Model Predictive Control And Dynamic Behaviour Of CO₂ Compression And Purification Unit In Oxy-fuel Combustion Power Plants

As one of CO₂ capture,utilization and storage（CCUS）technologies,oxy-fuel combustion is a promising method to reduce CO₂ emissions in fossil fuel fired power plants.Typically,oxy-fuel combustion uses oxygen rather than air to combust with fuels and recycles one part of flue gas to moderate temperatu re profile in furnace.Other part of flue gas is sent to CO₂ compression and purification unit（CPU）for obtaining high purity CO₂ products.In this CPU system,low temperatur e double flash drums are applied to separate CO₂ from other impurities.However,due to intensive interactions among operating parameters,CPU system faces with the issue of low control reliability.Toward that end,this work aims to improve operating reli ability via control optimization.Based on autoregressive analysis model and stat e space model,two model predictive control（MPC）structures are proposed to optimize control actions and follow the optimal trajectory.In order to validate two new MPC schem es,dynamic behavior of operating parameters in CPU system is examined and compar ed to reference case（PID control）under two operating scenarios of flue gas flow rate and CO₂ concentration step changes.Subsequently,the coupling between Aspen plus dynamics and simulink software was carried out and the dynamic exergy analysis of the coupled CPU system was carried out.Several manipulated and control variables were selected according to the actual needs of the CPU system.Modelling was carried out by both sy stem identification method and control system design,and the autoregressive anal ysis model and state space model were obtained by screening and model predictive control was applied to both models.Through applying a 1%step change of flue gas CO ₂ concentration to CPU system,dynamic response of controlled variables（CO ₂ capture rate,CO₂ product purity and S-8 stream temperature）is obtained for three control structures（MPC-ARX,PID and MPC-SS）to examine the controllability.All controlled variables show some fluctuations but varied in an acceptable operating regimes.Integral square d error（ISE）is employed to evaluate the control reliability of different control strategies.MPC-SS exhibits 27.5%and 32.4%ISE values of PID case when flue gas CO ₂ concentration increases and decreases 1%,respectively.Moreover,MPC-SS performs smaller ISE value for CO₂ capture rate and S-8 temperature than PID control.Dynamic model verification by running APD in conjunction with Simulink via the AMSimulink module.The coupled linear model is validated using co-simulation,and the dynamic characteristics under the step of flue gas CO₂ concentration and flue gas flow are compared and analysed.In conjunction with the dynamic exergy theory,the Matlab software is used to design a calculation program to calculate and analyse the dynamic exergy of the coupled model.Simulation results show that,the CPU system is more sensitive to changes in flue gas CO ₂ concentration than to changes in flue gas flow,and the system saves energy during the rise in flue gas CO₂concentration.