Predictive Maintenance of Gas Turbine Air Inlet Systems for Enhanced Profitability as a Function of Environmental Conditions

摘要

Increasing trends in the digitalization of gas turbine plants allow significant opportunity for utilizing predictive maintenance methods to optimize engine performance. Digital twins have been used extensively to detect anomalies, prevent failures and encourage preventive maintenance activities. An active area of research is how to shift from preventative maintenance based on current conditions, to predictive maintenance based on future conditions. Close mapping of historical engine performance against air quality metrics and ambient weather conditions allow significant progress to be made towards true predictive maintenance, by providing a greater understanding of the condition of the turbine air inlet and compressor section. This allows for increasingly accurate predictions of future engine degradation due to air inlet pressure drop and compressor degradation to a fidelity useful for scheduling maintenance needs. An economic optimization can then be performed balancing the costs of the two engine degradation modes and the corrective actions that can be taken, namely air inlet pressure drop against filter replacement interval, and compressor degradation against compressor soak wash interval. This paper describes our experience in monitoring turbine performance to predict and optimize maintenance needs in combined cycle power plants.