This report describes work performed to evaluate and improve the 2018 version of EPIC. The first task includes a thorough evaluation of the three meshless-particle algorithms in EPIC (GPA, Nodal EFG, CPEM). A range of problems is evaluated to consider the effects of the three algorithms with and without conversion of elements into particles, conversion from hexahedral and tetrahedral elements, and a variety of sliding contact interfaces. The second task includes expansion of the initial OpenMP parallelization effort into new algorithms, and examination of potential barriers to scaling problems containing hundreds of millions of elements onto thousands of processor cores.
展开▼