Automated fitting of stress-strain data and texture for crystal plasticity codes such as VPSC, EPSC, CPFE, and VPFFT
Crystal plasticity and more specifically the implementations that account for twinning and dislocation evolution have several dozen parameters that must be fit. For a skilled, knowledgable worker, it requires weeks to months to configure a CP model to reproduce a robust dataset. Besides the large time commitment, the final parameters are likely not optimal.
This project couples Matlab's parallel environement, global optimization tool kit, and MTEX compatibility to quickly fit large datasets. The cost function currently handles fitting of the stress-strain response and will be extended in the future to include volume fraction of main texture components, general texture difference, and quantities such as twin fractions when determining the best fit parameters.