A Systematic Method of Sensor Selection for Gas Turbine Performance Degradation Estimation

Abstract As engine cycle increases, performance degradation becomes inevitable due to factors such as blade deformation, corrosion, and wear in engine components. Accurate estimation of gas turbine performance degradation is therefore critical to ensure safe operation and enable condition-based maintenance decisions. However, key parameters such as flow capacity and component efficiency degradation are not directly measurable and are typically estimated based on data from a limited number of gas path sensors. To date, there has been a lack of systematic research on the impact of sensor selection in estimating gas turbine performance degradation in the open literature. This gap has motivated the present study. This paper introduces an innovative approach utilizing the active subspace method for sensor selection. This method identifies the most influential directions in a high-dimensional space along which the degradation varies most. The effectiveness of the method was validated using simulation data from the Advanced Geared Turbofan (AGT30), a 30,000 lbf thrust engine representative of civil aviation. All simulations were performed using the open-source Toolbox for the Modeling and Analysis of Thermodynamic Systems (T-MATS). Results show that the active subspace method is a robust and efficient tool for identifying the principal direction of variation in degradation. A universal sensor set for estimating all component degradation or specific reduced sensor sets for individual degradation parameters can be expected. Lastly, measurement uncertainties must be considered when evaluating confidence in the estimated degradation.