Experiments, Theories, Models, and SimulationsExperiments, two-point theoretical closures, one-point turbulence models, and numerical simulations of turbulence all have important roles to play in advancing the field of turbulence. The information they provide is complementary, and none will replace any other. However, the evolution of simulations as a viable tool for turbulence research and prediction has changed the way in which the other tools are used.Experiments can do many things much faster than simulations and can cover ranges not attainable by simulations. Theoretical closures provide a framework for study of simple flows at high Reynolds numbers. Turbulence models are the most important tool for highly repetitive engineering analysis in moderately-demanding flows. Simulations give data that can be obtained in no other way, but only for limited conditions. Because of their general validity, simulations are going to become the primary tool for prediction of complex turbulent flows not amenable to prediction by simple turbulence models.The purpos e of this paper is to outline the general approach of direct numerical simulations and large-eddy simtflations of turbulent flows, discuss the advantages and problems of these methods, survey some of their important achievements, and forecast their future use and importance. For other recent reviews see Rogallo and Moin (1984) and Schumann and Friedrich (1987). For a recent review of CFD in general see Boris (1989).
2