Boundary layer development on low aspect ration compressor blading of large camber

Abstract

Modern methods of turbomachine through-flow calculation need estimates of flow blockage and energy loss along the stages. The latter estimates are made using boundary layer prediction schemes, which need to be tested under the conditions of three-dimensional flow and large blade surface curvature encountered in turbomachines. This thesis describes a study of boundary layer development on compressor blades of low aspect ratio and high camber, which represent a fairly extreme case of the type of flow involved.

Detailed measurements of boundary layer growth have been made in the presence of strong lateral contraction in a cascade of compressor blades of unit aspect ratio and a camber of fifty degrees. The contraction makes the boundary layer very much thicker than it would have been under two-dimensional conditions with the same pressure distribution. On the other hand, separation seems to be delayed, and very high values of shape factor are reached before turbulent separation takes place. In spite of the strong lateral contraction, the measured velocity profiles at mid-span correlate well with existing two-dimensional profile families.

As found by previous workers, the momentum thickness of the laminar boundary layer near the blade leading edge agrees well with prediction on a two-dimensional basis, but laminar flow persists well downstream of the predicted laminar separation point. This is confirmed by the measured velocity profiles as well as measurements of turbulent fluctuations, u u

. Transition agrees well with the two-dimensional correlation of Michel. It is noticed that the shape factor at transition is very close to 1.4 in every instance.

Preston tube measurements of skin friction give lower values than those estimated from the measured velocity profiles using various techniques. There is indirect evidence to suggest that the Preston tube reads low under the conditions of the experiment.

The best-rated two-dimensional turbulent boundary layer methods of the Stanford Conference and some later developments were evaluated for their ability to predict the experimental boundary layers. As expected, the momentum thickness growth is heavily underestimated, especially as the trailing edge is approached. Allowance for the lateral contraction improves the prediction, but due to the heavily three-dimensional nature of the flow, it is very difficult to determine what exactly is the effective contraction at the portion of the surface on which the boundary layer is measured. There does not seem to be a noticeable difference between the various computing methods used, schemes not allowing for the upstream history of the flow faring as well as the methods that do, as far as prediction of the momentum thickness is concerned.