This conference paper was submitted for presentation at the NAFEMS World Congress 2025, held in Salzburg, Austria from May 19–22, 2025.

Abstract

When finite element analysis (FEA) is applied to the process of powder bed fusion (PBF), there are numerous numerical options and choices that have to be made to obtain a satisfactory and effective FEA model that can produce the required results. The reported research here looks at the numerical considerations and choices and compares the numerical models and outcomes for each choice. It is important to consider the intended purpose for doing the numerical modelling work right from the onset and to embark on the numerical endeavour accordingly. The fidelity, complexity, size and speed of the numerical model should all be considered and planned carefully. In the reported work, the purpose behind the numerical modelling is presented. This is followed by looking at numerical considerations including the level and location of mesh refinement, numerical control parameters and how they can increase numerical efficiency, time increment sizes and how they correspond to process parameters and mesh refinement. Another worthwhile consideration, which can significantly impact the size of the output file, is to appropriately specify the required set of output parameters and the location where they are needed. A thermal FEA model is generated and meshed for the intended purpose. It is a small cuboid made of an aluminium alloy that has the top surface covered by a thin layer of the same metal in the powder state to allow the deposition in a single pass produced by a narrow laser beam tool path. Two meshing strategies are investigated, varying the level and location of mesh refinement. Comparisons between the two meshing strategies allow conclusions to be made about the most appropriate way to mesh such FEA models. The time increment may also have a direct impact on the accuracy of the temperature history and the numerical convergence of the model. A range of time increments is investigated for that reason. Furthermore, control parameters can be critical for the accuracy of the numerical results. Making them unnecessarily strict can make the model too slow to run without any improvement to the results. The control parameters are investigated for that reason, and recommendations are made that can save substantial run times without compromising the accuracy. Final conclusions are presented on how to improve the efficiency of this type of FEA.