The use of simple computational means to determine the performance of cascades of turbine blades is attractive because it can quickly and economically yield results that can be used for optimization of classes of blades. Fully viscous flow computations are not at the point where they are economical for use in a routine way, and most computational methods lack the resolution to determine shock losses in the transonic flow regime. There is still a need for approaches that combine computation and empiricism. We describe approaches that combine quasi-three dimensional inviscid codes and boundary layer methods for blade passage flow with empirical approaches for shock losses and base pressure deficits to predict the losses in cascades of blades. Downstream mixing losses are handled by a distributed variable approach that uses the inviscid and boundary layer results to determine the distribution of variables at the trailing edge plane. The method gives accurate predictions for the set of distinctly different steam turbine blades for which it was run, and forms the basis for the development of rule-based turbine blade design.

This content is only available via PDF.