Adaptive sampling, in hierarchical simulation

We propose art adaptive sampling methodology for hierarchical multi-scale simulation. The method Utilizes a moving kriging interpolation to significantly reduce the number of evaluations of finer-scale response functions to provide essential constitutive information to a coarser-scale simulation model. The Underlying interpolation scheme is unstructured and adaptive to handle the transient nature of a simulation. To handle the dynamic construction and searching of a potentially large set of finer-scale response data, we employ a dynamic metric-tree database. We study the performance of our adaptive sampling methodology for a two-level multi-scale model involving a coarse-scale finite element simulation and a fine-scale crystal plasticity-based constitutive law. Copyright (c) 2008 John Wiley & Soils, Ltd.