OCR Text |
Show 1993 AFRC INTERNATIONAL SYMPOSIUM Implementation of a Software CEM™ for NOx Emissions Monitoring October 19,1993 John Havener - Senior Consultant - Pavilion Technologies Inc. Keith Terhune- Chemical Engineer - Eastman Chemical Company Introduction Industry is struggling to come to grips with the high cost of compliance with the 1990 Clean Air Act Amendments (CAAA) which require continuous emissions monitoring ( C E M ) systems in order to provide accurate and reliable measurements of sulfur dioxide (SO2), nitrogen oxides (NOx), carbon monoxide (CO) and opacity. C E M s will be used to ensure compliance with emissions standards and mandated reductions, and will track allowances in the new market-based SO2 emissions-trading program. As a result of these mandates, C E M systems are becoming an integral part of utility and industrial operations. Most U.S. companies are installing C E M s as part of the cost of doing business. The initial capital cost of installing hardware CEMs is very high, typically in the $150K-$300K range. 40 C F R Part 75 mandates 9 0 % up-time for C E M s and has established strict regulatory standards requiring that substitute data be estimated and reported during any period that the C E M is unavailable. Severe penalties are imposed on a company whose C E M s fail, and governmental bodies consider all emissions outside the mandated 9 0 % up-time as out of compliance. The results are potentially high fines and negative publicity. In many cases this leads to the installation of redundant systems, at even higher cost. In addition to the capital cost, the maintenance cost of C E M operation is staggering. The instruments are delicate and must be recalibrated frequently. It is not unusual to require additional full time employees to keep a facility's C E M s operating within mandated relative accuracy limits. To avoid these high costs, industry has for some time attempted to develop sufficiently accurate parametric emissions monitoring methods (PEMS) that would use correlation with existing sensors to continuously predict emissions on-line. However, the accuracy of these methods have been inadequate to meet federal mandates, and regulators have resisted industry pressure to allow P E M S use as a replacement for C E M S . This problem has been overcome. Working cooperatively, the Arkansas Eastman Division of Eastman Chemical Company and Pavilion Technologies Inc. developed, installed, and certified a highly reliable, low cost C E M alternative to predict N O x emissions from Arkansas Eastman's No. 5 gas fired boiler located at the Batesville, Arkansas plant site. In a period of 60 days total elapsed time, a software based P E M S was developed and permitted. This technique uses the ultimate in non-linear regression utilizing advanced adaptive technologies (fuzzy logic, neural networks and chaotic systems theory). This lower cost solution method to the emissions monitoring problem is now available commercially from Pavilion Technologies and Callidus Technologies Inc. and is called the Software CEM™. This P E M S system offers high accuracy, superior performance, expanded function, and minimal maintenance cost when compared to hardware C E M s . As a result of the initial success and the considerable savings, industry groups such as the Texas Chemical Council, and the Chemical Manufactures Association have championed this technique to state and federal environmental regulators. Regulations in Arkansas, and Texasliave recently been revised to allow use of the Software CEM to replace hardware Pg. 1 III-15 |