Model predictive control of a catalytic reverse flow reactor

A control strategy of flow reversal with hot gas withdrawal for heat recovery and its application in mitigation and utilization of ventilation air methane in a reverse flow reactor

“…(20). Due to the intermittent operation of air dilution, the theoretical maximum switching time should be in the range between 0.5L cat /(1 + 2a)V fr and 0.5L cat /V fr .…”

“…To maintain the auto-thermal operation, different control algorithms have been proposed and multiple schemes have been investigated [15]. Great progresses have been made on the control of the CRFR recently, including feed forward control [16], model predictive control (MPC) [12,13,[17][18][19][20], linear quadratic regulator (LQR) control [18,21,22], robust control [23] and simple logic control [14,15,[24][25][26][27]. Apart from operating the CRFR auto-thermally, reclaiming a fraction of heat from methane combustion is also very tempting for the huge emission of VAM.…”

Online correction MPC strategy for spatially-distributed system based on PCA method

“…where equation (9) is the online correction model, equation (10) refers to the errors between the real output and predictive output of the low-dimensional models at a previous time,…”

“…The spatial-temporal coupling and infinite dimensionality of the systems make them very difficult for analysis, modeling and control [2][3][4][5][6][7]. Model predictive control (MPC) is a powerful model-based control method to deal with process control problems [8][9][10][11][12]. It utilizes a dynamic process model and solves an optimal control policy that takes into account constraints on the inputs and outputs, and the effect of unmeasured disturbances.…”

Model-based predictive control for spatially-distributed systems using dimensional reduction models

“…Because the best process identification is difficult, the modeling error occurs when using linear models [20−24] . Although the feedback in model predictive control (MPC) can reduce the impact of the discrepancy between the process and the predictive behavior, MPC is not designed to explicitly handle model mismatch [25,26] .…”