Gas turbines are faced with new challenges of increasing flexibility in their operation while reducing their life cycle costs, leading to new research priorities and challenges. One of these challenges involves the establishment of high fidelity, accurate, and computationally efficient engine performance simulation, diagnosis, and prognosis schemes, which will be able to handle and address the gas turbine's ever-growing flexible and dynamic operational characteristics. Predicting accurately the performance of gas turbines depends on detailed understanding of the engine components behavior that is captured by component performance maps. The limited availability of these maps due to their proprietary nature has been commonly managed by adapting default generic maps in order to match the targeted off-design or engine degraded measurements. Although these approaches might be suitable in small range of operating conditions, further investigation is required to assess the capabilities of such methods for use in gas turbine diagnosis under dynamic transient conditions. The diversification of energy portfolio and introduction of distributed generation in electrical energy production have created need for such studies. The reason is not only the fluctuation in energy demand but also more importantly the fact that renewable energy sources, which work with conventional fossil fuel based sources, supply the grid with varying power that depend, for example, on solar irradiation. In this paper, modeling methods for the compressor and turbine maps are presented for improving the accuracy and fidelity of the engine performance prediction and diagnosis. The proposed component map fitting methods simultaneously determine the best set of equations for matching the compressor and the turbine map data. The coefficients that determine the shape of the component map curves have been analyzed and tuned through a nonlinear multi-objective optimization scheme in order to meet the targeted set of engine measurements. The proposed component map modeling methods are developed in the object oriented matlab/simulink environment and integrated with a dynamic gas turbine engine model. The accuracy of the methods is evaluated for predicting multiple component degradations of an engine at transient operating conditions. The proposed adaptive diagnostics method has the capability to generalize current gas turbine performance prediction approaches and to improve performance-based diagnostic techniques.
Industrial Gas Turbine Performance: Compressor Fouling and On-Line Washing
