Data Structures for Efficient and Integrated Simulation of Multi-Physics Processes in Complex Geometries

Advanced data types for simulation of discrete and continuum dynamics in complex 3D geometries are proposed. These data types form the basis of a model-development environment, MulPhys, based on 3D libraries for manipulation of geometrical primitives, object-oriented multi-domain modeling paradigm, continuum solvers on unstructured meshes, and discrete particle solvers. The numerical schemes are based on the methods of continuum and discrete mechanics, such as the control-volume method for the solution of transient flow problems on unstructured meshes, the finite-element method for the solution of elasticity problems for linear, planar and 3D elements, and the Lagrangian particle method for the solution of particle motion in continuum media. The interaction of fluids, structures and particles is accomplished through the coupling of the solvers in a unified multi-solver modeling paradigm. Several prototype example cases of complex continuum mechanical systems are considered. It is proposed to refine and optimize the existing data types to better represent the multi-physics problems in a multi-processor computational environment. At the same time, further extensions and generalization of the data types and multi-modeling schemes are proposed so as to better suite the problems with strong non-locality features, and multi-scale processes.