In Selective Laser Melting (SLM) processes, residual stress formation can substantially affect mechanical prop­ erties and geometric accuracy of the produced part. The process parameters together with the adopted laser scan strategy should be optimized in order to reduce residual stresses in the solidified artifact. Employing traditional numerical methods, i.e., boundary-conforming mesh finite elements, can be computationally very expensive when complex geometrical features are involved, since the capability of capturing solidified part shapes depends on the element size. Immersed boundary methods present a valid alternative to effectively predict residual stresses employing a rather coarse discretization without loosing accuracy in the solidified geometry represen­ tation. In the present work, the finite cell method, an immersed boundary method, is employed to predict re­ sidual stresses using a high-fidelity thermo-mechanical model resolving the melt-pool length scale. The thermal model is first validated with respect to experimental measurements from the literature, while the mechanical results are compared to boundary-conforming mesh analysis for the same geometrical accuracy. Finally, the numerical method is applied to study the influence on residual stresses of two different laser scan strategies on a 10-layer SLM structure. The proposed immersed boundary approach can lead to a remarkable improvement in terms of computational time and memory usage when the mesh generation step is the main burden of a thermomechanical SLM process simulation.     «

In Selective Laser Melting (SLM) processes, residual stress formation can substantially affect mechanical prop­ erties and geometric accuracy of the produced part. The process parameters together with the adopted laser scan strategy should be optimized in order to reduce residual stresses in the solidified artifact. Employing traditional numerical methods, i.e., boundary-conforming mesh finite elements, can be computationally very expensive when complex geometrical features are involved, since t...     »