Contribution by ICSTM: Modelling generic 2d and 3d separated flows using anisotropy-resolving turbulence closures

Abstract

The contribution of Imperial College London has been targeted, principally, towards advanced anisotropy-resolving turbulence closures and their application to a range of physically complex “generic” flows in which 2d and 3d separation from continuous surfaces is the main linking feature. The closures included various non-linear eddy-viscosity models and explicit algebraic Reynolds-stress models as well as two full Reynolds-stress-transport models. The range of flows investigated extended from 2d separation in an asymmetric diffuser to the highly complex 3d flows around a 3d hill and a generic car body. For all geometries but one — the NACA 0012 aerofoil at moderate incidence — representatives from all three anisotropy-resolving model categories were investigated. One major conclusion derived from the studies is that observations made in statistically 2d flows often do not translate to much more complex 3d flows that feature massive separation and strong vortical transverse motion. While anisotropy-resolving closures may give a superior representation of the response of the stress field to different types of strain, they cannot account (in common with all other RANS models) for the complex dynamics associated with large-scale unsteadiness, and it appears that this remains a major obstacle to a satisfactory predictive performance of these models in highly separated conditions. In contrast, attached and mildly separated flows benefit from the much stronger fundamental rigour offered by elaborate anisotropy-resolving closures, but here too, the predictive performance varies greatly, and much depends on the precise details of the closure and its calibration.