 |
Center for Nonlinear Studies |
We are developing computational techniques for the prediction of general compressible multimaterial flows in contexts that involve multiscale dynamics. Examples of high-speed multimaterial flows include explosions in dusty gases and particle-laden fluids where the particles may be other fluids (droplets), spall of materials in munition deployment due to void growth and coalescence, detonation initiation in a porous/heterogeneous condensed explosive etc. Simulation of such interactions is a challenging task because the dynamics at the system level is intimately linked to that at the (unresolved particle-phase) meso-scale. This scale separation indicates that in order to simulate system behavior, information from the meso-scale needs to be modeled and presented to the continuum (large-scale) formulation as source terms or other homogenized property relationships. Techniques have been developed to perform (separately) simulations at the meso- and macro-scales and transmit information between the scales. This talk will focus on examples of the computations of shock-wave interactions with a variety of systems containing multimaterials (solid-solid, solid-gas, liquid-gas mixtures). The capability of performing meso-scale (direct numerical) simulations with multitudes of particles will be shown. Having developed the techniques for handling phenomena at each scale, the challenge lies in devising the exchange of information between the scales. The approach that is being developed to handle this information exchange relies on data abstraction and compression using multiscale wavelet transforms and model assimilation using artificial neural networks. These techniques and their applicability to scale identification and inter-scale communication will be described in the context of modeling of particle-laden gas flows in the presence of shocks.
Host: Mikhail Shashkov. shashkov@lanl.gov, 667-4400