Parallel computation of 2-D potential and euler equations on transputer arrays

Abstract

The transonic full potential and the Euler equations are solved on a distributed-memory parallel computer with 16 transputers. The method of lines approach is adopted with a second order centered difference spatial discretization and a rational Runge–Kutta time integration. For both potential and Euler equations, artificial viscosity is introduced to capture the shock wave. In the parallelization strategy, the domain decomposition approach is used in which the whole physical domain is divided into a number of smaller subdomains called blocks. One line of overlapping auxiliary grids surrounding a given number of blocks is used to make available the information required in the determination of flow variables at the block boundaries. Basically each block is treated by a different processor. The results of applications to transonic flows in a 2-D channel with 10 % circular bump and subsonic M∞ =0.38 flow over a circular cylinder demonstrate a high speedup of the present algorithm on the transputer arrays.