An Improved Correlation for Turbine Endwall Loss

Abstract To develop high performance turbomachinery, we need accurate preliminary design correlations to guide us towards optimal system and component configurations. Endwall (or secondary) loss has a significant impact on turbine efficiency, but remains challenging to predict. Most existing correlations are based on well-conditioned linear cascade experiments, but these generally rely on limited databases and do not include the effect of turbine-realistic inlet conditions, which can increase loss significantly. This paper develops a new correlation to help address these deficiencies. The new method is physics-based and tuned to a large database of linear cascade computations with varying inlet conditions. Endwall loss is decomposed into Wetted Area Loss and Secondary Flow Loss. The Wetted Area Loss represents dissipation in the endwall boundary layers, in a similar manner to the approach of Denton [1]. The Secondary Flow Loss correlates with the secondary vorticity predicted by classical theory, e.g. Hawthorne [2], and is strongly dependent on the thickness and shape of the incoming endwall boundary layer. Validation with literature cascade experiments shows that the new correlation reduces the average error by > 30% compared to the best existing method, which was fit to these data, and more accurately predicts sensitivity to design parameters. The new method predicts a realistic increase in loss when moving from cascade to turbine-like inlet conditions, and thus helps to reconcile the historical mismatch between endwall losses in cascades and real turbines.