Experimental and numerical investigation on flow and heat transfer in large-scale, turbine cooling, representative, rib-roughened channels

This paper deals with the application of a three-dimensional Navier-Stokes solver for the prediction of steady viscous compressible flow and heat transfer in a square channel with one rib-roughened wall. The computation results are compared with detailed experiments carried out at the von Karman Institute. The two-dimensional computations agree rather well with the experiments for the prediction of the aerodynamics, even if the recirculation length is overestimated. In this case, a k-l turbulence model seems to be sufficient. However, heat transfer between the ribs is poorly matched except when a thermal ASM (algebraic stress model) turbulence model (GGDH, or generalized gradient diffusion hypothesis), which computes the u_iθ (velocity-temperature) correlations by algebraic equations, is used. The three-dimensional computations capture the correct position of the reattachment point with the k-l turbulence model. It is nevertheless necessary to use the ASM turbulence model to find vortices turning the correct way in the cross-sections. These are indeed secondary flows of the second kind which are mainly due to turbulence anisotropy when the ribs are inclined at 90° to the flow direction.