PSI - Issue 42

Harry O. Psihoyos et al. / Procedia Structural Integrity 42 (2022) 299–306 Author name / Structural Integrity Procedia 00 (2019) 000 – 000

300

2

1. Introduction Selective Laser Melting (SLM) is an Additive Manufacturing (AM) process in which metallic parts with complex geometries, canbe fabricated. In SLM process, a single ormultiple laser heat beam sources are utilized to selectively melt and fuse metallic powder together. The pattern of the heat source is guided from the input 3Dmodel and when the process of a powder layer is finished, the bed moves down to be recoated with another powder layer and the process is repeated in layer-by-layer manner until the building of the whole part (Frazier, 2014). Due to the typical melt track widths of 0.1-0.2 mm acquired with the SLM process, the intricate geometric characteristics of the input digita l models can be fabricated with high precision and moderate-to-high surface finish (du Plessis et a l., 2022). Therefore, SLM process has ga ined the a ttention of many industria l sectors over the last two decades, including automotive, medical andaerospace (Blakey-Milner et a l., 2021). One of the ma in disadvantages of the SLM process is the development of residual stresses during the fabrication of the part due to the high temperature gradients and the rapid heatingandcooling ra tes associated with the process (Olleak & Xi, 2020b). Although, residual stresses can be relieved by post processing techniques, they may lead to premature failure of the part during its fabrication (Li et a l., 2017). Thus, the prediction of residual stresses in SLM parts with reliable models is of grea t importance to prevent failure and ma terial waste. Because of the huge computational cost of the deta iled modelling of the SLM process for the fabrication of a part, many modelling approaches have been proposed in the litera ture. All of these models use approximations to simplify the analysis, so the direct modelling of each individual laser pass is no longer required (Gouge et a l., 2019). Despite their simplifica tions, a ll approaches a im to accurately predict the residual stresses in an efficient way. To this end, the utilized meshing strategy is very important to achievea combination of minimum computational times with accurate results. The commonest mesh strategy is themeshingwith hexahedral elements, also knownas cartesianmesh scheme, eitherwith uniform element size throughout thesimulationof the buildingof a part orwith adaptive meshing resizing technique (Denlinger et a l., 2014).More recently, layered tetrahedralmeshinghas beenproposedas ana lternative to cartesianmesh scheme due to its ability tobetter discretize the intricategeometric features of components. Previously utilized in finite element models for fracture mechanics analyses (Nejati et al., 2015), tetrahedral mesh scheme has been integrated both in inherent strain (Baiges et al., 2021) and thermomechanical modelling methods for the SLM simula tionof parts (Olleak &Xi, 2020a, 2021). However, a detailed investigation of the accuracy and efficiency of layered tetrahedralmeshinghas not presented in the litera ture with the exception of the work of Weber et a l. (2020). In the present work, a comparison between cartesian and layered tetrahedral mesh stra tegies and a detailed investiga tion of accuracy and characteristics of layered tetrahedral meshing in SLM simulations are presented. The results of the models of the different mesh schemes were compared with available from litera ture experimental data of residual stra insto assess their accuracy.Moreover, a ll the models are evaluated in terms of computational time and element quality, in order toprovide guidelines for the selection of thesuitable mesh schemefor theefficient simulation of the SLM process of a fabrication of a realistic part. 2. Experimental data In order to evaluate theaccuracyof themodels basedon theexamined mesh strategies, the predicted results of each model are compared with available experimental data from the literature. The experimental measurements concern the distributionof residual strains of an IN625 single cantilever beamand theywere characterized by X-raydiffraction method. The present data are part of a set of experimental campaign series conducted by NIST to support modelling va lida tion purposes (Phan et al., 2019). The geometry of cantilever beam specimen is presented in the Fig. 1. As it can be seen, the cantilever beam has some intrica te geometric characteristics, designed on purpose to provoke distortions. The specimen was built in EOS M270 machine and the main process parameters were 195 W for laser power, 600 mm/s for scanning speed, 20 μ m for layer thickness, 100 μ m for hatch spacingand17s for layer deposition time. The scanpatterndirection was 45˚ from thex - directionwith contour scanningand 90˚ rotationper layer. More deta ils about process conditions, the dimensions and the experimental procedure for the determination of residual stra ins canbe found in the work of Phanet a l. (2019).