OCR Text |
Show CEMs. This technique has now been applied to reciprocating engines, ethylene furnaces, and gas fired turbines, in addition to the original Eastman Chemicals project on a gas fired boiler. Investigations have begun on coal fired boilers as well. Scores of similar Software CEM systems are being initiated in non-attainment areas. The savings generated is typically 1/3 to 1/2 the cost of the hardware C E M installations. This presentation describes the original project, the methodology, and the software used. The SOFTWARE CEM Solution The Software CEM takes advantage of existing investments in data collection and process instrumentation to eliminate the need for a new C E M instrument by using Process Insights™, an adaptive software package, to predict the emissions from process data. The solution is composed of three straight-forward steps. First, stack emissions test data are gathered using portable emissions testing trailers. Second, this data is combined with existing historical data to create a data set for input to the modeling software. Third, using Process Insights, nonlinear models are generated to predict emission rates that would be measured if a hardware C E M were actually in place. These models are very accurate, and have gained acceptance by state regulatory bodies for C E M service. The methodology can be employed for N O x , S02, C O , opacity, and other emission measurements. Arkansas Eastman Problem Description and Background Arkansas Eastman is a division of Eastman Chemical Company, located in Batesville, Arkansas. Eastman Chemical is a major producer of plastics and chemicals with global sales in excess of 2.5 billion per year. The Arkansas Eastman facility is the smallest of the Eastman Chemical plant sites with total employment of 720. Arkansas Eastman produces specialty organic chemicals for integration with consumer products. On May 14, 1991 Arkansas Eastman was granted permission to construct and operate No.5 Boiler, a 221 MBTU/hr natural gas fired boiler, which requires a P S D permit. The boiler was also subject to N S PS regulations under 40CFR 60 Subpart Db. Initially, Arkansas Eastman proposed a PEMS system based on a mathematical model developed from a literature search. The accuracy of this model was not found to be acceptable to the regulators. However, Arkansas Eastman installed state-of-the-art computer technology early in 1993 that provided the ability to predict N O x emission from the No.5 Boiler with much greater accuracy and reliability than the originally proposed model. The new state-of-the-art technology was the Process Insights modeling system from Pavilion Technologies of Austin, Texas. This tool employs a graphical user interface that insulates the user from the complexities of fuzzy logic, neural networks, and chaotic systems theory enabling non Al specialists to make use of the power of these advanced methods. On March 19, 1993, Arkansas Eastman notified the Arkansas Department of Pollution Control and Ecology ( A D P C & E ) that it would agree to pursuing a 60-day schedule to confirm the feasibility and accuracy of the proposed Software CEM. The A D P C & E required that in the event of unsuccessful testing of the Software CEM, a hardware C E M must be operational within 120 days, allowing a total of 180 days for a continuous emission monitoring to be installed on Boiler No. 5. Preliminary testing using U S E P A approved Reference method technology ( R M 7E and 3A) indicated major improvements over the previously proposed mathematical model. Pg.2 III-15 |