Experiment-based validation and uncertainty quantification of coupled multi-scale plasticity models

摘要: Anisotropic materials, such as zirconium require the inclusion of evolution crystal structure in finite element model representations mechanical behavior, naturally leading to coupled meso- and macro-scale plasticity models. For achieving models, partitioned techniques where isolated models that resolve system behavior at different scales are through iterative exchange inputs outputs widely used. In this treatment, a provides strain information meso-scale visco-plastic self-consistent represent micro-scale properties. These properties then returned for new stress calculations. During process, biases uncertainties inherent within constituent predictions propagate between constituents. This propagation creates need multi-scale approach experiment-based validation uncertainty quantification, which separate effect experiments conducted each constituent’s domain test validity independent constituents their respective integral-effect executed validate entire system. paper authors implement utilizing both experiments. Results demonstrate training an error bias separate-effect means appropriately bias-correcting during coupling iterations results improved predictive capability. improvement is demonstrated use