In this paper, a turbine on-line performance calculation system is presented. The system was implemented on a 575 MW unit of the Israel Electric Corporation and has been in operation for one year. The system was developed jointly by IEC and Berman Engineering Ltd. The main feature of the described system is the precision of the turbine heat rate calculation. This increased precision of the turbine heat rate calculation was accomplished by utilizing sophisticated statistical techniques, such as parametric and nonparametric regression, robust estimation, special filtration methods, autocorrelation methods, and uncertainty estimation methods. This high precision allows using the calculated heat rate as the main input to the turbine diagnostic system. The selection of turbine heat rate as the main diagnostic input is due to its high sensitivity to efficiency deviations of each turbine subsystem (turbine internal efficiency, condenser cleanliness, regenerative heaters' cleanliness, etc.). However, despite this high sensitivity, the turbine heat rate cannot be used directly without implementing the sophisticated statistical techniques mentioned above because: 1. relatively small variation of the calculated heat rate over the entire turbine load range (only about 3%); 2. the presence of systematic and random measurement errors; 3. low signaloise ratio as a result of the above items. In order to develop the techniques mentioned above, a detailed study of the error characteristics and error propagation was carried out. This study defined the problems which had to be solved in order to achieve an acceptably high precision of the calculation results.
展开▼
