关键词: 间歇精馏, 样条插值模型, 模型预测控制

Abstract: Batch distillation is a widely used separation process in chemical engineering.Due to the comparatively small investment and flexibility in production, batch distillation has found various industrial applications in biopharmaceuticals, fine chemicals production and foods processing.To guarantee high product quality, modeling and advanced process control for batch distillation have received significant attention. As an unsteady state dynamic process with strong nonlinearity, however, it is still not easy for engineers to design a control system on the basis of a simple model to guarantee good control performance for batch distillation.In this paper, by using the batch distillation model in Aspen Batch Distillation（ABD)as prototype, the process data of tower residual liquid and distillate compositions under different reflux ratios were obtained firstly.Based on the process data, then, the distillate volume and concentration of the process were formulated by the steady-state spline interpolation models（SIM).To compensate the dynamic error caused by variation of reflux ratios, a simple dynamic model was identified and combined with the steady-state SIM, resulting in a simple dynamic SIM for batch distillation.The comparison of the responses of the proposed SIM and ABD model to the time-varying reflux ratio indicated the applicability and precision of the proposed SIM.By using the SIM as the prediction model, a model predictive control （MPC)algorithm was further proposed for the concentration control of batch distillation.Numerical simulations demonstrated the applicability and robustness of the proposed control scheme for batch distillation with different feed compositions.The proposed control scheme lays a solid foundation for the further studies on online optimization of batch distillation.

Key words: batch distillation, spline interpolation model, model predictive control