Abstract
This chapter demonstrates high parallel efficiency of the coupled method by applying it to two kinds of unsteady flows: one is the flow around a 2D square cylinder and the other is the flow around a 3D cube. A Fujitsu API000 is used for calculation. The efficiency roughly depends on the number of control volumes allocated to one processor. Furthermore, the chapter proposes the Boundary Fitted Coordinate (B.F.C) version of the coupled method. The B.F.C version has not been implemented on a parallel computer, but its parallel efficiency is expected to be comparable to that of the Cartesian version. In the coupled method—as an incompressible Navier–Stokes solver suitable for massively parallel computing of very fine granularity—on the Cartesian, 5 or 7 velocity components and one pressure value corresponding to one staggered cell are simultaneously updated. When a parallel computer with n processors is used, the computational time, t is roughly proportional to the square of the number of grid points N 2 and to the reciprocal of the net computer performance. This situation in the future is difficult to imagine at present, but it may happen dome day. The focus is on the way to manage the very fine granularity that can be seen in such a situation. The coupled method discussed in the chapter is characterized by use of the staggered grid. It implies the Cartesian. However, it is very important to simulate flows of complicated geometry in terms of practical application, and expansion of the coupled method to the B.F.C is necessary.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call