Computational study of combustor dilution flow interaction with turbine vanes

Higher efficiency and greater performance in gas-turbine engines can be achieved by increasing the combustion temperature but is limited by durability concerns for downstream hardware. Large-scale dilution cooling flows can be injected in the combustor to promote mixing and reduce incoming temperatures, but this generates spatial and temporal nonuniformities for the first row of turbine vanes. The lack of uniformity is exacerbated as combustor designs are reduced in length to reduce overall engine weight, but little is known about how the vane surface conditions are affected. This work computationally modeled the flow in a previously studied nonreacting combustor simulator, using both time-average (steady Reynolds-averaged Navier–Stokes, RANS) and time-dependent (delayed detached-eddy simulation, DDES) analyses. The effect of several dilution hole configurations on the adiabatic wall temperature and heat transfer coefficient of the first vane were studied in the time-average approach. Configurations in which the dilution jets were closer to the vane resulted in some vane surface cooling but also significant flow nonuniformity at the turbine inlet, high vane temperature gradients, and increased heat flux to the vane. Heat transfer coefficient augmentation on the vane pressure side was more than three times higher than without upstream dilution, for dilution jets close to the vane. The time-dependent DDES analysis predicted a larger degree of mixing and more uniform vane temperatures, relative to the RANS analysis, but also indicated anisotropy of the turbulence entering the turbine, which cannot be correctly captured by a RANS approach. Refinement of the time-dependent analysis in a region around the dilution jets did not significantly change the turbine inlet flowfield and had only a minor impact on the predicted vane surface temperature. Overall, the results indicate that nonuniform flow in the combustor can significantly impact vane temperature but should be modeled using scale-resolving simulations for best accuracy.