Unraveling the influence of residence time distribution on the performance of reactive distillation – process optimization and experimental validation

This work is the first to explore the influence and illustrate the impact of the stage average residence time distribution (SARTD) in the reaction section on the performance of reactive distillation (RD) systems, providing guidance for the design of the internals and industrial applications of RD processes. The experimental validation and process optimization were performed using the ethyl levulinate production as case study. Kinetics of the esterification reaction between levulinic acid (LA) and ethanol was explored using an ion
exchange resin as catalyst, and then fitted by a pseudo-homogeneous model, providing basic data for subsequent RD pilot-scale experiment and process simulations. The RD experiment
was firstly successfully carried out for two catalyst-loading conditions (uniform or non uniform per stage) to illustrate the impact of SARDT on the RD performance. The experimental results show that the conversion of LA increases by about 10% after catalyst loading redistribution. A genetic algorithm (GA) with global convergence and fast convergence speed was used to realize the optimization of SARTD instead of using arbitrary settings. The objective function to be minimized was the reboiler duty per unit product (RDP). The optimization results show that the conversion of LA increases by 3.06% (percentage
points) and RDP decreases by 10.09% if the SARTD is optimally redistributed. Based on this, it can be predicted that the total catalyst-loading can be reduced by reasonable catalyst redistribution in the industrial applications of RD. In addition, a new method for optimal design of SARTD is proposed, paving the way for optimizing other RD processes.