The development of multiscale models for dynamic response of solids and fluids under strong loadings is now viewed as a promising approach to enabling predictive simulation. Validation of these models is particularly challenging, in that one must examine various unit (typically sub-grid) processes that are used to build the models and assess their accuracy in the context of an integrated simulation. In this talk we describe the development of well-instrumented benchmark experiments that are designed and carried out in close connection with supporting simulations. The idea is to ensure that the experimental results are directly relevant to the validation of the associated multiscale modeling. and that experimental results can be used to sharpen simulation (and vice versa). A related goal is to provide real but reasonable challenges to validated simulation. Three such experiments and the associated simulations are reviewed: generation of compressible turbulence and mixing in a converging shock tube, evolution of polycrystalline texture in copper undergoing high rate shear and compression, and fracture of thin metal tubes by gas phase detonation waves.
(Work supported under the Academic Strategic Alliance Program of the Advanced Simulation and Computing Program at NNSA)