Improvement of turbine blade lifetime assessment by more accurate estimation of the thermal boundary conditions

Abstract

The known studies in the area of gas turbine lifetime prediction do not result in the algorithms for on-line engine monitoring. This article introduces and investigates a new method for developing ''light'' mathematical models to estimate static thermal boundary conditions for gas turbine hot elements. In contrast to the previous developments, these models allow on-line lifetime monitoring of such elements. The blade of the high-pressure turbine of a two-spool free turbine power plant was chosen as a test case. The models of blade boundary conditions were developed based on well-known thermodynamic relations and a steady-state nonlinear physics-based model of this engine. Many candidate models are analyzed in the article, and the best models are selected by their accuracy and robustness to engine faults using instrumental and truncation errors as criteria. The instrumentation errors are induced by measurement inaccuracy of gas path variables used. For the analysis of the model robustness, the truncation errors are computed. They appear when performance of an engine deviates from a baseline due to normal degradation of the engine and because of its faults. The gas path parameters under healthy and faulty engine health conditions are simulated by the thermodynamic model. These simulated quantities are used as the input data to perform the comparison of the candidate models. The final accuracy analysis shows that the proposed method improves the estimates of the thermal boundary conditions. As a result, prediction of an engine lifetime becomes significantly more accurate. The article also determines the positive effect of the compressor discharge temperature sensor. When it is installed in addition to a standard gas path measurement system, the accuracy of the measurement-based lifetime prediction grows drastically.