Temperature modeling and control of the sulfuric acid plant in an industrial smelter

Abstract

Roasting, smelting and converting are pyrometallurgical techniques used to eliminate gangue rock from sulfide ores and produce a saleable metal product. Due to the large amount of sulfur present in sulfide ores, the off-gas produced from pyrometallurgical processing is laden with sulfur dioxide (SO2). Production of sulfuric acid (H2SO4), from off-gas laden with SO2 is one of the methods used at smelters to reduce the amount of SO2 being released into the atmosphere. The acid plant consists mainly of a catalytic converter, absorption towers and a network of heat exchangers. SO2 is converted to sulfur trioxide (SO3) which is absorbed to produce a sulfuric acid product. Efficient oxidation of SO2 occurs within a tight temperature range so heat exchangers equipped with bypass valves are used to regulate the temperature throughout the acid plant. In this dissertation, a steady-state model is developed from fundamental steady-state mass and energy balances. The steady-state model provides a relation between the process variables and the temperature of the outlet streams of the heat exchangers. Unknown variables are estimated using industrial operating data. The steady-state model is used to investigate the effect of process variables on the controlled variables. The disturbance variables that have the largest effect on the process are the feed gas flow rate and the SO2 concentration. The results provide useful information since with recent process modifications that are part of the Clean AER project, variations in the feed gas flow rate and SO2 concentration will increase. The effects of the manipulated variables were also investigated which provides a foundation of understanding for process control. The results of the investigation of the effect of process variables on the controlled variables were quantified by calculating the steady-state gains. The dynamics of the process were investigated through analyzing the industrial operating data. The output variables do not vary simultaneously with changes in the input variables. The correlation coefficients were determined for the variables. The correlation coefficient between variables provides an estimate of how much influence the input variables have on the output variables. Delayed correlation analysis was performed to explore the process dynamics. Dynamic models were identified using industrial operating data with and without prior information supplied using the System Identification toolbox in Matlab. Providing the steadystate gains and an estimate of the process time constants to the System Identification toolbox greatly improved the identified model. The dynamic model was validated by comparing the model-estimated output and the output from industrial operating data. Temperature control within the acid plant is a multiple-input-multiple-output control problem. Bristol’s Relative Gain Array and Singular Value Analysis are used to determine the most effective pairing of variables. A feedforward-feedback control scheme for temperature regulation is explored. Simulations for major disturbances, such as flow rate and SO2 concentration of the feed gas, are carried out using two alternative controller pairings. The results of the simulations are reviewed and the advantages of each controller pairing are discussed.