The aerodynamic performance of a turbine vane was measured in a linear cascade. These measurements were conducted for exit-true chord Reynolds numbers between 150,000 and 1,800,000. The vane surface rms roughness-to-true chord ratio was approximately 2 × 10−4. Measurements were made for exit Mach numbers between 0.3 and 0.9 to achieve different loading distributions. Measurements were made at three different inlet turbulence levels. High and intermediate turbulence levels were generated using two different blown grids. The turbulence was low when no grid was present. The wide range of Reynolds numbers was chosen so that, at the lower Reynolds numbers the rough surfaces would be hydraulically smooth. The primary purpose of the tests was to provide data to verify CFD predictions of surface roughness effects on aerodynamic performance. Data comparisons are made using a two-dimensional Navier-Stokes analysis. Both two-equation and algebraic roughness turbulence models were used. A model is proposed to account for the increase in loss due to roughness as the Reynolds number increases.

Comparisons with experimental heat transfer and surface pressures were made for seven turbine vane and blade geometries using a quasi-three-dimensional thin-layer Navier-Stokes analysis. Comparisons are made for cases with both separated and unseparated flow over a range of Reynolds numbers and freestream turbulence intensities. The analysis used a modified Baldwin-Lomax turbulent eddy viscosity model. Modifications were made to account for the effects of: 1) freestream turbulence on both transition and leading edge heat transfer; 2) strong favorable pressure gradients on re-laminarization; and 3) variable turbulent Prandtl number on heat transfer. In addition, the effect on heat transfer of the near-wall model of Deissler is compared with the Van Driest model.

Results are shown for a three-dimensional Navier-Stokes analysis of both the flow and the surface heat transfer for turbine applications. Heat transfer comparisons are made with the experimental shocktunnel data of Dunn and Kirn, and with the data of Blair for the rotor of the large scale rotating turbine. The analysis was done using the steady-state, three-dimensional, thin-layer Navier-Stokes code developed by Chlrna, which uses a multistage Runge-Kutta scheme with implicit residual smoothing. An algebraic mixing length turbulence model is used to calculate turbulent eddy viscosity. The variation in heat transfer due to variations in grid parameters is examined. The effects of rotation, tip clearance, and inlet boundary layer thickness variation on the predicted blade and endwall heat transfer are examined. Nomenclature Cp-Pressure coei_cient (P-P{N)/(PI_N-PSXIT) P-Pressure Re-Unit Reynolds number C-Chord s Fractionalsurfacedistance St-Stanton number based on inletconditions z Fractionalchordwise distance y Distance normal to surface _+-Normalized distancefrom surface 6 Full boundary layer thickness u Kinematic viscosity p Density fMember AIAA This paper isdeclared a work of the U.S. Government and isnot subjectto copyrightprotectionin the United States. Subscripts ¢-Gas total IN-Blade row inlet EXIT " Blade row hub exit w Wall 1-First gridlinefrom surface Superscript ' Total Form Approved REPORT DOCUMENTATION PAGE I OMB No. 0704-0188 i Public reporting burden "for this collection of information is estimated to average i hour per response, including the time for reviewing instructions, searching existing data sources, gathering and maintaining the data needed, and completing and reviewing the collection of information. Send comments regarding this bu_en est_ate or any other aspect o! this collection of information, including suggestions for reducing this burden, to Washington Headquatlers Services, Directorate for information Operations and Reports,

The use of a Navier–Stokes analysis to predict the change in turbine efficiency resulting from changes in blade surface roughness or incidence flow angles is discussed. The results of a midspan Navier–Stokes analysis are combined with those from a quasi-three-dimensional flow analysis code to predict turbine performance. A quasi-three-dimensional flow analysis code was used to determine turbine performance over a range of incidence flow angles. This analysis was done for a number of incidence loss models. The change in loss due to changes in incidence flow computed from the Navier–Stokes analysis is compared with the results obtained using the empirical loss models. The Navier–Stokes analysis was also used to determine the effects of surface roughness using a mixing length turbulence model, which incorporated the roughness height. The validity of the approach used was verified by comparisons with experimental data for a turbine with both smooth and rough blades tested over a wide range of blade incidence flow angles.

Midspan aerodynamic measurements for a three vane-four passage linear turbine vane cascade are given. The vane axial chord was 4.45cm. Surface pressures and loss coefficients were measured at exit Mach numbers of 0.3, 0.7, and 0.9. Reynolds number was varied by a factor of six at the two highest Mach numbers, and by a factor often at the lowest Mach number. Measurements were made with and without a turbulence grid. Inlet turbulence intensities were less than 1% and greater than 10%. Length scales were also measured. Pressurized air fed the test section, and exited to a low pressure exhaust system. Maximum inlet pressure was two atmospheres. The minimum inlet pressure for an exit Mach number of 0.9 was one-third of an atmosphere, and at a Mach number of 0.3, the minimum pressure was half this value. The purpose of the test was to provide data for verification of turbine vane aerodynamic analyses, especially at low Reynolds numbers. Predictions obtained using a Navier-Stokes analysis with an algebraic turbulence model are also given.