No abstract

Formic acid production through methyl formate hydrolysis has been shown to be energy and capital cost intensive, and its performance could be significantly promoted by process intensification. However, integration of reaction and separation exhibits a complex nonlinear behavior, which requires corresponding optimization and control to be effectively addressed before further industrial implementation. In the present work, optimization was first performed for a conventional reactive distillation (RD) process and reactive dividing wall column (RDWC) by coupling genetic algorithm and rigorous simulations, in which a user-defined model was incorporated to take kinetics into account. Although the results demonstrate that RDWC is inferior to RD in terms of economic criteria, it provides the basis for proposing new easy-to-operate configurations in the subsequent part 2 of this series. Then multiloop proportional–integral (PI) control structures and linear model predictive control (MPC) schemes were designed for the conventional RD process and RDWC, respectively. The performances of two control structures were compared subject to feed disturbance, by using quantitative indexes such as oscillation, settling time, overshoot, and integral of the squared error (ISE). The dynamic response validates that MPC outperforms classical multiloop control schemes and could tackle the excessive overshoot deficiency in PI control for both the conventional RD process and RDWC.

A reactive dividing wall column (RDWC) integrates reactive distillation and multiproduct separation together, leading to the realization of process intensification. However, the reluctance to use it is due to the uncontrollable vapor split, which is self-regulated according to the flow resistance on each side of the partition wall. For some cases, the pressure of reaction and multicomponent separation is different, which results in an energy penalty if we directly integrate these units together, for example, formic acid (FA) production through methyl formate (MF) as presented in the part I of this series (Ind. Eng. Chem. Res. 2020, 59, 22215). To tackle these obstacles, a new reactive dividing wall column without the uncontrollable vapor split (NV-RDWC) by converting the bidirectional vapor–liquid thermal coupling to liquid-only transfer stream is proposed in this work. Optimization was first carried out by coupling the genetic algorithm (GA) and rigorous simulation. On the basis of the optimal solution, a detailed comparison was conducted between the conventional reactive distillation process, RDWC, and NV-RDWC, and the results show the superiority of NV-RDWC. Then two multiloop proportional-integral (PI) control structures and a model predictive control (MPC) for NV-RDWC were developed, respectively, to investigate their control performance. The dynamic response in the face of feed disturbance shows that MPC could give superior control performance for this complex coupled process.