Performance Characteristic Modeling for 2D Compressor Cascades

Abstract

Abstract The flow field in a compressor cascade is very complex owing to the highly 3D, turbulent, and viscous properties. However, in the through-flow analysis method, the viscosity effects are taken into account using empirical models. These models were based on experimental results for early blades. However, the blade types in modern compressors are quite different from those in older compressors. Therefore, it is not possible to predict the performance of modern compressors using the old empirical models. In this study, several multiple circular-arc (MCA) blades commonly used in modern compressors were simulated. After the simulations, a database of the cascade performance was built. Based on this database, some new models were established to predict the performance of modern cascades using regression analysis methods and an artificial neural network (ANN) method. The accuracy of all these new models is high enough for use in engineering applications.