CFD Based External Heat Transfer Coefficient Predictions on a Transonic Film-Cooled Gas Turbine Guide Vane : A Computational Fluid Dynamics Study on the Von Karman Institute LS94 Test Case

Abstract: The turbine inlet guide vanes of a gas-turbine are subjected to extreme hot gas temperatures which increases the risk of mechanical failure and overall reduces the component lifespan. Hence, it is of great interest for gas-turbine manufacturers to establish methods for accurately estimating the temperature distribution along the vane surface. Due to the three-dimensional nature of turbine flow, it is of interest to establish Computational Fluid Dynamics (CFD) methodology which capture these three-dimensional effects. This thesis is one in a collection of theses conducted at Siemens Energy AB on the subject. Previous studies have investigated and validated the implementation of RANS simulations on non-cooled turbine vanes and endwalls. In this study, the focus is on studying a film cooled vane and establishing one RANS as well as one hybrid modelling strategy for heat transfer coefficient (HTC) predictions. The HTC prediction capabilities are compared and validated against experimental data presented in the doctoral thesis by Fabrizio Fontaneto on the LS94 vane at Von Karman Institute. The chosen RANS modelling method was the Shear Stress Transport (SST) k-ω turbulence model, with γ-Reθ transition modelling, based on the findings by Enico (2021) and Daugulis (2022). The model proved capable in estimating the HTC well on mainly the suction side of the vane. The pressure side HTC was largely under-predicted, a common issue with the SST model also seen in the previous theses as well as the hybrid simulations. The strength of the SST k-ω turbulence model, with γ-Reθ transition modelling, is in accurately capturing the HTC magnitude, most likely due to the well-predicted turbulence intensity decay at the inlet. However, it largely under-predicts the HTC along the suction side film-coolant layer, implying that it may be over-estimating the film-cooling capabilities. The hybrid model chosen was the Scale Resolving Hybrid (SRH) model, with underlying RANS SST k-ω. Compared to RANS, hybrid results were under-estimated, seemingly offset from the experimental data by a constant 200 units along the entire vane midspan. This is likely due to the inaccurate turbulence intensity presented in the SRH simulations, which decays quickly along the inlet compared to RANS and experimental data. Yet still, the hybrid model showed potential in capturing certain results not seen with RANS, such as the secondary flow effects by the vane endwalls, as well as arguably capturing the general HTC trend at midspan seen in the experimental data. Additionally, the section of severely under-predicted HTC by the suction side film-coolant seen with RANS is not present in the hybrid results. Although the hybrid model has proven promising in many aspects, in its current state it is not a viable method for HTC predictions due to its general under-prediction of HTC. Largely, the authors suspect this is due to the undesirably coarse mesh around the cooling holes, which leads to RANS computation in regions where SRH is desired. Thus, improvements would need to be made to the model, where, for example, implementing a zonal hybrid RANS-LES model would be an option. Considering the hybrid model in its current state, RANS is the preferred method, especially when considering the greater computational cost and the labor associated with hybrid simulations which were experienced during this study. In conclusion, it is evident that the correct capture of inlet turbulence intensity decay as well as suitable mesh refinement by the cooling holes are crucial for obtaining the correct magnitudes of HTC, and thus, the capture of it should be of utmost priority in future work within the field.