Comparison of a Conventional Thermal Analysis of a Turbine Cascade to a Full Conjugate Heat Transfer Computation

Abstract

Recent development in commercial CFD codes offers possibilities to include the solid body in order to perform conjugate heat transfer computations for complex geometries. The current paper aims to analyse the differences between a conjugate heat transfer computation and conventional uncoupled approaches where a heat transfer coefficient is first derived from a flow solution and then taken as boundary condition for a thermal conduction analysis of the solid part. Whereas the thermal analyses are done with a Rolls-Royce in-house finite element code, the CFD as well as the conjugate heat transfer computation are done using the new version 8 of the commercial code Fine Turbo from Numeca International. The analysed geometry is a turbine cascade that was tested by VKI in Brussels within the European FP6 project AITEB 2. First, the paper presents the aerodynamic results. The pure flow solutions are validated against pressure measurements of the cascade test. Then, the heat transfer from flow computations with wall temperature boundary conditions is compared to the measured heat transfer. Once validated, the heat transfer coefficients are used as boundary condition for three uncoupled thermal analyses of the blade to predict its surface temperatures in a steady state. The results are then compared to a conjugate heat transfer method. Therefore, a mesh of the solid blade was added to the validated flow computation. The paper will present and compare the results of conventional uncoupled thermal analyses with different strategies for the wall boundary condition to results of a conjugate heat transfer computation. As it turns out, the global results are similar but especially the over-tip region with its complex geometry and flow structure and where effective cooling is crucial shows remarkable differences because the conjugate heat transfer solution predicts lower blade tip temperatures. This will be explained by the missing coupling between the fluid and the solid domain.