An optimization tool for gas turbine engine diagnostics

Abstract

A major challenge faced by the Gas Turbine industry, both the users and the manufacturers is the reduction of life cycle costs and safe running of a gas turbine. A reduction in the costs can be achieved by reducing the development time while the engine is in the development stage and reducing operating costs for in service engines. One of the ways of achieving these would be the use of sophisticated performance analysis and diagnostic techniques. Techniques for such purposes of diagnosis have developed a great deal over the last three decades. The initial work was on gas path analysis, followed by use of conventional techniques such as Kalman filters and Least squares algorithm for gas path analysis. The last decade has seen a lot of work on the use of intelligent systems such as neural networks, fuzzy logic and expert systems for such purposes. Though improvements have been made over the years, but all these techniques have major drawbacks, which make their use in the current stage of development very unlikely. The use of genetic algorithm based optimization technique for diagnostics of well instrumented engines (development engines) was successfully made at Cranfield University. The present work presents a technique for fault diagnostics of engines that are relatively poorly instrumented. The work presents how the task is achieved by the use of multiple operating point analysis and the use of a genetic algorithm based optimization technique for optimization of an objective function that depends on the measurements and the corresponding value for changed performance and power setting parameters obtained from the thermodynamic performance model of the engine. The main issues that have been addressed are the choice and number of operating points and also the development of the multi objective optimization technique. The technique is able to accurately identify the faulty components and quantify the fault. The fault is expressed in terms of a change in efficiency and capacity of the various components. The optimization also carries out Sensor fault detection, isolation and accommodation .The technique has been tested on a number of engine types using simulated data. These engines have been chosen to cover a wide range of instrumentation suites. The advantages, drawbacks and the suggested method of application have also been presented.