A Hybrid Approach for Process Optimization of Distillation Reflux Condition using First Principle Models and Least Squares Regression

概要

Distillation columns are conventionally controlled at fixed reflux ratio to maintain the quality of the overhead product. If the cooling temperature of condenser becomes cold, the reboiler heat increases to cope with the internal reflux flow. Also, the loss of product recovery happens at the bottom of the column during the control delay of maintaining the bottom temperature. The reflux ratio is adequately adjusted by based on a proposed hybrid model analysis using the fist principle model (FPM) and least squares regression. According to the sensitivity case analysis with FPM on benzene / toluene binary distillation column, the reflux rate is substantially estimated by the multivariate least squares regression using the following explanatory variables; 1) feed flow rate, 2) column temperature in certain of stages, 3) column pressure, 4) sub-cooling temperature of condenser. To confirm the effectiveness of the hybrid approach, the dynamic simulation study cases of benzene / toluene binary distillation column are carried out. The simulation results show the effectiveness of the hybrid approach to reduce the energy consumption of reboiler heat duty and loss prevention of product.

収録
Computer Aided Chemical Engineering Vol.44, pp.229-234