Adp 001005

j An adaptive grid, finite-volume method has been used to solve the NavierStokes equations for complete (forebody and afterbody) flowfields around blunt belies. The code, which is applicable for axisymmetric cr two-dimensional flows, allows the mesh to adjust during the computation to provide a closer spacing of mesh points in regions of high gradients, thus minimizing the number of required computational points. The solution technique is explicit, utilizing a maximum time-step advancement at each grid point to accelerate convergence to the steady state. The code has been fully vectorized for efficient solution on the CYBER 203 computer. A very flexible rezoning routine is used to concentrate mesh points anywhere in the field, either by a user-defined weighting function or by allowing high gradient regions to adjust the grid. The grid adjustment routine is implicit in nature and represents a very small portion of the total computational cost. Currently, the code runs in approximately 1.6 > W?? 0, ' c 0 < seconds per grid point per iteration. INTRODUCTION The finite-volufcK; method of numerically solving systems of conservation laws has been successfully applied to a wide variety of problems in fluid mechanics. Its ability to maintain conservation of mass, momentum, and energy from cell to cell, even in rather complex nonorthogonal coordinates, makes it particularly attractive for use with adaptive grid techniques. For the purposes cf this paper, a finite-volume formulation (FVF) is defined as a discrete approximation to a conservation law written in integral form which (1) uniquely defines control volumes in such a way that control volumes (ceils) do not overlap nor are gaps left in physical space and (2) uniquely defines fluxes and forces acting through coll walls so that summability without residue (conservation) is guaranteed. It differs from a finite difference formulation IFDF) only in th» «av a prcdlsa i» approached. For example, given a system of conservation laws we might consider the FDF A» a discrete approximation to the