Assessment of element type on fatigue damage prediction using crystal plasticity finite element simulations — UC Berkeley

Fatigue predictions using microstructure-sensitive computational frameworks such as crystal plasticity finite element (CPFE) simulations have gained traction in recent years as a means of reducing overreliance on extensive experimental testing. However, to build confidence in CPFE-based fatigue results, it is essential to characterize and understand all potential sources of variability inherent to such simulations. While factors such as regularization domains, input parameters, boundary conditions, and loading conditions have been investigated in literature, the effect of element type on fatigue damage predictions has not been previously studied, despite the widespread use of linear tetrahedral (C3D4) and hexahedral (C3D8) elements. This work addresses the gap by systematically evaluating the influence of element type on CPFE-based fatigue predictions. Known numerical artifacts associated with each element type, specifically, volumetric locking in C3D4 elements and stair-stepping at grain boundaries in voxelated C3D8 elements, are examined in the context of the location of hotspot and fatigue life predictions. Additionally, specific cases are compared against reduced (C3D8R) and higher ordered (C3D10) element types. The findings demonstrate that despite these inherent differences, both element types result in the same location of fatigue hotspot and yield comparable fatigue life predictions, suggesting that element type does not introduce significant bias into CPFE-based fatigue assessments when appropriate modeling practices are followed.