다.

    풍력 프로파일 power 법칙을 사용하여 10m에서 7.5m / s 이상 및 이하의 다양한 높이에 대한 속도를 추정했습니다. 경계에서 속도를 적용하는 이 방법은 경계를 따라 지형 변화를 허용하지 않습니다. 기준 속도는 서쪽, 남서부 및 남풍 방향에 대해 해발 10 미터에서 할당되었습니다. 동풍의 경우, 속도는 뒤쪽 (Y- 최대) 경계에서 경사 10 미터 위의 기준 높이에 할당되었습니다.

    풍력 프로파일 power 법칙은 z 방향으로 최대 360m까지 모든 미터에서 계산되었습니다. 속도는 메쉬 크기와 동일한 간격으로 평균화되었습니다. 속도가 할당된 높이 간격은 2, 4, 6, 8, 10, 20, 70, 181, 270 및 360 미터입니다. 속도 프로파일을 설정 한 후 각 높이 간격에 대한 값은 네 가지 풍향 (서쪽, 남서쪽, 남쪽 및 동쪽) 각각에 대해 X 및 Y 구성 요소로 세분화되었습니다. 초기 조건은 메쉬 블록의 외부면에 할당되어 비축에 도달하기 전에 속도 프로파일이 개발 될 수있는 충분한 수평 공간을 남겼습니다.

    풍력 프로필 power 법칙은 다음과 같습니다.

    \ displaystyle {{u} _ {x}} = {{u} _ {r}} {{\ left ({\ frac {{{{z} _ {x}}}} {{{{z} _ { r}}}}} \ right)} ^ {\ propto}}, 여기서

    U x  = 높이에서의 풍속 x
    U r  = 기준 높이에서의 풍속
    Z x  = 높이 x
    Z r  = 기준 높이
    α = 1/7 ‐ 대기 안정성 계수

    지형

    3 개의 지형파일인 스테레오리소그래피 (STL) 파일이 생성되어 모델에 통합되었습니다. 개별 파일은 지형, 창고 및 기둥에 해당합니다. 옵션 A와 B에 대해 다른 STL 파일이 생성되었습니다.

    메싱

    모델 도메인은 각 풍향에 대해 조정되었습니다. 메쉬 크기는 옵션 A의 경우 240 만에서 330 만 셀, 옵션 B의 경우 130 만 셀입니다. 정확하게 기둥 근처에 높이 2m, 길이 4m, 너비 4m의 셀 크기를 사용했습니다. 해당 지역의 속도를 계산합니다.

    경계 조건

    비축 시뮬레이션에는 네 가지 경계 유형이 사용되었습니다. 모든 풍향에 대해 상단 경계 (Z-max)가 정체 압력으로 지정되었습니다. 바람의 방향에 따라 두 개의 측벽이 유출 경계 조건으로 지정되었습니다. 나머지 두 측벽에는 그리드 오버레이 경계가 지정되었습니다. 그리드 오버레이를 사용하면 초기 조건의 속도를 모델에 입력 할 수 있습니다. 중첩 된 블록을 사용하여 원하는 메시 해상도와 배율을 만들었습니다. 내포된 블록 사이의 경계면에서 대칭 경계 조건이 사용되었습니다. 대칭을 사용하면 블록간에 정보를 전송할 수 있습니다. 그림 1은 서쪽 풍향 (y 방향)에 대한 경계 조건 설정을 보여줍니다. 다른 풍향의 경우 경계 조건을 적용하는 데 유사한 방법이 사용되었습니다.

    경계 조건 서쪽 풍향
    그림 1. 서쪽 풍향의 경계 조건

    장벽

    FLOW-3D 의 배플 기능은 비축된 곳의 주변에 바람 장벽을 만드는데 사용되었습니다. 옵션 A와 B의 배플은 높이가 30 미터였으며 지형을 따라 여러 부분으로 구성되었습니다. 모델링된 장벽은 본질적으로 다공성입니다. 34 %의 다공성 값 (즉, 34 % 개방 면적) 및 해당 속도 대 압력 강하 값은 장벽 제조업체에서 얻었습니다. FLOW-3D의  모델과 연관된 흐름 다공성 손실이 지정될 수있는 배플 알고리즘을 사용합니다. 배플은 무한히 얇고 부피를 차지하지 않습니다.

    시뮬레이션 결과

    옵션 A

    옵션 A의 경우 풍속 7.5m / s에 대한 장벽이 있거나없는 4 가지 풍향을 분석하고 시뮬레이션했습니다.

    바람의 방향배리어없는 최대 속도 (m / s)배리어가있는 최대 속도 (m / s)최대 속도 감소
    서부13.58611.27817 %
    남서부13.04510.79617 %
    남쪽12.35212.122 %
    동쪽9.768.59712 %

    각 시뮬레이션의 최대 속도와 장벽과 장벽이 없는 경우 사이의 최대 속도 감소는 위의 표 1에 나와 있습니다. 장벽은 남풍의 최대 속도에 가장 적은 영향을 미칩니다. 옵션 A에 대한 장벽 추가로 최대 속도가 2 % 감소했습니다. 장벽은 서풍 또는 남서풍이있는 전체 파일 케이스의 속도에 가장 큰 영향을 미쳤습니다. 최대 속도는 서풍과 남서풍 모두에서 17 % 감소했습니다.

    옵션 B

    옵션 B의 경우 풍속 7.5m / s에 대한 장벽이 있거나없는 네 가지 풍향을 분석하고 시뮬레이션했습니다.

    그림 2. 옵션 A : 장벽이없는 서풍의 비축량에서 계산 된 속도 크기
    장벽이있는 속도 크기 서풍
    그림 3. 옵션 A : 장벽이있는 서풍 방향의 비축에서 계산 된 속도 크기
    바람의 방향배리어없는 최대 속도 (m / s)배리어 포함 최대 속도 (m / s)최대 속도 감소
    서부15.9711.3629 %
    남서부15.149.2139 %
    남쪽13.410.124 %
    동쪽12.787.1544 %
    그림 4. 옵션 B. 장벽이없는 동풍의 비축량에서 계산 된 속도 크기
    그림 5. 옵션 B : 장벽이있는 동풍의 비축에서 계산 된 속도 크기

    결론

    모델 결과는 비축물 주변에 장벽을 추가하는 것이 속도를 줄이고 비산먼지 배출을 방지하는데 도움이 된다는 것을 분명히 보여주었습니다. 현장 주변의 장벽 추가와 관련된 비용이 있지만, 이 옵션은 먼지 배출량을 줄임으로써 환경 규범을 달성하는 데 도움이 될 것입니다. 모델 결과를 보면 FLOW-3D가 비산먼지 방출을 연구하기 위한 정확하고 신뢰할 수 있는 도구로 사용될 수 있다는 것이 분명합니다. 추가 설계 변경과 철골 배치의 새로운 옵션이 제안될 경우 FLOW-3D에서 쉽게 모델링하여 비용 및 환경적으로 효과적인 최적의 구성을 결정할 수 있습니다.

    Filling / 충진

    Filling / 충진

    재료 비용을 줄이고 사이클 시간을 개선하기 위해 소비재 회사는 슬로 싱, 튀기 및 공기 혼입을 포함한 많은 자유 표면 유체 문제를 처리해야합니다.

    Predicting Entrained Air in a Bottle Filling Example

    혼입된 공기는 생산 라인에서 컨테이너가 채워질 때 액체의 부피를 증가시킬 수 있습니다. 아래 왼쪽 이미지는 높이가 약 20cm인 병에 1.2 초 동안 채우는 것을 보여줍니다. 색상 음영은 액체에서 공기의 부피 비율을 나타냅니다. 병에서 짧은 시간과 높은 수준의 혼합으로 인해 공기가 표면으로 올라와 빠져 나갈 시간이 없었습니다. 그러나 오른쪽 이미지에서 볼 수 있듯이 약 1.7 초의 추가 시간이 지나면 표면으로 상승하는 공기로 인한 액체 부피 감소가 명확하게 보입니다. FLOW-3D의 드리프트 플럭스 모델을 사용하면 액체의 기포와 같은 성분을 분리하여 분리 할 수 있습니다.

    Air entrainment (left) and separation of air and liquid (right)

    In by 9, out by 5 – Rapid evaluation of Tide® bottle filling

    FLOW-3D를 사용하여 새로운 Tide 병 디자인의 채우기를 모델링하는 방법을 설명하는이 기사는 The Procter and Gamble Company의 기술 부문 책임자 인 John McKibben이 기고했습니다.

    오전 9시에 긴급한 이메일을 받았다고 상상해보십시오.

    새로운 Tide® 병 디자인 중 하나가 핸들을 채우고 충전 장비에 문제가있을 수 있음을 방금 깨달았습니다. 프로토 타입 병도없고 몇 주 동안도 없을 것입니다. 디자이너와 소비자는 디자인의 모양을 좋아하지만 그것이 채우는 방식은 우리 생산 시설의 쇼 스토퍼가 될 수 있습니다.
    이 상황을 접했을 때 저는 3D 지오메트리 (그림 1)의 스테레오 리소그래피 (.stl) 파일을 요청하여 응답을 시작했고 제가 할 수있는 일을 확인했습니다. FLOW-3D는 .stl 파일을 사용하여 지오메트리를 입력 할 수 있으며 채우기에 대한 자유 표면 문제를 해결할 수 있어야한다는 것을 알고있었습니다. 나는 이것이 잠재적 인 문제에 대한 좋은 질적 이해를 제공 할 것으로 기대했지만,이 응용 프로그램에 대해 얼마나 정확한지에 대해서는 약간 불확실했습니다.

    Setting up and Running the Simulation

    오후 1 시경에 지오메트리 파일, 유량 및 유체 속성을 받았습니다. 몇 시간 내에 시뮬레이션이 실행되어 예비 결과를 제공했습니다. 저는 제 고객을 초대하여 결과를 간단히 살펴 보았고 그는 “보스의 상사”도 함께 살펴 보았습니다. 그래서 저녁 5 시까 지 예비 결과를보고 원래 우려 사항이 문제가 아니라고 판단했습니다.

    그러나 결과는 몇 가지 다른 질문을 제기했습니다. 핸들을 채우면 유입되는 유체 분사가 많이 분리되었습니다. 나는 이것이 동반 된 공기와 거품의 양을 증가시킬 것이라는 것을 알고 있었다 (우리는 결국 세탁 세제를 채우고있다). FLOW-3D 공기 혼입 모델을 테스트하기로 결정했습니다. 이 모델은 원래 난류 제트 용으로 개발되었으며,이 층류 문제를 볼 때 얼마나 잘 수행 될지 확신 할 수 없었습니다.

    Figure 2: Filled results
    Experimental comparison of bottle filling model with and without the air entrainment model, courtesy of The Procter & Gamble Company.

    그림 2는 공기 유입 모델이 있거나없는 병 충전 모델의 결과를 보여줍니다. 혼입 된 공기가 포함되면 충전 레벨이 크게 증가합니다. 혼입 된 공기가 병 상단에서 액체를 밀어 내지는 않지만 공기 혼입 정확도를 확인해야 할만큼 충분히 가깝습니다.

    그림 3은 몇 주 후에 실행 된 실험의 이미지와 공기 혼입 수준을 비교합니다 (시제품 병이 제공되었을 때). 제트 분리 및 충진 수준의 정 성적 일치는 우수하며 시뮬레이션이 병 설계를 선별하기에 충분히 정확하다는 것을 확인했습니다.

    FLOW-3D 튜토리얼 V12

    FLOW-3D 튜토리얼 V12

    빠른 시작

    이 튜토리얼 매뉴얼은 FLOW-3D 처음 사용하는 사용자에게 그래픽 사용자 인터페이스(GUI)의 주요 구성 요소를 쉽게 익히도록 하고, 다양한 시뮬레이션의 설정 및 실행 방법을 안내하기 위한 것입니다.

    이 매뉴얼에 있는 실습과정은 FLOW-3D의 기본 사항을 다루기 위한 것입니다. 이 매뉴얼에서 제시하는 문제는 다양한 주제를 설명하고, 발생할 수 있는 많은 질문을 해결하기 위해 선정되었습니다. 이 매뉴얼의 실습과정은 FLOW-3D실행하는 컴퓨터에 앉아 사용하는 것이 가장 좋습니다.

    CFD 사용 철학에 대한 간단한 섹션 다음에는 중요 파일과 시뮬레이션 파일을 실행하는 방법이 소개되어 있습니다. 이 소개 섹션 다음에는 모델 설정, 시뮬레이션 실행 및 포스트 프로세스, Simulation Manager 탐색 방법에 대한 설명이 있습니다. 이러한 각 단계에 대한 자세한 내용은 모델 설정, 컴퓨팅 결과 및 후처리 장에서 확인할 수 있습니다.

    1.CFD 사용에 대한 철학

    CFD (Computational Fluid Dynamics)는 유체 흐름(질량, 운동량 및 에너지 보존)에 대한 지배 방정식의 컴퓨터 솔루션입니다. 지정된 지배방정식은 이론 장에 설명된 Numerical방법을 사용하여 이산화되고 계산됩니다.

    CFD 소프트웨어를 사용하는 것은 여러 면에서 실험을 설정하는 것과 유사합니다. 실제 상황을 시뮬레이션하기 위해 실험을 올바르게 설정하지 않으면, 그 결과는 실제 상황을 반영하지 않습니다. 같은 방법으로 수치 모델이 실제 상황을 정확하게 나타내지 않으면, 그 결과는 실제 상황을 반영하지 않습니다. 사용자는 어떤 것이 중요한지, 어떻게 표현해야 하는지를 결정해야 합니다. 시작하기 전에 다음과 같은 질문을 하는 것이 중요합니다.

    • CFD 계산에서 무엇을 알고 싶습니까?
    • 중요한 현상을 포착하기 위해 규모와 Mesh는 어떻게 설계되어야 하는가?
    • 실제 물리적 상황을 가장 잘 나타내는 경계 조건은 무엇입니까?
    • 어떤 종류의 유체를 사용해야합니까?
    • 이 문제에 어떤 유체 특성이 중요합니까?
    • 다른 어떤 물리적 현상이 중요합니까?
    • 초기 유체 상태는 어떻게 됩니까?
    • 어떤 단위 시스템을 사용해야합니까?

    모델링 되는 문제가 실제 상황을 가능한 한 유사하게 나타내는지 확인하는 것이 중요합니다. 사용자는 복잡한 시뮬레이션 작업을 해결 가능한 부분으로 나누는 것이 좋습니다.

    복잡한 물리 효과를 추가하기 전에, 간단하고 쉽게 이해할 수 있는 근사값으로 점차적으로 시작하여 프로세스 진행하십시오. 간단한 손 계산(베르누이 방정식, 에너지 균형, 파동
    전파, 경계층 성장 등)은 물리 및 매개 변수를 선택하는데 도움이 되고, 결과와 비교할 수 있는 점검항목을 제공합니다.

    CFD의 장단점을 이해하면 분석을 진행하는데 도움이 될 수 있습니다. CFD는 다음과 같은 경우 탁월한 분석 옵션입니다.

    • 기하 구조, 물리학 또는 필요한 상세 수준으로 인해 표준 엔지니어링 계산이 유용하지 않은 경우가 많습니다.
    • 실제 실험은 비용이 많이 소요됩니다.
    • 실험에서 수집할 수 있는 것보다 유체흐름에 대한 자세한 정보가 필요한 경우 유용합니다.
    • 위험하거나 적대적인 조건, 확장이 잘되지 않는 프로세스 등으로 인해 정확한 실험 측정을 하기가 어려운 경우
    • 복잡한 흐름 정보에 대한 커뮤니케이션

    CFD는 다음과 같은 경우에 덜 효과적입니다.

    • 솔루션이 계산 리소스가 매우 많이 소요되거나, 도메인 크기를 줄이기 위한 가정 또는 해결되지 않은 물리적 현상을 설명하기 위한 반 임계 모델이 필요한 경우
    • CFD 시뮬레이션에 대한 입력이 되는 중요한 물리적 현상이 알려지지 않은 경우
    • 물리적 현상이 잘 이해되지 않거나 매우 복잡한 경우

    CFD를 사용할 때 명심해야 할 몇 가지 중요한 참고 사항이 있습니다.

    • CFD는 규정된 초기 및 경계 조건에 따라 지정된 지배 방정식의 수치해석 솔루션입니다. 따라서 모델 설정, 즉 어떤 방정식을 풀어야 하는지, 재료 특성, 초기 조건 및 경계 조건이, 가능한 한 물리적 상황과 최대한 일치해야 합니다.
    • 방정식의 수치 해는 일반적으로 어떤 종류의 근사치를 필요로 합니다. 물리적 모델에 대한 가정과 해결방법을 검토한 후 사용하는 것이 좋습니다.
    • 디지털 컴퓨터는 숫자가 유한 정밀도로 이진수로 표시되는 방식으로 인해 반올림 오류가 발생합니다. 이는 문제를 악화시키기 때문에 매우 근소한 숫자의 차이를 계산해야 하는 상황을 피하십시오. 이러한 상황의 예는 시뮬레이션 도메인이 원점에서 멀리 떨어져 있을 때입니다.

     

    2.중요한 파일

    FLOW-3D 시뮬레이션과 관련된 많은 파일이 있습니다. 가장 중요한 것들이 아래에 설명되어 있습니다. 모든 prepin.* 파일의 명칭에서 prepin는 파일 형식을 의미하며, 별표시* 위치는 시뮬레이션 이름을 의미합니다. ( : prepin.example_simulation.)

    • ·prepin.*: 시뮬레이션용 입력 파일입니다. 시뮬레이션 설정을 설명하는 모든 입력 변수가 포함되어 있습니다.
    • ·prpgrf.*: 이것은 전 처리기 출력 파일입니다. 여기에는 계산된 초기 조건이 포함되며 시뮬레이션을 실행하기 전에 설정을 확인하는 데 사용될 수 있습니다.
    • ·flsgrf.*: 솔버 출력 파일입니다. 시뮬레이션의 최종 결과가 포함됩니다.
    • ·prperr.*, report.*, prpout.*: 이 파일들은 Preprocessor Diagnostic Files.
    • ·hd3err.*, hd3msg.*, hd3out.*: 이 파일들은 Solver Diagnostic Files.

    모든 시뮬레이션 파일은 단일 폴더에 함께 유지하므로, 설명이 될 수 있는 시뮬레이션 이름을 사용하는 것이 좋습니다. 그러나 매우 긴 파일 이름은 운영 체제에 따라 문제가 될 수 있습니다.

    노트

    • 시뮬레이션 이름이 inp(즉, 입력 파일이 있다면 prepin.inp) 출력 및 진단 파일은 모두 .dat이름을 갖습니다. 예: flsgrf.dat.
    • 모든 입력 파일은 네트워크 위치의 컴퓨터 대신 로컬 디렉토리에 저장하는 것이 좋습니다. 이것은 솔버가 더 빠르게 실행되고 GUI의 응답 속도가 빨라지며 실행중인 시뮬레이션을 방해하는 네트워크 문제 가능성을 제거합니다.

    3.시뮬레이션 관리자

    FLOW-3D 시뮬레이션 관리자의 탭은 주로 시뮬레이션을 실행할 수 있도록 시뮬레이션 환경을 구성하고 실행 시뮬레이션에 대한 상태 정보를 표시하는데 사용됩니다.

    작업 공간 (Workspaces)

    작업 공간(Workspaces)Simulation Manager의 필수 부분이며 파일을 FLOW-3D에서 처리하는 방식입니다. 기본적으로 시뮬레이션을 포함하고 구성하는 폴더입니다. 몇 가지 예를 들면 시뮬레이션과 또 다른 작업 공간인 검증 사례를 포함하도록 할 수 있습니다:

    포트폴리오의 작업 공간

    새로운 작업 공간 만들기

    튜토리얼에서는 작성하려는 시뮬레이션을 포함할 작업 공간(Workspaces)을 작성하십시오.

    1.File -> New workspace 이동

    2.작업 공간 이름으로 Tutorial를 입력하십시오.

    3.기본 위치는 현재 사용자의 홈 디렉토리에 있습니다. 다른 곳에서 찾을 수 있지만 기본 위치가 우리의 목적에 적합합니다.

    4.하위 디렉토리를 사용하여 작업 공간 이름 만들기 확인란을 선택합니다. 이렇게 하면 파일 시스템에서 작업 공간에 대한 새로운 하위 디렉토리가 만들어져 시뮬레이션 파일을 훨씬 쉽게 구성할 수 있습니다.

    새로운 작업 공간 만들기

    5.확인을 눌러 새 작업 공간을 작성하십시오. 이제 포트폴리오에 표시됩니다.

    새로운 작업 공간 만들기

    작업 공간 닫기

    포트폴리오를 정리하고 탐색하기 쉽도록 필요 없는 작업공간을 닫는 것이 편리합니다. 작업 공간을 닫으면 포트폴리오에서 해당 작업 공간만 제거됩니다. 그러나, 컴퓨터에서 작업 공간을 삭제하지는 않습니다.

    작업 공간을 닫으려면

    1.기존 작업 공간을 마우스 오른쪽 버튼으로 클릭하고 작업 Close Workspace 선택하십시오. 또는 포트폴리오에서 작업 공간을 선택 (왼쪽 클릭) 하고 Delete 키를 누를 수 있습니다.

    2.작업 공간을 닫을 것인지 묻는 메세지가 표시됩니다. 예를 선택하십시오.

    3.포트폴리오는 더 이상 닫힌 작업 공간을 포함하지 않습니다.

    기존 작업 공간 열기

    오래된 작업 공간을 열어야 할 때가 있을 것입니다. 예를 들어, 새 프로젝트에 유사한 시뮬레이션을 작성하기 전에 기존 시뮬레이션의 설정을 검토할 수 있습니다. 기존 작업 공간을 열려면

    1.File -> Open Workspace를 선택하십시오

    2.작업 공간 파일이 있는 디렉토리를 찾으십시오. Tutorial.FLOW-3D_Workspace.

    작업 공간 열기

    3.작업 공간을 로드 하려면 OK누르십시오.

    작업 공간에서 시뮬레이션 작업

    작업 공간을 사용하는 방법을 알았으니, 여기에 시뮬레이션을 추가해 봅시다.

    Example를 추가하십시오

    작업 공간에 작업 시뮬레이션을 추가하는 가장 간단한 방법은 포함된 예제 시뮬레이션 중 하나를 추가하는 것입니다. FLOW-3D의 다양한 기능을 사용하는
    방법을 보여주기 위해 설계된 간단하고 빠른 시뮬레이션입니다. 기존 작업 공간에 예제를 추가하려면 다음을 수행하십시오.

    1.포트폴리오에서 원하는 작업 공간을 강조 표시하십시오

    2.File -> Add example 선택하십시오. 또는 작업공간을 마우스 오른쪽 버튼으로 클릭하고 예제 추가선택할 수 있습니다.

    3.예제 대화 상자에서 예제를 선택하고 열기를 누르십시오. 자연 대류(Natural Convection) 예제를 선택했습니다.

    시뮬레이션 예제 추가

    4.새 시뮬레이션 대화 상자가 열립니다.

    5.디렉토리가 작업 공간 위치에 있는지 확인하는 것이 좋으므로 기본 시뮬레이션 이름과 위치를 잘 확인하는 것이 좋습니다. FLOW-3D는 모든 시뮬레이션 파일을 이 작업 공간 디렉토리의 별도 하위 디렉토리에 배치하여 파일 구성을 쉽게 만들어 줍니다.

    6.시뮬레이션을 위한 단위 시스템을 선택하십시오. 표준 단위 시스템이 권장되지만 각 단위를 독립적으로 선택하기 위해 사용자 지정 단위 시스템을 선택할 수 있습니다.

    7.확인을 눌러 새 시뮬레이션을 작업 공간에 추가하십시오.

    작업 공간에서의 시뮬레이션

    작업 공간에서 시뮬레이션 제거

    작업 공간에서 시뮬레이션을 제거해야 하는 경우가 있습니다 (이는 작업 공간에서 시뮬레이션을 제거만 하며, 컴퓨터에서 시뮬레이션을 삭제하지는 않습니다). 작업 공간에서 시뮬레이션을 제거하려면 다음을 수행하십시오.

    1.작업 공간에서 기존 시뮬레이션을 마우스 오른쪽 버튼으로 클릭하고 (이 경우 이전 섹션에서 추가 한 예제 사용) 시뮬레이션 제거를 선택하십시오. 또는 작업 공간에서 시뮬레이션을 선택 (왼쪽 클릭)하고 Delete 키를 누를 수 있습니다.

    2.작업 공간에는 더 이상 시뮬레이션이 포함되지 않습니다.

    모든 작업 공간 및 디스크에서 시뮬레이션 삭제

    작업 공간에서 시뮬레이션을 제거하는 것 외에도 디스크에서 모든 시뮬레이션 파일을 삭제해야 할 수도 있습니다. 작업 공간에서 시뮬레이션을 제거하고 디스크에서 시뮬레이션
    파일을 삭제하려면 다음을 수행하십시오.

    1.작업 공간에서 기존 시뮬레이션을 마우스 오른쪽 단추로 클릭하고 (이 경우 이전 섹션에서 추가 한 예제 사용) 모든 작업 공간 및 디스크에서 시뮬레이션
    삭제를
    선택하십시오.

    2.시뮬레이션 디렉토리에서 삭제할 파일을 선택할 수 있는 창이 나타납니다. 삭제할 파일을 선택한 다음 확인을 눌러 해당 파일을 삭제하거나 취소를 눌러 작업을 중단하십시오.

    3.OK를 선택한 경우 선택한 작업 공간은 더 이상 시뮬레이션을 포함하지 않습니다. 선택한 작업 공간의 모든 시뮬레이션 파일은 디렉토리에서 삭제됩니다.

    경고

    이 작업은 취소할 수 없으므로 계속하기 확인 후 파일을 삭제해야 합니다.

    작업 공간에 기존 시뮬레이션 추가

    기존 시뮬레이션을 작업 공간에 추가하려면 다음을 수행하십시오.

    1.열린 작업 공간을 마우스 오른쪽 버튼으로 클릭하고 기존 시뮬레이션 추가 선택합니다. 작업 공간을 선택한 다음 File->Add Existing Simulation 을 선택할 수도 있습니다.

    2.prepin.*파일 위치로 이동하여 열기를 선택하십시오.

    작업 공간에 기존 시뮬레이션 추가

    3.시뮬레이션이 이제 작업 공간에 나타납니다.

    작업 공간에 새로운 시뮬레이션 추가

    대부분의 경우 기존 시뮬레이션을 사용하는 대신 새 시뮬레이션을 작성하게 됩니다. 작업 공간에 새로운 시뮬레이션을 추가하려면:

    1.기존 작업 공간을 마우스 오른쪽 버튼으로 클릭하고 새 시뮬레이션 추가 선택하십시오.

    2.시뮬레이션 이름을 입력하라는 message가 표시됩니다. 이 예제에서는 heat transfer example 불러오십시오.

    3.그런 다음 드롭다운 목록을 사용하여 시뮬레이션을 위한 단위 시스템을 결정합니다. 사용 가능한 옵션은 질량, 길이, 시간, 전기요금
    각각 g, cm, s, coul기준의 Kg, m, s, CGS입니다. 또한 엔지니어링 단위도 사용할 수 있으며, slug, ft, s의 기초 단위가 있지만, 전기
    충전을 위한 단위는 없습니다. 이러한 옵션 중 어느 것도 해당되지 않는 경우, 질량, 길이, 시간 및 전기요금에 대한 기준 등을 사용자 정의하여 사용자 지정 단위 시스템을 사용할 수 있습니다.

    4.온도 단위는 드롭다운 목록을 사용하여 지정해야 합니다. 사용 가능한 옵션은 SI CGS 단위의 경우 Celsius
    Kelvin, 엔지니어링 단위의 경우 Fahrenheit Rankine입니다. Custom units(사용자 정의 단위) 옵션을 선택한 경우, 사용 가능한 온도 단위는 질량
    및 길이에 대해 선택한 기본 단위에 따라 변경됩니다.

    노트

    새 시뮬레이션의 시뮬레이션 단위는 신중하게 선택하십시오. 일단 설정하면 단위를 변경할 수 없습니다.

    5.이 시뮬레이션에 사용된 템플릿이 기본 템플릿이 됩니다. 템플릿은 포함된 설정을 새 시뮬레이션에 적용하는 저장된 값 세트입니다. 다른 템플릿을 사용해야하는 경우
    찾아보기 아이콘 (
    browse_icon_v12)을 클릭하여 사용 가능한 템플릿 목록에서 선택하십시오.

    6.기본 시뮬레이션 이름과 위치는 디렉토리가 작업 공간 위치에 있는지 확인하는 것이 좋습니다. FLOW-3D는 모든 시뮬레이션 파일을 이 작업 공간 디렉토리의 별도 하위 디렉토리에 배치하여 파일 구성을 훨씬 쉽게 만듭니다. 시뮬레이션을 다른 위치에 저장하려면 찾아보기 아이콘 ( browse_icon_v12)을 사용하여 원하는 위치로 이동하십시오.

    7.확인을 클릭하여 작업 공간에 새 시뮬레이션을 추가하십시오.

    heat transfer example

    새로운 시뮬레이션 추가

    다른 옵션

    우리는 지금 이러한 옵션을 사용하지 않는 동안, 이 시뮬레이션을 마우스 오른쪽 버튼으로 클릭하여 추가 옵션에 대한 액세스를 제공합니다.

    일반적으로 사용되는 Add Simulation Copy… 그리고 Add Restart Simulation…을 추가합니다. 첫 번째 옵션은 기존 시뮬레이션의 사본을
    작성하고, 두 번째 옵션은 기존 시뮬레이션을 복사하고 원래 시뮬레이션의 결과를 다시 시작 시뮬레이션의 초기 조건으로 사용하도록 다시 시작 옵션을 구성합니다.

    추가 정보

    재시작 시뮬레이션에 대한 자세한 내용은 도움말에서 모델 설정 장의 재시작 섹션을 참조하십시오.

    전처리 및 시뮬레이션 실행

    시뮬레이션 전처리

    시뮬레이션 전처리는 초기 조건을 계산하고 입력 파일에서 일부 진단 테스트를 실행합니다. 문제가 올바르게 구성되었는지 확인하거나 전 처리기의 진단 정보가 필요한 경우에
    유용합니다. 시뮬레이션을 실행하기 전에 전처리할 필요가 없습니다. 시뮬레이션을 전처리 하려면

    1.작업 공간에서 시뮬레이션을 마우스 오른쪽 버튼으로 클릭하고 Preprocess Simulation->Local 선택합니다. 이 경우 입력 파일 heat transfer example이 아직 완전히 정의되지 않았으므로 작업 공간에서 예제 문제를 선택하십시오.

    2.전처리 프로세스가 시작되고 Simulation Manager 하단의 텍스트 창에 일부 정보가 인쇄된 후 성공적으로 완료됩니다. 포트폴리오에서 시뮬레이션 이름 옆의 아이콘도 시뮬레이션이 성공적으로 처리되었음을 나타내도록 변경됩니다.

    추가 정보

    자세한 내용은 도움말의 컴퓨팅 결과 장의 전처리 섹션을 참조하십시오.

    시뮬레이션 실행

    시뮬레이션을 실행하면 입력 파일에 정의된 문제에 대한 지배 방정식(물리적 모델, 형상, 초기 조건, 경계 조건 등)이 해석됩니다. 시뮬레이션을 실행하려면

    1.작업 공간에서 시뮬레이션을 마우스 오른쪽 버튼으로 클릭하고 Run Simulation->Local을 선택하십시오. 이 경우 입력 파일 heat transfer example이 아직 완전히
    정의되지 않았으므로 작업 공간에서 예제 문제를 선택하십시오.

    2.솔버가 시작되고 시뮬레이션 관리자 하단의 텍스트 창에 일부 정보가 인쇄되고 플롯이 업데이트 된 후 성공적으로 완료됩니다. 포트폴리오에서 시뮬레이션 이름 옆의
    아이콘도 시뮬레이션이 성공적으로 실행되었음을 나타내도록 변경됩니다. 또한 솔버가 실행되는 동안 큐에 시뮬레이션이 나타나는 것을 볼 수 있으며, 완료되면 사라집니다
    .

    추가 정보

    시뮬레이션 실행 및 진단 읽기에 대한 자세한 내용은 도움말의 컴퓨팅 결과 장에서 솔버 실행 섹션을 참조하십시오.

    작업 공간에서 모든 시뮬레이션 실행

    작업 공간을 마우스 오른쪽 버튼으로 클릭하고 Simulate Workspace->Local을 선택하여 작업 공간에서 모든 시뮬레이션을 실행할 수도 있습니다.

    추가 정보

    자세한 내용은 컴퓨팅 결과 장에서 솔버 실행 섹션을 참조하십시오.

    대기열

    사전 처리 또는 실행에 작업이 제출되면 큐의 맨 아래에 시뮬레이션이 자동으로 추가됩니다. 그런 다음 솔버에 사용 가능한 라이센스 및 계산 리소스가 있으면 시뮬레이션이 사전 처리되거나 실행됩니다. 대기열에 있지만 아직 전처리 또는 실행되지 않은 시뮬레이션은 대기열 맨 아래의 컨트롤을 사용하여 대기열에서 다시 정렬하거나 대기열에서 제거할 수 있습니다.

    추가 정보

    자세한 내용은 컴퓨팅 결과 장을 참조하십시오.

    파일 시스템에서 파일 찾기

    어떤 이유로 구조물 파일에 액세스해야 하는 경우 (아마 *.STL 폴더에 파일을 배치해야 함) 표시된 파일 경로를 시뮬레이션 입력 파일로 클릭하여 파일 시스템의 해당 위치로 이동할 수 있습니다.

    파일 링크

    4.모델 설정

    Model Setup(모델 설정) 탭은 시뮬레이션 관리자에서 현재 선택한 시뮬레이션에 대한 입력 매개 변수를 정의하는 곳입니다. 여기에는 전역설정, 물리학 모델, 유체,
    기하학, 메싱, 구성요소 특성, 초기 조건, 경계 조건, 출력 옵션 및 숫자가 포함된다.

    이 섹션은 물에 잠긴 모래(; 파랑)의 바닥에서 가열된 구리 블록(; 빨간색)에 의해 발생하는 열 기둥(아래)을 보여주는 간단한 시뮬레이션 설정 방법을 안내합니다.

    예제 문제

    이 튜토리얼은 방법이나 모델이 어떻게 작동하는지, 옵션을 선택한 이유 등에 대한 포괄적인 논의를 의도한 것이 아니며, 이 특정 시뮬레이션을 설정하기 위해 수행해야 할 사항에
    대한 간략한 개요일 뿐입니다. 여기서 행해지는 것에 대한 방법/모델과 추론의 세부사항은 사용 설명서의 다른 장에서 확인할 수 있습니다.

    시작하려면 새 작업 공간을 작성하고 새 시뮬레이션을 추가하십시오. 이를 수행하는 방법에 대한 지침은 새 작업 공간 작성 및 작업 공간에 새 시뮬레이션 추가를 참조하십시오.

    탐색

    모델 설정은 주로 빨간색으로 표시된 처음 9 개의 아이콘의 탐색을 통해 수행됩니다. 각 아이콘은 시뮬레이션의 특정 측면을 구성하기 위한 위젯을 엽니다. Global에서 시작하여 Numerics로 끝나는 다음 섹션은 각 위젯의 목적을 보여줍니다.

    시뮬레이션의 다양한 측면을 정의하기위한 탐색 아이콘

    통제 수단

    다음은 FLOW-3D 사용자 인터페이스의 그래픽 디스플레이 영역에서 사용되는 마우스 컨트롤입니다.

    행동

    버튼/

    동작

    기술

    회전

    왼쪽

    길게 클릭

    마우스 왼쪽 버튼을 클릭 한 채로 Meshing & Geometry 창에서
    마우스를 움직입니다. 그에 따라 모델이 회전합니다.

    중간 버튼/스크롤

    스크롤/클릭 한
    상태

    마우스를 앞뒤로 움직여 확대/축소하려면 가운데 휠을 굴리거나 마우스 가운데 버튼을 클릭
    한 상태로 유지하십시오.

    우측

    길게 클릭

    마우스 오른쪽 버튼을 클릭 한 채로 창에서 마우스를 움직입니다. 모델이 마우스와 함께 움직입니다.

    객체에 초점 설정

    해당 없음

    객체 위에 커서를 놓기

    커서를 개체 위로 가져 가면 마우스 오른쪽 버튼 클릭 메뉴를
    통해 추가 조작을 위해 개체가 활성화됩니다. 개체가 활성화되면 강조 표시됩니다. Meshing & Geometry 탭에서 Tools->
    Mouse Hover
    Selection
    환경 설정 이 활성화된 경우에만
    수행됩니다.

    선택

    왼쪽

    더블 클릭

    객체를 두 번 클릭하면 마우스 오른쪽 버튼 메뉴를 통해 추가
    조작을 위해 객체를 선택하고 활성화합니다. Meshing
    & Geometry
    탭에서 Tools
    ->Mouse Hover Selection 환경 설정 이
    비활성화 된 경우에만 활성화됩니다.

    액세스 객체 속성

    우측

    딸깍 하는 소리

    강조 표시된 객체를 마우스 오른쪽 버튼으로 클릭하면 객체
    식별, 표시/숨기기, 활성화/비활성화, 투명도 조정 등의 옵션 목록이 표시됩니다.

    커서 좌표 반환 (프로브)

    왼쪽

    Shift + 클릭

    Shift 키를 누르면 커서가 대상으로 바뀝니다. Shift 키를 누른 상태에서 클릭하면 화면의 왼쪽 하단에 표시된 표면의 좌표가 표시됩니다.

    피벗 점 배치

    왼쪽

    cntrl + 클릭

    Ctrl 키를 누르고 있으면 커서가 피벗 아이콘으로 바뀝니다. Ctrl 키를 누른 상태에서 클릭하여 피벗 점을 설정하십시오. 뷰가
    피벗 점을 중심으로 회전합니다. 토글 사용자 정의 피벗 피벗 점을 끕니다.
    보기 창 위의 버튼을 누릅니다.

    도움이
    되는 툴바 옵션도 있습니다. 옵션의 목적을 찾으려면 아이콘 위로 마우스를 가져갑니다.

    메시 및 지오메트리 탭의 컨트롤

    글로벌

    이 매뉴얼에 대한 시뮬레이션을 만들려면 원하는 작업 공간을 마우스 오른쪽 단추로 클릭하고 새 시뮬레이션 추가를 선택하십시오. 매뉴얼 섹션의 새 시뮬레이션 추가 작업 공간에 설명된 대로 이름을 ‘heat transfer example’로 지정하고 작업 공간에 추가하십시오. SI Kelvin을 각각 단위 시스템과 온도로 선택합니다. 일단 설정되면
    시뮬레이션을 위한 단위는 변경할 수 없다는 점을 기억하십시오.

    글로벌 아이콘 f3d_global_icon을 클릭하여 글로벌 위젯을 여십시오. 여기에서 정의된 단위가 표시되고 시뮬레이션 완료 시간이 설정됩니다. 이 시뮬레이션의 경우 완료 시간을 200 초로 설정하십시오. 시뮬레이션에 대한 중요한 세부 정보는 여기 노트 필드에도 추가할 수 있습니다.

    글로벌 탭 예를 들어 문제

    추가 정보

    자세한 내용은 모델 설정 장의 전역 섹션을 참조하십시오.

    물리

    물리 f3d_models_icon아이콘을 클릭하여 물리 위젯을 엽니다.

    모델 선택을위한 물리 위젯

    이 문제의 경우, 하나의 유체, 자유 표면, 경계 및 비압축/제한 압축의 기본 설정이 모두 정확합니다.

    관련 물리 메커니즘(, 추가 지배 방정식 또는 지배 방정식 용어)은 물리 위젯에서 정의됩니다. 모델을 활성화하려면 해당 모델의 아이콘을 마우스 왼쪽 버튼으로 클릭하고활성화 선택하십시오. 이 시뮬레이션을 위해서는 다음 모델을 활성화해야 합니다.

    ·Density evaluation(밀도 평가): 이 모델은 열 기둥을 생성하는 밀도 변화를 설명합니다. 다른 양(: 온도 또는 스칼라)의 함수로 평가된 밀도를 선택하고 Include volumetric thermal expansion 상자를 선택하십시오.

    문제 평가를위한 밀도 평가 모델

    ·Gravity and non-inertial reference frame(중력 및 비 관성 기준 프레임): 중력을 나타내는 힘이 추가되므로 Z 중력 성분에 -9.81을 입력하십시오.

    예를 들어 중력 모델

    ·
    Heat transfer(열 전달): 이 모델은 유체와 고체 물체 사이의 열 전달을 설명합니다. 이 시뮬레이션의 경우 First order for the Fluid internal Energy advection를 선택하고 Fluid to solid heat transfer를 활성화하려면 확인란을 선택하십시오. 나머지 옵션은 기본값으로 두어야합니다.

    열전달 모델 예 : 문제

    ·
    Viscosity and turbulence(점성 및 난류): 이 모델은 유체의 점성 응력을 설명합니다. Viscous flow 옵션을 선택하고 나머지 옵션은 기본값으로 두십시오.

    예를 들어 문제의 점도 모델

    추가 정보

    자세한 내용은 모델 설정 장의 물리 섹션을 참조하십시오.

    유체

    유체의 속성은 모델 설정 탭의 유체 위젯에 정의되어 있습니다. 유체 위젯은 수직 도구 모음에서 Fluids f3d_fluids_icon f3d_fluids_icon아이콘을 클릭하여 액세스할 수 있습니다. 먼저 유체 옵션 1 이 속성 옵션으로 선택되어 있는지 확인하십시오. 유체 1의 속성은 수동으로 입력할 수 있지만 일반적인 유체의 속성을 설정하는 빠른 방법은 재료 속성로드 버튼Matdatbas을 클릭하여 재료 데이터베이스에서 유체를 로드하는 것입니다. 다음으로, 원하는 재료를 탐색하십시오. 이 경우 Fluids->Liquids->Water_at_20_C를 선택하고 Load를 클릭하십시오.

    이 시뮬레이션에는 데이터베이스에 없는 특성인 체적 열 팽창 계수가 필요합니다. 밀도 하위 탭에서 207e-6을 입력하십시오. 최종 속성 세트는 다음과 같아야 합니다.

    유체 특성 (예 : 문제)

    추가 정보

    자세한 내용은 모델 설정 장의 유체 섹션을 참조하십시오.

    Geometry(기하)

    기하형상 f3d_geometry_icon아이콘을 클릭하여 물리 위젯을 엽니다.

    이 시뮬레이션을 위해 생성해야 하는 두 가지 형상은 구리 블록과 모래층이 있습니다. 둘 다 프리미티브를 사용하여 작성합니다. 보다 현실적인 시뮬레이션은 Primitives, Stereolithography(STL) Geometry File (s)/또는 Raster File (s)을 사용하여 지오메트리를 정의할 수 있습니다.

    구리 블록을 만들려면 먼저 지정된 상자 형상 아이콘을 클릭하여 작성합니다. 구리 블록을 x y 방향 원점에서 +/- 2cm 연장하고 z 방향으로 0-4cm 연장합니다. 나머지 옵션은 그대로 두고 블럭을 솔리드로 만들고 새 구성 요소에 추가합니다.

    예제 문제에 대한 구리 블록 정의

    하위 구성 요소 정의를 마치고 구성 요소 정의로 이동하려면 확인을 선택하십시오. 자동으로 열린 구성요소 추가 대화상자에서 Type as General(솔리드)을 그대로 두고 Name(이름) 필드에 Copper block을 입력한 다음 OK(확인)를 선택하여 구성요소 정의를 완료하십시오.

    상자아이콘을 다시 클릭하여 베드 하위 구성 요소를 작성하십시오. 아래 표시된 범위를 사용하고 컴포넌트에 추가 선택 사항을 새 컴포넌트(2)로 설정하십시오.

    예를 들어 침대 문제 정의

    하위 구성 요소 정의를 마치고 구성 요소 정의로 이동하려면 확인을 선택하십시오. 대화 형으로 이름 필드에서Bed를 입력한 후 구성요소 정의를 마칩니다. 최종 형상은 다음과 같이 표시됩니다.

    예제 문제에 대한 형상 정의

    새 구성 요소를 추가하면 가로 및 세로 방향으로 그래픽 표시 창에 길이 스케일이 자동으로 생성됩니다. 눈금자 도구를 사용하여 생성된 기하학적 객체의 범위를 빠르게 측정할 수 있습니다.

    노트

    표시 영역에는 지오메트리 모양 정의만 표시되므로 객체가 솔리드인지 구멍인지에 대한 정보는 표시되지 않습니다. 즐겨 찾기옵션을 사용하여 Mesh 후에 나중에 수행할 수 있습니다.

    추가 정보

    자세한 내용은 도움말 모델 설정 장의 형상 섹션을 참조하십시오.

    구성 요소 속성

    열전달 모델은 고체 구성 요소의 전도 방정식을 해결하기 위해 재료 특성이 필요합니다. 이러한 속성은 이 아이콘f3d_geometry_icon을 클릭하여 구성 요소 속성 위젯에서 설정합니다.

    구성 요소 특성 위젯

    각 구성 요소에는 솔리드 특성 및 표면 특성이 정의 되어 있어야합니다. 구리 블록에 대해 이를 설정하려면 먼저 형상 위젯에서 구성 요소 1: copper block 요소를 선택하십시오. 그런 다음 컴포넌트 특성 위젯에서 솔리드 특성을 선택하고 다음과 같이 특성을 정의하십시오.

    구리 블록 고체 특성

    여기에서 두 번째 구성 요소(베드)에 대해 설명된 구성 요소 특성 정의를 위한 대체 방법을 사용할 수 있습니다. 이 방법에서는 구성 요소 2: 베드 구성 요소를 클릭하고 재료 필드 옆에 있는 재료 특성로드 Matdatbas 아이콘을 선택하여 시작합니다. 다음으로 재료를 탐색합니다. 이 경우 Solids->Sands->Sand_Quartz 선택하고 Load를 선택하십시오.

    베드 솔리드 속성

    추가 정보

    l 자세한 내용은 모델 설정 장의 유체 섹션을 참조하십시오.

    l 주어진 물리적 모델에 필요한 속성에 대한 자세한 내용은 모델 참조 장을 참조하십시오.

    Meshing(메싱)

    Mesh Mesh 위젯에서 생성 및 정의되며, 위젯을 통해 액세스 할 수 있습니다. f3d_mesh_icon아이콘을 눌러 add_iconMesh를 추가합니다. Mesh의 범위를 형상에 빠르게 적용하려면 형상에 맞추기 라디오 버튼을 선택하고 오프셋 라디오 버튼을 백분율로 유지합니다. 블록 속성에서 셀 크기를 0.004로 설정하십시오.

    메시 블록을 형상에 맞추기

    Mesh 상단은 z 방향으로 위쪽으로 확장해야 합니다. Z-Direciton 탭을 선택하고 Mesh Plane 2 0.2를 입력합니다.

    z 높이 조정

    이 시뮬레이션은 2D가 될 것입니다. 동일한 프로세스에 따라 Y 방향 범위를 -0.005 0.005 로 설정하십시오. 그리고 합계 셀을 1로 설정하십시오.

    y 메쉬 평면 조정

    최종 Mesh는 그래픽 디스플레이 창 바로 위의 Mesh->Flow Mesh->View 모드 드롭 다운 메뉴에서 옵션을 변경하여 다른 방식으로 볼 수 있습니다. 그리드 라인 마다 그리드 선을 표시합니다 옵션은 Mesh Plane의 옵션만 표시됩니다 Plane Mesh 및 개요 옵션은 Mesh의 범위를 보여줍니다.

    또한 솔버가 Mesh의 최종 지오메트리를 인식하는 방법은 FAVOR TM 알고리즘을 사용하여 형상 정의를 면적 분수 및 부피 분수로 변환합니다. 이렇게 하려면 즐겨 찾기아이콘을 클릭한 다음 생성을 선택하십시오.

    호의

    잠시 후 회색 영역이 고체 물질을 나타내는 아래와 같은 형상을 표시해야 합니다.

    선호하는 결과

    추가 정보

    l Mesh에 대한 자세한 내용은 모델 설정 장의 Mesh 섹션을 참조하십시오.

    l FAVORTM FAVORize
    옵션에 대한 자세한 내용은 모델 설정 즐겨 찾기장의 Reviewing the FAVORized Geometry and Mesh 섹션을 참조하십시오.

    경계 조건

    FLOW-3D는 구성 요소 유형 및 활성 물리적 모델에 기초한 구성 요소에 적절한 경계 조건을 자동으로 적용합니다. 그러나 경계 조건 위젯에서 Mesh 블록면의 경계 조건은 각 Mesh 블록에 대해 수동으로 설정해야 합니다(f3d_bc_icon ).

    이 매뉴얼의 경우 경계 조건 중 3 가지가 경계조건( X Min , X Max, Z Max 경계)을 기본 대칭 조건조건부터 변경해야 합니다.

    ·X Min :

    o경계 조건 위젯의 경계 섹션 아래에 있는 X Min 목록을 클릭하십시오. Type에서 경계 유형을 Velocity로 설정하고 X 속도에 대해 0.001을 입력하십시오.

    XMIN 경계 조건

    ·다음으로, 유체 분율 사용에서 유체 표고 사용으로 드롭다운 상자를 변경하고 유체 높이를 0.15로 설정하십시오.

    ·마지막으로 온도를 298K로 설정하십시오.

    XMIN 경계 조건

    ·
    X Max :

    o경계 조건 위젯의 경계 섹션 아래에 있는 X 최대 목록을 클릭하십시오. 경계 유형을 압력으로 설정하고 압력에 대해 0을 입력하십시오.

    o다음으로, 유체 분율 사용에서 유체 높이 사용으로 드롭다운 상자를 변경하고 유체 높이를 0.15로 설정하십시오.

    o마지막으로 온도를 298K로 맞춥니다.

    oXMAX 경계 조건

    ·
    Z 최대 :

    o경계 조건 위젯의 경계 섹션 아래에 있는 Z 최대 목록을 클릭하십시오. 경계 유형을 압력으로 설정하고 압력에 대해 0을 입력하십시오.

    o다음으로 유체 분율을 0.0으로 설정하십시오.

    o마지막으로 온도를 298K로 맞춘다.

    ZMAX 경계 조건

    추가 정보

    자세한 내용은 모델 설정 장의 Mesh 경계 조건 섹션을 참조하십시오.

    초기 조건

    도메인 내부의 솔리드 객체(구성 요소)와 유체 모두에 대해 초기 조건을 설정해야 합니다.

    ·
    구성 요소 :이 시뮬레이션에서 솔리드 객체에 필요한 유일한 초기 조건은 초기 온도입니다. 이것은 각 구성 요소에 대한 위젯에 설정되어 있는 구성 요소 속성에 대해 수행한 것과 유사한 방식으로 구성 요소를 등록합니다. 구성 요소 속성을 설정할 때 이전과 동일한 방법으로 구성 요소 1의 초기 온도를 350K로 설정하고 구성 요소 2의 초기 온도를 298K로 설정하십시오.

    유체 초기 조건

    유체: 유체의 초기 조건을 설정하기 위해 조금 더 설정해야 합니다. 이 경우 유체 구성, 온도, 속도 및 압력 분포를 모두 설정해야 합니다. 유체 초기 조건은 초기 위젯을 설정하고 초기 f3d_initial_icon를 클릭하면 열립니다.

    f3d_initial_icon 아이콘을 선택한 후 유체 목록에서 압력을 선택하고 온도를 298K로 설정합니다. x, y, z 속도를 0.0으로 설정하십시오.

    유체 초기 조건

    다음으로, 높이/볼륨 목록과 유체 높이 사용 드롭다운 버튼을 선택합니다. 유체 높이를 0.15로 설정하십시오.

    유체 초기 조건 계속

    추가 정보

    자세한 내용은 모델 설정 장의 초기 조건 섹션을 참조하십시오.

    출력

    FLOW-3D 옵션에는 결과 파일에 기록될 데이터와 출력 위젯에서 발견된 빈도를 제어하는 7가지 데이터 유형이 있습니다. 출력 f3d_output_icon 아이콘을 클릭합니다.

    다른 데이터 유형은 다음과 같습니다.

    ·Restart: 모든 흐름 변수. 기본 출력 주기는 시뮬레이션 시간의 1/10입니다.

    ·Selected: 사용자가 선택한 흐름 변수 만. 기본 출력 주기는 시뮬레이션 시간의 1/100입니다.

    ·History: 하나의 변수와 시간의 변화를 보여주는 데이터. 예는 시간 단계 크기, 평균 운동 에너지, 배플에서의 유속 등을 포함합니다. 기본 출력 주기 = 시뮬레이션 시간의 1/100.

    ·Short print: hd3msg.*파일에 텍스트 진단 데이터가 기록 됩니다. 기본 출력 주기는 시뮬레이션 시간의 1/100입니다.

    ·Long print : hd3out.*파일에 텍스트 진단 데이터가 기록 됩니다. 기본 출력 주기는 시뮬레이션 시간의 1/10입니다.

    ·Solidification: 응고 모델이 활성화 된 경우에만 사용 가능합니다.

    ·FSI TSE: 변형 가능한 솔리드에 대한 추가 출력 옵션.

    일반적으로 이 시뮬레이션에는 기본 출력 속도가 적합합니다. 그러나 Selected Data의 일부 추가 구성은 유용합니다. Selected data interval 0.5로 설정한 다음 Fluid 온도, Fluid velocity, Macroscopic density Wall 온도 옆에 있는 상자를 선택합니다. 그러면 이러한 값이 0.5초마다 출력됩니다.

    출력 탭 설정

    추가 정보

    자세한 내용은 모델 설정 장의 출력 섹션을 참조하십시오.

    Numerics

    기본 Numerics 옵션은 대부분의 시뮬레이션에서 잘 작동하므로 기본 옵션에서 벗어나야 하는 충분한 이유가 없는 경우에는 현재 그대로 두는 것이 가장 좋습니다.

    이것으로 모델 설정 섹션에서 시작된 예제 문제의 설정을 마칩니다. 이제 실행할 준비가 되었으므로 전처리 및 시뮬레이션 실행의 단계에 따라 시뮬레이션을 실행하십시오.

    추가 정보

    자세한 내용은 모델 설정 장의 Numerics 옵션 섹션을 참조하십시오.

    일반 시뮬레이션 설정 점검 목록

    시뮬레이션을 설정하는 데 필요한 단계에 대한 개략적인 개요가 아래에 나와 있습니다. 이 목록은 포괄적인 목록이 아닙니다. 일반적인 단계, 고려해야 할 몇 가지 중요한 사항 및 권장되는 설정 순서를 간단히 설명하는 안내서일 뿐입니다.

    시작하기 전에

    1.물리적 문제의 다이어그램을 그리기 및 주석 달기 : 이 다이어그램에는 기하학적 치수, 유체의 위치, 관련 힘, 움직이는 물체의 속도, 관련 열 전달 메커니즘 등이 포함되어야 합니다. 완성된 다이어그램은 문제에 대한 모든 관련 엔지니어링 정보로 인한 물리적 문제에 대한 이미지여야 합니다.

    2.모델링 접근법 결정: 주석이 달린 다이어그램을 가이드로 사용하여 문제점에 접근하는 방법을 결정 : 문제가 되는 유체의 수, 혼화 가능한 경우, 하나 이상의 유체에서 방정식을 풀어야하는 경우 및 압축성이 중요한지 파악하여 시작하십시오. 그런 다음 어떤 물리적 메커니즘이 중요한지 결정하십시오. 이러한 각 옵션 (: 유체 유형, 열 전달 메커니즘 등)에 대한 관련 엔지니어링 정보를 다이어그램에 추가하십시오. 물리적 메커니즘이 포함되거나 무시된 이유를 정당화하려고 합니다. 이를 통해 시뮬레이션 프로세스 초기에 오류를 수정하는 데 시간이 거의 걸리지 않는 초기에 실수를 잡을 수 있습니다.

    3.다이어그램에 계산 영역을 그리고, 계산 영역의 가장자리에 있는 물리적 상황 설명 : 경계의 물리적 상황을 가장 잘 나타내는 경계 조건 유형을 기록합니다. 사용 가능한 경계 조건 유형이 경계의 물리적 상황에 대한 합리적인 근사치가 아닌 경우 이 경계를 다른 곳으로 이동해야 합니다.

    모델 설정 : 일반

    1.문제, 시뮬레이션의 목적, 사례 번호 등을 설명하는 메모를 추가하십시오. 메모는 향후 사용자 또는 나중에 참조할 수 있도록 설정을 설명하고 정당화하는 데 도움이 됩니다. 시뮬레이션의 목적, 분석 방법 등을 논의해야합니다.

    2.사용할 솔버와 프로세서 수를 선택하십시오.

    3.단위 시스템 선택: 소규모 문제를 모델링 할 때는 작은 단위 ( : mm-gm-msec)사용하고 규모가 큰 문제는 큰 단위 ( : SI)를 사용하십시오. 이를 통해 기계 정밀도로 인한 반올림 오류를 방지할 수 있습니다.

    4.유체 수, 인터페이스 추적 옵션 및 유량 모드를 선택하십시오. 주석이 달린 다이어그램을 이 단계의 지침으로 사용하십시오. 유체의 수는 질량, 운동량 및 에너지 보존을 관장하는 방정식이 유체 분율 f> 0″ width=”44″ height=”17″ border=”0″><!--[endif]--></span>(</span>유체<span lang= 1을 나타내는) 또는 유체 분획 f \ geq 0(유체 1 및 유체 2)이 있는 영역에서 해결되는지 여부를 나타냅니다. 인터페이스
    추적 옵션은 유체 분율의 변화가 급격한지 또는 확산되어야 하는지 여부를 정의하는 반면, 흐름 모드는 f = 0두 유체 문제에서 처리되는 영역을 정의합니다.

    5.마감 조건 정의: 시뮬레이션 종료 시점을 선택합니다. 시간, 채우기 비율 또는 기타 정상 상태 측정을 기반으로 할 수 있습니다.

    6.기존 결과에서 시뮬레이션을 다시 시작하는 방법 정의 (선택 사항): 기존 결과 파일에서 시뮬레이션을 다시 시작할 때 다시 시작 옵션이 적용됩니다. 재시작 옵션은 재시작 소스 파일에서 가져온 정보와 시뮬레이션의 초기 조건을 사용하여 재설정되는 정보를 정의합니다.

    모델 설정 : 물리

    1.주석이 달린 다이어그램을 기반으로 관련 실제 모델 활성화

    모델 설정 : 유체

    1.유체의 속성 정의 1: 주석이 달린 다이어그램을 가이드로 사용하여 활성 물리적 모델에 대한 적절한 물리적 속성을 정의하십시오.

    2.유체 2의 속성 정의 (사용하는 경우): 주석이 달린 다이어그램을 가이드로 사용하여 활성 물리적 모델에 적절한 물리적 속성을 정의하십시오.

    3.인터페이스의 속성 정의: f = 1 f = 0의 영역 사이의 인터페이스 속성을 정의하십시오. 여기에는 표면 장력, 상 변화 및 확산에 대한 특성이 포함됩니다.

    모델 설정 : Mesh 및 형상

    1.모든 STL 파일의 오류 점검: ADmesh, netfabb Studio 또는 유사한 프로그램을 사용하여 모든 STL 파일의 오류를 점검하십시오. 이는 모델 설정에 시간을 소비하기 전에 형상
    정의와 관련된 문제를 파악하는 데 도움이 됩니다.

    2.모든 하위 구성 요소 및 구성 요소 가져 오기 및 정의 : 주석이 달린 다이어그램에 설명 된 대로 실제 사례와 일치하도록 3D 솔리드 형상을 정의합니다. 최종 결과는 물리적 형상의 정확한 복제본이어야 합니다. 각 부분에 설명적인 이름을 사용하고 대량 소스가 될 구성 요소를 포함하십시오.

    3.모든 구성 요소의 속성 정의: 주석이 달린 다이어그램에 그려진 내용을 기반으로 각 구성 요소의 모든 재료 속성, 표면 속성, 모션 속성 등을 정의합니다. 경계 조건이 정의될 때까지 질량 소스 특성을 정의하기를 기다리십시오.

    4.스프링과 로프 및 각각에 대한 관련 속성을 정의합니다.

    5.주석이 달린 다이어그램에 설명된 시뮬레이션 도메인과 일치하도록 Mesh를 정의하십시오. 도메인의 모서리가 다이어그램에서 식별된 위치에 있는지 확인하십시오. 또한 인터페이스 (셀이 있는 셀과 셀이 f = 1다른 셀 이 있는 셀)를 식별하려면 세 개의 셀이 필요합니다.f = 0 ). 최소 5 개의 셀이 예상되는 가장 얇은 연속 영역에 맞도록 충분히 작은 셀을 사용하십시오. f = 1 f = 0 .

    6.지오메트리를 정의하는 모든 배플 정의

    7.경계 조건, 질량 소스, 질량 모멘텀 소스, 밸브 및 벤트 정의: 경계 조건 (질량 소스, 질량 모멘텀 소스, 밸브 및 벤트 포함)은 모든 방정식을 풀기 위해 주어진 위치에서 솔루션을 규정합니다. 주석이 달린 다이어그램을 사용하여 각 경계 (또는 소스 등)에 지정된 내용이 유동 솔루션, 열 전달 솔루션, 전위 등에 대한 현실과 일치하는지 확인하십시오.

    8.유체 및 구성 요소의 초기 조건을 정의합니다. 초기 조건은 모든 방정식 (유량 솔루션, 열 전달 솔루션, 전위 등)에 대해 모든 영역에서 솔루션을 규정합니다.t = 0 .주석이
    달린 다이어그램을 사용하여 초기 조건에 지정된 내용이 현재 현실에 대한 근사치인지 확인하십시오. 유체 영역뿐만 아니라 구성 요소의 초기 조건을 설정해야 합니다.

    9.모든 측정 장치 정의 (샘플링 볼륨, 플럭스 표면 및 히스토리 프로브)

    모델 설정 : 출력

    1.출력 기준 (시간, 채우기 비율 또는 응고된 비율)을 선택하십시오.

    2.재시작 데이터에 추가할 출력을 선택하십시오.

    3.선택한 데이터에 기록할 정보를 선택하십시오.

    4.재시작, 선택, 히스토리, 짧은 인쇄 및 긴 인쇄 데이터의 출력 속도 정의 : 기본 속도는 재시작 및 긴 인쇄 데이터의 경우 (10개 출력)/(시뮬레이션 종료 시간) 및 선택한 기록, 짧은 인쇄 데이터의 경우 (100개 출력)/(시뮬레이션 종료 시간)입니다.

    모델 설정 : 숫자

    1.기본값이 아닌 필수 숫자 옵션을 선택 FLOW-3D의 숫자 옵션은 고급 사용자를 대상으로 하며, 지배 방정식을 해결하는 데 사용되는 숫자 근사치 및 방법을 상당히 제어할 수 있습니다. 이러한 옵션 중 일부를 잘못 사용하면 솔루션에 문제가 발생할 수 있으므로 일반적으로 이 옵션의 기능을 먼저 이해하고 조정의 정당성을 갖추지 않고는 이러한 설정을 조정하지 않습니다.

    5.FLOW-3D에서 후 처리

    이 섹션에서는 FLOW-3D에 통합된 포스트 프로세서를 사용하는 방법에 대해 설명합니다. 보다 강력한 외부 포스트프로세서 FlowSight에 대한 튜토리얼은 FlowSight 설명서를 참조하십시오. 또한 이 섹션에서는 Flow Over A Weir 예제 문제를 실행하여 생성된 결과 파일을 사용합니다. 이 예제 문제를 실행하는 방법에 대한 지침은 예제 추가 및 시뮬레이션 사전 처리 및 실행을 참조합니다.

    FlowSight 사용에 대한 기본 참조는 FlowSight Help->helpLocal Help 메뉴에서 액세스하는 FlowSight 사용자 설명서입니다.

    추가 정보

    기존 플롯

    기존 플롯은 솔버가 자동으로 생성하는 사전 정의된 플롯입니다. 사용자 정의 플롯은 아래의 사용자 정의 플롯 섹션에 설명되어 있습니다.

    1.분석 탭을 클릭하십시오. FLOW-3D 결과 대화 상자가 표시됩니다; 메세지가 나타나지 않으면 (분석 탭이 열림) 결과 파일 열기를 선택하여 동일한 대화 상자를 엽니다.

    2.기존 라디오 버튼을 선택하십시오. 데이터 파일 경로 상자에 두 가지 유형의 파일이 표시됩니다 (있는 경우). 이름이 prpplt.*있는 파일 에는 전처리 flsplt.*기에 의해 자동으로 작성된 플롯이 포함되고 이름이 있는 파일에는 입력 파일에 사전 지정된 플롯 뿐만 아니라 후 처리기에 의해 자동으로 작성된 플롯이 포함됩니다.

    3. 확인을 선택 flsplt.Flow_Over_A_Weir하고 클릭하십시오. 그러면 디스플레이 탭이 자동으로 열립니다.

    기존 결과 대화 상자

    4.사용 가능한 플롯 목록이 오른쪽에 나타납니다. 목록에서 해당 플롯의 이름을 클릭하면 특정 플롯을 볼 수 있습니다. 플롯 26 이 아래에 나와 있습니다.

    기존 플롯보기

    커스텀 플롯

    1.분석 탭으로 돌아갑니다. 대화 상자를 열려면 결과 파일 열기를 선택하십시오.

    2.전체 출력 파일을 보려면 사용자 정의 단일 선택 단추를 선택하십시오. 전체 출력 파일에는 prpgrf.*파일과 파일이 포함됩니다 flsgrf.*. 시뮬레이션이 실행되었으므로 전 처리기 출력 파일이 삭제되어 flsgrf파일에 통합되었습니다.

    3.flsgrf.Flow_Over_A_Weir대화 상자 에서 파일을 선택하고 확인을 클릭하십시오.

    FLOW-3D 결과 대화 상자

    이제 분석 탭이 표시됩니다. 시뮬레이션 결과를 시각화 하는 방법에는 여러 가지가 있습니다. 사용 가능한 플롯 유형은 다음과 같습니다.

    ·Custom : 이 매뉴얼 의 FLSINP 파일을 사용하여 플롯합니다. 사용자 정의 섹션의 출력 코드를 사용하여 출력 플롯을 수동으로 수정하는 데 사용할 수 있습니다. 이것은 고급 옵션입니다.

    ·프로브 : 개별 셀, 경계, 구성 요소 및 도메인 전체(전역) 변수 대 시간에 대한 그래픽 및 텍스트 출력을 표시합니다. 자세한 내용은 프로브 플롯 프로브 : 특정 시점의 데이터와 시간 을 참조하십시오.

    ·1-D : 셀 데이터는 X, Y 또는 Z 방향의 셀 라인을 따라 볼 수 있습니다. 플롯 제한은 공간 및 시간에 모두 적용할 수 있습니다. 자세한 내용은 1-D 플롯 1-D : 라인을 따른 데이터 시간 을 참조하십시오.

    ·2-D : 셀 데이터는 XY, YZ 또는 XZ 평면에서 볼 수 있습니다. 플롯 제한은 공간 및 시간에 모두 적용할 수 있습니다. 속도 벡터 및 입자를 추가할 수 있습니다. 자세한 내용은 2 차원 플롯 2 차원 : 평면의 데이터와 시간의 데이터 를 참조하십시오.

    ·3-D : 유체와 고체의 표면 플롯을 생성하고 셀 데이터로 채색 할 수 있습니다. 속도 벡터, 입자 (있는 경우) 및 유선과 같은 추가 정보를 추가할 수 있습니다. 플롯 제한은 공간 및 시간에 모두 적용할 수 있습니다. 자세한 내용은 3D 플롯 3D : 표면의 데이터 시간 을 참조하십시오.

    ·텍스트 출력 : cell-by-cell 재시작, 선택 및 응고 데이터를 텍스트 파일에 쓸 수 있습니다. 자세한 내용은 텍스트 출력 텍스트 : ASCII 형식의 공간 데이터 출력 시간 을 참조하십시오.

    ·중립 파일 : 재시작 및 선택된 데이터는 별도의 텍스트 파일에 정의 된 지정된 지점(보간 또는 셀 중심)에서 출력 될 수 있습니다. 자세한 내용은 중립 파일 : 사용자 정의 좌표에서의 공간 데이터 출력 시간 을 참조하십시오.

    ·FSI TSE : 유한 요소 유체 / 고체 상호 작용 및 열 응력 진화 물리학 패키지에서 출력됩니다. 자세한 내용은 FSI / TSE : 표면의 구조 데이터와 시간 을 참조하십시오.

    3 차원 도표

    1.Analyze -> 3-D 탭을 선택하십시오.

    2.Iso-surface = Fraction of fluid 선택하십시오. 이것은 표면을 그리는 데 사용되는 변수입니다. 선택한 등면 변수에 대한 등고선 값 기준을 충족하는 모든 셀을 통해 표면이 그려집니다. 유체의 분율이 기본값이며 유체 표면이 표시됩니다.

    등 면형

    3.색상 변수 = 압력을 선택하십시오. 이 선택은 등위면의 색을 지정하는 데 사용되는 변수를 결정합니다 (이 경우 유체 표면은 압력에 의해 색이 그려집니다).

    색상 변수 유형

    4.Component iso-surface overlay = Solid volume 선택하십시오. 솔리드 볼륨 은 유체와 함께 솔리드 구성 요소를 표시합니다. 이전 단계에서는 체적 분수의 보완을 등위면으로 선택하여 이 작업을 수행했지만 이 옵션을 사용하면 유체와 고체 표면을 동시에 플롯 할 수 있습니다.

    등표면 옵션

    5.이동 시간 프레임의 최소 및 최대 위치들 (0 내지 1.25 )에 슬라이더 위치.

    시간대 옵션

    6.렌더 버튼을 클릭하여 디스플레이 탭으로 전환하고 t = 0.0에서 1.25 초 사이에 일련의 11 플롯을 생성하여 압력에 의해 채색된 유체 표면과 위어 구조를 보여줍니다. 데이터 다시 시작 이 선택되었으므로 11 개의 플롯이 있습니다.

    7.사용 가능한 플롯이 사용 가능한 시간 프레임 목록에 나열됩니다. 다음을 클릭하여 시간 프레임 사이를 이동하거나 시간 프레임을 두 번 클릭하여 표시하십시오. 첫 번째 및 마지막 시간 프레임은 다음과 같아야 합니다.

    위어 구조 렌더링

    8.Analyze -> 3-D 탭으로 돌아가서 Data Source 그룹에서 Selected data 라디오 버튼을 선택하십시오.

    데이터 소스

    9.시간 프레임 선택기의 두 슬라이더가 모두 오른쪽에 있으므로 마지막 시간 프레임 만 생성됩니다. 사용 가능한 시간 프레임이 많고 렌더링하는데 시간이 오래 걸리므로 선택한 데이터를 선택하면 인터페이스에서 자동으로 수행됩니다. 사용 가능한 모든 시간 프레임을 렌더링 하려면 왼쪽 슬라이더를 Time Frame Min = 0 으로 이동하십시오.

    10. 렌더링 버튼을 클릭하십시오. 몇 초 안에 뷰가 디스플레이 창으로 전환되고 101 개의 플롯이 사용 가능한 시간 프레임 목록에 나열됩니다. 시간 프레임 사이를 이동하려면 다음을 반복해서 클릭하십시오.

    대칭 흐름 표시

    위어 중심 아래로 대칭 평면을 사용하여 시뮬레이션을 설정했으므로 위어 구조의 절반만 시뮬레이션되고 표시됩니다. 프리젠테이션 목적으로 대칭 모델의 두 반쪽을 모두 표시할
    수 있습니다.

    1.아래와 같이 Analyze -> 3-D 탭으로 돌아가서 Open Symmetry Boundaries 확인란을 선택하십시오.

    열린 대칭 경계

    2.렌더링을 클릭하십시오. 유체 표면이 디스플레이 탭의 대칭 경계에서 열린 상태로 나타납니다.

    3.화면 위의 도구 모음 메뉴에서 도구 -> 대칭을 선택하십시오.

    4.대화 상자에서 Y 방향 확인란을 선택하여 Y = 0 평면에서 결과를 미러링합니다.

    대조

    5.적용 닫기를 선택하십시오.

    6.마지막 시간 프레임을 두 번 클릭하십시오. 디스플레이는 아래와 같이 전체 위어 구조를 보여줍니다.

    전체 위어 구조

    3 차원 애니메이션 만들기

    다음 단계는 3 차원 유체 표면의 애니메이션을 만드는 것입니다. 애니메이션은 사용 가능한 시간 프레임 목록의 프레임에서 만든 동영상입니다. 애니메이션의 시각적 효과를 향상시키려면 모든 프레임에 공통 색상 스케일을 적용하는 것이 좋습니다.

    1.분석 -> 3-D 탭으로 돌아갑니다.

    2.윤곽 제한 그룹 상자에서 전역 라디오 버튼을 모두 선택하십시오.

    윤곽 제한

    3.렌더 클릭 하여 다시 그리고 디스플레이 탭으로 돌아갑니다.

    4.도구 -> 대칭 -> Y 방향 -> 적용 선택을 반복하여 Y = 0 평면에서 결과를 반영합니다.

    5.선택 도구 -> 애니메이션 -> 러버 밴드 캡처를 다음과 같이 선택 확인 Mesh지가 나타납니다 그것을 읽은 후.

    러버 밴드 캡처

    6.마우스 왼쪽 버튼을 클릭 한 상태에서 드래그하여 애니메이션을 적용할 화면 부분을 선택하십시오. 선택한 영역 주위에 선택 상자가 나타납니다.

    X, Y, 너비 및 높이 상자

    7.디스플레이 창 위에서 빨간색 캡처 버튼을 선택하십시오. 애니메이션을 시작하는 대화 상자가 나타납니다.

    8.애니메이션의 기본 이름은 out.avi입니다. 아래에 표시된 것처럼 보다 구체적인 이름이 권장됩니다.

    9.기본 프레임 속도는 초당 10 프레임입니다. 이 시뮬레이션의 마감 시간은 1.25 초이고, 일정한 시간 간격으로 100 개의 플롯이 있으므로실제속도는 초당 80 프레임입니다. 너무 빠를 수 있으므로 대신 5 입력 하고 확인을 누르십시오.

    AVI 캡처

    10. 각 시간 프레임이 표시 창에 렌더링 되고 비트 맵 파일이 시뮬레이션 디렉토리에 작성됩니다. 이 프로세스가 완료되면 다음 대화 상자가 나타납니다.

    생성 된 이미지 소스 파일

    1. 프로세스의 다음 단계를 시작하려면 확인 버튼을 클릭하십시오. 새로운 프로세스 (BMP2VAI.exe)가 시작되고 압축 방법을 선택할 수 있는 새로운 비디오 압축 창이 나타납니다. 다른 창 뒤에 숨겨져 있으면 앞으로 가져옵니다.
    2. 애니메이션의 기본 압축은 압축되지 않습니다. 파일 크기가 너무 커서 뷰어에 로드 할 수 없으므로 대부분의 애니메이션에는 권장되지 않습니다. Windows를 사용하는 경우 Microsoft Video 1, Linux를 사용하는 경우 Cinepak 선택하십시오. 여기에서 선택하는 것은 컴퓨터에서 사용할 수 있는 비디오 코덱과 비디오를 표시하는 데 사용하는 기계에서 사용할 수 있는 것입니다.
    3. 애니메이션 속도가 데이터 속도에 의해 제한되지 않도록 데이터 속도 확인란을 선택 취소하십시오.
    비디오 압축

    1. 압축 프로세스를 시작하려면 확인을 클릭하십시오. 압축이 완료되면 다음 대화 상자가 나타납니다.
    AVI 파일 생성

    1. 확인을 클릭하십시오. 애니메이션 프로세스가 완료되었습니다.
    2. Windows 탐색기에서 .avi 파일을 찾는 가장 빠른 방법 은 시뮬레이션 관리자 탭으로 이동하여 시뮬레이션 입력 파일 링크를 클릭하는 것 입니다.
    3. .avi파일 을 두 번 클릭하여 애니메이션을 재생 하십시오. 이전에 선택한 압축 형식을 읽을 수 있는 올바른 코덱이 설치되어 있지 않으면 오픈 소스 다중 코덱 비디오 플레이어 설치를 고려하십시오.

    2 차원 도표

    1.Analyze -> 2-D 탭을 선택하십시오. 이 시뮬레이션의 결과를 보는 데 가장 유용한 평면은 평면 Y = 0.0에있는 위어 중심선의 XZ 평면입니다.

    2.XZ 평면 라디오 버튼을 선택하십시오.

    3.Y 제한 슬라이더를 모두 Y = 0.25 (Y = 0.0에 가장 가까운 셀 중심 y 좌표)로 드래그 합니다. 또한 동일한 위치가 J = 2 로 식별되어 해당 셀이 도메인에서 두 번째임을 나타냅니다. 첫 번째 셀 (J = 1) Mesh 외부에 있으며 경계
    조건 속성을 계산하는 데 사용됩니다. 기본
    윤곽 변수는 압력이며 기본 속도 벡터는 기본적으로 선택됩니다. 솔리드 형상은 모든 2D 플롯과 함께 자동으로 표시되므로 3D 플롯과 같이 활성화 할 필요가 없습니다.

    4.벡터 옵션을 클릭하고 X = 2 Z = 2 입력하십시오. 벡터는 이제 다른 모든 셀에 플롯 됩니다. 벡터 옵션을 적용하려면 확인을 선택하십시오.

    벡터 옵션

    5.Y = 0 평면에서 2 차원 압력 플롯의 시간 시퀀스를 생성하려면 렌더링을 클릭하십시오. T = 0.0 (왼쪽) 인 다음과 유사한 그래픽이 나타납니다. T = 0.125 (중간); 그리고 T = 1.25 (오른쪽).

    2D 결과

    6.디스플레이 화면의 오른쪽 상단에 있는 형식 버튼을 선택하십시오.

    형식 옵션

    7.선 색상, 벡터 길이 및 화살촉 크기 변경과 같은 다양한 옵션을 시험해보십시오. 변경 사항을 보려면 적용을 선택하십시오. 완료되면 재설정 확인을 선택하여
    기본 설정으로 돌아가서 대화 상자를 닫습니다. 모든 플롯에 대해 선호하는 옵션 세트가 있는 경우
    저장 버튼을 선택하여 저장할 수 있습니다.

    1 차원 도표

    1. 분석 -> 1-D 탭을 선택하십시오. 이 탭에서는 하나 이상의 플롯 시간에서 셀 행을 따라 압력, 유체 깊이, 유체 상승 및 속도와 같은 셀별 출력 변수의 꺾은 선형
      차트 플롯을 사용할 수 있습니다.
    2. 데이터 소스 로 선택을 선택합니다. 사용 가능한 변수는 이제 더 빈번한 플로팅을 위해 선택된 변수 만 표시합니다.
    3. 자유 변수 표고데이터 변수 로 선택하십시오. 유압 데이터출력 탭에서 선택되었으므로 사용할 수 있습니다.
    ID 그래픽을 위해 선택된 데이터

    1. 이 시뮬레이션의 흐름 방향은 주로 x 축과 평행하므로 X 방향을 선택하십시오.
    2. Y 방향 슬라이더를 0.25(J = 2)로 이동하여 Y 방향에서 흐름 중심선에 가장 가까운 셀이 표시됩니다.
    3. 기본적으로 전체 X 범위가 표시됩니다. 플롯의 범위를 제한하려는 경우 X 방향 슬라이더를 이동할 수 있습니다. Z 방향 슬라이더의 위치는 주어진 x, y 위치에서 z 셀의 각 열에 대해 하나의 자유 표면 높이만 기록되므로 중요하지 않습니다. 시간 프레임 슬라이더는 0초와 1.25초여야 합니다.
    흘러가는 방향

    1. 렌더링을 클릭하십시오. t = 0.0에서 t = 1.25s까지의 시리즈 플롯이 디스플레이 탭의 플롯 목록에 나열됩니다. 이러한 플롯을 볼 수 있는 여러 가지 모드가
      있습니다. 기본 모드는
      단일 모드이며 형식 버튼 아래의 드롭 다운 상자에 표시됩니다.
    기본 단일 모드

    1. 다양한 시간에 유체 표면 높이의 플롯을 비교하려면 드롭 다운 상자에서 오버레이 모드를 선택하십시오.
    2. 오른쪽 창에서 플롯 1, 13 101 선택하려면 클릭하십시오. 플롯 이름에는 또한 기록된 시간이 표시됩니다 (t = 0.0, 0.15s 1.25 ). 출력은 아래와 같이 나타납니다.
    자유 표면 고도

    1. 이 플롯을 비트 맵 또는 포스트 스크립트 파일에 저장하려면 출력 버튼을 선택하십시오.
    2. 확인 화면에 플롯 오버레이 플롯을 캡처하는 확인란을 (그리고 단 하나의 출력 파일을).
    3. 쓰기 버튼을 선택하여 이미지 파일을 만듭니다.
    4. 결과 이미지 파일은 시뮬레이션 디렉토리에 있으며 (시뮬레이션 관리자 탭 에서이 파일을 찾는 방법을 기억하십시오) 이름이 지정한 plots_on_screen.bmp됩니다.
    출력 사진

    프로브 플롯

    1.
    분석 -> 프로브 탭을 선택하십시오. 시간 기록 플롯은이 탭에서 변수 대 시간의 라인 그래프 또는 텍스트 출력으로 생성됩니다. FLOW-3D 에는 데이터 소스 그룹에서 선택되는 세 가지 유형의 시간 종속 데이터가 있습니다.

    ·공간 데이터 : 재시작 선택된 데이터 소스. 단일 x, y, z 셀 중심 좌표의 시간 종속 값이 표시됩니다. 값은 시간과 관련하여 통합되거나 시간과 관련하여 차별화되거나 이동 평균 (시간)으로 통합될 수 있습니다.

    ·일반 history 데이터 :. 글로벌 수량은 시간에 따라 다릅니다. 일반적인 양은 평균 운동 에너지, 시간 단계 및 대류 볼륨 오류입니다. 또한 이 데이터 유형에는 모델 설정 -> 메싱 및 지오메트리 탭에서 이러한 옵션을 선택한 경우 지정된 측정 위치(배플, 샘플링 볼륨, 히스토리 프로브)의 모든 데이터와 이동 또는 정지 상태의 솔리드 및 스프링/로프를
    위한 통합 출력이 포함됩니다.

    ·Mesh-dependent data : 메쉬 경계에서 시간에 따른 수량(계산 또는 사용자 지정)입니다. 일반적인 수량은 경계에서의 유량 및 경계에서의 지정된 유체 높이입니다.

    2.데이터 원본에서 일반 기록 라디오 버튼을 선택합니다. X, Y Z 데이터 점 슬라이더가 회색으로 바뀝니다. 이는 일반 기록 데이터가 특정 셀과 연결되어 있지 않기 때문입니다.

    3.목록에서 질량  평균 유체 평균 운동 에너지를 선택하십시오.

    그래픽 데이터 출력

    4. 단위를 선택하여 플로팅 단위 대화 상자를 엽니다.

    5. 플롯에 단위 표시를 선택하십시오.

    6. SI, CGS, slugs/feet/seconds 또는 pounds/inches/seconds를 선택하여 원하는 단위 시스템으로 결과를 변환하고 출력합니다. 장치를 표시하고 변환하려면 모델 설정 -> 일반 탭에서 장치 시스템을 선택해야 합니다. 이전 단계에서 이 항목을 확인했으며, 지오메트리 및 유체 특성은 centimeters/grams/seconds 시스템에서 지정되었습니다.

    플로팅 단위

    7.Plotting Units 대화 상자를 닫으려면 OK를 선택하십시오.

    8.데이터의 그래픽 출력을 생성하려면 렌더를 선택하십시오. 출력은 시간에 따른 영역의 모든 유체에 대한 질량 평균 평균 운동 에너지를 보여줍니다. 이전 단계에서 선택한 사항에 따라 단위 레이블과 함께 그림이 나타납니다. 플롯은 총 운동 에너지가 일부 평균값 주위에서 진동하고 있음을 나타냅니다. 진동이 작아짐에 따라 시뮬레이션은 정상 상태 흐름에 접근합니다.

    프로브 MKE 출력

    9.분석 -> 프로브 탭으로 돌아갑니다.

    10. 출력 양식 그룹에서 텍스트를 선택하여 그래프를 텍스트 데이터로 출력한 다음 렌더링을 다시 선택하십시오.

    출력 형태

    11. 나타나는 텍스트 대화 상자에서 다른 이름으로 저장 버튼을 선택하여 출력을 텍스트 파일로 저장할 수 있습니다.

    12. 출력 창을 닫으려면 계속을 선택하십시오.

    텍스트 출력

    1.Analyze -> Text Output 탭을 선택하십시오.

    2.텍스트 출력 은 셀별 데이터 ( 다시 시작 또는 선택됨 ) 만 출력 할 수 있고 (구성 요소, 측정 스테이션 또는 글로벌 데이터 없음) 둘 이상의 셀을 선택할 수 있다는 점을 제외하고 프로브 탭 과 동일한 방식으로 작동합니다. 각 플롯 시간에 대한 출력 데이터. 셀은 슬라이더를 사용하여 3D 블록에서 선택됩니다. 기본 공간 범위는 전체 도메인으로 설정됩니다.

    3.직접 텍스트 데이터를 출력해보십시오.

     

    FLOW-3D TruVOF는 미국 및 기타 국가에서 등록 상표입니다.

    Coating field – Roll Coating

    Roll Coating (롤 코팅)

    • 응용
      – 접착제
      – 밀폐제
      – 섬유 산업
    • 공정 파라미터
      – 롤 속도
      – 기질 속도
      – 유동학
    • 품질 관리
      – 코팅 두께
      – 결함 최소화

    • 손쉬운 설정의 시뮬레이션
      – STL 가져 오기 또는 기본 요소로 생성
      – 간단한 직사각형 격자

    롤의 속도가 코팅에 미치는 영향

    • 전형적으로 유입구에 코팅액이 적당하게 있는 상황
    • 롤의 회전이 역으로 작동하는 상황
    • 유입구에 코팅액이 적게 들어오는 상황

    공기가 유입된 롤 코팅


    FLOW-3D What’s New Ver.12.0

    FLOW-3D v12는 그래픽 사용자 인터페이스 (GUI)의 설계 및 기능에서 매우 큰 변화를 이룬 제품으로 모델 설정을 단순화하고 사용자 워크 플로를 향상시킵니다. 최첨단 Immersed Boundary Method(침수경계 방법)은 FLOW-3D v12 솔루션의 정확성을 높여줍니다. 다른 주요 기능으로는 슬러지 침강 모델, 2-Fluid 2-Temperature 모델 및 Steady State Accelerator가 있으며,이를 통해 사용자는 자유 표면 흐름을 더욱 빠르게 모델링 할 수 있습니다.

    Physical and Numerical Model

    Immersed boundary method

    힘과 에너지 손실에 대한 정확한 예측은 고체 주위의 흐름과 관련된 많은 엔지니어링 문제를 모델링하는 데 중요합니다. 새 릴리스 FLOW-3D v12에는 이러한 문제점 해결을 위해 설계된 새로운 고스트 셀 기반 Immersed Boundary Method (IBM)가 있습니다. IBM은 내 외부 흐름 해석을 위해, 벽 근처에서 보다 정확한 해를 제공하여 드래그 앤 리프트 힘의 계산을 향상시킵니다.힘과 에너지 손실의 정확한 예측은 고체 주위의 흐름을 포함하는 많은 공학적 문제를 모델링 하는데 중요합니다.

    Two-field temperature for the two-fluid model

    2 유체 열전달 모델은 각 유체에 대한 에너지 전달 방정식을 분리하기 위해 확장되었습니다. 각 유체는 이제 자체 온도 변수를 가지므로 인터페이스 근처의 열 및 물질 전달 솔루션의 정확도가 향상됩니다. 인터페이스에서의 열전달은 이제 시간의 표 함수가 될 수 있는 사용자 정의 열전달 계수에 의해 제어됩니다.

    블로그 보기

    Sludge settling model

    새로운 슬러지 정착 모델은 수처리 애플리케이션에 부가되어 사용자들이 수 처리 탱크와 클래리퍼의 고형 폐기물 역학을 모델링 할 수 있게 해 줍니다. 침전 속도가 분산상의 액적 크기의 함수 인 드리프트-플럭스 모델과 달리, 침전 속도는 슬러지 농도의 함수이며 기능 및 표 형식으로 입력 할 수 있습니다.

    개발노트 읽기

    Steady-state accelerator for free surface flows

    이름에서 알 수 있듯이 정상 상태 가속기는 정상 상태 솔루션에 대한 접근을 빠르게합니다.
    이것은 작은 진폭 중력과 모세관 표면파를 감쇠시킴으로써 달성되며 자유 표면 흐름에만 적용 할 수 있습니다.

    개발노트 읽기

    Void particles

    Void particles 가 기포 및 상 변화 모델에 추가되었습니다. Void particles는 붕괴 된 Void 영역을 나타내며, 항력 및 압력을 통해 유체와 상호 작용하는 작은 기포로 작용합니다. 주변 유체 압력에 따라 크기가 변하고 시뮬레이션이 끝날 때의 최종 위치는 공기 유입 가능성을 나타냅니다.

    Sediment scour model

    퇴적물 수송 및 침식 모델은 정확성과 안정성을 향상시키기 위해 정비되었습니다. 특히 퇴적물 종의 질량 보존이 크게 개선되었습니다.

    개발 노트 읽기>

    Outflow pressure boundary condition

    고정 압력 경계 조건에는 압력 및 유체 분율을 제외한 모든 유량이 해당 경계의 상류의 유량 조건을 반영하는 ‘유출’옵션이 포함됩니다. 유출 압력 경계 조건은 고정 압력 및 연속 경계 조건의 하이브리드입니다.

    Moving particle sources

    시뮬레이션 중에 입자 소스를 이동할 수 있습니다. 시간에 따른 병진 및 회전 속도는 표 형식으로 정의됩니다. 입자 소스의 운동은 소스에서 방출 된 입자의 초기 속도에 추가됩니다.

    Variable center of gravity

    기변 무게중심은 중력 및 비관 성 기준 프레임 모델에서, 시간의 함수로서 무게 중심의 위치는 외부 파일에서 테이블로서 정의 될 수있다. 이 기능은 연료를 소비하고 분리 단계를 수행하는 로켓과 같은 모형을 모델링 할 때 유용합니다.

    공기 유입 모델

    가장 간단한 부피 기반 공기 유입 모델 옵션이 기존 질량 기반 모델로 대체되었습니다. 질량 기반 모델은 부피와 달리 주변 유체 압력에 따라 부피가 변화하는 동안 흡입된 공기량이 보존되기 때문에 물리학적 모델입니다.

    Tracer diffusion

    유동 표면에서 생성된 추적 물질은 분자 및 난류 확산 과정에 의해 확산될 수 있으며, 예를 들어 실제 오염 물질의 동작을 모방한다.

    Model Setup

    Simulation units

    온도를 포함하여 단위 시스템은 완전히 정의해야하는데 표준 단위 시스템이 제공됩니다. 또한 사용자는 다양한 옵션 중에서 질량, 시간 및 길이 단위를 정의 할 수 있으므로 사용자 정의가 가능한 편리한 단위를 사용할 수 있습니다. 사용자는 압력이 게이지 또는 절대 단위로 정의되는지 여부도 지정해야합니다. 기본 시뮬레이션 단위는 기본 설정에서 설정할 수 있습니다. 단위를 완전히 정의하면 FLOW-3D 가 물리량의 기본값을 정의하고 범용 상수를 설정하여 사용자가 요구하는 작업량을 최소화 할 수 있습니다.

    Shallow water model

    Manning’s roughness in shallow water model

    Manning의 거칠기 계수는 지형 표면의 전단 응력 평가를 위해 천수(shallow water) 모델에서 구현되었습니다. 표면 결함의 크기를 기반으로 기존 거칠기 모델을 보완하며 이 모델과 함께 사용할 수 있습니다. 표준 거칠기와 마찬가지로 매닝 계수는 구성 요소 또는 하위 구성 요소의 속성이거나 지형 래스터 데이터 세트에서 가져올 수 있습니다.

    Mesh generation

    하단 및 상단 경계 좌표의 정의만으로 수직 방향의 메시 설정이 단순화되었습니다.

    Component transformations

    사용자는 이제 여러 하위 구성 요소로 구성된 구성 요소에 회전, 변환 및 스케일링 변환을 적용하여 복잡한 형상 어셈블리 설정 프로세스를 단순화 할 수 있습니다. GMO (General Moving Object) 구성 요소의 경우, 이러한 변환을 구성 요소의 대칭 축과 정렬되도록 신체에 맞는 좌표계에 적용 할 수 있습니다.

    Changing the number of threads at runtime

    시뮬레이션 중에 솔버가 사용하는 스레드 수를 변경하는 기능이 런타임 옵션 대화 상자에 추가되어 사용 가능한 스레드를 추가하거나 다른 태스크에 자원이 필요한 경우 스레드 수를 줄일 수 있습니다.

    Probe-controlled heat sources

    활성 시뮬레이션 제어가 형상 구성 요소와 관련된 heat sources로 확장되었습니다. 히스토리 프로브로 열 방출을 제어 할 수 있습니다.

    Time-dependent temperature at sources     

    질량 및 질량 / 운동량 소스의 유체 온도는 이제 테이블 입력을 사용하여 시간의 함수로 정의 할 수 있습니다.

    Emissivity coefficients

    공극으로의 복사 열 전달을위한 방사율 계수는 이제 사용자가 방사율과 스테판-볼츠만 상수를 지정하도록 요구하지 않고 직접 정의됩니다. 후자는 이제 단위 시스템을 기반으로 솔버에 의해 자동으로 설정됩니다.

    Output

    • 등속 필드 솔버 옵션을 사용할 때 유량 속도를 선택한 데이터 로 출력 할 수 있습니다 .
    • 벽 접착력으로 인한 지오메트리 구성 요소의 토크 는 기존 벽 접착력의 출력 외에도 일반 이력 데이터에 별도의 수량으로 출력됩니다.
    • 난류 모델 출력이 요청 될 때 난류 에너지 및 소산과 함께 전단 속도 및 y +가 선택된 데이터로 자동 출력됩니다 .
    • 공기 유입 모델 출력에 몇 가지 수량이 추가되었습니다. 자유 표면을 포함하는 모든 셀에서 혼입 된 공기 및 빠져 나가는 공기의 체적 플럭스가 재시작 및 선택된 데이터로 출력되어 사용자에게 공기가 혼입 및 탈선되는 위치 및 시간에 대한 자세한 정보를 제공합니다. 전체 계산 영역 및 각 샘플링 볼륨 에 대해이 두 수량의 시간 및 공간 통합 등가물 이 일반 히스토리 로 출력됩니다.
    • 솔버의 출력 파일 flsgrf 의 최종 크기 는 시뮬레이션이 끝날 때보 고됩니다.
    • 2 유체 시뮬레이션의 경우, 기존의 출력 수량 유체 체류 시간 및 유체 가 이동 한 거리는 이제 유체 # 1 및 # 2와 유체의 혼합물에 대해 별도로 계산됩니다.
    • 질량 입자의 경우 각 종의 총 부피와 질량이 계산되어 전체 계산 영역, 샘플링 볼륨 및 플럭스 표면에 대한 일반 히스토리 로 출력되어 입자 종 수에 대한 현재 출력을 보완합니다.
    • 예를 들어 사용자가 가스 미순환을 식별하고 연료 탱크의 환기 시스템을 설계하는 데 도움이 되도록 마지막 국부적 가스 압력이 옵션 출력량으로 추가되었습니다. 이 양은 유체가 채워지기 전에 셀의 마지막 간극 압력을 기록하며, 단열 버블 모델과 함께 사용됩니다.

    New Customizable Source Routines

    사용자 정의 가능한 새로운 소스 루틴이 추가되었으며 사용자의 개발 환경에서 액세스 할 수 있습니다.

    소스 루틴 이름설명
    cav_prod_cal캐비 테이션 생산 및 확산 속도
    sldg_uset슬러지 정착 속도
    phchg_mass_flux증발 및 응축에 의한 질량 흐름
    flhtccl유체#1과#2사이의 열 전달 계수
    dsize_cal2상 유동에서의 동적 낙하 크기 모델의 충돌 및 이탈율
    elstc_custom.점탄성 유체에 대한 응력 방정식의 소스 용어

    Brand New User Interface

    FLOW-3D의 사용자 인터페이스가 완전히 재설계되어 사용자의 작업 흐름을 획기적으로 간소화하는 최신의 타일 구조를 제공합니다.

    Dock widgets 설정

    Physics, Fluids, Mesh 및 FAVOR ™를 포함한 모든 설정 작업이 형상 창 주위의 dock widgets으로 변환되어 모델 설정을 단일 탭으로 압축 할 수 있습니다. 이 전환을 통해 이전 버전의 복잡한 트리가 훨씬 깔끔하고 효율적인 메뉴 표시로 바뀌어 모델 설정 탭을 떠나지 않고도 모든 매개 변수에 쉽게 액세스 할 수 있습니다.

    New Model Setup icons
    With our new Model Setup design comes new icons, representing each step of the setup process.
    New Physics icons
    Our Physics icons are designed to be easily differentiated from one another at a glance, while providing clear visual representation of each model’s purpose and use.

    RSS feed

    새 RSS 피드부터 FLOW-3D v12.0 의 시뮬레이션 관리자 탭이 개선되었습니다 . FLOW-3D 를 시작하면 사용자에게 Flow Science의 최신 뉴스, 이벤트 및 블로그 게시물이 표시됩니다.

    Configurable simulation monitor

    시뮬레이션을 실행할 때 중요한 작업은 모니터링입니다. FLOW-3Dv12.0에서는 사용자가 시뮬레이션을 더 잘 모니터링할 수 있도록 Simulation Manager의 플로팅 기능이 향상되었습니다. 사용자는 시뮬레이션 런타임 그래프를 통해 모니터링할 사용 가능한 모든 일반 기록 데이터 변수를 선택하고 각 그래프에 여러 변수를 추가할 수 있습니다. 이제 런타임에서 사용할 수 있는 일반 기록 데이터는 다음과 같습니다.

    • 최소/최대 유체 온도
    • 프로브 위치의 온도
    • 유동 표면 위치에서의 유량
    • 시뮬레이션 진단(예:시간 단계, 안정성 한계)
    Runtime plots of the flow rate at the gates of the large dam / Large dam with flux surfaces at the gates

    Conforming mesh visualization

    사용자는 이제 새로운 FAVOR ™ 독 위젯을 통해 적합한 메쉬 블록을 시각화 할 수 있습니다 .

    Large raster and STL data

    데이터를 처리하는 데 걸리는 시간으로 인해 큰 형상 데이터를 처리하는 것은 어려울 수 있습니다. 대형 지오메트리 데이터를 처리하는 데 여전히 상당한 시간이 소요될 수 있지만 FLOW-3D는 이제 이러한 대형 데이터 세트를 백그라운드 작업으로로드하여 사용자가 데이터를 처리하는 동안 완벽하게 응답하고 중단없는 인터페이스에서 계속 작업 할 수 있습니다.

    Permanent Mold Workspace | FLOW-3D CAST

    영구 금형 주조의 장점

    • 높은 생산률에 적합
    • 모래에 비해 복잡한 금형에 용이하고 표면 조도 및 치수 정확도가 높음
    • 재료 보존으로 인한 수율 향상 및 금형 관련 결함 발생이 감소

    영구 금형의 workflow


    재료와 구성요소의 선택

    • 모든 금속 & 주형의 재질은 사용자로부터 제작될 수 있음.
    • 재질의 데이터는 갖추어짐.

    CAD → MESH

    • FLOW-3D CAST는 사용자가 만든 stl파일에 알맞게 쉽고 자동으로 격자를 생성해줌.
    • FAVOR = Fractional Area-Volume Obstacle Representation
    • 격자의 성질의 조정없이 빠르고 쉽게 새로운 모형을 업로드

    응고 모델


    출력 선택 & 후처리 과정

    • 정확한 출력 변수를 정의
    • FlowSight로 고품질의 시각적 데이터를 쉽게 렌더링

    지속 가능한 건축물 LEED 인증 획득한 건축사례

    USGBC LEED4.1

    LEED (Leadership in Energy and Environmental Design)는 제 3자가 친환경 건축물 인증을 제공하는 자발적 인증 시스템입니다. LEED 자발적 참여는 리더십, 혁신, 환경 보호 및 사회적 책임을 보여 줍니다. LEED는 건물 소유자와 운영자에게 건물의 성능과 수익에 즉시 영향을 미치는데 필요한 도구를 제공하는 동시에 건물 거주자에게 건강한 실내 공간을 제공합니다.

    FLOW-3D는 보고타(콜롬비아)의 사무실 건물에서 “IEQ-Credit2–환기 증가”라는 신뢰를 얻는 데 큰 도움을 주었습니다. 이러한 인정을 받기 위해서는 실외 공기가 ASHRAE의 표준 비율인 30%를 초과한다는 것을 증명해야만 합니다. 이 건물에서 실외 공기는 태양 광선에 의해, 가열되는 지붕 위의 2개의 유리 굴뚝에 의해 발생되는 온도 차이에 의해 발생하는 열 부력의 영향으로 제공됩니다. 이것은 바람이 불지 않는 조건에서 이루어져야 합니다.

    The model has been easily created using Google SketchUp® and exported directly to FLOW-3D in STL format. The STL model was imported as STL baffles in FLOW-3D.

    외부 에너지 시뮬레이션 소프트웨어를 통해 공간 및 열 하중(또는 손실)의 초기 조건을 파악했습니다. 이 소프트웨어는 태양열 복사, 열 관성, 단열, 내부 부하, 단열, 유리 및 건물의 열 동작을 정의하는 기타 모든 매개 변수를 고려합니다. 열 하중은 FLOW-3D에 포함되어 있어 CFD시뮬레이션이 수행될 때 최종 온도 분포를 제공합니다.

    <좌측>건물 내부의 공기 흐름이 간소화됩니다.     <우측>건물 내부 온도 분포

    왼쪽 그림에서 2개의 열 굴뚝은 바람이 전혀 없는 상태에서도 실외 공기 유입을 유발합니다. 환기 효율, 시간당 공기 변화, 국소 난류, 공기 잔류 시간은 설계 프로세스에서 검사할 수 있는 변수 중 일부입니다. 오른쪽 그림에서 FLOW-3D는 건물 내부의 온도 분포를 제공하여 설계 과정에서 열적 쾌적성 평가를 수행할 수 있도록 합니다. 그릴 크기는 모든 공간에서 원하는 쾌적한 온도를 얻을 수 있도록 조정할 수 있습니다.

    외부 공기 온도와 기압은 경계 조건(흡입 및 배출 공기 그릴)에서 설정되었습니다. 나머지는 건물 내부에서 상세한 공기 이동을 분석한 것입니다. 시뮬레이션은 에너지가 정상 상태 조건이 달성될 때까지 수행되었다.

    건물 내 공기를 올바르게 분배하고 적절한 쾌적한 온도를 확보하기 위해 건물 구조와 그릴 치수를 조정해야 했습니다. 이 과정을 반복한 후 모델을 검증하여 LEED 인증 조건을 충족한 사례입니다.

    정밀주조품의 수축 결함 예측

    정밀 주조품의 수축 결함 예측

    정밀 주조 공정은 가장 오래된 주조 공정 중 하나로 기원전 4000년 이후에 보편화되었습니다. 이 과정은 용해된 금속을 소모품(왁스)패턴으로 생성된 세라믹 쉘에 주입하는 과정을 수반합니다. 일찍이 그것은 금, 은, 구리와 청동 합금으로 장신구와 우상을 만드는데 사용되었습니다.

    정밀 주조공정은 1897년 아이오와 주, 위원회 블러프스의 Barabas Frederick Philbrook이 묘사한 대로 치과의사들이 왕관과 인레이를 만들기 위해 그것을 사용하기 시작한 19세기 말 현대 산업공정으로 사용되기 시작했습니다. 1940년대에는 제2차 세계대전 당시 기존 방법으로는 형성될 수 없거나 지나치게 많은 가공이 필요한 특수 합금의 정밀 순모형 제조 기술에 대한 수요로 인해 투자 주조 공정이 증가하였습니다.

    오늘날 정밀 주조 공정은 표면 마감 및 치수 정확도가 우수하여 거의 순 형태에 가까운 철, 비철 및 초합금의 소형 산업용 부품을 생산하는데 주로 사용됩니다.

    정밀 주조 공정은 다음 네 가지 주요 단계로 구성됩니다.

    • 왁스 패턴 생성 후, 패턴 클러스터 또는 ‘트리’를 만들기 위해 게이트 시스템으로 청소 및 조립합니다.
    • 나무는 세라믹 쉘을 얻기 위해 미세 모래와 Course한 모래 입자의 슬러리로 번갈아 코팅됩니다.
    • 용기는 건조되고, 왁스를 녹이기 위해 가열되며, 강도를 높이고 주입 준비합니다.
    • 마침내 주조 합금이 용해되어 예열된 쉘에 주입됩니다. 응고 후에 쉘이 파손되어 주조 부품을 얻습니다.

    Figure 1. Solid model of the casting geometry

    정밀 주조 공정에서 얻은 부품은 많은 중요한 용도에 사용되므로 내부적인 결함이 없어야 합니다. 정밀 주조 공정에서 발생하는 주요 결함은 세라믹 포함, 균열, 변형, 플래시, 주탕불량, 수축, 슬래그 포함, 탕경계등입니다. 얻은 주조물의 품질을 예측하려면 금속-몰드 열 전달계수, 주입 온도 등 다양한 주조 공정 매개 변수의 영향을 연구해야 합니다. 즉, 쉘 두께 및 쉘 열 전달계수가 그것입니다. 현대 컴퓨터 시스템 및 시뮬레이션 소프트웨어의 출현과 함께 금형 충진 및 응고 시뮬레이션은 주조공장에서 결함을 예측하고 설계를 최적화하는데 점점 더 많이 사용되고 있습니다.

    이 연구의 주요 목적은 정밀 주조 공정에서 주요 요소인 복사 열 전달과 정밀 주조 공정에 고유한 쉘 금형이 FLOW-3D에서 효과적으로 구현될 수 있는지를 조사하는 것입니다. FLOW-3D를 사용하여 간단한 형상을 위한 정밀 주조공정의 주입 및 응고 시뮬레이션을 수행함으로써 두 구성요소의 서로 다른 효과를 조사합니다. 다양한 위치에서 얻은 온도의 수치는 문헌 [1]에보고 된 실험 결과로 검증됩니다. 복사 열 전달계수, 쉘 몰드 두께, 탕구 및 게이트의 위치에 대한 영향도 조사했습니다.

    Shell mold

    Figure 2. Shell mold

    Methodology

    현재 연구에서 사용된 계산 형상은 그림 1에 나와 있습니다. 쉘 몰드는 다음 단계를 사용하여 작성되었습니다.

    • complement 1로 형상을 FLOW-3D로 가져오고 지정된 셀 크기로 가져온 형상을 중심으로 메쉬 블록을 작성합니다.
    • “complement”유형의 component1의 첫 번째 하위 구성 요소를 만들어 하위 구성 요소 외부의 모든 항목을 메쉬의 범위까지 확고하게 만듭니다.
    • 솔리드 데이터베이스에서 이 솔리드 블록의 금형 재질 특성을 정의하십시오.
    • 솔리드 특성 GUI의 구성 요소 특성에서 “Thermal penetration depth”를 정의하는 옵션이 있습니다. 여기서 쉘 두께 값을 정의 할 수 있습니다.
    • 이제 전처리기를 실행하십시오.
    • Analyze 탭>3D 탭으로 이동 한 다음 이전 단계에서 생성 한 prpgrf 파일을 엽니다. ‘Iso-surface’와 ‘color variable’에서 “thermally active component volume”을 선택하고 “Render”을 선택하십시오.
    • Display에 이제 형상의 셸 부분 만 표시됩니다.
    • 개체 목록 (창의 왼쪽 하단)에서 “component 1″을 선택하고 “component 1″을 마우스 오른쪽 단추로 클릭 한 다음 “stl로 내보내기”를 선택하여 이 곡면을 STL 파일로 저장하십시오.
    Two mesh blocks

    Figure 3. The view of the two mesh blocks for the creation of a void with discretization

    쉘 몰드 용 STL 파일을 만든 후에, 이 파일을 component 1로 새 시뮬레이션으로 가져오고 이전에 작성한 주조 형상을 하위 구성 요소로 가져오고 유형을 ‘hole’으로 선택합니다. 쉘 몰드와 함께 주조 형상이 그림 2에 나와 있습니다. 이것은 우리의 계산 영역으로 사용됩니다. 다음은 계산 영역을 cubical/rectangular셀로 분할하기 위한 메쉬를 만드는 것입니다. 메쉬 블록을 작성하여 FLOW-3D에서 메쉬를 생성합니다. 현재의 작업을 위해 2.5mm의 고정된 셀 크기가 선택된 그림 3에 표시된 균일한 메쉬 옵션을 선택했습니다. 입력 위치 주변에 메시 블록 2가 사용되는 현재 시뮬레이션을 위해 메시 블록 2개가 생성되었습니다. 쉘과 주변 공기 사이의 30°C에서의 열 전달을 고려하여 쉘 주위에 보이드 영역이 정의됩니다. 이 영역은 ‘heat transfer type 1’이 있는 보이드 영역으로 선택되며 셸과 주변 공기 사이에 열 전달 계수 값이 지정됩니다. heat transfer type 1은 방사선을 포함한 종합 열 전달 계수가 됩니다.
    쉘 주형에 선택된 재료는 zircon이며 열 특성은 Sabau and Vishwanathan에 의해 수행된 실험에서 얻을 수 있습니다[2]. 표 1은 연구에 사용된 재료에 대해 지정된 값을 보여 줍니다.

    MATERIALPROPERTYVALUEUNIT
    Fluid –AluminiumA356 alloyDensity 2437kg/m³
    Thermal conductivity116.8W/(m K)
    Specific heat1074J/(kg K)
    Latent heat433.22kJ/m³
    Liquidus temperature6080C
    Solidus temperature552.40C
    Zircon MoldThermal conductivity1.09W/(m K)
    Specific heat* Density1.63E+06J/( m³ 

    Initial and boundary conditions used are show in Table 2.

    Mold temperature430°C
    Melt pouring temperature680°C
    Filling time7 s
    Interface heat transfer coefficient850 W/m2K
    Heat transfer coefficient between ambient and mold (radiation effect)30 -100 W/m2K

    Table 2. Initial and boundary conditions used for the simulation

    Sprue basin에 들어가는 용융물의 초기 속도와 온도는 메시 블록 2의 상단 경계에서 속도 경계 조건으로 주어집니다. 기본적으로 다른 모든 경계는 대칭 유형으로 설정됩니다.

    Results & Discussion

    Validation with reported experimental results

    Experimental and numerical comparison

    충전 및 응고 동안 냉각 곡선을 얻기 위한 실험에서 Sabuet.al[1]에 의해 선택된 네 개의 위치가 검증 목적으로 사용되었습니다. 그들은 C1, C2, S11, S12및 S21로 언급됩니다. C1과 C2지점은 주물의 플레이트의 중심에 있으며 S11, S12및 S21은 모두 쉘에 위치합니다. 이러한 위치에서의 온도 변화는 그림 4와 같습니다.
    온도 프로파일의 수치 및 실험결과의 차이가 허용한계 안에 있음을 알 수 있습니다. probe points C1과 C2의 경우, 수치와 실험 결과 사이의 차이는 응고 중에 5%, 응고 후 냉각 시 12% 이내입니다. 쉘의 점에 대한 수치 결과는 실험 결과보다 약 5% 높습니다. 이는 쉘 재료에 열 물리학적 특성을 할당할 때 발생하는 가정과 쉘 열 전달 계수의 값 때문일 수 있습니다.

    Fill sequence & solidification pattern for two different sprue locations

    2 개의 상이한 탕구 위치에서 용탕 충전 순서는 5a 및 5b에 나와 있습니다. 최종 탕구가 더 많은 splashing을 생성하므로 결함으로 이어질 수 있습니다. 탕구가 중간에 놓여지면 흐름은 보다 균일 해지고 두 주조 단면에서 비슷한 온도 분포를 보입니다. 50 % 응고 후의 온도 프로파일의 2D 도면은 두 경우 모두 그림 5c 및 5d에 나와 있습니다. 수축 위치에서 볼 때 두 탕구 위치가 결함을 일으키는 것은 분명합니다.

    Fill sequence at different time intervals when the sprue is located at one end
    Figure 5a. Fill sequence at different time intervals when the sprue is located at one end

     

    Fill sequence at different time intervals when the sprue is located in the middle
    Figure 5b. Fill sequence at different time intervals when the sprue is located in the middle

    2D temperature profile after 50% solidification when the sprue is located at one end
    Figure 5c. 2D temperature profile after 50% solidification when the sprue is located at one end
    2D temperature profile after 50% solidification when the sprue is located in the middle
    Figure 5d. 2D temperature profile after 50% solidification when the sprue is located in the middle
    Effect of shell thickness

    정밀 주조에 대한 쉘 두께의 효과를 연구하기 위해 두께가 7.2, 10, 15 및 20 mm인 주물을 선정하였습니다. 그림 6a 및 6b는 주조품의 특정 위치에서 냉각 곡선을 나타내며, 이는 C1으로 나타내고 쉘 몰드 내의 특정 위치에 있으며, 응고 중에 S11로 나타납니다. 세라믹 쉘의 두께가 7.2 mm에서 15 mm로 증가하면 냉각 속도가 감소하여 응고 시간이 길어지는 것을 볼 수 있습니다.

    Effect of shell heat transfer coefficient

    쉘 열 전달 계수는 열이 쉘 몰드의 외부 벽에서 방사선을 통해 주변 공기로 열을 방출하는 속도를 나타냅니다. 이 효과를 조사하기 위해 열 전달 계수의 값을 20에서 80W/m2K까지 다양하게 했습니다. 7a 및 7b로부터, h의 변화는 주조 재료 및 쉘의 냉각 속도에 중요한 영향을 미친다는 것을 알 수 있습니다. 열 전달 계수가 20에서 80W/m2K로 증가하면 C1에서의 응고 시간이 812 초에서 334 초 (약 44 %)로 감소되었음을 알 수 있습니다. 따라서, h의 값을 변화시키는 것은 주물의 미세 구조에 영향을 미칩니다.

    Temperature profile 1
    Figure 6a. Temperature profile at location C1 (casting) for the casting geometry where the sprue is located at one end for various shell thickness values
    Temperature profile 2
    Figure 6b. Temperature profile at location S11 (shell) for the casting geometry where the sprue is located at one end for various shell thickness values
    Temperature profile at location C1
    Figure 7a. Temperature profile at location C1 (casting) for the casting geometry where the sprue is located at one end for various heat transfer coefficient values between the shell mold & ambient
    Temperature profile at location S11
    Figure 7b. Temperature profile at location S11 (shell) for the casting geometry where the sprue is located at one end for various heat transfer coefficient values between the shell mold & ambient

    Conclusions

    정밀 주조 공정의 몰드 충진 및 응고 시뮬레이션은 FLOW-3D를 사용하여 수행되었습니다. 주조 공정에 대한 주조 매개변수의 영향을 연구하기 위해 파라메트릭 연구가 수행되었습니다. 본 연구에서 다음과 같은 결론을 도출 할 수 있습니다.

    • FLOW-3D는 멀티 캐비티 몰드의 주입 및 응고 모델링이 가능합니다. 프로브 위치의 예측 온도 프로파일은 실험 데이터의 허용오차 이내였다.
    • 쉘 두께의 경우, 두 경우 모두 셸의 임계 두께가 있으며, 그 이상으로 열 전달 특성이 역행하는 것으로 확인되었습니다. 셸 두께가 증가함에 따라 응고 시간이 임계 두께까지 증가하여 감소하기 시작했습니다. 원래 형상의 경우 임계 두께는 15~20mm인 반면 수정된 형상의 경우 10mm와 15mm 사이에 있다.
    • 쉘과 대기 사이의 열 전달 계수 h는 열 전달 특성에 가장 큰 영향을 미치는 것으로 나타났습니다. h가 20에서 80W/m2K로 4 배 증가할 때 탕구의 중심에서 응고 시간이 40 % 이상 감소했습니다.

    References

    Sabau, A.S., Numerical Simulation of the Investment Casting Process, Transactions of the American Foundry Society, vol. 113, Paper No. 05-160, 2005.

    Sabau, A.S., and Viswanathan, S., Thermophysical Properties of Zircon and Fused Silica-based Shells used in the Investment Casting ProcessTransactions of the American Foundry Society, vol. 112, Paper No. 04-081, 2004.

    FLOW-3D Home

    

    수치해석(CFD)이 필요하십니까? 아마 FLOW-3D 는 귀하가 찾으시는 분야에 가장 적합한 최적의 수치해석 소프트웨어일 것입니다.
    천천히 당사의 홈페이지 내용을 살펴보시면 FLOW-3D 의 기술적인 강점과 어떤 분야에 어떻게 적용하여 효과를 볼 수 있는지 알 수 있습니다.FLOW-3D 는 범용 3 차원 수치해석(CFD) 소프트웨어로, 특히 자유표면(자유수면)을 가진 유동흐름을 정확하게 예측하는 분야에서는 타의 추종을 불허하는 성능을 자랑합니다. FLOW-3D 는 수 많은 물리적 유동현상에 대한 시뮬레이션 모델을 제공하여 설계 및 운영단계에서 엔지니어에게 귀중한 통찰력을 제공할 수 있는 세계적인 CFD 소프트웨어입니다.FLOW-3D 는 해석에 필요한 모든 기능을 제공하는 풀 패키지 소프트웨어로, 격자 및 결과 분석에 추가 비용이 필요 없습니다.
    또한 기존의 CAD 시스템이나 타 소프트웨어에서 생성된 모델데이터를 STL로 읽을 수 있기 때문에 기존 모델데이터를 쉽게 활용할 수 있습니다.
    FLOW-3D 의 가장 큰 특징은 단 몇번의 조작만으로 격자를 생성할 수 있으며, 초보자도 쉽게 시작할 수 있고, 매우 뛰어난 정확성을 가지고 있다는 점 입니다.

    FLOW-3DFDM (Finite Difference Method : 유한차분법)에 따라 비정상 흐름을 해석하는범용 3 차원 CFD 소프트웨어로, 비압축성 및 압축성을 고려한 2차원 / 3차원 열 유동 문제, 상 변화, 다양한 점성 현상 및 유체–구조 연성 등의 다중 물리 문제를 해석할 수 있으며, 특히 자유표면 및 두 유체사이 계면에 대한 고속 · 고정밀도 해석에 탁월합니다.

    수치해석과 관련하여 궁금하신 사항은 언제든지 부담없이 문의 해주십시오.
    감사합니다.

    FLOW-3D 제품 안내
    FLOW-3D 제품 안내

    FLOW-3D MP버전 안내
    FLOW-3D MP버전 안내

    FLOW-3D Cast 버전 안내
    FLOW-3D Cast 버전 안내

    Technical Resource

    FLOW-3D 당사의 주력 제품으로 강력하고 매우 정확한 다중 물리 전산 유체역학(CFD) 패키지 프로그램…
    Read More >>
    FLOW-3D/MP 는 매우 큰 영역 또는 긴 runtime 문제를 해결하기 위해 고성능 컴퓨팅을 사용할 수…
    Read More >>
    FLOW-3D Cast 는 다양한 주조 공정의 충전 및 응고, 결함 분포 예측이 가능한 3차원 유동해석 프로그램…
    Read More >>
    FLOW-3D 제품에 대한 기술자료와 이론 및 논문 등 다양한 기술자료를 제공합…
    Read More >>
    신규소식 기술자료

    Fig. 9. Rear view at t ¼ 240 s and Q ¼ 1,300 L=s.

    Numerical Model for a Nineteenth-Century Hydrometric Module

    Fig. 9. Rear view at t ¼ 240 s and Q ¼ 1,300 L=s. 이 소개자료는 "Numerical Model for a Nineteenth-Century Hydrometric Module"논문의 소개자료임. 연구 목적 본 연구는 19세기에 건설된 ...

    FLOW-3D WELD/AM 2025R1 무료 웨비나

    다가오는 4월 30일(수) 오후 2시부터 3시까지, 최신 버전인 FLOW-3D WELD/FLOW-3D AM 2025R1을 소개하는 무료 웨비나가 진행됩니다. 이번 웨비나에서는✅ FLOW-3D WELD와 FLOW-3D AM의 기능 및 적용 사례✅ 최신 기능 및 개선 ...
    한국해양과학기술협의회 공동학술대회

    2025 한국해양과학기술협의회 공동학술대회

    한국해양과학기술협의회 공동학술대회 2025년도 한국해양과학기술협의회 공동학술대회가 2025.05.08.~2025.05.10. 부산 BEXCO에서 개최됩니다. 명 칭: 2025년도 한국해양과학기술협의회 공동학술대회 일 시: 2025년 5월 8일(목) - 10일(토) 장 소: 부산 BEXCO 주 최: 한국해양과학기술협의회 주 관: ...
    Figure 13 | 3D illustration of Fr variation in the old stilling basin at (a) 129.10 m tailwater, (b) 129.70 m tailwater, and (c) 130.30 m tailwater. In the new stilling basin at (d) 129.10 m tailwater, (e) 129.70 m tailwater, and (f) 130.30 m tailwater

    Hydraulic investigation of stilling basins of the barrage before and after remodelling using FLOW-3D

    Figure 13 | 3D illustration of Fr variation in the old stilling basin at (a) 129.10 m tailwater, (b) 129.70 m tailwater, and (c) 130.30 m tailwater. In the new ...
    Fig.(6) view of an 3D modeling

    Simulation of Flow on Bottom Turn out Structures with Flow 3D

    이 소개자료는 February 2014, Bulletin of Environment, Pharmacology and Life Sciences에 수록된 Simulation of Flow on Bottom Turn out Structures with Flow 3D 논문에 대한소개자료입니다. Fig.(6) view of an 3D ...
    Figure 9. Simulation results (packed sediment height net change) after the steady-state

    Sacrificial Piles as Scour Countermeasures in River Bridges: A Numerical Study

    해당 소개자료는 "Civil & Environmental Engineering and Construction Faculty Publications"에서 발표한 "Sacrificial Piles as Scour Countermeasures in River Bridges A Numerical Study using FLOW-3DNumerical Study using FLOW-3D " 논문에 대한 ...
    Figure 4.18 scour development at time = 360 min and discharge 0.057 m3/sec

    SIMULATION OF LOCAL SCOUR AROUND A GROUP OF BRIDGE PIER USING FLOW-3D SOFTWARE

    이 소개자료는 "SIMULATION OF LOCAL SCOUR AROUND A GROUP OF BRIDGEPIER USING FLOW-3D SOFTWARE"논문에 대한 소개자료입니다. Figure 4.18 scour development at time = 360 min and discharge 0.057 m3/sec 연구 ...
    Figure 5. 3D view of scour under square tide conditions (every 300 s).

    CFD simulation of local scour in complex piers under tidal flow

    Figure 5. 3D view of scour under square tide conditions (every 300 s). 이 소개자료는 CFD simulation of local scour in complex piers under tidal flow 논문에 대한 소개자료입니다. 연구 ...
    Figure 10 – Scour pit around pile group with different pile cap installation levels

    Numerical simulation of geotechnical effects on local scour in inclined pier group with Flow-3D software

    이 소개자료는 Water Resources Engineering Journal Spring 2022. Vol 15. Issue 52에 개제된 Numerical simulation of geotechnical effects on local scour in inclined pier group with Flow-3D software 논문에 대한 ...
    Figure 7. Modelling results of velocity magnitude of the embankment at time interval (a) 60 s, (b) 100 s, and (c) 140 s.

    A hydrodynamic model of an embankment breaching due to overtopping flow using FLOW-3D

    본 소개자료는 2021, IOP Conf. Ser.: Earth Environ. Sci. 920 012036에 발표된 A hydrodynamic model of an embankment breaching due to overtopping flow using FLOW-3D 논문에 대한 소개자료입니다. Figure 7 ...
    Figure 9: 3D maximum sour around single pier (no initial flow)

    LOCAL SCOUR ANALYSIS AROUND SINGLE PIER AND GROUP OF PIERS IN TANDEM ARRANGEMENT USING FLOW 3D

    연구 목적 본 연구는 교각 주변의 국부 세굴 현상을 예측하기 위해 FLOW-3D 소프트웨어를 사용하여 단일 교각과 직렬 배치된 다중 교각 주변의 맑은 물 세굴을 수치적으로 모의실험하고, 그 결과를 기존 실험실 ...
    Figure 6 | (a) Contaminant concentration distribution (gr/L) at 13 cm distance from channel bed; (b) contaminant concentration distribution (gr/L) at 17 cm distance from the channel bed.

    Numerical simulation of pollution transport and hydrodynamic characteristics throughthe river confluence using FLOW 3D

    이 소개 자료는 "Water Supply Vol 22 No 10"에 게재된 "Numerical simulation of pollution transport and hydrodynamic characteristics throughthe river confluence using FLOW 3D" 논문을 기반으로 작성되었습니다. Figure 6 | ...
    Graphical Abstract

    Numerical Investigation of Hydraulic Jump for Different Stilling Basins Using FLOW-3D

    FLOW-3D를 이용한 다양한 정수지(Stilling Basin)에서의 수력 도약(Hydraulic Jump) 수치적 연구 Graphical Abstract 연구 배경 및 목적 문제 정의 Taunsa Barrage(파키스탄)의 정수지는 기존의 USBR Type-III Basin을 개량한 형태로, 충격 바플(Impact Baffle)과 ...
    Figure 8. Numerical simulation results for the gate discharge test conditions, Case 1. (a) Case 1 surface velocity distribution. (b) Case 1 longitudinal velocity distribution of gate center.

    FLOW-3D Model Development for the Analysis of the Flow Characteristics of Downstream Hydraulic Structures

    이 소개자료는 Sustainability에서 발표한 FLOW-3D Model Development for the Analysis of the Flow Characteristics of Downstream Hydraulic Structures 논문에 대한 소개자료입니다. Figure 8. Numerical simulation results for the gate discharge ...
    Fig. 9 Velocity vectors and the plunging flow for Model A (Q 0.035 m3/s; l 0.685 m; w 0.141 m)

    An Investigation on Hydraulic Aspects of Rectangular Labyrinth Pooland Weir Fishway Using FLOW-3D

    본 소개 자료는 "Arabian Journal for Science and Engineering"에서 발행한 "An Investigation on Hydraulic Aspects of Rectangular Labyrinth Pool and Weir Fishway Using FLOW-3D" 논문을 기반으로 합니다. Fig. 9 Velocity ...
    Figure 7. Numerical flow3D profiles for, a) Sediment suspension Concentration profiles at initial motion, b) Sediment suspension Concentration profiles after initial motion, c) Shear stress, d) Flow velocity

    Comparison Between Numerical Flow3D Software and Laboratory Data for Sediment Incipient Motion

    FLOW-3D 소프트웨어와 실험 데이터를 이용한 퇴적물 초기 이동(Sediment Incipient Motion) 비교 연구 Figure 7. Numerical flow3D profiles for, a) Sediment suspension Concentration profiles at initial motion, b) Sediment suspension Concentration ...
    Fig. 9. Improvement Measures for Tunnel Plan

    Flow-3D 모형을 이용한 충주댐 보조여수로 계획 평가

    FLOW-3D를 이용한 충주댐 보조여수로 설치 계획 평가 및 개선안 제시 Fig. 9. Improvement Measures for Tunnel Plan 연구 배경 및 목적 문제 정의 기후변화로 인해 극한 강우 발생 빈도가 증가하고 ...
    Fig. 3. Ship wave patterns in theory (upper) and by FLOW-3D (lower)

    변수심에서의 항주파 파형 예측 및 FLOW-3D에 의한 검증

    본 소개 자료는 "Journal of the Korean Society of Civil Engineers"에서 발행한 "변수심에서의 항주파 파형 예측 및 FLOW-3D에 의한 검증"논문을 기반으로 합니다. Fig. 3. Ship wave patterns in theory (upper) ...
    Figure 4. Modeling of variant 1 with the movement of waves in the port water area

    FLOW-3D를 이용한 항만 수역 배치 설계의 타당성 분석

    본 소개 자료는 'IOP Conference Series: Materials Science and Engineering'에서 발행한 'FLOW-3D software for substantiation the layout of the port water area' 논문을 기반으로 합니다. Figure 4. Modeling of variant ...
    Figure_1._Flow_velocity_on_seawall_in_A1_modeling.

    FLOW-3D를 이용한 다양한 조건에서의 해안 방파제 유속 변화 모델링

    본 소개 자료는 'Open Journal of Marine Science'에서 발행한 'Modeling of the Changes in Flow Velocity on Seawalls under Different Conditions Using FLOW-3D Software' 논문을 기반으로 합니다. Figure_1._Flow_velocity_on_seawall_in_A1_modeling. 1. 서론 ...
    Fig. 10 Transverse scour hole profles for six cases

    FLOW-3D를 이용한 에어포일 컬러(AFC) 적용 유무에 따른 교각 주변 국부 세굴 수치 시뮬레이션

    본 소개 자료는 'Environmental Fluid Mechanics'에서 발행한 'Numerical simulation of local scour around the pier with and without airfoil collar (AFC) using FLOW-3D' 논문을 기반으로 합니다. Fig. 10 Transverse scour ...
    Fig. 6. Air core forming process display.

    FLOW-3D를 이용한 와류 침전지의 수면 프로파일 및 와류 구조 수치 시뮬레이션

    본 소개 논문은 Journal of Marine Science and Technology에서 발행한 논문 "NUMERICAL SIMULATIONS OF WATER SURFACE PROFILES AND VORTEX STRUCTURE IN A VORTEX SETTLING BASIN BY USING FLOW-3D"의 연구 내용입니다 ...
    Fig. 9. Scour phenomenon around jacket substructure(Case 1)

    FLOW-3D를 이용한 해상 자켓구조물 주변의 세굴 수치모의 실험

    본 소개 논문은 한국해안·해양공학회논문집에서 발행한 논문 "FLOW-3D를 이용한 해상 자켓구조물 주변의 세굴 수치모의 실험"의 연구 내용입니다. 1. 서론 해상풍력 터빈 및 해상 플랫폼과 같은 구조물의 설치가 증가하면서 세굴(Scour) 현상이 중요한 ...
    Fig. 9 Velocity vectors for Q = 0.0181 m3 /s in the area of the broad-crested weir.

    FLOW-3D를 이용한 사다리꼴 넓은 마루 위어 유동의 수치 모델링

    본 소개 논문은 Engineering Applications of Computational Fluid Mechanics에서 발행한 논문 "Numerical Modeling of Flow Over Trapezoidal Broad-Crested Weir"의 연구 내용입니다. Fig. 9 Velocity vectors for Q = 0.0181 m3/s ...
    그림 11. 연직방향 유속분포(교각전면부)

    FLOW-3D를 이용한 교각 주변 흐름의 수치해석

    본 소개 자료는 논문 "FLOW-3D를 이용한 교각주변 흐름의 수치해석"의 연구 내용입니다. 그림 11. 연직방향 유속분포(교각전면부) 1. 서론 최근 수리구조물 설계에서 3차원 수치해석 기법이 널리 사용되며, 교각 주변의 유동 특성 분석이 ...
    Figure 8. Wave formation and propagation in the impact area using the second-order approach for the density evaluation. Observation gauges P1, P2, and P3 are set to verify the water surface elevation and flow speed. Their trends are shown in the graphs for different grid resolutions (R: 5, 10, 20 m). More accurate results are obtained using the grid resolution of 5 m (sky-blue line, R5).

    1958년 리투야 베이 쓰나미 – 사전 해저 지형 재구성 및 FLOW-3D를 이용한 3D 수치 모델링

    본 소개 자료는 Nat. Hazards Earth Syst. Sci에 게재된 논문 "The 1958 Lituya Bay tsunami – pre-event bathymetry reconstruction and 3D numerical modelling utilising the computational fluid dynamics software Flow-3D"의 ...
    Fig. 5 Fluid behavior in liquid propellant dischargesimulation

    FLOW-3D를 이용한 표면장력 탱크용 메시 스크린 모델링

    본 소개 자료는 한국추진공학회 2017년도 춘계학술대회 논문집 에 게재된 논문 "Flow-3d를 이용한 표면장력 탱크용메시스크린모델링"의 소개 내용입니다. Fig. 5 Fluid behavior in liquid propellant discharge simulation 1. 서론 우주비행체의 미소 중력 ...
    Fig. 6-Shear stress distribution upstream of the orifice for different depths

    Modeling Longitudinal and Transverse Velocity Profiles Upstream of an Orifice Using the FLOW-3D Model

    본 소개 자료는 Irrigation Sciences and Engineering (JISE)에서 발행한 "Modeling Longitudinal and Transverse Velocity Profiles Upstream of an Orifice Using the FLOW-3D Model" 논문의 연구 내용을 담고 있습니다. Fig. 6-Shear ...
    Fig.5- View of a simulated congressional overflow

    Studying the effect of shape changes in plan of labyrinth weir on increasing flow discharge coefficient using Flow-3D numerical model

    본 소개 자료는 Irrigation Sciences and Engineering (JISE)에서 발행한 "Studying the effect of shape changes in plan of labyrinth weir on increasing flow discharge coefficient using Flow-3D numerical model" 논문의 ...
    Fig. 3 Vane V0 induced circulation downstream of vane (x = 65.5 cm), flow Froude number of Fr = 0.16

    Performance Evaluation of Submerged Vanes by Flow-3D Numerical Model

    본 소개 자료는 Iranian Hydraulic Association Journal of Hydraulics에서 발행한 "Performance Evaluation of Submerged Vanes by Flow-3D Numerical Model" 논문의 연구 내용을 담고 있습니다. Fig. 3 Vane V0 induced circulation ...

    Landslide-Induced Wave Hazard

    Landslide-Induced Wave Hazard 

    Figure 1. The outskirts of Chungtangh village

    인도 Sikkim에 위치한 The Teesta III Hydropower Project는 가파르고 좁은 히말라야 계곡에 위치한 60m의 Concrete Face Rockfill Dam (CFRD)이 포함되어 있습니다. 이 계곡은 지진 활동이 활발하며 가파른 경사면은 산사태를 발생시킬 수 있습니다. 댐 상류 저수지의 산사태로 CFRD를 범람할 수 있다는 우려가 있었습니다. 몇 초 이상 과도하게 지속되면 오버플로우로 인해 CFRD가 잘못될 수 있습니다. 비록 댐이 무너지지 않았지만, 여전히 Chungtangh에 있는 상류쪽 작은 마을은 홍수가 날 것이라는 우려가 있었습니다.

    Teesta강 계곡의 가장 가파른 경사면은 댐의 바로 상류에 위치해 있는데, 댐의 산사태가 가장 일어날 가능성이 높은 지역입니다. 이 분석의 목적은 저수지에 대한 산사태를 시뮬레이션하고 그 결과로 발생하는 파도가 댐에 넘치는지 여부를 결정하는 것이었습니다.

    Moving Objects Model Used to Simulate Landslide                                      

    Tecsult는 저수지의 침전물과 퇴적물을 모델링하는데 성공적이었기 때문에 FLOW-3D를 선택하여 이를 시뮬레이션하였습니다. 저수지의 시뮬레이션은 시작점으로 사용되었습니다. FLOW-3D의 Moving Objects모델은 산사태를 시뮬레이션하는데 사용되었으며 VOF모델은 웨이브 생성을 시뮬레이션하는 데 사용되었습니다.

    저수지의 산사태를 추정하기 위해서는 여러가지 방법이 고려되었습니다. 경험적 방법은 흔히 산사태가 발생한 파도를 평가하는데 사용되지만, 이러한 방법은 여러가지 면에서 부족합니다. 이러한 방법은 근접 필드 또는 스플래시 영역에 대한 정보를 제공하지 않습니다. 댐은 슬라이드 면과 매우 가깝기 때문에 스플래시 영역을 아는 것이 중요했습니다. CFRD는 몇 초 이상 overflow를 견딜 수 없었습니다. FLOW-3D는 미끄러운 지형 질량과 물 사이의 완전 결합된 상호 작용을 계산하여 시나리오를 3 차원에서 시뮬레이션하는 방법을 제공합니다.

    이 문제를 시뮬레이션하기 위해 간단하고 작은 크기의 자유 낙하 블록으로 구성된 실험과 비교하였습니다. 이 경우는 아래 동영상에 나와 있습니다. 그 결과로 생긴 파도 높이는 그 실험과 잘 맞았습니다.

    이 모델의 STL파일은 FLOW-3D로 직접 가져옵니다. 예상 산사태 지역의 크기는 지질 정보와 주변 산사태 관측치를 바탕으로 결정되었습니다. 30,000m³, 100m높이의 산사태가 310만 셀의 메쉬로 시뮬레이션 되었습니다. 높이가 1m인 측면 3m의 균일한 셀을 사용했습니다. 최대 슬라이딩 속도는 진입 지점에서 23m/s에 도달했습니다. 파도는 높이 8m, 속도 10m/s로 댐에 도달하여 몇 초 동안 범람했습니다. 그 결과로 상류 마을에서는 홍수가 나타나지 않았습니다.

    Figure 3. Prediction of wave height in the splash zone and near field in a small reservoir, with refraction.

    Figure 4. Wave heights plotted against each other

    Figure 5. Downstream view of TEEST III dam and water intake CATIA model

    Conclusions

    이 작업의 주된 관심사는 댐의 범람으로 인해 댐과 Chungtangh 마을이 파괴될 수 있었다는 것입니다. 그러나 시뮬레이션에 따르면 댐은 잠시 동안만 범람했고 파도는 마을에 닿지 않았습니다. Chungtangh마을은 강 위에 충분히 높기 때문에, 그것을 범람시키기 위해서는 상당한 파도의 높이가 필요할 것입니다.

     

    Aerial Landslide Generated Wave Simulations

    Aerial Landslide Generated Wave Simulations

    Aerial Landslide Generated Wave(ALGW)는 수역에 영향을 미치는 빠른 슬라이드의 결과이다. 이것은 암석에 의해 생성된 작은 파도 이거나, 3000만 입방 미터의 암석으로 인한 500m를 초과하는 파도 일 수도 있다.
    공학적 관점에서 보면 ALGW는 근접한 해안을 따라 인간이 거주하는 인구/자산이 있는 수역에서 발생할 때 큰 관심을 가진다. 여기서 파동이 발생하면 해안선이 파손되고 홍수가 날수 있으며, 댐붕괴로 인한 사망까지 일으킬 수 있다(Müller-Salzburg, 1987). 결과적으로, ALGW에 의해 야기되는 최대 파도 상승을 예측하는 것은 경제적, 환경적, 안전상의 이유로 매우 중요합니다.
    안타깝게도 분석적인 솔루션이 없는 매우 복잡한 문제로, 유체 역학적인 측면에서뿐만 아니라 지질학적인 관점(즉, 크기/기하학적인 슬라이드의 밀도 프로파일)에서도 마찬가지입니다. 이와 같이, 대부분의 현장 별 ALGW 최대 파형 예측은 확장된 물리적 모델을 사용하여 평가되었다. 일부는 전산유체역학(CFD) 소프트웨어를 기반으로 할 수도 있지만 비용이 많이 들며, 특히 풀 스케일 3차원 문제의 경우 정확성에 대한 논쟁의 대상이 되고 있습니다.
    그러나 컴퓨터 하드웨어와 CFD소프트웨어가 계속 발전함에 따라 이제 CFD를 사용하여 ALGW를 실제로 시뮬레이션할 수 있게 되었습니다. 이와 같이 본 연구는 고 충실도의 물리적 모델 데이터를 FLOW-3D와 비교하여 ALGW를 CFD시뮬레이션을 검증하기 위한 지속적인 노력으로 진척시키는 것을 목표로 한다.
    다음 절에서는 실제 및 수치 모델 설정에 대한 개요를 제공한다. 뿐만 아니라, 생성된 데이터와 간단한 비교를 제공한다.

    Experimental Setup
    물리적 실험은 Northwest Hydraulic Consultants 노스 밴쿠버, 캐나다 실험실에서 만들어졌고 실험을 거쳤다. 그것은 30° 경사의 서쪽 벽을 가진 0.5미터 폭의 수로, 45°의 경사진 동쪽 벽, 그리고 두개의 북쪽과 남쪽 측면에 수직 벽, 그리고 1.025m의 수평 단면을 가진 0.610m 너비의 수로로 구성되었다. ALGW를 생성하고 평가하기 위해, 45° 경사 노즈를 가진 0.177×0.305×0.305m의 아크릴 박스를 사용한 6초 시험을 사용했다.
    이 슬라이드를 놓았을 때, 슬라이드는 (중력에 의해) 0.607m 심층수에 충돌하기 전에 서쪽 경사면에서 0.768m 아래로 이동했다. 그 후, 물을 통해 또 다른 1.05m를 이동하여 정지 블록을 치기 시작했다. 슬라이드 가속 및 변위뿐만 아니라 파고 높이는 6 초 실험 전체에 대해 100Hz의 주파수에서 기록되었다. 이 데이터를 수집하는 데 사용 된 도구는 다음과 같다.

    • 컴퓨터화된 데이터 수집 시스템
    • 슬라이드의 시간에 따라 이동 한 거리를 측정하는 문자열 가변 저항기
    • 슬라이드 가속도를 측정하는 1 차원 가속도계
    • 물의 주요 본체 내에 배치 된 3 개의 1 차원 커패시턴스 웨이브 – 프로브
    • 웨이브 런업을 캡처하기 위해 동쪽 경사면을 따라 사용되는 저항 사다리꼴 웨이브 프로브
    • 타이밍 스위치 캡처 슬라이드 릴리스 시간 사용
    • 흑백 비디오 카메라

    테스트가 반복 가능하고 오작동이 발생하지 않았는지 확인하기 위해 테스트를 5 번 반복하고 각 장비에 대해 평균을 구했다.

    Numerical Model Setup
    물리적 실험의 전산화 된 3 차원 모델을 제작한 STL 파일을 FLOW-3D로 가져왔다. 일단 FLOW-3D에 들어간 3D 모델은 약 1,370 만개의 0.0075m 크기의 정사각형 셀로 이산화되었고, 벽을 둘러싸고있는 6 개의면 각각에 ‘wall’경계가 사용되었다.
    슬라이드를 일반적인 이동 물체로 설정하고, 물리 모델로부터 수집 된 데이터(즉, 가속 및 변위 데이터의 후 처리)에 기초하여 속도가 주어졌다. 동서면 경사면의 표면 거칠기는 0.00025m으로 설정되었다. 모델링 된 유체는 293k의 물이었고, 동적 RNG 난류 모델이 기본 설정과 함께 사용되었다(implicit pressure solve; and, explicit viscous stress, free surface pressure, advection, moving object/fluid coupling solvers).
    물리적 모델과 마찬가지로 FLOW-3D는 6 초의 시간을 시뮬레이트하지만 실제 모델과 같이 매 0.01 초가 아닌 0.02 초마다 데이터를 저장하였다(데이터 관리 관점에서 선택하였음).

    Result

    FLOW-3D 실험의 결과는 그림에 나와 있다. 4개의 웨이브 각각에 대해 실험 시간 동안 파고를 보여준다. 이와 같이, 제시된 파도 높이는 단순히 flume을 통해 전파되는 파도의 구현(즉, 2 차원의 경우에서 볼 수있는 것)이 아니라 오히려 여러 파도의 상호 작용으로 인한 파도 높이를 초래한다.

    • 슬라이드 충격시 발생하는 충격파(1차 신호)
    • 슬라이드 뒤의 충격파 충돌(2차 신호)
    • 북쪽, 동쪽, 서쪽 및 남쪽 벽에서의 웨이브 반사(3차 신호)

    또한 길이 방향의 FLOW-3D 데이터(중심선에서)를 실제 모델 비디오 위에 겹쳐서 자유 표면의 FLOW-3D 글로벌 예측을 평가했다. 이것은 아래의 동영상에서 볼 수 있다.
    그림과 위의 비디오를 보면 FLOW-3D 데이터가 웨이브 프로브 1, 2 및 3의 경우 물리적 데이터를 매우 잘 일치한다는 것을 알 수 있다. 하지만 웨이브 프로브 4에 대해서는 정확도가 떨어진다.
    FLOW-3D 시간 데이터와 관련된 오류는 각 웨이브 프로브에 대한 RMSE (root-mean-square-error)를 취하여 평가된다.

    Discussion
    이 조사에서 실제 모델의 고 충실도 데이터는 ALGW로 최대 파도 상승에 대한 FLOW-3D 예측과 비교되었다. RNG 모형의 기본 설정을 사용하여 FLOW-3D는 주요 수역 내에서 파고를 정확하게 재현 할 수 있었다. 그러나 최대 파동은 약 43%가 넘었다.
    최대 웨이브 런업을 줄이기 위해 몇 가지 대안인 FLOW-3D 물리 설정이 사용되었다. 그러나 43 % 이하로 떨어지는 것은 불가능했다. 이러한 대체 시뮬레이션에 대한 주목할만한 관찰은 다음과 같다.

    • first-order momentum advection scheme의 0.01m 메쉬는 최대 파동 상승 오차가 96% 인 반면 동일하게 0.0075m 메쉬의 오차는 130%였다. 그러나 second-order로 변경하면 0.01 m 및 0.0075 m 메시의 경우 각각 55% 및 43%의 오차가 발생한다. 또한 메쉬 셀 크기를 0.005m으로 줄이면 80%의 오차가 발생한다.
    • 이 테스트 케이스에서 가장 중요한 매개 변수는 momentum advection scheme이다. 평균적으로 second-order를 사용하면 first-order대비 오차가 약 50% 감소한다.
    • FLOW-3D의 MP 버전을 사용하여 0.005m의 메쉬 셀 크기를 사용해야 한다. 해석 시 CPU 시간은 33 시간이었다. 비교를 위해 FLOW-3D의 SMP 버전은 0.0075m의 메쉬 셀 크기로 시뮬레이션을 실행하는 데 26시간이 필요했지만 MP 버전은 4.5시간 밖에 걸리지 않았다.

    [1] 3.5GHz 8 코어 AMD FX-8320 프로세서에서 약 6초의 시뮬레이션 시간이 대략 26시간 소요되었다.

    References
    Fritz, H. M., Hager, W. H., & Minor, H.-E. (2004). Near Field Characteristics of Landslide Generated Impulse Waves. Journal of Waterway, Port, Coastal & Ocean Engineering, 130(6), 287–302. doi:10.1061/(ASCE)0733-950X(2004)130:6(287)
    Miller, D. J. (1960). Giant Waves in Lituya Bay Alaska (Geological Survey Professional Paper No. 354-C). Washington, D.C.: United States Government Printing Office.
    Müller-Salzburg, L. (1987). The Vajont catastrophe— A personal review. Engineering Geology, 24(1–4), 423–444. doi:10.1016/0013-7952(87)90078-0

    Prediction of Shrinkage Defects During Investment Casting Process

    Indianapolis Storm-Water System

    하수도 시스템은 액션영화의 도피 루트로 사용되지 않는 한 흥미롭지 않을 것입니다. 폭우로 인해 이산화탄소 수치가 올라갈 때까지 여러분은 그것에 대해 생각조차 하지 않을 것입니다. 불행하게도, 770개 이상의 오래 된 미국 도시들 아래에 있는 하수구 시스템은 심한 폭풍으로 오염 문제를 일으킵니다. 이러한 구형 설계는 하수 및 폭풍 유실을 위한 비용 효율적인 단일 스타일 파이프를 사용했으며 연결된 파이프로 강 및 호수에 하수를 내보냅니다(CSO).

    1994년 미국 환경보호청(EPA)은 주로 북동부 및 그레이트 레이크 지역의 관련 지방 자치 단체들에게 CSO관련 문제를 줄이거나 제거하도록 하는 정책을 발표했습니다. (2000년 “Clean Water Act”의 일부로 법률화된 정책). 인디애나 폴리스(Indianapolis)는 가벼운 비 폭풍으로 인해 하수 오물의 백업 및 범람이 발생할 수 있는 도시 중 하나였으므로, 주요 건설 조건에서 2025년까지 문제를 해결하는 것이 필요하였습니다.

    인디애나 폴리스는 국제 디자인 회사인 AECOM에 Citizens Energy Group이 건설하고 있는 3개의 깊은 암석 저장 터널 중 첫 번째를 설계할 것을 요청했습니다. 총 25마일인 이 시스템은 대규모 지하 펌프장과 기존의 하수구에서 CSO를 수직으로 떨어뜨리는 연결 구조물을 포함합니다. 첫 번째 터널의 경우, 강우가 가라 앉은 후에 3 개의 커다란 강하 구조물이 CSO를 저장 터널로 전환하여 후속 처리를 수행했습니다.

    프로젝트를 해결하기 위해 AECOM은 여러 가능한 낙하 구조물 설계의 동작을 시뮬레이션하기 위해 FLOW-3D를 선택하여, 구축 및 평가 예산이 책정 된 물리적 모델에 대한 재 작업의 필요성을 최소화했습니다. 테스트 결과는 예측 값과 일치하였으므로 재설계가 필요하지 않았습니다. 또한, 이제 AECOM은 유압 설계작업의 첫 번째 단계를 일반적으로 CFD시뮬레이션을 사용합니다.

    Large Scale Project on a Tight Delivery Schedule

    촉박한 납품 일정에 따른 대규모 프로젝트

    20세기에 건설된 하수 처리장은 주거용, 상업용, 환경유출물의 유출로 무엇을 해야 할 것인지에 대한 새로운 인식을 가져다 주었습니다. CSO 방전은 정상적으로 운영되는 동안 처리시설로 직접 이동되며 모든 과정이 양호하게 운영됩니다. 불행하게도, 대규모 폭풍이 발생하는 동안, 발전소들의 초과 용량문제를 피하기 위해 인근 수역으로 과도한 유량을 방출합니다. 이들 배출은 기름과 살충제, 야생동물 배설물에 이르기까지 다양한 오염 물질을 포함합니다.
    고무적인 성공의 신호로, 1990 년대에 착공된 새로운 CSO 분리, 저장 및 처리 시설로 오염의 영향에 대해 67 %의 개선을 이루었지만, 여전히 많은 연구가 이루어져야 합니다. 인디애나 폴리스의 경우, 인디애나 폴리스시 공공사업부가 CSO 장기 통제계획을 준비한 2008년에 그러한 노력이 시작되었습니다. 정상적인 처리 공장에서 처리 할 수 있을 때까지 오버플로우가 발생하는 “저장 및 운송”접근법의 핵심은 인디애나 폴리스 터널 저장 시스템 또는 인디애나라고 합니다.

    이 시스템의 첫번째 단계는 딥 록 터널 커넥터(DRTC)라고 불리는 1억 8천만달러 가치의 프로젝트입니다. DRTC는 길이 7마일의 18피트 직경의 지하 터널로, 기존의 인디애나 폴리스의 3개의 서버 대 계층 유출 연결의 흐름 경로를 다시 만들 것입니다(그림 1). 목표는 과잉 강우 유출을 기존 하수구와 새 터널 사이의 낙하 구조를 통해 이들 대피소에서 거대한 터널로 안전하게 재배치하고, 폭풍 후 처리를 위해 처리장으로 펌핑 될 수있을 때까지 유지합니다.

    Fig. 1. City of Indianapolis Deep Rock Tunnel Connector (DRTC), a “storage and transport” concept being built to handle combined sewage overflow (CSO) during heavy storms. Three vertical drop structures will capture this flow and divert it downwards to 18-foot-diameter storage tunnels running more than 250 feet underground; the tunnels store the CSO until sewage treatment plant capacity becomes available. (Image courtesy Citizens Energy Group)

    평균적으로 지표면 아래 250피트 깊이에서, DRTC는 건설과 궁극적인 운영 동안 위의 주변 지역에 대한 혼란을 최소화하도록 설계되었습니다. 그러나 이 프로젝트의 규모와 복잡성은 AECOM의 과제에 긴급성을 더했습니다. 세 장소 각각에 대한 가능한 낙하 구조 설계와 평가, 구조물 설계의 60%를 7개월 이내에 마무리 지었습니다.

    이러한 구조물의 목적은 표준 도시 하수 시스템에서 깊은 저장 터널로 하수 흐름을 전달하는 동시에, 효율적 손실( 느린 속도 또는 백업)과 장기적인 도심을 방지하는 것입니다. 각 섹션의 크기와 모양이 유입 흐름의 볼륨 및 속도와 세심하게 일치하지 않을 경우 발생할 수 있는 구조적 손상입니다.
    AECOM의 수석 기술 전문가인 라이언 에디슨 컨설턴트는 계약의 스케줄링 요구 사항이 유효성 검사를 위해서는, 단 하나의 모델에만 물리적 건물과 테스트 활동을 제한할 것이라는 것을 알게되었습니다. 다른 주요 건설 프로젝트에 15년간 FLOW-3D 시뮬레이션 소프트웨어를 사용해 왔기 때문에, 난류, 과전압 및 에너지 낭비를 예측하는 능력은 충분하지 않고 디자인 프로젝트에 적합하다고 자신했습니다. 또한 여러 검증(what-if) 시나리오를 실행하기 위한 소프트웨어 옵션을 통해 설계 세부 사항을 다시 실행해야 하는 위험을 최소화할 수 있었습니다. 변경 사항이 적용될 경우 상당한 이점은 여러개의 병렬 시공 트랙이 있는 프로젝트에 있습니다.
    시간 제약에도 불구하고, 에디슨은 특히 이 도전에 만족했습니다. 왜냐하면 “CFD로 드롭 구조 설계를 만들고 물리학에서 이것들은 너무 큰 구조이기 때문입니다.”라고 그는 말합니다. 그것들은 CFD는 실제로 사용되지 않는데 보통 물리적 모델이나 손으로 계산하는 것으로 이루어집니다.

    DRTC 프로젝트를 위해서, 그는 먼저 시뮬레이션된 작동 조건에 대해서 컴퓨터 설계를 테스트할 것입니다. 에디슨은 3차원의 일시적이고 격동적인 흐름 조건을 모델링 할 수 있는 소프트웨어 패키지인 FLOW-3D를 사용했습니다. 각 설계에 대한 계산 메쉬를 변경하지 않고도 여러 설계 지오 메트리를 모델링 할 수 있는 기능이였습니다.
    시뮬레이션 데이터로 무장한 에디슨은 그 결과를 아이오와 대학교 II. 시설에서 시험한 1:10 크기의 물리적 모델의 작동 데이터와 비교하였습니다. (후자는 원래 아이오와 유압 연구소라고 불렸지만, 지금은 그룹의 다양한 범위를 반영하여 IIHR-Hydroscience & Engineering으로 알려져 있습니다.)

    Zeroing in on the Drop-Structure Challenge

    드롭 구조 과제에서 영점 조정

    가장 제한적인 DRTC 사이트의 지오 메트리는 CSO 008로 지정된 레귤레이터에서 발생합니다. 기존 CSO 레귤레이터(기울기 약 75피트 아래)를 새 18피트 직경의 수집 터널과 연결하려면, 이 위치에서 150피트 이상의 수직 방향 주행이 필요합니다. 각 낙하 구조에 7백만달러 이상이 소요되는 경우, 프로젝트 관리자들은 물리적 모델이 구축된 후 비용과 시간이 많이 소요되는 재설계가 필요한 가능성을 낮추려고 애썼습니다.

    역사적으로 낙하 구조는 이전 프로젝트를 적용하여 설계된 후 축소 모델로 구축되었으며, 테스트만으로도 6개월 이상이 소요될 수 있습니다. 가속화된 이 프로젝트에서, 2009년 가을에 시작한 AECOM의 초기 과제는 두가지 표준 개념 중에서 하나를 선택하는 것이었습니다. 포장-파운드 스타일과 접선 vortex버전, 둘 다 시속 35마일의 폭풍이 몰아치는 물 속에서 속도를 늦추고 통제하기 위해서 직접 계산 및 FLOW-3D에서 결정한 일반 구조 직경 및 구성 요소 크기를 사용한 초기 CFD분석으로, AECOM은 시공 가능성 및 비용 고려 사항을 평가하는 데 사용했습니다.
    CSO 008의 현장 요구 사항과 비용 효율성을 고려할 때, 시 당국과 AECOM은 접선 소용돌이 낙하 구조를 선택했습니다. 이 설계의 핵심 요소는 흐름을 먼저 환상적인 제트로 유도한 다음, vortex 유도 나선형 흐름을 생성하는 테이퍼(확대) 접근 채널에 의해 공급되는 수직 튜브(드롭 샤프트)입니다. 이 통제 된 하강은 속도가 느려지고 하루 3 억 갤런 (mgd) 이상에 이르는 흐름을 안전하게 처리합니다. 스토리지 터널의 파괴적인 난류를 방지하는 것이 핵심 목표이므로 드롭 샤프트 흐름의 사전 차단이 설계의 핵심입니다.

    구조 자체는 6 개의 주요 부분으로 구성됩니다. 1) 접근 채널 (기존의 하수 터널에서 나온 것), 2) 수평 흐름을 넓히고 수직 드롭 샤프트로 수평 흐름을 전달하는 직사각형 전이 테이퍼 채널, 3) 드롭 샤프트 자체 4) 탈 기실 (유량을 수평 방향으로 방향을 바꾸고 공기 유입을 감소시키는), 5) 수직 공기 배출구를 통해 낙하에서 유입 된 공기를 제거하고 적하 유체의 공기 코어가 열려 있고 6) 탈기 챔버와 저장 터널 챔버를 연결하는 파이프 (adit) (그림 2).

    Fig. 2. CAD diagram of proposed Indianapolis DRTC combined sewage overflow (CSO) vertical drop structure, showing approach channel, taper channel and vortex dropshaft. Using FLOW-3D CFD analysis software, AECOM simulated the flow behavior, gaining confidence in the system performance prior to physical model testing. (Image courtesy AECOM)
    Prediction of Shrinkage Defects During Investment Casting Process

    This article was contributed by Dr. S. Savithri, Senior Principal Scientist at CSIR-NIIST

     

    인베스트먼트 주조공정은 가장 오래된 주조 공정 중 하나로 기원전 4000년 이후에 보편화되었습니다. 이 과정은 용해된 금속을 소모품패턴으로 생성된 세라믹 쉘에 주입하는 과정을 수반합니다. 일찍이 그것은 금, 은, 구리와 청동 합금으로 장신구와 우상을 만드는데 사용되었습니다.

    인베스트먼트 주조공정은 1897년 아이오와 주, 위원회 블러프스의 Barabas Frederick Philbrook이 묘사한 대로 치과의사들이 왕관과 인레이를 만들기 위해 그것을 사용하기 시작한 19세기 말 현대 산업공정으로 사용되기 시작했다. 1940년대에는 제2차 세계대전 당시 기존 방법으로는 형성될 수 없거나 지나치게 많은 가공이 필요한 특수 합금의 정밀 순모형 제조 기술에 대한 수요로 인해 투자 주조 공정이 증가하였다.

    오늘날 투자 주조 공정은 표면 마감 및 치수 정확도가 우수하여 거의 순 형태에 가까운 철, 비철 및 초합금의 소형 산업용 부품을 생산하는데 주로 사용됩니다.

    인베스트먼트 주조 공정은 다음 네 가지 주요 단계로 구성됩니다.

    • 왁스 패턴 생성 후, 패턴 클러스터를 만들기 위해 게이트 시스템으로 청소 및 조립합니다.
    • 나무는 세라믹 쉘을 얻기 위해 미세 모래와 Course한 모래 입자의 슬러리로 번갈아 코팅됩니다.
    • 용기는 건조되고, 왁스를 녹이기 위해 가열되며, 강도를 높이고 주입 준비합니다.
    • 마침내 주조 합금이 용해되어 예열된 쉘에 주입됩니다. 응고 후에 쉘이 파손되어 주조 부품을 얻습니다.

    Figure 1. Solid model of the casting geometry

    인베스트먼트 주조 공정에서 얻은 부품은 많은 중요한 용도에 사용되므로 내부적인 결함이 없어야 합니다. 투자 주조 공정에서 발생하는 주요 결함은 세라믹 포함, 균열, 변형, 플래시, 주탕불량, 수축, 슬래그 포함, 탕경계등입니다. 얻은 주조물의 품질을 예측하려면 금속-몰드 열 전달계수, 주입 온도 등 다양한 주조 공정 매개 변수의 영향을 연구해야 합니다. 즉, 쉘 두께 및 쉘 열 전달계수가 그것입니다. 현대 컴퓨터 시스템 및 시뮬레이션 소프트웨어의 출현과 함께 금형 충진 및 응고 시뮬레이션은 주조공장에서 결함을 예측하고 설계를 최적화하는데 점점 더 많이 사용되고 있습니다.

    이 연구의 주요 목적은 투자 주조 공정에서 주요 요소인 복사 열 전달과 인베스트먼트 주조공정에 고유한 쉘 금형이 FLOW-3D에서 효과적으로 구현될 수 있는지를 조사하는 것입니다. FLOW-3D를 사용하여 간단한 형상을 위한 인베스트먼트 주조공정의 주입 및 응고 시뮬레이션을 수행함으로써 두 구성요소의 서로 다른 효과를 조사합니다. 다양한 위치에서 얻은 온도의 수치는 문헌 [1]에보고 된 실험 결과로 검증됩니다. 복사 열 전달계수, 쉘 몰드 두께, 탕구 및 게이트의 위치에 대한 영향도 조사했습니다.

    Figure 2. Shell mold

     

    Methodology

    현재 연구에서 사용된 계산 형상은 그림 1에 나와 있습니다. 쉘 몰드는 다음 단계를 사용하여 작성되었습니다.

    • 구성 요소 1로 형상을 FLOW-3D로 가져오고 지정된 셀 크기로 가져온 형상을 중심으로 메쉬 블록을 작성합니다.
    • “보완”유형의 component1의 첫 번째 하위 구성 요소를 만들어 하위 구성 요소 외부의 모든 항목을 메쉬의 범위까지 확고하게 만듭니다.
    • 솔리드 데이터베이스에서 이 솔리드 블록의 금형 재질 특성을 정의하십시오.
    • 솔리드 특성 GUI의 구성 요소 특성에서 “열 침투 깊이”를 정의하는 옵션이 있습니다. 여기서 쉘 두께 값을 정의 할 수 있습니다.
    • 이제 전처리기를 실행하십시오.
    • 분석 탭> 3D 탭으로 이동 한 다음 이전 단계에서 생성 한 prpgrf 파일을 엽니다. ‘Iso-surface’와 ‘color variable’에서 “열 활성화 구성 요소 볼륨”을 선택하고 “렌더링”을 선택하십시오.
    • Display에 이제 형상의 셸 부분 만 표시됩니다.
    • 개체 목록 (창의 왼쪽 하단)에서 “구성 요소 1″을 선택하고 “구성 요소 1″을 마우스 오른쪽 단추로 클릭 한 다음 “stl로 내보내기”를 선택하여 이 곡면을 STL 파일로 저장하십시오.

    Figure 3. The view of the two mesh blocks for the creation of a void with discretization

    쉘 몰드 용 STL 파일을 만든 후 파일을 구성 요소 1로 새 시뮬레이션으로 가져오고 이전에 작성한 주조 형상을 하위 구성 요소로 가져오고 유형을 ‘hole’으로 선택합니다. 쉘 몰드와 함께 주조 형상이 그림 2에 나와 있습니다. 이것은 우리의 계산 영역으로 사용됩니다. 다음은 계산 영역을 cubical/rectangular셀로 분할하기 위한 메쉬를 만드는 것입니다. 메쉬 블록을 작성하여 FLOW-3D에서 메쉬를 생성합니다. 현재의 작업을 위해 우리는 2.5mm의 고정된 셀 크기가 선택된 그림 3에 표시된 균일한 메쉬 옵션을 선택했습니다. 입력 위치 주변에 메시 블록 2가 사용되는 현재 시뮬레이션을 위해 메시 블록 2개가 생성되었습니다. 쉘과 주변 공기 사이의 30°C에서의 열 전달을 고려하여 쉘 주위에 보이드 영역이 정의됩니다. 이 영역은 ‘열 전달 유형 1’이 있는 보이드 영역으로 선택되며 셸과 주변 공기 사이에 열 전달 계수 값이 지정됩니다. 열 전달 유형 1은 방사선을 포함한 종합 열 전달 계수가 됩니다.

    쉘 주형에 선택된 재료는 zircon이며 열 특성은 Sabau and Vishwanathan에 의해 수행된 실험에서 얻을 수 있습니다[2]. 표 1은 연구에 사용된 재료에 대해 지정된 값을 보여 줍니다.

    MATERIALPROPERTY VALUEUNIT
    Fluid –AluminiumA356

    alloy

    Density  2437kg/m³
    Thermal conductivity116.8W/(mK)
    Specific heat 1074J/(kgK)
    Latent heat 433.22kJ/m³
    Liquidus temperature608°C
    Solidus temperature552.4°C
    Zircon MoldThermal conductivity1.09W/(mK)
    Specific heat* Density1.63E+06J/( m³K)

    Initial and boundary conditions used are show in Table 2.      

     

    Mold temperature 430°C
    Melt pouring temperature 680°C
    Filling time 7 s
    Interface heat transfer coefficient 850 W/m2K
    Heat transfer coefficient between ambient and mold (radiation effect)30 -100 W/m2K

    Table 2. Initial and boundary conditions used for the simulation

     

    탕구저에 들어가는 용융물의 초기 속도와 온도는 메시 블록 2의 상단 경계에서 속도 경계 조건으로 주어집니다. 기본적으로 다른 모든 경계는 대칭 유형으로 설정됩니다.

     

    Results & Discussion

    Validation with reported experimental results

    충전 및 응고 동안 냉각 곡선을 얻기 위한 실험에서 Sabuet.al[1]에 의해 선택된 네 개의 위치가 검증 목적으로 사용되었습니다. 그들은 C1, C2, S11, S12및 S21로 언급됩니다. C1과 C2지점은 주물의 플레이트의 중심에 있으며 S11, S12및 S21은 모두 쉘에 위치합니다. 이러한 위치에서의 온도 변화는 그림 4와 같습니다.

    온도 프로파일의 수치 및 실험결과의 차이가 허용한계 안에 있음을 알 수 있습니다. 프로브 포인트 C1과 C2의 경우, 수치와 실험 결과 사이의 차이는 응고 중에 5%, 응고 후 냉각 시 12% 이내입니다. 쉘의 점에 대한 수치 결과는 실험 결과보다 약 5% 높습니다. 이는 쉘 재료에 열 물리학적 특성을 할당할 때 발생하는 가정과 쉘 열 전달 계수의 값 때문일 수 있습니다.

     

    Fill sequence & solidification pattern for two different sprue locations

    두 가지 다른 스프 루 위치의 채우기 순서 및 응고 패턴

    2 개의 상이한 탕구 위치에 주물충전 순서는5a 및5b에 나와 있습니다. 최종 탕구가 더 많은 스플라인을 생성하므로 결함으로 이어질 수 있습니다. 탕구가 중간에 놓여지면 흐름은 보다 균일 해지고 두 주조 단면에서 비슷한 온도 분포를 보입니다. 50 % 응고 후의 온도 프로파일의 2D 도면은 두 경우 모두 그림 5c 및 5d에 나와 있습니다. 수축 위치에서 볼 때 두 탕구 위치가 결함을 일으키는 것은 분명합니다.

    Figure 5a. Fill sequence at different time intervals when the sprue is located at one end

    Figure 5b. Fill sequence at different time intervals when the sprue is located in the middle

    Figure 5c. 2D temperature profile after 50% solidification when the sprue is located at one end

    Figure 5d. 2D temperature profile after 50% solidification when the sprue is located in the middle

    Effect of shell thickness

    인베스트먼트 주조에 대한 쉘 두께의 효과를 연구하기 위해 두께가 7.2, 10, 15 및 20 mm인 주물을 선정하였습니다. 그림 6a 및 6b는 주조품의 특정 위치에서 냉각 곡선을 나타내며, 이는 C1으로 나타내고 쉘 몰드 내의 특정 위치에 있으며, 응고 중에 S11로 나타납니다. 세라믹 쉘의 두께가 7.2 mm에서 15 mm로 증가하면 냉각 속도가 감소하여 응고 시간이 길어지는 것을 볼 수 있습니다.

    Effect of shell heat transfer coefficient

    셸 열 전달 계수는 열이 셸 금형의 외부 벽에서 방사선을 통해 주변 공기로 열을 방출하는 속도를 나타냅니다. 이 효과를 조사하기 위해 열 전달 계수의 값을 20에서 80W/m2K까지 다양하게 했습니다. 7a 및 7b로부터, h의 변화는 주조 재료 및 쉘의 냉각 속도에 중요한 영향을 미친다는 것을 알 수 있습니다. 열 전달 계수가 20에서 80W/m2K로 증가하면 C1에서의 응고 시간이 812 초에서 334 초 (약 44 %)로 감소되었음을 알 수 있습니다. 따라서, h의 값을 변화시키는 것은 주물의 미세 구조에 영향을 미칩니다.

    Figure 6a. Temperature profile at location C1 (casting) for the casting geometry where the sprue is located at one end for various shell thickness values

     

    F

    Figure 6b. Temperature profile at location S11 (shell) for the casting geometry where the sprue is located at one end for various shell thickness values

    Figure 7a. Temperature profile at location C1 (casting) for the casting geometry where the sprue is located at one end for various heat transfer coefficient values between the shell mold & ambient

    Figure 7b. Temperature profile at location S11 (shell) for the casting geometry where the sprue is located at one end for various heat transfer coefficient values between the shell mold & ambient

    Conclusions

    인베스트먼트 주조 공정의 몰드 충진 및 응고 시뮬레이션은 FLOW-3D를 사용하여 수행되었습니다. 주조 공정에 대한 주조 매개변수의 영향을 연구하기 위해 파라메트릭 연구가 수행되었습니다. 본 연구에서 다음과 같은 결론을 도출 할 수 있습니다.

    • FLOW-3D는 멀티 캐비티 몰드의 주입 및 응고 모델링이 가능합니다. 프로브 위치의 예측 온도 프로파일은 실험 데이터의 허용오차 이내였다.
    • 쉘 두께의 경우, 두 경우 모두 셸의 임계 두께가 있으며, 그 이상으로 열 전달 특성이 역행하는 것으로 확인되었습니다. 셸 두께가 증가함에 따라 응고 시간이 임계 두께까지 증가하여 감소하기 시작했습니다. 원래 형상의 경우 임계 두께는 15~20mm인 반면 수정된 형상의 경우 10mm와 15mm 사이에 있다.
    • 쉘과 대기 사이의 열 전달 계수 h는 열 전달 특성에 가장 큰 영향을 미치는 것으로 나타났습니다. h가 20에서 80W/m2K로 4 배 증가할 때 탕구의 중심에서 응고 시간이 40 % 이상 감소했습니다.

    References

    Sabau, A.S., Numerical Simulation of the Investment Casting Process, Transactions of the American Foundry Society, vol. 113, Paper No. 05-160, 2005.

    Sabau, A.S., and Viswanathan, S., Thermophysical Properties of Zircon and Fused Silica-based Shells used in the Investment Casting ProcessTransactions of the American Foundry Society, vol. 112, Paper No. 04-081, 2004.

     
    Design and CFD Analysis

    설계 및 CFD분석

    일반적인 소용돌이 설계는 널리 받아들여지고 있지만, 각 낙하 구조는 최적의 접선 흐름 특성을 보장하기 위해 인디애나 폴리스의 위상에 맞는 적절한 크기를 가져야 했습니다. 특히, 가능한 설계에 대한 AECOM의 계획은 세가지 목표를 가지고 있었습니다. 결합된 접근법과 테이퍼 채널을 짧은 길이로 제한하는 현장, 고유의 제약이 있었는지를 결정합니다. 허용 가능하지만 접근 방식에서 과도한 난류 조건이 발생하지 않았습니다. 테이퍼 채널에 안정적인 흐름 조건이 존재하는지 확인하고 다양한 흐름 조건에서 흐름 안정성을 평가했고, 논리적 기준점은 밀워키 인라인 스토리지 프로젝트라고 불리는 잘 알려지고 문서화된 시스템이었습니다.

    Edison은 DRTC 프로젝트 규모에 맞춰 H-4로 지정된 Milwaukee 드롭 구조 설계를 기반으로 초기 설계를 기반으로했습니다.
    166 피트의 기본 낙하 길이를 포함하고 체적 유량, 벽, 대칭 및 기타 초기 매개 변수를 지정하는 FLOW-3D 분석을 설정합니다.
    그는 우리가 CFD를 통해 발견한 것은 밀워키에서 이 디자인을 사용하면 우리의 어플리케이션에 잘 맞지 않는다는 것이라고 말합니다. FLOW-3D는 이것을 보여 주고 있었기 때문에 CFD를 사용하여 변형을 시도하고 우리의 수정된 디자인을 고안했습니다.
    더 넓은 접근 경로, 더 넓은 테이퍼 및/또는 더 깊은 테이퍼 깊이를 사용한 수정은 에디슨은 FLOW-3D에서 각 변동 사항을 설정하는 것이 매우 빠르다고 말합니다. (그림 3,4,5). 개선의 진전은 고무적이었습니다. 시뮬레이션 결과의 높은 수준은 심지어 절삭(침식)을 개선하기 위해 드롭 축의 바닥에 의문스러운 플레이트가 수직 흐름이 수평으로 전환되는 난류 분리 및 감소가되도록 기능을 추가하도록 설득했습니다.

    Figs. 3, 4 and 5. Tangential drop structure flow simulated with FLOW-3D. Structure dimensions were optimized through multiple design iterations. (Image courtesy AECOM)

    9번째 설계 변동에 대한 FLOW-3D 출력 동작인 V9는 접근 섹션을 확장했으며, 모든 흐름 볼륨 레벨에서 300mg/d까지 양호한 흐름 안정성을 보였으며 유압식 점프는 없었습니다. 그리고 양호한 Froude numners(유체 움직임에 미치는 중력의 영향을 나타내기 위해 사용되는 치수 없는 수량), 2010년 2월부터 AECOM이 물리적 시험과 검증을 위해 선택하였습니다(그림 6). 그 계획은 아이오와 연구소의 시험 결과에 기초하여 CFD와 최적화를 추가하는 것이였습니다.

    Fig. 6. Scale model (1:10) of vertical drop structure, tested at University of Iowa IIHR Hydroscience & Engineering facility. (Image courtesy AECOM)

    에디슨은 V9에서 결정된 치수 매개 변수에 대해 그 디자인을 아이오와 주에 가져가서 CFD를 이용해 만들었는데 완벽하게 작동했습니다. (II.)직원들은 실제로 무언가를 설치한 것은 이번이 처음이며, 변경하라고 말할 만한 것이 아무것도 없다고 말했습니다. 측정된 데이터는 드롭 샤프트 연결 구조 내의 수면 높이, Adit내 공기 침투의 정량, 벤트 샤프트 위로 공기 흐름을 포함했습니다. 흐름이 증가함에 따라 와류량이 증가함에 따라 축 벽에 부착되어 탈산소까지 원활하게 회전하는 모습이 포착되었습니다(그림 7).

    에디슨은 후속 실험을 위해 여러번 시험장을 돌아다녔습니다. 물리적 모델이 처음부터 올바르게 작동했기 때문에 시험 프로그램을 확장할 시간이 있었습니다. “재미 있는 것은 환기구를 움직이는 것과 같이 우리가 궁금했던 것들을 탐구해서 지적으로 그것을 가지고 놀 시간이 있었다는 것입니다.” 에디슨은 예정보다 앞서 있었기 때문에 잔여 프로젝트 시간을 이용해 탈염소와 adit 내의 유압 장치를 조사할 수 있었습니다.

    Fig. 7. Operation of scale-model vertical drop structure, showing test run of 300 million gallons per day (mgd). Flow vortex development shows good rotation and attachment to the shaft wall all the way down to the de-aeration chamber. No design modifications were necessary to the simulated design. (Image courtesy AECOM)

    Final Results

    AECOM은 2010년 7월 DRTC에 대한 전반적인 작업을 마쳤습니다. 2013년 3월부터 18구경 터널을 굴착하기 시작했고, CSO드롭 구조 3개(CFD로 설계된 나머지 2개의 구조물만 있음)는 모두 현재 공사 중입니다.

    에디슨의 의견으로는, 토목 공학은 전체적으로 CFD를 채택하는 데 느린 편이었습니다. 이를 입증하기 위해 그는 인천 국제 공항을 처음 방문한 당시 접선 소용돌이 모형의 소위 “묘지”에서 본것을 기술했습니다. 그러나 그는 이들을 다시 처리해야 했다고 말했습니다.  그는 유압 설계를 위한 시뮬레이션 사용으로 판매되는 것을 권장하고 있습니다.

    에디슨은 DRTC노력을 요약하면서 “정말 재미 있었습니다. 물리적 모델링이 필요한 위치에 대해 더 자세히 알아보았고, 그렇다면 어떤 경우에는 순수한 RAID기반 설계를 수행할 수 있습니다. 많은 DRTC작업들이 그것의 증거입니다. 물리적 모델은 실제로 필요하지 않았지만 검증을 통해 위험을 줄일 수 있었습니다. 프로젝트에서 이 두가지를 모두 수행할 수 있었다는 것은 믿을 수 없는 일입니다.”라고 말했습니다.

    This article first appeared in WaterWorld Magazine.

    CFD + Physical Modeling Results

    CFD + Physical Modeling Results

    This material was provided by Kevin Sydor, M.Sc., P.Eng., Section Head, Hydrotechnical and Oceanographic Studies, Water Resources Engineering; Manitoba Hydro; Joe Groeneveld, Western Canada Discipline Practice Lead – Hydrotechnical, Hatch Ltd.; Graham Holder, Consultant, LaSalle; D.G. Murray, P.Eng., M.Sc., Discipline Practice Lead – Hydrotechnical, Hatch Ltd.

     

    10년이 넘는 기간 동안 Manitoba Hydro는 Flow-3D의 힘으로 수력 발전소 설계의 복잡성을 해결해 왔습니다. 최근 Manitoba Hydro는 급류, 다중 채널, 그리고 natural contours을 포함한 복잡한 장소에서 제안된 Keeyask생성에 대한 사전연구에 집중해 왔습니다. FLOW-3D사용 이전에는 초기 설계를 토대로 시뮬레이션과 물리적 모델링의 결합 결과가 서로의 성능을 검증하고 향상시키는 통합 연구를 수행했습니다.

    Water velocities (m/s) as determined in CFD simulation at left, compared with photo of physical model in operation at right, for Stage 1 Cofferdam operation at a construction length of 450m.

     

    실제 발전소와 제철소를 건설하기 위해서는 두 단계의 강 유역이 필요했습니다. Manitoba Hydro는 임시 코퍼 댐 건설 중 물리적 조건이 변화함에 따라 다양한 지역에서의 수위와 속도가 어떻게 변할 것인지를 추정하는 시뮬레이션을 수행했습니다. 그런 다음, 그들은 연안 공사, 우회 구조, 하천 폐쇄 및 배수로의 1/120 축척모델에서 측정된 결과와, 배수로 구역의 1/50 축척모델에서 측정된 결과를 비교했습니다. 1/120 축척모델의 연산에서 관찰된 수치는 수정되었고, CFD시뮬레이션 내 경계 조건을 나타내는 STL모델의 변경 사항으로 세부 사항이 피드백 되었습니다. 여러 가지의 상세한 공정은 물리적 축척 모형의 거동을 약 5%이내에서 예측했을 뿐만 아니라 공사비를 절감할 수 있는 설계 변경 사항도 찾아냈습니다.

     

    Setting up and Calibrating the CFD Model

    Simulation of final Keeyask spillway structure, verifying water velocities (m/s) to compare with physical scale model operation.

     

    CFD모델은 약 3km x 2km의 영역을 커버하였으며, 탐지 속도 경계로 설정된 경계 조건을 통해 상류 쪽으로의 흐름을 제어하고 하류 쪽 끝의 지속적인 유출 경계를 설정하였습니다. 설계자들은 교각, 교대, 여수로 구조 및 코퍼댐과 같은 기하학적 객체의 STL AutoCAD파일을 가져와 물리적 경계를 나타낸 다음 매개 변수를 정의 했습니다.

    강 급류의 특성과 레일 통로의 평행 부분을 통해 생성되는 예상 유량 범위를 모두 수용하기 위해 CFD모델이 다시 정규화되도록 설정되었습니다. 일반화된 최소 잔류 방법에 기초한 회전 난류 모드 및 implicit의 압력-속도 솔버를 설정했습니다. 메쉬는 데카르트 좌표로 설정되었고 보다 정밀한 메쉬 처리가 필요한 영역에서 grid를 다듬기 위해 중첩된 메쉬 블록을 사용했습니다. 배수로 구조 주위 영역의 격자 간격은 1m x 1m x 1m로 설정되었습니다. 즉, 배수로 및 배수로 용마루의 형상을 포함하는 데 필요했던 것입니다.

    시뮬레이션의 목적은 건설 일정상 다양한 지점에서의 방전 용량, 수위, 속도 및 흐름 패턴, 다양한 위치, 경로 게이트(부분에서 완전히 열림)등을 추정하는 것이었습니다.  이 계산된 값들은 코페르담 건축에서 암석 덩어리에 필요한 돌의 크기를 결정하는데 중요합니다. 건축의 모든 단계에서 그들을 제자리와 하류로 이동시키는 항력에 저항할 만큼 암석들은 충분히 커야 합니다.

    Excellent agreement in flow-rate prediction of spillway behavior between numeric and measured physical model values.

     

    Physical Modeling

    수력 발전소 설계로 인해 처음에는 제대로 하지 못하는 일이 너무 많습니다. 중요한 지형에서의 용량, 압력, 속도 및 배수로 게이트 동작(완전 개방)을 검증하기 위해서는 중요한 흐름 영역의 규모 물리적 모델을 구축해야 합니다. Manitoba Hydro는 LaSalle Consulting Group에 1/120 스케일의 하나의 포괄적 인 레이아웃과 1/50 스케일의 2 개의 전체구역과 2 개의 절반 구역을 가진 방수 모델을 구축할 것을 요청했습니다.

     

    Integrated Modeling Results

    실제 모델의 크기에 대한 힘의 이동 관계를 살펴보면, 바위 크기 예측에 대한 시뮬레이션은 약간 보존적입니다. 그러나 초기 수위 데이터 곡선은 시뮬레이션과 물리적 모델 행동 사이에 일치를 보여주었고 추가 시험을 위한 단계를 설정했습니다. 모델에서 코퍼댐이 서서히 생성됨에 따라 후속 수위 CFD시뮬레이션 결과를 정확하게 예측했음을 보여주었습니다.

    완성된 코퍼댐의 테스트에 따르면 제어 구조가 아닌 채널 입구에서 흐름이 제어되고 있는 것으로 나타났습니다. 이것은 원하는 것보다 높은 상류수위가 나타났습니다. 접근 채널의 입구를 낮추도록 물리적 모델을 재구성하여 CFD에 사용된 고도를 반영했습니다. 출입구가 더 낮게 발굴되어, 수로의 왼쪽 둑을 따라 굴착하는 것은 입구 근처의 작은 지역에서만 필요했습니다.

     

    Conclusion

    Manitoba Hydro는 CFD모델링이 미래의 수력 발전소뿐만 아니라 Keeyask 발전소의 건설과 운영을 계획하는 데에도 여러 가지 이점을 제공한다는 사실을 발견했습니다 두 가지 접근법의 결과간에 매우 잘 일치했을 뿐만 아니라 FLOW-3D 시뮬레이션과 스케일 모델 테스트를 결합하면 두 가지 설계 옵션의 유효성을 개선하는 반복적 인 방법이 제공되었습니다. 또한 시뮬레이션을 통해 사용자는 실제 사용할 수 있는 값의 수가 제한되어 있지 않고 CFD모델 도메인 내의 어디서나 속도, 수위 및 유량을 쉽고 빠르게 추출할 수 있습니다.

    Topo2STL을 이용한 지형 파일로 stl 생성방법

    유틸리티 Topo2STL을 이용한 지형 도면 파일에서 3차원 모델링 STL로 생성하는 방법

    FLOW-3D 해석 모델을 작성할 때 별도의 3차원 모델러 없이, FLOW-3D 의 유틸리티인 Topo2STL을 이용하여 지형 도면 파일을 3차원 모델링 파일인 STL로 생성하는 방법을 간략히 설명하고자 한다.

    정의 방법

    1) 지형 도면 파일(*.dwg)을 DraftSight(오토캐드도 가능함.)에서 불러온다.

    2) 파일>다른이름으로 저장>파일형식을 *.dxf 파일로 저장한다.

    3. CAD2Shape 프로그램을 이용하여 저장한 *.dxf 파일을 불러온다.

    (CAD2Shape 다운로드 경로 : http://www.guthcad.com/cad2shape.htm)

    4. Filters>Entities에서 Points, Lines, Polylines 선택한다.

    5. Output Fomat > Output Data Fomat에서 다른 불필요한 정보 없이 Raw로 선택한다.

    6. *.xyz 로 저장한다.

    7. 각 좌표간에 쉼표를 없애기 위해 엑셀에서 *.xyz 파일을 열어 탭으로 분리된 *.txt 로 저장한다.

     

    8. 저장한 *.txt 파일의 확장자를 *.topo로 변경한다.

    9. FLOW-3D를 실행하여 Utilities>Topo2STL 을 실행하거나, 시작프로그램>FLOW-3D v11.x에서 Topo2STL을 실행한다.

       

    10. *.topo 파일을 연다.

    11. 정보를 확인 후 *.stl로 convert 한다.

    12. convert가 완료되면 창을 닫고 *.stl 파일을 확인한다.

    FLOW-3D CAST 사양

    FLOW-3D CAST Feature


    Active Simulation Control

    실행중인 해석의 제어 파라미터는 History probes에서 사용자가 정의한 조건에 따라, 런타임 동안에 자동으로 변경 될 수 있습니다. History probes에 의해 기록된 시뮬레이션 변수는 경계 조건, mass source 및 General Moving Object 기능을 이용하여, 시간에 따른 개체의 동작을 제어하기 위해 사용될 수있습니다. 예를 들어, 고압다이캐스팅 해석에서 게이트에 설정한 History probes에 유체가 도달하면, 그 정보를 캡처하는 데이터 출력 주파수를 증가시켜 플런저의 속도를 고속으로 자동 전환 될 수있습니다. 고압다이캐스팅 해석은 유체가 게이트에 도달 할 때 자동으로 고속 전환됩니다. 이 프로세스는 새로운 실행 시뮬레이션 제어 기능을 통해 자동으로 진행됩니다. 저속 구간에서 플런저의 움직임은 trigger 슬리브의 용융물에 혼입되는 공기의 양을 최소화하기 위해 Barkhudarov 방법 1을 사용하여 계산됩니다. 이 결과는 훨씬 더 높은 품질의 주조품이 나올수 있도록 설계하는데 도움이 될 수 있습니다. Read the development note > Read the blog post >

    Batch Postprocessing & Report Generation

    Batch 후처리 및 보고서 생성은 해석 결과 분석시 사용자의 해석 처리 시간을 절약하기 위해 개발되었습니다. Batch 후처리는, 해석이 완료된 후, 사용자가 애니메이션, 시나리오, 그래프, 텍스트 데이터 시리즈를 정의하여 자동으로 생성되도록 할 수 있습니다. 그래픽 요청은 백그라운드에서 FlowSight를 실행하여 처리되도록 FLOW-3D Cast에 정의되어 있습니다. 원하는 해석 결과를 생성할 수 있는 컨텍스트 파일을 사용하면 Batch 후처리 기능을 사용할 수 있습니다. Batch 후처리가 완료되면, 사용자는 쉽게 자신의 관리자, 동료, 또는 클라이언트에 보낼 수있는 HTML5 형식의 완벽한 기능을 갖춘 보고서를 만들 수 있습니다. 이미지 및 동영상도 보고서에 포함 할 수 있고, 사용자는 텍스트, 캡션, 참고 문헌의 형식을 완벽하게 제어 하고 유지할 수 있습니다. Read the blog post >

    Metal Casting Models

    Squeeze Pin Model

    스퀴즈 핀은 주조시 주입 공급이 어려운 영역에서, 응고하는 동안 금속 수축을 보상하기 위해 사용되는 실제의 다이 캐스팅 머신의 동작을 모델링하는 해석을 할 수 있습니다. 스퀴즈 핀은 선택된 표면에 cylinderical squeeze pin을 추가하여, STL 파일 또는 대화식으로 생성 될 수 있습니다. Read the development note >

    Intensification Pressure Model

    새로운 플런저 타입 형상이 추가 되었습니다. 강화된 압력 조건으로 macro-shrinkage 와 micro-porosity 제거를 지정할 수 있습니다.

    Thermal Die Cycling model

    FLOW-3D Cast v4.1's full process thermal die cycling model

    다이싸이클링 (Thermal die cycling, TDC) 모델에 새로운 두 가지의 단계가 추가되었습니다. 금형이 열린 상태에서 제품이 여전히 금형 내부에 있는 ejection 단계와, 금형이 닫혔지만 사출 바로전의 preparation 단계가 추가되었습니다. 또한, 마지막 싸이클만이 아닌 모든 금형 싸이클 모두 수렴된 결과를 전달하기 위해 TDC 솔버가 성능 손실 없이 최적화 되었습니다. Read the blog post >

    Valves and Vents

    Modeling valves and vents in FLOW-3D Cast v4.1

    밸브와 밴트의 외부 압력과 온도는 이제 사용자가 다이 캐스팅 공정에서 충진중에 보다 실제적인 동작을 정의 할 수 있도록, 시간의 표 함수로서 정의 할 수있습니다. 밸브 및 벤트의 압력 및 온도는 프로세스 설계 단계에서 유용한 제품 내부에 설정된 프로브에 의해 제어 될 수 있습니다.

    PQ2 Diagram

    PQ2다이어그램의 사용은 사용자가 더 나은 슬리브의 플런저 실제 움직임과 유사하게 적용 할 수 있습니다. 새로운 기능은 실제 공정 변수가 아직 알려져 있지 않았을 때 다이캐스팅 설계 단계 중에 특히 유용합니다. Read the blog post >

    Cooling Channels

    냉각 채널은 금형 각각의 냉각 유로에 의해 제거되거나 추가된 열의 총량에 의해 제어 될 수 있습니다. Read the development note >

    Air Entrainment Model

    Air entrainment 모델에 compressibility를 입력하는 새로운 옵션이 추가되었습니다. 고압 다이캐스팅의 충진 공정과 같은 경우, 공기 압축성은 유체 압력의 변화로 인한 유체의 흐름에 중요한 인자가 됩니다.
     

    Cavitation Model

    캐비테이션 모델은 유동 조건의 더 넓은 범위에 걸쳐 유체의 캐비테이션 거동을 나타내도록 개선되었습니다. 캐비테이션 생성에 대한 새로운 옵션은 경험적 관계를 기반으로, 기존의 일정한 속도로 생성되는 방식에서 보완되었습니다. 새로운 passive gas model 옵션은 open bubbles이 아닌 유체내에 cavitationg gas를 추적하여, 계산에 필요한 격자와 계산시간을 줄일 수 있습니다. Read the development note >

    Two-fluid Phase Change Model

    Two-fluid phase change model 은 과냉각을 포함하도록 확장되었습니다. 일정한 과냉각 온도를 정의하고 가스 온도가 응축이 일어나기 전에 포화점 이하로 내려갈 수 있게 함으로써 구현됩니다.

    Simulation Results and Analysis

    Simulation Results File Editor

    사용자가 FLOW-3D Cast v4.1 결과 파일들을 병합 및 제거 할 수 있는 편집 유틸리티

    Linking flsgrf.* files

    Restart 해석 결과 파일들(flsgrf.*)은 FlowSight 에서 하나의 연속적인 애니메이션 결과를 표시하기 위해 restart source 결과로 링크될 수 있습니다.

    Fluid/wall Contact Time

    A new spatial quantity has been added to the solution output that stores the time that metal spent in contact with each geometric component, as well as the time spent by each component with metal.

    용탕이 각 geometry 컴포넌트를 접촉한 시간과 각 컴포넌트가 용탕과의 접촉 시간을 나타내는 새로운 공간적 양이 해석 아웃풋에 추가 되었습니다.

    Performance and Usability

    Calculators

    열전달 계수, 열 침투 깊이, 밸브 손실 계수, 슬리브에 용탕량(깊이), 플런저의 속도를 계산할 수 있는 Calculators 기능이 Model Setup 창에서 바로 가능해졌습니다. 또한 유틸리티 메뉴에서도 가능합니다.

    Thermal Die Cycling

    Heat transfer database in FLOW-3D Cast v4.1

    열전달 계수 데이터베이스와 각 싸이클 단계들이 입력되어있어 간편하게 다이싸이클링 해석을 하실 수 있습니다.

    GMRES Pressure Solver

    GMRES pressure solver의 속도가 솔버 데이터 구조의 최적화로 인해 2배까지 향상되었습니다. 이로 인해 메모리 사용량이 20% 미만으로 증가할 수 있습니다. Read the blog post >

    Sampling Volumes

    Sampling volume 기능은 STL로 정의할 수 있습니다. 각 sampling volume에 의해 계산된 양들의 목록은 유체의 부피, 최대/최소 온도, 파티클의 갯수와 같은 전체 해석 영역에 대해 모두 같은 양이 되도록 확장되었습니다.

     

    FSI/TSE Model

    구조분석 모델의 성능이 부분적인 coupling으로 해석 솔버의 병렬화와 최적화를 통해 향상되었습니다.

    Workspaces

    Workspaces 를 이전에 설치된 FLOW-3D에서 가져올 수 있습니다. Workspaces 와 사용자가 선택한 시뮬레이션들을 복사할 수 있습니다.

    Expanded Simulation Pre-check

    Simulation pre-check 기능은 preprocessor checks를 포함하고, 문제가 발생하는 경우 링크됩니다.

    Improved Transparency

    Depth-peeling 옵션은 transparent geometries 를 좀 더 잘 표현하고, v4.0보다 10배 빨라졌습니다.

    Interactive Tools

    Baffles, history probes, void/fluid pointers, valves, mass-momentum sources, squeeze pins에 대한 새로운 대화형 생성 기능이 추가되었습니다. 또한 probing과 clipping 도구들이 대화형으로 개선되었습니다.

    General Enable/Disable

    모든 objects (e.g., mesh blocks)은 활성화/비활성화 할 수 있습니다.

    Estimated Remaining Simulation Time

    솔버 메세지 파일에 short-print로 추정된 잔여 해석 시간이 추가 되었습니다.

    Tabular Data

    테이블 형식의 데이터에서 선택된 데이터를 마우스 오른쪽 버튼을 클릭하여 csv파일 또는 외부 파일에 복사, 저장할 수 있습니다.

    1 23-10 Michael R. Barkhudarov, Minimizing Air Entrainment, The Canadian Die Caster, June 2010

    FLOW-3D 제품소개

    About FLOW-3D


    FLOW-3D 2022R2
    FLOW-3D 2022R2

    FLOW-3D 개발 회사

    Flow Science Inc Logo Green.svg
    IndustryComputational Fluid Dynamics Software
    Founded1980
    FounderDr. C.W. “Tony” Hirt
    Headquarters
    Santa Fe, New Mexico, USA
    United States
    Key people
    Dr. Amir Isfahani, President & CEO
    ProductsFLOW-3D, FLOW-3D CAST, FLOW-3D AM, FLOW-3D CLOUD, FlowSight
    ServicesCFD consultation and services

    FLOW-3D 개요

    FLOW-3D는 미국 뉴멕시코주(New Mexico) 로스알라모스(Los Alamos)에 있는 Flow Scicence, Inc에서 개발한 범용 전산유체역학(Computational Fluid Dynamics) 프로그램입니다. 로스알라모스 국립연구소의 수치유체역학 연구실에서 F.Harlow, B. Nichols 및 T.Hirt 등에 의해 개발된 MAC(Marker and Cell) 방법과 SOLA-VOF 방식을 기초로 하여, Hirt 박사가 1980년에 Flow Science, Inc사를 설립하여 계속 프로그램을 발전시켰으며 1985년부터 FLOW-3D를 전세계에 배포하였습니다.

    유체의 3차원 거동 해석을 수행하는데 사용되는 CFD모형은 몇몇 있으나, 유동해석에 적용할 물리모델 선정은 해석의 정밀도와 밀접한 관계가 있으므로, 해석하고자 하는 대상의 유동 특성을 분석하여 신중하게 결정하여야 합니다.

    FLOW-3D는 자유표면(Free Surface) 해석에 있어서 매우 정확한 해석 결과를 제공합니다. 해석방법은 자유표면을 포함한 비정상 유동 상태를 기본으로 하며, 연속방정식, 3차원 운동량 보전방정식(Navier-Stokes eq.) 및 에너지 보존방정식 등을 적용할 수 있습니다.

    FLOW-3D는 유한차분법을 사용하고 있으며, 유한요소법(FEM, Finite Element Method), 경계요소법(Boundary Element Method)등을 포함하여 자유표면을 포함하는 유동장 해석(Fluid Flow Analysis)에서 공기와 액체의 경계면을 정밀하게 표현 가능합니다.

    유체의 난류 해석에 대해서는 혼합길이 모형, 난류 에너지 모형, RNG(Renormalized Group Theory)  k-ε 모형, k-ω 모형, LES 모형 등 6개 모형을 적용할 수 있으며, 자유표면 해석을 위하여 VOF(Volume of Fluid) 방정식을 사용하고, 격자 생성시 사용자가 가장 쉽게 만들 수 있는 직각형상격자는 형상을 더욱 정확하게 표현하기 위해 FAVOR(Fractional Area Volume Obstacle Representation) 기법을 각 방정식에 적용하고 있습니다.

    FLOW-3D는 비압축성(Incompressible Fluid Flow), 압축성 유체(Compressible Fluid Flow)의 유동현상 뿐만 아니라 고체와의 열전달 현상을 해석할 수 있으며, 비정상 상태의 해석을 기본으로 합니다.

    FLOW-3D v12.0은 모델 설정을 간소화하고 사용자 워크 플로우를 개선하는 GUI(그래픽 사용자 인터페이스)의 설계 및 기능에 있어 중요한 변화를 가져왔습니다. 최첨단 Immersed Boundary Method는 FLOW-3Dv12.0솔루션의 정확도를 높여 줍니다. 다른 특징적인 주요 개발에는 슬러지 안착 모델, 2-유체 2-온도 모델, 사용자가 자유 표면 흐름을 훨씬 더 빠르게 모델링 할 수 있는 Steady State Accelerator등이 있습니다.

    물리 및 수치 모델

    Immersed Boundary Method

    힘과 에너지 손실에 대한 정확한 예측은 솔리드 바디 주변의 흐름과 관련된 많은 엔지니어링 문제를 모델링하는 데 중요합니다. FLOW-3D v12.0의 릴리스에는 이러한 문제 해결을 위해 설계된 새로운 고스트 셀 기반 Immersed Boundary Method (IBM)가 포함되어 있습니다. IBM은 내부 및 외부 흐름을 위해 벽 근처 해석을 위해 보다 정확한 솔루션을 제공하여 드래그 앤 리프트 힘의 계산을 개선합니다.

    Two-field temperature for the two-fluid model

    2유체 열 전달 모델은 각 유체에 대한 에너지 전달 공식을 분리하도록 확장되었습니다. 이제 각 유체에는 고유한 온도 변수가 있어 인터페이스 근처의 열 및 물질 전달 솔루션의 정확도를 향상시킵니다. 인터페이스에서의 열 전달은 시간의 표 함수가 될 수 있는 사용자 정의 열 전달 계수에 의해 제어됩니다.

    슬러지 침전 모델 / Sludge settling model

    중요 추가 기능인 새로운 슬러지 침전 모델은 도시 수처리 시설물 응용 분야에 사용하면 수처리 탱크 및 정화기의 고형 폐기물 역학을 모델링 할 수 있습니다. 침전 속도가 확산된 위상의 방울 크기에 대한 함수인 드리프트-플럭스 모델과 달리, 침전 속도는 슬러지 농도의 함수이며 기능적인 형태와 표 형태로 모두 입력 할 수 있습니다.

    Steady-state accelerator for free surface flows

    이름이 암시하듯이, 정상 상태 가속기는 안정된 상태의 솔루션에 대한 접근을 가속화합니다. 이는 작은 진폭의 중력과 모세관 현상을 감쇠하여 이루어지며 자유 표면 흐름에만 적용됩니다.

    꾸준한 상태 가속기

    Void particles

    보이드 입자가 버블 및 위상 변경 모델에 추가되었습니다. 보이드 입자는 항력과 압력 힘을 통해 유체와 상호 작용하는 작은 기포의 역할을 하는 붕괴된 보이드 영역을 나타냅니다. 주변 유체 압력에 따라 크기가 변경되고 시뮬레이션이 끝난 후 최종 위치는 공기 침투 가능성을 나타냅니다.

    Sediment scour model

    침전물의 정확성과 안정성을 향상시키기 위해 침전물의 운반과 침식 모델을 정밀 조사하였다. 특히, 침전물 종에 대한 질량 보존이 크게 개선되었습니다.

    Outflow pressure boundary condition

    고정 압력 경계 조건에는 이제 압력 및 유체 비율을 제외한 모든 유량이 해당 경계의 상류에 있는 흐름 조건을 반영하는 ‘유출’ 옵션이 포함됩니다. 유출 압력 경계 조건은 고정 압력 및 연속성 경계 조건의 혼합입니다.

    Moving particle sources

    시뮬레이션 중에 입자 소스는 이동할 수 있습니다. 시간에 따른 변환 및 회전 속도는 표 형식으로 정의됩니다. 입자 소스의 운동은 소스에서 방출 된 입자의 초기 속도에 추가됩니다.

    Variable center of gravity

    중력 및 비 관성 기준 프레임 모델에서 시간 함수로서의 무게 중심의 위치는 외부 파일의 표로 정의할 수 있습니다. 이 기능은 연료를 소모하는 로켓을 모델링하고 단계를 분리할 때 유용합니다.

    공기 유입 모델

    가장 간단한 부피 기반 공기 유입 모델 옵션이 기존 질량 기반 모델로 대체되었습니다.  질량 기반 모델은 부피와 달리 주변 유체 압력에 따라 부피가 변화하는 동안 흡입된 공기량이 보존되기 때문에 물리학적 모델입니다.

    Air entrainment model in FLOW-3D v12.0

    Tracer diffusion / 트레이서 확산

    유동 표면에서 생성된 추적 물질은 분자 및 난류 확산 과정에 의해 확산될 수 있으며, 예를 들어 실제 오염 물질의 거동을 모방합니다.

    모델 설정

    시뮬레이션 단위

    이제 온도를 포함하여 단위계 시스템을 완전히 정의해야 합니다. 표준 단위 시스템이 제공됩니다. 또한 사용자는 선택한 옵션에서 질량, 시간 및 길이 단위를 정의하여 편리하며, 사용자 정의된 단위를 사용할 수 있습니다. 사용자는 또한 압력이 게이지 단위로 정의되는지 절대 단위로 정의되는지 여부를 지정해야 합니다. 기본 시뮬레이션 단위는 Preferences(기본 설정)에서 설정할 수 있습니다. 단위를 완벽하게 정의하면 FLOW-3D는 물리적 수량에 대한 기본 값을 정의하고 범용 상수를 설정할 수 있으므로 사용자가 필요로 하는 작업량을 최소화할 수 있습니다.

    Shallow water model

    천수(shallow water) 모델에서 매닝의 거칠기

    Manning의 거칠기 계수는 지형 표면의 전단 응력 평가를 위해 천수(shallow water) 모델에서 구현되었습니다. 표면 결함의 크기를 기반으로 기존 거칠기 모델을 보완하며이 모델과 함께 사용할 수 있습니다. 표준 거칠기와 마찬가지로 매닝 계수는 구성 요소 또는 하위 구성 요소의 속성이거나 지형 래스터 데이터 세트에서 가져올 수 있습니다.

    메시 생성

    하단 및 상단 경계 좌표의 정의만으로 수직 방향의 메시 설정이 단순화되었습니다.

    구성 요소 변환

    사용자는 이제 여러 하위 구성 요소로 구성된 구성 요소에 회전, 변환 및 스케일링 변환을 적용하여 복잡한 형상 어셈블리 설정 프로세스를 단순화 할 수 있습니다. GMO (General Moving Object) 구성 요소의 경우, 이러한 변환을 구성 요소의 대칭 축과 정렬되도록 신체에 맞는 좌표계에 적용 할 수 있습니다.

    런타임시 스레드 수 변경

    시뮬레이션 중에 솔버가 사용하는 스레드 수를 변경하는 기능이 런타임 옵션 대화 상자에 추가되어 사용 가능한 스레드를 추가하거나 다른 태스크에 자원이 필요한 경우 스레드 수를 줄일 수 있습니다.

    프로브 제어 열원

    활성 시뮬레이션 제어가 형상 구성 요소와 관련된 heat sources로 확장되었습니다.  history probes로 열 방출을 제어 할 수 있습니다.

    소스에서 시간에 따른 온도

    질량 및 질량/모멘트 소스의 유체 온도는 이제 테이블 입력을 사용하여 시간의 함수로 정의 할 수 있습니다.

    방사율 계수

    공극으로의 복사 열 전달을위한 방사율 계수는 이제 사용자가 방사율과 스테판-볼츠만 상수를 지정하도록 요구하지 않고 직접 정의됩니다. 후자는 이제 단위 시스템을 기반으로 솔버에 의해 자동으로 설정됩니다.

    Output

    • 등속 필드 솔버 옵션을 사용할 때 유량 속도를 선택한 데이터로 출력 할 수 있습니다.
    • 벽 접착력으로 인한 지오메트리 구성 요소의 토크는 기존 벽 접착력 출력과 함께 별도의 수량으로 일반 이력 데이터에 출력됩니다.
    • 난류 모델 출력이 요청 될 때 난류 에너지 및 소산과 함께 전단 속도 및 y +가 선택된 데이터로 자동 출력됩니다.
    • 공기 유입 모델 출력에 몇 가지 수량이 추가되었습니다. 자유 표면을 포함하는 모든 셀에서 혼입 된 공기 및 빠져 나가는 공기의 체적 플럭스가 재시작 및 선택된 데이터로 출력되어 사용자에게 공기가 혼입 및 탈선되는 위치 및 시간에 대한 자세한 정보를 제공합니다. 전체 계산 영역 및 각 샘플링 볼륨 에 대해이 두 수량의 시간 및 공간 통합 등가물이 일반 히스토리 로 출력됩니다.
    • 솔버의 출력 파일 flsgrf 의 최종 크기는 시뮬레이션이 끝날 때 보고됩니다.
    • 2 유체 시뮬레이션의 경우, 기존의 출력 수량 유체 체류 시간 및 유체 가 이동 한 거리는 이제 유체 # 1 및 # 2와 유체의 혼합물에 대해 별도로 계산됩니다.
    • 질량 입자의 경우, 각 종의 총 부피 및 질량이 계산되어 전체 계산 영역, 샘플링 볼륨 및 플럭스 표면에 대한 일반 히스토리 로 출력되어 입자 종 수에 대한 현재 출력을 보완합니다.
    • 최종 로컬 가스 압력 은 사용자가 가스 포획을 식별하고 연료 탱크의 배기 시스템 설계를 지원하는 데 도움이되는 선택적 출력량으로 추가되었습니다. 이 양은 유체로 채워지기 전에 셀의 마지막 공극 압력을 기록하며 단열 버블 모델과 함께 사용됩니다.

    새로운 맞춤형 소스 루틴

    새로운 사용자 정의 가능 소스 루틴이 추가되었으며 사용자의 개발 환경에서 액세스 할 수 있습니다.

    소스 루틴 이름기술
    cav_prod_calCavitation 생성과 소산 비율
    sldg_uset슬러지 침전 속도
    phchg_mass_flux증발 및 응축으로 인한 질량 플럭스
    flhtccl유체 # 1과 # 2 사이의 열전달 계수
    dsize_cal2 상 흐름에서 동적 액적 크기 모델의 응집 및 분해 속도
    elstc_custom점탄성 유체에 대한 응력 방정식의 Source Terms

    새로운 사용자 인터페이스

    FLOW-3D 사용자 인터페이스는 완전히 새롭게 디자인되어 현대적이고 평평한 구조로 사용자의 작업 흐름을 획기적으로 간소화합니다.

    Setup dock widgets

    Physics, Fluids, Mesh 및 FAVOR ™를 포함한 모든 설정 작업이 지오 메트리 윈도우 주변에서 독 위젯으로 변환되어 모델 설정을 단일 탭으로 요약할 수 있습니다. 이러한 전환으로 인해 이전 버전의 복잡한 접이식 트리가 훨씬 깨끗하고 효율적인 메뉴 프레젠테이션으로 대체되어 사용자는 ModelSetup탭을 떠나지 않고도 모든 매개 변수에 쉽게 액세스 할 수 있습니다.

    New Model Setup icons

    새로운 모델 설정 디자인에는 설정 프로세스의 각 단계를 나타내는 새로운 아이콘이 있습니다.

    Model setup icons - FLOW-3D v12.0

    New Physics icons

    RSS feed

    새 RSS 피드부터 FLOW-3D v12.0의 시뮬레이션 관리자 탭이 개선되었습니다. FLOW-3D 를 시작하면 사용자에게 Flow Science의 최신 뉴스, 이벤트 및 블로그 게시물이 표시됩니다.

    RSS feed - FLOW-3D

    Configurable simulation monitor

    시뮬레이션을 실행할 때 중요한 작업은 모니터링입니다. FLOW-3Dv1.0에서는 사용자가 시뮬레이션을 더 잘 모니터링할 수 있도록 SimulationManager의 플로팅 기능이 향상되었습니다. 사용자는 시뮬레이션 런타임 그래프를 통해 모니터링할 사용 가능한 모든 일반 기록 데이터 변수를 선택하고 각 그래프에 여러 변수를 추가할 수 있습니다. 이제 런타임에서 사용할 수 있는 일반 기록 데이터는 다음과 같습니다.

    • 최소/최대 유체 온도
    • 프로브 위치의 온도
    • 유동 표면 위치에서의 유량
    • 시뮬레이션 진단(예:시간 단계, 안정성 한계)
    출입문에 유동 표면이 있는 대형 댐
    Runtime plots of the flow rate at the gates of the large dam

    Conforming 메쉬 시각화

    용자는 이제 새로운 FAVOR ™ 독 위젯을 통해 적합한 메쉬 블록을 시각화 할 수 있습니다.Visualize conforming mesh blocks

    Large raster and STL data

    데이터를 처리하는 데 걸리는 시간 때문에 큰 지오 메트리 데이터를 처리하는 것은 수고스러울 수 있습니다. 대형 지오 메트리 데이터를 처리하는 데는 여전히 상당한 시간이 걸릴 수 있지만, FLOW-3D는 이제 이러한 대규모 데이터 세트를 백그라운드 작업으로 로드하여 사용자가 데이터를 처리하는 동안 완전히 응답하고 중단 없는 인터페이스에서 작업을 계속할 수 있습니다

    The Sedimentation Scour Model [침전 세굴(쇄굴) 모델]

    1. Introduction
    The three-dimensional sediment scour model for non-cohesive soils was first introduced to FLOW-3D in Version 8.0 to simulate sediment erosion and deposition (Brethour, 2003). It was coupled with the three-dimensional fluid dynamics and considered entrainment, drifting and settling of sediment grains. In Version 9.4 the model was improved by introducing bedload transport and multiple sediment species (Brethour and Burnham, 2010). Although applications were successfully simulated, a major limitation of the model was the approximate treatment of the interface between the packed and suspended sediments. The packed bed was represented by scalars rather than FAVORTM (Fractional Area Volume Obstacle Representation, the standard treatment for solid components in FLOW-3D). As a result, limited information about the packed bed interface was available. That made accurate calculation of bed shear stress, a critical factor determining the model accuracy, challenging.

    In this work, the 3D sediment scour model is mostly redeveloped and rewritten. The model is still fully coupled with fluid flow, allows multiple non-cohesive species and considers entrainment, deposition, bedload transport and suspended load transport. The fundamental difference from the old model is that the packed bed is described by the FAVORTM technique. At each time step, area and volume fractions describing the packed sediments are calculated throughout the domain. In the mesh cells at the bed interface, the location, orientation and area of the interface are calculated and used to determine the bed shear stress, the critical Shields parameter, the erosion rate and the bedload transport rate. Bed shear stress is evaluated using the standard wall function with consideration of bed surface roughness that is related to the median grain size d50. A sub-mesh method is developed and implemented to calculate bedload transport. Computation of erosion considers entrainment and deposition simultaneously in addition to bedload transport.

    Furthermore, a shallow-water sediment scour model is developed in this work by adapting the new 3D model. It is coupled with the 2D shallow water flows to calculate depth-averaged properties for both suspended and packed sediments. Its main differences from the 3D model are 1) the settling velocity of grains is calculated using an existing equation instead of the drift-flux approach in the 3D model, and 2) turbulent bed shear stress is calculated using a well-accepted quadratic law rather than the log wall function. The drag coefficient for the bed shear stress is either user-given or locally evaluated using the water depth and the bed surface roughness that is proportional to d50 of the bed material. The following sections present the sediment theory used in the model and application and validation cases.