A key technology for the application of LES to high–Re flows of engineer-
ing interest is an appropriate wall modeling strategy bridging the near–wall
region and thus avoiding the expensive DNS–like resolution. Standard wall
models such as Schumann’s model [1] work well for attached flows, e.g. plane
channel flow, where the averaged quantities required can be calculated. The
model reads: τ w = [ |τ w | / |v tan| ] v tan. Hence, a phase coincidence of the
instantaneous wall shear stress τ w and the tangential velocity v tan at the
wall–nearest grid point is assumed. More complex 3–D flows with large pres-
sure gradients or local separation are not well reproduced. For such flows the
mean values required are not accessible and the validity of the law of the
wall used is questionable. The model of Werner and Wengle [2] (WW) reads
τ w = f (|vtan|) vtan where f (|vtan|) is expressed by a law of the wall applied
directly to the instantaneous quantities. Therefore, this model is also not able
to predict separated flows correctly. The main reason is the dependence of
τ w on additional quantities such as the pressure gradient ∇p. Thus, attempts
were made to take ∇p into account, e.g. by Hoffmann and Benocci [3].