Analysis of nonlinear gas turbine models using influence coefficients

Abstract

The limited availability of gas turbine data, especially fault data, and the high costs and risks of experimenting with faults in test benches cause the lack of data to form a representative fault classification for gas turbine diagnostics. These circumstances explain the need of models that can simulate the faults. The utility of the simulated data for the diagnostics depends on the accuracy of fault simulation at different operating modes. The present paper analyses random errors of and an operating conditions influence on a gas turbine fault description. The analysis is applied to the thermodynamic models of a turboshaft and a turbofan of the well-known commercial software GasTurb 12. Big data containing measured quantities with the influence of fault parameters and operation conditions were generated with this software. Then the matrixes that determine the influence of faults and operating conditions were calculated to analyze the accuracy and behavior of the models. The results show that the engine models are accurate enough and the influence of operation conditions on the fault action is significant in contrast to some other engine models.