Synthesis of ethyl tert-buty ether (ETBE), a high-performance fuel additive, through reactive distillation (RD) is an attractive route, while its operation and control are exceptionally difficult due to its functional combination and complex dynamics. Modern control technology greatly relies on good process models, while a reasonable RD model is too complex for control design. Moreover, RD contains considerable uncertainties that cannot be well described in process modelling. Alleviating the model requirement, this work aims to maintain the product purity in RD of ETBE through developing a pattern-based predictive control (PPC) scheme. Process dynamics, control structure, nonlinear transformation, feature pattern extraction, and pattern-based prediction and its incorporation with a conventional linear controller are discussed. Case studies show the effectiveness of the proposed method.

Reactive distillation (RD) is a favorable alternative to the conventional series of reactionseparation processes. However, the control of RD is challenging because of its complex dynamics resulting from the integrated functionality of reaction and separation. A linear proportionalintegral-derivative algorithm with fixed parameters has been shown not to be satisfactory to handle the high directionality of the process gain. It needs to be retuned adequately over a wide range of operating conditions. In this work, model gain scheduling is investigated for an ethyl tert-butyl ether reactive distillation column. It employs a set of simple local models, which cover relevant operating conditions and cope with the nonlinear characteristics of the process. The simple models are then integrated by using a proper switching scheme. Simulation results have shown that the proposed control strategy outperforms the standard proportional-integral control in both set-point tracking and disturbance rejection.

No abstract

No abstract

No abstract