SUMMARY

Computationally efficient structure-property (S-P) linkages are essential for the accelerated development and deployment of structural materials. This represents a major challenge for polycrystalline materials, which exhibit rich heterogeneous topologies at multiple structure/length scales and produce a wide range of properties. The comprehensive evaluation of these properties potentially requires thousands of simulations using costly models. In the proposed research, computationally efficient protocols will be developed to predict the mechanical properties of polycrystalline materials using Materials Knowledge Systems (MKS). For the MKS approach, physics-based kernels (calibrated with microstructures and their associated responses) capture the microstructure-sensitive response of a materials system. Once calibrated, the kernels may be used to predict the responses and properties of new microstructures at orders of magnitude lower cost. It is believed that generalized spherical harmonics (GSH) as a Fourier basis for functions defined on the orientation space will lead to a compact and computationally efficient representation of the desired S-P linkages for polycrystalline materials. In this work novel protocols will be developed and demonstrated for the prediction of bulk properties including yield strength and elastic modulus, and for the ranking of fatigue resistance in alpha-Ti microstructures using datasets produced through crystal plasticity finite element (CPFE) simulations.