The coupled dynamics of interfacial fluid phases and unconstrained solid particles during the binder jet 3D printing process govern the final quality and performance of the resulting components. The present work proposes a computational fluid dynamics (CFD) and discrete element method (DEM) framework capable of simulating the complex interfacial fluid–particle interaction that occurs when binder microdroplets are deposited into a powder bed. The CFD solver uses a volume-of-fluid (VOF) method for capturing liquid–gas multifluid flows and relies on block-structured adaptive mesh refinement (AMR) to localize grid refinement around evolving fluid–fluid interfaces. The DEM module resolves six degrees of freedom particle motion and accounts for particle contact, cohesion, and rolling resistance. Fully-resolved CFD-DEM coupling is achieved through a fictitious domain immersed boundary (IB) approach. An improved method for enforcing three-phase contact lines with a VOF-IB extension technique is introduced. We present several simulations of binder jet primitive formation using realistic process parameters and material properties. The DEM particle systems are experimentally calibrated to reproduce the cohesion behavior of physical nickel alloy powder feedstocks. We demonstrate the proposed model’s ability to resolve the interdependent fluid and particle dynamics underlying the process by directly comparing simulated primitive granules with one-to-one experimental counterparts obtained from an in-house validation apparatus. This computational framework provides unprecedented insight into the fundamental mechanisms of binder jet 3D printing and presents a versatile new approach for process parameter optimization and defect mitigation that avoids the inherent challenges of experiments.
바인더 젯 3D 프린팅 공정 중 계면 유체 상과 구속되지 않은 고체 입자의 결합 역학이 결과 구성 요소의 최종 품질과 성능을 좌우합니다. 본 연구는 바인더 미세액적이 분말층에 증착될 때 발생하는 복잡한 계면 유체-입자 상호작용을 시뮬레이션할 수 있는 전산유체역학(CFD) 및 이산요소법(DEM) 프레임워크를 제안합니다.
CFD 솔버는 액체-가스 다중유체 흐름을 포착하기 위해 VOF(유체량) 방법을 사용하고 블록 구조 적응형 메쉬 세분화(AMR)를 사용하여 진화하는 유체-유체 인터페이스 주위의 그리드 세분화를 국지화합니다. DEM 모듈은 6개의 자유도 입자 운동을 해결하고 입자 접촉, 응집력 및 구름 저항을 설명합니다.
완전 분해된 CFD-DEM 결합은 가상 도메인 침지 경계(IB) 접근 방식을 통해 달성됩니다. VOF-IB 확장 기술을 사용하여 3상 접촉 라인을 강화하는 향상된 방법이 도입되었습니다. 현실적인 공정 매개변수와 재료 특성을 사용하여 바인더 제트 기본 형성에 대한 여러 시뮬레이션을 제시합니다.
DEM 입자 시스템은 물리적 니켈 합금 분말 공급원료의 응집 거동을 재현하기 위해 실험적으로 보정되었습니다. 우리는 시뮬레이션된 기본 과립과 내부 검증 장치에서 얻은 일대일 실험 대응물을 직접 비교하여 프로세스의 기본이 되는 상호 의존적인 유체 및 입자 역학을 해결하는 제안된 모델의 능력을 보여줍니다.
이 계산 프레임워크는 바인더 제트 3D 프린팅의 기본 메커니즘에 대한 전례 없는 통찰력을 제공하고 실험에 내재된 문제를 피하는 공정 매개변수 최적화 및 결함 완화를 위한 다용도의 새로운 접근 방식을 제시합니다.
Introduction
Binder jet 3D printing (BJ3DP) is a powder bed additive manufacturing (AM) technology capable of fabricating geometrically complex components from advanced engineering materials, such as metallic superalloys and ultra-high temperature ceramics [1], [2]. As illustrated in Fig. 1(a), the process is comprised of many repetitive print cycles, each contributing a new cross-sectional layer on top of a preceding one to form a 3D CAD-specified geometry. The feedstock material is first delivered from a hopper to a build plate and then spread into a thin layer by a counter-rotating roller. After powder spreading, a print head containing many individual inkjet nozzles traverses over the powder bed while precisely jetting binder microdroplets onto select regions of the spread layer. Following binder deposition, the build plate lowers by a specified layer thickness, leaving a thin void space at the top of the job box that the subsequent powder layer will occupy. This cycle repeats until the full geometries are formed layer by layer. Powder bed fusion (PBF) methods follow a similar procedure, except they instead use a laser or electron beam to selectively melt and fuse the powder material. Compared to PBF, binder jetting offers several distinct advantages, including faster build rates, enhanced scalability for large production volumes, reduced machine and operational costs, and a wider selection of suitable feedstock materials [2]. However, binder jetted parts generally possess inferior mechanical properties and reduced dimensional accuracy [3]. As a result, widescale adoption of BJ3DP to fabricate high-performance, mission-critical components, such as those common to the aerospace and defense sectors, is contingent on novel process improvements and innovations [4].
A major obstacle hindering the advancement of BJ3DP is our limited understanding of how various printing parameters and material properties collectively influence the underlying physical mechanisms of the process and their effect on the resulting components. To date, the vast majority of research efforts to uncover these relationships have relied mainly on experimental approaches [5], [6], [7], [8], [9], [10], [11], [12], [13], [14], [15], [16], [17], [18], [19], which are often expensive and time-consuming and have inherent physical restrictions on what can be measured and observed. For these reasons, there is a rapidly growing interest in using computational models to circumvent the challenges of experimental investigations and facilitate a deeper understanding of the process’s fundamental phenomena. While significant progress has been made in developing and deploying numerical frameworks aimed at powder spreading [20], [21], [22], [23], [24], [25], [26], [27] and sintering [28], [29], [30], [31], [32], simulating the interfacial fluid–particle interaction (IFPI) in the binder deposition stage is still in its infancy. In their exhaustive review, Mostafaei et al. [2] point out the lack of computational models capable of resolving the coupled fluid and particle dynamics associated with binder jetting and suggest that the development of such tools is critical to further improving the process and enhancing the quality of its end-use components.
We define IFPI as a multiphase flow regime characterized by immiscible fluid phases separated by dynamic interfaces that intersect the surfaces of moving solid particles. As illustrated in Fig. 1(b), an elaborate IFPI occurs when a binder droplet impacts the powder bed in BJ3DP. The momentum transferred from the impacting droplet may cause powder compaction, cratering, and particle ejection. These ballistic disturbances can have deleterious effects on surface texture and lead to the formation of large void spaces inside the part [5], [13]. After impact, the droplet spreads laterally on the bed surface and vertically into the pore network, driven initially by inertial impact forces and then solely by capillary action [33]. Attractive capillary forces exerted on mutually wetted particles tend to draw them inward towards each other, forming a packed cluster of bound particles referred to as a primitive [34]. A single-drop primitive is the most fundamental building element of a BJ3DP part, and the interaction leading to its formation has important implications on the final part characteristics, such as its mechanical properties, resolution, and dimensional accuracy. Generally, binder droplets are deposited successively as the print head traverses over the powder bed. The traversal speed and jetting frequency are set such that consecutive droplets coalesce in the bed, creating a multi-drop primitive line instead of a single-drop primitive granule. The binder must be jetted with sufficient velocity to penetrate the powder bed deep enough to provide adequate interlayer binding; however, a higher impact velocity leads to more pronounced ballistic effects.
A computational framework equipped to simulate the interdependent fluid and particle dynamics in BJ3DP would allow for unprecedented observational and measurement capability at temporal and spatial resolutions not currently achievable by state-of-the-art imaging technology, namely synchrotron X-ray imaging [13], [14], [18], [19]. Unfortunately, BJ3DP presents significant numerical challenges that have slowed the development of suitable modeling frameworks; the most significant of which are as follows:
1.Incorporating dynamic fluid–fluid interfaces with complex topological features remains a nontrivial task for standard mesh-based CFD codes. There are two broad categories encompassing the methods used to handle interfacial flows: interface tracking and interface capturing [35]. Interface capturing techniques, such as the popular volume-of-fluid (VOF) [36] and level-set methods [37], [38], are better suited for problems with interfaces that become heavily distorted or when coalescence and fragmentation occur frequently; however, they are less accurate in resolving surface tension and boundary layer effects compared to interface tracking methods like front-tracking [39], arbitrary Lagrangian–Eulerian [40], and space–time finite element formulations [41]. Since interfacial forces become increasingly dominant at decreasing length scales, inaccurate surface tension calculations can significantly deteriorate the fidelity of IFPI simulations involving <100 μm droplets and particles.
2.Dynamic powder systems are often modeled using the discrete element method (DEM) introduced by Cundall and Strack [42]. For IFPI problems, a CFD-DEM coupling scheme is required to exchange information between the fluid and particle solvers. Fully-resolved CFD-DEM coupling suggests that the flow field around individual particle surfaces is resolved on the CFD mesh [43], [44]. In contrast, unresolved coupling volume averages the effect of the dispersed solid phase on the continuous fluid phases [45], [46], [47], [48]. Comparatively, the former is computationally expensive but provides detailed information about the IFPI in question and is more appropriate when contact line dynamics are significant. However, since the pore structure of a powder bed is convoluted and evolves with time, resolving such solid–fluid interfaces on a computational mesh presents similar challenges as fluid–fluid interfaces discussed in the previous point. Although various algorithms have been developed to deform unstructured meshes to accommodate moving solid surfaces (see Bazilevs et al. [49] for an overview of such methods), they can be prohibitively expensive when frequent topology changes require mesh regeneration rather than just modification through nodal displacement. The pore network in a powder bed undergoes many topology changes as particles come in and out of contact with each other, constantly closing and opening new flow channels. Non-body-conforming structured grid approaches that rely on immersed boundary (IB) methods to embed the particles in the flow field can be better suited for such cases [50]. Nevertheless, accurately representing these complex pore geometries on Cartesian grids requires extremely high mesh resolutions, which can impose significant computational costs.
3.Capillary effects depend on the contact angle at solid–liquid–gas intersections. Since mesh nodes do not coincide with a particle surface when using an IB method on structured grids, imposing contact angle boundary conditions at three-phase contact lines is not straightforward.
While these issues also pertain to PBF process modeling, resolving particle motion is generally less crucial for analyzing melt pool dynamics compared to primitive formation in BJ3DP. Therefore, at present, the vast majority of computational process models of PBF assume static powder beds and avoid many of the complications described above, see, e.g., [51], [52], [53], [54], [55], [56], [57], [58], [59]. Li et al. [60] presented the first 2D fully-resolved CFD-DEM simulations of the interaction between the melt pool, powder particles, surrounding gas, and metal vapor in PBF. Following this work, Yu and Zhao [61], [62] published similar melt pool IFPI simulations in 3D; however, contact line dynamics and capillary forces were not considered. Compared to PBF, relatively little work has been published regarding the computational modeling of binder deposition in BJ3DP. Employing the open-source VOF code Gerris [63], Tan [33] first simulated droplet impact on a powder bed with appropriate binder jet parameters, namely droplet size and impact velocity. However, similar to most PBF melt pool simulations described in the current literature, the powder bed was fixed in place and not allowed to respond to the interacting fluid phases. Furthermore, a simple face-centered cubic packing of non-contacting, monosized particles was considered, which does not provide a realistic pore structure for AM powder beds. Building upon this approach, we presented a framework to simulate droplet impact on static powder beds with more practical particle size distributions and packing arrangements [64]. In a study similar to [33], [64], Deng et al. [65] used the VOF capability in Ansys Fluent to examine the lateral and vertical spreading of a binder droplet impacting a fixed bimodal powder bed with body-centered packing. Li et al. [66] also adopted Fluent to conduct 2D simulations of a 100 μm diameter droplet impacting substrates with spherical roughness patterns meant to represent the surface of a simplified powder bed with monosized particles. The commercial VOF-based software FLOW-3D offers an AM module centered on process modeling of various AM technologies, including BJ3DP. However, like the above studies, particle motion is still not considered in this codebase. Ur Rehman et al. [67] employed FLOW-3D to examine microdroplet impact on a fixed stainless steel powder bed. Using OpenFOAM, Erhard et al. [68] presented simulations of different droplet impact spacings and patterns on static sand particles.
Recently, Fuchs et al. [69] introduced an impressive multipurpose smoothed particle hydrodynamics (SPH) framework capable of resolving IFPI in various AM methods, including both PBF and BJ3DP. In contrast to a combined CFD-DEM approach, this model relies entirely on SPH meshfree discretization of both the fluid and solid governing equations. The authors performed several prototype simulations demonstrating an 80 μm diameter droplet impacting an unconstrained powder bed at different speeds. While the powder bed responds to the hydrodynamic forces imparted by the impacting droplet, the particle motion is inconsistent with experimental time-resolved observations of the process [13]. Specifically, the ballistic effects, such as particle ejection and bed deformation, were drastically subdued, even in simulations using a droplet velocity ∼ 5× that of typical jetting conditions. This behavior could be caused by excessive damping in the inter-particle contact force computations within their SPH framework. Moreover, the wetted particles did not appear to be significantly influenced by the strong capillary forces exerted by the binder as no primitive agglomeration occurred. The authors mention that the objective of these simulations was to demonstrate their codebase’s broad capabilities and that some unrealistic process parameters were used to improve computational efficiency and stability, which could explain the deviations from experimental observations.
In the present paper, we develop a novel 3D CFD-DEM numerical framework for simulating fully-resolved IFPI during binder jetting with realistic material properties and process parameters. The CFD module is based on the VOF method for capturing binder–air interfaces. Surface tension effects are realized through the continuum surface force (CSF) method with height function calculations of interface curvature. Central to our fluid solver is a proprietary block-structured AMR library with hierarchical octree grid nesting to focus enhanced grid resolution near fluid–fluid interfaces. The GPU-accelerated DEM module considers six degrees of freedom particle motion and includes models based on Hertz-Mindlin contact, van der Waals cohesion, and viscoelastic rolling resistance. The CFD and DEM modules are coupled to achieve fully-resolved IFPI using an IB approach in which Lagrangian solid particles are mapped to the underlying Eulerian fluid mesh through a solid volume fraction field. An improved VOF-IB extension algorithm is introduced to enforce the contact angle at three-phase intersections. This provides robust capillary flow behavior and accurate computations of the fluid-induced forces and torques acting on individual wetted particles in densely packed powder beds.
We deploy our integrated codebase for direct numerical simulations of single-drop primitive formation with powder beds whose particle size distributions are generated from corresponding laboratory samples. These simulations use jetting parameters similar to those employed in current BJ3DP machines, fluid properties that match commonly used aqueous polymeric binders, and powder properties specific to nickel alloy feedstocks. The cohesion behavior of the DEM powder is calibrated based on the angle of repose of the laboratory powder systems. The resulting primitive granules are compared with those obtained from one-to-one experiments conducted using a dedicated in-house test apparatus. Finally, we demonstrate how the proposed framework can simulate more complex and realistic printing operations involving multi-drop primitive lines.
Section snippets
Mathematical description of interfacial fluid–particle interaction
This section briefly describes the governing equations of fluid and particle dynamics underlying the CFD and DEM solvers. Our unified framework follows an Eulerian–Lagrangian approach, wherein the Navier–Stokes equations of incompressible flow are discretized on an Eulerian grid to describe the motion of the binder liquid and surrounding gas, and the Newton–Euler equations account for the positions and orientations of the Lagrangian powder particles. The mathematical foundation for
CFD solver for incompressible flow with multifluid interfaces
This section details the numerical methodology used in our CFD module to solve the Navier–Stokes equations of incompressible flow. First, we introduce the VOF method for capturing the interfaces between the binder and air phases. This approach allows us to solve the fluid dynamics equations considering only a single continuum field with spatial and temporal variations in fluid properties. Next, we describe the time integration procedure using a fractional-step projection algorithm for
DEM solver for solid particle dynamics
This section covers the numerical procedure for tracking the motion of individual powder particles with DEM. The Newton–Euler equations (Eqs. (10), (11)) are ordinary differential equations (ODEs) for which many established numerical integrators are available. In general, the most challenging aspects of DEM involve processing particle collisions in a computationally efficient manner and dealing with small time step constraints that result from stiff materials, such as metallic AM powders. The
Unified CFD-DEM solver
The preceding sections have introduced the CFD and DEM solution algorithms separately. Here, we discuss the integrated CFD-DEM solution algorithm and related details.
Binder jet process modeling and validation experiments
In this section, we deploy our CFD-DEM framework to simulate the IFPI occurring during the binder droplet deposition stage of the BJ3DP process. The first simulations attempt to reproduce experimental single-drop primitive granules extracted from four nickel alloy powder samples with varying particle size distributions. The experiments are conducted with a dedicated in-house test apparatus that allows for the precision deposition of individual binder microdroplets into a powder bed sample. The
Conclusions
This paper introduces a coupled CFD-DEM framework capable of fully-resolved simulation of the interfacial fluid–particle interaction occurring in the binder jet 3D printing process. The interfacial flow of binder and surrounding air is captured with the VOF method and surface tension effects are incorporated using the CSF technique augmented by height function curvature calculations. Block-structured AMR is employed to provide localized grid refinement around the evolving liquid–gas interface.
CRediT authorship contribution statement
Joshua J. Wagner: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Software, Visualization, Writing – original draft, Writing – review & editing. C. Fred Higgs III: Conceptualization, Funding acquisition, Investigation, Methodology, Project administration, Resources, Supervision, Writing – original draft, Writing – review & editing.
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.
Acknowledgments
This work was supported by a NASA Space Technology Research Fellowship, United States of America, Grant No. 80NSSC19K1171. Partial support was also provided through an AIAA Foundation Orville, USA and Wilbur Wright Graduate Award, USA . The authors would like to gratefully acknowledge Dr. Craig Smith of NASA Glenn Research Center for the valuable input he provided on this project.
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.
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.ArticleCASGoogle Scholar
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.ArticleCASGoogle Scholar
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.ArticleCASGoogle Scholar
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.ArticleGoogle Scholar
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.ArticleCASGoogle Scholar
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.ArticleCASADSGoogle Scholar
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.ArticleGoogle Scholar
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.CASGoogle Scholar
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.ArticleGoogle Scholar
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.ArticleGoogle Scholar
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.ArticleGoogle Scholar
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.ArticleGoogle Scholar
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.ArticleCASGoogle Scholar
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.ArticleCASGoogle Scholar
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.ArticleCASGoogle Scholar
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.ArticleCASGoogle Scholar
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.ArticleCASPubMedPubMed CentralGoogle Scholar
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.ArticleCASGoogle Scholar
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.CASGoogle Scholar
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.ArticleCASGoogle Scholar
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.ArticleGoogle Scholar
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.ArticleCASGoogle Scholar
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.ArticleCASGoogle Scholar
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.ArticleGoogle Scholar
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.ArticleCASGoogle Scholar
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.ArticleCASGoogle Scholar
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.ArticleCASADSGoogle Scholar
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.ArticleCASGoogle Scholar
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.ArticleGoogle Scholar
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.ArticleGoogle Scholar
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.ArticleCASGoogle Scholar
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.ArticleGoogle Scholar
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.ArticleCASGoogle Scholar
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.ArticleCASGoogle Scholar
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.ArticleGoogle Scholar
Microstructural defects in laser powder bed fusion (LPBF) metallic materials are correlated with processing parameters. A multi-physics model and a crystal plasticity framework are employed to predict microstructure growth in molten pools and assess the impact of manufacturing defects on plastic damage parameters. Criteria for optimising the LPBF process are identified, addressing microstructural defects and tensile properties of LPBF Hastelloy X at various volumetric energy densities (VED). The results show that higher VED levels foster a specific Goss texture {110} <001>, with irregular lack of fusion defects significantly affecting plastic damage, especially near the material surface. A critical threshold emerges between manufacturing defects and grain sizes in plastic strain accumulation. The optimal processing window for superior Hastelloy X mechanical properties ranges from 43 to 53 J/mm3 . This work accelerates the development of superior strengthductility alloys via LPBF, streamlining the trial-and-error process and reducing associated costs.
References [1] DebRoy T, Wei HL, Zuback JS, et al. Additive manufacturing of metallic components – process, structure and properties. Prog Mater Sci. 2018;92:112–224. doi:10. 1016/j.pmatsci.2017.10.001 [2] Mostafaei A, Ghiaasiaan R, Ho IT, et al. Additive manufacturing of nickel-based superalloys: A state-of-the-art review on process-structure-defect-property relationship. Prog Mater Sci. 2023;136:101108. doi:10.1016/j.pmatsci. 2023.101108 [3] Akande IG, Oluwole OO, Fayomi OSI, et al. Overview of mechanical, microstructural, oxidation properties and high-temperature applications of superalloys. Mater Today Proc. 2021;43:2222–2231. doi:10.1016/j.matpr. 2020.12.523 [4] Sanchez S, Smith P, Xu Z, et al. Powder bed fusion of nickel-based superalloys: a review. Int J Machine Tools Manuf. 2021;165:103729. doi:10.1016/j.ijmachtools.2021. 103729 [5] Xie Y, Teng Q, Shen M, et al. The role of overlap region width in multi-laser powder bed fusion of Hastelloy X superalloy. Virtual Phys Prototyp. 2023;18(1):e2142802. doi:10.1080/17452759.2022.2142802 [6] Yuan W, Chen H, Cheng T, et al. Effects of laser scanning speeds on different states of the molten pool during selective laser melting: simulation and experiment. Mater Des. 2020;189:108542. doi:10.1016/j.matdes.2020. 108542 [7] He X, Kong D, Zhou Y, et al. Powder recycling effects on porosity development and mechanical properties of Hastelloy X alloy during laser powder bed fusion process. Addit Manuf. 2022;55:102840. doi:10.1016/j. addma.2022.102840 [8] Sanaei N, Fatemi A. Defects in additive manufactured metals and their effect on fatigue performance: a stateof-the-art review. Prog Mater Sci. 2021;117:100724. doi:10.1016/j.pmatsci.2020.100724 [9] Pourbabak S, Montero-Sistiaga ML, Schryvers D, et al. Microscopic investigation of as built and hot isostatic pressed Hastelloy X processed by selective laser melting. Mater Charact. 2019;153:366–371. doi:10.1016/j. matchar.2019.05.024 [10] He X, Wang L, Kong D, et al. Recrystallization effect on surface passivation of Hastelloy X alloy fabricated by laser powder bed fusion. J Mater Sci Technol. 2023;163:245–258. doi:https://doi.org/10.1016j.jmst. 2023.06.003. [11] Sabzi HE, Maeng S, Liang X, et al. Controlling crack formation and porosity in laser powder bed fusion: alloy design and process optimisation. Addit Manuf. 2020;34:101360. doi:10.1016/j.addma.2020.101360 [12] Yu C, Chen N, Li R, et al. Selective laser melting of GH3536 superalloy: microstructure, mechanical properties, and hydrocyclone manufacturing. Adv Powder Mater. 2023:
doi:10.1016/j.apmate.2023.100134 [13] Ye C, Zhang C, Zhao J, et al. Effects of post-processing on the surface finish, porosity, residual stresses, and fatigue performance of additive manufactured metals: a review. J Mater Eng Perform. 2021;30(9):6407–6425. doi:10. 1007/s11665-021-06021-7 [14] Zhang W, Zheng Y, Liu F, et al. Effect of solution temperature on the microstructure and mechanical properties of Hastelloy X superalloy fabricated by laser directed energy deposition. Mater Sci Eng A. 2021;820:141537. doi:10. 1016/j.msea.2021.141537 [15] Lehmann T, Rose D, Ranjbar E, et al. Large-scale metal additive manufacturing: a holistic review of the state of the art and challenges. Int Mater Rev. 2021;67(4):410–459. doi:10.1080/09506608.2021.1971427
[16] Wu S, Hu Y, Yang B, et al. Review on defect characterization and structural integrity assessment method of additively manufactured materials. J Mech Eng. 2021;57 (22):3–34. doi:10.3901/JME.2021.22.003
[17] Keller C, Mokhtari M, Vieille B, et al. Influence of a rescanning strategy with different laser powers on the microstructure and mechanical properties of Hastelloy X elaborated by powder bed fusion. Mater Sci Eng A. 2021;803:140474. doi:10.1016/j.msea.2020.140474
[18] Keshavarzkermani A, Marzbanrad E, Esmaeilizadeh R,et al. An investigation into the effect of process parameters on melt pool geometry, cell spacing, and grain refinement during laser powder bed fusion. Optics & Laser Technol. 2019;116:83–91. doi:10.1016/j.optlastec. 2019.03.012
[19] Watring DS, Benzing JT, Hrabe N, et al. Effects of laserenergy density and build orientation on the structureproperty relationships in as-built Inconel 718 manufactured by laser powder bed fusion. Addit Manuf. 2020;36:101425. doi:10.1016/j.addma.2020.101425
[20] Xiao H, Liu X, Xiao W, et al. Influence of molten-pool cooling rate on solidification structure and mechanical property of laser additive manufactured Inconel 718. J Mater Res Technol. 2022;19:4404–4416. doi:10.1016/j. jmrt.2022.06.162
[21] Wang J, Zhu R, Liu Y, et al. Understanding melt pool characteristics in laser powder bed fusion: An overview of single- and multi-track melt pools for process optimization. Adv Powder Mater. 2023;2(4):100137. doi:10.1016/j. apmate.2023.100137
[22] Li Z, Deng Y, Yao B, et al. Effect of laser scan speed on pool size and densification of selective laser melted CoCr alloy under constant laser energy density. Laser Optoelectronics Progress. 2022;59(7):0736001. doi:10. 3788/LOP202259.0736001
[23] Zhang J, Yuan W, Song B, et al. Towards understanding metallurgical defect formation of selective laser melted wrought aluminum alloys. Adv Powder Mater. 2022;1 (4):100035. doi:10.1016/j.apmate.2022.100035
[24] Rui H, Meiping W, Chen C, et al. Effects of laser energy density on microstructure and corrosion resistance of FeCrNiMnAl high entropy alloy coating. Optics & Laser Technol. 2022;152:108188. doi:https://doi.org/10.1016j. optlastec.2022.108188.
[25] Zhao Y, Sun W, Wang Q, et al. Effect of beam energy density characteristics on microstructure and mechanical properties of nickel-based alloys manufactured by laser directed energy deposition. J Mater Process Technol. 2023;319:118074. doi:10.1016/j.jmatprotec.2023.118074
[26] Tan P, Zhou M, Tang C, et al. Multiphysics modelling of powder bed fusion for polymers. Virtual Phys Prototyp. 2023;18(1):e2257191. doi:10.1080/17452759.2023. 2257191
[27] Tan P, Shen F, Shian Tey W, et al. A numerical study on the packing quality of fibre/polymer composite powder for powder bed fusion additive manufacturing. Virtual Phys Prototyp. 2021;16(sup1):S1–S18. doi:10.1080/17452759. 2021.1922965
[28] Kusano M, Watanabe M. Microstructure control of Hastelloy X by geometry-induced elevation of sample temperature during a laser powder bed fusion process. Mater Des. 2022;222:111016. doi:10.1016/j.matdes.2022. 111016
[29] Lee YS, Zhang W. Modeling of heat transfer, fluid flow and solidification microstructure of nickel-base superalloy fabricated by laser powder bed fusion. Addit Manuf. 2016;12:178–188. doi:10.1016/j.addma.2016.05.003
[30] Lv F, Liang HX, Xie DQ, et al. On the role of laser in situ remelting into pore elimination of Ti-6Al-4V components fabricated by selective laser melting. J Alloys Compd. 2021;854:156866. doi:10.1016/j.jallcom.2020.156866
[31] Prithivirajan V, Sangid MD. The role of defects and critical pore size analysis in the fatigue response of additively manufactured IN718 via crystal plasticity. Mater Des. 2018;150:139–153. doi:10.1016/j.matdes.2018.04.022
[32] Huang Y. A user-material subroutine incroporating single crystal plasticity in the ABAQUS finite element program. Cambridge: Harvard University Press; 1991.
[33] Pilgar CM, Fernandez AM, Lucarini S, et al. Effect of printing direction and thickness on the mechanical behavior of SLM fabricated Hastelloy-X. Int J Plasticity. 2022;153:103250. doi:10.1016/j.ijplas.2022.103250
[34] Garlea E, Choo H, Sluss CC, et al. Variation of elastic mechanical properties with texture, porosity, and defect characteristics in laser powder bed fusion 316L stainless steel. Mater Sci Eng A. 2019;763:138032. doi:10.1016/j. msea.2019.138032
[35] Sanchez-Mata O, Wang X, Muñiz-Lerma JA, et al. Dependence of mechanical properties on crystallographic orientation in nickel-based superalloy Hastelloy X fabricated by laser powder bed fusion. J Alloys Compd. 2021;865:158868. doi:10.1016/j.jallcom.2021. 158868
[36] Gu H, Wei C, Li L, et al. Multi-physics modelling of molten
pool development and track formation in multi-track, multi-layer and multi-material selective laser melting. Int J Heat Mass Transf. 2020;151:119458. doi:10.1016/j. ijheatmasstransfer.2020.119458
[37] Johnson L, Mahmoudi M, Zhang B, et al. Assessing printability maps in additive manufacturing of metal alloys. Acta Mater. 2019;176:199–210. doi:10.1016/j.actamat. 2019.07.005
[38] Wang S, Ning J, Zhu L, et al. Role of porosity defects in metal 3D printing: formation mechanisms, impacts on properties and mitigation strategies. Mater Today. 2022;59:133–160. doi:10.1016/j.mattod.2022.08.014
[39] Guo Y, Collins DM, Tarleton E, et al. Measurements of stress fields near a grain boundary: exploring blocked arrays of dislocations in 3D. Acta Mater. 2015;96:229–doi:10.1016/j.actamat.2015.05.041 [40] Kong D, Dong C, Ni X, et al. Hetero-deformation-induced stress in additively manufactured 316L stainless steel. Mater Res Lett. 2020;8(10):390–397. doi:10.1080/ 21663831.2020.1775149
Solute segregation significantly affects material properties and is a critical issue in the laser powder-bed fusion (LPBF) additive manufacturing (AM) of Ni-based superalloys. To the best of our knowledge, this is the first study to demonstrate a computational thermal-fluid dynamics (CtFD) simulation coupled multi-phase-field (MPF) simulation with a multicomponent-composition model of Ni-based superalloy to predict solute segregation under solidification conditions in LPBF. The MPF simulation of the Hastelloy-X superalloy reproduced the experimentally observed submicron-sized cell structure. Significant solute segregations were formed within interdendritic regions during solidification at high cooling rates of up to 108 K s-1, a characteristic feature of LPBF. Solute segregation caused a decrease in the solidus temperature (TS), with a reduction of up to 30.4 K, which increases the risk of liquation cracks during LPBF. In addition, the segregation triggers the formation of carbide phases, which increases the susceptibility to ductility dip cracking. Conversely, we found that the decrease in TS is suppressed at the melt-pool boundary regions, where re-remelting occurs during the stacking of the layer above. Controlling the re-remelting behavior is deemed to be crucial for designing crack-free alloys. Thus, we demonstrated that solute segregation at the various interfacial regions of Ni-based multicomponent alloys can be predicted by the conventional MPF simulation. The design of crack-free Ni-based superalloys can be expedited by MPF simulations of a broad range of element combinations and their concentrations in multicomponent Ni-based superalloys.
Additive manufacturing (AM) technologies have attracted considerable attention as they allow us to easily build three-dimensional (3D) parts with complex geometries. Among the wide range of available AM techniques, laser powder-bed fusion (LPBF) has emerged as a preferred technique for metal AM [1], [2], [3], [4], [5]. In LPBF, metal products are built layer-by-layer by scanning laser, which fuse metal powder particles into bulk solids.
Significant attempts have been made to integrate LPBF techniques within the aerospace industry, with a particular focus on weldable Ni-based superalloys, such as IN718 [6], [7], [8], IN625 [9], [10], and Hastelloy-X (HX) [11], [12], [13], [14]. Non-weldable alloys, such as IN738LC [15], [16] and CMSX-4 [1], [17] are also suitable for their sufficient creep resistance under higher temperature conditions. However, non-weldable alloys are difficult to build using LPBF because of their susceptibility to cracking during the process. In general, a macro solute-segregation during solidification is suppressed by the rapid cooling conditions (up to 108 K s-1) unique to the LPBF process [18]. However, the solute segregation still occurs in the interdendritic regions that are smaller than the micrometer scale [5], [19], [20], [21]; these regions are suggested to be related to the hot cracks in LPBF-fabricated parts. Therefore, an understanding of solute segregation is essential for the fabrication of reliable LPBF-fabricated parts while avoiding cracks.
The multiphase-field (MPF) method has gained popularity for modeling the microstructure evolution and solute segregation under rapid cooling conditions [5], [20], [21], [22], [23], [24], [25], [26], [27], [28]. Moreover, quantifiable predictions have been achieved by combining the MPF method with temperature distribution analysis methods such as the finite-element method (FEM) [20] and computational thermal-fluid dynamics (CtFD) simulations [28]. These aforementioned studies have used binary-approximated multicomponent systems, such as Ni–Nb binary alloys, to simulate IN718 alloys. While MPF simulations using binary alloy systems can effectively reproduce microstructure formations and segregation behaviors, the binary approximation might be affected by the chemical interactions between the removed solute elements in the target multicomponent alloy. The limit of absolute stability predicted by the Mullins-Sekerka theory [29] is also crucial because the limit velocity is close to the solidification rate in the LPBF process and is different in multicomponent and binary-approximated systems. The difference between the solidus and liquidus temperatures, ΔT0, directly determines the absolute stability according to the Mullins-Sekerka theory. For example, the ΔT0 values of IN718 and its binary-approximated Ni–5 wt.%Nb alloy are 134 K [28] and 71 K [30], respectively. The solidification rate compared to the limit of absolute stability, i.e., the relative non-equilibrium of solidification, changes by simplification of the system. It is therefore important to use the composition of the actual multicomponent system in such simulations. However, to the best of our knowledge, there is no MPF simulation using a multicomponent model coupled with a temperature analysis simulation to predict solute segregation in a Ni-based superalloy.
In this study, we demonstrate that the conventional MPF model can reproduce experimentally observed dendritic structures by performing a phase-field simulation using the temperature distribution obtained by a CtFD simulation of a multicomponent Ni-based alloy (conventional solid-solution hardening-type HX). The MPF simulation revealed that the segregation behavior of solute elements largely depends on the regions of the melt pool, such as the cell boundary, the interior of the melt-pool boundary, and heat-affected regions. The sensitivities of the various interfaces to liquation and solidification cracks are compared based on the predicted concentration distributions. Moreover, the feasibility of using the conventional MPF model for LPBF is discussed in terms of the absolute stability limit.
2. Methods
2.1. Laser-beam irradiation experiments
Rolled and recrystallized HX ingots with dimensions of 20 × 50 × 10 mm were used as the specimens for laser-irradiation experiments. The specimens were irradiated with a laser beam scanned along straight lines of 10 mm in length using a laser AM machine (EOS 290 M, EOS) equipped with a 400 W Yb-fiber laser. Irradiation was performed with a beam power of P = 300 W and a scanning speed of V = 600 mm s-1, which are the conditions generally used in the LPBF fabrication of Ni-based superalloy [8]. The corresponding line energy was 0.5 J mm-1. The samples were cut perpendicular to the beam-scanning direction for cross-sectional observation using a field-emission scanning electron microscope (FE-SEM, JEOL JSM 6500). Crystal orientation analysis was performed by electron backscatter diffraction (EBSD). The sizes of each crystal grain and their aspect ratios were evaluated by analyzing the EBSD data.
2.2. CtFD simulation
CtFD simulations of the laser-beam irradiation of HX were performed using a 3D thermo-fluid analysis software (Flow Science FLOW-3D® with Flow-3D Weld module). A Gaussian heat source model was used, in which the irradiation intensity distribution of the beam is regarded as a symmetrical Gaussian distribution over the entire beam. The distribution of the beam irradiation intensity is expressed by the following equation.(1)q̇=2ηPπR2exp−2r2R2.
Here, P is the power, R is the effective beam radius, r is the actual beam radius, and η is the beam absorption rate of the substrate. To improve the accuracy of the model, η was calculated by assuming multiple reflections using the Fresnel equation:(2)�=1−121+1−�cos�21+1+�cos�2+�2−2�cos�+2cos2��2+2�cos�+2cos2�.
ε is the Fresnel coefficient and θ is the incident angle of the laser. A local laser melt causes the vaporization of the material and results in a high vapor pressure. This vapor pressure acts as a recoil pressure on the surface, pushing the weld pool down. The recoil pressure is reproduced using the following equation.(3)precoil=Ap0exp∆HLVRTV1−TVT.
Here, p0 is the atmospheric pressure, ∆HLV is the latent heat of vaporization, R is the gas constant, and TV is the boiling point at the saturated vapor pressure. A is a ratio coefficient that is generally assumed to be 0.54, indicating that the recoil pressure due to evaporation is 54% of the vapor pressure at equilibrium on the liquid surface.
Table 1 shows the parameters used in the simulations. Most parameters were evaluated using an alloy physical property calculation software (Sente software JMatPro v11). The values in a previously published study [31] were used for the emissivity and the Stefan–Boltzmann constant, and the values for pure Ni [32] were used for the heat of vaporization and vaporization temperatures. The Fresnel coefficient, which determines the beam absorption efficiency, was used as a fitting parameter to reproduce the morphology of the experimentally observed melt region, and a Fresnel coefficient of 0.12 was used in this study.
The dimensions of the computational domain of the numerical model were 4.0 mm in the beam-scanning direction, 0.4 mm in width, and 0.3 mm in height. A uniform mesh size of 10 μm was applied throughout the computational domain. The boundary condition of continuity was applied to all boundaries except for the top surface. The temperature was initially set to 300 K. P and V were set to their experimental values, i.e., 300 W and 600 mm s-1, respectively. Solidification conditions based on the temperature gradient, G, the solidification rate, R, and the cooling rate were evaluated, and the obtained temperature distribution was used in the MPF simulations.
2.3. MPF simulation
Two-dimensional MPF simulations weakly coupled with the CtFD simulation were performed using the Microstructure Evolution Simulation Software (MICRESS) [33], [34], [35], [36], [37] with the TQ-Interface for Thermo-Calc [38]. A simplified HX alloy composition of Ni-21.4Cr-17.6Fe-0.46Mn-8.80Mo-0.39Si-0.50W-1.10Co-0.08 C (mass %) was used in this study. The Gibbs free energy and diffusion coefficient of the system were calculated using the TCNI9 thermodynamic database [39] and the MOBNi5 mobility database [40]. Τhe equilibrium phase diagram calculated using Thermo-Calc indicates that the face-centered cubic (FCC) and σ phases appear as the equilibrium solid phases [19]. However, according to the time-temperature-transformation (TTT) diagram [41], the phases are formed after the sample is maintained for tens of hours in a temperature range of 1073 to 1173 K. Therefore, only the liquid and FCC phases were assumed to appear in the MPF simulations. The simulation domain was 5 × 100 μm, and the grid size Δx and interface width were set to 0.025 and 0.1 µm, respectively. The interfacial mobility between the solid and liquid phases was set to 1.0 × 10-8 m4 J-1 s-1. Initially, one crystalline nucleus with a [100] crystal orientation was placed at the left bottom of the simulation domain, with the liquid phase occupying the remainder of the domain. The model was solidified under the temperature field distribution obtained by the CtFD simulation. The concentration distribution and crystal orientation of the solidified model were examined. The primary dendrite arm space (PDAS) was compared to the experimental PDAS measured by the cross-sectional SEM observation.
In an actual LPBF process, solidified layers are remelted and resolidified during the stacking of the one layer above, thereby greatly affecting solute element distributions in those regions. Therefore, remelting and resolidification simulations were performed to examine the effect of remelting on solute segregation. The solidified model was remelted and resolidified by applying a time-dependent temperature field shifted by 60 μm in the height direction, assuming reheating during the stacking of the upper layer (i.e., the upper 40 μm region of the simulation box was remelted and resolidified). The changes in the composition distribution and formed microstructure were investigated.
3. Results
3.1. Experimental observation of melt pool
Fig. 1 shows a cross-sectional optical microscopy image and corresponding inverse pole figure (IPF) orientation maps obtained from the laser-melted region of HX. The dashed line indicates the fusion line. A deep melted region was formed by keyhole-mode melting due to the vaporization of the metal and resultant recoil pressure. Epitaxial growth from the unmelted region was observed. Columnar crystal grains with an average diameter of 5.46 ± 0.32 μm and an aspect ratio of 3.61 ± 0.13 appeared at the melt regions (Figs. 1b–1d). In addition, crystal grains growing in the z direction could be observed in the lower center.
Fig. 2a shows a cross-sectional backscattering electron image (BEI) obtained from the laser-melted region indicated by the black square in Fig. 1a. The bright particles with a diameter of approximately 2 μm observed outside the melt pool. It is well known that M6C, M23C6, σ, and μ precipitate phases are formed in Hastelloy-X [41]. These precipitates mainly consisted of Mo, Cr, Fe, and Ni; The μ and M6C phases are rich in Mo, while the σ and M23C6 phases are rich in Cr. The SEM energy dispersive X-ray spectroscopy analysis suggested that the bright particles are the stable precipitates as shown in Fig. S2 and Table S1. Conversely, there are no carbides in the melt pool. This suggests that the cooling rate is extremely high during LPBF, which prevents the formation of a stable carbide during solidification. Figs. 2b–2f show magnified BEI images at different height positions indicated in Fig. 2a. Bright regions are observed between the cells, which become fragmentary at the center of the melt pool, as indicated by the yellow arrow heads in Figs. 2e and 2f.
3.2. CtFD simulation
Figs. 3a–3c show snapshots of the CtFD simulation of HX at 2.72 ms, with the temperature indicated in color. A melt pool with an elongated teardrop shape formed and keyhole-mode melting was observed at the front of the melt region. The cooling rate, temperature gradient (G), and solidification rate (R) were evaluated from the temporal change in the temperature distribution of the CtFD simulation results. The z-position of the solid/liquid interface during the melting and solidification processes is shown in Fig. 3d. The interface goes down rapidly during melting and then rises during solidification. The MPF simulation of the microstructure formation during solidification was performed using the temperature distribution. Moreover, the microstructure formation process during the fabrication of the upper layer was investigated by remelting and resolidifying the solidified layer using the same temperature distribution with a 60 μm upward shift, corresponding to the layer thickness commonly used in the LPBF of Ni-based superalloys.
Figs. 4a–4c show the changes in the cooling rate, temperature gradient, and solidification rate in the center line of the melt pool parallel to the z direction. To output the solidification conditions at the solid/liquid interface in the melt pool, only the data of the mesh where the solid phase ratio was close to 0.5 were plotted. Solidification occurred where the cooling rate was in the range of 2.1 × 105–1.6 × 106 K s-1, G was in the range of 3.6 × 105–1.9 × 107 K m-1, and R was in the range of 8.2 × 10−2–6.3 × 10−1 m s-1. The cooling rate was the highest near the fusion line and decreased as the interface approached the center of the melt region (Fig. 4a). G also exhibited the highest value in the regions near the fusion line and decreased throughout the solid/liquid interface toward the center of the melt pool (Fig. 4b). R had the lowest value near the fusion line and increased as the interface approached the center of the melt region (Fig. 4c).
3.3. MPF simulations coupled with CtFD simulation
MPF simulations of solidification, remelting, and resolidification were performed using the temperature-time distribution obtained by the CtFD simulation. Fig. 5 shows the MPF solidified models colored by phase and Mo concentration. All the computational domains show the FCC phase after the solidification (Fig. 5a). Dendrites grew parallel to the heat flow direction, and solute segregations were observed in the interdendritic regions. At the bottom of the melt pool (Fig. 5d), planar interface growth occurred before the formation of primary dendrites. The bottom of the melt pool is the turning point of the solid/liquid interface from the downward motion in melting to the upward motion in solidification. Thus, the solidification rate at the boundary is zero, and is extremely low immediately above the molt-pool boundary. Here, the lower limit of the solidification rate (R) for dendritic growth can be represented by the constitutional supercooling criterion [29], Vcs = (G × DL) / ΔT, and planar interface growth occurs at R < Vcs. DL and ΔT denote the diffusion coefficient in the liquid and the equilibrium freezing range, respectively. The results suggest that planar interface growth occurs at the bottom of the melt pool, resulting in a dark region with a different solute element distribution. Some of the primary dendrites were diminished by competition with other dendrites. In addition, secondary dendrite arms could be seen in the upper regions (Fig. 5c), where solidification occurred at a lower cooling rate. The fragmentation of the solute segregation near the secondary dendrite arms is similar to that observed in the experimental melt pool shown in Figs. 2e and 2f, and the secondary dendrite arms are suggested to have appeared at the center of the melt region. Fig. 6 shows the PDASs measured from the MPF simulation models, compared to the experimental PDASs measured by the cross-sectional SEM observation of the laser-melted regions (Fig. 2). The PDAS obtained by the MPF simulation become larger as the solidification progress. Ghosh et al. [21] evident by the phase-field method that the PDAS decreases as the cooling rate increases under the rapid cooling conditions obtained by the finite element analysis. In this study, the cooling rate was decreased as the interface approached the center of the melt region (Fig. 4a), and the trends in PDAS changes with respect to cooling rate is same as the reported trend [21]. The simulated trends of the PDAS with the position in the melt pool agreed well with the experimental trends. However, all PDASs in the simulation were larger than those observed in the experiment at the same positions. Ode et al. [42] reported that PDAS differences between 2D and 3D MPF simulations can be represented by PDAS2D = 1.12 × PDAS3D owing to differences in the effects of the interfacial energy and diffusivity. We also performed 2D and 3D MPF simulations under the solidification conditions of G = 1.94 × 107 K m-1 and R = 0.82 m s-1 (Fig. S1), and found that the PDAS from the 2D MPF simulation was 1.26 times larger than that from the 3D simulation. Therefore, the cell structure obtained by the CtFD simulation coupled with the 2D MPF simulation agreed well with the experimental results over the entire melt pool region considering the dimensional effects.
Fig. 7b1 and 7c1 show the concentration profiles of the solidified model along the growth direction indicated by dashed lines in Fig. 7a. The differences in concentrations from the alloy composition are also shown in Fig. 7b2 and 7c2. Cr, Mo, C, Mn, and W were segregated to the interdendritic regions, while Si, Fe, and Co were depressed. The solute segregation behavior agrees with the experimentally observation [43] and the prediction by the Scheil-Gulliver simulation [19]. Segregation occurred to the highest degree in Mo, while the ratio of segregation to the alloy composition was remarkable in C. The concentration fluctuations correlated with the position in the melt pool and decreased at the center of the melt pool, which was suggested to correspond to the lower cooling rate in this region. Conversely, droplets that appeared between secondary dendrite arms in the upper regions of the simulation domain exhibited a locally high segregation of solute elements, with the same amount of segregation as that at the bottom of the melt pool.
3.4. Remelting and resolidification simulation
The solidified model was subjected to remelting and resolidification conditions by shifting the temperature profile upward by 60 µm to reveal the effect of reheating on the solute segregation behavior. Figs. 8a and 8b shows the simulation domains of the HX model after resolidification, colored by phase and Mo concentration. The magnified MPF models during the resolidification of the regions indicated by rectangles in Figs. 8a and 8b are also shown as Figs. 8c and 8d. Dendrites grew from the bottom of the remelted region, with the segregation of solute elements occurring in the interdendritic regions. The entire domain become the FCC phase after the resolidification, as shown in Fig. 8a. The bottom of the remelted regions exhibited a different microstructure, and Mo was depressed at the remelted regions, rather than the interdendritic regions. The different solute segregation behavior [44] and the microstructure formation [45] at the melt pool boundary is also observed in LPBF manufactured 316 L stainless steel. We found that this microstructure was formed by further remelting during the resolidification process, which is shown in Fig. 9. Here, the solidified HX model was heated, and the interdendritic regions were preferentially melted while concentration fluctuations were maintained (Fig. 9a1 and 9a2). Subsequently, planer interface growth occurs near the melt pool boundary where the solidification rate is almost zero, and the dendrites outside of the boundary are grown epitaxially (Fig. 9b1 and 9b2). However, these remelted again because of the temperature rise (Fig. 9c1 and 9c2, and the temperature-time profile shown in Fig. 9e). The remelted regions then cooled and solidified with the abnormal solute segregations (Fig. 9d1 and 9d2). Then, dendrite grows from amplified fluctuations under the solidification rate larger than the criterion of constitutional supercooling (Fig. 9d1, 9d2, and Fig. 8d). It has been reported [46], [47] that temperature rising owning to latent heat affects microstructure formation: phase-field simulations of a Ni–Al binary alloy suggest that the release of latent heat during solidification increases the average temperature of the system [46] and strongly influences the solidification conditions [47]. In this study, the release of latent heat during solidification is considered in CtFD simulations for calculating the temperature distribution, and the temperature increase is suggested to have also occurred due to the release of latent heat.
Fig. 10b1 and 10c1 show the solute element concentration line profiles of the resolidified model along the growth direction indicated by dashed lines in Fig. 10a. Fig. 10b2 and 10c2 show the corresponding differences in concentration from the alloy composition. The segregation behavior of solute elements at the interdendritic regions (Fig. 10b1 and 10b2) was the same as that in the solidified model (Figs. 7b1 and 7b2). Here, Cr, Mo, C, Mn, and W were segregated to the interdendritic regions, while Si, Fe, and Co were depressed. However, the concentration fluctuations at the interdendritic regions were larger than those in the solidified model. Moreover, the segregation of the outside of the melt pool, i.e., the heat-affected zone, was remarkable throughout remelting and resolidification. Different segregation behaviors were observed in the re-remelted region: Mo, Si, Mn, and W were segregated, while Ni, Fe, and Co were depressed. These solute segregations caused by remelting are expected to heavily influence the crack behavior.
4. Discussion
4.1. Effect of segregation of solute elements on liquation cracking susceptibility
Strong solute segregation was observed between the interdendritic regions of the solidified alloy (Fig. 7). In addition, the solute segregation behavior was significantly affected by remelting and resolidification and varied across the alloy. Solute segregation can be categorized by the regions shown in Fig. 11a1–11a4, namely the cell boundary (Fig. 11a1), interior of the melt-pool boundary (Fig. 11a2), re-remelted regions (Fig. 11a3), and heat-affected regions (Fig. 11a4). The concentration profiles of these regions are shown in Fig. 11b1–11b4. Solute segregation was the highest in the cell boundary region. The solute segregation in the heat-affected region was almost the same as that in the cell boundary region, but seemed to have been attenuated by reheating during remelting and resolidification. The interior of the melt-pool boundary region also had the same tendency for solute segregation. However, the amount of Cr segregation was smaller than that of Mo. A decrease in the Cr concentration was also mitigated, and the concentration remained the same as that in the alloy composition. Fig. 11c1–11c4 show the chemical potentials of the solute elements for the FCC phase at 1073 K calculated using the compositions of those interfacial regions. All the interfacial regions showed non-constant chemical potentials for each element along the perpendicular direction, but the fluctuations of the chemical potentials differed by the type of interfaces. In particular, the fluctuation of the chemical potential of C at the cell boundary region was the largest, suggesting it can be relaxed easily by heat treatment. On the other hand, the fluctuations of the other elements in all the regions were small. The solute segregations are most likely to remain after the heat treatment and are supposed to affect the cracking susceptibilities.
The solidus temperatures TS, the difference between the liquidus and solidus temperatures (i.e., the brittle temperature range (BTR)), and the fractions of the equilibrium precipitate phases at 1073 K of the interfacial regions were calculated as the liquation, solidification, and ductility dip cracking susceptibilities, respectively. At the cell boundary (Fig. 12a1), interior of the melt-pool boundary (Fig. 12a1), and heat-affected regions (Fig. 12a1), the internal and interfacial regions exhibited higher and lower TS compared to that of the alloy composition, respectively. The lowest Ts was obtained with the composition at the cell boundary region, which is the largest solute-segregated region. It has been suggested that strong segregations of solute elements in LPBF lead to liquation cracks [16]. This study also supports this suggestion, and liquation cracks are more likely to occur at the interfacial regions indicated by predicting the solute segregation behavior using the MPF model. Additionally, the BTRs of the cell boundary, interior of the melt-pool boundary, and heat-affected regions were wider at the interdendritic regions, and solidification cracks were also likely to occur in these regions. Moreover, within the solute segregation regions, the fraction of the precipitate phases in these interfacial regions was larger than that calculated using the alloy composition (Fig. 12c1, 12c2, and 12c4). This indicates that ductility dip cracking is also likely to occur at the cell boundary, interior of the melt-pool boundary, and in heat-affected regions. Contrarily, we found that the re-remelted region exhibited a higher TS and smaller BTR even in the interfacial region (Fig. 12a3 and 12b3), where the solute segregation behavior was different from that of the other regions. In addition, the re-remelting region exhibited less precipitation compared with the other segregated regions (Fig. 12c3). The re-remelting caused by the latent heat can attenuate solute segregation, prevent Ts from decreasing, decrease the BTR, and decrease the amount of precipitate phases. Alloys with a large amount of latent heat are expected to increase the re-remelting region, thereby decreasing the susceptibility to liquation and ductility dip cracks due to solute element segregation. This can be a guide for designing alloys for the LPBF process. As mentioned in Section 3.4, the microstructure [45] and the solute segregation behavior [44] at the melt pool boundary of LPBF-manufactured 316 L stainless steel are observed, and they are different from that of the interdendritic regions. Experimental observations of the solute segregation behavior in the LPBF-fabricated Ni-based alloys are currently underway.
4.2. Applicability of the conventional MPF simulation to microstructure formation under LPBF
As the solidification growth rate increases, segregation coefficients approach 1, and the fluctuation of the solid/liquid interface is suppressed by the interfacial tension. The interface growth occurs in a flat fashion instead of having a cellular morphology at a velocity above the absolute stability limit, Ras, predicted by the Mullins-Sekerka theory [29]: Ras = (ΔT0DL) / (k Γ) where ΔT0, DL, k, and Γ are the difference between the liquidus and solidus temperatures, equilibrium segregation coefficient, the diffusivity of liquid, and the Gibbs-Thomson coefficient, respectively.
The Ras of HX was calculated using the equation and the thermodynamic parameters obtained by the TCNI9 thermodynamic database [39]. The calculated Ras of HX was 3.9 m s-1 and is ten times larger than that of the Ni–Nb alloy (approximately 0.4 m s-1) [20]. The HX alloy was solidified under R values in the range of 8.2 × 10−2–6.3 × 10−1 m s-1. The theoretically calculated criterion is larger than the evaluated R, and is in agreement with the experiment in which dendritic growth is observed in the melt pool (Fig. 5). In contrast, Karayagiz et al. [20] reported that the R of the Ni–Nb binary alloy under LPBF was as high as approximately 2 m s-1, and planar interface growth was observed to be predominant under the high-growth-rate conditions. These experimentally observed microstructures agree well with the prediction by the Mullins-Sekerka theory about the relationship between the morphology and solidification rates.
In this study, the solidification microstructure formed by the laser-beam irradiation of an HX multicomponent Ni-based superalloy was reproduced by a conventional MPF simulation, in which the system was assumed to be in a quasi-equilibrium condition. Boussinot et al. [24] also suggested that the conventional phase-field model can be applied to simulate the microstructure of an IN718 multicomponent Ni-based superalloy in LPBF. In contrast, Kagayaski et al. [20] suggested that the conventional MPF simulation cannot be applied to the solidification of the Ni-Nb binary alloy system and that the finite interface dissipation model proposed by Steinbach et al. [48], [49] is necessary to simulate the high solidification rates observed in LPBF. The difference in the applicability of the conventional MPF method to HX and Ni–Nb binary alloys is presumed to arise from the differences in the non-equilibrium degree of these systems under the high solidification rates of LPBF. The results suggest that Ras can be used as a simple index to apply the conventional MPF model for solidification in LPBF. Solidification becomes a non-equilibrium process as the solidification rate approaches the limit of absolute stability, Ras. In this study, the solidification of the HX multicomponent system occurred under a relatively low solidification rate compared to Ras, and the microstructure of the conventional MPF model was successfully reproduced in the physical experiment. However, note that the limit of absolute stability predicted by the Mullins-Sekerka theory was originally proposed for solidification in a binary alloy system, and further investigation is required to consider its applicability to multicomponent alloy systems. Moreover, the fast solidification, such as in the LPBF process, causes segregation coefficient approaching a value of 1 [20], [21], [25] corresponds to a diffusion length that is on the order of the atomic interface thickness. When the segregation coefficient approaches 1, solute undercooling disappears; hence, there is no driving force to amplify fluctuations regardless of whether interfacial tension is present. This phenomenon should be further investigated in future studies.
5. Conclusions
We simulated solute segregation in a multicomponent HX alloy under the LPBF process by an MPF simulation using the temperature distributions obtained by a CtFD simulation. We set the parameters of the CtFD simulation to match the melt pool shape formed in the laser-irradiation experiment and found that solidification occurred under high cooling rates of up to 1.6 × 106 K s-1.
MPF simulations using the temperature distributions from CtFD simulation could reproduce the experimentally observed PDAS and revealed that significant solute segregation occurred at the interdendritic regions. Equilibrium thermodynamic calculations using the alloy compositions of the segregated regions when considering crack sensitivities suggested a decrease in the solidus temperature and an increase in the amount of carbide precipitation, thereby increasing the susceptibility to liquation and ductility dip cracks in these regions. Notably, these changes were suppressed at the melt-pool boundary region, where re-remelting occurred during the stacking of the layer above. This effect can be used to achieve a novel in-process segregation attenuation.
Our study revealed that a conventional MPF simulation weakly coupled with a CtFD simulation can be used to study the solidification of multicomponent alloys in LPBF, contrary to the cases of binary alloys investigated in previous studies. We discussed the applicability of the conventional MPF model to the LPBF process in terms of the limit of absolute stability, Ras, and suggested that alloys with a high limit velocity, i.e., multicomponent alloys, can be simulated using the conventional MPF model even under the high solidification velocity conditions of LPBF.
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
Acknowledgments
This work was partly supported by the Cabinet Office, Government of Japan, Cross-ministerial Strategic Innovation Promotion Program (SIP), “Materials Integration for Revolutionary Design System of Structural Materials,” (funding agency: The Japan Science and Technology Agency), by JSPS KAKENHI Grant Numbers 21H05018 and 21H05193, and by CREST Nanomechanics: Elucidation of macroscale mechanical properties based on understanding nanoscale dynamics for innovative mechanical materials (Grant Number: JPMJCR2194) from the Japan Science and Technology Agency (JST). The authors would like to thank Mr. H. Kawabata and Mr. K. Kimura for their technical support with the sample preparations and laser beam irradiation experiments.
[1]M. Ramsperger, R.F. Singer, C. KörnerMicrostructure of the nickel-base superalloy CMSX-4 fabricated by selective electron beam meltingMetall. Mater. Trans. A Phys. Metall. Mater. Sci., 47 (2016), pp. 1469-1480, 10.1007/s11661-015-3300-y View at publisher This article is free to access.View in ScopusGoogle Scholar
[3]K. Hagihara, T. NakanoControl of anisotropic crystallographic texture in powder bed fusion additive manufacturing of metals and ceramics—A reviewJOM, 74 (2021), pp. 1760-1773, 10.1007/s11837-021-04966-7 View at publisher This article is free to access.Google Scholar
[6]N. Raghavan, R. Dehoff, S. Pannala, S. Simunovic, M. Kirka, J. Turner, N. Carlson, S.S. BabuNumerical modeling of heat-transfer and the influence of process parameters on tailoring the grain morphology of IN718 in electron beam additive manufacturingActa Mater, 112 (2016), pp. 303-314, 10.1016/j.actamat.2016.03.063View PDFView articleView in ScopusGoogle Scholar
[7]N. Raghavan, S. Simunovic, R. Dehoff, A. Plotkowski, J. Turner, M. Kirka, S. BabuLocalized melt-scan strategy for site specific control of grain size and primary dendrite arm spacing in electron beam additive manufacturingActa Mater, 140 (2017), pp. 375-387, 10.1016/j.actamat.2017.08.038View PDFView articleView in ScopusGoogle Scholar
[9]T. Keller, G. Lindwall, S. Ghosh, L. Ma, B.M. Lane, F. Zhang, U.R. Kattner, E.A. Lass, J.C. Heigel, Y. Idell, M.E. Williams, A.J. Allen, J.E. Guyer, L.E. LevineApplication of finite element, phase-field, and CALPHAD-based methods to additive manufacturing of Ni-based superalloysActa Mater, 139 (2017), pp. 244-253, 10.1016/j.actamat.2017.05.003View PDFView articleView in ScopusGoogle Scholar
[10]F. Zhang, L.E. Levine, A.J. Allen, M.R. Stoudt, G. Lindwall, E.A. Lass, M.E. Williams, Y. Idell, C.E. CampbellEffect of heat treatment on the microstructural evolution of a nickel-based superalloy additive-manufactured by laser powder bed fusionActa Mater, 152 (2018), pp. 200-214, 10.1016/j.actamat.2018.03.017View PDFView articleView in ScopusGoogle Scholar
[12]H. Wang, L. Chen, B. Dovgyy, W. Xu, A. Sha, X. Li, H. Tang, Y. Liu, H. Wu, M.S. PhamMicro-cracking, microstructure and mechanical properties of Hastelloy-X alloy printed by laser powder bed fusion: As-built, annealed and hot-isostatic pressedAddit. Manuf., 39 (2021), Article 101853, 10.1016/j.addma.2021.101853View PDFView articleView in ScopusGoogle Scholar
[14]H. Kitano, M. Tsujii, M. Kusano, A. Yumoto, M. WatanabeEffect of plastic strain on the solidification cracking of Hastelloy-X in the selective laser melting processAddit. Manuf., 37 (2020), Article 101742, 10.1016/j.addma.2020.101742View at publisher Google Scholar
[15]J. Xu, H. Gruber, D. Deng, R.L. Peng, J.J. MoverareShort-term creep behavior of an additive manufactured non-weldable Nickel-base superalloy evaluated by slow strain rate testingActa Mater, 179 (2019), pp. 142-157, 10.1016/j.actamat.2019.08.034View PDFView articleGoogle Scholar
[16]P. Kontis, E. Chauvet, Z. Peng, J. He, A.K. da Silva, D. Raabe, C. Tassin, J.J. Blandin, S. Abed, R. Dendievel, B. Gault, G. MartinAtomic-scale grain boundary engineering to overcome hot-cracking in additively-manufactured superalloysActa Mater, 177 (2019), pp. 209-221, 10.1016/j.actamat.2019.07.041View PDFView articleView in ScopusGoogle Scholar
[17]C. Körner, M. Ramsperger, C. Meid, D. Bürger, P. Wollgramm, M. Bartsch, G. EggelerMicrostructure and mechanical properties of CMSX-4 single crystals prepared by additive manufacturingMetall. Mater. Trans. A Phys. Metall. Mater. Sci., 49 (2018), pp. 3781-3792, 10.1007/s11661-018-4762-5 View at publisher This article is free to access.View in ScopusGoogle Scholar
[18]T. Ishimoto, R. Ozasa, K. Nakano, M. Weinmann, C. Schnitter, M. Stenzel, A. Matsugaki, T. Nagase, T. Matsuzaka, M. Todai, H.S. Kim, T. NakanoDevelopment of TiNbTaZrMo bio-high entropy alloy (BioHEA) super-solid solution by selective laser melting, and its improved mechanical property and biocompatibilityScr. Mater., 194 (2021), Article 113658, 10.1016/j.scriptamat.2020.113658View PDFView articleView in ScopusGoogle Scholar
[20]K. Karayagiz, L. Johnson, R. Seede, V. Attari, B. Zhang, X. Huang, S. Ghosh, T. Duong, I. Karaman, A. Elwany, R. ArróyaveFinite interface dissipation phase field modeling of Ni–Nb under additive manufacturing conditionsActa Mater, 185 (2020), pp. 320-339, 10.1016/j.actamat.2019.11.057View PDFView articleView in ScopusGoogle Scholar
[21]S. Ghosh, L. Ma, N. Ofori-Opoku, J.E. GuyerOn the primary spacing and microsegregation of cellular dendrites in laser deposited Ni-Nb alloysModel. Simul. Mater. Sci. Eng., 25 (2017), 10.1088/1361-651X/aa7369View at publisher Google Scholar
[22]S. Nomoto, M. Segawa, M. WatanabeNon- and quasi-equilibrium multi-phase field methods coupled with CALPHAD database for rapid-solidification microstructural evolution in laser powder bed additive manufacturing conditionMetals, 11 (2021), p. 626, 10.3390/met11040626View at publisher View in ScopusGoogle Scholar
[26]M. Okugawa, Y. Ohigashi, Y. Furishiro, Y. Koizumi, T. NakanoEquiaxed grain formation by intrinsic heterogeneous nucleation via rapid heating and cooling in additive manufacturing of aluminum-silicon hypoeutectic alloyJ. Alloys Compd., 919 (2022), Article 165812, 10.1016/j.jallcom.2022.165812View PDFView articleView in ScopusGoogle Scholar
[27]M. Okugawa, Y. Furushiro, Y. KoizumiEffect of rapid heating and cooling conditions on microstructure formation in powder bed fusion of Al-Si hypoeutectic alloy: A Phase-Field StudyMaterials, 15 (2022), p. 6092, 10.3390/ma15176092View at publisher View in ScopusGoogle Scholar
[36]C. Kumara, A. Ramanathan Balachandramurthi, S. Goel, F. Hanning, J. MoverareToward a better understanding of phase transformations in additive manufacturing of Alloy 718Materialia, 13 (2020), Article 100862View PDFView articleView in ScopusGoogle Scholar
[37]B. Böttger, M. ApelPhase-field simulation of the formation of new grains by fragmentation during melting of an ABD900 superalloyIOP Conf. Ser. Mater. Sci. Eng., 1281 (2023), Article 012008, 10.1088/1757-899x/1281/1/012008 View at publisher This article is free to access.Google Scholar
[44]K. Sato, S. Takagi, S. Ichikawa, T. Ishimoto, T. NakanoMicrostructure and Solute Segregation around the Melt-Pool Boundary of Orientation-Controlled 316L Austenitic Stainless Steel Produced by Laser Powder Bed FusionMaterials, 16 (2023), 10.3390/ma16010218View at publisher Google Scholar
[45]Y. Liu, K. Nose, M. Okugawa, Y. Koizumi, T. NakanoFabrication and process monitoring of 316L stainless steel by laser powder bed fusion with µ-helix scanning strategy and narrow scanning line intervalsMater. Trans., 64 (2023), pp. 1135-1142, 10.2320/matertrans.MT-ME2022006View at publisherView in ScopusGoogle Scholar
[46]H. Schaar, I. Steinbach, M. TegelerNumerical study of epitaxial growth after partial remelting during selective electron beam melting in the context of ni–alMetals, 11 (2021), pp. 2-13, 10.3390/met11122012View at publisher Google Scholar
[47]M. Uddagiri, O. Shchyglo, I. Steinbach, B. Wahlmann, C. KoernerPhase-Field Study of the History-Effect of Remelted Microstructures on Nucleation During Additive Manufacturing of Ni-Based SuperalloysMetall. Mater. Trans. A Phys. Metall. Mater. Sci., 54 (2023), pp. 1825-1842, 10.1007/s11661-023-07004-0 View at publisher This article is free to access.View in ScopusGoogle Scholar
Unintended end-of-process depression (EOPD) commonly occurs in laser powder bed fusion (LPBF), leading to poor surface quality and lower fatigue strength, especially for many implants. In this study, a high-fidelity multi-physics meso-scale simulation model is developed to uncover the forming mechanism of this defect. A defect-process map of the EOPD phenomenon is obtained using this simulation model. It is found that the EOPD formation mechanisms are different under distinct regions of process parameters. At low scanning speeds in keyhole mode, the long-lasting recoil pressure and the large temperature gradient easily induce EOPD. While at high scanning speeds in keyhole mode, the shallow molten pool morphology and the large solidification rate allow the keyhole to evolve into an EOPD quickly. Nevertheless, in the conduction mode, the Marangoni effects along with a faster solidification rate induce EOPD. Finally, a ‘step’ variable power strategy is proposed to optimise the EOPD defects for the case with high volumetric energy density at low scanning speeds. This work provides a profound understanding and valuable insights into the quality control of LPBF fabrication.
의도하지 않은 공정 종료 후 함몰(EOPD)은 LPBF(레이저 분말층 융합)에서 흔히 발생하며, 특히 많은 임플란트의 경우 표면 품질이 떨어지고 피로 강도가 낮아집니다. 본 연구에서는 이 결함의 형성 메커니즘을 밝히기 위해 충실도가 높은 다중 물리학 메조 규모 시뮬레이션 모델을 개발했습니다.
이 시뮬레이션 모델을 사용하여 EOPD 현상의 결함 프로세스 맵을 얻습니다. EOPD 형성 메커니즘은 공정 매개변수의 별개 영역에서 서로 다른 것으로 밝혀졌습니다.
키홀 모드의 낮은 스캔 속도에서는 오래 지속되는 반동 압력과 큰 온도 구배로 인해 EOPD가 쉽게 유발됩니다. 키홀 모드에서 높은 스캐닝 속도를 유지하는 동안 얕은 용융 풀 형태와 큰 응고 속도로 인해 키홀이 EOPD로 빠르게 진화할 수 있습니다.
그럼에도 불구하고 전도 모드에서는 더 빠른 응고 속도와 함께 마랑고니 효과가 EOPD를 유발합니다. 마지막으로, 낮은 스캐닝 속도에서 높은 체적 에너지 밀도를 갖는 경우에 대해 EOPD 결함을 최적화하기 위한 ‘단계’ 가변 전력 전략이 제안되었습니다.
이 작업은 LPBF 제조의 품질 관리에 대한 심오한 이해와 귀중한 통찰력을 제공합니다.
References
[1] Zhang C, Li Z, Zhang J, et al. Additive manufacturing of magnesium matrix composites: comprehensive review of recent progress and research perspectives. J Mag Alloys. 2023. doi:10.1016/j.jma.2023.02.005 [2] Webster S, Lin H, Carter III FM, et al. Physical mechanisms in hybrid additive manufacturing: a process design framework. J Mater Process Technol. 2022;291:117048. doi:10. 1016/j.jmatprotec.2021.117048 [3] Wang S, Ning J, Zhu L, et al. Role of porosity defects in metal 3D printing: formation mechanisms, impacts on properties and mitigation strategies. Mater Today. 2022. doi:10.1016/j.mattod.2022.08.014 [4] Wei C, Li L. Recent progress and scientific challenges in multi-material additive manufacturing via laser-based powder bed fusion. Virtual Phys Prototyp. 2021;16 (3):347–371. doi:10.1080/17452759.2021.1928520 [5] Lin X, Wang Q, Fuh JYH, et al. Motion feature based melt pool monitoring for selective laser melting process. J Mater Process Technol. 2022;303:117523. doi:10.1016/j. jmatprotec.2022.117523 [6] Gockel J, Sheridan L, Koerper B, et al. The influence of additive manufacturing processing parameters on surface roughness and fatigue life. Int J Fatigue. 2019;124:380–388. doi:10.1016/j.ijfatigue.2019.03.025 [7] Nicoletto G. Influence of rough as-built surfaces on smooth and notched fatigue behavior of L-PBF AlSi10Mg. Addit Manuf. 2020;34:101251. doi:10.1016/j. addma.2020.101251 [8] Spece H, Yu T, Law AW, et al. 3D printed porous PEEK created via fused filament fabrication for osteoconductive orthopaedic surfaces. J Mech Behav Biomed Mater. 2020;109:103850. doi:10.1115/1.0004270v [9] Andrukhov O, Huber R, Shi B, et al. Proliferation, behavior, and differentiation of osteoblasts on surfaces of different microroughness. Dent Mater. 2016;32(11):1374–1384. doi:10.1016/j.dental.2016.08.217 [10] Dai N, Zhang LC, Zhang J, et al. Corrosion behavior of selective laser melted Ti-6Al-4 V alloy in NaCl solution. Corros Sci. 2016;102:484–489. doi:10.1016/j.corsci.2015. 10.041 [11] Li EL, Wang L, Yu AB, et al. A three-phase model for simulation of heat transfer and melt pool behaviour in laser powder bed fusion process. Powder Technol. 2021;381:298–312. doi:10.1016/j.powtec.2020.11.061 [12] Liao B, Xia RF, Li W, et al. 3D-printed ti6al4v scaffolds with graded triply periodic minimal surface structure for bone tissue engineering. J Mater Eng Perform. 2021;30:4993– 5004. doi:10.1007/s11665-021-05580-z [13] Li E, Zhou Z, Wang L, et al. Melt pool dynamics and pores formation in multi-track studies in laser powder bed fusion process. Powder Technol. 2022;405:117533. doi:10.1016/j.powtec.2022.117533 [14] Guo L, Geng S, Gao X, et al. Numerical simulation of heat transfer and fluid flow during nanosecond pulsed laser processing of Fe78Si9B13 amorphous alloys. Int J Heat Mass Transfer. 2021;170:121003. doi:10.1016/j.ijheatma sstransfer.2021.121003 [15] Guo L, Li Y, Geng S, et al. Numerical and experimental analysis for morphology evolution of 6061 aluminum alloy during nanosecond pulsed laser cleaning. Surf Coat Technol. 2022;432:128056. doi:10.1016/j.surfcoat. 2021.128056 [16] Li S, Liu D, Mi H, et al. Numerical simulation on evolution process of molten pool and solidification characteristics of melt track in selective laser melting of ceramic powder. Ceram Int. 2022;48(13):18302–18315. doi:10. 1016/j.ceramint.2022.03.089 [17] Aboulkhair NT, Maskery I, Tuck C, et al. On the formation of AlSi10Mg single tracks and layers in selective laser melting: microstructure and nano-mechanical properties. J Mater Process Technol. 2016;230:88–98. doi:10.1016/j. jmatprotec.2015.11.016 [18] Thijs L, Kempen K, Kruth JP, et al. Fine-structured aluminium products with controllable texture by selective laser melting of pre-alloyed AlSi10Mg powder. Acta Mater. 2013;61(5):1809–1819. doi:10.1016/j.actamat.2012.11.052 [19] Qiu C, Adkins NJE, Attallah MM. Microstructure and tensile properties of selectively laser-melted and of HIPed laser-melted Ti–6Al–4 V. Mater Sci Eng A. 2013;578:230–239. doi:10.1016/j.msea.2013.04.099 [20] Kazemi Z, Soleimani M, Rokhgireh H, et al. Melting pool simulation of 316L samples manufactured by selective laser melting method, comparison with experimental results. Int J Therm Sci. 2022;176:107538. doi:10.1016/j. ijthermalsci.2022.107538 [21] Cao L. Workpiece-scale numerical simulations of SLM molten pool dynamic behavior of 316L stainless steel. Comput Math Appl. 2021;96:209–228. doi:10.1016/j. camwa.2020.04.020 [22] Liu B, Fang G, Lei L, et al. Predicting the porosity defects in selective laser melting (SLM) by molten pool geometry. Int J Mech Sci. 2022;228:107478. doi:10.1016/j.ijmecsci. 2022.107478 [23] Ur Rehman A, Pitir F, Salamci MU. Full-field mapping and flow quantification of melt pool dynamics in laser powder bed fusion of SS316L. Materials. 2021;14(21):6264. doi:10. 3390/ma14216264 [24] Chia HY, Wang L, Yan W. Influence of oxygen content on melt pool dynamics in metal additive manufacturing: high-fidelity modeling with experimental validation. Acta Mater. 2023;249:118824. doi:10.1016/j.actamat. 2023.118824 [25] Cheng B, Loeber L, Willeck H, et al. Computational investigation of melt pool process dynamics and pore formation in laser powder bed fusion. J Mater Eng Perform. 2019;28:6565–6578. doi:10.1007/s11665-019- 04435-y [26] Li X, Guo Q, Chen L, et al. Quantitative investigation of gas flow, powder-gas interaction, and powder behavior under different ambient pressure levels in laser powder bed fusion. Int J Mach Tools Manuf. 2021;170:103797. doi:10.1016/j.ijmachtools.2021.103797 [27] Wu Y, Li M, Wang J, et al. Powder-bed-fusion additive manufacturing of molybdenum: process simulation, optimization, and property prediction. Addit Manuf. 2022;58:103069. doi:10.1016/j.addma.2022.103069 [28] Wu S, Yang Y, Huang Y, et al. Study on powder particle behavior in powder spreading with discrete element method and its critical implications for binder jetting additive manufacturing processes. Virtual Phys Prototyp. 2023;18(1):e2158877. doi:10.1080/17452759.2022.2158877 [29] Klassen A, Schakowsky T, Kerner C. Evaporation model for beam based additive manufacturing using free surface lattice Boltzmann methods. J Phys D Appl Phys. 2014;47 (27):275303. doi:10.1088/0022-3727/47/27/275303 [30] Cao L. Mesoscopic-scale numerical simulation including the influence of process parameters on slm single-layer multi-pass formation. Metall Mater Trans A. 2020;51:4130–4145. doi:10.1007/s11661-020-05831-z [31] Zhuang JR, Lee YT, Hsieh WH, et al. Determination of melt pool dimensions using DOE-FEM and RSM with process window during SLM of Ti6Al4V powder. Opt Laser Technol. 2018;103:59–76. doi:10.1016/j.optlastec.2018. 01.013 [32] Li Y, Gu D. Thermal behavior during selective laser melting of commercially pure titanium powder: numerical simulation and experimental study. Addit Manuf. 2014;1–4:99–109. doi:10.1016/j.addma.2014.09.001 [33] Dai D, Gu D. Thermal behavior and densification mechanism during selective laser melting of copper matrix composites: simulation and experiments. Mater Des. 2014;55 (0):482–491. doi:10.1016/j.matdes.2013.10.006 [34] Wang S, Zhu L, Dun Y, et al. Multi-physics modeling of direct energy deposition process of thin-walled structures: defect analysis. Comput Mech. 2021;67:c1229– c1242. doi:10.1007/s00466-021-01992-9 [35] Wu J, Zheng J, Zhou H, et al. Molten pool behavior and its mechanism during selective laser melting of polyamide 6 powder: single track simulation and experiments. Mater Res Express. 2019;6. doi:10.1088/2053-1591/ab2747 [36] Cho JH, Farson DF, Milewski JO, et al. Weld pool flows during initial stages of keyhole formation in laser welding. J Phys D Appl Phys. 2009;42. doi:10.1088/0022- 3727/42/17/175502 [37] Sinha KN. Identification of a suitable volumetric heat source for modelling of selective laser melting of Ti6Al4V powder using numerical and experimental validation approach. Int J Adv Manuf Technol. 2018;99:2257–2270. doi:10.1007/s00170-018-2631-4 [38] Fu CH, Guo YB. Three-dimensional temperature gradient mechanism in selective laser melting of Ti-6Al-4V. J Manuf Sci Eng. 2014;136(6):061004. doi:10.1115/1.4028539 [39] Ansari P, Rehman AU, Pitir F, et al. Selective laser melting of 316 l austenitic stainless steel: detailed process understanding using multiphysics simulation and experimentation. Metals. 2021;11(7):1076. doi:10.3390/met11071076 [40] Zhao C, Shi B, Chen S, et al. Laser melting modes in metal powder bed fusion additive manufacturing. Rev Mod Phys. 2022;94(4):045002. doi:10.1103/revmodphys.94. 045002 [41] Bertoli US, Wolfer AJ, Matthews MJ, et al. On the limitations of volumetric energy density as a design parameter for selective laser melting. Mater Des. 2017;113:331–340. doi:10.1016/j.matdes.2016.10.037 [42] Dash A, Kamaraj A. Prediction of the shift in melting mode during additive manufacturing of 316L stainless steel. Mater Today Commun. 2023: 107238. doi:10.1016/j. mtcomm.2023.107238 [43] Majeed M, Khan HM, Rasheed I. Finite element analysis of melt pool thermal characteristics with passing laser in SLM process. Optik. 2019;194:163068. doi:10.1016/j.ijleo. 2019.163068
In order to comprehensively reveal the evolutionary dynamics of the molten pool and the state of motion of the fluid during the high-precision laser powder bed fusion (HP-LPBF) process, this study aims to deeply investigate the specific manifestations of the multiphase flow, solidification phenomena, and heat transfer during the process by means of numerical simulation methods. Numerical simulation models of SS316L single-layer HP-LPBF formation with single and double tracks were constructed using the discrete element method and the computational fluid dynamics method. The effects of various factors such as Marangoni convection, surface tension, vapor recoil, gravity, thermal convection, thermal radiation, and evaporative heat dissipation on the heat and mass transfer in the molten pool have been paid attention to during the model construction process. The results show that the molten pool exhibits a “comet” shape, in which the temperature gradient at the front end of the pool is significantly larger than that at the tail end, with the highest temperature gradient up to 1.69 × 108 K/s. It is also found that the depth of the second track is larger than that of the first one, and the process parameter window has been determined preliminarily. In addition, the application of HP-LPBF technology helps to reduce the surface roughness and minimize the forming size.
Laser powder bed fusion (LPBF) has become a research hotspot in the field of additive manufacturing of metals due to its advantages of high-dimensional accuracy, good surface quality, high density, and high material utilization.1,2 With the rapid development of electronics, medical, automotive, biotechnology, energy, communication, and optics, the demand for microfabrication technology is increasing day by day.3 High-precision laser powder bed fusion (HP-LPBF) is one of the key manufacturing technologies for tiny parts in the fields of electronics, medical, automotive, biotechnology, energy, communication, and optics because of its process characteristics such as small focal spot diameter, small powder particle size, and thin powder layup layer thickness.4–13 Compared with LPBF, HP-LPBF has the significant advantages of smaller focal spot diameter, smaller powder particle size, and thinner layer thickness. These advantages make HP-LPBF perform better in producing micro-fine parts, high surface quality, and parts with excellent mechanical properties.
HP-LPBF is in the exploratory stage, and researchers have already done some exploratory studies on the focal spot diameter, the amount of defocusing, and the powder particle size. In order to explore the influence of changing the laser focal spot diameter on the LPBF process characteristics of the law, Wildman et al.14 studied five groups of different focal spot diameter LPBF forming 316L stainless steel (SS316L) processing effect, the smallest focal spot diameter of 26 μm, and the results confirm that changing the focal spot diameter can be achieved to achieve the energy control, so as to control the quality of forming. Subsequently, Mclouth et al.15 proposed the laser out-of-focus amount (focal spot diameter) parameter, which characterizes the distance between the forming plane and the laser focal plane. The laser energy density was controlled by varying the defocusing amount while keeping the laser parameters constant. Sample preparation at different focal positions was investigated, and their microstructures were characterized. The results show that the samples at the focal plane have finer microstructure than those away from the focal plane, which is the effect of higher power density and smaller focal spot diameter. In order to explore the influence of changing the powder particle size on the characteristics of the LPBF process, Qian et al.16 carried out single-track scanning simulations on powder beds with average powder particle sizes of 70 and 40 μm, respectively, and the results showed that the melt tracks sizes were close to each other under the same process parameters for the two particle-size distributions and that the molten pool of powder beds with small particles was more elongated and the edges of the melt tracks were relatively flat. In order to explore the superiority of HP-LPBF technology, Xu et al.17 conducted a comparative analysis of HP-LPBF and conventional LPBF of SS316L. The results showed that the average surface roughness of the top surface after forming by HP-LPBF could reach 3.40 μm. Once again, it was verified that HP-LPBF had higher forming quality than conventional LPBF. On this basis, Wei et al.6 comparatively analyzed the effects of different laser focal spot diameters on different powder particle sizes formed by LPBF. The results showed that the smaller the laser focal spot diameter, the fewer the defects on the top and side surfaces. The above research results confirm that reducing the laser focal spot diameter can obtain higher energy density and thus better forming quality.
LPBF involves a variety of complex systems and mechanisms, and the final quality of the part is influenced by a large number of process parameters.18–24 Some research results have shown that there are more than 50 factors affecting the quality of the specimen. The influencing factors are mainly categorized into three main groups: (1) laser parameters, (2) powder parameters, and (3) equipment parameters, which interact with each other to determine the final specimen quality. With the continuous development of technologies such as computational materials science and computational fluid dynamics (CFD), the method of studying the influence of different factors on the forming quality of LPBF forming process has been shifted from time-consuming and laborious experimental characterization to the use of numerical simulation methods. As a result, more and more researchers are adopting this approach for their studies. Currently, numerical simulation studies on LPBF are mainly focused on the exploration of molten pool, temperature distribution, and residual stresses.
Finite element simulation based on continuum mechanics and free surface fluid flow modeling based on fluid dynamics are two common approaches to study the behavior of LPBF molten pool.25–28 Finite element simulation focuses on the temperature and thermal stress fields, treats the powder bed as a continuum, and determines the molten pool size by plotting the elemental temperature above the melting point. In contrast, fluid dynamics modeling can simulate the 2D or 3D morphology of the metal powder pile and obtain the powder size and distribution by certain algorithms.29 The flow in the molten pool is mainly affected by recoil pressure and the Marangoni effect. By simulating the molten pool formation, it is possible to predict defects, molten pool shape, and flow characteristics, as well as the effect of process parameters on the molten pool geometry.30–34 In addition, other researchers have been conducted to optimize the laser processing parameters through different simulation methods and experimental data.35–46 Crystal growth during solidification is studied to further understand the effect of laser parameters on dendritic morphology and solute segregation.47–54 A multi-scale system has been developed to describe the fused deposition process during 3D printing, which is combined with the conductive heat transfer model and the dendritic solidification model.55,56
Relevant scholars have adopted various different methods for simulation, such as sequential coupling theory,57 Lagrangian and Eulerian thermal models,58 birth–death element method,25 and finite element method,59 in order to reveal the physical phenomena of the laser melting process and optimize the process parameters. Luo et al.60 compared the LPBF temperature field and molten pool under double ellipsoidal and Gaussian heat sources by ANSYS APDL and found that the diffusion of the laser energy in the powder significantly affects the molten pool size and the temperature field.
The thermal stresses obtained from the simulation correlate with the actual cracks,61 and local preheating can effectively reduce the residual stresses.62 A three-dimensional thermodynamic finite element model investigated the temperature and stress variations during laser-assisted fabrication and found that powder-to-solid conversion increases the temperature gradient, stresses, and warpage.63 Other scholars have predicted residual stresses and part deflection for LPBF specimens and investigated the effects of deposition pattern, heat, laser power, and scanning strategy on residual stresses, noting that high-temperature gradients lead to higher residual stresses.64–67
In short, the process of LPBF forming SS316L is extremely complex and usually involves drastic multi-scale physicochemical changes that will only take place on a very small scale. Existing literature employs DEM-based mesoscopic-scale numerical simulations to investigate the effects of process parameters on the molten pool dynamics of LPBF-formed SS316L. However, a few studies have been reported on the key mechanisms of heating and solidification, spatter, and convective behavior of the molten pool of HP-LPBF-formed SS316L with small laser focal spot diameters. In this paper, the geometrical properties of coarse and fine powder particles under three-dimensional conditions were first calculated using DEM. Then, numerical simulation models for single-track and double-track cases in the single-layer HP-LPBF forming SS316L process were developed at mesoscopic scale using the CFD method. The flow genesis of the melt in the single-track and double-track molten pools is discussed, and their 3D morphology and dimensional characteristics are discussed. In addition, the effects of laser process parameters, powder particle size, and laser focal spot diameter on the temperature field, characterization information, and defects in the molten pool are discussed.
II. MODELING
A. 3D powder bed modeling
HP-LPBF is an advanced processing technique for preparing target parts layer by layer stacking, the process of which involves repetitive spreading and melting of powders. In this process, both the powder spreading and the morphology of the powder bed are closely related to the results of the subsequent melting process, while the melted surface also affects the uniform distribution of the next layer of powder. For this reason, this chapter focuses on the modeling of the physical action during the powder spreading process and the theory of DEM to establish the numerical model of the powder bed, so as to lay a solid foundation for the accuracy of volume of fluid (VOF) and CFD.
1. DEM
DEM is a numerical technique for calculating the interaction of a large number of particles, which calculates the forces and motions of the spheres by considering each powder sphere as an independent unit. The motion of the powder particles follows the laws of classical Newtonian mechanics, including translational and rotational,38,68–70 which are expressed as follows:����¨=���+∑��ij,
(1)����¨=∑�(�ij×�ij),
(2)
where �� is the mass of unit particle i in kg, ��¨ is the advective acceleration in m/s2, And g is the gravitational acceleration in m/s2. �ij is the force in contact with the neighboring particle � in N. �� is the rotational inertia of the unit particle � in kg · m2. ��¨ is the unit particle � angular acceleration in rad/s2. �ij is the vector pointing from unit particle � to the contact point of neighboring particle �.
Equations (1) and (2) can be used to calculate the velocity and angular velocity variations of powder particles to determine their positions and velocities. A three-dimensional powder bed model of SS316L was developed using DEM. The powder particles are assumed to be perfect spheres, and the substrate and walls are assumed to be rigid. To describe the contact between the powder particles and between the particles and the substrate, a non-slip Hertz–Mindlin nonlinear spring-damping model71 was used with the following expression:�hz=��������+��[(�����ij−�eff����)−(�����+�eff����)],
(3)
where �hz is the force calculated using the Hertzian in M. �� and �� are the radius of unit particles � and � in m, respectively. �� is the overlap size of the two powder particles in m. ��, �� are the elastic constants in the normal and tangential directions, respectively. �ij is the unit vector connecting the centerlines of the two powder particles. �eff is the effective mass of the two powder particles in kg. �� and �� are the viscoelastic damping constants in the normal and tangential directions, respectively. �� and �� are the components of the relative velocities of the two powder particles. ��� is the displacement vector between two spherical particles. The schematic diagram of overlapping powder particles is shown in Fig. 1.
Schematic diagram of overlapping powder particles.
Because the particle size of the powder used for HP-LPBF is much smaller than 100 μm, the effect of van der Waals forces must be considered. Therefore, the cohesive force �jkr of the Hertz–Mindlin model was used instead of van der Waals forces,72 with the following expression:�jkr=−4��0�*�1.5+4�*3�*�3,
(4)1�*=(1−��2)��+(1−��2)��,
(5)1�*=1��+1��,
(6)
where �* is the equivalent Young’s modulus in GPa; �* is the equivalent particle radius in m; �0 is the surface energy of the powder particles in J/m2; α is the contact radius in m; �� and �� are the Young’s modulus of the unit particles � and �, respectively, in GPa; and �� and �� are the Poisson’s ratio of the unit particles � and �, respectively.
2. Model building
Figure 2 shows a 3D powder bed model generated using DEM with a coarse powder geometry of 1000 × 400 × 30 μm3. The powder layer thickness is 30 μm, and the powder bed porosity is 40%. The average particle size of this spherical powder is 31.7 μm and is normally distributed in the range of 15–53 μm. The geometry of the fine powder was 1000 × 400 × 20 μm3, with a layer thickness of 20 μm, and the powder bed porosity of 40%. The average particle size of this spherical powder is 11.5 μm and is normally distributed in the range of 5–25 μm. After the 3D powder bed model is generated, it needs to be imported into the CFD simulation software for calculation, and the imported geometric model is shown in Fig. 3. This geometric model is mainly composed of three parts: protective gas, powder bed, and substrate. Under the premise of ensuring the accuracy of the calculation, the mesh size is set to 3 μm, and the total number of coarse powder meshes is 1 704 940. The total number of fine powder meshes is 3 982 250.
Geometric modeling of the powder bed computational domain: (a) coarse powder, (b) fine powder.
B. Modeling of fluid mechanics simulation
In order to solve the flow, melting, and solidification problems involved in HP-LPBF molten pool, the study must follow the three governing equations of conservation of mass, conservation of energy, and conservation of momentum.73 The VOF method, which is the most widely used in fluid dynamics, is used to solve the molten pool dynamics model.
1. VOF
VOF is a method for tracking the free interface between the gas and liquid phases on the molten pool surface. The core idea of the method is to define a volume fraction function F within each grid, indicating the proportion of the grid space occupied by the material, 0 ≤ F ≤ 1 in Fig. 4. Specifically, when F = 0, the grid is empty and belongs to the gas-phase region; when F = 1, the grid is completely filled with material and belongs to the liquid-phase region; and when 0 < F < 1, the grid contains free surfaces and belongs to the mixed region. The direction normal to the free surface is the direction of the fastest change in the volume fraction F (the direction of the gradient of the volume fraction), and the direction of the gradient of the volume fraction can be calculated from the values of the volume fractions in the neighboring grids.74 The equations controlling the VOF are expressed as follows:𝛻����+�⋅(��→)=0,
(7)
where t is the time in s and �→ is the liquid velocity in m/s.
The material parameters of the mixing zone are altered due to the inclusion of both the gas and liquid phases. Therefore, in order to represent the density of the mixing zone, the average density �¯ is used, which is expressed as follows:72�¯=(1−�1)�gas+�1�metal,
(8)
where �1 is the proportion of liquid phase, �gas is the density of protective gas in kg/m3, and �metal is the density of metal in kg/m3.
2. Control equations and boundary conditions
Figure 5 is a schematic diagram of the HP-LPBF melting process. First, the laser light strikes a localized area of the material and rapidly heats up the area. Next, the energy absorbed in the region is diffused through a variety of pathways (heat conduction, heat convection, and surface radiation), and this process triggers complex phase transition phenomena (melting, evaporation, and solidification). In metals undergoing melting, the driving forces include surface tension and the Marangoni effect, recoil due to evaporation, and buoyancy due to gravity and uneven density. The above physical phenomena interact with each other and do not occur independently.
Laser heat sourceThe Gaussian surface heat source model is used as the laser heat source model with the following expression:�=2�0����2exp(−2�12��2),(9)where � is the heat flow density in W/m2, �0 is the absorption rate of SS316L, �� is the radius of the laser focal spot in m, and �1 is the radial distance from the center of the laser focal spot in m. The laser focal spot can be used for a wide range of applications.
Energy absorptionThe formula for calculating the laser absorption �0 of SS316L is as follows:�0=0.365(�0[1+�0(�−20)]/�)0.5,(10)where �0 is the direct current resistivity of SS316L at 20 °C in Ω m, �0 is the resistance temperature coefficient in ppm/°C, � is the temperature in °C, and � is the laser wavelength in m.
Heat transferThe basic principle of heat transfer is conservation of energy, which is expressed as follows:𝛻𝛻𝛻�(��)��+�·(��→�)=�·(�0����)+��,(11)where � is the density of liquid phase SS316L in kg/m3, �� is the specific heat capacity of SS316L in J/(kg K), 𝛻� is the gradient operator, t is the time in s, T is the temperature in K, 𝛻�� is the temperature gradient, �→ is the velocity vector, �0 is the coefficient of thermal conduction of SS316L in W/(m K), and �� is the thermal energy dissipation term in the molten pool.
Molten pool flowThe following three conditions need to be satisfied for the molten pool to flow:
Conservation of mass with the following expression:𝛻�·(��→)=0.(12)
Conservation of momentum (Navier–Stokes equation) with the following expression:𝛻𝛻𝛻𝛻���→��+�(�→·�)�→=�·[−pI+�(��→+(��→)�)]+�,(13)where � is the pressure in Pa exerted on the liquid phase SS316L microelement, � is the unit matrix, � is the fluid viscosity in N s/m2, and � is the volumetric force (gravity, atmospheric pressure, surface tension, vapor recoil, and the Marangoni effect).
Surface tension and the Marangoni effectThe effect of temperature on the surface tension coefficient is considered and set as a linear relationship with the following expression:�=�0−��dT(�−��),(14)where � is the surface tension of the molten pool at temperature T in N/m, �� is the melting temperature of SS316L in K, �0 is the surface tension of the molten pool at temperature �� in Pa, and σdσ/ dT is the surface tension temperature coefficient in N/(m K).In general, surface tension decreases with increasing temperature. A temperature gradient causes a gradient in surface tension that drives the liquid to flow, known as the Marangoni effect.
Metal vapor recoilAt higher input energy densities, the maximum temperature of the molten pool surface reaches the evaporation temperature of the material, and a gasification recoil pressure occurs vertically downward toward the molten pool surface, which will be the dominant driving force for the molten pool flow.75 The expression is as follows:��=0.54�� exp ���−���0���,(15)where �� is the gasification recoil pressure in Pa, �� is the ambient pressure in kPa, �� is the latent heat of evaporation in J/kg, �0 is the gas constant in J/(mol K), T is the surface temperature of the molten pool in K, and Te is the evaporation temperature in K.
Solid–liquid–gas phase transitionWhen the laser hits the powder layer, the powder goes through three stages: heating, melting, and solidification. During the solidification phase, mutual transformations between solid, liquid, and gaseous states occur. At this point, the latent heat of phase transition absorbed or released during the phase transition needs to be considered.68 The phase transition is represented based on the relationship between energy and temperature with the following expression:�=�����,(�<��),�(��)+�−����−����,(��<�<��)�(��)+(�−��)����,(��<�),,(16)where �� and �� are solid and liquid phase density, respectively, of SS316L in kg/m3. �� and �� unit volume of solid and liquid phase-specific heat capacity, respectively, of SS316L in J/(kg K). �� and ��, respectively, are the solidification temperature and melting temperature of SS316L in K. �� is the latent heat of the phase transition of SS316L melting in J/kg.
3. Assumptions
The CFD model was computed using the commercial software package FLOW-3D.76 In order to simplify the calculation and solution process while ensuring the accuracy of the results, the model makes the following assumptions:
It is assumed that the effects of thermal stress and material solid-phase thermal expansion on the calculation results are negligible.
The molten pool flow is assumed to be a Newtonian incompressible laminar flow, while the effects of liquid thermal expansion and density on the results are neglected.
It is assumed that the surface tension can be simplified to an equivalent pressure acting on the free surface of the molten pool, and the effect of chemical composition on the results is negligible.
Neglecting the effect of the gas flow field on the molten pool.
The mass loss due to evaporation of the liquid metal is not considered.
The influence of the plasma effect of the molten metal on the calculation results is neglected.
It is worth noting that the formulation of assumptions requires a trade-off between accuracy and computational efficiency. In the above models, some physical phenomena that have a small effect or high difficulty on the calculation results are simplified or ignored. Such simplifications make numerical simulations more efficient and computationally tractable, while still yielding accurate results.
4. Initial conditions
The preheating temperature of the substrate was set to 393 K, at which time all materials were in the solid state and the flow rate was zero.
5. Material parameters
The material used is SS316L and the relevant parameters required for numerical simulations are shown in Table I.46,77,78
TABLE I.
SS316L-related parameters.
Property
Symbol
Value
Density of solid metal (kg/m3)
�metal
7980
Solid phase line temperature (K)
��
1658
Liquid phase line temperature (K)
��
1723
Vaporization temperature (K)
��
3090
Latent heat of melting ( J/kg)
��
2.60×105
Latent heat of evaporation ( J/kg)
��
7.45×106
Surface tension of liquid phase (N /m)
�
1.60
Liquid metal viscosity (kg/m s)
��
6×10−3
Gaseous metal viscosity (kg/m s)
�gas
1.85×10−5
Temperature coefficient of surface tension (N/m K)
��/�T
0.80×10−3
Molar mass ( kg/mol)
M
0.05 593
Emissivity
�
0.26
Laser absorption
�0
0.35
Ambient pressure (kPa)
��
101 325
Ambient temperature (K)
�0
300
Stefan–Boltzmann constant (W/m2 K4)
�
5.67×10−8
Thermal conductivity of metals ( W/m K)
�
24.55
Density of protective gas (kg/m3)
�gas
1.25
Coefficient of thermal expansion (/K)
��
16×10−6
Generalized gas constant ( J/mol K)
R
8.314
III. RESULTS AND DISCUSSION
With the objective of studying in depth the evolutionary patterns of single-track and double-track molten pool development, detailed observations were made for certain specific locations in the model, as shown in Fig. 6. In this figure, P1 and P2 represent the longitudinal tangents to the centers of the two melt tracks in the XZ plane, while L1 is the transverse profile in the YZ plane. The scanning direction is positive and negative along the X axis. Points A and B are the locations of the centers of the molten pool of the first and second melt tracks, respectively (x = 1.995 × 10−4, y = 5 × 10−7, and z = −4.85 × 10−5).
A series of single-track molten pool simulation experiments were carried out in order to investigate the influence law of laser power as well as scanning speed on the HP-LPBF process. Figure 7 demonstrates the evolution of the 3D morphology and temperature field of the single-track molten pool in the time period of 50–500 μs under a laser power of 100 W and a scanning speed of 800 mm/s. The powder bed is in the natural cooling state. When t = 50 μs, the powder is heated by the laser heat and rapidly melts and settles to form the initial molten pool. This process is accompanied by partial melting of the substrate and solidification together with the melted powder. The molten pool rapidly expands with increasing width, depth, length, and temperature, as shown in Fig. 7(a). When t = 150 μs, the molten pool expands more obviously, and the temperature starts to transfer to the surrounding area, forming a heat-affected zone. At this point, the width of the molten pool tends to stabilize, and the temperature in the center of the molten pool has reached its peak and remains largely stable. However, the phenomenon of molten pool spatter was also observed in this process, as shown in Fig. 7(b). As time advances, when t = 300 μs, solidification begins to occur at the tail of the molten pool, and tiny ripples are produced on the solidified surface. This is due to the fact that the melt flows toward the region with large temperature gradient under the influence of Marangoni convection and solidifies together with the melt at the end of the bath. At this point, the temperature gradient at the front of the bath is significantly larger than at the end. While the width of the molten pool was gradually reduced, the shape of the molten pool was gradually changed to a “comet” shape. In addition, a slight depression was observed at the top of the bath because the peak temperature at the surface of the bath reached the evaporation temperature, which resulted in a recoil pressure perpendicular to the surface of the bath downward, creating a depressed region. As the laser focal spot moves and is paired with the Marangoni convection of the melt, these recessed areas will be filled in as shown in Fig. 7(c). It has been shown that the depressed regions are the result of the coupled effect of Marangoni convection, recoil pressure, and surface tension.79 By t = 500 μs, the width and height of the molten pool stabilize and show a “comet” shape in Fig. 7(d).
Single-track molten pool process: (a) t = 50 ��, (b) t = 150 ��, (c) t = 300 ��, (d) t = 500 ��.
Figure 8 depicts the velocity vector diagram of the P1 profile in a single-track molten pool, the length of the arrows represents the magnitude of the velocity, and the maximum velocity is about 2.36 m/s. When t = 50 μs, the molten pool takes shape, and the velocities at the two ends of the pool are the largest. The variation of the velocities at the front end is especially more significant in Fig. 8(a). As the time advances to t = 150 μs, the molten pool expands rapidly, in which the velocity at the tail increases and changes more significantly, while the velocity at the front is relatively small. At this stage, the melt moves backward from the center of the molten pool, which in turn expands the molten pool area. The melt at the back end of the molten pool center flows backward along the edge of the molten pool surface and then converges along the edge of the molten pool to the bottom center, rising to form a closed loop. Similarly, a similar closed loop is formed at the front end of the center of the bath, but with a shorter path. However, a large portion of the melt in the center of the closed loop formed at the front end of the bath is in a nearly stationary state. The main cause of this melt flow phenomenon is the effect of temperature gradient and surface tension (the Marangoni effect), as shown in Figs. 8(b) and 8(e). This dynamic behavior of the melt tends to form an “elliptical” pool. At t = 300 μs, the tendency of the above two melt flows to close the loop is more prominent and faster in Fig. 8(c). When t = 500 μs, the velocity vector of the molten pool shows a stable trend, and the closed loop of melt flow also remains stable. With the gradual laser focal spot movement, the melt is gradually solidified at its tail, and finally, a continuous and stable single track is formed in Fig. 8(d).
Vector plot of single-track molten pool velocity in XZ longitudinal section: (a) t = 50 ��, (b) t = 150 ��, (c) t = 300 ��, (d) t = 500 ��, (e) molten pool flow.
In order to explore in depth the transient evolution of the molten pool, the evolution of the single-track temperature field and the melt flow was monitored in the YZ cross section. Figure 9(a) shows the state of the powder bed at the initial moment. When t = 250 μs, the laser focal spot acts on the powder bed and the powder starts to melt and gradually collects in the molten pool. At this time, the substrate will also start to melt, and the melt flow mainly moves in the downward and outward directions and the velocity is maximum at the edges in Fig. 9(b). When t = 300 μs, the width and depth of the molten pool increase due to the recoil pressure. At this time, the melt flows more slowly at the center, but the direction of motion is still downward in Fig. 9(c). When t = 350 μs, the width and depth of the molten pool further increase, at which time the intensity of the melt flow reaches its peak and the direction of motion remains the same in Fig. 9(d). When t = 400 μs, the melt starts to move upward, and the surrounding powder or molten material gradually fills up, causing the surface of the molten pool to begin to flatten. At this time, the maximum velocity of the melt is at the center of the bath, while the velocity at the edge is close to zero, and the edge of the melt starts to solidify in Fig. 9(e). When t = 450 μs, the melt continues to move upward, forming a convex surface of the melt track. However, the melt movement slows down, as shown in Fig. 9(f). When t = 500 μs, the melt further moves upward and its speed gradually becomes smaller. At the same time, the melt solidifies further, as shown in Fig. 9(g). When t = 550 μs, the melt track is basically formed into a single track with a similar “mountain” shape. At this stage, the velocity is close to zero only at the center of the molten pool, and the flow behavior of the melt is poor in Fig. 9(h). At t = 600 μs, the melt stops moving and solidification is rapidly completed. Up to this point, a single track is formed in Fig. 9(i). During the laser action on the powder bed, the substrate melts and combines with the molten state powder. The powder-to-powder fusion is like the convergence of water droplets, which are rapidly fused by surface tension. However, the fusion between the molten state powder and the substrate occurs driven by surface tension, and the molten powder around the molten pool is pulled toward the substrate (a wetting effect occurs), which ultimately results in the formation of a monolithic whole.38,80,81
Evolution of single-track molten pool temperature and melt flow in the YZ cross section: (a) t = 0 ��, (b) t = 250 ��, (c) t = 300 ��, (d) t = 350 ��, (e) t = 400 ��, (f) t = 450 ��, (g) t = 500 ��, (h) t = 550 ��, (i) t = 600 ��.
The wetting ability between the liquid metal and the solid substrate in the molten pool directly affects the degree of balling of the melt,82,83 and the wetting ability can be measured by the contact angle of a single track in Fig. 10. A smaller value of contact angle represents better wettability. The contact angle α can be calculated by�=�1−�22,
(17)
where �1 and �2 are the contact angles of the left and right regions, respectively.
Relevant studies have confirmed that the wettability is better at a contact angle α around or below 40°.84 After measurement, a single-track contact angle α of about 33° was obtained under this process parameter, which further confirms the good wettability.
B. Double-track simulation
In order to deeply investigate the influence of hatch spacing on the characteristics of the HP-LPBF process, a series of double-track molten pool simulation experiments were systematically carried out. Figure 11 shows in detail the dynamic changes of the 3D morphology and temperature field of the double-track molten pool in the time period of 2050–2500 μs under the conditions of laser power of 100 W, scanning speed of 800 mm/s, and hatch spacing of 0.06 mm. By comparing the study with Fig. 7, it is observed that the basic characteristics of the 3D morphology and temperature field of the second track are similar to those of the first track. However, there are subtle differences between them. The first track exhibits a basically symmetric shape, but the second track morphology shows a slight deviation influenced by the difference in thermal diffusion rate between the solidified metal and the powder. Otherwise, the other characteristic information is almost the same as that of the first track. Figure 12 shows the velocity vector plot of the P2 profile in the double-track molten pool, with a maximum velocity of about 2.63 m/s. The melt dynamics at both ends of the pool are more stable at t = 2050 μs, where the maximum rate of the second track is only 1/3 of that of the first one. Other than that, the rest of the information is almost no significant difference from the characteristic information of the first track. Figure 13 demonstrates a detailed observation of the double-track temperature field and melts flow in the YZ cross section, and a comparative study with Fig. 9 reveals that the width of the second track is slightly wider. In addition, after the melt direction shifts from bottom to top, the first track undergoes four time periods (50 μs) to reach full solidification, while the second track takes five time periods. This is due to the presence of significant heat buildup in the powder bed after the forming of the first track, resulting in a longer dynamic time of the melt and an increased molten pool lifetime. In conclusion, the level of specimen forming can be significantly optimized by adjusting the laser power and hatch spacing.
Evolution of double-track molten pool temperature and melt flow in the YZ cross section: (a) t = 2250 ��, (b) t = 2300 ��, (c) t = 2350 ��, (d) t = 2400 ��, (e) t = 2450 ��, (f) t = 2500 ��, (g) t = 2550 ��, (h) t = 2600 ��, (i) t = 2650 ��.
In order to quantitatively detect the molten pool dimensions as well as the remolten region dimensions, the molten pool characterization information in Fig. 14 is constructed by drawing the boundary on the YZ cross section based on the isothermal surface of the liquid phase line. It can be observed that the heights of the first track and second track are basically the same, but the depth of the second track increases relative to the first track. The molten pool width is mainly positively correlated with the laser power as well as the scanning speed (the laser line energy density �). However, the remelted zone width is negatively correlated with the hatch spacing (the overlapping ratio). Overall, the forming quality of the specimens can be directly influenced by adjusting the laser power, scanning speed, and hatch spacing.
Double-track molten pool characterization information on YZ cross section.
In order to study the variation rule of the temperature in the center of the molten pool with time, Fig. 15 demonstrates the temperature variation curves with time for two reference points, A and B. Among them, the red dotted line indicates the liquid phase line temperature of SS316L. From the figure, it can be seen that the maximum temperature at the center of the molten pool in the first track is lower than that in the second track, which is mainly due to the heat accumulation generated after passing through the first track. The maximum temperature gradient was calculated to be 1.69 × 108 K/s. When the laser scanned the first track, the temperature in the center of the molten pool of the second track increased slightly. Similarly, when the laser scanned the second track, a similar situation existed in the first track. Since the temperature gradient in the second track is larger than that in the first track, the residence time of the liquid phase in the molten pool of the first track is longer than that of the second track.
Temperature profiles as a function of time for two reference points A and B.
C. Simulation analysis of molten pool under different process parameters
In order to deeply investigate the effects of various process parameters on the mesoscopic-scale temperature field, molten pool characteristic information and defects of HP-LPBF, numerical simulation experiments on mesoscopic-scale laser power, scanning speed, and hatch spacing of double-track molten pools were carried out.
1. Laser power
Figure 16 shows the effects of different laser power on the morphology and temperature field of the double-track molten pool at a scanning speed of 800 mm/s and a hatch spacing of 0.06 mm. When P = 50 W, a smaller molten pool is formed due to the lower heat generated by the Gaussian light source per unit time. This leads to a smaller track width, which results in adjacent track not lapping properly and the presence of a large number of unmelted powder particles, resulting in an increase in the number of defects, such as pores in the specimen. The surface of the track is relatively flat, and the depth is small. In addition, the temperature gradient before and after the molten pool was large, and the depression location appeared at the biased front end in Fig. 16(a). When P = 100 W, the surface of the track is flat and smooth with excellent lap. Due to the Marangoni effect, the velocity field of the molten pool is in the form of “vortex,” and the melt has good fluidity, and the maximum velocity reaches 2.15 m/s in Fig. 16(b). When P = 200 W, the heat generated by the Gaussian light source per unit time is too large, resulting in the melt rapidly reaching the evaporation temperature, generating a huge recoil pressure, forming a large molten pool, and the surface of the track is obviously raised. The melt movement is intense, especially the closed loop at the center end of the molten pool. At this time, the depth and width of the molten pool are large, leading to the expansion of the remolten region and the increased chance of the appearance of porosity defects in Fig. 16(c). The results show that at low laser power, the surface tension in the molten pool is dominant. At high laser power, recoil pressure is its main role.
Simulation results of double-track molten pool under different laser powers: (a) P = 50 W, (b) P = 100 W, (c) P = 200 W.
Table II shows the effect of different laser powers on the characteristic information of the double-track molten pool at a scanning speed of 800 mm/s and a hatch spacing of 0.06 mm. The negative overlapping ratio in the table indicates that the melt tracks are not lapped, and 26/29 indicates the melt depth of the first track/second track. It can be seen that with the increase in laser power, the melt depth, melt width, melt height, and remelted zone show a gradual increase. At the same time, the overlapping ratio also increases. Especially in the process of laser power from 50 to 200 W, the melting depth and melting width increased the most, which increased nearly 2 and 1.5 times, respectively. Meanwhile, the overlapping ratio also increases with the increase in laser power, which indicates that the melting and fusion of materials are better at high laser power. On the other hand, the dimensions of the molten pool did not change uniformly with the change of laser power. Specifically, the depth-to-width ratio of the molten pool increased from about 0.30 to 0.39 during the increase from 50 to 120 W, which further indicates that the effective heat transfer in the vertical direction is greater than that in the horizontal direction with the increase in laser power. This dimensional response to laser power is mainly affected by the recoil pressure and also by the difference in the densification degree between the powder layer and the metal substrate. In addition, according to the experimental results, the contact angle shows a tendency to increase and then decrease during the process of laser power increase, and always stays within the range of less than 33°. Therefore, in practical applications, it is necessary to select the appropriate laser power according to the specific needs in order to achieve the best processing results.
TABLE II.
Double-track molten pool characterization information at different laser powers.
Laser power (W)
Depth (μm)
Width (μm)
Height (μm)
Remolten region (μm)
Overlapping ratio (%)
Contact angle (°)
50
16
54
11
/
−10
23
100
26/29
74
14
18
23.33
33
200
37/45
116
21
52
93.33
28
2. Scanning speed
Figure 17 demonstrates the effect of different scanning speeds on the morphology and temperature field of the double-track molten pool at a laser power of 100 W and a hatch spacing of 0.06 mm. With the gradual increase in scanning speed, the surface morphology of the molten pool evolves from circular to elliptical. When � = 200 mm/s, the slow scanning speed causes the material to absorb too much heat, which is very easy to trigger the overburning phenomenon. At this point, the molten pool is larger and the surface morphology is uneven. This situation is consistent with the previously discussed scenario with high laser power in Fig. 17(a). However, when � = 1600 mm/s, the scanning speed is too fast, resulting in the material not being able to absorb sufficient heat, which triggers the powder particles that fail to melt completely to have a direct effect on the bonding of the melt to the substrate. At this time, the molten pool volume is relatively small and the neighboring melt track cannot lap properly. This result is consistent with the previously discussed case of low laser power in Fig. 17(b). Overall, the ratio of the laser power to the scanning speed (the line energy density �) has a direct effect on the temperature field and surface morphology of the molten pool.
Simulation results of double-track molten pool under different scanning speed: (a) � = 200 mm/s, (b) � = 1600 mm/s.
Table III shows the effects of different scanning speed on the characteristic information of the double-track molten pool under the condition of laser power of 100 W and hatch spacing of 0.06 mm. It can be seen that the scanning speed has a significant effect on the melt depth, melt width, melt height, remolten region, and overlapping ratio. With the increase in scanning speed, the melt depth, melt width, melt height, remelted zone, and overlapping ratio show a gradual decreasing trend. Among them, the melt depth and melt width decreased faster, while the melt height and remolten region decreased relatively slowly. In addition, when the scanning speed was increased from 200 to 800 mm/s, the decreasing speeds of melt depth and melt width were significantly accelerated, while the decreasing speeds of overlapping ratio were relatively slow. When the scanning speed was further increased to 1600 mm/s, the decreasing speeds of melt depth and melt width were further accelerated, and the un-lapped condition of the melt channel also appeared. In addition, the contact angle increases and then decreases with the scanning speed, and both are lower than 33°. Therefore, when selecting the scanning speed, it is necessary to make reasonable trade-offs according to the specific situation, and take into account the factors of melt depth, melt width, melt height, remolten region, and overlapping ratio, in order to achieve the best processing results.
TABLE III.
Double-track molten pool characterization information at different scanning speeds.
Scanning speed (mm/s)
Depth (μm)
Width (μm)
Height (μm)
Remolten region (μm)
Overlapping ratio (%)
Contact angle (°)
200
55/68
182
19/32
124
203.33
22
1600
13
50
11
/
−16.67
31
3. Hatch spacing
Figure 18 shows the effect of different hatch spacing on the morphology and temperature field of the double-track molten pool under the condition of laser power of 100 W and scanning speed of 800 mm/s. The surface morphology and temperature field of the first track and second track are basically the same, but slightly different. The first track shows a basically symmetric morphology along the scanning direction, while the second track shows a slight offset due to the difference in the heat transfer rate between the solidified material and the powder particles. When the hatch spacing is too small, the overlapping ratio increases and the probability of defects caused by remelting phenomenon grows. When the hatch spacing is too large, the neighboring melt track cannot overlap properly, and the powder particles are not completely melted, leading to an increase in the number of holes. In conclusion, the ratio of the line energy density � to the hatch spacing (the volume energy density E) has a significant effect on the temperature field and surface morphology of the molten pool.
Simulation results of double-track molten pool under different hatch spacings: (a) H = 0.03 mm, (b) H = 0.12 mm.
Table IV shows the effects of different hatch spacing on the characteristic information of the double-track molten pool under the condition of laser power of 100 W and scanning speed of 800 mm/s. It can be seen that the hatch spacing has little effect on the melt depth, melt width, and melt height, but has some effect on the remolten region. With the gradual expansion of hatch spacing, the remolten region shows a gradual decrease. At the same time, the overlapping ratio also decreased with the increase in hatch spacing. In addition, it is observed that the contact angle shows a tendency to increase and then remain stable when the hatch spacing increases, which has a more limited effect on it. Therefore, trade-offs and decisions need to be made on a case-by-case basis when selecting the hatch spacing.
TABLE IV.
Double-track molten pool characterization information at different hatch spacings.
Hatch spacing (mm)
Depth (μm)
Width (μm)
Height (μm)
Remolten region (μm)
Overlapping ratio (%)
Contact angle (°)
0.03
25/27
82
14
59
173.33
30
0.12
26
78
14
/
−35
33
In summary, the laser power, scanning speed, and hatch spacing have a significant effect on the formation of the molten pool, and the correct selection of these three process parameters is crucial to ensure the forming quality. In addition, the melt depth of the second track is slightly larger than that of the first track at higher line energy density � and volume energy density E. This is mainly due to the fact that a large amount of heat accumulation is generated after the first track, forming a larger molten pool volume, which leads to an increase in the melt depth.
D. Simulation analysis of molten pool with powder particle size and laser focal spot diameter
Figure 19 demonstrates the effect of different powder particle sizes and laser focal spot diameters on the morphology and temperature field of the double-track molten pool under a laser power of 100 W, a scanning speed of 800 mm/s, and a hatch spacing of 0.06 mm. In the process of melting coarse powder with small laser focal spot diameter, the laser energy cannot completely melt the larger powder particles, resulting in their partial melting and further generating excessive pore defects. The larger powder particles tend to generate zigzag molten pool edges, which cause an increase in the roughness of the melt track surface. In addition, the molten pool is also prone to generate the present spatter phenomenon, which can directly affect the quality of forming. The volume of the formed molten pool is relatively small, while the melt depth, melt width, and melt height are all smaller relative to the fine powder in Fig. 19(a). In the process of melting fine powders with a large laser focal spot diameter, the laser energy is able to melt the fine powder particles sufficiently, even to the point of overmelting. This results in a large number of fine spatters being generated at the edge of the molten pool, which causes porosity defects in the melt track in Fig. 19(b). In addition, the maximum velocity of the molten pool is larger for large powder particle sizes compared to small powder particle sizes, which indicates that the temperature gradient in the molten pool is larger for large powder particle sizes and the melt motion is more intense. However, the size of the laser focal spot diameter has a relatively small effect on the melt motion. However, a larger focal spot diameter induces a larger melt volume with greater depth, width, and height. In conclusion, a small powder size helps to reduce the surface roughness of the specimen, and a small laser spot diameter reduces the minimum forming size of a single track.
Simulation results of double-track molten pool with different powder particle size and laser focal spot diameter: (a) focal spot = 25 μm, coarse powder, (b) focal spot = 80 μm, fine powder.
Table V shows the maximum temperature gradient at the reference point for different powder sizes and laser focal spot diameters. As can be seen from the table, the maximum temperature gradient is lower than that of HP-LPBF for both coarse powders with a small laser spot diameter and fine powders with a large spot diameter, a phenomenon that leads to an increase in the heat transfer rate of HP-LPBF, which in turn leads to a corresponding increase in the cooling rate and, ultimately, to the formation of finer microstructures.
TABLE V.
Maximum temperature gradient at the reference point for different powder particle sizes and laser focal spot diameters.
Laser power (W)
Scanning speed (mm/s)
Hatch spacing (mm)
Average powder size (μm)
Laser focal spot diameter (μm)
Maximum temperature gradient (×107 K/s)
100
800
0.06
31.7
25
7.89
11.5
80
7.11
IV. CONCLUSIONS
In this study, the geometrical characteristics of 3D coarse and fine powder particles were first calculated using DEM and then numerical simulations of single track and double track in the process of forming SS316L from monolayer HP-LPBF at mesoscopic scale were developed using CFD method. The effects of Marangoni convection, surface tension, recoil pressure, gravity, thermal convection, thermal radiation, and evaporative heat dissipation on the heat and mass transfer in the molten pool were considered in this model. The effects of laser power, scanning speed, and hatch spacing on the dynamics of the single-track and double-track molten pools, as well as on other characteristic information, were investigated. The effects of the powder particle size on the molten pool were investigated comparatively with the laser focal spot diameter. The main conclusions are as follows:
The results show that the temperature gradient at the front of the molten pool is significantly larger than that at the tail, and the molten pool exhibits a “comet” morphology. At the top of the molten pool, there is a slightly concave region, which is the result of the coupling of Marangoni convection, recoil pressure, and surface tension. The melt flow forms two closed loops, which are mainly influenced by temperature gradients and surface tension. This special dynamic behavior of the melt tends to form an “elliptical” molten pool and an almost “mountain” shape in single-track forming.
The basic characteristics of the three-dimensional morphology and temperature field of the second track are similar to those of the first track, but there are subtle differences. The first track exhibits a basically symmetrical shape; however, due to the difference in thermal diffusion rates between the solidified metal and the powder, a slight asymmetry in the molten pool morphology of the second track occurs. After forming through the first track, there is a significant heat buildup in the powder bed, resulting in a longer dynamic time of the melt, which increases the life of the molten pool. The heights of the first track and second track remained essentially the same, but the depth of the second track was greater relative to the first track. In addition, the maximum temperature gradient was 1.69 × 108 K/s during HP-LPBF forming.
At low laser power, the surface tension in the molten pool plays a dominant role. At high laser power, recoil pressure becomes the main influencing factor. With the increase of laser power, the effective heat transfer in the vertical direction is superior to that in the horizontal direction. With the gradual increase of scanning speed, the surface morphology of the molten pool evolves from circular to elliptical. In addition, the scanning speed has a significant effect on the melt depth, melt width, melt height, remolten region, and overlapping ratio. Too large or too small hatch spacing will lead to remelting or non-lap phenomenon, which in turn causes the formation of defects.
When using a small laser focal spot diameter, it is difficult to completely melt large powder particle sizes, resulting in partial melting and excessive porosity generation. At the same time, large powder particles produce curved edges of the molten pool, resulting in increased surface roughness of the melt track. In addition, spatter occurs, which directly affects the forming quality. At small focal spot diameters, the molten pool volume is relatively small, and the melt depth, the melt width, and the melt height are correspondingly small. Taken together, the small powder particle size helps to reduce surface roughness, while the small spot diameter reduces the forming size.
REFERENCES
S. L. Sing and W. Y. Yeong , “ Laser powder bed fusion for metal additive manufacturing: Perspectives on recent developments,” Virtual Phys. Prototyping. 15, 359–370 (2020).https://doi.org/10.1080/17452759.2020.1779999 Google ScholarCrossref
A. M. Khorasani , I. G. Jithin , J. K. Veetil , and A. H. Ghasemi , “ A review of technological improvements in laser-based powder bed fusion of metal printers,” Int. J. Adv. Manuf. Technol. 108, 191–209 (2020).https://doi.org/10.1007/s00170-020-05361-3 Google ScholarCrossref
Y. Qin , A. Brockett , Y. Ma , A. Razali , J. Zhao , C. Harrison , W. Pan , X. Dai , and D. Loziak , “ Micro-manufacturing: Research, technology outcomes and development issues,” Int. J. Adv. Manuf. Technol. 47, 821–837 (2010).https://doi.org/10.1007/s00170-009-2411-2 Google ScholarCrossref
B. Nagarajan , Z. Hu , X. Song , W. Zhai , and J. Wei , “ Development of micro selective laser melting: The state of the art and future perspectives,” Engineering. 5, 702–720 (2019).https://doi.org/10.1016/j.eng.2019.07.002 Google ScholarCrossref
Y. Wei , G. Chen , W. Li , Y. Zhou , Z. Nie , J. Xu , and W. Zhou , “ Micro selective laser melting of SS316L: Single tracks, defects, microstructures and thermal/mechanical properties,” Opt. Laser Technol. 145, 107469 (2022).https://doi.org/10.1016/j.optlastec.2021.107469 Google ScholarCrossref
Y. Wei , G. Chen , W. Li , M. Li , Y. Zhou , Z. Nie , and J. Xu , “ Process optimization of micro selective laser melting and comparison of different laser diameter for forming different powder,” Opt. Laser Technol. 150, 107953 (2022).https://doi.org/10.1016/j.optlastec.2022.107953 Google ScholarCrossref
H. Zhiheng , B. Nagarajan , X. Song , R. Huang , W. Zhai , and J. Wei , “ Formation of SS316L single tracks in micro selective laser melting: Surface, geometry, and defects,” Adv. Mater. Sci. Eng. 2019, 9451406.https://doi.org/10.1155/2019/9451406 Crossref
B. Nagarajan , Z. Hu , S. Gao , X. Song , R. Huang , M. Seita , and J. Wei , “ Effect of in-situ laser remelting on the microstructure of SS316L fabricated by micro selective laser melting,” in Advanced Surface Enhancement, edited by Sho Itoh and Shashwat Shukla , Lecture Notes in Mechanical Engineering ( Springer Singapore, Singapore, 2020), pp. 330–336. Google ScholarCrossref
H. Zhiheng , B. Nagarajan , X. Song , R. Huang , W. Zhai , and J. Wei , “ Tailoring surface roughness of micro selective laser melted SS316L by in-situ laser remelting,” in Advanced Surface Enhancement, edited by Sho Itoh and Shashwat Shukla , Lecture Notes in Mechanical Engineering ( Springer Singapore, Singapore, 2020), pp. 337–343. Google Scholar
J. Fu , Z. Hu , X. Song , W. Zhai , Y. Long , H. Li , and M. Fu , “ Micro selective laser melting of NiTi shape memory alloy: Defects, microstructures and thermal/mechanical properties,” Opt. Laser Technol. 131, 106374 (2020).https://doi.org/10.1016/j.optlastec.2020.106374 Google ScholarCrossref
E. Abele and M. Kniepkamp , “ Analysis and optimisation of vertical surface roughness in micro selective laser melting,” Surf. Topogr.: Metrol. Prop. 3, 034007 (2015).https://doi.org/10.1088/2051-672X/3/3/034007 Google ScholarCrossref
S. Qu , J. Ding , J. Fu , M. Fu , B. Zhang , and X. Song , “ High-precision laser powder bed fusion processing of pure copper,” Addit. Manuf. 48, 102417 (2021).https://doi.org/10.1016/j.addma.2021.102417 Google ScholarCrossref
Y. Wei , G. Chen , M. Li , W. Li , Y. Zhou , J. Xu , and Z. wei , “ High-precision laser powder bed fusion of 18Ni300 maraging steel and its SiC reinforcement composite materials,” J. Manuf. Process. 84, 750–763 (2022).https://doi.org/10.1016/j.jmapro.2022.10.049 Google ScholarCrossref
B. Liu , R. Wildman , T. Christopher , I. Ashcroft , and H. Richard , “ Investigation the effect of particle size distribution on processing parameters optimisation in selective laser melting process,” in 2011 International Solid Freeform Fabrication Symposium ( University of Texas at Austin, 2011). Google Scholar
T. D. McLouth , G. E. Bean , D. B. Witkin , S. D. Sitzman , P. M. Adams , D. N. Patel , W. Park , J.-M. Yang , and R. J. Zaldivar , “ The effect of laser focus shift on microstructural variation of Inconel 718 produced by selective laser melting,” Mater. Des. 149, 205–213 (2018).https://doi.org/10.1016/j.matdes.2018.04.019 Google ScholarCrossref
Y. Qian , Y. Wentao , and L. Feng , “ Mesoscopic simulations of powder bed fusion: Research progresses and conditions,” Electromachining Mould 06, 46–52 (2017).https://doi.org/10.3969/j.issn.1009-279X.2017.06.012 Google Scholar
J. Fu , S. Qu , J. Ding , X. Song , and M. W. Fu , “ Comparison of the microstructure, mechanical properties and distortion of stainless Steel 316L fabricated by micro and conventional laser powder bed fusion,” Addit. Manuf. 44, 102067 (2021).https://doi.org/10.1016/j.addma.2021.102067 Google ScholarCrossref
N. T. Aboulkhair , I. Maskery , C. Tuck , I. Ashcroft , and N. M. Everitt , “ The microstructure and mechanical properties of selectively laser Melted AlSi10Mg: The effect of a conventional T6-like heat treatment,” Mater. Sci. Eng. A 667, 139–146 (2016).https://doi.org/10.1016/j.msea.2016.04.092 Google ScholarCrossref
S. Y. Chen , J. C. Huang , C. T. Pan , C. H. Lin , T. L. Yang , Y. S. Huang , C. H. Ou , L. Y. Chen , D. Y. Lin , H. K. Lin , T. H. Li , J. S. C. Jang , and C. C. Yang , “ Microstructure and mechanical properties of open-cell porous Ti-6Al-4V fabricated by selective laser melting,” J. Alloys Compd. 713, 248–254 (2017).https://doi.org/10.1016/j.jallcom.2017.04.190 Google ScholarCrossref
Y. Bai , Y. Yang , D. Wang , and M. Zhang , “ Influence mechanism of parameters process and mechanical properties evolution mechanism of Maraging steel 300 by selective laser melting,” Mater. Sci. Eng. A 703, 116–123 (2017).https://doi.org/10.1016/j.msea.2017.06.033 Google ScholarCrossref
Y. Bai , Y. Yang , Z. Xiao , M. Zhang , and D. Wang , “ Process optimization and mechanical property evolution of AlSiMg0.75 by selective laser melting,” Mater. Des. 140, 257–266 (2018).https://doi.org/10.1016/j.matdes.2017.11.045 Google ScholarCrossref
Y. Liu , M. Zhang , W. Shi , Y. Ma , and J. Yang , “ Study on performance optimization of 316L stainless steel parts by high-efficiency selective laser melting,” Opt. Laser Technol. 138, 106872 (2021).https://doi.org/10.1016/j.optlastec.2020.106872 Google ScholarCrossref
D. Gu , Y.-C. Hagedorn , W. Meiners , G. Meng , R. J. S. Batista , K. Wissenbach , and R. Poprawe , “ Densification behavior, microstructure evolution, and wear performance of selective laser melting processed commercially pure titanium,” Acta Mater. 60, 3849–3860 (2012).https://doi.org/10.1016/j.actamat.2012.04.006 Google ScholarCrossref
N. Read , W. Wang , K. Essa , and M. M. Attallah , “ Selective laser melting of AlSi10Mg alloy: Process optimisation and mechanical properties development,” Mater. Des. 65, 417–424 (2015).https://doi.org/10.1016/j.matdes.2014.09.044 Google ScholarCrossref
I. A. Roberts , C. J. Wang , R. Esterlein , M. Stanford , and D. J. Mynors , “ A three-dimensional finite element analysis of the temperature field during laser melting of metal powders in additive layer manufacturing,” Int. J. Mach. Tools Manuf. 49(12–13), 916–923 (2009).https://doi.org/10.1016/j.ijmachtools.2009.07.004 Google ScholarCrossref
K. Dai and L. Shaw , “ Finite element analysis of the effect of volume shrinkage during laser densification,” Acta Mater. 53(18), 4743–4754 (2005).https://doi.org/10.1016/j.actamat.2005.06.014 Google ScholarCrossref
K. Carolin , E. Attar , and P. Heinl , “ Mesoscopic simulation of selective beam melting processes,” J. Mater. Process. Technol. 211(6), 978–987 (2011).https://doi.org/10.1016/j.jmatprotec.2010.12.016 Google ScholarCrossref
F.-J. Gürtler , M. Karg , K.-H. Leitz , and M. Schmidt , “ Simulation of laser beam melting of steel powders using the three-dimensional volume of fluid method,” Phys. Procedia 41, 881–886 (2013).https://doi.org/10.1016/j.phpro.2013.03.162 Google ScholarCrossref
P. Meakin and R. Jullien , “ Restructuring effects in the rain model for random deposition,” J. Phys. France 48(10), 1651–1662 (1987).https://doi.org/10.1051/jphys:0198700480100165100 Google ScholarCrossref
J-m Wang , G-h Liu , Y-l Fang , and W-k Li , “ Marangoni effect in nonequilibrium multiphase system of material processing,” Rev. Chem. Eng. 32(5), 551–585 (2016).https://doi.org/10.1515/revce-2015-0067 Google ScholarCrossref
W. Ye , S. Zhang , L. L. Mendez , M. Farias , J. Li , B. Xu , P. Li , and Y. Zhang , “ Numerical simulation of the melting and alloying processes of elemental titanium and boron powders using selective laser alloying,” J. Manuf. Process. 64, 1235–1247 (2021).https://doi.org/10.1016/j.jmapro.2021.02.044 Google ScholarCrossref
U. S. Bertoli , A. J. Wolfer , M. J. Matthews , J.-P. R. Delplanque , and J. M. Schoenung , “ On the limitations of volumetric energy density as a design parameter for selective laser melting,” Mater. Des. 113, 331–340 (2017).https://doi.org/10.1016/j.matdes.2016.10.037 Google ScholarCrossref
W. E. King , H. D. Barth , V. M. Castillo , G. F. Gallegos , J. W. Gibbs , D. E. Hahn , C. Kamath , and A. M. Rubenchik , “ Observation of keyhole-mode laser melting in laser powder-bed fusion additive manufacturing,” J. Mater. Process. Technol. 214(12), 2915–2925 (2014).https://doi.org/10.1016/j.jmatprotec.2014.06.005 Google ScholarCrossref
L. Cao , “ Numerical simulation of the impact of laying powder on selective laser melting single-pass formation,” Int. J. Heat Mass Transfer 141, 1036–1048 (2019).https://doi.org/10.1016/j.ijheatmasstransfer.2019.07.053 Google ScholarCrossref
L. Huang , X. Hua , D. Wu , and F. Li , “ Numerical study of keyhole instability and porosity formation mechanism in laser welding of aluminum alloy and steel,” J. Mater. Process. Technol. 252, 421–431 (2018).https://doi.org/10.1016/j.jmatprotec.2017.10.011 Google ScholarCrossref
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. 145, 105992 (2019).https://doi.org/10.1016/j.ijthermalsci.2019.105992 Google ScholarCrossref
J.-H. Cho and S.-J. Na , “ Theoretical analysis of keyhole dynamics in polarized laser drilling,” J. Phys. D: Appl. Phys. 40(24), 7638 (2007).https://doi.org/10.1088/0022-3727/40/24/007 Google ScholarCrossref
W. Ye , “ Mechanism analysis of selective laser melting and metallurgy process based on base element powder of titanium and boron,” Ph.D. dissertation ( Nanchang University, 2021). Google Scholar
R. Ammer , M. Markl , U. Ljungblad , C. Körner , and U. Rüde , “ Simulating fast electron beam melting with a parallel thermal free surface lattice Boltzmann method,” Comput. Math. Appl. 67(2), 318–330 (2014).https://doi.org/10.1016/j.camwa.2013.10.001 Google ScholarCrossref
H. Chen , Q. Wei , S. Wen , Z. Li , and Y. Shi , “ Flow behavior of powder particles in layering process of selective laser melting: Numerical modeling and experimental verification based on discrete element method,” Int. J. Mach. Tools Manuf. 123, 146–159 (2017).https://doi.org/10.1016/j.ijmachtools.2017.08.004 Google ScholarCrossref
F. Verhaeghe , T. Craeghs , J. Heulens , and L. Pandelaers , “ A pragmatic model for selective laser melting with evaporation,” Acta Mater. 57(20), 6006–6012 (2009).https://doi.org/10.1016/j.actamat.2009.08.027 Google ScholarCrossref
C. H. Fu and Y. B. Guo , “ Three-dimensional temperature gradient mechanism in selective laser melting of Ti-6Al-4V,” J. Manuf. Sci. Eng. 136(6), 061004 (2014).https://doi.org/10.1115/1.4028539 Google ScholarCrossref
Y. Xiang , Z. Shuzhe , L. Junfeng , W. Zhengying , Y. Lixiang , and J. Lihao , “ Numerical simulation and experimental verification for selective laser single track melting forming of Ti6Al4V,” J. Zhejiang Univ. (Eng. Sci.) 53(11), 2102–2109 + 2117 (2019).https://doi.org/10.3785/j.issn.1008-973X.2019.11.007 Google Scholar
Q. He , H. Xia , J. Liu , X. Ao , and S. Lin , “ Modeling and numerical studies of selective laser melting: Multiphase flow, solidification and heat transfer,” Mater. Des. 196, 109115 (2020).https://doi.org/10.1016/j.matdes.2020.109115 Google ScholarCrossref
L. Cao , “ Mesoscopic-scale numerical simulation including the influence of process parameters on SLM single-layer multi-pass formation,” Metall. Mater. Trans. A 51, 4130–4145 (2020).https://doi.org/10.1007/s11661-020-05831-z Google ScholarCrossref
L. Cao , “ Mesoscopic-scale numerical investigation including the influence of process parameters on LPBF multi-layer multi-path formation,” Comput. Model. Eng. Sci. 126(1), 5–23 (2021).https://doi.org/10.32604/cmes.2021.014693 Google ScholarCrossref
H. Yin and S. D. Felicelli , “ Dendrite growth simulation during solidification in the LENS process,” Acta Mater. 58(4), 1455–1465 (2010).https://doi.org/10.1016/j.actamat.2009.10.053 Google ScholarCrossref
P. Nie , O. A. Ojo , and Z. Li , “ Numerical modeling of microstructure evolution during laser additive manufacturing of a nickel-based superalloy,” Acta Mater. 77, 85–95 (2014).https://doi.org/10.1016/j.actamat.2014.05.039 Google ScholarCrossref
Z. Liu and H. Qi , “ Effects of substrate crystallographic orientations on crystal growth and microstructure formation in laser powder deposition of nickel-based superalloy,” Acta Mater. 87, 248–258 (2015).https://doi.org/10.1016/j.actamat.2014.12.046 Google ScholarCrossref
L. Wei , L. Xin , W. Meng , and H. Weidong , “ Cellular automaton simulation of the molten pool of laser solid forming process,” Acta Phys. Sin. 64(01), 018103–018363 (2015).https://doi.org/10.7498/aps.64.018103 Google ScholarCrossref
R. Acharya , J. A. Sharon , and A. Staroselsky , “ Prediction of microstructure in laser powder bed fusion process,” Acta Mater. 124, 360–371 (2017).https://doi.org/10.1016/j.actamat.2016.11.018 Google ScholarCrossref
M. R. Rolchigo and R. LeSar , “ Modeling of binary alloy solidification under conditions representative of additive manufacturing,” Comput. Mater. Sci. 150, 535–545 (2018).https://doi.org/10.1016/j.commatsci.2018.04.004 Google ScholarCrossref
S. Geng , P. Jiang , L. Guo , X. Gao , and G. Mi , “ Multi-scale simulation of grain/sub-grain structure evolution during solidification in laser welding of aluminum alloys,” Int. J. Heat Mass Transfer 149, 119252 (2020).https://doi.org/10.1016/j.ijheatmasstransfer.2019.119252 Google ScholarCrossref
W. L. Wang , W. Q. Liu , X. Yang , R. R. Xu , and Q. Y. Dai , “ Multi-scale simulation of columnar-to-equiaxed transition during laser selective melting of rare earth magnesium alloy,” J. Mater. Sci. Technol. 119, 11–24 (2022).https://doi.org/10.1016/j.jmst.2021.12.029 Google ScholarCrossref
Q. Xia , J. Yang , and Y. Li , “ On the conservative phase-field method with the N-component incompressible flows,” Phys. Fluids 35, 012120 (2023).https://doi.org/10.1063/5.0135490 Google ScholarCrossref
Q. Xia , G. Sun , J. Kim , and Y. Li , “ Multi-scale modeling and simulation of additive manufacturing based on fused deposition technique,” Phys. Fluids 35, 034116 (2023).https://doi.org/10.1063/5.0141316 Google ScholarCrossref
A. Hussein , L. Hao , C. Yan , and R. Everson , “ Finite element simulation of the temperature and stress fields in single layers built without-support in selective laser melting,” Mater. Des. 52, 638–647 (2013).https://doi.org/10.1016/j.matdes.2013.05.070 Google ScholarCrossref
J. Ding , P. Colegrove , J. Mehnen , S. Ganguly , P. M. Sequeira Almeida , F. Wang , and S. Williams , “ Thermo-mechanical analysis of wire and arc additive layer manufacturing process on large multi-layer parts,” Comput. Mater. Sci. 50(12), 3315–3322 (2011).https://doi.org/10.1016/j.commatsci.2011.06.023 Google ScholarCrossref
Y. Du , X. You , F. Qiao , L. Guo , and Z. Liu , “ A model for predicting the temperature field during selective laser melting,” Results Phys. 12, 52–60 (2019).https://doi.org/10.1016/j.rinp.2018.11.031 Google ScholarCrossref
X. Luo , M. Liu , L. Zhenhua , H. Li , and J. Shen , “ Effect of different heat-source models on calculated temperature field of selective laser melted 18Ni300,” Chin. J. Lasers 48(14), 1402005–1402062 (2021).https://doi.org/10.3788/CJL202148.1402005 Google ScholarCrossref
J. F. Li , L. Li , and F. H. Stott , “ Thermal stresses and their implication on cracking during laser melting of ceramic materials,” Acta Mater. 52(14), 4385–4398 (2004).https://doi.org/10.1016/j.actamat.2004.06.005 Google ScholarCrossref
P. Aggarangsi and J. L. Beuth , “ Localized preheating approaches for reducing residual stress in additive manufacturing,” paper presented at the 2006 International Solid Freeform Fabrication Symposium, The University of Texas in Austin on August 14–16, 2006.
K. Dai and L. Shaw , “ Thermal and mechanical finite element modeling of laser forming from metal and ceramic powders,” Acta Mater. 52(1), 69–80 (2004).https://doi.org/10.1016/j.actamat.2003.08.028 Google ScholarCrossref
A. H. Nickel , D. M. Barnett , and F. B. Prinz , “ Thermal stresses and deposition patterns in layered manufacturing,” Mater. Sci. Eng. A 317(1–2), 59–64 (2001).https://doi.org/10.1016/S0921-5093(01)01179-0 Google ScholarCrossref
M. F. Zaeh and G. Branner , “ Investigations on residual stresses and deformations in selective laser melting,” Prod. Eng. 4(1), 35–45 (2010).https://doi.org/10.1007/s11740-009-0192-y Google ScholarCrossref
P. Bian , J. Shi , Y. Liu , and Y. Xie , “ Influence of laser power and scanning strategy on residual stress distribution in additively manufactured 316L steel,” Opt. Laser Technol. 132, 106477 (2020).https://doi.org/10.1016/j.optlastec.2020.106477 Google ScholarCrossref
B. M. Marques , C. M. Andrade , D. M. Neto , M. C. Oliveira , J. L. Alves , and L. F. Menezes , “ Numerical analysis of residual stresses in parts produced by selective laser melting process,” Procedia Manuf. 47, 1170–1177 (2020).https://doi.org/10.1016/j.promfg.2020.04.167 Google ScholarCrossref
W. Mu , “ Numerical simulation of SLM forming process and research and prediction of forming properties,” MA thesis ( Anhui Jianzhu University, 2022). Google Scholar
Y. Zhang , “ Multi-scale multi-physics modeling of laser powder bed fusion process of metallic materials with experiment validation,” Ph.D. dissertation ( Purdue University, 2018). Google Scholar
Y. Qian , “ Mesoscopic simulation studies of key processing issues for powder bed fusion technology,” Ph.D. dissertation ( Tsinghua University, 2019). Google Scholar
N. V. Brilliantov , S. Frank , J.-M. Hertzsch , and T. Pöschel , “ Model for collisions in granular gases,” Phys. Rev. E 53(5), 5382–5392 (1996).https://doi.org/10.1103/PhysRevE.53.5382 Google ScholarCrossref
Z. Xiao , “ Research on microscale selective laser melting process of high strength pure copper specimens,” MA thesis ( Hunan University, 2022). Google Scholar
Z. Li , K. Mukai , M. Zeze , and K. C. Mills , “ Determination of the surface tension of liquid stainless steel,” J. Mater. Sci. 40(9–10), 2191–2195 (2005).https://doi.org/10.1007/s10853-005-1931-x Google ScholarCrossref
R. Scardovelli and S. Zaleski , “ Analytical relations connecting linear interfaces and volume fractions in rectangular grids,” J. Comput. Phys. 164(1), 228–237 (2000).https://doi.org/10.1006/jcph.2000.6567 Google ScholarCrossref
D.-W. Cho , W.-I. Cho , and S.-J. Na , “ Modeling and simulation of arc: Laser and hybrid welding process,” J. Manuf. Process. 16(1), 26–55 (2014).https://doi.org/10.1016/j.jmapro.2013.06.012 Google ScholarCrossref 76.Flow3D. Version 11.1.0: User Manual ( FlowScience, Santa Fe, NM, USA, 2015).
Y. Tian , L. Yang , D. Zhao , Y. Huang , and J. Pan , “ Numerical analysis of powder bed generation and single track forming for selective laser melting of ss316l stainless steel,” J. Manuf. Process. 58, 964–974 (2020).https://doi.org/10.1016/j.jmapro.2020.09.002 Google ScholarCrossref
C. Tang , K. Q. Le , and C. H. Wong , “ Physics of humping formation in laser powder bed fusion,” Int. J. Heat Mass Transfer 149, 119172 (2020).https://doi.org/10.1016/j.ijheatmasstransfer.2019.119172 Google ScholarCrossref
L. Cao , “ Mesoscopic-scale simulation of pore evolution during laser powder bed fusion process,” Comput. Mater. Sci. 179, 109686 (2020).https://doi.org/10.1016/j.commatsci.2020.109686 Google ScholarCrossref
R. Li , J. Liu , Y. Shi , W. Li , and W. Jiang , “ Balling behavior of stainless steel and nickel powder during selective laser melting process,” Int. J. Adv. Manuf. Technol. 59(9–12), 1025–1035 (2012).https://doi.org/10.1007/s00170-011-3566-1 Google ScholarCrossref
S. A. Khairallah and A. Anderson , “ Mesoscopic simulation model of selective laser melting of stainless steel powder,” J. Mater. Process. Technol. 214(11), 2627–2636 (2014).https://doi.org/10.1016/j.jmatprotec.2014.06.001 Google ScholarCrossref
J. Liu , D. Gu , H. Chen , D. Dai , and H. Zhang , “ Influence of substrate surface morphology on wetting behavior of tracks during selective laser melting of aluminum-based alloys,” J. Zhejiang Univ. Sci. A 19(2), 111–121 (2018).https://doi.org/10.1631/jzus.A1700599 Google ScholarCrossref
L. Li , J. Li , and T. Fan , “ Phase-field modeling of wetting and balling dynamics in powder bed fusion process,” Phys. Fluids 33, 042116 (2021).https://doi.org/10.1063/5.0046771 Google ScholarCrossref
X. Nie , Z. Hu , H. Zhu , Z. Hu , L. Ke , and X. Zeng , “ Analysis of processing parameters and characteristics of selective laser melted high strength Al-Cu-Mg alloys: from single tracks to cubic samples,” J. Mater. Process. Technol. 256, 69–77 (2018).https://doi.org/10.1016/j.jmatprotec.2018.01.030 Google ScholarCrossref
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.
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.
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.
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%
Ni
Cr
Co
Al
Mo
W
Ti
Nb
C
B
Zr
Ta
Others
Bulk
59.12
17.47
8.48
7.00
1.01
0.81
3.96
0.49
0.47
0.05
0.09
0.56
0.46
γ matrix
Top
50.48
32.91
11.59
1.94
1.39
0.82
0.44
0.8
0.03
0.03
0.02
–
0.24
Mid
50.37
32.61
11.93
1.79
1.54
0.89
0.44
0.1
0.03
0.02
0.02
0.01
0.23
Bot
48.10
34.57
12.08
2.14
1.43
0.88
0.48
0.08
0.04
0.03
0.01
–
0.12
Primary γ′
Top
72.17
2.51
3.44
12.71
0.25
0.39
7.78
0.56
–
0.03
0.02
0.05
0.08
Mid
71.60
2.57
3.28
13.55
0.42
0.68
7.04
0.73
–
0.01
0.03
0.04
0.04
Bot
72.34
2.47
3.86
12.50
0.26
0.44
7.46
0.50
0.05
0.02
0.02
0.03
0.04
Secondary γ′
Mid
70.42
4.20
3.23
14.19
0.63
1.03
5.34
0.79
0.03
–
0.04
0.04
0.05
Bot
69.91
4.06
3.68
14.32
0.81
1.04
5.22
0.65
0.05
–
0.10
0.02
0.11
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. 1, Fig. 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.
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.
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 sites. Table 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.
γ′ = 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].
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.
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.
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).
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. 9, Fig. 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.
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.
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′.
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. 9, Fig. 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 conductivity, specific 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.
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.
[1]T. Debroy, H.L. Wei, J.S. Zuback, T. Mukherjee, J.W. Elmer, J.O. Milewski, A.M. Beese, A. Wilson-heid, A. De, W. ZhangAdditive manufacturing of metallic components – process, structure and propertiesProg. Mater. Sci., 92 (2018), pp. 112-224, 10.1016/j.pmatsci.2017.10.001View PDFView articleView in ScopusGoogle Scholar
[4]N.K. Adomako, J.J. Lewandowski, B.M. Arkhurst, H. Choi, H.J. Chang, J.H. KimMicrostructures and mechanical properties of multi-layered materials composed of Ti-6Al-4V, vanadium, and 17–4PH stainless steel produced by directed energy depositionAddit. Manuf., 59 (2022), Article 103174, 10.1016/j.addma.2022.103174View PDFView articleView in ScopusGoogle Scholar
[5]H. Wang, Z.G. Zhu, H. Chen, A.G. Wang, J.Q. Liu, H.W. Liu, R.K. Zheng, S.M.L. Nai, S. Primig, S.S. Babu, S.P. Ringer, X.Z. LiaoEffect of cyclic rapid thermal loadings on the microstructural evolution of a CrMnFeCoNi high-entropy alloy manufactured by selective laser meltingActa Mater., 196 (2020), pp. 609-625, 10.1016/J.ACTAMAT.2020.07.006View PDFView articleView in ScopusGoogle Scholar
[10]S.S. Al-Bermani, M.L. Blackmore, W. Zhang, I. ToddThe origin of microstructural diversity, texture, and mechanical properties in electron beam melted Ti-6Al-4VMetall. Mater. Trans. A Phys. Metall. Mater. Sci., 41 (2010), pp. 3422-3434, 10.1007/s11661-010-0397-xView article View in ScopusGoogle Scholar
[13]H. Helmer, A. Bauereiß, R.F. Singer, C. KörnerErratum to: ‘Grain structure evolution in Inconel 718 during selective electron beam melting’ (Materials Science & Engineering A (2016) 668 (180–187 (S0921509316305536) (10.1016/j.msea.2016.05.046))Mater. Sci. Eng. A., 676 (2016), p. 546, 10.1016/j.msea.2016.09.016View PDFView articleView in ScopusGoogle Scholar
[16]N. Haghdadi, E. Whitelock, B. Lim, H. Chen, X. Liao, S.S. Babu, S.P. Ringer, S. PrimigMultimodal γ′ precipitation in Inconel-738 Ni-based superalloy during electron-beam powder bed fusion additive manufacturingJ. Mater. Sci., 55 (2020), pp. 13342-13350, 10.1007/s10853-020-04915-wView article View in ScopusGoogle Scholar
[17]B. Lim, H. Chen, Z. Chen, N. Haghdadi, X. Liao, S. Primig, S.S. Babu, A. Breen, S.P. RingerMicrostructure–property gradients in Ni-based superalloy (Inconel 738) additively manufactured via electron beam powder bed fusionAddit. Manuf. (2021), Article 102121, 10.1016/j.addma.2021.102121View PDFView articleView in ScopusGoogle Scholar
[18]P. Karimi, E. Sadeghi, P. Åkerfeldt, J. Ålgårdh, J. AnderssonInfluence of successive thermal cycling on microstructure evolution of EBM-manufactured alloy 718 in track-by-track and layer-by-layer designMater. Des., 160 (2018), pp. 427-441, 10.1016/j.matdes.2018.09.038View PDFView articleView in ScopusGoogle Scholar
[19]E. Chauvet, P. Kontis, E.A. Jägle, B. Gault, D. Raabe, C. Tassin, J.J. Blandin, R. Dendievel, B. Vayre, S. Abed, G. MartinHot cracking mechanism affecting a non-weldable Ni-based superalloy produced by selective electron Beam MeltingActa Mater., 142 (2018), pp. 82-94, 10.1016/j.actamat.2017.09.047View PDFView articleView in ScopusGoogle Scholar
[20]M. Ramsperger, R.F. Singer, C. KörnerMicrostructure of the nickel-base superalloy CMSX-4 fabricated by selective electron beam meltingMetall. Mater. Trans. A Phys. Metall. Mater. Sci., 47 (2016), pp. 1469-1480, 10.1007/s11661-015-3300-y View PDF This article is free to access.View in ScopusGoogle Scholar
[21]B. Zhang, P. Wang, Y. Chew, Y. Wen, M. Zhang, P. Wang, G. Bi, J. WeiMechanical properties and microstructure evolution of selective laser melting Inconel 718 along building direction and sectional dimensionMater. Sci. Eng. A, 794 (2020), Article 139941, 10.1016/j.msea.2020.139941View PDFView articleView in ScopusGoogle Scholar
[22]C. Körner, M. Ramsperger, C. Meid, D. Bürger, P. Wollgramm, M. Bartsch, G. EggelerMicrostructure and mechanical properties of CMSX-4 single crystals prepared by additive manufacturingMetall. Mater. Trans. A Phys. Metall. Mater. Sci., 49 (2018), pp. 3781-3792, 10.1007/s11661-018-4762-5 View PDF This article is free to access.View in ScopusGoogle Scholar
[23]B. Lim, H. Chen, K. Nomoto, Z. Chen, A.I. Saville, S. Vogel, A.J. Clarke, A. Paradowska, M. Reid, S. Primig, X. Liao, S.S. Babu, A.J. Breen, S.P. RingerAdditively manufactured Haynes-282 monoliths containing thin wall struts of varying thicknessesAddit. Manuf., 59 (2022), Article 103120, 10.1016/j.addma.2022.103120View PDFView articleView in ScopusGoogle Scholar
[24]C.L.A. Leung, S. Marussi, R.C. Atwood, M. Towrie, P.J. Withers, P.D. LeeIn situ X-ray imaging of defect and molten pool dynamics in laser additive manufacturingNat. Commun., 9 (2018), pp. 1-9, 10.1038/s41467-018-03734-7 View PDF This article is free to access.View in ScopusGoogle Scholar
[25]C. Zhao, K. Fezzaa, R.W. Cunningham, H. Wen, F. De Carlo, L. Chen, A.D. Rollett, T. SunReal-time monitoring of laser powder bed fusion process using high-speed X-ray imaging and diffractionSci. Rep., 7 (2017), pp. 1-11, 10.1038/s41598-017-03761-2 View PDF This article is free to access.View in ScopusGoogle Scholar
[26]C. Kenel, D. Grolimund, X. Li, E. Panepucci, V.A. Samson, D.F. Sanchez, F. Marone, C. LeinenbachIn situ investigation of phase transformations in Ti-6Al-4V under additive manufacturing conditions combining laser melting and high-speed micro-X-ray diffractionSci. Rep., 7 (2017), pp. 1-10, 10.1038/s41598-017-16760-0 View PDF This article is free to access.Google Scholar
[27]W.L. Bevilaqua, J. Epp, H. Meyer, J. Dong, H. Roelofs, A. da, S. Rocha, A. RegulyRevealing the dynamic transformation of austenite to bainite during uniaxial warm compression through in-situ synchrotron X-ray diffractionMetals, 11 (2021), pp. 1-14, 10.3390/met11030467View article View in ScopusGoogle Scholar
[28]B. Wahlmann, E. Krohmer, C. Breuning, N. Schell, P. Staron, E. Uhlmann, C. KörnerIn situ observation of γ′ phase transformation dynamics during selective laser melting of CMSX-4Adv. Eng. Mater., 23 (2021), 10.1002/adem.202100112 View PDF This article is free to access.Google Scholar
[35]M.J. Anderson, J. Benson, J.W. Brooks, B. Saunders, H.C. BasoaltoPredicting precipitation kinetics during the annealing of additive manufactured Inconel 625 componentsIntegr. Mater. Manuf. Innov., 8 (2019), pp. 154-166, 10.1007/S40192-019-00134-7/FIGURES/11 View PDFThis article is free to access.View in ScopusGoogle Scholar
[36]H.C. Basoalto, C. Panwisawas, Y. Sovani, M.J. Anderson, R.P. Turner, B. Saunders, J.W. BrooksA computational study on the three-dimensional printability of precipitate-strengthened nickel-based superalloysProc. R. Soc. A, 474 (2018), 10.1098/RSPA.2018.0295View article Google Scholar
[37]K. McNamara, Y. Ji, F. Lia, P. Promoppatum, S.C. Yao, H. Zhou, Y. Wang, L.Q. Chen, R.P. MartukanitzPredicting phase transformation kinetics during metal additive manufacturing using non-isothermal Johnson-Mehl-Avrami models: application to Inconel 718 and Ti-6Al-4VAddit. Manuf., 49 (2022), Article 102478, 10.1016/J.ADDMA.2021.102478View PDFView articleView in ScopusGoogle Scholar
[39]A. Drexler, B. Oberwinkler, S. Primig, C. Turk, E. Povoden-karadeniz, A. Heinemann, W. Ecker, M. StockingerMaterials Science & Engineering A Experimental and numerical investigations of the γ ″ and γ ′ precipitation kinetics in Alloy 718Mater. Sci. Eng. A., 723 (2018), pp. 314-323, 10.1016/j.msea.2018.03.013View PDFView articleView in ScopusGoogle Scholar
[44]A.V. Sotov, A.V. Agapovichev, V.G. Smelov, V.V. Kokareva, M.O. Dmitrieva, A.A. Melnikov, S.P. Golanov, Y.M. AnurovInvestigation of the IN-738 superalloy microstructure and mechanical properties for the manufacturing of gas turbine engine nozzle guide vane by selective laser meltingInt. J. Adv. Manuf. Technol., 107 (2020), pp. 2525-2535, 10.1007/s00170-020-05197-xView article View in ScopusGoogle Scholar
[49]S. Sanchez, P. Smith, Z. Xu, G. Gaspard, C.J. Hyde, W.W. Wits, I.A. Ashcroft, H. Chen, A.T. ClarePowder Bed Fusion of nickel-based superalloys: a reviewInt. J. Mach. Tools Manuf., 165 (2021), 10.1016/j.ijmachtools.2021.103729 View PDF This article is free to access.Google Scholar
[50]C.L.A. Leung, R. Tosi, E. Muzangaza, S. Nonni, P.J. Withers, P.D. LeeEffect of preheating on the thermal, microstructural and mechanical properties of selective electron beam melted Ti-6Al-4V componentsMater. Des., 174 (2019), Article 107792, 10.1016/j.matdes.2019.107792View PDFView articleView in ScopusGoogle Scholar
[51]S. Griffiths, H. Ghasemi Tabasi, T. Ivas, X. Maeder, A. De Luca, K. Zweiacker, R. Wróbel, J. Jhabvala, R.E. Logé, C. LeinenbachCombining alloy and process modification for micro-crack mitigation in an additively manufactured Ni-base superalloyAddit. Manuf., 36 (2020), 10.1016/j.addma.2020.101443View article Google Scholar
[52]P. Soille, L. Vincent Pierre Soille, L.M. Vincent, Determining watersheds in digital pictures via flooding simulations, Https://Doi.Org/10.1117/12.24211. 1360 (1990) 240–250. https://doi.org/10.1117/12.24211.Google Scholar
[53]ASTM Standard Test Method for Microindentation Hardness of MaterialsKnoop and Vickers Hardness of Materials 1, Annu. B ASTM Stand. i 2010 1 42.Google Scholar
[56]B. Sonderegger, E. KozeschnikGeneralized nearest-neighbor broken-bond analysis of randomly oriented coherent interfaces in multicomponent Fcc and Bcc structuresMetall. Mater. Trans. A Phys. Metall. Mater. Sci., 40 (2009), pp. 499-510, 10.1007/S11661-008-9752-6/FIGURES/8View articleView in ScopusGoogle Scholar
[58]B. Sonderegger, E. KozeschnikInterfacial energy of diffuse phase boundaries in the generalized broken-bond approachMetall. Mater. Trans. A Phys. Metall. Mater. Sci., 41 (2010), pp. 3262-3269, 10.1007/S11661-010-0370-8/FIGURES/4View articleView in ScopusGoogle Scholar
[67]P. Strunz, M. Petrenec, J. Polák, U. Gasser, G. FarkasFormation and dissolution of’ precipitates in IN792 superalloy at elevated temperaturesMetals, 6 (2016), 10.3390/met6020037View article Google Scholar
[68]A.R.P. Singh, S. Nag, J.Y. Hwang, G.B. Viswanathan, J. Tiley, R. Srinivasan, H.L. Fraser, R. BanerjeeInfluence of cooling rate on the development of multiple generations of γ′ precipitates in a commercial nickel base superalloyMater. Charact., 62 (2011), pp. 878-886, 10.1016/j.matchar.2011.06.002View PDFView articleView in ScopusGoogle Scholar
[69]E. Balikci, A. Raman, R. MirshamsMicrostructure evolution in polycrystalline IN738LC in the range 1120 to 1250C, Zeitschrift FuerMet, 90 (1999), pp. 132-140View in ScopusGoogle Scholar
[71]Ł. Rakoczy, M. Grudzień-Rakoczy, F. Hanning, G. Cempura, R. Cygan, J. Andersson, A. Zielińska-LipiecInvestigation of the γ′ precipitates dissolution in a Ni-based superalloy during stress-free short-term annealing at high homologous temperaturesMetall. Mater. Trans. A Phys. Metall. Mater. Sci., 52 (2021), pp. 4767-4784, 10.1007/s11661-021-06420-4 View PDF This article is free to access.View in ScopusGoogle Scholar
[73]F. Masoumi, D. Shahriari, M. Jahazi, J. Cormier, A. DevauxKinetics and Mechanisms of γ′ Reprecipitation in a Ni-based SuperalloySci. Rep., 6 (2016), pp. 1-16, 10.1038/srep28650View articleGoogle Scholar
[74]A.R.P. Singh, S. Nag, S. Chattopadhyay, Y. Ren, J. Tiley, G.B. Viswanathan, H.L. Fraser, R. BanerjeeMechanisms related to different generations of γ′ precipitation during continuous cooling of a nickel base superalloyActa Mater., 61 (2013), pp. 280-293, 10.1016/j.actamat.2012.09.058View PDFView articleView in ScopusGoogle Scholar
[76]N.D. Souza, M.C. Hardy, B. Roebuck, W.E.I. Li, G.D. West, D.M. Collins, On the Rate Dependence of Precipitate Formation and Dissolution in a Nickel-Base Superalloy, Metall. Mater. Trans. A. (n.d.). https://doi.org/10.1007/s11661–022-06680–8.Google Scholar
[79]H. Huang, G. Liu, H. Wang, A. Ullah, B. HuDissolution behavior and kinetics of γ′ phase during solution treatment in powder metallurgy nickel-based superalloyMetall. Mater. Trans. A Phys. Metall. Mater. Sci., 51 (2020), pp. 1075-1084, 10.1007/s11661-019-05581-7View article View in ScopusGoogle Scholar
[80]A.J. Goodfellow, E.I. Galindo-Nava, K.A. Christofidou, N.G. Jones, T. Martin, P.A.J. Bagot, C.D. Boyer, M.C. Hardy, H.J. StoneGamma prime precipitate evolution during aging of a model nickel-based superalloyMetall. Mater. Trans. A Phys. Metall. Mater. Sci., 49 (2018), pp. 718-728, 10.1007/s11661-017-4336-y View PDF This article is free to access.View in ScopusGoogle Scholar
[81]T.P. Gabb, D.G. Backman, D.Y. Wei, D.P. Mourer, D. Furrer, A. Garg, D.L. Ellis, #947;’ Form. a Nickel-Base Disk Superalloy 2012 405 414 doi: 10.7449/2000/superalloys_2000_405_414.Google Scholar
[82]A. PlatiModelling of γ precipitation in superalloys University of CambridgeMater. Sci. (2003), p. 73Google Scholar
[91]F. Theska, W.F. Tse, B. Schulz, R. Buerstmayr, S.R. Street, M. Lison-Pick, S. PrimigReview of microstructure–mechanical property relationships in cast and wrought ni-based superalloys with boron, carbon, and zirconium microalloying additionsAdv. Eng. Mater. (2022), p. 2201514, 10.1002/ADEM.202201514 View PDF This article is free to access.Google Scholar
[93]L. Zhang, Y. Li, Q. Zhang, S. ZhangMicrostructure evolution, phase transformation and mechanical properties of IN738 superalloy fabricated by selective laser melting under different heat treatmentsMater. Sci. Eng. A, 844 (2022), Article 142947, 10.1016/j.msea.2022.142947View PDFView articleView in ScopusGoogle Scholar
[94]J.C. Franco-Correa, E. Martínez-Franco, J.M. Alvarado-Orozco, L.A. Cáceres-Díaz, D.G. Espinosa-Arbelaez, J.A. VilladaEffect of conventional heat treatments on the microstructure and microhardness of IN718 obtained by wrought and additive manufacturingJ. Mater. Eng. Perform., 30 (2021), pp. 7035-7045, 10.1007/s11665-021-06138-9View article View in ScopusGoogle Scholar
[96]N. Li, M.J. Anderson, H.C. BasoaltoAutomated stereology and uncertainty quantification considering spherical non-penetrating dispersionsPage 464.Cryst 2023, Vol. 13 (13) (2023), p. 464, 10.3390/CRYST13030464View article Google Scholar
[97]W.T. Loomis, J.W. Freeman, D.L. SponsellerInfluence of molybdenum on the γ′- phase in experimental nickelbase superalloysMet. Trans., 3 (1972), pp. 989-1000Google Scholar
[98]A.S. Shaikh Development of a γ’ Precipitation Hardening Ni-Base Superalloy for Additive Manufacturing Thesis 2018 102.〈https://odr.chalmers.se/handle/20.500.12380/255645%0Ahttps://www.researchgate.net/profile/Abdul_Shaafi_Shaikh2/publication/326226200%0Ahttps://drive.google.com/open?id=1BIez-aJyBTjnSgazzzTvv3jlrYpFC0N-〉.Google Scholar
[102]P.N. Quested, R.F. Brooks, L. Chapman, R. Morrell, Y. Youssef, K.C. MillsMeasurement and estimation of thermophysical properties of nickel based superalloysMater. Sci. Technol., 25 (2009), pp. 154-162, 10.1179/174328408×361454View articleView in ScopusGoogle Scholar
The elimination of internal macro-defects is a key issue in Ti–6Al–4V alloys fabricated via powder bed fusion using electron beams (PBF-EB), wherein internal macro-defects mainly originate from the virgin powder and inappropriate printing parameters. This study compares different types powders by combining support vector machine techniques to determine the most suitable powder for PBF-EB and to predict the processing window for the printing parameters without internal macro-defects. The results show that powders fabricated via plasma rotating electrode process have the best sphericity, flowability, and minimal porosity and are most suitable for printing. Surface roughness criterion was also applied to determine the quality of the even surfaces, and support vector machine was used to construct processing maps capable of predicting a wide range of four-dimensional printing parameters to obtain macro-defect-free samples, offering the possibility of subsequent development of Ti–6Al–4V alloys with excellent properties. The macro-defect-free samples exhibited good elongation, with the best overall mechanical properties being the ultimate tensile strength and elongation of 934.7 MPa and 24.3%, respectively. The elongation of the three macro-defect-free samples was much higher than that previously reported for additively manufactured Ti–6Al–4V alloys. The high elongation of the samples in this work is mainly attributed to the elimination of internal macro-defects.
Introduction
Additive manufacturing (AM) technologies can rapidly manufacture complex or custom parts, reducing process steps and saving manufacturing time [[1], [2], [3], [4]], and are widely used in the aerospace, automotive, and other precision industries [5,6]. Powder bed fusion using an electron beam (PBF-EB) is an additive manufacturing method that uses a high-energy electron beam to melt metal powders layer by layer to produce parts. In contrast to selective laser melting, PBF-EB involves the preparation of samples in a high vacuum environment, which effectively prevents the introduction of impurities such as O and N. It also involves a preheating process for the print substrate and powder, which reduces residual thermal stress on the sample and subsequent heat treatment processes [[2], [3], [4],7]. Due to these features and advantages, PBF-EB technology is a very important AM technology with great potential in metallic materials. Moreover, PBF-EB is the ideal AM technology for the manufacture of complex components made of many alloys, such as titanium alloys, nickel-based superalloys, aluminum alloys and stainless steels [[2], [3], [4],8].
Ti–6Al–4V alloy is one of the prevalent commercial titanium alloys possessing high specific strength, excellent mechanical properties, excellent corrosion resistance, and good biocompatibility [9,10]. It is widely used in applications requiring low density and excellent corrosion resistance, such as the aerospace industry and biomechanical applications [11,12]. The mechanical properties of PBF-EB-processed Ti–6Al–4V alloys are superior to those fabricated by casting or forging, because the rapid cooling rate in PBF-EB results in finer grains [[12], [13], [14], [15], [16], [17], [18]]. However, the PBF-EB-fabricated parts often include internal macro-defects, which compromises their mechanical properties [[19], [20], [21], [22]]. This study focused on the elimination of macro-defects, such as porosity, lack of fusion, incomplete penetration and unmelted powders, which distinguishes them from micro-defects such as vacancies, dislocations, grain boundaries and secondary phases, etc. Large-sized fusion defects cause a severe reduction in mechanical strength. Smaller defects, such as pores and cracks, lead to the initiation of fatigue cracking and rapidly accelerate the cracking process [23]. The issue of internal macro-defects must be addressed to expand the application of the PBF-EB technology. The main studies for controlling internal macro-defects are online monitoring of defects, remelting and hot isostatic pressing (HIP). The literatures [24,25] report the use of infrared imaging or other imaging techniques to identify defects, but the monitoring of smaller sized defects is still not adequate. And in some cases remelting does not reduce the internal macro-defects of the part, but instead causes coarsening of the macrostructure and volatilization of some metal elements [23]. The HIP treatment does not completely eliminate the internal macro-defects, the original defect location may still act as a point of origin of the crack, and the subsequent treatment will consume more time and economic costs [23]. Therefore, optimizing suitable printing parameters to avoid internal macro-defects in printed parts at source is of great industrial value and research significance, and is an urgent issue in PBF-EB related technology.
There are two causes of internal macro-defects in the AM process: gas pores trapped in the virgin powder and the inappropriate printing parameters [7,23]. Gui et al. [26] classify internal macro-defects during PBF-EB process according to their shape, such as spherical defects, elongated shape defects, flat shape defects and other irregular shape defects. Of these, spherical defects mainly originate from raw material powders. Other shape defects mainly originate from lack of fusion or unmelted powders caused by unsuitable printing parameters, etc. The PBF-EB process requires powders with good flowability, and spherical powders are typically chosen as raw materials. The prevalent techniques for the fabrication of pre-alloyed powders are gas atomization (GA), plasma atomization (PA), and the plasma rotating electrode process (PREP) [27,28]. These methods yield powders with different characteristics that affect the subsequent fabrication. The selection of a suitable powder for PBF-EB is particularly important to produce Ti–6Al–4V alloys without internal macro-defects. The need to optimize several printing parameters such as beam current, scan speed, line offset, and focus offset make it difficult to eliminate internal macro-defects that occur during printing [23]. Most of the studies [11,12,22,[29], [30], [31], [32], [33]] on the optimization of AM processes for Ti–6Al–4V alloys have focused on samples with a limited set of parameters (e.g., power–scan speed) and do not allow for the guidance and development of unknown process windows for macro-defect-free samples. In addition, process optimization remains a time-consuming problem, with the traditional ‘trial and error’ method demanding considerable time and economic costs. The development of a simple and efficient method to predict the processing window for alloys without internal macro-defects is a key issue. In recent years, machine learning techniques have increasingly been used in the field of additive manufacturing and materials development [[34], [35], [36], [37]]. Aoyagi et al. [38] recently proposed a novel and efficient method based on a support vector machine (SVM) to optimize the two-dimensional process parameters (current and scan speed) and obtain PBF-EB-processed CoCr alloys without internal macro-defects. The method is one of the potential approaches toward effective optimization of more than two process parameters and makes it possible for the machine learning techniques to accelerate the development of alloys without internal macro-defects.
Herein, we focus on the elimination of internal macro-defects, such as pores, lack of fusion, etc., caused by raw powders and printing parameters. The Ti–6Al–4V powders produced by three different methods were compared, and the powder with the best sphericity, flowability, and minimal porosity was selected as the feedstock for subsequent printing. The relationship between the surface roughness and internal macro-defects in the Ti–6Al–4V components was also investigated. The combination of SVM and surface roughness indices (Sdr) predicted a wider four-dimensional processing window for obtaining Ti–6Al–4V alloys without internal macro-defects. Finally, we investigated the tensile properties of Ti–6Al–4V alloys at room temperature with different printing parameters, as well as the corresponding microstructures and fracture types.
Section snippets
Starting materials
Three types of Ti–6Al–4V alloy powders, produced by GA, PA, and PREP, were compared. The particle size distribution of the powders was determined using a laser particle size analyzer (LS230, Beckman Coulter, USA), and the flowability was measured using a Hall flowmeter (JIS-Z2502, Tsutsui Scientific Instruments Co., Ltd., Japan), according to the ASTM B213 standard. The powder morphology and internal macro-defects were determined using scanning electron microscopy (SEM, JEOL JCM-6000) and X-ray
Comparison of the characteristics of GA, PA, and PREP Ti–6Al–4V powders
The particle size distributions (PSDs) and flowability of the three types of Ti–6Al–4V alloy powders produced by GA, PA, and PREP are shown in Fig. 2. Although the average particle sizes are similar (89.4 μm for GA, 82.5 μm for PA, and 86.1μm for PREP), the particle size range is different for the three types of powder (6.2–174.8 μm for GA, 27.3–139.2 μm for PA, and 39.4–133.9 μm for PREP). The flowability of the GA, PA, and PREP powders was 30.25 ± 0.98, 26.54 ± 0.37, and 25.03 ± 0.22 (s/50
Conclusions
The characteristics of the three types of Ti–6Al–4V alloy powders produced via GA, PA, and PREP were compared. The PREP powder with the best sphericity, flowability, and low porosity was found to be the most favorable powder for subsequent printing of Ti–6Al–4V alloys without internal macro-defects. The quantitative criterion of Sdr <0.015 for even surfaces was also found to be applicable to Ti–6Al–4V alloys. The process maps of Ti–6Al–4V alloys include two regions, high beam current/scan speed
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.
Acknowledgments
This study was based on the results obtained from project JPNP19007, commissioned by the New Energy and Industrial Technology Development Organization (NEDO). This work was also supported by JSPS KAKENHI (Proposal No. 21K03801) and the Inter-University Cooperative Research Program (Proposal nos. 18G0418, 19G0411, and 20G0418) of the Cooperative Research and Development Center for Advanced Materials, Institute for Materials Research, Tohoku University. It was also supported by the Council for
에피택셜 과 등축 응고 사이의 경쟁은 적층 제조에서 실행되는 레이저 용융 동안 CMSX-4 단결정 초합금에서 조사되었습니다. 단일 트랙 레이저 스캔은 레이저 출력과 스캐닝 속도의 여러 조합으로 방향성 응고된 CMSX-4 합금의 분말 없는 표면에서 수행되었습니다. EBSD(Electron Backscattered Diffraction) 매핑은 새로운 방향의 식별을 용이하게 합니다. 영역 분율 및 공간 분포와 함께 융합 영역 내에서 핵을 형성한 “스트레이 그레인”은 충실도가 높은 전산 유체 역학 시뮬레이션을 사용하여 용융 풀 내의 온도 및 유체 속도 필드를 모두 추정했습니다. 이 정보를 핵 생성 모델과 결합하여 용융 풀에서 핵 생성이 발생할 확률이 가장 높은 위치를 결정했습니다. 금속 적층 가공의 일반적인 경험에 따라 레이저 용융 트랙의 응고된 미세 구조는 에피택셜 입자 성장에 의해 지배됩니다. 더 높은 레이저 스캐닝 속도와 더 낮은 출력이 일반적으로 흩어진 입자 감소에 도움이 되지만,그럼에도 불구하고 길쭉한 용융 풀에서 흩어진 입자가 분명했습니다.
The competition between epitaxial vs. equiaxed solidification has been investigated in CMSX-4 single crystal superalloy during laser melting as practiced in additive manufacturing. Single-track laser scans were performed on a powder-free surface of directionally solidified CMSX-4 alloy with several combinations of laser power and scanning velocity. Electron backscattered diffraction (EBSD) mapping facilitated identification of new orientations, i.e., “stray grains” that nucleated within the fusion zone along with their area fraction and spatial distribution. Using high-fidelity computational fluid dynamics simulations, both the temperature and fluid velocity fields within the melt pool were estimated. This information was combined with a nucleation model to determine locations where nucleation has the highest probability to occur in melt pools. In conformance with general experience in metals additive manufacturing, the as-solidified microstructure of the laser-melted tracks is dominated by epitaxial grain growth; nevertheless, stray grains were evident in elongated melt pools. It was found that, though a higher laser scanning velocity and lower power are generally helpful in the reduction of stray grains, the combination of a stable keyhole and minimal fluid velocity further mitigates stray grains in laser single tracks.
Introduction
니켈 기반 초합금은 고온에서 긴 노출 시간 동안 높은 인장 강도, 낮은 산화 및 우수한 크리프 저항성을 포함하는 우수한 특성의 고유한 조합으로 인해 가스 터빈 엔진 응용 분야에서 광범위하게 사용됩니다. CMSX-4는 특히 장기 크리프 거동과 관련하여 초고강도의 2세대 레늄 함유 니켈 기반 단결정 초합금입니다. [ 1 , 2 ]입계의 존재가 크리프를 가속화한다는 인식은 가스 터빈 엔진의 고온 단계를 위한 단결정 블레이드를 개발하게 하여 작동 온도를 높이고 효율을 높이는 데 기여했습니다. 이러한 구성 요소는 사용 중 마모될 수 있습니다. 즉, 구성 요소의 무결성을 복원하고 단결정 미세 구조를 유지하는 수리 방법을 개발하기 위한 지속적인 작업이 있었습니다. [ 3 , 4 , 5 ]
적층 제조(AM)가 등장하기 전에는 다양한 용접 공정을 통해 단결정 초합금에 대한 수리 시도가 수행되었습니다. 균열 [ 6 , 7 ] 및 흩어진 입자 [ 8 , 9 ] 와 같은 심각한 결함 이 이 수리 중에 자주 발생합니다. 일반적으로 “스트레이 그레인”이라고 하는 응고 중 모재의 방향과 다른 결정학적 방향을 가진 새로운 그레인의 형성은 니켈 기반 단결정 초합금의 수리 중 유해한 영향으로 인해 중요한 관심 대상입니다. [ 3 , 10 ]결과적으로 재료의 단결정 구조가 손실되고 원래 구성 요소에 비해 기계적 특성이 손상됩니다. 이러한 흩어진 입자는 특정 조건에서 에피택셜 성장을 대체하는 등축 응고의 시작에 해당합니다.
떠돌이 결정립 형성을 완화하기 위해 이전 작업은 용융 영역(FZ) 내에서 응고하는 동안 떠돌이 결정립 형성에 영향을 미치는 수지상 응고 거동 및 처리 조건을 이해하는 데 중점을 두었습니다. [ 11 , 12 , 13 , 14 ] 연구원들은 단결정 합금의 용접 중에 표류 결정립 형성에 대한 몇 가지 가능한 메커니즘을 제안했습니다. [ 12 , 13 , 14 , 15 ]응고 전단에 앞서 국부적인 구성 과냉각은 이질적인 핵 생성 및 등축 결정립의 성장을 유발할 수 있습니다. 또한 용융 풀에서 활발한 유체 흐름으로 인해 발생하는 덴드라이트 조각화는 용융 풀 경계 근처에서 새로운 결정립을 형성할 수도 있습니다. 두 메커니즘 모두에서, 표류 결정립 형성은 핵 생성 위치에 의존하며, 차이점은 수상 돌기 조각화는 수상 돌기 조각이 핵 생성 위치로 작용한다는 것을 의미하는 반면 다른 메커니즘은 재료, 예 를 들어 산화물 입자에서 발견되는 다른 유형의 핵 생성 위치를 사용한다는 것을 의미합니다. 잘 알려진 바와 같이, 많은 주물에 대한 반대 접근법은 TiB와 같은 핵제의 도입을 통해 등축 응고를 촉진하는 것입니다.22알루미늄 합금에서.
헌법적 과냉 메커니즘에서 Hunt [ 11 ] 는 정상 상태 조건에서 기둥에서 등축으로의 전이(CET)를 설명하는 모델을 개발했습니다. Gaumann과 Kurz는 Hunt의 모델을 수정하여 단결정이 응고되는 동안 떠돌이 결정립이 핵을 생성하고 성장할 수 있는 정도를 설명했습니다. [ 12 , 14 ] 이후 연구에서 Vitek은 Gaumann의 모델을 개선하고 출력 및 스캐닝 속도와 같은 용접 조건의 영향에 대한 보다 자세한 분석을 포함했습니다. Vitek은 또한 실험 및 모델링 기술을 통해 표류 입자 형성에 대한 기판 방향의 영향을 포함했습니다. [ 3 , 10 ]일반적으로 높은 용접 속도와 낮은 출력은 표류 입자의 양을 최소화하고 레이저 용접 공정 중 에피택셜 단결정 성장을 최대화하는 것으로 나타났습니다. [ 3,10 ] 그러나 Vitek은 덴드라이트 조각화를 고려하지 않았으며 그의 연구는 불균질 핵형성이 레이저 용접된 CMSX -4 단결정 합금에서 표류 결정립 형성을 이끄는 주요 메커니즘임을 나타냅니다. 현재 작업에서 Vitek의 수치적 방법이 채택되고 금속 AM의 급속한 특성의 더 높은 속도와 더 낮은 전력 특성으로 확장됩니다.
AM을 통한 금속 부품 제조 는 지난 10년 동안 급격한 인기 증가를 목격했습니다. [ 16 ] EBM(Electron Beam Melting)에 의한 CMSX-4의 제작 가능성은 자주 조사되었으나 [ 17 , 18 , 19 , 20 , 21 ] CMSX의 제조 및 수리에 대한 조사는 매우 제한적이었다. – 4개의 단결정 구성요소는 레이저 분말 베드 융합(LPBF)을 사용하며, AM의 인기 있는 하위 집합으로, 특히 표류 입자 형성을 완화하는 메커니즘과 관련이 있습니다. [ 22 ]이러한 조사 부족은 주로 이러한 합금 시스템과 관련된 처리 문제로 인해 발생합니다. [ 2 , 19 , 22 , 23 , 24 ] 공정 매개변수( 예: 열원 전력, 스캐닝 속도, 스폿 크기, 예열 온도 및 스캔 전략)의 엄격한 제어는 완전히 조밀한 부품을 만들고 유지 관리할 수 있도록 하는 데 필수적입니다. 단결정 미세구조. [ 25 ] EBM을 사용하여 단결정 합금의 균열 없는 수리가 현재 가능하지만 [ 19 , 24 ] 표류 입자를 생성하지 않는 수리는 쉽게 달성할 수 없습니다.[ 23 , 26 ]
이 작업에서 LPBF를 대표하는 조건으로 레이저 용융을 사용하여 단결정 CMSX-4에서 표류 입자 완화를 조사했습니다. LPBF는 스캐닝 레이저 빔을 사용하여 금속 분말의 얇은 층을 기판에 녹이고 융합합니다. 층별 증착에서 레이저 빔의 사용은 급격한 온도 구배, 빠른 가열/냉각 주기 및 격렬한 유체 흐름을 경험하는 용융 풀을 생성 합니다 . 이것은 일반적으로 부품에 결함을 일으킬 수 있는 매우 동적인 물리적 현상으로 이어집니다. [ 28 , 29 , 30 ] 레이저 유도 키홀의 동역학( 예:, 기화 유발 반동 압력으로 인한 위상 함몰) 및 열유체 흐름은 AM 공정에서 응고 결함과 강하게 결합되고 관련됩니다. [ 31 , 32 , 33 , 34 ] 기하 구조의 급격한 변화가 발생하기 쉬운 불안정한 키홀은 다공성, 볼링, 스패터 형성 및 흔하지 않은 미세 구조 상을 포함하는 유해한 물리적 결함을 유발할 수 있습니다. 그러나 키홀 진화와 유체 흐름은 자연적으로 다음을 통해 포착 하기 어렵 습니다 .전통적인 사후 특성화 기술. 고충실도 수치 모델링을 활용하기 위해 이 연구에서는 전산유체역학(CFD)을 적용하여 표면 아래의 레이저-물질 상호 작용을 명확히 했습니다. [ 36 ] 이것은 응고된 용융물 풀의 단면에 대한 오랫동안 확립된 사후 특성화와 비교하여 키홀 및 용융물 풀 유체 흐름 정량화를 실행합니다.
CMSX-4 구성 요소의 레이저 기반 AM 수리 및 제조를 위한 적절한 절차를 개발하기 위해 적절한 공정 창을 설정하고 응고 중 표류 입자 형성 경향에 대한 예측 기능을 개발하는 것부터 시작합니다. 다중 합금에 대한 단일 트랙 증착은 분말 층이 있거나 없는 AM 공정에서 용융 풀 형상 및 미세 구조의 정확한 분석을 제공하는 것으로 나타났습니다. [ 37 , 38 , 39 ]따라서 본 연구에서는 CMSX-4의 응고 거동을 알아보기 위해 분말을 사용하지 않는 단일 트랙 레이저 스캔 실험을 사용하였다. 이는 CMSX-4 단결정의 LPBF 제조를 위한 예비 실험 지침을 제공합니다. 또한 응고 모델링은 기존 용접에서 LPBF와 관련된 급속 용접으로 확장되어 표류 입자 감소를 위한 최적의 레이저 용융 조건을 식별했습니다. 가공 매개변수 최적화를 위한 추가 지침을 제공하기 위해 용융물 풀의 매우 동적인 유체 흐름을 모델링했습니다.
재료 및 방법
단일 트랙 실험
방전 가공(EDM)을 사용하여 CMSX-4 방향성 응고 단결정 잉곳으로부터 샘플을 제작했습니다. 샘플의 최종 기하학은 치수 20의 직육면체 형태였습니다.××20××6mm. 6개 중 하나⟨ 001 ⟩⟨001⟩잉곳의 결정학적 방향은 레이저 트랙이 이 바람직한 성장 방향을 따라 스캔되도록 절단 표면에 수직으로 위치했습니다. 단일 레이저 용융 트랙은 EOS M290 기계를 사용하여 분말이 없는 샘플 표면에 만들어졌습니다. 이 기계는 최대 출력 400W, 가우시안 빔 직경 100의 이터븀 파이버 레이저가 장착된 LPBF 시스템입니다. μμ초점에서 m. 실험 중에 직사각형 샘플을 LPBF 기계용 맞춤형 샘플 홀더의 포켓에 끼워 표면을 동일한 높이로 유지했습니다. 이 맞춤형 샘플 홀더에 대한 자세한 내용은 다른 곳에서 설명합니다. 실험 은 아르곤 퍼지 분위기에서 수행되었으며 예열은 적용되지 않았습니다 . 단일 트랙 레이저 용융 실험은 다양한 레이저 출력(200~370W)과 스캔 속도(0.4~1.4m/s)에서 수행되었습니다.
성격 묘사
레이저 스캐닝 후, 레이저 빔 스캐닝 방향에 수직인 평면에서 FZ를 통해 다이아몬드 톱을 사용하여 샘플을 절단했습니다. 그 후, 샘플을 장착하고 220 그릿 SiC 페이퍼로 시작하여 콜로이드 실리카 현탁액 광택제로 마무리하여 자동 연마했습니다. 결정학적 특성화는 20kV의 가속 전압에서 TESCAN MIRA 3XMH 전계 방출 주사 전자 현미경(SEM)에서 수행되었습니다. EBSD 지도는0.4μm _0.4μ미디엄단계 크기. Bruker 시스템을 사용하여 EBSD 데이터를 정리하고 분석했습니다. EBSD 클린업은 그레인을 접촉시키기 위한 그레인 확장 루틴으로 시작한 다음 인덱스되지 않은 회절 패턴과 관련된 검은색 픽셀을 해결하기 위해 이웃 방향 클린업 루틴으로 이어졌습니다. 용융 풀 형태를 분석하기 위해 단면을 광학 현미경으로 분석했습니다. 광학 특성화의 대비를 향상시키기 위해 10g CuSO로 구성된 Marbles 시약의 변형으로 샘플을 에칭했습니다.44, 50mL HCl 및 70mL H22영형.
응고 모델링
구조적 과냉 기준에 기반한 응고 모델링을 수행하여 표유 입자의 성향 및 분포에 대한 가공 매개변수의 영향을 평가했습니다. 이 분석 모델링 접근 방식에 대한 자세한 내용은 이전 작업에서 제공됩니다. [ 3 , 10 ] 참고문헌 3 에 기술된 바와 같이 , 기본 재료의 결정학적 배향을 가진 용융 풀에서 총 표유 입자 면적 분율의 변화는 최소이므로 기본 재료 배향의 영향은 이 작업에서 고려되지 않았습니다. 우리의 LPBF 결과를 이전 작업과 비교하기 위해 Vitek의 작업에서 사용된 수학적으로 간단한 Rosenthal 방정식 [ 3 ]또한 레이저 매개변수의 함수로 용융 풀의 모양과 FZ의 열 조건을 계산하기 위한 기준으로 여기에서 채택되었습니다. Rosenthal 솔루션은 열이 일정한 재료 특성을 가진 반무한 판의 정상 상태 점원을 통해서만 전도를 통해 전달된다고 가정하며 일반적으로 다음과 같이 표현 됩니다 [ 40 , 41 ] .
티=티0+η피2 파이케이엑스2+와이2+지2———-√경험치[- 브이(엑스2+와이2+지2———-√− 엑스 )2α _] ,티=티0+η피2파이케이엑스2+와이2+지2경험치[-V(엑스2+와이2+지2-엑스)2α],(1)
여기서 T 는 온도,티0티0본 연구에서 313K( 즉 , EOS 기계 챔버 온도)로 설정된 주변 온도, P 는 레이저 빔 파워, V 는 레이저 빔 스캐닝 속도,ηη는 레이저 흡수율, k 는 열전도율,αα베이스 합금의 열확산율입니다. x , y , z 는 각각 레이저 스캐닝 방향, 가로 방향 및 세로 방향의 반대 방향과 정렬된 방향입니다 . 이 직교 좌표는 참조 3 의 그림 1에 있는 시스템을 따랐습니다 . CMSX-4에 대한 고상선 온도(1603K)와 액상선 온도(1669K)의 등온선 평균으로 응고 프런트( 즉 , 고체-액체 계면)를 정의했습니다. [ 42 , 43 , 44 ] 시뮬레이션에 사용된 열물리적 특성은 표 I 에 나열되어 있습니다.표 I CMSX-4의 응고 모델링에 사용된 열물리적 특성
어디θθ는 스캔 방향과 응고 전면의 법선 방향( 즉 , 최대 열 흐름 방향) 사이의 각도입니다. 이 연구의 용접 조건과 같은 제한된 성장에서 수지상 응고 전면은 고체-액체 등온선의 속도로 성장하도록 강제됩니다.V티V티. [ 46 ]
응고 전선이 진행되기 전에 새로 핵 생성된 입자의 국지적 비율ΦΦ, 액체 온도 구배 G 에 의해 결정 , 응고 선단 속도V티V티및 핵 밀도N0N0. 고정된 임계 과냉각에서 모든 입자가 핵형성된다고 가정함으로써△티N△티N, 등축 결정립의 반경은 결정립이 핵 생성을 시작하는 시점부터 주상 전선이 결정립에 도달하는 시간까지의 성장 속도를 통합하여 얻습니다. 과냉각으로 대체 시간d (ΔT_) / dt = – _V티G디(△티)/디티=-V티G, 열 구배 G 사이의 다음 관계 , 등축 입자의 국부적 부피 분율ΦΦ, 수상 돌기 팁 과냉각ΔT _△티, 핵 밀도N0N0, 재료 매개변수 n 및 핵생성 과냉각△티N△티N, Gäumann 외 여러분 에 의해 파생되었습니다 . [ 12 , 14 ] Hunt의 모델 [ 11 ] 의 수정에 기반함 :
계산을 단순화하기 위해 덴드라이트 팁 과냉각을 전적으로 구성 과냉각의 것으로 추정합니다.△티씨△티씨, 멱법칙 형식으로 근사화할 수 있습니다.△티씨= ( _V티)1 / 엔△티씨=(ㅏV티)1/N, 여기서 a 와 n 은 재료 종속 상수입니다. CMSX-4의 경우 이 값은a = 1.25 ×106ㅏ=1.25×106 s K 3.4m− 1-1,엔 = 3.4N=3.4, 그리고N0= 2 ×1015N0=2×1015미디엄− 3,-삼,참고문헌 3 에 의해 보고된 바와 같이 .△티N△티N2.5K이며 보다 큰 냉각 속도에서 응고에 대해 무시할 수 있습니다.106106 K/s. 에 대한 표현ΦΦ위의 방정식을 재배열하여 해결됩니다.
As proposed by Hunt,[11] a value of Φ≤0.66Φ≤0.66 pct represents fully columnar epitaxial growth condition, and, conversely, a value of Φ≥49Φ≥49 pct indicates that the initial single crystal microstructure is fully replaced by an equiaxed microstructure. To calculate the overall stray grain area fraction, we followed Vitek’s method by dividing the FZ into roughly 19 to 28 discrete parts (depending on the length of the melt pool) of equal length from the point of maximum width to the end of melt pool along the x direction. The values of G and vTvT were determined at the center on the melt pool boundary of each section and these values were used to represent the entire section. The area-weighted average of ΦΦ over these discrete sections along the length of melt pool is designated as Φ¯¯¯¯Φ¯, and is given by:
Φ¯¯¯¯=∑kAkΦk∑kAk,Φ¯=∑kAkΦk∑kAk,
(6)
where k is the index for each subsection, and AkAk and ΦkΦk are the areas and ΦΦ values for each subsection. The summation is taken over all the sections along the melt pool. Vitek’s improved model allows the calculation of stray grain area fraction by considering the melt pool geometry and variations of G and vTvT around the tail end of the pool.
수년에 걸쳐 용융 풀 현상 모델링의 정확도를 개선하기 위해 많은 고급 수치 방법이 개발되었습니다. 우리는 FLOW-3D와 함께 고충실도 CFD를 사용했습니다. FLOW-3D는 여러 물리 모델을 통합하는 상용 FVM(Finite Volume Method)입니다. [ 47 , 48 ] CFD는 유체 운동과 열 전달을 수치적으로 시뮬레이션하며 여기서 사용된 기본 물리 모델은 레이저 및 표면력 모델이었습니다. 레이저 모델에서는 레이 트레이싱 기법을 통해 다중 반사와 프레넬 흡수를 구현합니다. [ 36 ]먼저, 레이저 빔은 레이저 빔에 의해 조명되는 각 그리드 셀을 기준으로 여러 개의 광선으로 이산화됩니다. 그런 다음 각 입사 광선에 대해 입사 벡터가 입사 위치에서 금속 표면의 법선 벡터와 정렬될 때 에너지의 일부가 금속에 의해 흡수됩니다. 흡수율은 Fresnel 방정식을 사용하여 추정됩니다. 나머지 에너지는 반사광선 에 의해 유지되며 , 반사광선은 재료 표면에 부딪히면 새로운 입사광선으로 처리됩니다. 두 가지 주요 힘이 액체 금속 표면에 작용하여 자유 표면을 변형시킵니다. 금속의 증발에 의해 생성된 반동 압력은 증기 억제를 일으키는 주요 힘입니다. 본 연구에서 사용된 반동 압력 모델은피아르 자형= 특급 _{ B ( 1- _티V/ 티) }피아르 자형=ㅏ경험치{비(1-티V/티)}, 어디피아르 자형피아르 자형는 반동압력, A 와 B 는 재료의 물성에 관련된 계수로 각각 75와 15이다.티V티V는 포화 온도이고 T 는 키홀 벽의 온도입니다. 표면 흐름 및 키홀 형성의 다른 원동력은 표면 장력입니다. 표면 장력 계수는 Marangoni 흐름을 포함하기 위해 온도의 선형 함수로 추정되며,σ =1.79-9.90⋅10− 4( 티− 1654케이 )σ=1.79-9.90⋅10-4(티-1654년케이)엔엠− 1-1. [ 49 ] 계산 영역은 베어 플레이트의 절반입니다(2300 μμ미디엄××250 μμ미디엄××500 μμm) xz 평면 에 적용된 대칭 경계 조건 . 메쉬 크기는 8입니다. μμm이고 시간 단계는 0.15입니다. μμs는 계산 효율성과 정확성 간의 균형을 제공합니다.
결과 및 논의
용융 풀 형태
이 작업에 사용된 5개의 레이저 파워( P )와 6개의 스캐닝 속도( V )는 서로 다른 29개의 용융 풀을 생성했습니다.피- 브이피-V조합. P 와 V 값이 가장 높은 것은 그림 1 을 기준으로 과도한 볼링과 관련이 있기 때문에 본 연구에서는 분석하지 않았다 .
단일 트랙 용융 풀은 그림 1 과 같이 형상에 따라 네 가지 유형으로 분류할 수 있습니다 [ 39 ] : (1) 전도 모드(파란색 상자), (2) 키홀 모드(빨간색), (3) 전환 모드(마젠타), (4) 볼링 모드(녹색). 높은 레이저 출력과 낮은 스캐닝 속도의 일반적인 조합인 키홀 모드에서 용융물 풀은 일반적으로 너비/깊이( W / D ) 비율이 0.5보다 훨씬 큰 깊고 가느다란 모양을 나타냅니다 . 스캐닝 속도가 증가함에 따라 용융 풀이 얕아져 W / D 가 약 0.5인 반원형 전도 모드 용융 풀을 나타냅니다. W / D _전환 모드 용융 풀의 경우 1에서 0.5 사이입니다. 스캐닝 속도를 1200 및 1400mm/s로 더 높이면 충분히 큰 캡 높이와 볼링 모드 용융 풀의 특징인 과도한 언더컷이 발생할 수 있습니다.
힘과 속도의 함수로서의 용융 풀 깊이와 너비는 각각 그림 2 (a)와 (b)에 표시되어 있습니다. 용융 풀 폭은 기판 표면에서 측정되었습니다. 그림 2 (a)는 깊이가 레이저 출력과 매우 선형적인 관계를 따른다는 것을 보여줍니다. 속도가 증가함에 따라 깊이 대 파워 곡선의 기울기는 꾸준히 감소하지만 더 높은 속도 곡선에는 약간의 겹침이 있습니다. 이러한 예상치 못한 중첩은 종종 용융 풀 형태의 동적 변화를 유발하는 유체 흐름의 영향과 레이저 스캔당 하나의 이미지만 추출되었다는 사실 때문일 수 있습니다. 이러한 선형 동작은 그림 2 (b) 의 너비에 대해 명확하지 않습니다 . 그림 2(c)는 선형 에너지 밀도 P / V 의 함수로서 용융 깊이와 폭을 보여줍니다 . 선형 에너지 밀도는 퇴적물의 단위 길이당 에너지 투입량을 측정한 것입니다. [ 50 ] 용융 풀 깊이는 에너지 밀도에 따라 달라지며 너비는 더 많은 분산을 나타냅니다. 동일한 에너지 밀도가 준공 부품의 용융 풀, 미세 구조 또는 속성에서 반드시 동일한 유체 역학을 초래하지는 않는다는 점에 유의하는 것이 중요합니다. [ 50 ]
레이저 흡수율 평가
레이저 흡수율은 LPBF 조건에서 재료 및 가공 매개변수에 따라 크게 달라진다는 것은 잘 알려져 있습니다. [ 31 , 51 , 52 ] 적분구를 이용한 전통적인 흡수율의 직접 측정은 일반적으로 높은 비용과 구현의 어려움으로 인해 쉽게 접근할 수 없습니다. [ 51 ] 그 외 . [ 39 ] 전도 모드 용융 풀에 대한 Rosenthal 방정식을 기반으로 경험적 레이저 흡수율 모델을 개발했지만 기본 가정으로 인해 키홀 용융 풀에 대한 정확한 예측을 제공하지 못했습니다. [ 40 ] 최근 간외 . [ 53 ] Ti–6Al–4V에 대한 30개의 고충실도 다중 물리 시뮬레이션 사례를 사용하여 레이저 흡수에 대한 스케일링 법칙을 확인했습니다. 그러나 연구 중인 특정 재료에 대한 최소 흡수(평평한 용융 표면의 흡수율)에 대한 지식이 필요하며 이는 CMSX-4에 대해 알려지지 않았습니다. 다양한 키홀 모양의 용융 풀에 대한 레이저 흡수의 정확한 추정치를 얻기가 어렵기 때문에 상한 및 하한 흡수율로 분석 시뮬레이션을 실행하기로 결정했습니다. 깊은 키홀 모양의 용융 풀의 경우 대부분의 빛을 가두는 키홀 내 다중 반사로 인해 레이저 흡수율이 0.8만큼 높을 수 있습니다. 이것은 기하학적 현상이며 기본 재료에 민감하지 않습니다. [ 51, 52 , 54 ] 따라서 본 연구에서는 흡수율의 상한을 0.8로 설정하였다. 참고 문헌 51 에 나타낸 바와 같이 , 전도 용융 풀에 해당하는 최저 흡수율은 약 0.3이었으며, 이는 이 연구에서 합리적인 하한 값입니다. 따라서 레이저 흡수율이 스트레이 그레인 형성에 미치는 영향을 보여주기 위해 흡수율 값을 0.55 ± 0.25로 설정했습니다. Vitek의 작업에서는 1.0의 고정 흡수율 값이 사용되었습니다. [ 3 ]
퓨전 존 미세구조
그림 3 은 200~300W 및 600~300W 및 600~300W 범위의 레이저 출력 및 속도로 9가지 다른 처리 매개변수에 의해 생성된 CMSX-4 레이저 트랙의 yz 단면 에서 취한 EBSD 역극점도와 해당 역극점도를 보여 줍니다. 각각 1400mm/s. EBSD 맵에서 여러 기능을 쉽게 관찰할 수 있습니다. 스트레이 그레인은 EBSD 맵에서 그 방향에 해당하는 다른 RGB 색상으로 나타나고 그레인 경계를 묘사하기 위해 5도의 잘못된 방향이 사용되었습니다. 여기, 그림 3 에서 스트레이 그레인은 대부분 용융 풀의 상단 중심선에 집중되어 있으며, 이는 용접된 단결정 CMSX-4의 이전 보고서와 일치합니다. [ 10 ]역 극점도에서, 점 근처에 집중된 클러스터⟨ 001 ⟩⟨001⟩융합 경계에서 유사한 방향을 유지하는 단결정 기반 및 에피택셜로 응고된 덴드라이트를 나타냅니다. 그러나 흩어진 곡물은 식별할 수 있는 질감이 없는 흩어져 있는 점으로 나타납니다. 단결정 기본 재료의 결정학적 방향은 주로⟨ 001 ⟩⟨001⟩비록 샘플을 절단하는 동안 식별할 수 없는 기울기 각도로 인해 또는 단결정 성장 과정에서 약간의 잘못된 방향이 있었기 때문에 약간의 편차가 있지만. 용융 풀 내부의 응고된 수상 돌기의 기본 방향은 다시 한 번⟨ 001 ⟩⟨001⟩주상 결정립 구조와 유사한 에피택셜 성장의 결과. 그림 3 과 같이 용융 풀에서 수상돌기의 성장 방향은 하단의 수직 방향에서 상단의 수평 방향으로 변경되었습니다 . 이 전이는 주로 온도 구배 방향의 변화로 인한 것입니다. 두 번째 전환은 CET입니다. FZ의 상단 중심선 주변에서 다양한 방향의 흩어진 입자가 관찰되며, 여기서 안쪽으로 성장하는 수상돌기가 서로 충돌하여 용융 풀에서 응고되는 마지막 위치가 됩니다.
더 깊은 키홀 모양을 특징으로 하는 샘플에서 용융 풀의 경계 근처에 침전된 흩어진 입자가 분명합니다. 이러한 새로운 입자는 나중에 모델링 섹션에서 논의되는 수상돌기 조각화 메커니즘에 의해 잠재적으로 발생합니다. 결정립이 강한 열 구배에서 핵을 생성하고 성장한 결과, 대부분의 흩어진 결정립은 모든 방향에서 동일한 크기를 갖기보다는 장축이 열 구배 방향과 정렬된 길쭉한 모양을 갖습니다. 그림 3 의 전도 모드 용융 풀 흩어진 입자가 없는 것으로 입증되는 더 나은 단결정 품질을 나타냅니다. 상대적으로 낮은 출력과 높은 속도의 스캐닝 레이저에 의해 생성된 이러한 더 얕은 용융 풀에서 최소한의 결정립 핵형성이 발생한다는 것은 명백합니다. 더 큰 면적 분율을 가진 스트레이 그레인은 고출력 및 저속으로 생성된 깊은 용융 풀에서 더 자주 관찰됩니다. 국부 응고 조건에 대한 동력 및 속도의 영향은 후속 모델링 섹션에서 조사할 것입니다.
응고 모델링
서론에서 언급한 바와 같이 연구자들은 단결정 용접 중에 표류 결정립 형성의 가능한 메커니즘을 평가했습니다. [ 12 , 13 , 14 , 15 , 55 ]논의된 가장 인기 있는 두 가지 메커니즘은 (1) 응고 전단에 앞서 구성적 과냉각에 의해 도움을 받는 이종 핵형성 및 (2) 용융물 풀의 유체 흐름으로 인한 덴드라이트 조각화입니다. 첫 번째 메커니즘은 광범위하게 연구되었습니다. 이원 합금을 예로 들면, 고체는 액체만큼 많은 용질을 수용할 수 없으므로 응고 중에 용질을 액체로 거부합니다. 결과적으로, 성장하는 수상돌기 앞에서 용질 분할은 실제 온도가 국부 평형 액상선보다 낮은 과냉각 액체를 생성합니다. 충분히 광범위한 체질적으로 과냉각된 구역의 존재는 새로운 결정립의 핵형성 및 성장을 촉진합니다. [ 56 ]전체 과냉각은 응고 전면에서의 구성, 동역학 및 곡률 과냉각을 포함한 여러 기여의 합입니다. 일반적인 가정은 동역학 및 곡률 과냉각이 합금에 대한 용질 과냉각의 더 큰 기여와 관련하여 무시될 수 있다는 것입니다. [ 57 ]
서로 다른 기본 메커니즘을 더 잘 이해하려면피- 브이피-V조건에서 응고 모델링이 수행됩니다. 첫 번째 목적은 스트레이 그레인의 전체 범위를 평가하는 것입니다(Φ¯¯¯¯Φ¯) 처리 매개 변수의 함수로 국부적 표류 입자 비율의 변화를 조사하기 위해 (ΦΦ) 용융 풀의 위치 함수로. 두 번째 목적은 금속 AM의 빠른 응고 동안 응고 미세 구조와 표류 입자 형성 메커니즘 사이의 관계를 이해하는 것입니다.
그림 4 는 해석적으로 시뮬레이션된 표류 입자 비율을 보여줍니다.Φ¯¯¯¯Φ¯세 가지 레이저 흡수율 값에서 다양한 레이저 스캐닝 속도 및 레이저 출력에 대해. 결과는 스트레이 그레인 면적 비율이 흡수된 에너지에 민감하다는 것을 보여줍니다. 흡수율을 0.30에서 0.80으로 증가시키면Φ¯¯¯¯Φ¯약 3배이며, 이 효과는 저속 및 고출력 영역에서 더욱 두드러집니다. 다른 모든 조건이 같다면, 흡수된 전력의 큰 영향은 평균 열 구배 크기의 일반적인 감소와 용융 풀 내 평균 응고율의 증가에 기인합니다. 스캐닝 속도가 증가하고 전력이 감소함에 따라 평균 스트레이 그레인 비율이 감소합니다. 이러한 일반적인 경향은 Vitek의 작업에서 채택된 그림 5 의 파란색 영역에서 시뮬레이션된 용접 결과와 일치합니다 . [ 3 ] 더 큰 과냉각 구역( 즉, 지 /V티G/V티영역)은 용접 풀의 표유 입자의 면적 비율이 분홍색 영역에 해당하는 LPBF 조건의 면적 비율보다 훨씬 더 크다는 것을 의미합니다. 그럼에도 불구하고 두 데이터 세트의 일반적인 경향은 유사합니다. 즉 , 레이저 출력이 감소하고 레이저 속도가 증가함에 따라 표류 입자의 비율이 감소합니다. 또한 그림 5 에서 스캐닝 속도가 LPBF 영역으로 증가함에 따라 표유 입자 면적 분율에 대한 레이저 매개변수의 변화 효과가 감소한다는 것을 추론할 수 있습니다. 그림 6 (a)는 그림 3 의 EBSD 분석에서 나온 실험적 표류 결정립 면적 분율 과 그림 4 의 해석 시뮬레이션 결과를 비교합니다.. 열쇠 구멍 모양의 FZ에서 정확한 값이 다르지만 추세는 시뮬레이션과 실험 데이터 모두에서 일관되었습니다. 키홀 모양의 용융 풀, 특히 전력이 300W인 2개는 분석 시뮬레이션 예측보다 훨씬 더 많은 양의 흩어진 입자를 가지고 있습니다. Rosenthal 방정식은 일반적으로 열 전달이 순전히 전도에 의해 좌우된다는 가정으로 인해 열쇠 구멍 체제의 열 흐름을 적절하게 반영하지 못하기 때문에 이러한 불일치가 실제로 예상됩니다. [ 39 , 40 ] 그것은 또한 그림 4 의 발견 , 즉 키홀 모드 동안 흡수된 전력의 증가가 표류 입자 형성에 더 이상적인 조건을 초래한다는 것을 검증합니다. 그림 6 (b)는 실험을 비교Φ¯¯¯¯Φ¯수치 CFD 시뮬레이션Φ¯¯¯¯Φ¯. CFD 모델이 약간 초과 예측하지만Φ¯¯¯¯Φ¯전체적으로피- 브이피-V조건에서 열쇠 구멍 조건에서의 예측은 분석 모델보다 정확합니다. 전도 모드 용융 풀의 경우 실험 값이 분석 시뮬레이션 값과 더 가깝게 정렬됩니다.
모의 온도 구배 G 분포 및 응고율 검사V티V티분석 모델링의 쌍은 그림 7 (a)의 CMSX-4 미세 구조 선택 맵에 표시됩니다. 제공지 /V티G/V티( 즉 , 형태 인자)는 형태를 제어하고지 ×V티G×V티( 즉 , 냉각 속도)는 응고된 미세 구조의 규모를 제어하고 , [ 58 , 59 ]지 -V티G-V티플롯은 전통적인 제조 공정과 AM 공정 모두에서 미세 구조 제어를 지원합니다. 이 플롯의 몇 가지 분명한 특징은 등축, 주상, 평면 전면 및 이러한 경계 근처의 전이 영역을 구분하는 경계입니다. 그림 7 (a)는 몇 가지 선택된 분석 열 시뮬레이션에 대한 미세 구조 선택 맵을 나타내는 반면 그림 7 (b)는 수치 열 모델의 결과와 동일한 맵을 보여줍니다. 등축 미세구조의 형성은 낮은 G 이상 에서 명확하게 선호됩니다.V티V티정황. 이 플롯에서 각 곡선의 평면 전면에 가장 가까운 지점은 용융 풀의 최대 너비 위치에 해당하는 반면 등축 영역에 가까운 지점의 끝은 용융 풀의 후면 꼬리에 해당합니다. 그림 7 (a)에서 대부분의지 -V티G-V티응고 전면의 쌍은 원주형 영역에 속하고 점차 CET 영역으로 위쪽으로 이동하지만 용융 풀의 꼬리는 다음에 따라 완전히 등축 영역에 도달하거나 도달하지 않을 수 있습니다.피- 브이피-V조합. 그림 7 (a) 의 곡선 중 어느 것도 평면 전면 영역을 통과하지 않지만 더 높은 전력의 경우에 가까워집니다. 저속 레이저 용융 공정을 사용하는 이전 작업에서는 곡선이 평면 영역을 통과할 수 있습니다. 레이저 속도가 증가함에 따라 용융 풀 꼬리는 여전히 CET 영역에 있지만 완전히 등축 영역에서 멀어집니다. CET 영역으로 떨어지는 섹션의 수도 감소합니다.Φ¯¯¯¯Φ¯응고된 물질에서.
그만큼지 -V티G-V티CFD 모델을 사용하여 시뮬레이션된 응고 전면의 쌍이 그림 7 (b)에 나와 있습니다. 세 방향 모두에서 각 점 사이의 일정한 간격으로 미리 정의된 좌표에서 수행된 해석 시뮬레이션과 달리, 고충실도 CFD 모델의 출력은 불규칙한 사면체 좌표계에 있었고 G 를 추출하기 전에 일반 3D 그리드에 선형 보간되었습니다. 그리고V티V티그런 다음 미세 구조 선택 맵에 플롯됩니다. 일반적인 경향은 그림 7 (a)의 것과 일치하지만 이 방법으로 모델링된 매우 동적인 유체 흐름으로 인해 결과에 더 많은 분산이 있었습니다. 그만큼지 -V티G-V티분석 열 모델의 쌍 경로는 더 연속적인 반면 수치 시뮬레이션의 경로는 용융 풀 꼬리 모양의 차이를 나타내는 날카로운 굴곡이 있습니다(이는 G 및V티V티) 두 모델에 의해 시뮬레이션됩니다.
유체 흐름을 통합한 응고 모델링
수치 CFD 모델을 사용하여 유동 입자 형성 정도에 대한 유체 흐름의 영향을 이해하고 시뮬레이션 결과를 분석 Rosenthal 솔루션과 비교했습니다. 그림 8 은 응고 매개변수 G 의 분포를 보여줍니다.V티V티,지 /V티G/V티, 그리고지 ×V티G×V티yz 단면에서 x 는 FLOW-3D에서 (a1–d1) 분석 열 모델링 및 (a2–d2) FVM 방법을 사용하여 시뮬레이션된 용융 풀의 최대 폭입니다. 그림 8 의 값은 응고 전선이 특정 위치에 도달할 때 정확한 값일 수도 있고 아닐 수도 있지만 일반적인 추세를 반영한다는 의미의 임시 가상 값입니다. 이 프로파일은 출력 300W 및 속도 400mm/s의 레이저 빔에서 시뮬레이션됩니다. 용융 풀 경계는 흰색 곡선으로 표시됩니다. (a2–d2)의 CFD 시뮬레이션 용융 풀 깊이는 342입니다. μμm, 측정 깊이 352와 잘 일치 μμ일치하는 길쭉한 열쇠 구멍 모양과 함께 그림 1 에 표시된 실험 FZ의 m . 그러나 분석 모델은 반원 모양의 용융 풀을 출력하고 용융 풀 깊이는 264에 불과합니다. μμ열쇠 구멍의 경우 현실과는 거리가 멀다. CFD 시뮬레이션 결과에서 열 구배는 레이저 반사 증가와 불안정한 액체-증기 상호 작용이 발생하는 증기 함몰의 동적 부분 근처에 있기 때문에 FZ 하단에서 더 높습니다. 대조적으로 해석 결과의 열 구배 크기는 경계를 따라 균일합니다. 두 시뮬레이션 결과 모두 그림 8 (a1) 및 (a2) 에서 응고가 용융 풀의 상단 중심선을 향해 진행됨에 따라 열 구배가 점차 감소합니다 . 응고율은 그림 8 과 같이 경계 근처에서 거의 0입니다. (b1) 및 (b2). 이는 경계 영역이 응고되기 시작할 때 국부 응고 전면의 법선 방향이 레이저 스캐닝 방향에 수직이기 때문입니다. 이것은 드라이브θ → π/ 2θ→파이/2그리고V티→ 0V티→0식에서 [ 3 ]. 대조적으로 용융 풀의 상단 중심선 근처 영역에서 응고 전면의 법선 방향은 레이저 스캐닝 방향과 잘 정렬되어 있습니다.θ → 0θ→0그리고V티→ 브이V티→V, 빔 스캐닝 속도. G 와 _V티V티값이 얻어지면 냉각 속도지 ×V티G×V티및 형태 인자지 /V티G/V티계산할 수 있습니다. 그림 8 (c2)는 용융 풀 바닥 근처의 온도 구배가 매우 높고 상단에서 더 빠른 성장 속도로 인해 냉각 속도가 용융 풀의 바닥 및 상단 중심선 근처에서 더 높다는 것을 보여줍니다. 지역. 그러나 이러한 추세는 그림 8 (c1)에 캡처되지 않았습니다. 그림 8 의 형태 요인 (d1) 및 (d2)는 중심선에 접근함에 따라 눈에 띄게 감소합니다. 경계에서 큰 값은 열 구배를 거의 0인 성장 속도로 나누기 때문에 발생합니다. 이 높은 형태 인자는 주상 미세구조 형성 가능성이 높음을 시사하는 반면, 중앙 영역의 값이 낮을수록 등축 미세구조의 가능성이 더 크다는 것을 나타냅니다. Tanet al. 또한 키홀 모양의 용접 풀 [ 59 ] 에서 이러한 응고 매개변수의 분포 를 비슷한 일반적인 경향으로 보여주었습니다. 그림 3 에서 볼 수 있듯이 용융 풀의 상단 중심선에 있는 흩어진 입자는 낮은 특징을 나타내는 영역과 일치합니다.지 /V티G/V티그림 8 (d1) 및 (d2)의 값. 시뮬레이션과 실험 간의 이러한 일치는 용융 풀의 상단 중심선에 축적된 흩어진 입자의 핵 생성 및 성장이 등온선 속도의 증가와 온도 구배의 감소에 의해 촉진됨을 보여줍니다.
그림 9 는 유체 속도 및 국부적 핵형성 성향을 보여줍니다.ΦΦ300W의 일정한 레이저 출력과 400, 800 및 1200mm/s의 세 가지 다른 레이저 속도에 의해 생성된 3D 용융 풀 전체에 걸쳐. 그림 9 (d)~(f)는 로컬ΦΦ해당 3D 보기에서 밝은 회색 평면으로 표시된 특정 yz 단면의 분포. 이 yz 섹션은 가장 높기 때문에 선택되었습니다.Φ¯¯¯¯Φ¯용융 풀 내의 값은 각각 23.40, 11.85 및 2.45pct입니다. 이들은 그림 3 의 실험 데이터와 비교하기에 적절하지 않을 수 있는 액체 용융 풀의 과도 값이며Φ¯¯¯¯Φ¯그림 6 의 값은 이 값이 고체-액체 계면에 가깝지 않고 용융 풀의 중간에서 취해졌기 때문입니다. 온도가 훨씬 낮아서 핵이 생존하고 성장할 수 있기 때문에 핵 형성은 용융 풀의 중간이 아닌 고체-액체 계면에 더 가깝게 발생할 가능성이 있습니다.
그림 3 (a), (d), (g), (h)에서 위쪽 중심선에서 멀리 떨어져 있는 흩어진 결정립이 있었습니다. 그들은 훨씬 더 높은 열 구배와 더 낮은 응고 속도 필드에 위치하기 때문에 과냉각 이론은 이러한 영역에서 표류 입자의 형성에 대한 만족스러운 설명이 아닙니다. 이것은 떠돌이 결정립의 형성을 야기할 수 있는 두 번째 메커니즘, 즉 수상돌기의 팁을 가로지르는 유체 흐름에 의해 유발되는 수상돌기 조각화를 고려하도록 동기를 부여합니다. 유체 흐름이 열 구배를 따라 속도 성분을 갖고 고체-액체 계면 속도보다 클 때, 주상 수상돌기의 국지적 재용융은 용질이 풍부한 액체가 흐물흐물한 구역의 깊은 곳에서 액상선 등온선까지 이동함으로써 발생할 수 있습니다. . [ 55] 분리된 수상돌기는 대류에 의해 열린 액체로 운반될 수 있습니다. 풀이 과냉각 상태이기 때문에 이러한 파편은 고온 조건에서 충분히 오래 생존하여 길 잃은 입자의 핵 생성 사이트로 작용할 수 있습니다. 결과적으로 수상 돌기 조각화 과정은 활성 핵의 수를 효과적으로 증가시킬 수 있습니다.N0N0) 용융 풀 [ 15 , 60 , 61 ] 에서 생성된 미세 구조에서 표류 입자의 면적을 증가시킵니다.
그림 9 (a) 및 (b)에서 반동 압력은 용융 유체를 아래쪽으로 흐르게 하여 결과 흐름을 지배합니다. 유체 속도의 역방향 요소는 V = 400 및 800mm/s에 대해 각각 최대값 1.0 및 1.6m/s로 더 느려집니다 . 그림 9 (c)에서 레이저 속도가 더 증가함에 따라 증기 침하가 더 얕고 넓어지고 반동 압력이 더 고르게 분포되어 증기 침강에서 주변 영역으로 유체를 밀어냅니다. 역류는 최대값 3.5m/s로 더 빨라집니다. 용융 풀의 최대 너비에서 yz 단면 의 키홀 아래 평균 유체 속도는 그림에 표시된 경우에 대해 0.46, 0.45 및 1.44m/s입니다.9 (a), (b) 및 (c). 키홀 깊이의 변동은 각 경우의 최대 깊이와 최소 깊이의 차이로 정의되는 크기로 정량화됩니다. 240 범위의 강한 증기 내림 변동 μμm은 그림 9 (a)의 V = 400mm/s 경우에서 발견 되지만 이 변동은 그림 9 (c)에서 16의 범위로 크게 감소합니다.μμ미디엄. V = 400mm/s인 경우 의 유체장과 높은 변동 범위는 이전 키홀 동역학 시뮬레이션과 일치합니다. [ 34 ]
따라서 V = 400mm/s 키홀 케이스의 무질서한 변동 흐름이 용융 풀 경계를 따라 응고된 주상 수상돌기에서 분리된 조각을 구동할 가능성이 있습니다. V = 1200mm/s의 경우 강한 역류 는 그림 3 에서 관찰되지 않았지만 동일한 효과를 가질 수 있습니다. . 덴드라이트 조각화에 대한 유체 유동장의 영향에 대한 이 경험적 설명은 용융 풀 경계 근처에 떠돌이 입자의 존재에 대한 그럴듯한 설명을 제공합니다. 분명히 하기 위해, 우리는 이 가설을 검증하기 위해 이 현상에 대한 직접적인 실험적 관찰을 하지 않았습니다. 이 작업에서 표유 입자 면적 분율을 계산할 때 단순화를 위해 핵 생성 모델링에 일정한 핵 생성 수 밀도가 적용되었습니다. 이는 그림 9 의 표류 입자 영역 비율 이 수지상정 조각화가 발생하는 경우 이러한 높은 유체 흐름 용융 풀에서 발생할 수 있는 것, 즉 강화된 핵 생성 밀도를 반영하지 않는다는 것을 의미합니다.
위의 이유로 핵 형성에 대한 수상 돌기 조각화의 영향을 아직 배제할 수 없습니다. 그러나 단편화 이론은 용접 문헌 [ 62 ] 에서 검증될 만큼 충분히 개발되지 않았 으므로 부차적인 중요성만 고려된다는 점에 유의해야 합니다. 1200mm/s를 초과하는 레이저 스캐닝 속도는 최소한의 표류 결정립 면적 분율을 가지고 있음에도 불구하고 분명한 볼링을 나타내기 때문에 단결정 수리 및 AM 처리에 적합하지 않습니다. 따라서 낮은 P 및 높은 V 에 의해 생성된 응고 전면 근처에서 키홀 변동이 최소화되고 유체 속도가 완만해진 용융 풀이 생성된다는 결론을 내릴 수 있습니다., 처리 창의 극한은 아니지만 흩어진 입자를 나타낼 가능성이 가장 적습니다.
마지막으로 단일 레이저 트랙의 응고 거동을 조사하면 에피택셜 성장 동안 표류 입자 형성을 더 잘 이해할 수 있다는 점에 주목하는 것이 중요합니다. 우리의 현재 결과는 최적의 레이저 매개변수에 대한 일반적인 지침을 제공하여 최소 스트레이 그레인을 달성하고 단결정 구조를 유지합니다. 이 가이드라인은 250W 정도의 전력과 600~800mm/s의 스캔 속도로 최소 흩어진 입자에 적합한 공정 창을 제공합니다. 각 처리 매개변수를 신중하게 선택하면 과거에 스테인리스강에 대한 거의 단결정 미세 구조를 인쇄하는 데 성공했으며 이는 CMSX-4 AM 빌드에 대한 가능성을 보여줍니다. [ 63 ]신뢰성을 보장하기 위해 AM 수리 프로세스를 시작하기 전에 보다 엄격한 실험 테스트 및 시뮬레이션이 여전히 필요합니다. 둘 이상의 레이저 트랙 사이의 상호 작용도 고려해야 합니다. 또한 레이저, CMSX-4 분말 및 벌크 재료 간의 상호 작용이 중요하며, 수리 중에 여러 층의 CMSX-4 재료를 축적해야 하는 경우 다른 스캔 전략의 효과도 중요한 역할을 할 수 있습니다. 분말이 포함된 경우 Lopez-Galilea 등 의 연구에서 제안한 바와 같이 분말이 주로 완전히 녹지 않았을 때 추가 핵 생성 사이트를 도입하기 때문에 단순히 레이저 분말과 속도를 조작하여 흩어진 입자 형성을 완화하기 어려울 수 있습니다 . [ 22 ]결과적으로 CMSX-4 단결정을 수리하기 위한 레이저 AM의 가능성을 다루기 위해서는 기판 재료, 레이저 출력, 속도, 해치 간격 및 층 두께의 조합을 모두 고려해야 하며 향후 연구에서 다루어야 합니다. CFD 모델링은 2개 이상의 레이저 트랙 사이의 상호작용과 열장에 미치는 영향을 통합할 수 있으며, 이는 AM 빌드 시나리오 동안 핵 생성 조건으로 단일 비드 연구의 지식 격차를 해소할 것입니다.
결론
LPBF 제조의 특징적인 조건 하에서 CMSX-4 단결정 의 에피택셜(기둥형) 대 등축 응고 사이의 경쟁을 실험적 및 이론적으로 모두 조사했습니다. 이 연구는 고전적인 응고 개념을 도입하여 빠른 레이저 용융의 미세 구조 특징을 설명하고 응고 조건과 표유 결정 성향을 예측하기 위해 해석적 및 수치적 고충실도 CFD 열 모델 간의 비교를 설명했습니다. 본 연구로부터 다음과 같은 주요 결론을 도출할 수 있다.
단일 레이저 트랙의 레이저 가공 조건은 용융 풀 형상, 레이저 흡수율, 유체 흐름 및 키홀 요동, 입자 구조 및 표류 입자 형성 민감성에 강한 영향을 미치는 것으로 밝혀졌습니다.
레이저 용접을 위해 개발된 이론적인 표유 결정립 핵형성 분석이 레이저 용융 AM 조건으로 확장되었습니다. 분석 모델링 결과와 단일 레이저 트랙의 미세구조 특성화를 비교하면 예측이 전도 및 볼링 조건에서 실험적 관찰과 잘 일치하는 반면 키홀 조건에서는 예측이 약간 과소하다는 것을 알 수 있습니다. 이러한 불일치는 레이저 트랙의 대표성이 없는 섹션이나 유체 속도 필드의 변화로 인해 발생할 수 있습니다. CFD 모델에서 추출한 열장에 동일한 표유 입자 계산 파이프라인을 적용하면 연구된 모든 사례에서 과대평가가 발생하지만 분석 모델보다 연장된 용융 풀의 실험 데이터와 더 정확하게 일치합니다.
이 연구에서 두 가지 표류 결정립 형성 메커니즘인 불균일 핵형성 및 수상돌기 조각화가 평가되었습니다. 우리의 결과는 불균일 핵형성이 용융 풀의 상단 중심선에서 새로운 결정립의 형성으로 이어지는 주요 메커니즘임을 시사합니다.지 /V티G/V티정권.
용융 풀 경계 근처의 흩어진 입자는 깊은 키홀 모양의 용융 풀에서 독점적으로 관찰되며, 이는 강한 유체 흐름으로 인한 수상 돌기 조각화의 영향이 이러한 유형의 용융 풀에서 고려하기에 충분히 강력할 수 있음을 시사합니다.
일반적으로 더 높은 레이저 스캐닝 속도와 더 낮은 전력 외에도 안정적인 키홀과 최소 유체 속도는 또한 흩어진 입자 형성을 완화하고 레이저 단일 트랙에서 에피택셜 성장을 보존합니다.
References
R.C. Reed: The Superalloys: Fundamentals and Applications, Cambridge University Press, Cambridge, 2006, pp.17–20.BookGoogle Scholar
A. Basak, R. Acharya, and S. Das: Metall. Mater. Trans. A, 2016, vol. 47A, pp. 3845–59.ArticleGoogle Scholar
R. Vilar and A. Almeida: J. Laser Appl., 2015, vol. 27, p. S17004.ArticleGoogle Scholar
T. Kalfhaus, M. Schneider, B. Ruttert, D. Sebold, T. Hammerschmidt, J. Frenzel, R. Drautz, W. Theisen, G. Eggeler, O. Guillon, and R. Vassen: Mater. Des., 2019, vol. 168, p. 107656.ArticleCASGoogle Scholar
S.S. Babu, S.A. David, J.W. Park, and J.M. Vitek: Sci. Technol. Weld. Join., 2004, vol. 9, pp. 1–12.ArticleCASGoogle Scholar
L. Felberbaum, K. Voisey, M. Gäumann, B. Viguier, and A. Mortensen: Mater. Sci. Eng. A, 2001, vol. 299, pp. 152–56.ArticleGoogle Scholar
S. Mokadem, C. Bezençon, J.M. Drezet, A. Jacot, J.D. Wagnière, and W. Kurz: TMS Annual Meeting, 2004, pp. 67–76.
J.M. Vitek: ASM Proc. Int. Conf. Trends Weld. Res., vol. 2005, pp. 773–79.
B. Kianian: Wohlers Report 2017: 3D Printing and Additive Manufacturing State of the Industry, Annual Worldwide Progress Report, Wohlers Associates, Inc., Fort Collins, 2017.Google Scholar
M. Ramsperger, L. Mújica Roncery, I. Lopez-Galilea, R.F. Singer, W. Theisen, and C. Körner: Adv. Eng. Mater., 2015, vol. 17, pp. 1486–93.ArticleCASGoogle Scholar
A.B. Parsa, M. Ramsperger, A. Kostka, C. Somsen, C. Körner, and G. Eggeler: Metals, 2016, vol. 6, pp. 258-1–17.ArticleGoogle Scholar
C. Körner, M. Ramsperger, C. Meid, D. Bürger, P. Wollgramm, M. Bartsch, and G. Eggeler: Metall. Mater. Trans. A, 2018, vol. 49A, pp. 3781–92.ArticleGoogle Scholar
D. Bürger, A. Parsa, M. Ramsperger, C. Körner, and G. Eggeler: Mater. Sci. Eng. A, 2019, vol. 762, p. 138098,ArticleGoogle Scholar
R. Cunningham, C. Zhao, N. Parab, C. Kantzos, J. Pauza, K. Fezzaa, T. Sun, and A.D. Rollett: Science, 2019, vol. 363, pp. 849–52.ArticleCASGoogle Scholar
B. Fotovvati, S.F. Wayne, G. Lewis, and E. Asadi: Adv. Mater. Sci. Eng., 2018, vol. 2018, p. 4920718.ArticleGoogle Scholar
P.-J. Chiang, R. Jiang, R. Cunningham, N. Parab, C. Zhao, K. Fezzaa, T. Sun, and A.D. Rollett: in Advanced Real Time Imaging II, pp. 77–85.
J. Ye, S.A. Khairallah, A.M. Rubenchik, M.F. Crumb, G. Guss, J. Belak, and M.J. Matthews: Adv. Eng. Mater., 2019, vol. 21, pp. 1–9.ArticleGoogle Scholar
C. Zhao, Q. Guo, X. Li, N. Parab, K. Fezzaa, W. Tan, L. Chen, and T. Sun: Phys. Rev. X, 2019, vol. 9, p. 021052.CASGoogle Scholar
S.A. Khairallah, A.T. Anderson, A. Rubenchik, and W.E. King: Acta Mater., 2016, vol. 108, pp. 36–45.ArticleCASGoogle Scholar
N. Kouraytem, X. Li, R. Cunningham, C. Zhao, N. Parab, T. Sun, A.D. Rollett, A.D. Spear, and W. Tan: Appl. Phys. Rev., 2019, vol. 11, p. 064054.ArticleCASGoogle Scholar
T. DebRoy, H. Wei, J. Zuback, T. Mukherjee, J. Elmer, J. Milewski, A. Beese, A. Wilson-Heid, A. De, and W. Zhang: Prog. Mater. Sci., 2018, vol. 92, pp. 112–224.ArticleCASGoogle Scholar
I. Yadroitsev, A. Gusarov, I. Yadroitsava, and I. Smurov: J. Mater. Process. Technol., 2010, vol. 210, pp. 1624–31.ArticleCASGoogle Scholar
S. Ghosh, L. Ma, L.E. Levine, R.E. Ricker, M.R. Stoudt, J.C. Heigel, and J.E. Guyer: JOM, 2018, vol. 70, pp. 1011–16.ArticleCASGoogle Scholar
Y. He, C. Montgomery, J. Beuth, and B. Webler: Mater. Des., 2019, vol. 183, p. 108126.ArticleCASGoogle Scholar
D. Rosenthal: Weld. J., 1941, vol. 20, pp. 220–34.Google Scholar
M. Tang, P.C. Pistorius, and J.L. Beuth: Addit. Manuf., 2017, vol. 14, pp. 39–48.CASGoogle Scholar
R.E. Aune, L. Battezzati, R. Brooks, I. Egry, H.J. Fecht, J.P. Garandet, M. Hayashi, K.C. Mills, A. Passerone, P.N. Quested, E. Ricci, F. Schmidt-Hohagen, S. Seetharaman, B. Vinet, and R.K. Wunderlich: Proc. Int.Symp. Superalloys Var. Deriv., 2005, pp. 467–76.
B.C. Wilson, J.A. Hickman, and G.E. Fuchs: JOM, 2003, vol. 55, pp. 35–40.ArticleCASGoogle Scholar
J.J. Valencia and P.N. Quested: ASM Handb., 2008, vol. 15, pp. 468–81.Google Scholar
H.L. Wei, J. Mazumder, and T. DebRoy: Sci. Rep., 2015, vol. 5, pp. 1–7.Google Scholar
N. Raghavan, R. Dehoff, S. Pannala, S. Simunovic, M. Kirka, J. Turner, N. Carlson, and S.S. Babu: Acta Mater., 2016, vol. 112, pp. 303–14.ArticleCASGoogle Scholar
R. Lin, H. Wang, F. Lu, J. Solomon, and B.E. Carlson: Int. J. Heat Mass Transf., 2017, vol. 108, pp. 244–56.ArticleCASGoogle Scholar
M. Bayat, A. Thanki, S. Mohanty, A. Witvrouw, S. Yang, J. Thorborg, N.S. Tiedje, and J.H. Hattel: Addit. Manuf., 2019, vol. 30, p. 100835.CASGoogle Scholar
K. Higuchi, H.-J. Fecht, and R.K. Wunderlich: Adv. Eng. Mater., 2007, vol. 9, pp. 349–54.ArticleCASGoogle Scholar
Q. Guo, C. Zhao, M. Qu, L. Xiong, L.I. Escano, S.M.H. Hojjatzadeh, N.D. Parab, K. Fezzaa, W. Everhart, T. Sun, and L. Chen: Addit. Manuf., 2019, vol. 28, pp. 600–09.Google Scholar
J. Trapp, A.M. Rubenchik, G. Guss, and M.J. Matthews: Appl. Mater. Today, 2017, vol. 9, pp. 341–49.ArticleGoogle Scholar
M. Schneider, L. Berthe, R. Fabbro, and M. Muller: J. Phys. D, 2008, vol. 41, p. 155502.ArticleGoogle Scholar
Z. Gan, O.L. Kafka, N. Parab, C. Zhao, L. Fang, O. Heinonen, T. Sun, and W.K. Liu: Nat. Commun., 2021, vol. 12, p. 2379.ArticleCASGoogle Scholar
B.J. Simonds, E.J. Garboczi, T.A. Palmer, and P.A. Williams: Appl. Phys. Rev., 2020, vol. 13, p. 024057.ArticleCASGoogle Scholar
J. Dantzig and M. Rappaz: Solidification, 2nd ed., EPFL Press, Lausanne, 2016, pp. 483–532.Google Scholar
W. Tiller, K. Jackson, J. Rutter, and B. Chalmers: Acta Metall., 1953, vol. 1, pp. 428–37.ArticleCASGoogle Scholar
D. Zhang, A. Prasad, M.J. Bermingham, C.J. Todaro, M.J. Benoit, M.N. Patel, D. Qiu, D.H. StJohn, M. Qian, and M.A. Easton: Metall. Mater. Trans. A, 2020, vol. 51A, pp. 4341–59.ArticleGoogle Scholar
F. Yan, W. Xiong, and E.J. Faierson: Materials, 2017, vol. 10, p. 1260.ArticleGoogle Scholar
J.M. Vitek, S.A. David, and L.A. Boatner: Sci. Technol. Weld. Join., 1997, vol. 2, pp. 109–18.ArticleCASGoogle Scholar
X. Wang, J.A. Muñiz-Lerma, O. Sanchez-Mata, S.E. Atabay, M.A. Shandiz, and M. Brochu: Prog. Addit. Manuf., 2020, vol. 5, pp. 41–49.ArticleGoogle Scholar
레이저 금속 적층 제조(AM) 공정의 낮은 에너지 효율은 대규모 산업 생산에서 잠재적인 지속 가능성 문제입니다. 레이저 용융을 위한 에너지 효율의 명시적 조사는 용융 금속의 불투명한 특성으로 인해 매우 어려운 용융 풀 치수 및 증기 내림의 직접적인 특성화를 요구합니다.
여기에서 우리는 현장 고속 고에너지 x-선 이미징에 의해 Al6061의 레이저 분말 베드 융합(LPBF) 동안 증기 강하 및 용융 풀 형성에 대한 TiC 나노 입자의 효과에 대한 직접적인 관찰 및 정량화를 보고합니다. 정량 결과를 바탕으로, 우리는 Al6061의 LPBF 동안 TiC 나노 입자가 있거나 없을 때 레이저 용융 에너지 효율(여기서 재료를 용융하는 데 필요한 에너지 대 레이저 빔에 의해 전달되는 에너지의 비율로 정의)을 계산했습니다.
결과는 TiC 나노 입자를 Al6061에 추가하면 레이저 용융 에너지 효율이 크게 증가한다는 것을 보여줍니다(평균 114% 증가, 312에서 521% 증가). W 레이저 출력, 0.4m /s 스캔 속도). 체계적인 특성 측정, 시뮬레이션 및 x-선 이미징 연구를 통해 우리는 처음으로 세 가지 메커니즘이 함께 작동하여 레이저 용융 에너지 효율을 향상시킨다는 것을 확인할 수 있었습니다.
(1) TiC 나노 입자를 추가하면 흡수율이 증가합니다. (2) TiC 나노입자를 추가하면 열전도율이 감소하고, (3) TiC 나노입자를 추가하면 더 낮은 레이저 출력에서 증기 억제 및 다중 반사를 시작할 수 있습니다(즉, 키홀링에 대한 레이저 출력 임계값을 낮춤).
여기서 보고한 Al6061의 LPBF 동안 레이저 용융 에너지 효율을 증가시키기 위해 TiC 나노입자를 사용하는 방법 및 메커니즘은 보다 에너지 효율적인 레이저 금속 AM을 위한 공급원료 재료의 개발을 안내할 수 있습니다.
The low energy efficiency of the laser metal additive manufacturing (AM) process is a potential sustainability concern for large-scale industrial production. Explicit investigation of the energy efficiency for laser melting requires the direct characterization of melt pool dimension and vapor depression, which is very difficult due to the opaque nature of the molten metal. Here we report the direct observation and quantification of effects of the TiC nanoparticles on the vapor depression and melt pool formation during laser powder bed fusion (LPBF) of Al6061 by in-situ high-speed high-energy x-ray imaging. Based on the quantification results, we calculated the laser melting energy efficiency (defined here as the ratio of the energy needed to melt the material to the energy delivered by the laser beam) with and without TiC nanoparticles during LPBF of Al6061. The results show that adding TiC nanoparticles into Al6061 leads to a significant increase of laser melting energy efficiency (114% increase on average, 521% increase under 312 W laser power, 0.4 m/s scan speed). Systematic property measurement, simulation, and x-ray imaging studies enable us, for the first time, to identify that three mechanisms work together to enhance the laser melting energy efficiency: (1) adding TiC nanoparticles increases the absorptivity; (2) adding TiC nanoparticles decreases the thermal conductivity, and (3) adding TiC nanoparticles enables the initiation of vapor depression and multiple reflection at lower laser power (i.e., lowers the laser power threshold for keyholing). The method and mechanisms of using TiC nanoparticles to increase the laser melting energy efficiency during LPBF of Al6061 we reported here may guide the development of feedstock materials for more energy efficient laser metal AM.
316-L 스테인리스강의 레이저 분말 베드 융합 중 콜드 스패터 형성의 충실도 높은 수치 모델링
W.E. ALPHONSO1*, M. BAYAT1 and J.H. HATTEL1 *Corresponding author 1Technical University of Denmark (DTU), 2800, Kgs, Lyngby, Denmark
ABSTRACT
L-PBF(Laser Powder Bed Fusion)는 금속 적층 제조(MAM) 기술로, 기존 제조 공정에 비해 부품 설계 자유도, 조립품 통합, 부품 맞춤화 및 낮은 툴링 비용과 같은 여러 이점을 산업에 제공합니다.
전기 코일 및 열 관리 장치는 일반적으로 높은 전기 및 열 전도성 특성으로 인해 순수 구리로 제조됩니다. 따라서 순동의 L-PBF가 가능하다면 기하학적으로 최적화된 방열판과 자유형 전자코일을 제작할 수 있습니다.
그러나 L-PBF로 조밀한 순동 부품을 생산하는 것은 적외선에 대한 낮은 광 흡수율과 높은 열전도율로 인해 어렵습니다. 기존의 L-PBF 시스템에서 조밀한 구리 부품을 생산하려면 적외선 레이저의 출력을 500W 이상으로 높이거나 구리의 광흡수율이 높은 녹색 레이저를 사용해야 합니다.
적외선 레이저 출력을 높이면 후면 반사로 인해 레이저 시스템의 광학 구성 요소가 손상되고 렌즈의 열 광학 현상으로 인해 공정이 불안정해질 수 있습니다. 이 작업에서 FVM(Finite Volume Method)에 기반한 다중 물리학 중간 규모 수치 모델은 Flow-3D에서 개발되어 용융 풀 역학과 궁극적으로 부품 품질을 제어하는 물리적 현상 상호 작용을 조사합니다.
녹색 레이저 열원과 적외선 레이저 열원은 기판 위의 순수 구리 분말 베드에 단일 트랙 증착을 생성하기 위해 개별적으로 사용됩니다.
용융 풀 역학에 대한 레이저 열원의 유사하지 않은 광학 흡수 특성의 영향이 탐구됩니다. 수치 모델을 검증하기 위해 단일 트랙이 구리 분말 베드에 증착되고 시뮬레이션된 용융 풀 모양과 크기가 비교되는 실험이 수행되었습니다.
녹색 레이저는 광흡수율이 높아 전도 및 키홀 모드 용융이 가능하고 적외선 레이저는 흡수율이 낮아 키홀 모드 용융만 가능하다. 레이저 파장에 대한 용융 모드의 변화는 궁극적으로 기계적, 전기적 및 열적 특성에 영향을 미치는 열 구배 및 냉각 속도에 대한 결과를 가져옵니다.
Laser Powder Bed Fusion (L-PBF) is a Metal Additive Manufacturing (MAM) technology which offers several advantages to industries such as part design freedom, consolidation of assemblies, part customization and low tooling cost over conventional manufacturing processes. Electric coils and thermal management devices are generally manufactured from pure copper due to its high electrical and thermal conductivity properties. Therefore, if L-PBF of pure copper is feasible, geometrically optimized heat sinks and free-form electromagnetic coils can be manufactured. However, producing dense pure copper parts by L-PBF is difficult due to low optical absorptivity to infrared radiation and high thermal conductivity. To produce dense copper parts in a conventional L-PBF system either the power of the infrared laser must be increased above 500W, or a green laser should be used for which copper has a high optical absorptivity. Increasing the infrared laser power can damage the optical components of the laser systems due to back reflections and create instabilities in the process due to thermal-optical phenomenon of the lenses. In this work, a multi-physics meso-scale numerical model based on Finite Volume Method (FVM) is developed in Flow-3D to investigate the physical phenomena interaction which governs the melt pool dynamics and ultimately the part quality. A green laser heat source and an infrared laser heat source are used individually to create single track deposition on pure copper powder bed above a substrate. The effect of the dissimilar optical absorptivity property of laser heat sources on the melt pool dynamics is explored. To validate the numerical model, experiments were conducted wherein single tracks are deposited on a copper powder bed and the simulated melt pool shape and size are compared. As the green laser has a high optical absorptivity, a conduction and keyhole mode melting is possible while for the infrared laser only keyhole mode melting is possible due to low absorptivity. The variation in melting modes with respect to the laser wavelength has an outcome on thermal gradient and cooling rates which ultimately affect the mechanical, electrical, and thermal properties.
Keywords
Pure Copper, Laser Powder Bed Fusion, Finite Volume Method, multi-physics
References
[1] L. Jyothish Kumar, P. M. Pandey, and D. I. Wimpenny, 3D printing and additive manufacturing technologies. Springer Singapore, 2018. doi: 10.1007/978-981-13-0305-0. [2] T. DebRoy et al., “Additive manufacturing of metallic components – Process, structure and properties,” Progress in Materials Science, vol. 92, pp. 112–224, 2018, doi: 10.1016/j.pmatsci.2017.10.001. [3] C. S. Lefky, B. Zucker, D. Wright, A. R. Nassar, T. W. Simpson, and O. J. Hildreth, “Dissolvable Supports in Powder Bed Fusion-Printed Stainless Steel,” 3D Printing and Additive Manufacturing, vol. 4, no. 1, pp. 3–11, 2017, doi: 10.1089/3dp.2016.0043. [4] J. L. Bartlett and X. Li, “An overview of residual stresses in metal powder bed fusion,” Additive Manufacturing, vol. 27, no. January, pp. 131–149, 2019, doi: 10.1016/j.addma.2019.02.020. [5] I. H. Ahn, “Determination of a process window with consideration of effective layer thickness in SLM process,” International Journal of Advanced Manufacturing Technology, vol. 105, no. 10, pp. 4181–4191, 2019, doi: 10.1007/s00170-019-04402-w.
[6] R. McCann et al., “In-situ sensing, process monitoring and machine control in Laser Powder Bed Fusion: A review,” Additive Manufacturing, vol. 45, no. May, 2021, doi: 10.1016/j.addma.2021.102058. [7] M. Bayat et al., “Keyhole-induced porosities in Laser-based Powder Bed Fusion (L-PBF) of Ti6Al4V: High-fidelity modelling and experimental validation,” Additive Manufacturing, vol. 30, no. August, p. 100835, 2019, doi: 10.1016/j.addma.2019.100835. [8] M. Bayat, S. Mohanty, and J. H. Hattel, “Multiphysics modelling of lack-of-fusion voids formation and evolution in IN718 made by multi-track/multi-layer L-PBF,” International Journal of Heat and Mass Transfer, vol. 139, pp. 95–114, 2019, doi: 10.1016/j.ijheatmasstransfer.2019.05.003. [9] S. D. Jadhav, L. R. Goossens, Y. Kinds, B. van Hooreweder, and K. Vanmeensel, “Laserbased powder bed fusion additive manufacturing of pure copper,” Additive Manufacturing, vol. 42, no. March, 2021, doi: 10.1016/j.addma.2021.101990. [10] S. D. Jadhav, S. Dadbakhsh, L. Goossens, J. P. Kruth, J. van Humbeeck, and K. Vanmeensel, “Influence of selective laser melting process parameters on texture evolution in pure copper,” Journal of Materials Processing Technology, vol. 270, no. January, pp. 47–58, 2019, doi: 10.1016/j.jmatprotec.2019.02.022. [11] H. Siva Prasad, F. Brueckner, J. Volpp, and A. F. H. Kaplan, “Laser metal deposition of copper on diverse metals using green laser sources,” International Journal of Advanced Manufacturing Technology, vol. 107, no. 3–4, pp. 1559–1568, 2020, doi: 10.1007/s00170- 020-05117-z. [12] L. R. Goossens, Y. Kinds, J. P. Kruth, and B. van Hooreweder, “On the influence of thermal lensing during selective laser melting,” Solid Freeform Fabrication 2018: Proceedings of the 29th Annual International Solid Freeform Fabrication Symposium – An Additive Manufacturing Conference, SFF 2018, no. December, pp. 2267–2274, 2020. [13] M. Bayat, V. K. Nadimpalli, D. B. Pedersen, and J. H. Hattel, “A fundamental investigation of thermo-capillarity in laser powder bed fusion of metals and alloys,” International Journal of Heat and Mass Transfer, vol. 166, p. 120766, 2021, doi: 10.1016/j.ijheatmasstransfer.2020.120766. [14] H. Chen, Q. Wei, Y. Zhang, F. Chen, Y. Shi, and W. Yan, “Powder-spreading mechanisms in powder-bed-based additive manufacturing: Experiments and computational modeling,” Acta Materialia, vol. 179, pp. 158–171, 2019, doi: 10.1016/j.actamat.2019.08.030. [15] S. K. Nayak, S. K. Mishra, C. P. Paul, A. N. Jinoop, and K. S. Bindra, “Effect of energy density on laser powder bed fusion built single tracks and thin wall structures with 100 µm preplaced powder layer thickness,” Optics and Laser Technology, vol. 125, May 2020, doi: 10.1016/j.optlastec.2019.106016. [16] G. Nordet et al., “Absorptivity measurements during laser powder bed fusion of pure copper with a 1 kW cw green laser,” Optics & Laser Technology, vol. 147, no. April 2021, p. 107612, 2022, doi: 10.1016/j.optlastec.2021.107612. [17] M. Hummel, C. Schöler, A. Häusler, A. Gillner, and R. Poprawe, “New approaches on laser micro welding of copper by using a laser beam source with a wavelength of 450 nm,” Journal of Advanced Joining Processes, vol. 1, no. February, p. 100012, 2020, doi: 10.1016/j.jajp.2020.100012. [18] M. Hummel, M. Külkens, C. Schöler, W. Schulz, and A. Gillner, “In situ X-ray tomography investigations on laser welding of copper with 515 and 1030 nm laser beam sources,” Journal of Manufacturing Processes, vol. 67, no. April, pp. 170–176, 2021, doi: 10.1016/j.jmapro.2021.04.063. [19] L. Gargalis et al., “Determining processing behaviour of pure Cu in laser powder bed fusion using direct micro-calorimetry,” Journal of Materials Processing Technology, vol. 294, no. March, p. 117130, 2021, doi: 10.1016/j.jmatprotec.2021.117130. [20] A. Mondal, D. Agrawal, and A. Upadhyaya, “Microwave heating of pure copper powder with varying particle size and porosity,” Journal of Microwave Power and Electromagnetic Energy, vol. 43, no. 1, pp. 4315–43110, 2009, doi: 10.1080/08327823.2008.11688599.
•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.
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
W.E. Alphonso1, M.Bayat1,*, M. Baier 2, S. Carmignato2, J.H. Hattel1 1Department of Mechanical Engineering, Technical University of Denmark (DTU), Lyngby, Denmark 2Department of Management and Engineering – University of Padova, Padova, Italy
ABSTRACT
L-PBF(Laser Powder Bed Fusion)는 레이저 열원을 사용하여 선택적으로 통합되는 분말 층으로 복잡한 3D 금속 부품을 만드는 금속 적층 제조(MAM) 기술입니다. 처리 영역은 수십 마이크로미터 정도이므로 L-PBF를 다중 규모 제조 공정으로 만듭니다.
기체 기공의 형성 및 성장 및 용융되지 않은 분말 영역의 생성은 다중물리 모델에 의해 예측할 수 있습니다. 또한 이러한 모델을 사용하여 용융 풀 모양 및 크기, 온도 분포, 용융 풀 유체 흐름 및 입자 크기 및 형태와 같은 미세 구조 특성을 계산할 수 있습니다.
이 작업에서는 용융, 응고, 유체 흐름, 표면 장력, 열 모세관, 증발 및 광선 추적을 통한 다중 반사를 포함하는 스테인리스 스틸 316-L에 대한 충실도 다중 물리학 중간 규모 수치 모델이 개발되었습니다. 완전한 실험 설계(DoE) 방법을 사용하는 통계 연구가 수행되었으며, 여기서 불확실한 재료 특성 및 공정 매개변수, 즉 흡수율, 반동 압력(기화) 및 레이저 빔 크기가 용융수지 모양 및 크기에 미치는 영향을 분석했습니다.
또한 용융 풀 역학에 대한 위에서 언급한 불확실한 입력 매개변수의 중요성을 강조하기 위해 흡수율이 가장 큰 영향을 미치고 레이저 빔 크기가 그 뒤를 잇는 주요 효과 플롯이 생성되었습니다. 용융 풀 크기에 대한 반동 압력의 중요성은 흡수율에 따라 달라지는 용융 풀 부피와 함께 증가합니다.
모델의 예측 정확도는 유사한 공정 매개변수로 생성된 단일 트랙 실험과 시뮬레이션의 용융 풀 모양 및 크기를 비교하여 검증됩니다.
더욱이, 열 렌즈 효과는 레이저 빔 크기를 증가시켜 수치 모델에서 고려되었으며 나중에 결과적인 용융 풀 프로파일은 모델의 견고성을 보여주기 위한 실험과 비교되었습니다.
Laser Powder Bed Fusion (L-PBF) is a Metal Additive Manufacturing (MAM) technology where a complex 3D metal part is built from powder layers, which are selectively consolidated using a laser heat source. The processing zone is in the order of a few tenths of micrometer, making L-PBF a multi-scale manufacturing process. The formation and growth of gas pores and the creation of un-melted powder zones can be predicted by multiphysics models. Also, with these models, the melt pool shape and size, temperature distribution, melt pool fluid flow and its microstructural features like grain size and morphology can be calculated. In this work, a high fidelity multi-physics meso-scale numerical model is developed for stainless steel 316-L which includes melting, solidification, fluid flow, surface tension, thermo-capillarity, evaporation and multiple reflection with ray-tracing. A statistical study using a full Design of Experiments (DoE) method was conducted, wherein the impact of uncertain material properties and process parameters namely absorptivity, recoil pressure (vaporization) and laser beam size on the melt pool shape and size was analysed. Furthermore, to emphasize on the significance of the above mentioned uncertain input parameters on the melt pool dynamics, a main effects plot was created which showed that absorptivity had the highest impact followed by laser beam size. The significance of recoil pressure on the melt pool size increases with melt pool volume which is dependent on absorptivity. The prediction accuracy of the model is validated by comparing the melt pool shape and size from the simulation with single track experiments that were produced with similar process parameters. Moreover, the effect of thermal lensing was considered in the numerical model by increasing the laser beam size and later on the resultant melt pool profile was compared with experiments to show the robustness of the model.
CONCLUSION
In this work, a high-fidelity multi-physics numerical model was developed for L-PBF using the FVM method in Flow-3D. The impact of uncertainty in the input parameters including absorptivity, recoil pressure and laser beam size on the melt pool is addressed using a DoE method. The DoE analysis shows that absorptivity has the highest impact on the melt pool. The recoil pressure and laser beam size only become significant once absorptivity is 0.45. Furthermore, the numerical model is validated by comparing the predicted melt pool shape and size with experiments conducted with similar process parameters wherein a high prediction accuracy is achieved by the model. In addition, the impact of thermal lensing on the melt pool dimensions by increasing the laser beam spot size is considered in the validated numerical model and the resultant melt pool is compared with experiments.
REFERENCES
[1] T. Bonhoff, M. Schniedenharn, J. Stollenwerk, P. Loosen, Experimental and theoretical analysis of thermooptical effects in protective window for selective laser melting, Proc. Int. Conf. Lasers Manuf. LiM. (2017) 26–29. https://www.wlt.de/lim/Proceedings2017/Data/PDF/Contribution31_final.pdf. [2] L.R. Goossens, Y. Kinds, J.P. Kruth, B. van Hooreweder, On the influence of thermal lensing during selective laser melting, Solid Free. Fabr. 2018 Proc. 29th Annu. Int. Solid Free. Fabr. Symp. – An Addit. Manuf. Conf. SFF 2018. (2020) 2267–2274. [3] J. Shinjo, C. Panwisawas, Digital materials design by thermal-fluid science for multi-metal additive manufacturing, Acta Mater. 210 (2021) 116825. https://doi.org/10.1016/j.actamat.2021.116825. [4] Z. Zhang, Y. Huang, A. Rani Kasinathan, S. Imani Shahabad, U. Ali, Y. Mahmoodkhani, E. Toyserkani, 3- Dimensional heat transfer modeling for laser powder-bed fusion additive manufacturing with volumetric heat sources based on varied thermal conductivity and absorptivity, Opt. Laser Technol. 109 (2019) 297–312. https://doi.org/10.1016/j.optlastec.2018.08.012. [5] M. Bayat, A. Thanki, S. Mohanty, A. Witvrouw, S. Yang, J. Thorborg, N.S. Tiedje, J.H. Hattel, Keyholeinduced porosities in Laser-based Powder Bed Fusion (L-PBF) of Ti6Al4V: High-fidelity modelling and experimental validation, Addit. Manuf. 30 (2019) 100835. https://doi.org/10.1016/j.addma.2019.100835. [6] M. Bayat, S. Mohanty, J.H. Hattel, Multiphysics modelling of lack-of-fusion voids formation and evolution in IN718 made by multi-track/multi-layer L-PBF, Int. J. Heat Mass Transf. 139 (2019) 95–114. https://doi.org/10.1016/j.ijheatmasstransfer.2019.05.003. [7] J. Metelkova, Y. Kinds, K. Kempen, C. de Formanoir, A. Witvrouw, B. Van Hooreweder, On the influence of laser defocusing in Selective Laser Melting of 316L, Addit. Manuf. 23 (2018) 161–169. https://doi.org/10.1016/j.addma.2018.08.006.
M. Bayat* , V. K. Nadimpalli, J. H. Hattel 1Department of Mechanical Engineering, Technical University of Denmark (DTU), Produktionstorvet 425, Kgs. 2800, Lyngby, Denmark
ABSTRACT
L-PBF(Laser Powder Bed Fusion)는 다양한 산업 분야에서 많은 관심을 받았으며, 주로 기존 제조 기술을 사용하여 만들 수 없었던 복잡한 토폴로지 최적화 구성 요소를 구현하는 잘 알려진 능력 덕분입니다. . 펄스 L-PBF(PL-PBF)에서 레이저의 시간적 프로파일은 주기 지속 시간과 듀티 주기 중 하나 또는 둘 다를 수정하여 변조할 수 있습니다. 따라서 레이저의 시간적 프로파일은 향후 적용을 위해 이 프로세스를 더 잘 제어할 수 있는 길을 열어주는 새로운 프로세스 매개변수로 간주될 수 있습니다. 따라서 이 작업에서 우리는 레이저의 시간적 프로파일을 변경하는 것이 PL-PBF 공정에서 용융 풀 조건과 트랙의 최종 모양 및 형상에 어떻게 영향을 미칠 수 있는지 조사하는 것을 목표로 합니다. 이와 관련하여 본 논문에서는 CFD(Computational Fluid Dynamics) 소프트웨어 패키지인 Flow-3D를 기반으로 하는 316-L 스테인리스강 PL-PBF 공정의 다중물리 수치 모델을 개발하고 이 모델을 사용하여 열과 유체를 시뮬레이션합니다. 다양한 펄스 모드에서 공정 과정 중 용융 풀 내부에서 발생하는 유동 조건. 따라서 고정된 레이저 듀티 사이클(50%)이 있는 레이저 주기 지속 시간이 용융 풀의 모양과 크기 및 최종 트랙 형태에 미치는 영향을 연구하기 위해 매개변수 연구가 수행됩니다. 더 긴 주기 기간에서 더 많은 재료가 더 큰 용융 풀 내에서 변위됨에 따라 용융 풀의 후류에 더 눈에 띄는 혹이 형성되며, 동시에 더 심각한 반동 압력을 받습니다. 또한 시뮬레이션에서 50% 듀티 사이클에서 1000μs에서 형성된 보다 대칭적인 용융 풀과 비교하여 400μs 사이클 주기에서 더 긴 용융 풀이 형성된다는 것이 관찰되었습니다. 풀 볼륨은 1000μs의 경우 더 큽니다. 매개변수 연구는 연속 트랙과 파손된 트랙 PL-PBF 사이의 경계를 설명하며, 여기서 연속 트랙은 항상 소량의 용융 재료를 유지함으로써 유지됩니다.
English Abstract
Laser Powder Bed Fusion (L-PBF) has attracted a lot of attention from various industrial sectors and mainly thanks to its well-proven well-known capacity of realizing complex topology-optimized components that have so far been impossible to make using conventional manufacturing techniques. In Pulsed L-PBF (PL-PBF), the laser’s temporal profile can be modulated via modifying either or both the cycle duration and the duty cycle. Thus, the laser’s temporal profile could be considered as a new process parameter that paves the way for a better control of this process for future applications. Therefore, in this work we aim to investigate how changing the laser’s temporal profile can affect the melt pool conditions and the final shape and geometry of a track in the PL-PBF process. In this respect, in this paper a multiphysics numerical model of the PL-PBF process of 316-L stainless steel is developed based on the computational fluid dynamics (CFD) software package Flow-3D and the model is used to simulate the heat and fluid flow conditions occurring inside the melt pool during the course of the process at different pulsing modes. Thus, a parametric study is carried out to study the influence of the laser’s cycle duration with a fixed laser duty cycle (50 %) on the shape and size of the melt pool and the final track morphology. It is noticed that at longer cycle periods, more noticeable humps form at the wake of the melt pool as more material is displaced within bigger melt pools, which are at the same time subjected to more significant recoil pressures. It is also observed in the simulations that at 50 % duty cycle, longer melt pools form at 400 μs cycle period compared to the more symmetrical melt pools formed at 1000 μs, primarily because of shorter laser off-times in the former, even though melt pool volume is bigger for the 1000 μs case. The parameteric study illustrates the boundary between a continuous track and a broken track PL-PBF wherein the continuous track is retained by always maintaining a small volume of molten material.
CONCLUSIONS
In this work a CFD model of the modulated PL-PBF process of stainless steel 316-L is developed in the commercial software package Flow-3D. The model involves physics such as solidification, melting, evaporation, convection, laser-material interaction, capillarity, Marangoni effect and the recoil pressure effect. In the current study, a parametric study is carried out to understand how the change in the cycle period duration affects the melt pool’s thermo-fluid conditions during the modulated PL-PBF process. It is observed that at the pulse mode with 50 % duty cycle and 400 μs cycle period, an overlapped chain of humps form at the wake of the melt pool and at a spatial frequency of occurrence of about 78 μm. Furthermore and as expected, it is noted that the melt pool volume, the size of the hump as well as the crater size at the end of the track, increase with increase in the cycle period duration, as more material is re-deposited at the back of the melt pool and that itself is caused by more pronounced recoil pressures. Moreover, it is noticed that due to the short off-time period of the laser in the 400 μs cycle period case, there is always an amount of liquid metal left from the previous cycle, at the time the new cycle starts. This is found to be the main reason why longer and elongated melt pools form at 400 μs cycle period, compared to the bigger, shorter and more symmetrical-like melt pools forming at the 1000 μs case. In this study PL-PBF single tracks including the broken track and the continuous track examples were studied to illustrate the boundary of this transition at a given laser scan parameter setting. At higher scan speeds, it is expected that the Plateau–Rayleigh instability will compete with the pulsing behavior to change the transition boundary between a broken and continuous track, which is suggested as future work from this study.
REFERENCES
[1] T. Craeghs, L. Thijs, F. Verhaeghe, J.-P. Kruth, J. Van Humbeeck, A study of the microstructural evolution during selective laser melting of Ti–6Al–4V, Acta Mater. 58 (2010) 3303–3312. https://doi.org/10.1016/j.actamat.2010.02.004. [2] J. Liu, A.T. Gaynor, S. Chen, Z. Kang, K. Suresh, A. Takezawa, L. Li, J. Kato, J. Tang, C.C.L. Wang, L. Cheng, X. Liang, A.C. To, Current and future trends in topology optimization for additive manufacturing, (2018) 2457–2483. [3] M. Bayat, W. Dong, J. Thorborg, A.C. To, J.H. Hattel, A review of multi-scale and multi-physics simulations of metal additive manufacturing processes with focus on modeling strategies, Addit. Manuf. 47 (2021). https://doi.org/10.1016/j.addma.2021.102278. [4] A. Foroozmehr, M. Badrossamay, E. Foroozmehr, S. Golabi, Finite Element Simulation of Selective Laser Melting process considering Optical Penetration Depth of laser in powder bed, Mater. Des. 89 (2016) 255–263. https://doi.org/10.1016/j.matdes.2015.10.002. [5] Y.S. Lee, W. Zhang, Modeling of heat transfer, fluid flow and solidification microstructure of nickel-base superalloy fabricated by laser powder bed fusion, Addit. Manuf. 12 (2016) 178–188. https://doi.org/10.1016/j.addma.2016.05.003. [6] S.A. Khairallah, A.T. Anderson, A. Rubenchik, 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. 108 (2016) 36–45. https://doi.org/10.1016/j.actamat.2016.02.014. [7] M. Bayat, A. Thanki, S. Mohanty, A. Witvrouw, S. Yang, J. Thorborg, N.S. Tiedje, J.H. Hattel, Keyholeinduced porosities in Laser-based Powder Bed Fusion (L-PBF) of Ti6Al4V: High-fidelity modelling and experimental validation, Addit. Manuf. 30 (2019). https://doi.org/10.1016/j.addma.2019.100835. [8] A. Charles, M. Bayat, A. Elkaseer, L. Thijs, J.H. Hattel, S. Scholz, Elucidation of dross formation in laser powder bed fusion at down-facing surfaces: phenomenon-oriented multiphysics simulation and experimental validation, Addit. Manuf. Under revi (2021). [9] M. Bayat, V.K. Nadimpalli, D.B. Pedersen, J.H. Hattel, A fundamental investigation of thermo-capillarity in laser powder bed fusion of metals and alloys, Int. J. Heat Mass Transf. 166 (2021) 120766. https://doi.org/10.1016/j.ijheatmasstransfer.2020.120766. [10] J.D. Roehling, S.A. Khairallah, Y. Shen, A. Bayramian, C.D. Boley, A.M. Rubenchik, J. Demuth, N. Duanmu, M.J. Matthews, Physics of large-area pulsed laser powder bed fusion, Addit. Manuf. 46 (2021) https://doi.org/10.1016/j.addma.2021.102186. [11] M. Zheng, L. Wei, J. Chen, Q. Zhang, J. Li, S. Sui, G. Wang, W. Huang, Surface morphology evolution during pulsed selective laser melting: Numerical and experimental investigations, Appl. Surf. Sci. 496 (2019) 143649. https://doi.org/10.1016/j.apsusc.2019.143649. [12] M. Bayat, V.K. Nadimpalli, D.B. Pedersen, J.H. Hattel, A fundamental investigation of thermo-capillarity in laser powder bed fusion of metals and alloys, Int. J. Heat Mass Transf. 166 (2021). https://doi.org/10.1016/j.ijheatmasstransfer.2020.120766.
레이저 분말 베드 퓨전(L-PBF) 적층 제조(AM)는 우수한 기계적 특성으로 그물 모양에 가까운 복잡한 부품을 생산할 수 있습니다. 그러나 빌드 실패 및 다공성과 같은 결함으로 이어지는 원치 않는 잔류 응력 및 왜곡이 L-PBF의 광범위한 적용을 방해하고 있습니다.
L-PBF의 잠재력을 최대한 실현하기 위해 잔류 변형, 용융 풀 및 다공성 형성을 예측하는 다중 규모 모델링 방법론이 개발되었습니다. L-PBF의 잔류 변형 및 응력을 부품 규모에서 예측하기 위해 고유 변형 방법을 기반으로 하는 다중 규모 프로세스 모델링 프레임워크가 제안됩니다.
고유한 변형 벡터는 마이크로 스케일에서 충실도가 높은 상세한 다층 프로세스 시뮬레이션에서 추출됩니다. 균일하지만 이방성인 변형은 잔류 왜곡 및 응력을 예측하기 위해 준 정적 평형 유한 요소 분석(FEA)에서 레이어별로 L-PBF 부품에 적용됩니다.
부품 규모에서의 잔류 변형 및 응력 예측 외에도 분말 규모의 다중물리 모델링을 수행하여 공정 매개변수, 예열 온도 및 스패터링 입자에 의해 유도된 용융 풀 변동 및 결함 형성을 연구합니다. 이러한 요인과 관련된 용융 풀 역학 및 다공성 형성 메커니즘은 시뮬레이션 및 실험을 통해 밝혀졌습니다.
제안된 부품 규모 잔류 응력 및 왜곡 모델을 기반으로 경로 계획 방법은 큰 잔류 변형 및 건물 파손을 방지하기 위해 주어진 형상에 대한 레이저 스캐닝 경로를 조정하기 위해 개발되었습니다.
연속 및 아일랜드 스캐닝 전략을 위한 기울기 기반 경로 계획이 공식화되고 공식화된 컴플라이언스 및 스트레스 최소화 문제에 대한 전체 감도 분석이 수행됩니다. 이 제안된 경로 계획 방법의 타당성과 효율성은 AconityONE L-PBF 시스템을 사용하여 실험적으로 입증되었습니다.
또한 기계 학습을 활용한 데이터 기반 프레임워크를 개발하여 L-PBF에 대한 부품 규모의 열 이력을 예측합니다. 본 연구에서는 실시간 열 이력 예측을 위해 CNN(Convolutional Neural Network)과 RNN(Recurrent Neural Network)을 포함하는 순차적 기계 학습 모델을 제안합니다.
유한 요소 해석과 비교하여 100배의 예측 속도 향상이 달성되어 실제 제작 프로세스보다 빠른 예측이 가능하고 실시간 온도 프로파일을 사용할 수 있습니다.
Laser powder bed fusion (L-PBF) additive manufacturing (AM) is capable of producing complex parts near net shape with good mechanical properties. However, undesired residual stress and distortion that lead to build failure and defects such as porosity are preventing broader applications of L-PBF. To realize the full potential of L-PBF, a multiscale modeling methodology is developed to predict residual deformation, melt pool, and porosity formation. To predict the residual deformation and stress in L-PBF at part-scale, a multiscale process modeling framework based on inherent strain method is proposed.
Inherent strain vectors are extracted from detailed multi-layer process simulation with high fidelity at micro-scale. Uniform but anisotropic strains are then applied to L-PBF part in a layer-by-layer fashion in a quasi-static equilibrium finite element analysis (FEA) to predict residual distortion and stress. Besides residual distortion and stress prediction at part scale, multiphysics modeling at powder scale is performed to study the melt pool variation and defect formation induced by process parameters, preheating temperature and spattering particles. Melt pool dynamics and porosity formation mechanisms associated with these factors are revealed through simulation and experiments.
Based on the proposed part-scale residual stress and distortion model, path planning method is developed to tailor the laser scanning path for a given geometry to prevent large residual deformation and building failures. Gradient based path planning for continuous and island scanning strategy is formulated and full sensitivity analysis for the formulated compliance- and stress-minimization problem is performed.
The feasibility and effectiveness of this proposed path planning method is demonstrated experimentally using the AconityONE L-PBF system. In addition, a data-driven framework utilizing machine learning is developed to predict the thermal history at part-scale for L-PBF.
In this work, a sequential machine learning model including convolutional neural network (CNN) and recurrent neural network (RNN), long shortterm memory unit, is proposed for real-time thermal history prediction. A 100x prediction speed improvement is achieved compared to the finite element analysis which makes the prediction faster than real fabrication process and real-time temperature profile available.
Bibliography
[1] I. Astm, ASTM52900-15 Standard Terminology for Additive Manufacturing—General Principles—Terminology, ASTM International, West Conshohocken, PA 3(4) (2015) 5. [2] W.E. King, A.T. Anderson, R.M. Ferencz, N.E. Hodge, C. Kamath, S.A. Khairallah, A.M. Rubenchik, Laser powder bed fusion additive manufacturing of metals; physics, computational, and materials challenges, Applied Physics Reviews 2(4) (2015) 041304. [3] W. Yan, Y. Lu, K. Jones, Z. Yang, J. Fox, P. Witherell, G. Wagner, W.K. Liu, Data-driven characterization of thermal models for powder-bed-fusion additive manufacturing, Additive Manufacturing (2020) 101503. [4] K. Dai, L. Shaw, Thermal and stress modeling of multi-material laser processing, Acta Materialia 49(20) (2001) 4171-4181. [5] K. Dai, L. Shaw, Distortion minimization of laser-processed components through control of laser scanning patterns, Rapid Prototyping Journal 8(5) (2002) 270-276. [6] S.S. Bo Cheng, Kevin Chou, Stress and deformation evaluations of scanning strategy effect in selective laser melting, Additive Manufacturing (2017). [7] C. Fu, Y. Guo, Three-dimensional temperature gradient mechanism in selective laser melting of Ti-6Al-4V, Journal of Manufacturing Science and Engineering 136(6) (2014) 061004. [8] P. Prabhakar, W.J. Sames, R. Dehoff, S.S. Babu, Computational modeling of residual stress formation during the electron beam melting process for Inconel 718, Additive Manufacturing 7 (2015) 83-91. [9] A. Hussein, L. Hao, C. Yan, R. Everson, Finite element simulation of the temperature and stress fields in single layers built without-support in selective laser melting, Materials & Design (1980-2015) 52 (2013) 638-647. [10] P.Z. Qingcheng Yang, Lin Cheng, Zheng Min, Minking Chyu, Albert C. To, articleFinite element modeling and validation of thermomechanicalbehavior of Ti-6Al-4V in directed energy deposition additivemanufacturing, Additive Manufacturing (2016). [11] E.R. Denlinger, J. Irwin, P. Michaleris, Thermomechanical Modeling of Additive Manufacturing Large Parts, Journal of Manufacturing Science and Engineering 136(6) (2014) 061007. [12] E.R. Denlinger, M. Gouge, J. Irwin, P. Michaleris, Thermomechanical model development and in situ experimental validation of the Laser Powder-Bed Fusion process, Additive Manufacturing 16 (2017) 73-80. [13] V.J. Erik R Denlinger, G.V. Srinivasan, Tahany EI-Wardany, Pan Michaleris, Thermal modeling of Inconel 718 processed with powder bed fusionand experimental validation using in situ measurements, Additive Manufacturing 11 (2016) 7-15. [14] N. Patil, D. Pal, H.K. Rafi, K. Zeng, A. Moreland, A. Hicks, D. Beeler, B. Stucker, A Generalized Feed Forward Dynamic Adaptive Mesh Refinement and Derefinement Finite Element Framework for Metal Laser Sintering—Part I: Formulation and Algorithm Development, Journal of Manufacturing Science and Engineering 137(4) (2015) 041001. [15] D. Pal, N. Patil, K.H. Kutty, K. Zeng, A. Moreland, A. Hicks, D. Beeler, B. Stucker, A Generalized Feed-Forward Dynamic Adaptive Mesh Refinement and Derefinement FiniteElement Framework for Metal Laser Sintering—Part II: Nonlinear Thermal Simulations and Validations, Journal of Manufacturing Science and Engineering 138(6) (2016) 061003. [16] N. Keller, V. Ploshikhin, New method for fast predictions of residual stress and distortion of AM parts, Solid Freeform Fabrication Symposium, Austin, Texas, 2014, pp. 1229-1237. [17] S.A. Khairallah, A.T. Anderson, A. Rubenchik, W.E. King, Laser powder-bed fusion additive manufacturing: Physics of complex melt flow and formation mechanisms of pores, spatter, and denudation zones, Acta Materialia 108 (2016) 36-45. [18] M.J. Matthews, G. Guss, S.A. Khairallah, A.M. Rubenchik, P.J. Depond, W.E. King, Denudation of metal powder layers in laser powder bed fusion processes, Acta Materialia 114 (2016) 33-42. [19] A.A. Martin, N.P. Calta, S.A. Khairallah, J. Wang, P.J. Depond, A.Y. Fong, V. Thampy, G.M. Guss, A.M. Kiss, K.H. Stone, Dynamics of pore formation during laser powder bed fusion additive manufacturing, Nature communications 10(1) (2019) 1987. [20] R. Shi, S.A. Khairallah, T.T. Roehling, T.W. Heo, J.T. McKeown, M.J. Matthews, Microstructural control in metal laser powder bed fusion additive manufacturing using laser beam shaping strategy, Acta Materialia (2019). [21] S.A. Khairallah, A.A. Martin, J.R. Lee, G. Guss, N.P. Calta, J.A. Hammons, M.H. Nielsen, K. Chaput, E. Schwalbach, M.N. Shah, Controlling interdependent meso-nanosecond dynamics and defect generation in metal 3D printing, Science 368(6491) (2020) 660-665. [22] W. Yan, W. Ge, Y. Qian, S. Lin, B. Zhou, W.K. Liu, F. Lin, G.J. Wagner, Multi-physics modeling of single/multiple-track defect mechanisms in electron beam selective melting, Acta Materialia 134 (2017) 324-333. [23] S. Shrestha, Y. Kevin Chou, A Numerical Study on the Keyhole Formation During Laser Powder Bed Fusion Process, Journal of Manufacturing Science and Engineering 141(10) (2019). [24] S. Shrestha, B. Cheng, K. Chou, An Investigation into Melt Pool Effective Thermal Conductivity for Thermal Modeling of Powder-Bed Electron Beam Additive Manufacturing. [25] D. Rosenthal, Mathematical theory of heat distribution during welding and cutting, Welding journal 20 (1941) 220-234. [26] P. Promoppatum, S.-C. Yao, P.C. Pistorius, A.D. Rollett, A comprehensive comparison of the analytical and numerical prediction of the thermal history and solidification microstructure of Inconel 718 products made by laser powder-bed fusion, Engineering 3(5) (2017) 685-694. [27] M. Tang, P.C. Pistorius, J.L. Beuth, Prediction of lack-of-fusion porosity for powder bed fusion, Additive Manufacturing 14 (2017) 39-48. [28] T. Moran, P. Li, D. Warner, N. Phan, Utility of superposition-based finite element approach for part-scale thermal simulation in additive manufacturing, Additive Manufacturing 21 (2018) 215-219. [29] Y. Yang, M. Knol, F. van Keulen, C. Ayas, A semi-analytical thermal modelling approach for selective laser melting, Additive Manufacturing 21 (2018) 284-297. [30] B. Cheng, S. Shrestha, K. Chou, Stress and deformation evaluations of scanning strategy effect in selective laser melting, Additive Manufacturing 12 (2016) 240-251. [31] L.H. Ahmed Hussein, Chunze Yan, Richard Everson, Finite element simulation of the temperature and stress fields in single layers built without-support in selective laser melting, Materials and Design 52 (2013) 638-647. [32] H. Peng, D.B. Go, R. Billo, S. Gong, M.R. Shankar, B.A. Gatrell, J. Budzinski, P. Ostiguy, R. Attardo, C. Tomonto, Part-scale model for fast prediction of thermal distortion in DMLS additive manufacturing; Part 2: a quasi-static thermo-mechanical model, Austin, Texas (2016). [33] M.F. Zaeh, G. Branner, Investigations on residual stresses and deformations in selective laser melting, Production Engineering 4(1) (2010) 35-45. [34] C. Li, C. Fu, Y. Guo, F. Fang, A multiscale modeling approach for fast prediction of part distortion in selective laser melting, Journal of Materials Processing Technology 229 (2016) 703- 712. [35] C. Li, Z. Liu, X. Fang, Y. Guo, On the Simulation Scalability of Predicting Residual Stress and Distortion in Selective Laser Melting, Journal of Manufacturing Science and Engineering 140(4) (2018) 041013. [36] S. Afazov, W.A. Denmark, B.L. Toralles, A. Holloway, A. Yaghi, Distortion Prediction and Compensation in Selective Laser Melting, Additive Manufacturing 17 (2017) 15-22. [37] Y. Lee, W. Zhang, Modeling of heat transfer, fluid flow and solidification microstructure of nickel-base superalloy fabricated by laser powder bed fusion, Additive Manufacturing 12 (2016) 178-188. [38] L. Scime, J. Beuth, A multi-scale convolutional neural network for autonomous anomaly detection and classification in a laser powder bed fusion additive manufacturing process, Additive Manufacturing 24 (2018) 273-286. [39] L. Scime, J. Beuth, Using machine learning to identify in-situ melt pool signatures indicative of flaw formation in a laser powder bed fusion additive manufacturing process, Additive Manufacturing 25 (2019) 151-165. [40] X. Xie, J. Bennett, S. Saha, Y. Lu, J. Cao, W.K. Liu, Z. Gan, Mechanistic data-driven prediction of as-built mechanical properties in metal additive manufacturing, npj Computational Materials 7(1) (2021) 1-12. [41] C. Wang, X. Tan, S. Tor, C. Lim, Machine learning in additive manufacturing: State-of-theart and perspectives, Additive Manufacturing (2020) 101538. [42] J. Li, R. Jin, Z.Y. Hang, Integration of physically-based and data-driven approaches for thermal field prediction in additive manufacturing, Materials & Design 139 (2018) 473-485. [43] M. Mozaffar, A. Paul, R. Al-Bahrani, S. Wolff, A. Choudhary, A. Agrawal, K. Ehmann, J. Cao, Data-driven prediction of the high-dimensional thermal history in directed energy deposition processes via recurrent neural networks, Manufacturing letters 18 (2018) 35-39. [44] A. Paul, M. Mozaffar, Z. Yang, W.-k. Liao, A. Choudhary, J. Cao, A. Agrawal, A real-time iterative machine learning approach for temperature profile prediction in additive manufacturing processes, 2019 IEEE International Conference on Data Science and Advanced Analytics (DSAA), IEEE, 2019, pp. 541-550. [45] S. Clijsters, T. Craeghs, J.-P. Kruth, A priori process parameter adjustment for SLM process optimization, Innovative developments on virtual and physical prototyping, Taylor & Francis Group., 2012, pp. 553-560. [46] R. Mertens, S. Clijsters, K. Kempen, J.-P. Kruth, Optimization of scan strategies in selective laser melting of aluminum parts with downfacing areas, Journal of Manufacturing Science and Engineering 136(6) (2014) 061012. [47] J.-P. Kruth, J. Deckers, E. Yasa, R. Wauthlé, Assessing and comparing influencing factors of residual stresses in selective laser melting using a novel analysis method, Proceedings of the institution of mechanical engineers, Part B: Journal of Engineering Manufacture 226(6) (2012) 980-991. [48] Y. Lu, S. Wu, Y. Gan, T. Huang, C. Yang, L. Junjie, J. Lin, Study on the microstructure, mechanical property and residual stress of SLM Inconel-718 alloy manufactured by differing island scanning strategy, Optics & Laser Technology 75 (2015) 197-206. [49] E. Foroozmehr, R. Kovacevic, Effect of path planning on the laser powder deposition process: thermal and structural evaluation, The International Journal of Advanced Manufacturing Technology 51(5-8) (2010) 659-669. [50] L.H. Ahmed Hussein, Chunze Yan, Richard Everson, Finite element simulation of the temperature and stress fields in single layers built without-support in selective laser melting, Materials and Design (2013). [51] J.-P. Kruth, M. Badrossamay, E. Yasa, J. Deckers, L. Thijs, J. Van Humbeeck, Part and material properties in selective laser melting of metals, Proceedings of the 16th international symposium on electromachining, 2010, pp. 1-12. [52] L. Thijs, K. Kempen, J.-P. Kruth, J. Van Humbeeck, Fine-structured aluminium products with controllable texture by selective laser melting of pre-alloyed AlSi10Mg powder, Acta Materialia 61(5) (2013) 1809-1819. [53] D. Ding, Z.S. Pan, D. Cuiuri, H. Li, A tool-path generation strategy for wire and arc additive manufacturing, The international journal of advanced manufacturing technology 73(1-4) (2014) 173-183. [54] B.E. Carroll, T.A. Palmer, A.M. Beese, Anisotropic tensile behavior of Ti–6Al–4V components fabricated with directed energy deposition additive manufacturing, Acta Materialia 87 (2015) 309-320. [55] D. Ding, Z. Pan, D. Cuiuri, H. Li, A practical path planning methodology for wire and arc additive manufacturing of thin-walled structures, Robotics and Computer-Integrated Manufacturing 34 (2015) 8-19. [56] D. Ding, Z. Pan, D. Cuiuri, H. Li, S. van Duin, N. Larkin, Bead modelling and implementation of adaptive MAT path in wire and arc additive manufacturing, Robotics and Computer-Integrated Manufacturing 39 (2016) 32-42. [57] R. Ponche, O. Kerbrat, P. Mognol, J.-Y. Hascoet, A novel methodology of design for Additive Manufacturing applied to Additive Laser Manufacturing process, Robotics and ComputerIntegrated Manufacturing 30(4) (2014) 389-398. [58] D.E. Smith, R. Hoglund, Continuous fiber angle topology optimization for polymer fused fillament fabrication, Annu. Int. Solid Free. Fabr. Symp. Austin, TX, 2016. [59] J. Liu, J. Liu, H. Yu, H. Yu, Concurrent deposition path planning and structural topology optimization for additive manufacturing, Rapid Prototyping Journal 23(5) (2017) 930-942. [60] Q. Xia, T. Shi, Optimization of composite structures with continuous spatial variation of fiber angle through Shepard interpolation, Composite Structures 182 (2017) 273-282. [61] C. Kiyono, E. Silva, J. Reddy, A novel fiber optimization method based on normal distribution function with continuously varying fiber path, Composite Structures 160 (2017) 503-515. [62] C.J. Brampton, K.C. Wu, H.A. Kim, New optimization method for steered fiber composites using the level set method, Structural and Multidisciplinary Optimization 52(3) (2015) 493-505. [63] J. Liu, A.C. To, Deposition path planning-integrated structural topology optimization for 3D additive manufacturing subject to self-support constraint, Computer-Aided Design 91 (2017) 27- 45. [64] H. Shen, J. Fu, Z. Chen, Y. Fan, Generation of offset surface for tool path in NC machining through level set methods, The International Journal of Advanced Manufacturing Technology 46(9-12) (2010) 1043-1047. [65] C. Zhuang, Z. Xiong, H. Ding, High speed machining tool path generation for pockets using level sets, International Journal of Production Research 48(19) (2010) 5749-5766. [66] K.C. Mills, Recommended values of thermophysical properties for selected commercial alloys, Woodhead Publishing2002. [67] S.S. Sih, J.W. Barlow, The prediction of the emissivity and thermal conductivity of powder beds, Particulate Science and Technology 22(4) (2004) 427-440. [68] L. Dong, A. Makradi, S. Ahzi, Y. Remond, Three-dimensional transient finite element analysis of the selective laser sintering process, Journal of materials processing technology 209(2) (2009) 700-706. [69] J.J. Beaman, J.W. Barlow, D.L. Bourell, R.H. Crawford, H.L. Marcus, K.P. McAlea, Solid freeform fabrication: a new direction in manufacturing, Kluwer Academic Publishers, Norwell, MA 2061 (1997) 25-49. [70] G. Bugeda Miguel Cervera, G. Lombera, Numerical prediction of temperature and density distributions in selective laser sintering processes, Rapid Prototyping Journal 5(1) (1999) 21-26. [71] T. Mukherjee, W. Zhang, T. DebRoy, An improved prediction of residual stresses and distortion in additive manufacturing, Computational Materials Science 126 (2017) 360-372. [72] A.J. Dunbar, E.R. Denlinger, M.F. Gouge, P. Michaleris, Experimental validation of finite element modeling for laser powderbed fusion deformation, Additive Manufacturing 12 (2016) 108-120. [73] J. Goldak, A. Chakravarti, M. Bibby, A new finite element model for welding heat sources, Metallurgical and Materials Transactions B 15(2) (1984) 299-305. [74] J. Liu, Q. Chen, Y. Zhao, W. Xiong, A. To, Quantitative Texture Prediction of Epitaxial Columnar Grains in Alloy 718 Processed by Additive Manufacturing, Proceedings of the 9th International Symposium on Superalloy 718 & Derivatives: Energy, Aerospace, and Industrial Applications, Springer, 2018, pp. 749-755. [75] J. Irwin, P. Michaleris, A line heat input model for additive manufacturing, Journal of Manufacturing Science and Engineering 138(11) (2016) 111004. [76] M. Gouge, J. Heigel, P. Michaleris, T. Palmer, Modeling forced convection in the thermal simulation of laser cladding processes, International Journal of Advanced Manufacturing Technology 79 (2015). [77] J. Heigel, P. Michaleris, E. Reutzel, Thermo-mechanical model development and validation of directed energy deposition additive manufacturing of Ti–6Al–4V, Additive manufacturing 5 (2015) 9-19. [78] E.R. Denlinger, J.C. Heigel, P. Michaleris, Residual stress and distortion modeling of electron beam direct manufacturing Ti-6Al-4V, Proceedings of the Institution of Mechanical Engineers, Part B: Journal of Engineering Manufacture 229(10) (2015) 1803-1813. [79] X. Liang, Q. Chen, L. Cheng, Q. Yang, A. To, A modified inherent strain method for fast prediction of residual deformation in additive manufacturing of metal parts, 2017 Solid Freeform Fabrication Symposium Proceedings, Austin, Texas, 2017. [80] X. Liang, L. Cheng, Q. Chen, Q. Yang, A. To, A Modified Method for Estimating Inherent Strains from Detailed Process Simulation for Fast Residual Distortion Prediction of Single-Walled Structures Fabricated by Directed Energy Deposition, Additive Manufacturing 23 (2018) 471-486. [81] L. Sochalski-Kolbus, E.A. Payzant, P.A. Cornwell, T.R. Watkins, S.S. Babu, R.R. Dehoff, M. Lorenz, O. Ovchinnikova, C. Duty, Comparison of residual stresses in Inconel 718 simple parts made by electron beam melting and direct laser metal sintering, Metallurgical and Materials Transactions A 46(3) (2015) 1419-1432. [82] P. Mercelis, J.-P. Kruth, Residual stresses in selective laser sintering and selective laser melting, Rapid Prototyping Journal 12(5) (2006) 254-265. [83] N. Hodge, R. Ferencz, J. Solberg, Implementation of a thermomechanical model for the simulation of selective laser melting, Computational Mechanics 54(1) (2014) 33-51. [84] A.S. Wu, D.W. Brown, M. Kumar, G.F. Gallegos, W.E. King, An experimental investigation into additive manufacturing-induced residual stresses in 316L stainless steel, Metallurgical and Materials Transactions A 45(13) (2014) 6260-6270. [85] C. Li, J. liu, Y. Guo, Efficient predictive model of part distortion and residual stress in selective laser melting, Solid Freeform Fabrication 2016, 2017. [86] Y. Zhao, Y. Koizumi, K. Aoyagi, D. Wei, K. Yamanaka, A. Chiba, Molten pool behavior and effect of fluid flow on solidification conditions in selective electron beam melting (SEBM) of a biomedical Co-Cr-Mo alloy, Additive Manufacturing 26 (2019) 202-214. [87] J.-H. Cho, S.-J. Na, Implementation of real-time multiple reflection and Fresnel absorption of laser beam in keyhole, Journal of Physics D: Applied Physics 39(24) (2006) 5372. [88] Q. Guo, C. Zhao, M. Qu, L. Xiong, L.I. Escano, S.M.H. Hojjatzadeh, N.D. Parab, K. Fezzaa, W. Everhart, T. Sun, In-situ characterization and quantification of melt pool variation under constant input energy density in laser powder-bed fusion additive manufacturing process, Additive Manufacturing (2019). [89] E. Assuncao, S. Williams, D. Yapp, Interaction time and beam diameter effects on the conduction mode limit, Optics and Lasers in Engineering 50(6) (2012) 823-828. [90] R. Cunningham, C. Zhao, N. Parab, C. Kantzos, J. Pauza, K. Fezzaa, T. Sun, A.D. Rollett, Keyhole threshold and morphology in laser melting revealed by ultrahigh-speed x-ray imaging, Science 363(6429) (2019) 849-852. [91] W. Tan, N.S. Bailey, Y.C. Shin, Investigation of keyhole plume and molten pool based on a three-dimensional dynamic model with sharp interface formulation, Journal of Physics D: Applied Physics 46(5) (2013) 055501. [92] W. Tan, Y.C. Shin, Analysis of multi-phase interaction and its effects on keyhole dynamics with a multi-physics numerical model, Journal of Physics D: Applied Physics 47(34) (2014) 345501. [93] R. Fabbro, K. Chouf, Keyhole modeling during laser welding, Journal of applied Physics 87(9) (2000) 4075-4083. [94] Q. Guo, C. Zhao, M. Qu, L. Xiong, S.M.H. Hojjatzadeh, L.I. Escano, N.D. Parab, K. Fezzaa, T. Sun, L. Chen, In-situ full-field mapping of melt flow dynamics in laser metal additive manufacturing, Additive Manufacturing 31 (2020) 100939. [95] Y. Ueda, K. Fukuda, K. Nakacho, S. Endo, A new measuring method of residual stresses with the aid of finite element method and reliability of estimated values, Journal of the Society of Naval Architects of Japan 1975(138) (1975) 499-507. [96] M.R. Hill, D.V. Nelson, The inherent strain method for residual stress determination and its application to a long welded joint, ASME-PUBLICATIONS-PVP 318 (1995) 343-352. [97] H. Murakawa, Y. Luo, Y. Ueda, Prediction of welding deformation and residual stress by elastic FEM based on inherent strain, Journal of the society of Naval Architects of Japan 1996(180) (1996) 739-751. [98] M. Yuan, Y. Ueda, Prediction of residual stresses in welded T-and I-joints using inherent strains, Journal of Engineering Materials and Technology, Transactions of the ASME 118(2) (1996) 229-234. [99] L. Zhang, P. Michaleris, P. Marugabandhu, Evaluation of applied plastic strain methods for welding distortion prediction, Journal of Manufacturing Science and Engineering 129(6) (2007) 1000-1010. [100] M. Bugatti, Q. Semeraro, Limitations of the Inherent Strain Method in Simulating Powder Bed Fusion Processes, Additive Manufacturing 23 (2018) 329-346. [101] L. Cheng, X. Liang, J. Bai, Q. Chen, J. Lemon, A. To, On Utilizing Topology Optimization to Design Support Structure to Prevent Residual Stress Induced Build Failure in Laser Powder Bed Metal Additive Manufacturing, Additive Manufacturing (2019). [102] Q. Chen, X. Liang, D. Hayduke, J. Liu, L. Cheng, J. Oskin, R. Whitmore, A.C. To, An inherent strain based multiscale modeling framework for simulating part-scale residual deformation for direct metal laser sintering, Additive Manufacturing 28 (2019) 406-418. [103] S. Osher, J.A. Sethian, Fronts propagating with curvature-dependent speed: algorithms based on Hamilton-Jacobi formulations, Journal of computational physics 79(1) (1988) 12-49. [104] M.Y. Wang, X. Wang, D. Guo, A level set method for structural topology optimization, Computer methods in applied mechanics and engineering 192(1) (2003) 227-246. [105] G. Allaire, F. Jouve, A.-M. Toader, Structural optimization using sensitivity analysis and a level-set method, Journal of computational physics 194(1) (2004) 363-393. [106] Y. Wang, Z. Luo, Z. Kang, N. Zhang, A multi-material level set-based topology and shape optimization method, Computer Methods in Applied Mechanics and Engineering 283 (2015) 1570-1586. [107] P. Dunning, C. Brampton, H. Kim, Simultaneous optimisation of structural topology and material grading using level set method, Materials Science and Technology 31(8) (2015) 884-894. [108] P. Liu, Y. Luo, Z. Kang, Multi-material topology optimization considering interface behavior via XFEM and level set method, Computer methods in applied mechanics and engineering 308 (2016) 113-133. [109] J. Liu, Q. Chen, Y. Zheng, R. Ahmad, J. Tang, Y. Ma, Level set-based heterogeneous object modeling and optimization, Computer-Aided Design (2019). [110] J. Liu, Q. Chen, X. Liang, A.C. To, Manufacturing cost constrained topology optimization for additive manufacturing, Frontiers of Mechanical Engineering 14(2) (2019) 213-221. [111] Z. Kang, Y. Wang, Integrated topology optimization with embedded movable holes based on combined description by material density and level sets, Computer methods in applied mechanics and engineering 255 (2013) 1-13. [112] P.D. Dunning, H. Alicia Kim, A new hole insertion method for level set based structural topology optimization, International Journal for Numerical Methods in Engineering 93(1) (2013) 118-134. [113] J.A. Sethian, A fast marching level set method for monotonically advancing fronts, Proceedings of the National Academy of Sciences 93(4) (1996) 1591-1595. [114] J.A. Sethian, Level set methods and fast marching methods: evolving interfaces in computational geometry, fluid mechanics, computer vision, and materials science, Cambridge university press1999. [115] C. Le, J. Norato, T. Bruns, C. Ha, D. Tortorelli, Stress-based topology optimization for continua, Structural and Multidisciplinary Optimization 41(4) (2010) 605-620. [116] A. Takezawa, G.H. Yoon, S.H. Jeong, M. Kobashi, M. Kitamura, Structural topology optimization with strength and heat conduction constraints, Computer Methods in Applied Mechanics and Engineering 276 (2014) 341-361. [117] S. Hochreiter, J. Schmidhuber, Long short-term memory, Neural computation 9(8) (1997) 1735-1780. [118] A. Krizhevsky, I. Sutskever, G.E. Hinton, Imagenet classification with deep convolutional neural networks, Advances in neural information processing systems 25 (2012) 1097-1105. [119] K. Simonyan, A. Zisserman, Very deep convolutional networks for large-scale image recognition, arXiv preprint arXiv:1409.1556 (2014). [120] K. He, X. Zhang, S. Ren, J. Sun, Deep residual learning for image recognition, Proceedings of the IEEE conference on computer vision and pattern recognition, 2016, pp. 770-778. [121] O. Russakovsky, J. Deng, H. Su, J. Krause, S. Satheesh, S. Ma, Z. Huang, A. Karpathy, A. Khosla, M. Bernstein, Imagenet large scale visual recognition challenge, International journal of computer vision 115(3) (2015) 211-252. [122] S. Ren, K. He, R. Girshick, J. Sun, Faster r-cnn: Towards real-time object detection with region proposal networks, Advances in neural information processing systems 28 (2015) 91-99. [123] E.J. Schwalbach, S.P. Donegan, M.G. Chapman, K.J. Chaput, M.A. Groeber, A discrete source model of powder bed fusion additive manufacturing thermal history, Additive Manufacturing 25 (2019) 485-498. [124] D.G. Duffy, Green’s functions with applications, Chapman and Hall/CRC2015. [125] J. Martínez-Frutos, D. Herrero-Pérez, Efficient matrix-free GPU implementation of fixed grid finite element analysis, Finite Elements in Analysis and Design 104 (2015) 61-71. [126] F. Dugast, P. Apostolou, A. Fernandez, W. Dong, Q. Chen, S. Strayer, R. Wicker, A.C. To, Part-scale thermal process modeling for laser powder bed fusion with matrix-free method and GPU computing, Additive Manufacturing 37 (2021) 101732. [127] A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A.N. Gomez, Ł. Kaiser, I. Polosukhin, Attention is all you need, Advances in neural information processing systems, 2017, pp. 5998-6008. [128] J. Devlin, M.-W. Chang, K. Lee, K. Toutanova, Bert: Pre-training of deep bidirectional transformers for language understanding, arXiv preprint arXiv:1810.04805 (2018).
Keyhole Formation by Laser Drilling in Laser Powder Bed Fusion of Ti6Al4V Biomedical Alloy: Mesoscopic Computational Fluid Dynamics Simulation versus Mathematical Modelling Using Empirical Validation
Asif Ur Rehman 1,2,3,* ,† , Muhammad Arif Mahmood 4,* ,† , Fatih Pitir 1 , Metin Uymaz Salamci 2,3 , Andrei C. Popescu 4 and Ion N. Mihailescu 4
Abstract
LPBF(Laser Powder Bed fusion) 공정에서 작동 조건은 열 분포를 기반으로 레이저 유도 키홀 영역을 결정하는 데 필수적입니다. 얕은 구멍과 깊은 구멍으로 분류되는 이러한 영역은 LPBF 프로세스에서 확률과 결함 형성 강도를 제어합니다.
LPBF 프로세스의 핵심 구멍을 연구하고 제어하기 위해 수학적 및 CFD(전산 유체 역학) 모델이 제공됩니다. CFD의 경우 이산 요소 모델링 기법을 사용한 유체 체적 방법이 사용되었으며, 분말 베드 보이드 및 표면에 의한 레이저 빔 흡수를 포함하여 수학적 모델이 개발되었습니다.
동적 용융 풀 거동을 자세히 살펴봅니다. 실험적, CFD 시뮬레이션 및 분석적 컴퓨팅 결과 간에 정량적 비교가 수행되어 좋은 일치를 얻습니다.
LPBF에서 레이저 조사 영역 주변의 온도는 높은 내열성과 분말 입자 사이의 공기로 인해 분말층 주변에 비해 급격히 상승하여 레이저 횡방향 열파의 이동이 느려집니다. LPBF에서 키홀은 에너지 밀도에 의해 제어되는 얕고 깊은 키홀 모드로 분류될 수 있습니다. 에너지 밀도를 높이면 얕은 키홀 구멍 모드가 깊은 키홀 구멍 모드로 바뀝니다.
깊은 키홀 구멍의 에너지 밀도는 다중 반사와 키홀 구멍 내의 2차 반사 빔의 집중으로 인해 더 높아져 재료가 빠르게 기화됩니다.
깊은 키홀 구멍 모드에서는 온도 분포가 높기 때문에 액체 재료가 기화 온도에 가까우므로 얕은 키홀 구멍보다 구멍이 형성될 확률이 훨씬 높습니다. 온도가 급격히 상승하면 재료 밀도가 급격히 떨어지므로 비열과 융해 잠열로 인해 유체 부피가 증가합니다.
그 대가로 표면 장력을 낮추고 용융 풀 균일성에 영향을 미칩니다.
In the laser powder bed fusion (LPBF) process, the operating conditions are essential in determining laser-induced keyhole regimes based on the thermal distribution. These regimes, classified into shallow and deep keyholes, control the probability and defects formation intensity in the LPBF process. To study and control the keyhole in the LPBF process, mathematical and computational fluid dynamics (CFD) models are presented. For CFD, the volume of fluid method with the discrete element modeling technique was used, while a mathematical model was developed by including the laser beam absorption by the powder bed voids and surface. The dynamic melt pool behavior is explored in detail. Quantitative comparisons are made among experimental, CFD simulation and analytical computing results leading to a good correspondence. In LPBF, the temperature around the laser irradiation zone rises rapidly compared to the surroundings in the powder layer due to the high thermal resistance and the air between the powder particles, resulting in a slow travel of laser transverse heat waves. In LPBF, the keyhole can be classified into shallow and deep keyhole mode, controlled by the energy density. Increasing the energy density, the shallow keyhole mode transforms into the deep keyhole mode. The energy density in a deep keyhole is higher due to the multiple reflections and concentrations of secondary reflected beams within the keyhole, causing the material to vaporize quickly. Due to an elevated temperature distribution in deep keyhole mode, the probability of pores forming is much higher than in a shallow keyhole as the liquid material is close to the vaporization temperature. When the temperature increases rapidly, the material density drops quickly, thus, raising the fluid volume due to the specific heat and fusion latent heat. In return, this lowers the surface tension and affects the melt pool uniformity.
Keywords: laser powder bed fusion; computational fluid dynamics; analytical modelling; shallow and deep keyhole modes; experimental correlation
References
Kok, Y.; Tan, X.P.; Wang, P.; Nai, M.L.S.; Loh, N.H.; Liu, E.; Tor, S.B. Anisotropy and heterogeneity of microstructure and mechanical properties in metal additive manufacturing: A critical review. Mater. Des. 2018, 139, 565–586. [CrossRef]
Ansari, P.; Salamci, M.U. On the selective laser melting based additive manufacturing of AlSi10Mg: The process parameter investigation through multiphysics simulation and experimental validation. J. Alloys Compd. 2022, 890, 161873. [CrossRef]
Guo, N.; Leu, M.C. Additive manufacturing: Technology, applications and research needs. Front. Mech. Eng. 2013, 8, 215–243. [CrossRef]
Mohsin Raza, M.; Lo, Y.L. Experimental investigation into microstructure, mechanical properties, and cracking mechanism of IN713LC processed by laser powder bed fusion. Mater. Sci. Eng. A 2021, 819, 141527. [CrossRef]
Dezfoli, A.R.A.; Lo, Y.L.; Raza, M.M. Prediction of Epitaxial Grain Growth in Single-Track Laser Melting of IN718 Using Integrated Finite Element and Cellular Automaton Approach. Materials 2021, 14, 5202. [CrossRef]
Tiwari, S.K.; Pande, S.; Agrawal, S.; Bobade, S.M. Selection of selective laser sintering materials for different applications. Rapid Prototyp. J. 2015, 21, 630–648. [CrossRef]
Liu, F.H. Synthesis of bioceramic scaffolds for bone tissue engineering by rapid prototyping technique. J. Sol-Gel Sci. Technol. 2012, 64, 704–710. [CrossRef]
Ur Rehman, A.; Sglavo, V.M. 3D printing of geopolymer-based concrete for building applications. Rapid Prototyp. J. 2020, 26, 1783–1788. [CrossRef]
Ur Rehman, A.; Sglavo, V.M. 3D printing of Portland cement-containing bodies. Rapid Prototyp. J. 2021. ahead of print. [CrossRef]
Popovich, A.; Sufiiarov, V. Metal Powder Additive Manufacturing. In New Trends in 3D Printing; InTech: Rijeka, Croatia, 2016.
Jia, T.; Zhang, Y.; Chen, J.K.; He, Y.L. Dynamic simulation of granular packing of fine cohesive particles with different size distributions. Powder Technol. 2012, 218, 76–85. [CrossRef]
Ansari, P.; Ur Rehman, A.; Pitir, F.; Veziroglu, S.; Mishra, Y.K.; Aktas, O.C.; Salamci, M.U. Selective Laser Melting of 316L Austenitic Stainless Steel: Detailed Process Understanding Using Multiphysics Simulation and Experimentation. Metals 2021, 11, 1076. [CrossRef]
Ur Rehman, A.; Tingting, L.; Liao, W. 4D Printing; Printing Ceramics from Metals with Selective Oxidation. Patent No. W0/2019/052128, 21 March 2019.
Ullah, A.; Wu, H.; Ur Rehman, A.; Zhu, Y.; Liu, T.; Zhang, K. Influence of laser parameters and Ti content on the surface morphology of L-PBF fabricated Titania. Rapid Prototyp. J. 2021, 27, 71–80. [CrossRef]
Ur Rehman, A. Additive Manufacturing of Ceramic Materials and Combinations with New Laser Strategies. Master’s Thesis, Nanjing University of Science and Technology, Nanjing, China, 2017.
Wong, K.V.; Hernandez, A. A Review of Additive Manufacturing. ISRN Mech. Eng. 2012, 2012, 1–10. [CrossRef]
Körner, C. Additive manufacturing of metallic components by selective electron beam melting—A review. Int. Mater. Rev. 2016, 61, 361–377. [CrossRef]
Fayazfar, H.; Salarian, M.; Rogalsky, A.; Sarker, D.; Russo, P.; Paserin, V.; Toyserkani, E. A critical review of powder-based additive manufacturing of ferrous alloys: Process parameters, microstructure and mechanical properties. Mater. Des. 2018, 144, 98–128. [CrossRef]
Everton, S.K.; Hirsch, M.; Stavroulakis, P.I.; Leach, R.K.; Clare, A.T. Review of in-situ process monitoring and in-situ metrology for metal additive manufacturing. Mater. Des. 2016, 95, 431–445. [CrossRef]
Sing, S.L.; An, J.; Yeong, W.Y.; Wiria, F.E. Laser and electron-beam powder-bed additive manufacturing of metallic implants: A review on processes, materials and designs. J. Orthop. Res. 2016, 34, 369–385. [CrossRef] [PubMed]
Olakanmi, E.O.; Cochrane, R.F.; Dalgarno, K.W. A review on selective laser sintering/melting (SLS/SLM) of aluminium alloy powders: Processing, microstructure, and properties. Prog. Mater. Sci. 2015, 74, 401–477. [CrossRef]
Mahmood, M.A.; Popescu, A.C.; Hapenciuc, C.L.; Ristoscu, C.; Visan, A.I.; Oane, M.; Mihailescu, I.N. Estimation of clad geometry and corresponding residual stress distribution in laser melting deposition: Analytical modeling and experimental correlations. Int. J. Adv. Manuf. Technol. 2020, 111, 77–91. [CrossRef]
Mahmood, M.A.; Popescu, A.C.; Oane, M.; Ristoscu, C.; Chioibasu, D.; Mihai, S.; Mihailescu, I.N. Three-jet powder flow and laser–powder interaction in laser melting deposition: Modelling versus experimental correlations. Metals 2020, 10, 1113. [CrossRef]
King, W.; Anderson, A.T.; Ferencz, R.M.; Hodge, N.E.; Kamath, C.; Khairallah, S.A. Overview of modelling and simulation of metal powder bed fusion process at Lawrence Livermore National Laboratory. Mater. Sci. Technol. 2015, 31, 957–968. [CrossRef]
Gong, H.; Rafi, K.; Gu, H.; Starr, T.; Stucker, B. Analysis of defect generation in Ti-6Al-4V parts made using powder bed fusion additive manufacturing processes. Addit. Manuf. 2014, 1, 87–98. [CrossRef]
Frazier, W.E. Metal additive manufacturing: A review. J. Mater. Eng. Perform. 2014, 23, 1917–1928. [CrossRef]
Panwisawas, C.; Qiu, C.L.; Sovani, Y.; Brooks, J.W.; Attallah, M.M.; Basoalto, H.C. On the role of thermal fluid dynamics into the evolution of porosity during selective laser melting. Scr. Mater. 2015, 105, 14–17. [CrossRef]
Qian, Y.; Yan, W.; Lin, F. Parametric study and surface morphology analysis of electron beam selective melting. Rapid Prototyp. J. 2018, 24, 1586–1598. [CrossRef]
Panwisawas, C.; Perumal, B.; Ward, R.M.; Turner, N.; Turner, R.P.; Brooks, J.W.; Basoalto, H.C. Keyhole formation and thermal fluid flow-induced porosity during laser fusion welding in titanium alloys: Experimental and modelling. Acta Mater. 2017, 126, 251–263. [CrossRef]
Panwisawas, C.; Sovani, Y.; Turner, R.P.; Brooks, J.W.; Basoalto, H.C.; Choquet, I. Modelling of thermal fluid dynamics for fusion welding. J. Mater. Process. Technol. 2018, 252, 176–182. [CrossRef]
Martin, A.A.; Calta, N.P.; Hammons, J.A.; Khairallah, S.A.; Nielsen, M.H.; Shuttlesworth, R.M.; Sinclair, N.; Matthews, M.J.; Jeffries, J.R.; Willey, T.M.; et al. Ultrafast dynamics of laser-metal interactions in additive manufacturing alloys captured by in situ X-ray imaging. Mater. Today Adv. 2019, 1, 100002. [CrossRef]
Cunningham, R.; Zhao, C.; Parab, N.; Kantzos, C.; Pauza, J.; Fezzaa, K.; Sun, T.; Rollett, A.D. Keyhole threshold and morphology in laser melting revealed by ultrahigh-speed x-ray imaging. Science 2019, 363, 849–852. [CrossRef] [PubMed]
Tang, C.; Tan, J.L.; Wong, C.H. A numerical investigation on the physical mechanisms of single track defects in selective laser melting. Int. J. Heat Mass Transf. 2018, 126, 957–968. [CrossRef]
Mirkoohi, E.; Ning, J.; Bocchini, P.; Fergani, O.; Chiang, K.-N.; Liang, S. Thermal Modeling of Temperature Distribution in Metal Additive Manufacturing Considering Effects of Build Layers, Latent Heat, and Temperature-Sensitivity of Material Properties. J. Manuf. Mater. Process. 2018, 2, 63. [CrossRef]
Oane, M.; Sporea, D. Temperature profiles modeling in IR optical components during high power laser irradiation. Infrared Phys. Technol. 2001, 42, 31–40. [CrossRef]
Cleary, P.W.; Sawley, M.L. DEM modelling of industrial granular flows: 3D case studies and the effect of particle shape on hopper discharge. Appl. Math. Model. 2002, 26, 89–111. [CrossRef]
Parteli, E.J.R.; Pöschel, T. Particle-based simulation of powder application in additive manufacturing. Powder Technol. 2016, 288, 96–102. [CrossRef]
Cao, L. Numerical simulation of the impact of laying powder on selective laser melting single-pass formation. Int. J. Heat Mass Transf. 2019, 141, 1036–1048. [CrossRef]
Tian, Y.; Yang, L.; Zhao, D.; Huang, Y.; Pan, J. Numerical analysis of powder bed generation and single track forming for selective laser melting of SS316L stainless steel. J. Manuf. Process. 2020, 58, 964–974. [CrossRef]
Lee, Y.S.; Zhang, W. Modeling of heat transfer, fluid flow and solidification microstructure of nickel-base superalloy fabricated by laser powder bed fusion. Addit. Manuf. 2016, 12, 178–188. [CrossRef]
Tang, M.; Pistorius, P.C.; Beuth, J.L. Prediction of lack-of-fusion porosity for powder bed fusion. Addit. Manuf. 2017, 14, 39–48. [CrossRef]
Promoppatum, P.; Yao, S.C.; Pistorius, P.C.; Rollett, A.D. A Comprehensive Comparison of the Analytical and Numerical Prediction of the Thermal History and Solidification Microstructure of Inconel 718 Products Made by Laser Powder-Bed Fusion. Engineering 2017, 3, 685–694. [CrossRef]
Rosenthal, D. Mathematical Theory of Heat Distribution During Welding and Cutting. Weld. J. 1941, 20, 220–234.
Chen, Q.; Zhao, Y.Y.; Strayer, S.; Zhao, Y.Y.; Aoyagi, K.; Koizumi, Y.; Chiba, A.; Xiong, W.; To, A.C. Elucidating the Effect of Preheating Temperature on Melt Pool Morphology Variation in Inconel 718 Laser Powder Bed Fusion via Simulation and Experiment. Available online: https://www.sciencedirect.com/science/article/pii/S2214860420310149#bb8 (accessed on 30 April 2021).
Ur Rehman, A.; Pitir, F.; Salamci, M.U. Laser Powder Bed Fusion (LPBF) of In718 and the Impact of Pre-Heating at 500 and 1000 ◦C: Operando Study. Materials 2021, 14, 6683. [CrossRef] [PubMed]
Ur Rehman, A.; Pitir, F.; Salamci, M.U. Full-Field Mapping and Flow Quantification of Melt Pool Dynamics in Laser Powder Bed Fusion of SS316L. Materials 2021, 14, 6264. [CrossRef] [PubMed]
Gong, H.; Gu, H.; Zeng, K.; Dilip, J.J.S.; Pal, D.; Stucker, B.; Christiansen, D.; Beuth, J.; Lewandowski, J.J. Melt Pool Characterization for Selective Laser Melting of Ti-6Al-4V Pre-alloyed Powder. In Proceedings of the International Solid Freeform Fabrication Symposium, Austin, TX, USA, 10–12 August 2014; 2014; pp. 256–267.
Song, B.; Dong, S.; Liao, H.; Coddet, C. Process parameter selection for selective laser melting of Ti6Al4V based on temperature distribution simulation and experimental sintering. Int. J. Adv. Manuf. Technol. 2012, 61, 967–974. [CrossRef]
적층 제조(AM) 기술은 복잡한 형상의 3D 부품을 쉽게 만들고 미세 구조 제어를 통해 재료 특성을 크게 제어할 수 있기 때문에 많은 관심을 받았습니다. PBF(Powderbed fusion) 방식의 AM 공정에서는 금속 분말을 레이저나 전자빔으로 녹이고 응고시키는 과정을 반복하여 3D 부품을 제작합니다.
일반적으로 응고 미세구조는 Hunt[Mater. 과학. 영어 65, 75(1984)]. 그러나 CET 이론이 일반 316L 스테인리스강에서도 높은 G와 R로 인해 PBF형 AM 공정에 적용될 수 있을지는 불확실하다.
본 연구에서는 미세구조와 응고 조건 간의 관계를 밝히기 위해 전자빔 조사에 의해 유도된 316L 강의 응고 미세구조를 분석하고 CtFD(Computational Thermal-Fluid Dynamics) 방법을 사용하여 고체/액체 계면에서의 응고 조건을 평가했습니다.
CET 이론과 반대로 높은 G 조건에서 등축 결정립이 종종 형성되는 것으로 밝혀졌다. CtFD 시뮬레이션은 약 400 mm s-1의 속도까지 유체 흐름이 있음을 보여 주며 수상 돌기의 파편 및 이동의 영향으로 등축 결정립이 형성됨을 시사했습니다.
Additive manufacturing(AM)technologies have attracted much attention because it enables us to build 3D parts with complicated geometry easily and control material properties significantly via the control of microstructures. In the powderbed fusion(PBF)type AM process, 3D parts are fabricated by repeating a process of melting and solidifying metal powders by laser or electron beams. In general, the solidification microstructures can be predicted from solidification conditions defined by the combination of temperature gradient G and solidification rate R on the basis of columnar-equiaxed transition(CET)theory proposed by Hunt [Mater. Sci. Eng. 65, 75(1984)]. However, it is unclear whether the CET theory can be applied to the PBF type AM process because of the high G and R, even for general 316L stainless steel. In this study, to reveal relationships between microstructures and solidification conditions, we have analyzed solidification microstructures of 316L steel induced by electronbeam irradiation and evaluated solidification conditions at the solid/liquid interface using a computational thermal-fluid dynamics (CtFD)method. It was found that equiaxed grains were often formed under high G conditions contrary to the CET theory. CtFD simulation revealed that there is a fluid flow up to a velocity of about 400 mm s-1, and suggested that equiaxed grains are formed owing to the effect of fragmentations and migrations of dendrites.
Keywords
Additive Manufacturing, 316L Stainless Steel, Powder Bed Fusion, Electron Beam Melting, Computational Thermal Fluid Dynamics Simulation
References
1) M.C. Sow, T. De Terris, O. Castelnau, Z. Hamouche, F. Coste, R. Fabbro and P. Peyre: “Influence of beam diameter on Laser Powder
Bed Fusion(L-PBF)process”, Addit. Manuf. 36(2020), 101532. 2) J.C. Simmons, X. Chen, A. Azizi, M.A. Daeumer, P.Y. Zavalij, G. Zhou and S.N. Schiffres: “Influence of processing and microstructure on the local and bulk thermal conductivity of selective laser melted 316L stainless steel”, Addit. Manuf. 32(2020), 100996. 3) S. Dryepondt, P. Nandwana, P. Fernandez-Zelaia and F. List: “Microstructure and High Temperature Tensile properties of 316L Fabricated by Laser Powder-Bed Fusion”, Addit. Manuf. 37(2020), 101723. 4) S.H. Sun, T. Ishimoto, K. Hagihara, Y. Tsutsumi, T. Hanawa and T. Nakano: “Excellent mechanical and corrosion properties of austenitic stainless steel with a unique crystallographic lamellar microstructure via selective laser melting”, Scr. Mater. 159(2019), 89-93. 5) T. Ishimoto, S. Wu, Y. Ito, S.H. Sun, H. Amano and T. Nakano: “Crystallographic orientation control of 316L austenitic stainless steel via selective laser melting”, ISIJ Int. 60(2020), 1758-1764. 6) T. Ishimoto, K. Hagihara, K. Hisamoto, S.H. Sun and T. Nakano: “Crystallographic texture control of beta-type Ti-15Mo-5Zr3Al alloy by selective laser melting for the development of novel implants with a biocompatible low Young’s modulus”, Scr. Mater. 132(2017), 34-38. 7) X. Ding, Y. Koizumi, D. Wei and A. Chiba: “Effect of process parameters on melt pool geometry and microstructure development for electron beam melting of IN718: A systematic single bead analysis study”, Addit. Manuf. 26(2019), 215-226. 8) K. Karayagiz, L. Johnson, R. Seede, V. Attari, B. Zhang, X. Huang, S. Ghosh, T. Duong, I. Karaman, A. Elwany and R. Arróyave: “Finite interface dissipation phase field modeling of Ni-Nb under additive manufacturing conditions”, Acta Mater. 185(2020), 320-339. 9) M.M. Kirka, Y. Lee, D.A. Greeley, A. Okello, M.J. Goin, M.T. Pearce and R.R. Dehoff: “Strategy for Texture Management in Metals Additive Manufacturing”, JOM, 69(2017), 523-531. 10) S.S. Babu, N. Raghavan, J. Raplee, S.J. Foster, C. Frederick, M. Haines, R. Dinwiddie, M.K. Kirka, A. Plotkowski, Y. Lee and R.R. Dehoff: “Additive Manufacturing of Nickel Superalloys: Opportunities for Innovation and Challenges Related to Qualification”, Metall. Mater. Trans. A. 49(2018), 3764-3780. 11) M.R. Gotterbarm, A.M. Rausch and C. Körner: “Fabrication of Single Crystals through a μ-Helix Grain Selection Process during Electron Beam Metal Additive Manufacturing”, Metals, 10(2020), 313. 12) J.D.D. Hunt: “Steady state columnar and equiaxed growth of dendrites and eutectic”, Mater. Sci. Eng. 65(1984), 75-83. 13) S. Bontha, N.W. Klingbeil, P.A. Kobryn and H.L. Fraser: “Effects of process variables and size-scale on solidification microstructure in beam-based fabrication of bulky 3D structures”, Mater. Sci. Eng. A. 513-514(2009), 311-318. 14) J. Gockel and J. Beuth: “Understanding Ti-6Al-4V microstructure control in additive manufacturing via process maps”, 24th Int. SFF Symp. – An Addit. Manuf. Conf. SFF 2013.(2013), 666-674. 15) B. Schoinochoritis, D. Chantzis and K. Salonitis: “Simulation of metallic powder bed additive manufacturing processes with the finite element method: A critical review”, Proc. of Instit. Mech. Eng., Part B: J. Eng. Manuf. 231(2017), 96-117. 16)小泉雄一郎: “計算機シミュレーションを用いたAdditive Manufacturing プロセス最適化予測”, スマートプロセス学会誌, 8-4(2019), 132-138. 17) Y. Zhao, Y. Koizumi, K. Aoyagi, D. Wei, K. Yamanaka and A. Chiba: “Molten pool behavior and effect of fluid flow on solidification conditions in selective electron beam melting(SEBM)of a biomedical Co-Cr-Mo alloy”, Addit. Manuf. 26(2019), 202-214. 18) 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. 126(2018), 957-968. 19) Technical data for Iron, [Online]. Available: http://periodictable.com/ Elements/026/data.html. [Accessed: 8-Feb-2021]. 20) N. Raghavan, R. Dehoff, S. Pannala, S. Simunovic, M. Kirka, J. Turner, N. Carl-son and S.S. Babu: “Numerical modeling of heattransfer and the influence of process parameters on tailoring the grain morphology of IN718 in electron beam additive manufacturing”, Acta Mater. 112(2016), 303-314. 21) S. Morita, Y. Miki and K. Toishi: “Introduction of Dendrite Fragmentation to Microstructure Calculation by Cellular Automaton Method”, Tetsu-to-Hagane. 104(2018), 559-566. 22) H. Esaka and M. Tamura: “Model Experiment Using Succinonitrile on the Formation of Equiaxed Grains caused by Forced Convection”, Tetsu-to-Hagane. 86(2000), 252-258.
17-4 PH 스테인리스강의 레이저 분말 베드 융합: 열처리가 미세조직의 진화 및 기계적 특성에 미치는 영향에 대한 비교 연구
panelS.Saboonia, A.Chaboka, S.Fenga,e, H.Blaauwb, T.C.Pijperb,c, H.J.Yangd, Y.T.Peia aDepartment of Advanced Production Engineering, Engineering and Technology Institute Groningen, University of Groningen, Nijenborgh 4, 9747 AG, Groningen, The Netherlands bPhilips Personal Care, Oliemolenstraat 5, 9203 ZN, Drachten, The Netherlands cInnovation Cluster Drachten, Nipkowlaan 5, 9207 JA, Drachten, The Netherlands dShi-changxu Innovation Center for Advanced Materials, Institute of Metal Research, Chinese Academy of Sciences, 72 Wenhua Road, Shenyang 110016, P. R. China eSchool of Mechanical Engineering, University of Science and Technology Beijing, Beijing, 100083, P.R. China
Abstract
17-4 PH (precipitation hardening) stainless steel is commonly used for the fabrication of complicated molds with conformal cooling channels using laser powder bed fusion process (L-PBF). However, their microstructure in the as-printed condition varies notably with the chemical composition of the feedstock powder, resulting in different age-hardening behavior. In the present investigation, 17-4 PH stainless steel components were fabricated by L-PBF from two different feedstock powders, and subsequently subjected to different combinations of post-process heat treatments. It was observed that the microstructure in as-printed conditions could be almost fully martensitic or ferritic, depending on the ratio of Creq/Nieq of the feedstock powder. Aging treatment at 480 °C improved the yield and ultimate tensile strengths of the as-printed components. However, specimens with martensitic structures exhibited accelerated age-hardening response compared with the ferritic specimens due to the higher lattice distortion and dislocation accumulation, resulting in the “dislocation pipe diffusion mechanism”. It was also found that the martensitic structures were highly susceptible to the formation of reverted austenite during direct aging treatment, where 19.5% of austenite phase appeared in the microstructure after 15 h of direct aging. Higher fractions of reverted austenite activates the transformation induced plasticity and improves the ductility of heat treated specimens. The results of the present study can be used to tailor the microstructure of the L-PBF printed 17-4 PH stainless steel by post-process heat treatments to achieve a good combination of mechanical properties.
17-4 PH(석출 경화) 스테인리스강은 레이저 분말 베드 융합 공정(L-PBF)을 사용하여 등각 냉각 채널이 있는 복잡한 금형 제작에 일반적으로 사용됩니다. 그러나 인쇄된 상태의 미세 구조는 공급원료 분말의 화학적 조성에 따라 크게 달라지므로 시효 경화 거동이 다릅니다.
현재 조사에서 17-4 PH 스테인리스강 구성요소는 L-PBF에 의해 두 가지 다른 공급원료 분말로 제조되었으며, 이후에 다양한 조합의 후처리 열처리를 거쳤습니다. 인쇄된 상태의 미세구조는 공급원료 분말의 Creq/Nieq 비율에 따라 거의 완전히 마르텐사이트 또는 페라이트인 것으로 관찰되었습니다.
480 °C에서 노화 처리는 인쇄된 구성 요소의 수율과 극한 인장 강도를 개선했습니다. 그러나 마텐자이트 구조의 시편은 격자 변형 및 전위 축적이 높아 페라이트 시편에 비해 시효 경화 반응이 가속화되어 “전위 파이프 확산 메커니즘”이 발생합니다.
또한 마르텐사이트 구조는 직접 시효 처리 중에 복귀된 오스테나이트의 형성에 매우 민감한 것으로 밝혀졌으며, 여기서 15시간의 직접 시효 후 미세 조직에 19.5%의 오스테나이트 상이 나타났습니다.
복귀된 오스테나이트의 비율이 높을수록 변형 유도 가소성이 활성화되고 열처리된 시편의 연성이 향상됩니다. 본 연구의 결과는 기계적 특성의 우수한 조합을 달성하기 위해 후처리 열처리를 통해 L-PBF로 인쇄된 17-4 PH 스테인리스강의 미세 구조를 조정하는 데 사용할 수 있습니다.
Atom Probe Tomographic Characterization of nanoscale cu-rich Precipitates in 17-4 precipitate hardened stainless steel tempered at different temperatures
An analytical method to predict and compensate for residual stress-induced deformation in overhanging regions of internal channels fabricated using powder bed fusion
Nano/ultrafine grained austenitic stainless steel through the formation and reversion of deformation-induced martensite: Mechanisms, microstructures, mechanical properties, and TRIP effect
J.B. Ferguson, Benjamin F. Schultz, Dev Venugopalan1, Hugo F. Lopez, Pradeep K. Rohatgi, Kyu Cho, Chang-Soo Kim
On the Superposition of Strengthening Mechanisms in Dispersion Strengthened Alloys and Metal-Matrix Nanocomposites: Considerations of Stress and Energy
J.B. Speed School of Engineering, University of Louisville, Louisville, KY 40292, United States
Abstract
The surface morphology of parts made by the laser powder bed fusion (L-PBF) process is governed by the flow of the melt pool. The nature of the molten metal flow depends on the material properties, process parameters, and powder-bed particles, etc., and may result in potentially significant variations along a laser scanning path.
This study investigates the formation of transient and steady-state zones through a single-track l-PBF experiment using Inconel 625 powder. Single tracks with lengths of 1 mm and 2 mm were fabricated using 195 W laser power and scan speeds of 400 mm/s or 800 mm/s. The surface morphology of the track was analyzed using a white light interferometer (WLI), and an individual single track can be divided into three distinct zones based on the track width and height.
The initial transient region has a wider and taller solidified track geometry, the region near the end of a scan has a tapered profile with a decreasing track width and height, while the steady-state region in the middle has a smaller variation in the width and height.
A mesoscale numerical model was further developed using FLOW-3D to examine the formation of the transient and steady-state zones. At the start of a scan, strong flow occurs outward and backward, leading to the formation of a wider track with a bump. As the scan continues, the thermal gradient stabilizes, leading to a steady state, which resulted in a very small fluctuation in the width. Furthermore, the tapered end of the scan track is due to the half-lemniscate shape of the melt pool during laser scanning.
L-PBF(Laser Powder Bed fusion) 공정으로 만든 부품의 표면 형태는 용융 풀의 흐름에 따라 결정됩니다. 용융 금속 흐름의 특성은 재료 특성, 공정 매개변수 및 분말층 입자 등에 따라 달라지며 레이저 스캐닝 경로를 따라 잠재적으로 상당한 변동이 발생할 수 있습니다.
이 연구는 Inconel 625 분말을 사용하여 단일 트랙 l-PBF 실험을 통해 과도 및 정상 상태 영역의 형성을 조사합니다. 1 mm 및 2 mm 길이의 단일 트랙은 195 W 레이저 출력과 400 mm/s 또는 800 mm/s의 스캔 속도를 사용하여 제작되었습니다. 트랙의 표면 형태는 백색광 간섭계(WLI)를 사용하여 분석되었으며 개별 단일 트랙은 트랙 너비와 높이에 따라 3개의 별개 영역으로 나눌 수 있습니다.
초기 과도 영역은 더 넓고 더 높은 응고된 트랙 형상을 가지며, 스캔 끝 근처의 영역은 트랙 너비와 높이가 감소하는 테이퍼 프로파일을 갖는 반면, 중간의 정상 상태 영역은 너비와 높이에서 더 작은 변동을 가집니다. 신장. 중간 규모 수치 모델은 과도 및 정상 상태 영역의 형성을 조사하기 위해 FLOW-3D를 사용하여 추가로 개발되었습니다.
스캔이 시작될 때 강한 흐름이 바깥쪽과 뒤쪽으로 발생하여 범프가 있는 더 넓은 트랙이 형성됩니다. 스캔이 계속됨에 따라 열 구배가 안정화되어 정상 상태로 이어지며 폭의 변동이 매우 작습니다. 또한 스캔 트랙의 끝이 가늘어지는 것은 레이저 스캔 중 용융 풀의 반-렘니케이트 모양 때문입니다.
Keywords
Additive manufacturing, Laser powder bed fusion, Numerical modelling, Transient region
냉각 속도 및 온도 구배와 같은 FLOW-3D AM 데이터를 미세 구조 모델에 입력하여 결정 성장 및 수상 돌기 암 간격을 예측할 수 있습니다.
레이저 파우더 베드 융합으로 제작 된 니켈 기반 초합금의 열전달, 유체 흐름 및 응고 미세 구조 모델링
오하이오 주립 대학의 연구원들은 니켈 기반 초합금의 미세 구조 진화를 예측하기 위해 용융 풀과 고체 / 액체 인터페이스의 적절한 위치에서 열 구배 및 냉각 속도 데이터를 추출했습니다.
열 응력 | Thermal Stresses
FLOW-3D AM 시뮬레이션의 결과를 ABAQUS 또는 MSC NASTRAN과 같은 FEA 소프트웨어에 입력하여 추가 열 응력 분석을 실행할 수 있습니다. 여기에서 T- 조인트의 레이저 용접 시뮬레이션 결과를 추가 응력 분석을 위해 ABAQUS로 가져 오는 방법을 볼 수 있습니다. 마찬가지로 LPBF 시뮬레이션에서 응고 된 용융 풀 데이터의 결과를 사용하여 다른 FEA 소프트웨어에서 열 응력 및 왜곡 분석을 연구 할 수 있습니다.
Thermal Stresses Case Study
Directed Energy Deposition
DED (Directed Energy Deposition)는 레이저 또는 전자 빔과 같은 에너지 소스를 사용하여 가열 및 융합되는 와이어 또는 분말을 증착하여 부품을 만드는 적층 제조 공정입니다. FLOW-3D AM 은 분말 또는 와이어 이송 속도 및 크기 특성, 레이저 출력 및 스캔 속도와 같은 공정 매개 변수를 고려하여 DED 공정을 시뮬레이션 할 수 있습니다. 또한, 기판과 분말 재료의 서로 다른 합금에 대해 독립적 인 열 물리적 재료 특성을 정의하여 다중 재료 DED 프로세스를 시뮬레이션 할 수 있습니다.
레이저 물리학의 구현과 열 전달, 응고, 표면 장력, 차폐 가스 효과 및 반동 압력을 포함한 압력 효과를 통해 연구원은 결과 용접 비드의 강도 및 균일성에 대한 공정 매개 변수의 영향을 정확하게 분석 할 수 있습니다. 또한 이러한 시뮬레이션을 여러 레이어로 확장하여 후속 레이어 간의 융합을 분석 할 수 있습니다.