Improved gas turbine diagnostics towards an integrated prognostic approach wiht vibration and gas path analysis

Abstract

The degradation of a gas turbine engine in operation is inevitable, leading to losses in performance and eventually reduction in engine availability. Several methods like gas path analysis and vibration analysis have been developed to provide a means of identifying the onset of component degradation. Although both approaches have been applied individually with successes in identifying component faults; localizing complex faults and improving fault prediction confidence are some of the further benefits that can accrue from the integrated application of both techniques. Although, the link between gas path component faults and rotating mechanical component faults have been reported by several investigators, yet, gas path fault diagnostics and mechanical fault diagnostics are still treated as separated toolsets for gas turbine engine health monitoring. This research addresses this gap by laying a foundation for the integration of gas path analysis and vibration to monitor the effect of fouling in a gas turbine compressor. Previous work on the effect of compressor fouling on the gas turbine operation has been on estimating its impact on the gas turbine’s performance in terms of reduction in thermal efficiency and output power. Another methodology often used involves the determination of correlations to characterize the susceptibility and sensitivity of the gas turbine compressor to fouling. Although the above mentioned approaches are useful in determining the impact of compressor fouling on the gas turbine performance, they are limited in the sense that they are not capable of being used to access the interaction between the aerodynamic and rotordynamic domain in a fouled gas turbine compressor. In this work, a Greitzer-type compression system model is applied to predict the flow field dynamics of the fouled compressor. The Moore-Greitzer model is a lumped parameter model of a compressor operating between an inlet and exit ii duct which discharges to a plenum with a throttle to control the flow through the compression system. In a nutshell, the overall methodology applied in this work involves the interaction of four different models, which are: Moore-Greitzer compression system model, Al-Nahwi aerodynamic force model, 2D transfer matrix rotordynamic model and a gas turbine performance engine model. The study carried out in this work shows that as the rate of fouling increases, typified by a decrease in compressor massflow, isentropic efficiency and pressure ratio, there is a corresponding increase in the vibration amplitude at the compressor rotor first fundamental frequency. Also demonstrated in this work, is the application of a Moore-Greitzer type compressor model for the prediction of the inception of unstable operation in a compressor due to fouling. In modelling the interaction between the aerodynamic and rotordynamic domain in a fouled gas turbine compressor, linear simplifications have been adopted in the compression system model. A single term Fourier series has been used to approximate the resulting disturbed flow coefficient. This approximation is reasonable for weakly nonlinear systems such as compressor operating with either an incompressible inlet flow or low Mach number compressible inlet flow. To truly account for nonlinearity in the model, further recommendation for improvement includes using a second order or two-term Fourier series to approximate the disturbed flow coefficient. Further recommendation from this work include an extension of the rotordynamic analysis to include non-synchronous response of the rotor to an aerodynamic excitation and the application of the Greitzer type model for the prediction of the flow and pressure rise coefficient at the inlet of the compressor when fouled.