On the definition of the secondary flow in three-dimensional cascades

Abstract

Abstract

The present article proposes a novel methodology to evaluate secondary flows generated by the annulus boundary layers in complex cascades.

Unlike two-dimensional (2D) linear cascades, where the reference flow is commonly defined as that measured at midspan, the problem of the reference flow definition for annular or complex 3D linear cascades does not have a general solution up to the present time.

The proposed approach supports secondary flow analysis whenever exit streamwise vorticity produced by inlet endwall boundary layers is of interest. The idea is to compute the reference flow by applying slip boundary conditions at the endwalls in a viscous 3D numerical simulation, in which uniform total pressure is prescribed at the inlet. Thus the reference flow keeps the 3D nature of the actual flow except for the contribution of the endwall boundary layer vorticity. The resulting secondary field is then derived by projecting the 3D flow field (obtained from both an experiment and a fully viscous simulation) along the local reference flow direction; this approach can be proficiently applied to any complex geometry.

This method allows the representation of secondary velocity vectors with a better estimation of the vortex extension, since it offers the opportunity to visualize also the region of the vortices, which can be approximated as a potential type. Furthermore, a proficient evaluation of the secondary vorticity and deviation angle effectively induced by the annulus boundary layer is possible.

The approach was preliminarily verified against experimental data in linear cascades characterized by cylindrical blades, not reported for the sake of brevity, showing a very good agreement with the standard methodology based only on the experimental midspan flow field.

This article presents secondary flows obtained by the application of the proposed methodology on two annular cascades with cylindrical and 3D-designed blades, stressing the differences with other definitions. Both numerical and experimental results are considered.