Reconciled top-down and bottom-up hierarchical multiscale calibration of BCC FE crystal plasticity
Article
Tallman, AE, Swiler, LP, Wang, Y et al. (2017). Reconciled top-down and bottom-up hierarchical multiscale calibration of BCC FE crystal plasticity
. 15(6), 505-523. 10.1615/IntJMultCompEng.2017021859
Tallman, AE, Swiler, LP, Wang, Y et al. (2017). Reconciled top-down and bottom-up hierarchical multiscale calibration of BCC FE crystal plasticity
. 15(6), 505-523. 10.1615/IntJMultCompEng.2017021859
In this paper, a test for connections between models via parameter sets is developed. A set of parameters from the flow rule of a crystal plasticity model for bcc Fe is identified for connecting top-down and bottom-up information. The top-down calibration is performed using experimental measurements of single-crystal yield strength at multiple temperatures and crystallographic orientations, where a likelihood function in parameter space is informed using second-order regression surrogate modeling. A bottom-up calibration of the same model uses the parameter estimates from atomistic simulations to inform penalty functions. A constrained likelihood function incorporates the top-down and bottom-up information in one calibration of parameters. Decision making within hierarchical multiscale modeling is approached. The benefit to calibration precision brought by incorporating additional data from bottom up is considered against the uncertainty in the requisite multiscale connection. This trade-off is formulated into an empirical test of connections. Hypothetical decision making is demonstrated between multiple alternative bottom-up estimates.