Open AccessEstimation of NO{sub x} emissions from pulverized coal-fired utility boilers. Final report
Author(s) -
D.J. Wildman,
Scott M. Smouse
Publication year - 1995
Language(s) - English
Resource type - Reports
DOI - 10.2172/588588
Subject(s) - pulverized coal fired boiler , boiler (water heating) , combustion , coal , artificial neural network , engineering , process engineering , power station , software , waste management , coal combustion products , environmental science , computer science , chemistry , artificial intelligence , electrical engineering , organic chemistry , programming language
The formation of nitrogen oxides (NO{sub x}) during pulverized-coal combustion in utility boilers is governed by many factors, including the boiler`s design characteristics and operating conditions, and coal properties. Presently, no simple, reliable method is publicly available to estimate NO{sub x} emissions from any coal-fired boiler. A neural network back-propagation algorithm was previously developed using a small data set of boiler design characteristics and operating conditions, and coal properties for tangentially fired boilers. This initial effort yielded sufficient confidence in the use of neural network data analysis techniques to expand the data base to other boiler firing modes. A new neural network-based algorithm has been developed for all major pulverized coal-firing modes (wall, opposed-wall, cell, and tangential) that accurately predicts NO{sub x} emissions using 11 readily available data inputs. A sensitivity study, which was completed for all major input parameters, yielded results that agree with conventional wisdom and practical experience. This new algorithm is being used by others, including the Electric Power Research Institute (EPRI). EPRI has included the algorithm in its new software for making emissions compliance decisions, the Clean Air Technology Workstation