Dynamic real-time energy saving control of pressure-swing distillation based on artificial neural networks

Abstract

Parameter variation can increase economic costs and energy consumption in distillation process when disturbances occur. In this work, a dynamic optimization control model based on ANNs was developed. Composition disturbances in pressure-swing distillation with and without heat integration to separate isoamyl alcohol and toluene were regarded as the research object. The economic cost and energy consumption were effectively reduced by predicting the most optimal operating parameters during the disturbance process and controlling the process in real time. In the absence of heat integration, the energy consumption of the control structure based on ANNs (CS2) is reduced by 1.68% and 1.70% respectively compared to traditional control structure (CS1) during ± 20% composition disturbance. In partial heat integration, the energy consumption of the control structure based on ANNs (CS4) is reduced by 3.24% and 2.68% compared to traditional control structure (CS3). TAC is saved by 3.51% and 3.31% when reaching stability.