STUDIES OF ADSORPTION/DESORPTION IN PACKED BED OF MOLECULAR SIEVES

Abstract

Adsorption-desorption as a unit operation for the separation of adsorbable gases or vapours from inert or sparingly adsorbable carrier gases has been in use for a long time. The application of synthetic zeolites has further enhanced its utility. The present thesis concerns mainly with the development of a new desorption technique, which allows energy efficient recovery of adsorbate in pure form. In this technique, desorption of adsorbate is carried out by heating the adsorbent bed with the help of electrical heaters, imbedded in the bed. In order to study the proposed desorption technique, two types of experimental set-ups, called as Main and Auxiliary units, have been designed and fabricated. The first one (Main unit) is used to study the various aspects of the adsorption/desorption technique, including the experiments required for the estimation of thermal transport properties of adsorber beds and the second one (Auxiliary unit ) is used to carry out the single pellet adsorption studies, which may be used to prepare adsorption isotherms. The Main Unit consists of two types of adsorbers - small diameter (Laboratory size) and large diameter (Pilot plant size). Any one of them can be kept on-line for adsorption/desorption experiments. Water vapour-Zeolite (Type-13X) as adsorbateadsorbent system has been selected for the present study. To convert the proposed desorption technique in to a commercially viable process, experimental and theoretical investigations have been planned with the following objectives in mind. - Dry bed conduction studies in small diameter and large diameter adsorbers with a single coaxial heater and the estimation of thermal diffusivity of adsorber bed. - Dry bed conduction studies in large diameter adsorber using multiple parallel heaters. - Single pellet adsorption studies to estimate adsorption equilibrium constants and heat of adsorption. - Desorption studies in small diameter adsorber. Besides, it is well known that the molecular sieves possess a very low thermal conductivity. Therefore, the heaters, which are to be imbedded in the bed, should be able to work in a low conducting environment. However, such heaters were not available, therefore, the development of such heaters has also been included in the research programme. For simulating the dry bed conduction and desorption experiments, suitable mathematical models are required. Recently few models have been developed by Kothandapani et al.[2], Moharir et al.[3], and Moharir and Saraf [4-6] for various facets of proposed desorption technique. These XI are quite general in nature and can be used to simulate the conduction and desorption experiments in a small diameter adsorber with a single coaxial heater or in a large diameter adsorber with several patterns of multiple parallel heaters. These mathematical models have been used in the present studies. Dry bed conduction experiments have been carried out in small diameter as well as in large diameter adsorber beds with a single coaxial heater. In these experiments, transient radial temperature rise profiles have been obtained. These experimentally observed data have been used to estimate effective thermal diffusivity of adsorber bed. For this purpose, a collective error index, E, is defined as follows: e = a a (TexP - Tsim )2 ^ * i,n i,n i=l n=l T. has been obtained by numerically solving the mathematical 1 ,n -1 •* 3 model, due to Kothandapani et al.[2] and Moharir et al.[3] for conduction unaccompanied by desorption from a single coaxial heater, using the Semi-Implicit Crank Nicholson Finite Difference method for a given value of thermal diffusivity. The value of thermal diffusivity, which minimizes E, has been taken as its desired estimate. Collective error index, E, has been minimized using Fibonacci search. Average estimated value of effective thermal diffusivity in these experiments is of the order of 4.0 x 10 cm2/s. It may Xll be noted that in these computations, thermal conductivity of adsorbent bed has been assumed independent of temperature. Using the estimated value of effective thermal diffusivity, the temperature rise versus time profiles at various radial locations have been obtained by numerically solving the appropriate mathematical model for small diameter adsorber. These simulated profiles have been compared with those obtained experimentally. The match between the two is reasonably good except at the locations very near the heater or very near the bed surface. The mismatch near the heater surface can be attributed to the thermal inertia of the heater which has been neglected in the simulation model and the mismatch near the bed surface is due to the heat loss through the bed insulation. The dry bed conduction studies, carried out in large diameter adsorber with a single coaxial heater, have revealed that only adsorbent within a radius of 6 to 7 cm from the heater axis is effectively heated. This is a consequence of the poor thermal conductivity of the adsorbent bed. Thus the use of multiple parallel heaters will be essential in the large diameter industrial adsorbers. Similar type of comparison has also been made for dry bed conduction experiments in large diameter adsorber with two heater-patterns - Triangular (with quadrant symmetry) and Hexagonal. The mathematical model for this case involves a two-dimensional conduction equation and has been solved XI11 by using Alternating Direction Implicit (ADD technique. The simulated temperature rise versus time profiles match reasonably well with those, observed experimentally, and show a similar trend as observed in case of small diameter adsorber. This validates the applicability of mathematical model and testifies the efficacy of used heater-patterns. It has also been observed that the Triangular heater-pattern is more efficient in comparison to Hexagonal heater-pattern for heating the bed. Adsorption experiments on a single pellet of molecular sieve have been performed on the Auxiliary unit at three temperatures, 310, 32 3 and 333 K. The humidity of the feed (dry Nitrogen-Water vapour) range from 0 to 75% relative humidity (i.e. the adsorbate concentration in the feed from 0 to 19 x 10~7 g mol H20/c.c. of feed). This range of water vapour concentration in the feed covers the entire range of concentrations at which the packed bed desorption experiments are performed. It is observed from the developed adsorption isotherms that it is linear at higher temperature (3 33 K) but becomes more nonlinear as temperature decreases. Three commonly used adsorption isotherms, viz. Linear, Freundlich and Langmuir, have been fitted to experimentally obtained data by using linear regression analysis. Heat of adsorption has been estimated by considering only the visibly linear portion of the three isotherms at XIV 310, 323 and 333 K. The estimated equilibrium constants for three temperatures are then used in the van't Hoff relationship to obtain the heat of adsorption by linear regression analysis. The desorption studies have been carried out on the small diameter adsorber only using an imbedded coaxial heater. It is felt that once the desorption model is validated for desorption with single coaxial heater in small diameter adsorber, its extension to desorption with multiple parallel heaters in the large diameter adsorber will hopefully be valid as has been the case with dry bed conduction models with single coaxial and multiple parallel heaters. Such a general simulation model can then be used for industrial adsorber design. For each desorption experiment, recovery of adsorbate has been recorded with time. Under the conditions of experiments, single pellet adsorption studies show that the adsorption data are successfully represented by Langmuir adsorption isotherm. Based upon this, total recovery of adsorbate has been calculated for each experiment. The experimentally observed and calculated recoveries agree reasonably well. It is also observed that in the beginning, rate of recovery of adsorbate is low. It increases gradually, attains a maxima and then decreases with the passage of time. This observed desorption behaviour may be explained on the basis of the variation of diffusivity (of adsrobate XV through Zeolite crystals) with temperature. It has been concluded that the desorption phenomenon as visualized in the proposed desorption technique is simultaneously controlled by conduction as well as by diffusion. Some important observations can be made from the experimental studies, which are as follows: - The proposed desorption technique with imbedded heating source provides a near complete desorption in a reasonable time. - The desorption duration is smaller than the adsorption time required for complete saturation of the dry bed. This is important because it will allow a two bed system (one bed undergoing adsorption and another desorption) to function in a cyclic steady state in industrial applications. Also the shorter desorption time allows a period during which the bed may cool down to be subsequently used for adsorption. Simulation of the proposed desorption method with a single coaxial heater has also been carried out using the mathematical model proposed by Moharir and Saraf [6]. This model assumes the linearity of adsorption isotherm and conduction controlled desorption. Simulation has been attempted only for those experiments for which the assumption of the linearity of adsorption isotherm is approximately satisfied. XVI The simulation results indicate that the simulated desorption is considerably faster and is complete in about 2 hours in all the cases, selected for study, whereas the observed desorption is sluggish and takes about 6 to 8 hours for completion. This also supports our earlier observation that the outward diffusion of adsorbate from the zeolite crystals, alongwith conduction, seems to govern the desorption. The simulation model should, therefore, incorporate the diffusion considerations also. However, it is not attempted in the present thesis as it would require excessive computational efforts. Based upon the experimental and theoretical investigations, it is our view that the proposed desorption technique appears to be commercially viable. However, some more efforts are needed, which alongwith the main conclusions of the thesis are reported in the chapter on "Conclusions and Recommendations".