Axial-Flow Compressor Model Based on a Cascade Stacking Technique and Neural Networks

This paper introduces a cascade-stacking technique for the development of a gas turbine multi-stage axial-flow compressor model. A large database of stationary and rotating cascade performance is first obtained by quasi three-dimensional CFD simulations and used to train neural networks for the prediction of cascade performance under generalized conditions. Then the model directly calculates the operating point of a compressor having known geometry characteristics, including variable inlet guide/stator vane effects, as a function of mass flow rate and rotational speed. The model can also be used as a valuable preliminary design tool, obtaining geometry characteristics by imposing flow patterns.