In this paper a new analysis is proposed for the driving mechanisms and the
statistics for turbulent boundary layers at very high Reynolds numbers. It
differs from theories for moderate to low Reynolds numbers and is based on
the results of (linear) rapid distortion theory, and both laboratory and f
ield experimental data. The large-scale eddy structure near the wall in bou
ndary layers is distorted in several ways: by the strong mean shear, by the
blocking of the normal velocity component and by the moving internal shear
layers produced by large eddies as they impinge and scrape along the wall.
Elongated streamwise vortices are formed with length scales that are sever
al times the boundary layer height. An approximate stability argument sugge
sts that if the Reynolds number for the turbulence, Re-tau >> 10(4), these
internal layers are fully turbulent and that the large eddies can burst upw
ard where the vortical eddies interact. The forms of the main statistical q
uantities, such as variances, spectra, length scales, are derived in terms
of outer layer quantities using surface similarity and inhomogeneous linear
theory. These 'top-down' eddy-impingement, inner-layer/eddy-interaction/ej
ection mechanisms at very high Reynolds number are sensitive to changes in
surface conditions and to variations in pressure gradients. They may theref
ore require different techniques for their control from those used at lower
Reynolds number when boundary layers are driven by 'bottom-up' instability
/surface-interaction mechanisms. Furthermore, accurate numerical modelling
of boundary layers at high Reynolds number requires resolving surface proce
sses at very fine resolution. By inference, it is likely that there is some
residual 'top-down' influence, even at low Re-tau. (C) 2000 Editions scien
tifiques et medicales Elsevier SAS.