The purpose of this project is to develop parallel algorithms to treat a class of multiscale problems associated with solid particle dynamics in liquids. The focus is on full three-dimensional direct numerical simulation of the interactions of sufficiently large number of particles, thousands of particles, so that collective behavior might be inferred. This builds upon the previous research supported under the NSF-HPCC Grand Challenge Group Award where two-dimensional simulation had been successfully accomplished.
The particle movers algorithm uses a combined weak formulation in which the fluid and particle equations of motion are combined into one single equation set in which the interaction forces between the particles and fluid cancel, circumventing the need for explicit calculation of forces. Extension to three-dimensions as well as increasing the number of particles in order to obtain reliable sample size, requires a major overhaul of preconditions and solvers which will define new directions for computational fluid mechanics.
The multiscale phenomena addressed here are fundamental to the chemical process industry and oil exploration and recovery. Therefore, special considerations will be given to present the simulation results into forms which can be understood and used by industry. For example, collaboration with STIM-LAB a group of industrial partners in the oil and gas will be strengthened to develop a 3D design software package based on the research conducted under NSF support.
