OCR Text |
Show American Flame Research Committee International Symposium September 30,1996 Emission Control System (InEC Tlo4). This system has been installed at two coal fired electric generating plants to date: a 680 MW wall fired unit at NYSEG's Kintigh Station and an 80 MW tangentially fired unit at NYSEG's Goudey Station. This paper provides an overview of the InEC Tlo4 system and a summary of the results obtained thus far. The Technology The InECTlo4 system uses state-of-the-art computational intelligence techniques developed by LMCS to create highly accurate non -linear models of the NOx formation process and other processes such as CO or LOI, from actual boiler operating data. Since these models are developed from actual plant data, they are customized to each plant. The InEC Tlo4 system also uses controllable losses calculations to quantify power plant performance. Information from the process models and the plant performance calculations are used to determine the optimum control settings of the plant. The optimum control settings are those settings which result in the NOx output of the plant being equal to a preassigned NOx target, while not exceeding CO or LOI limits, and operating the plant at the highest level of performance given the constraints. Figure 1 shows a block diagram of the InEC Tlo4 system as it integrates into normal plant operations. The InEC Tlo4 system is trained on actual plant data collected by the system. Special testing is usually required to define the system response in operating regimes condusive to low NOx production. After collecting enough test and normal operation data, the NOx model architecture can be developed. This development process produces models that are optimized for high accuracy and good generalization over all operating conditions. After the models are developed they can be integrated with the rest of the system software and installed at the plant. As time goes on the plant is bound to experience changes due to normal wear, equipment failures, or equipment upgrades. To address this issue the InEC 1M system is capable of adapting to changes in the plant. The InEC Tlo4 system continually monitors its own performance by comparing estimated values to actual sensor values. When performance degradation is detected, the system updates its models to reflect the new system characteristics. The user interface of the InEC Tlo4 system is typically comprised of 10 or more graphical displays which provide real time and historical data. The primary display screen, shown in figure 2 is the NOx control screen. This screen was mainly developed for the boiler operator and displays the optimum control setting recommendations and historical plots of operating costs and NOx output. The top left area on figure 2 shows both the present incremental operating cost ($!hr) and the InEC 1M system calculated optimum cost, for the past 6 hours of operation. The top right area shows a plot of the NOx emissior: as reported by the Continuous Emission Monitoring (CEM) equipment, the InEC ™ predi ",d NOx emissions, and the desired NOx as determined by compliance and system requirements. Below the plots is a table of the present NOx control settings and the recommended NOx control settings as determined by the InEC ™ system. When the operator is required to reduce NOx output, he/she simply selects the GO button. The InEC ™ system then utilizes a proprietary algorithm which allows the system to extract -2- |