The Effect of the Casing Movement Relative to the Blades on the Tip Leakage Loss in Axial Flow Compressors

Abstract

This paper investigates the effect of the casing movement relative to the blades on the tip leakage loss generation mechanisms by using experimental results from a linear cascade test facility, and viscous numerical results. Traverse measurements in the pitch-wise and span-wise directions are made using a five-hole Pitot tube at the inlet and exit planes of a compressor linear cascade comprising seven equally-pitched blades. The blades are two-dimensionally stacked with a cross section representing a typical rear stage rotor of a highly loaded axial-flow compressor. A moving belt, driven by a motor and a pulley system, runs linearly at constant speed under the horizontally suspended cascade to simulate the relative motion of the blade and the casing. Tip clearance can be adjusted by changing the height of the blades. The experimental results, at 2% and 4% tip clearance to blade heights, indicate that the tip leakage loss decreases when the casing is in movement. The Reynolds-averaged Navier-Stokes numerical calculations with Spalart-Almaras turbulence closure model, run with the experimental boundary conditions, agree well with the test data, especially in terms of dependencies of the leakage loss magnitude on the relative movement between the blade and the casing. It is interesting that, contrary to the tendency in the leakage loss to decrease, the computed tip leakage mass flow rate increases with moving endwall. The computations show two distinct regions of high entropy creation rate near the blade tip. The first one is located close to the blade suction surface where the leakage flow leaves the clearance gap. The second one is located further from the suction surface and the entropy creation rate in this region decreases when the casing is in movement. This paper attempts to provide a qualitative analysis of the flow mechanisms involved in the entropy generation in the second regions. Finally Computations of a high loaded rotor show that the second region identified in the static cascade may also be present in the case of rotating cascades.