Algorithmic Trends in CFD in the 1990’s for Aerospace Flow Field Calculations

Abstract

The development of computational fluid dynamics (CFD) procedures has progressed extremely rapidly during the past two decades. The parallel rapid development in computer hardware resources and architectures has not only matched the explosive algorithm development but has indeed provided and continues to provide its impetus. Together the resources are now available for the numerical simulation of the flow about complex three dimensional aerospace configurations. Yet, in many ways the discipline of CFD is still in its infancy. Major decisions concerning its future direction need to decided and many impeding obstacles must be overcome before its evolution into a mature discipline for solving the equations of compressible viscous flow.

While the battles of the past concerning CFD centered on shock fitting verves shock capturing algorithms or finite element verses finite difference procedures, present and future battles will be concerned with structured multi-block element grids verses unstructured single element grids, indirect relaxation or approximately factored procedures verses direct solution procedures, and turbulence modeling verses direct simulation of turbulent phenomena. These items will be discussed in the paper with regard to the development of CFD for applications to aerospace problems of current and future engineering interest. It may appear premature to consider the item of direct simulation of turbulence as an alternative today. Indeed it is with respect to the prediction of turbulent flow past full aerospace configurations at flight conditions for the foreseeable future; but there are flows at considerably lower Reynolds numbers past elements of such configurations that could be conceivably attacked using this approach during the 1990’s and the CFD community should perhaps be gearing up for it now.