The computation of turbulent flows of industrial complexity by the finite element method—progress and prospects

Abstract

Abstract This paper is an expanded version of that delivered at the recent Sixth International Symposium on Finite Element Methods in Flow Problems, Antibes, France. It begins by reviewing the role of the finite element method (FEM) in turbulent flow simulation during recent years. The difficulties in incorporating sufficiently general descriptions of turbulence (i.e. two‐equation models) into successful finite‐element‐based Navier‐Stokes codes are examined and analysed in some depth. Current progress by various workers in overcoming these difficulties is reviewed and, by concentrating on one particular approach, it is demonstrated that the FEM has now matured into a powerful and flexible tool for solving two‐dimensional turbulent flows of industrial complexity. The applications presented highlight those features which render the FEM attractive in this field (viz., minimal false diffusion, arbitrary local refinement, boundary fitting capabilities and non‐structured grids). Finally, the prospects and challenges for the future are briefly discussed. In particular, the urgency and difficulty of constructing a competitive three‐dimensional capability which preserves these features is examined.