Modeling of turbine mass flow rate performances using the Taylor expansion

The turbine is a key component of many equipment and systems, such as air cycle refrigeration and gas-turbine engines. Existing turbine mass flow rate models need to be improved to increase the prediction accuracy and extrapolation performance for control and diagnosis-oriented simulation. This work proposes a novel methodology for building a regression model, which makes use of the Taylor series to expand functions to deal with variables with small variation and develops a single partly empirical model to present a component performance map. With the methodology, a general regression model of mass flow rates of inward radial turbines is built. Measured data of a turbocharger turbine and a simple air cycle machine turbine are used for the regression analysis to validate the methodology and model. Model predictions agree with measured data very well, proving that the proposed methodology and the model are highly reliable. Comparison of the proposed model with the best existing model searched shows that the present model reduces the mean absolute percentage error by more than 50%, and has much better extrapolation performance as well.