| Summary: | <p>The taxonomy of flow phenomena affecting turbine blade heat transfer has been widely researched and their specific effects established. However, the literature contains minimal consideration of circumferential length scales in the freestream turbulence intensity exiting the combustor despite experimental evidence for their existence in certain combustor geometries. The amplitude in these circumferential variations is demonstrated to vary up to 60% about the circumferential average. The distribution of the heat transfer coefficient on the surface of a turbine blade responds to variation in freestream turbulence intensity. Therefore it is necessary to study the effect of this circumferential variation in turbulence on turbine blade heat transfer.</p><p>This study utilized numerical methods to provide predictions of aerothermal conditions for eight high pressure turbine geometries and a NACA airfoil operating in transonic flow regimes. Included in the validation of the numerical solver were sen- sitivity studies of solver formulation and turbulence model selection. Comparative analysis using a NACA airfoil and high pressure turbine rotor blade for three solver formulations (density-based, pressure-based coupled, and pressure-based segregated) revealed that the pressure-based solvers retained improved computational efficiency and comparable aerothermal predictive performance relative to the density-based solver. However, the pressure-based solver required extensive native control at the start of the unsteady flow solutions in order to reach convergence, therefore rendering the density-based solver as the solver formulation of choice. In the parametric study of turbulence model sensitivity, the k − ω SST and Spalart-Allmaras models out-performed the others in predicting both surface pressure and heat transfer on two rotor blade profiles. In particular the k − ω SST model performed the best, predicting absolute levels of heat transfer to within 20% when the boundary-layer was fully turbulent. The two models also successfully captured the response of rotor heat transfer to variation in freestream turbulence intensity in steady flow. These turbulence models were then studied in simulations of two rotor profiles subjected to a passing upstream cylinder representing an upstream nozzle guide vane trailing edge. The two models were capable of resolving all of the unsteady flowfield features identified in previous experiments, as well as accurately predicting the time-averaged and unsteady rotor surface pressure conditions. The k − ω SST model outperformed the Spalart-Allmaras model in predicting both the time-averaged and unsteady heat transfer for the rotor subjected to a passing upstream bar. The accuracy of the k − ω SST model prediction of time-averaged rotor heat transfer was within 20% in fully turbulent regions of the rotor boundary-layer. Following final tuning, this high-performing turbulence model was used solely in the study of non-uniform inlet turbulence conditions.</p><p>The effectiveness of the code in predicting the aerodynamics and heat transfer in a range of validation studies provided justification for a numerical analysis of engine-representative turbulence intensity variation. This thesis employs circumferential freestream turbulence intensity variation in amplitudes ranging from 30% to 60%. Hot streaks were simulated at two ratios of maximum to minimum total temperature in the circumferential direction with values of 1.1 and 1.4. Combining the two distortions in phase allowed for the assessment of clocking on vane heat transfer. Combustor burner to vane count was simulated at 1:1 and 2:1. The predictions consistently showed the dependence of vane heat transfer coefficient distributions on turbulence distortion clocking to be on the order of 20%. This effect scaled on the amplitude of the circumferential turbulence variation. Simulations on a three-dimensional vane concurred with the two-dimensional results. These turbulence distortions propagated to the downstream rotor blade row, and altered rotor heat transfer unsteadiness. Overall, the engine-representative “turbulence streak” presents a new source of heat transfer that requires further experimental confirmation, creating another design variable to consider for turbine heat transfer engineers.</p>
|
|---|