Research On Economical NOx Control Technique Of Coal-fired Utility Boilers

With the developing demand of environmental protection, thermal powergenerators are being confronted with two requirements to reduce its operationcosts and to lower its emissions, especially NO_x emission, which is one of themain pollutants released from coal-fired boilers. Because of the irrationalityof the combustion organization and lack of advisable directions, most boilersoften deviate its optimal condition of operation designed for specific coal,which result in low efficiency and high NO_x emission. And these conditionsleave room for combustion optimization.Boiler is a multivariable system, it is hard to figure out the emissionproperty with a simple equation, because it is affected by many factors, suchas coal character, boiler's style, opertion parameters, etc. Some experimentscan roughly estemate the NO_x emission, but it needs too many experiments,and it is difficult to analyze the experimental data, because it only representsthe fixed condtion, the results without any generalization.In this paper, the formation and decomposition mechanism of nitrogenoxides were analyzed, factors impact the NO_x emissions were also discussed.With the help of orthogonal experimental design, the NO_x emission propertyand carbon content in fly ash of a 200MW utility boiler was experimentallyinvestigated. Taking advantage of the nonlinear dynamical and self-learningcharacteristics of artificial neural network, and training the network with LMalgorithm, the forecasting model of NO_x emission and carbon content in flyashes were set up. Based on these two models, an integrated model of highefficiency with low NO_x emission was also setup. All these models areverified through experimental data.The model about boiler is highly nonlinear. All the variables areconfined to a limited scope determined by the safety and stability of theboiler, thus, it is difficult to solve the combustion optimization model withtraditional method. Based on the comprehensively comparing some sorts ofalgorithms available for solving such problem and combing the characters ofthe models being proposed in this paper, a real coding based geneticalgorithms was chosen as numerical computing method. Some improvementsconcerning coding, operator designing etc. were made so as to make thealgorithm more suitable for the models. To meet the two requirements of highefficiency and low NO_x emission, a constant was used to combine theefficiency and emission factors in the fitness function, thus the model can betreated as a One-Object optimization problem. The solution shows that theproposed optimal algorithm may provide optimal control strategy to regulatethe operating parameters according to different optimal objects such as NO_xemission and boiler efficiency.