Modeling and Prediction of the Temperature Spread Variation for Evaluation of Gas Turbine Performance

This study presents modeling, analysis and prediction of the temperature spread variation for a 110 MW GT11N2-3VGV Alstom Gas Turbine at the Kpone Thermal Power Station (KTPS) for the purpose of optimised operation. The study focuses on the temperature spread variation around the mean point of the turbine section since uneven distribution of temperature around this section leads to severe deformation. A data-driven statistical analysis called Time Series Modeling was used as a tool for predicting the temperature spread variation of the gas turbine. Analysis of historical monitoring data of the industrial turbine informed the use of first order autoregressive model (AR (1)). The validity of the proposed autoregressive mode was tested using data covering three months logged from the Egatrol 8.0 software interface and the model was verified using residual analysis. The study revealed that the temperature spread at any instant follows an autoregression model of order one (1) with 95 % confidence limits and an error of 3.01. Hence, the AR (1) model can be used to predict the thermal health of the turbine as well as determine when maintenance should be conducted.