The computational scaling performance for different progressive damage and failure analysis (PDFA) problems of interest is examined using Abaqus/Explicit. Three models that range from a coupon with single-interface delamination to a structural element with a complex, several-interface, delamination migration damage mode were exercised in the study. By comparing run times across different numbers of cores, hardware, and models, the computational scaling performance is quantified. A run time reduction of ten times is achieved scaling from 24 to 512 cores. The computational scaling performance is relatively consistent among the different models, suggesting that the speed up performance that is obtained is characteristic of the class of models. By analyzing and extrapolating the data collected, the current state of computational tractability is defined and discussed.
展开▼
