| The endwall region in a turbine cascade passage is technologically important because it contributes a significant portion of the losses in a turbomachine. An experiment and a computation were carried out simultaneously to investigate how reliable a state-of-the-art Navier-Stokes code with a simple one-equation turbulence model is in predicting this complex flowfield. For the experiment, a large-scale, three-passage linear turbine cascade was built. The test rig attached to the end of a 30" x 30" cross-section, low-speed blower windtunnel. The experiment was conducted at a Reynolds number (based on axial chord and exit velocity) of 900,000. The mean flow within a representatively periodic passage was measured using Preston tubes, static pressure taps, three- and five-hole pressure probes and surface oil flow visualization. Total pressure loss contours, velocity magnitude contours, secondary flow vectors, and streamwise vorticity contours were plotted at nine planes.;An incompressible, three-dimensional, Reynolds-averaged Navier-Stokes code (INS3D-UP) was used to compute the flowfield. To account for the turbulence, the Baldwin-Barth one-equation model was employed. INS3D-UP computed the qualitative features of the flowfield very well, even with a fairly coarse grid. Quantitatively, the code predicted the main flow away from the endwall region very well, again even with a fairly coarse grid. In the endwall region, however, the results of the code differed significantly from those measured in the experiment. These differences were mainly due to the fact that the computation significantly overpredicted the strength of the secondary flows in the passage, as indicated by the greater skewing of the endwall boundary layer and the tighter roll-up and much greater strength of the passage vortex. |
