Abstract
One of the main reasons of the performance degradation of gas turbines is the axial compressor fouling due to air pollutants. Considering the fact that the fouling leads to high consumption of fuel, reducing of the axial compressor’s discharge air pressure and increasing of the exhaust temperature, thus designing a compressor degradation detection system will allow prevent such issues. Many gas turbine plants lose power due to dirty axial compressor blades, which can add up to 4% loss of power. In case of power plants, the power loosing could be observed by less megawatts produced by generator. But in case of gas compression stations the effect of power loosing could not be quickly detected, because there is not direct measurement of the discharge power produced by gas turbine. This article represents technique for detection of gas turbine axial compressor degradation in case of gas turbine driven natural gas compression units. Calculation of the centrifugal gas compressor power performed using proven methodology. Approach for evaluation of the gas turbine performance based on machine learning prediction model is shown. Adequacy of the model has been made to three weeks’ operation data of the 10 Megawatt class industrial gas turbine.