Optimal, active control of oxides of nitrogen (NO/sub x/) emissions from a natural gas-fired burner using a simple genetic algorithm

This paper describes the successful application of a genetic algorithm to the problem of active performance optimization of a non-linear, turbulent combustion process. Performance is measured in terms of combustion efficiency, /spl eta//sub c/, and the concentration of NO and NO/sub 2/ ([NO/sub x/]) emissions from a natural gas-fired research burner. Performance is affected by varying the mixing process through actuation of the excess air (EA) level and the swirl intensity (S') at the burner exit. The control scheme is a simple feedback loop, where the genetic algorithm functions as a specialized search operator, continuously monitoring [NO/sub x/] and /spl eta//sub c/ and updating EA and S' settings. The approach is shown to be effective in achieving and maintaining relatively high burner performance for a given burner geometry. The efficacy of the genetic algorithm is compared to a simpler search technique known as Powell's direction-set method. Although practical application of this technique will benefit from further development, the results point out some of the advantages of the genetic algorithm in this application.