Modeling And Control Of The Flue Gas Desulfurization Processes In A Circulating Fluidized Bed Scrubber

Flue gas desulfurization (FGD) in a circulating fluidized bed (CFB) scrubber is a novel technology that has the advantages of simple structure, low investment, small occupation space, high desulfurization efficiency, easy maintenance and stable operation, which is very attractive to small and medium sized power plant and boilers that require retrofits. The technology employs a circulating fluidized bed where most solid particle are recycled several times to prolong the reaction time between flue gas and the unreacted sorbent, which in turn impact the sorbent conversions and the removal efficiency. The circulating fluidized bed scrubbing process has been considered as one of the key developing and promoting flue gas desulfurization technology. Lots of experimental investigations have been reported on modeling of the flue gas desulfurization process, and the studies concerned with reaction kinetics or steady-state process behaviors become oversaturated. But so far the research on dynamic modeling and control of the circulating fluidized bed scrubber has been relatively delayed.By taking the experimental CFB scrubber, developed by the National Power Plant Combustion Engineering Technology Research Center, as the research object, this dissertation proposed a detailed first-principle model and a series of control strategies for the flue gas desulfurization process. The main contents are outlined as follows:(1) Based on mass balances and takes into account the hydrodynamic descriptions of the CFB scrubber, as well as the kinetics of the reaction between the calcium based sorbents and SO2, a detailed first-principle dynamic model and simulation analysis have been proposed for the first time. Both steady-state simulation and dynamic response analysis were carried out to investigate the major characteristics of the CFB scrubber. Sensitivity analysis was carried out through steady-state simulation to evaluate the influence of different operating conditions on the removal efficiency and the sorbent conversions for both fresh and recycled sorbents. The dynamic responses of major disturbance variables are investigated and illustrated the nonlinear and distributed nature of the process.(2) A soft-sensing hybrid model is proposed to solve the on-line measurement problem of the adiabatic saturation temperature in the flue gas desulfurization process in the CFB scrubber. The hybrid strategy combined the physical knowledge with the data infonnation of the desulfurization process. Simulation results showed that the estimation accuracy is greatly improved than that using traditional method. Just installing a humidity sensor at inlet will ensure the desulfurization plant running efficiently and safely.(3) A novel characteristics based identification algorithm was proposed for estimating the time-varying parameters for the distributed model of the CFB scrubber. Based on the method of characteristics, the partial differential equation was transformed into a system of ordinary differential equations. According to the analytical expression of the state variable, the model parameter was optimized and updated iteratively based on the latest measureable data sequence. Simulation results showed that the proposed identification algorithm achieved good performance and the time-varying parameters could rapidly converge to its true values.(4) The method of characteristics provides a convenient and efficient way to predict the future output from the current boundary value and can be used to develop a characteristic based MPC scheme. The hyperbolic distributed parameter system modeled by a partial differential equation (PDE) can be transformed into an equivalent ordinary differential equation (ODE) along the characteristic curve. Then the characteristic ODE can be integrated analytically to give the solution for the original PDE. The analytical expression of the distributed state variable allows prediction of the future outputs in high accuracy, which can be discredited as the predictive model to develop the off-line MPC control law. Simulation results of the outlet SO2 concentration regulation in a circulating fluidized bed scrubber indicate that the characteristic based MPC scheme can be successfully used to control hyperbolic distributed parameter systems, and generate better performance for setpoint tracking and disturbance rejection than the feedforward-feedback control scheme which is widely applied in the engineering fields.Finally, the potentioal further research directions and application prospect of the novel flue gas desulfurization technology is discussed in this dissertation.