Mathematical modeling, optimal design and control of an SCR reactor for NOx removal
- 作者：Chyi-Tsong Chen ; ctchen@fcu.edu.tw ; Wei-Lun Tan
- 关键词：SCR reactor ; NOx removal ; Mathematical modeling ; Parameter estimation ; Optimization ; Process control
- 刊名：Journal of the Taiwan Institute of Chemical Engineers
- 出版年：2012
- 期刊代码：P31_18761070
- 类别：ce
- 出版时间：May, 2012
- 卷：43
- 期：3
- 页码：409-419
- 文件大小：1650 K
摘要
The elimination of nitrogen oxides (NOx) is an important issue for global environment. This paper deals with the model development, optimal design and feedback control of an SCR (selective catalytic reduction) reactor for NOx removal. A 3D dynamic simulation model for use to investigate the reaction behavior and the transport phenomena in the catalytic filter of the SCR reactor is proposed. To estimate the model's kinetic parameters from experimental data, an optimization technique that integrates Taguchi method, a real-coded genetic algorithm and a neural network auxiliary model is developed. With the proposed dynamic model, we investigated the effects of the key parameters, such as the gas hourly space velocity, operating temperature and the amount of ammonium used, on the NOx conversion and NH3 slip phenomena. To improve the NOx abatement performance, the proposed optimization technique is then applied to search for a set of best operation conditions for the SCR reactor. The obtained results indicate that the optimized SCR can achieve the NOx reduction rate up to 99.93%, which is over 9%better in performance than the previously reported one in the literature. Besides, the optimal operating temperature is considerably lower and the emission of ammonium from the reactor is insignificant. Compared with conventional designs, the proposed design is much better in energy savings and is environment-friendly. To attenuate the negative effects of environmental disturbances on reactor's performance, we implemented a direct adaptive control strategy to the SCR reactor. The stability of the control system is theoretically guaranteed with a Lyapunov-based approach. Extensive simulation results show that the learning-type, nonlinear control strategy presents significantly much better NOx reduction performance than a conventional IMC-PI controller, especially when facing process uncertainties and disturbances.