Abstract: In the production of pure terephthalic acid (PTA), acetic acid (HAc) recovery system is one of the most important operation units. p-Xylene (PX) is oxidized to form PTA using HAc as the solvent, and water is generated in the reaction process, while there are unreacted reactant PX and by-product methyl acetate (MA) in the HAc solvent besides water. To reduce material consumption, the solvent mixture should be separated and concentrated for recycling by azeotropic distillation with an entrainer introduced into the system. How to realize the optimized operation of the process has always been the focus as well as a difficulty of the research. In recent years, most of the researches about the HAc dehydration system were dedicated to the issues of process synthesis, steady-state simulation and optimal design. A few works focused on the dynamic simulation and control of the process, but mainly aimed at the HAc dehydration tower. This paper is aimed at the dynamic simulation and control of another important tower of the system:entrainer recycle tower. Based on the steady-state model of our previous research, the dynamic simulation of entrainer recycle tower in the industrial acetic acid recovery system using n-propyl acetate as entrainer is carried out on the basis of Aspen Plus and Aspen Dynamics. Two kinds of dynamic control structures are designed for studying the dynamic control strategy to realize both high-efficiency and stability of the operation. The strategy 1 chooses sensitive board temperature signal to control the proportion of flow controllers, and the sensitivity of the plate temperature is used as a proportional controller's parameters to indirectly adjust. The strategy 2 chooses temperature-flow cascade controller to control heating through the measured sensitivity plate temperature directly, to adjust the separation tower by flow S3 flow. Then, the dynamic response is analyzed. In view of the simple structure of strategy 2, when the adjustment time and the stable value of strategy 1 are almost the same, the overshoot is smaller, indicating that the control structure is more stable, which is obviously a better choice for plant regulation. We find that direct control of heating stream through the temperature signal of sensitive plate is more efficient, which can shorten the transition process and reduce the corresponding material consumption and energy loss after disturbance appears. It provides a useful theoretical basis for the dynamic real-time optimization and advanced control of the recovery process, and the efficient optimized operation of the PTA plant.