Adaptive MPC Controller based Cascade Control of Distillation Column Parameters Estimation and Optimization

Abstract

One of the key element of many industrial process plants is a crude oil distillation system. This system requires heat for the vaporization of a mixture of feed to the distillation column in vapor form. In the existing process of distillation, an abnormal change of heat is exhibited due to improper control and monitoring of disturbances affecting the plant. This would result for undesired loss of product and product purity to be reduced. The parameters that are expected to be considered in the analysis, model and control of a distillation system described under the proposed study are inlet feed temperature, distillation column temperature, feed composition, internal liquid and vapor composition, feed flow rate, reboiler temperature and an external reflux temperature to the distillation tower.
Controlling distillation column parameters with Adaptive model predictive control allows for the determination of the predicted future instant values of the plant outputs. Using such control technique, the controlled plant outputs such as the temperature of distillation column, feed preheater, re boiler and the upper reflux is properly controlled and their parameters are estimated using recursive least squares approach in the entire process adaptation mechanism. From the analysis and optimization work made on the proposed system, the efficiency of the plant outputs in tracking their corresponding set point has improved based on the value of the transient system parameters as well as the value of relative volatility of feed mixture to the column. As per the finding in the analysis of the process, 95.4% and 93.5% improvement on the set point tracking and to the amount of evaporation liquid feed has been obtained respectively. On the other hand an improvement on transient parameters has been achieved to all plant outputs. As per the result obtained from the analysis, the peak overshoot, settling time and peak time of the system response has found to be less than 40% including the effect of measured disturbance to the plant. Hence, entire process variable optimization has been performed using the parameters of the model predictive controller to provide the proper degree of stability. Finally the proposed method of study has compared with other control strategies through which the performance of the proposed design has been ensured.