How to Create a Performance Model of a Gas Turbine From a Limited Amount of Information

Abstract

Gas turbine manufacturers develop complex performance simulation models for their products; these proprietary models are based on design information and the many measurements taken during engine development. For subcontractors in collaborative projects, for gas turbine users and outsiders there is often only a limited amount of data accessible for creating a performance model of the engine. User-friendly, accurate and fast PC-based engine simulation tools are available for anybody from several sources. With these tools it is possible to create from a limited amount of data full thermodynamic models. In this paper a methodology is presented which minimizes the effort needed for creating such models. It consists of four steps: Firstly a suitable cycle reference point is chosen and the model is tailored to the data of this point. Secondly compressor and turbine maps are added and scaled such that they fit exactly to the cycle reference point. In this step a second operating point is considered and the location of the cycle reference point in the component maps is adapted such that the simulation fits optimally to the given data of the second point. In a third step, the rest of the data are compared graphically with the simulation. Here many modelers fall in a trap: They plot the data versus spool speed as x-axis because speed is accurately measurable and regarded as reliable information. However, spool speed is — from the view of thermodynamics — a parameter of secondary importance. If the correlation of spool speed with corrected flow in the compressor map is incorrect — which is very probable at the beginning of the modeling process — then all graphics will show discrepancies. This makes the adaptation of the model to the data an extended iterative process. If one uses for the model checks a primary thermodynamic parameter — like corrected mass flow, overall pressure ratio or thrust respectively shaft power — as basis then the task is very much simplified. In the fourth and final step the speed values in the estimated compressor maps are adjusted. This has little effect on the matching accuracy of the previous steps, so the model is finished quickly. The procedure is demonstrated by creating a model for a two-spool turbojet which was tested over quite a range of operating conditions in an altitude test facility. Without much iteration a model is quickly created which matches all the measured data within the quoted uncertainty of the measurements.