Incorporating surface roughness into numerical modeling for predicting fatigue properties of L-PBF AlSi10Mg specimens

Abstract

Laser-powder bed fusion (L-PBF) is a popular additive manufacturing method for fabricating metal parts with complex geometries. However, one of the primary setbacks lies in the surface quality of the as-built components. Poor surface quality influences the dynamic fatigue properties of L-PBF parts. Experimental investigation of fatigue properties is time-intensive and expensive. Hence, computational modeling can be implemented to critically evaluate the impact of surface roughness on fatigue. In this work, a numerical modeling framework is developed that considers the influence of part-scale surface roughness on fatigue properties. To calibrate and validate the numerical modeling framework, mechanical investigations of miniature AlSi10Mg specimens are conducted in a custom-built tensile-fatigue testing apparatus. First, the tensile properties are evaluated to find out the reliability of the apparatus. Thereafter, the fatigue investigations of two sets of specimens are conducted: five specimens (FS1-FS5) for calibrating the numerical model and three specimens (VFS1-VFS3) for model validation. The numerical modeling framework involves the 3D reconstructed specimen geometries from computed tomography (CT) imaging of the VFS1-VFS3 specimens. The resultant 3D geometries are discretized using Ansys finite element software and analyzed to simulate the fatigue behavior. The numerical results of fatigue life and failure locations of the VFS1-VFS3 specimens are compared with the experimental observations. The comparison shows that the numerical results are within 5 % of the experimental observations. Additionally, the location of maximum strain localization in numerical modeling matches the crack initiation location. In conclusion, the numerical modeling can predict low cycle fatigue life and failure location of L-PBF AM specimens.