SUMMARY
Crystal Plasticity Finite Element (CPFE) simulations provide physics-based predictions of the plastic response in polycrystalline metals subjected to large plastic strains. Despite their demonstrated high fidelity in a number of applications, these approaches have not yet been adopted broadly by the metal working industry because of their extremely high computational cost. This dissertation develops suitable enhancements to the novel Materials Knowledge System (MKS) framework to define a reduced-order partitioning of the plastic response on a polycrystalline aggregate to regions within individual grains (i.e., localization of the plastic response). The reduced-order models developed using the enhanced MKS localization framework were found to provide highly accurate predictions of the mesoscale fields with 2-3 orders of magnitude savings in the computational cost. Furthermore, these computational savings were observed to be independent of the RVE size. As a result, the work performed in this dissertation paves the way forward for the development of computationally low-cost, fully-coupled, multiscale simulations of plastic deformations in polycrystalline metals.