(376g) Applying Evolutionary Algorithms in Optimizing Mixing in Selective Catalytic Reduction Systems Via Computational Fluid Dynamics (CFD)

One of the foremost applications of computational fluid dynamics modeling techniques over the past ten years has been in the area of optimizing Air Pollution Control (APC) systems used to reduce emissions from fossil fuel fired stationary sources such as power plants and steel mills. As the paper shall demonstrate, the optimization of static mixing devices is critical to the efficient performance of technologies that reduce NOx,. This paper shall describe the authors' experiences in optimizing the fluid flow performance of a Selective Catalytic Reduction (SCR) system that was installed on a 100MW biomass fired electricity generating boiler at the Gainesville Renewable Energy Center.

The combustion of fossil fuels is the leading contributor to global air emissions. A variety of APC technologies are used to limit these stack emissions, as required by the Clean Air Act and other legislation. Almost without exception, these technologies rely upon appropriately optimized fluid dynamic performance to ensure the required process performance. A common example of this reliance upon fluid dynamics optimization is that of SCR systems. SCR systems are dependent upon the effective homogenization of an injected reagent within the NOx containing flue gas to achieve high NOx reduction efficiency across the catalyst bed. The authors' development of numerical tools to evaluate the impact of non-homogeneous velocity and concentration profiles on NOx reduction performance have enabled a substantially more accurate assessment of the importance of this optimization and will be presented for the first time here. In addition, the specific combination of technologies that enable the effective homogenization including reagent injection equipment, static mixers and other flow distribution devices, will be discussed in light of the performance enhancements achieved. Of recent interest in this area of focus has been the coupling of CFD simulations to evolutionary search algorithms, such as genetic algorithms. The authors' progress in the application of this methodology will be discussed with an example given demonstrating Pareto front convergence in a constrained multi-objective search.