A nonlinear feedback linearizing control (FLC) strategy is proposed within the differential geometric framework for temperature control of a refinery debutanizer column. The distillation model is verified by real data. The FLC control algorithm usually consists of a transformer, a state estimator and an external linear controller. Here, two state estimators, namely extended Kalman filter (EKF) and short-cut model-based open-loop estimator (SMBOLE), have been developed to device the hybrid FLC-EKF and FLC-SMBOLE control systems, respectively. In order to avoid estimator design complexity as well as
computational burden, an ideal binary distillation model [light key (LK)/heavy key (HK)] has been used as an EKF predictor and open-loop estimator (OLE). In this article, a comparative study has been conducted between the FLC-EKF, FLC-SMBOLE and a classical dual-loop proportional integral derivative (PID) control structure. Simulation results show that despite the significant process/model mismatch, the proposed FLC controllers perform better than the PID control scheme.
Index Terms-Debutanizer, extended Kalman filter, feedback linearizing control, short-cut model-based open-loop estimator.
NOMENCLATUREBottom product flow rate, lbmol/hr. Distillate flow rate, lbmol/hr.
,Error to the external controllers.Proportional gain.Liquid flow rate, lbmol/hr.Tray liquid holdup, lbmol.Pressure, psi.Critical pressure, psi.Total pressure, psi.Vapor pressure, psi.Relative order.Reflux flow rate, lbmol/hr and covariance matrix of the measurement errors.Temperature, .Critical temperature, .