CFD-XDEM coupling approach towards melt pool simulations of selective laser melting

Within the domain of metal Additive Manufacturing (AM), the challenge of qualification emerges prominently. This challenge encapsulates the endeavor to establish a set of process parameters that can reliably yield consistent and repeatable production outcomes. While additive manufacturing technologies like selective laser melting (SLM) have gained widespread usage for crafting metal parts boasting intricate geometries and high precision, they are not exempt from critical concerns. Defects, most notably porosities, persist as a substantial hurdle. The origin of these imperfections lies in microscale phenomena inherent to the melting and solidification processes occurring during layer-by-layer fabrication. This study presents a Computational Fluid Dynamics-eXtended Discrete Element Method (CFD-XDEM) coupling to model the dynamics and thermodynamic interplay between the powder bed and melt pool during SLM. The XDEM model simulates various aspects of powder behavior, including deposition, heating via laser radiation, melting, shrinkage, and the associated transfer of mass, momentum, and energy between the particles and the surrounding liquid and gas. The CFD model is based on the Volume Of Fluid (VOF) method and simulates the formation and evolution of the melt pool, taking into account surface tension force, Marangoni flow, buoyancy-driven flow inside the melt pool, phase change (solidification and melting), and the laser radiation on the melt surface. A direct coupling establishes a bidirectional transfer of source term data between the XDEM and the CFD. This involves the exchange of information such as the mass source of molten metal, convective heat transfer between particles and the fluid mixture, as well as the drag forces acting between the liquid and the particles in both directions. This direct coupling is achieved through the incorporation of source terms within the equations of the XDEM and CFD models. The present study is currently undertaking a comprehensive validation of the proposed method throughout each stage of development. This validation involves comparing model results with experimental data and benchmark problems. To initiate the process, the Marangoni model is being i validated against benchmark problems. Subsequently, the laser model is being implemented to predict the results of a laser melting experiment on a metal block. As the CFD model is finalized, the coupling is concurrently being developed. To validate the reliability of heat, mass, and momentum transfer within the coupling, an experiment involving the melting of ice is being replicated. This experiment serves as a method to affirm the performance of the melting model. The outcomes of this experiment are providing validation for the CFD-XDEM coupling's performance. Moving forward, the model is being utilized to predict outcomes for a low-power SLM experiment involving a single layer, considering various laser scanning velocities. Impressively, the simulation outcomes are demonstrating excellent agreement with experimental data. This alignment is underlining the model's capacity to accurately forecast melt pool dimensions. Furthermore, the model is being extended to simulate a larger powder bed, enabling an examination of melt pool characteristics as well as heat transfer interactions with the powder particles. The model presented in this study offers several distinctive features: The phase change of the particles is explicitly solved in the XDEM model, with particles undergoing melting at the melting temperature, shrinking, and disappearing when they are completely melted. The XDEM model solves for conduction and radiation between adjacent particles, providing an advantage over continuous powder bed models that require an estimate of effective thermal conductivity. Moreover, The particles are modeled as one-dimensional elements instead of 3-dimensional CFD spherical geometries, which is anticipated to be computationally more efficient. The CFD model incorporates all the relevant physical phenomena to the dynamics of melt pool, including Marangoni flow, buoyancy-driven flow, and surface tension forces. The volumetric heat source for the laser radiation is adaptive to the geometry of melt pool. The proposed model offers a reliable and efficient method for predicting the behavior of melt pool, and it is expected to facilitate the optimization of SLM process parameters to reduce the defects and improve the quality of manufactured parts.