Two-dimensional, viscous, incompressible flow in complex geometries on a massively parallel processor (Q1201092)

In this paper a code to compute two-dimensional fluid flows in general geometries has been developed. The random vortex method is used and the code is implemented on the Connection Machine, a massively parallel processor with a hypercube interconnection network. The important issues of this procedure were: The design of efficient parallel \(N\)-body solvers. Accurate elliptic solvers in highly complex geometries. Efficient run-time data mappings between several processor topologies. Complex parallel boundary conditions to allow arbitrary geometries as input. Parallel real-time flow visualization for interactive flow diagnostics. Several examples, the flow in an expanding and contracting nozzle, the flow around an island and the flow around a piston/valve have been shown. The data from these computations have been visualized.