The turbulent boundary layer

Summary

Two experimental studies of turbulent boundary layers are described. The first was begun as part of a search for more accurate knowledge of the structure and motion of the large-scale eddy motions that appear to control the rate of entrainment of non-turbulent fluid and which may also play a dominant part in the constant-stress layer. The experimental technique depends on the measurement of mean products of the various components of the fluctuation velocities at two separated points in the flow, and a comparison of the measurements with those to be expected from the presence of hypothetical simple eddy structures. Within the inner, constant-stress layer, the measurements are consistent with the irregular occurrence of nearly two-dimensional jets, directed outwards from the viscous layer and of comparatively long and indeterminate length in the direction of mean flow. In the outer part of the layer the dominant large-scale motion appears to be jets of turbulent fluid which may arise from the release of the anisotropie Reynolds stresses set up by the shearing motion at the free boundary. The connection between these results and the hypothesis of wall similarity (“law of the wall”) is discussed.

The second part is concerned with the boundary layer on a flat plate of finite aspect ratio, in particular the flow in the immediate neighbourhood of the free edge parallel to the stream and its effect on the boundary layer as a whole. The measurements include mean velocities, turbulent intensities and local surface friction. For fully turbulent flow at effective Reynolds numbers over one million, well-developed eddies are found with their axes parallel to the edge and one on each side of the plate. These are caused by the cross-flow which must arise if a boundary layer is bounded by a free edge as a consequence of the well-known inequality of the normal Reynolds stresses in a boundary layer. Measurements at lower Reynolds numbers show that the laminar flow near the edge becomes unstable at a lower Reynolds number than the rest of the layer and may become turbulent before the remainder of the flow is in any way unstable. The interaction between the two parts of the flow has some very interesting features and it is hoped that these experiments may help in the interpretation of recent measurements of “turbulent” skin friction on flat plates at very low Reynolds numbers.