Experimental and Simulation Studies of Industrial Scale Finishing Reactive Distillation Column

Abstract

RD is the process developed by integrating two different operations (chemical reaction and multi-stage distillation) simultaneously in a single unit. Reactive distillation (RD) has been successfully implemented for the production of the Ethyl acetate (EtAc).The RD feature of shifting the equilibrium in the forward direction is useful for EtAc production by the reversible esterification reaction of acetic acid with ethanol. The main advantages of using the RD with a pre-reactor for the production of the EtAc is that, the existing plants based on the conventional process can be easily revamped to use the RD technology. Also, in case of RD with pre-reactor, a smaller RD column (as compared to complete RD process) is required which reduces the capital investment. At IOL Chemicals and Pharmaceuticals Ltd., the EtAc is produced by the esterification of HAc with EtOH in the presence of the homogeneous acid catalyst (para-toluene-sulfonic acid) using RD with pre-reactor. Most of the reaction takes place in the reactor and the reacted mixture is fed to the column where simultaneous removal of product improves the conversion and product yield. The work reported in this thesis was aimed at generating the experimental data from an industrial scale RD column (situated at IOL Chemicals and Pharmaceuticals Limited, Trident Complex, Barnala, Punjab, India) and to compare the simulation results with this RD column data. The provision for the collection of vapor and liquid samples from each tray was made during a scheduled shutdown of the plant. The simulation was carried out using the rate based model of Aspen PlusTM, as the column hardware information was readily available. A new set of thermodynamic parameters for the NRTL model was established for the highly the non-ideal liquid mixture, of this reaction system, by regression of the vapor-liquid equilibrium data available in literature. Predicted vapor and liquid composition profiles are in good agreement with plant data. The simulation results indicate that stages 20 to 30 are the most active stages of the column from the hydrodynamic point of view. Like a conventional distillation column, in this case too, an increase in organic reflux (keeping the distillate flow rate constant) causes an asymptotic increase in distillate purity (i.e., EtAc concentration). A linear increase in re-boiler duty and decrease in ethanol conversion was also observed with an increase in reflux flow rate. However, at high reflux flows, a higher EtAc concentration suppresses the forward reaction, thus, overall conversion in the column decreases. The simulation results further show that the column is having more number of trays (about 6 trays) in the rectification section than required. The effect of feed stage location is also discussed to utilize these extra trays. The results show that re-boiler and condenser duties decrease on moving the feed stage up in the column (from stage 39 to stage 33). Also, the ethanol conversion increases as the reaction zone in the column increases with the shifting of feed stage from stage 39 to stage 33. It is observed that the extra stages can be utilized in a more effective way by changing the feed location (moving the feed stage up in the column) so that reaction zone in the column increased; and by addition of a fresh feed of ethanol at stage 49 to improve the forward reaction. A fresh feed of ethanol (150 kg/h) is added in the down-comer of 49th stage and the feed from the pre-reactor was introduced at stage 34. The ethanol conversion in the RD column increased from 17.43 kg to 93.44 kg, the ethanol conversion (wt%) increased to 39.94 % from 18.76%. The specific energy (kcal/kg of EtAc) requirement of re-boiler reduced from 1115 to 1054, and condenser duty (kcal/kg of EtAc) reduced from 684 to 654. Comparison of the column composition profiles, for liquid and vapor phases, with EQ and rate-based model simulations show that, for all the components, the rate based model’s predictions are far more superior to the EQ model predictions. The rate-based model predicted lower EtAc generation rates, than the EQ model, as introduction of mass transfer resistance reduced the reaction rates.