Experimental investigation of disc cavity leakage flow and hub endwall contouring in a linear rotor cascade.

Abstract

Experimental and computational results which document mixing of passage flow and leakage flow in a rotor stage of a high pressure turbine are presented. Of specific focus are the effects of hub endwall contour geometries on mixing of the two flows and on film cooling coverage by the leakage flow over the endwall. The understanding of fluid physics in this area has received increased interest recently due to higher endwall heat loads as a result of new combustor designs. The setting is a linear, stationary cascade which represents many features of the actual engine, such as geometry, Reynolds number, approach flow turbulence level and scale, and leakage mass flow rates. Rotation, density gradient, and upstream airfoil row effects are not represented. Two hub endwall geometries which give quite different acceleration profiles in the airfoil row entry plane region are examined. The flow field in the leakage flow delivery plenum, important to the mixing process, is characterized by measurements and computation. The effects of leakage flow injection on the passage aerodynamic losses are also measured and computed. The loss pattern at the passage exit shows the effects of boundary layers on the pressure surface, the suction surface, and the two endwalls. Passage secondary flow features, such as remnants of the passage, horseshoe, and corner vortices are visible in the exit passage loss data. The effects of changes in leakage flow injection rate on the losses are found to be minimal for the cases studied. Measurements of adiabatic effectiveness on the contoured endwall show coverage only over the upstream portion of the passage, with concentration on the suction side. The effects of the horseshoe and corner vortices on mixing of the leakage and passage flows are evident in the effectiveness pattern. Computed effectiveness distributions show similar trends to those seen in the measurements; however, measured effectiveness values are generally lower than computed values, indicating the more rapid dissipation of turbulent transport in the computations than in reality. Comparison of effectiveness distributions shows that the contoured endwall geometry referred to as the “dolphin nose” leads to better overall film cooling coverage on the endwall.