Surface conformed linear mesh and data subdivision technique for large-scale flow simulation and visualization in Variable Intensity Computational Environment

Abstract

In this paper, we report the development of a parallel program to isotropically subdivide a 3-D hybrid unstructured coarse base mesh to generate a finer computational mesh without user interaction. Our Variable Intensity Computational Environment enables scientific computations using very large mesh yet allows users to interact with more manageable smaller mesh. The main motivation behind this study is to overcome the bottleneck in generating and processing of the computational meshes with billions of elements. First, we generate a coarse mesh using any unstructured mesh generator. Then, we subdivide the coarse mesh to the level of resolution needed for the simulations. Finally, we conform mesh nodes on solid surfaces to the original geometry since linear subdivision ignores surface curvatures. We use K–D tree search algorithm in the surface mapping. To deform interior mesh nodes due to the surface correction, we use the spring analogy method since deformations are very small. Surface correction is implemented in parallel using the Message Passing Interface. The new mesh obtained from the isotropic subdivision preserves mesh density distribution of the original coarse mesh. The mesh subdivision with surface correction is integral part of our Variable Intensity Computational Environment. Three test cases are used to demonstrate applicability of this method: a generic reentry vehicle, an Army projectile, and a sphere. Flow solutions are obtained using our compressible and incompressible Navier–Stokes CaMEL flow solvers with the Detached Eddy Simulation turbulence model. Flow solutions and mesh subdivisions are performed in a parallel cluster at the Jackson State University.