The flow over the high pressure blades of a gas turbine is disturbed by wakes and shock waves from the nozzle guide vanes upstream. These disturbances lead to increased heat transfer to the blade surfaces, the accurate prediction of which is an essential stage in the design process.
The Oxford Rotor experiment consists of a highly instrumented 0.5 m diameter shroudless turbine which is supplied with air from a piston tube during the 200 ms run time and simulates realistic engine Mach and Reynolds numbers. Previous experiments have measured blade surface pressures and heat transfer rates, and compared them with similar data from linear cascades.
The present work is designed to enable the accuracy of rotation terms in computational fluid dynamics (CFD) calculations to be assessed, by providing heat transfer data from the rotating frame in the absence of wakes. Flow disturbances were avoided by removing the nozzle guide vanes, the correct angle of incidence onto the rotor blades being achieved by rotating the rotor in the reverse direction. Blade surface heat fluxes were measured using thin film gauges. In the absence of the usual blade-passing fluctuations, the root-mean-square fluctuation in heat flux was typically only 7% of the DC level.
Nusselt numbers are compared with cascade data and CFD predictions from both a three-dimensional viscous Navier-Stokes equation solver and a two-dimensional boundary layer prediction. The low inlet turbulence level produced a long laminar region on the suction surface followed by sudden transition. CFD predictions of Nusselt number on this surface were very sensitive to the choice of boundary layer state, and the experimental level was approximately mid-way between predictions with a transitional intermittency distribution and those with a turbulent distribution. On the pressure surface the levels were approximately 25% below predicted levels, and possible reasons for this are considered.
