Published online by Cambridge University Press: 03 February 2016
The effect of tip geometry on discharge coefficient and heat transfer is investigated both experimentally and numerically using idealised models of an unshrouded rotor blade. A flat tip was compared with two squealer-type geometries (a cavity and suction-side squealer) under the transonic conditions expected in the gas turbine engine. Heat transfer measurements were performed using a transient liquid crystal technique while a duplicate test section was used for measuring the pressure field. Computations were carried out using an unstructured, fully compressible, three-dimensional RANS (Reynolds averaged Navier Stokes) solver. Initial computations performed using a low Reynolds number k-ε model demonstrated the inability of the model to predict the Nusselt number with reasonable accuracy. Further computations performed using a low Reynolds number k-ω model improved the predictions dramatically. The computed discharge coefficient and the average Nusselt number over the blade tip agreed well with the experiments. Three upstream-total to exit-static pressure ratios were used to create a range of engine-representative Mach numbers. Both experimental and numerical studies at the lower pressure ratio of 1·3 (exit Mach number ~ 0·65) established the cavity geometry as the best performer from an aerodynamic perspective by reducing the discharge through the tip. However, from the heat transfer perspective, both the peak Nusselt number and the average heat transfer to the tip were higher than the flat tip. At the higher pressure ratios of 1·85 and 2·27 (corresponding to exit Mach numbers ~ 0·98 and 1·12) the discharge coefficient and heat transfer to the tip increases. This paper explores the fluid dynamics associated with these flows and shows that the highest heat transfer is caused by reattachment and flow impingement. The fluid dynamic computations provide insight into the experimental measurements and were successfully compared with simple analytical models.