Flooding and offset-free nonlinear model predictive control of a high-purity industrial ethylene splitter using a hybrid model

This work aims to achieve flooding free control for an industrial ethylene splitter. The tower operates close to maximum capacity and experiences flooding due to feed disturbances. This work develops a replacement of the current control strategy with an offset-free Nonlinear Model Predictive Control (OF-NMPC) to improve the control and avoid flooding. OF-NMPC uses a hybrid model comprised of a Nonlinear Autoregressive Network (NARX) for predicting dynamics, and a first principles steady state model to calculate the tray liquid flow which is a constraint in OF-NMPC. The calculation predicts the steady-state internal flow which would occur at the current values of the manipulated variables, thus providing about 20 min time interval to avoid flooding via controller actions. An Aspen Dynamics model that closely mimics the plant is used to test the proposed OF-NMPC. The simulation results demonstrate the ability of the proposed OF-NMPC design to achieve flooding free control.