Flow computations using embedded boundary conditions on block structured Cartesian grid

Abstract

Computational Fluid Dynamics (CFD) has been increasingly relied on as an important tool for the aerodynamic design. However, mesh generation for 3D complex geometries is still the most time-consuming task in CFD. Recently, Cartesian grid methods have become popular because of their advantages of fast, robust, and automatic grid generation. Authors proposed a new approach based on a block-structured Cartesian grid approach, named Building-Cube Method (BCM, [5]). This method is aimed for large-scale, high-resolution computations of flows. In the original BCM, however, the wall boundary is defined by a staircase representation to keep the simplicity of the algorithm. Therefore, to resolve boundary layers, vast amounts of grid points are required especially for high-Reynolds number flows of practial applications. In this paper, a new embedded boundary treatment suitable for high-Reynolds viscous flows on BCM is proposed. The scalability performance on cluster computers is also investigated with this method. © 2010 Springer.