Gas Turbine Engine Performance Estimation and Prediction

Abstract

Modern health management approaches for gas turbine engines (GTE) aim to acquire precise information about the health state of the GTE components to optimize the maintenance decisions with respect to both the economy and safety. The task becomes more challenging for the GTE parts inaccessible to direct measurements with the available sensors of the GTE control system. This article-based thesis integrates a set of five coherent research work to address this problem.A detailed nonlinear thermodynamic model for single shaft GTEs is developed to predict the expected cycle parameters for the GTE in the healthy condition. In reality, the measured cycle parameters gradually deviate from the prediction due to performance deterioration. Physics-based performance indicators are defined based on the deviations in the measured performance parameters, compared to the respective model predictions.The indicators can effectively monitor the GTE performance deterioration in both shortterm and long-term regimes. In the next step, effect of the air humidity is taken into account to enhance the GTE model, and it is shown that the enhanced model can improve the performance monitoring by reducing the uncertainties.In order to separate the effects of different fault modes, an inference-based model is developed to predict the short-term recoverable performance deterioration due to the compressor fouling under different ambient and operating conditions. For the long-term non-recoverable performance deterioration due to the degradation mechanisms in the turbine hot section, two steps are undertaken; 1) a state estimation framework is developed for nonlinear/non-Gaussian systems with non-uniform time steps to track a degradation symptom of the turbine, i.e., loss of isentropic efficiency, using the observable performance indicators, and 2) the state estimation framework is extended to multidimensional dynamical systems with stochastic inputs for simultaneous tracking of two degradation symptoms, i.e., loss of isentropic efficiency and increase of the mass flow, using the observable parameters, provided by the GTE operating system.The developed techniques and frameworks are verified and validated, using a set of three-year operating data from an industrial GTE in a power plant. iii Preface This thesis is an integrated article-based thesis, consisted of the peer-reviewed journal articles and a conference paper, either published or currently under review. In all of the work presented henceforth, Houman Hanachi was the principal contributor in all areas of conceptualization, methodology development, implementation and validation, computer code writing, data preprocessing and analysis, as well as manuscript composition, including tables and figures. Professor Jie Liu provided the guidance and supervision of the overall project as well as timely and constructive feedbacks to all the research work. Dr. Avisekh Banerjee provided his ideas and guidance regarding the work as well as editorial suggestions. Dr. Ying Chen provided supportive explanatio...

An Engine Deterioration Model for Predicting Fuel Consumption Impact in a Regional Aircraft

“…The former is associated with dirt and dust accumulation (in the airfoils and flow passages) that causes the degradation of both efficiency and pumping capacity and is most often seen in the compression system, i.e., fan, low-pressure compressor (LPC), and HPC. This deterioration, as its name suggests, can be recovered by a proper, regular cleaning process [27], referred to as water washing (WW). For non-recoverable LT deterioration, the performance debit is permanent and can only be partially recovered by shop visit repairs.…”

Physics-Based Online Assessment of Surface Roughness Severity for High-Pressure Turbines

“…Figure 1-2 shows the scope of the research objectives with respect to GTE health monitoring as well as previous works. Previously, a series of system-level performance diagnostic and prognostic tools had been developed around the basic operational data of an industrial GTE by Hanachi et al [11], based on a nonlinear gas path analysis (GPA) approach. A description of the data is provided in Section 1.3.…”

“…The expected power is compared with the measured value, in order to capture the performance degradation. The thermodynamic model of the GTE components from Equation (2-1) to (2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13) can be arranged into a mathematically determined system of four independent equations, with four independent variables:…”

“…[55], which was obtained using a fixed five-day sampling period after each wash without distinguishing between operating conditions. was defined as the ratio of efficiency deviation to the actual efficiency, as in Equation (2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16). With an extended sample size and condition-based sampling criteria, assessed by the present interfacing process showed a relatively modest and gradual growth trend compared to that by Hanachi et al The difference was likely due to the following two reasons: (1) compared to sampling periods greater than 10 days, a five-day sampling period was still insufficient at gathering enough data to form a clear distribution of and (2) without distinguishing between operating conditions, i.e., , the distribution of will be influenced by the result entries associated with a significantly lower than , resulting in differences in corrected rotor speed and lossto-roughness sensitivity.…”