Abstract
Recent expansion of renewable power plants have transformed the role and operation of gas turbines to a great extent. From the base load operation era we are moving into a flexible and dynamic engine operation of gas turbines. In particular, aero derivative engines that have the capacity to start up, shut down in a short time frame are becoming quite popular for both hybrid power plant arrangements and distributed electricity generation.
Advances in computational intelligence, such as digital twins, have amplified the importance of condition monitoring, diagnostics and prognostics capabilities in the face of gas turbine operation. Given the dynamic operating profile of the gas turbines, it is of paramount importance to develop, tune and deploy engine models that are accurate and robust to accommodate their nonlinear behavior.
Performing diagnostics in transient conditions has recently gained attention, since the gas turbines are acting as partners of renewables and they have a supporting role. Among a family of diagnostics methods, one that has real time capabilities is based on zero-dimensional engine models. This paper present a novel diagnostic approach for determining the health of a gas turbine when it works in conjunction with a wind farm in hybrid power plant.
In contrary to our earlier works, where we have mathematically modeled component maps to derive the health of an engine, in this paper we propose a model-based diagnostic method without reconstructing component maps according to their degradation. Once the engine model is initially adapted to its clean condition, it is subsequently tuned in real time to reflect the changes in both the operation and degradation with respect to a benchmark engine model. Time evolving multiple component degradation scenarios for a gas turbine operating in conjunction with a wind farm, are simulated to test the accuracy and efficiency of the proposed method.
From a bank of simulated measurements, data trending is performed which facilitates the detection of degradation and provides useful conclusions about the health state of the engine. This diagnostic method is suitable for gas turbines that spend most of their life time in part-load and transient operation and it forms a simple and useful tool for operators in planning their assets’ maintenance in a computational efficient and accurate manner.
Advances in computational intelligence, such as digital twins, have amplified the importance of condition monitoring, diagnostics and prognostics capabilities in the face of gas turbine operation. Given the dynamic operating profile of the gas turbines, it is of paramount importance to develop, tune and deploy engine models that are accurate and robust to accommodate their nonlinear behavior.
Performing diagnostics in transient conditions has recently gained attention, since the gas turbines are acting as partners of renewables and they have a supporting role. Among a family of diagnostics methods, one that has real time capabilities is based on zero-dimensional engine models. This paper present a novel diagnostic approach for determining the health of a gas turbine when it works in conjunction with a wind farm in hybrid power plant.
In contrary to our earlier works, where we have mathematically modeled component maps to derive the health of an engine, in this paper we propose a model-based diagnostic method without reconstructing component maps according to their degradation. Once the engine model is initially adapted to its clean condition, it is subsequently tuned in real time to reflect the changes in both the operation and degradation with respect to a benchmark engine model. Time evolving multiple component degradation scenarios for a gas turbine operating in conjunction with a wind farm, are simulated to test the accuracy and efficiency of the proposed method.
From a bank of simulated measurements, data trending is performed which facilitates the detection of degradation and provides useful conclusions about the health state of the engine. This diagnostic method is suitable for gas turbines that spend most of their life time in part-load and transient operation and it forms a simple and useful tool for operators in planning their assets’ maintenance in a computational efficient and accurate manner.
| Original language | English |
|---|---|
| Title of host publication | ASME Turbo Expo 2020 |
| Subtitle of host publication | Proceedings of ASME Turbo Expo 2020 Turbomachinery Technical Conference and Exposition |
| Publication status | Published - 2020 |
