Dynamic Maximization of Oxygen Yield in an Elevated-Pressure Air Separation Unit using Multiple Model Predictive Control

摘要: In a typical air separation unit (ASU) utilizing either simple gaseous oxygen (GOX) cycle or pumped liquid (PLOX) cycle, the flowrate of nitrogen stream connecting high- and low-pressure columns has major impact on total yield. It is shown that this yield reaches maximum at certain optimal LN2 stream, creating challenging feedback controller design problem. To dynamically maximize while ASU undergoes load-change and/or process disturbance, multiple model predictive control (MMPC) algorithm proposed. any operating point ASU, MMPC algorithm, through model-weight calculation based plant measurements, naturally continuously selects dominant model(s) corresponding to current state, making control-move decisions approach point. This facilitates less energy consumption in form compressed feed-air compared ratio during load-swings. addition, since linear optimization problem solved each time step, involves much computational cost than (MPC) first-principles model.