Abstract
Aerodynamic loss models play an important role in the preliminary design of axial turbines. However, prediction accuracy and parameter sensitivity of traditional models being used for decades are problematic. Thus, this paper proposed a new loss model for predicting turbine aerodynamic performance at the design point. In order to construct the model, 228 turbine cascades for calibration were designed and evaluated by computational fluid dynamics (CFD). Based on massive CFD data, correlation of parameters and loss coefficients were investigated. Finally, new loss model expressions were built. The accuracy of the proposed novel model was verified by CFD for nine turbines including 15 stages. Numerical results show that the average efficiency deviation for the new model is 0.48% for stages and 0.009 for entropy loss coefficient of the blade rows. Compared with the Craig and Cox model and Kacker and Okapuu model, the presented new model performs better in predicting aerodynamic loss, especially in predicting the effect of parameters on the losses.
