COMPARISON OF TWO-EQUATION TURBULENCE MODELS FOR PREDICTION OF HEAT TRANSFER ON FILM-COOLED TURBINE BLADES

Abstract

A three-dimensional Navier-Stokes code has been used to compute the heat transfer coefficient on two film-cooled turbine blades, namely, the VKI rotor with six rows of cooling holes, including three rows on the shower head, and the CSX vane with nine rows of holes, including five rows on the shower head. Predictions of heat transfer coefficient at the Made surface using three two-equation turbulence models, specifically, Coakley's q-ω model, Chien's k-ϵ, and Wilcox's k-ω model with Mentor's modifications, have been compared with the experimental data of Camci and Arts [l] for the VKI rotor, and of Hylton et al. [2] for the C3X vane along with predictions using the Baldwin-Lomax (B-L) model taken from Garg and Gaugier [3]. It is found that for the cases considered here the two-equation models predict die blade heat transfer somewhat better than the B-L model except immediately downstream of the film-cooling holes on the suction surface of the VKI rotor, and over most of the suction surface of the C3X vane. However, all two-equation models require 40% more computer core than the B-L model for solution, and while the q-ω and k-ϵ models need 40% more computer time than the B-L model, the k-ω model requires at least 65% more time because of the slower rate of convergence. It is found that the heat transfer coefficient exhibits a strong spanwise as well as streamwise variation for both blades and all turbulence models.