Modelling and advanced control of a binary batch distillation pilot plant

Based on theoretical analysis of dynamic mass balance and vapor-liquid phase balance, a state-space model of a methanol/water binary batch distillation pilot plant was developed with an empirical temperature-composition relationship model to estimate the composition of the top distillate. As the nonlinearity and non-stationary characteristics of the batch process, the development and experimental test of four kinds of advanced control strategies including predictive control based on state space model, fuzzy logic control, gain scheduling control, hybrid control strategy combining gain scheduling and fuzzy logic control were carried out for the inferential composition control of the binary methanol/water batch distillation column top products. Batch distillation experimental studies showed that these advanced control strategies all gave good control results and a comparison of their performance with ordinary PI control was presented. The experimental results showed that all four advanced control methods were effective for the control of the binary batch column, and their performances were quite similar, but all were much better than ordinary PI control. The gain scheduling control strategy gave more accurate control result, and the model predictive control strategy achieved higher production rate.