Variable Fidelity Optimization of Film Cooling Hole Arrangements on High Pressure Turbine Considering Conjugate Heat Transfer

Abstract

Conjugate heat transfer (CHT) should be taken into account when analyzing heat transfer in high pressure turbine (HPT) to improve the accuracy of the simulation because film cooling is used to reduce the convective heat transfer into the blade and temperature potential which drives conduction by coolant film extracted from holes. However, Research of film cooling hole array optimization have been conducted under the adiabatic condition due to high computation burden in CHT analysis. Moreover, optimization considering CHT has been restrictively attempted on relatively simple problem such as single hole shape optimization problem. In this study, film cooling hole arrangement is optimized by considering CHT effects. In an effort to reduce computation load, EI-based efficient global optimization (EGO) algorithm coupled with hierarchical Kriging (HK) model is implemented. There, however, have been still several ambiguities to be clarified when applying the HK model to hole array optimization problem for practical utilization. The first is the existence of optimal high to low fidelity sample ratio to reduce the computational cost. The second is whether the HK model can produce converged result irrespective of the different high-to-low fidelity sample ratio. All the analyses of the ambiguity-clarification are conducted under the adiabatic condition. As a result, HK model not only produce the consistent and reliable optimization result but also reduce the CPU time at the similar level, 40%, irrespective of the different high-to-low fidelity sample ratios. Based on the adiabatic analysis-based optimization results, CHT-based optimization is conducted with 3-level HK model. Computation time decreases by 76.45% compared to CHT-only optimization. Furthermore, the film cooling hole arrangement shows substantially different configuration from adiabatic analysis-based optimization results. Specifically, although the second array shape is similar to the adiabatic result, the third hole arrangement has curvature towards the leading edge like a parabola. This resultant hole array configuration is shown to be the best combination of inner cooling components and film cooling hole locations, which make the inner and outer coolant cover the nozzle surface as much as possible. The detailed reasons why the hole arrangements are determined are discussed by investigating the pressure, velocity contours, and streamlines. As a result, the nozzle surface temperature decreases by 49.64 K and average overall film cooling effectiveness (ϕ) increases by 0.058 compared to those of baseline. The method used in this study is promising in terms of handling highly-nonlinear or high-computing required problems.