A STUDY ON THE EFFECT OF SUBDIVIDING AN ADSORPTION BED IN A SIMULATED COUNTER-CURRENT ADSORPTION SYSTEM

Abstract

Continuous separation of an aqueous binary mixture of carbohydrates was carried out on a simulated countercurrent unit. Silica gel was used as the adsorbent and water was used as the eluent throughout the experiments. The simulated countercurrent unit comprised of eight identical columns, each 1.4cm x 47.5cm length, connected in series. The columns were connected so that the system consisted of three sections. The mixture studied was dextran (mol. wt. 9400) and fructose. Basic data concerning the adsorption equilibrium constant and mass transfer had to be obtained to determine the separation factor of the two components. These data were used to determine the flow rates through each section of the simulated countercurrent unit. Studies were conducted with both pulse and step inputs to the system. Results of the pulse input experiments were analysed by the method of moments and data on the adsorption equilibrium constant and mass transfer coefficient was extracted. The adsorption equilibrium constant and mass transfer coefficient of dextran was found to be 0.24 cm3 / cm3 and 0.09 min-1 respectively while that of fructose was 0.91 cm3 /cm3 and 5.52 min-1 .The theoretical profile of the breakthrough curve (using data from the pulse response experiments) compared fairly well with the experimental results. Experiments were conducted to study the effect of subdivision of an adsorbent bed in countercurrent mode of operation. Purities of up to 89% were recorded for the extract stream and 100% were recorded for the raffinate stream. Recoveries of up to 95% were recorded for the extract stream and up to 68% for the raffinate stream. It was found that the purity of the extract stream was higher by about 10% when the adsorbent bed was subdivided. There was no change in the purity of the raffinate stream. It was also observed that the recovery of the products was the same irrespective of the subdivision of the bed. This corresponds to the theoretical prediction of improvement in purity with a greater subdivision of the bed. The other concept that was examined was the question of the middle of the switch profile corresponding to the time average concentration profile. Experimental results proved otherwise. This is very important because it means that the concentration measured at the middle of the switch is not necessarily a true representation of the concentration profile. The experimental results were compared with the theoretical profiles. The theoretical profile was obtained by solving the axially dispersed plug flow model through the method of orthogonal collocation. The resulting differential equations were solved through a standard integration routine. The theoretical profile matched well with the experimental profile. A discrepancy noted in an earlier work with respect to the three section flow scheme was not observed in this work. This has been attributed to a problem with the mass balance of the system. A dispersed plug flow predicts the experimental results quite satisfactorily. The cyclic process is attractive for binary mixtures since it comprises of a normal column which is easy to pack and to scale up.