Compressible turbulent flows: Aspects of prediction and analysis

Abstract

AbstractCompressible turbulent flows are an important element of high‐speed flight. Boundary layers developing along fuselage and wings of an aircraft and along engine compressor and turbine blades are compressible and mostly turbulent. The high‐speed flow around rockets and through rocket nozzles involves compressible turbulence and flow separation. Turbulent mixing and combustion in scramjet engines is another example where compressibility dominates the flow physics. Although compressible turbulent flows have attracted researchers since the fifties of the last century, they are not completely understood. Especially interactions between compressible turbulence and combustion lead to challenging, yet unsolved problems. Direct numerical simulation (DNS) and large‐eddy simulation (LES) represent modern powerful research tools which allow to mimic such flows in great detail and to analyze underlying physical mechanisms, even those which cannot be accessed by the experiment. The present lecture provides a short description of these tools and some of their numerical characteristics. It then describes DNS and LES results of fully‐developed channel and pipe flow and highlights effects of compressibility on the turbulence structure. The analysis of pressure fluctuations in such flows with isothermal cooled walls leads to the conclusion that the pressure‐strain correlation tensor decreases in the wall layer and that the turbulence anisotropy increases, since the mean density falls off relative to the incompressible flow case. Similar increases in turbulence anisotropy due to compressibility are observed in inert and reacting temporal mixing layers. The nature of the pressure fluctuations is however two‐facetted. While inert compressible mixing layers reveal wave‐propagation effects in the pressure and density fluctuations, compressible reacting mixing layers seem to generate pressure fluctuations that are controlled by the time‐rate of change of heat release and mean density variation as sources in the pressure Poisson equation rather than wave propagation phenomena. This opens new challenges to turbulence modellers and underlines the need to treat compressibility effects via the full Reynolds stress equations.