Intelligent control of wet scrubber system for gas-particle separations

Abstract

The growing interest in wet scrubber systems has been attributed to their important advantages when compared to other air pollution control systems. They can collect flammable and explosive particulate matter (PM) contaminants safely, absorb gaseous pollutants and mists and also can cool hot gas streams and remove odor. Among various wet scrubber systems, spray towers are considered to be simple, cheaper to construct and maintain, it is a system that needs small space requirements and can handle large volume of contaminants. Spray tower wet scrubbers are limited to the control of PM contaminants that are greater than 10µm. This is due to improper selection of the optimum liquid droplet size and effect of changes in PM size and mass concentration. But, the PM contaminants that represent a significant fraction of emissions from most industrial sources and form the regulated particles that penetrate the lower respiratory tract of human lungs are PM ˂ 10µm. Several attempts have been made to improve the performance of spray tower wet scrubber systems for controlling this size range, but most of the studies provided only analytical solutions such as model predictions of collection efficiencies, mass transfer between particle and liquid droplets, liquid droplet coagulations and heat transfer phenomena and reactions within the scrubber system. This research proposes two intelligent control techniques based on neural network model predictive control (NNMPC) and adaptive neuro-fuzzy inference system (ANFIS) approaches which has the advantage of improving the spray tower wet scrubber performance by manipulating the scrubbing liquid pump to provide the optimum droplet size that can effectively control the PM contaminants that are less than 10µm. ANSYS Fluent computational fluid dynamic (CFD) algorithm based on continuity, momentum and k-ε turbulence model were used to optimize the gas flow distributions within the proposed spray tower system. From the results, the velocity flow contours and vectors at the inlet, across the scrubbing chamber and the outlet shows a distributed flow and the velocity profiles have fully conformed to the recommended profile for turbulent flows in cylindrical pipes. This has shortened the experimental period and optimized the spray tower wet scrubber system. The intelligent controllers were developed using MATLAB® and it was observed that the ANFIS controller performs better than the NNMPC controller based on shorter settling time of 2s. A hardware system has been used to implement the ANFIS controller in real-time using Microdust sensor, variable speed pump and Arduino Duemilanove - MATLAB interface. The implementation indicated that the ANFIS controller has been able to achieve the desired objective of setting the PM contaminants that are less than 10µm below the World Health Organization (WHO) allowable PM concentration of 20μg/m3with an accuracy of 97.56%. The proposed intelligent spray tower wet scrubber was validated using simulations and experiments. The result shows a better performance of the scrubber system for the control of PM size between 1.87 – 4.06µm. This size range is within the category of PM2.5 and PM10 that are being emitted from most industries and the PM concentration that penetrates the human respiratory system.