The continuous monitoring of gas turbines in commercial power plant operation provides long-term engine data of field units. Evaluation of the engine performance from such data is challenging since, apart from variations of operating points and ambient conditions, the state of the engine is subject to change due to ageing of engine components. The installed measurement devices influence the analysis due to their accuracy, which may itself alter with time. Furthermore, the available measurements usually do not cover all necessary information for assessment of the engine performance. To overcome these issues, this paper describes a method to systematically evaluate long term operation data without the incorporation of engine design models that depict the design state of the engine, but do not cover performance changes when components are ageing. Key focus of the methodology is thereby to assess long-term emission performance in the most reliable manner.
The analysis applies a data reconciliation method to long-term operating data in order to model the engine performance including non-measured variables and to account for measurement inaccuracies. This procedure relies on redundancies in the data set due to available measurements and the identification of suitable additional constituting equations that are independent of component ageing. The resulting over-determined set of equations allows for performing a data set optimization with respect to a minimal cumulated deviation to the measurement values, which represents the most probable, real state of the engine. The paper illustrates the development and application of the method for analysing emission performance with long-term operating data of a commercial gas turbine combined cycle power plant.