DESORPTION STUDIES IN A PACKED BED OF MOLECULAR SIEVES USING IMBEDDED ELECTRICAL HEATERS

Abstract

Adsorption-Desorption is a widely used separation process in chemical and other process industries. The advent of zeolite molecular sieves has further enhanced its utility due to their specificity of separation by size. The various desorption processes may be classified as thermal swing, pressure swing, purge gas stripping, and displacement desorption. Thermal swing desorption process is used in numerous purification systems; its applications are growing at a steady rate for last three and a half decades. Zeolite molecular sieves being capable of retaining its adsorption capacity even at relatively high temperatures make a good combination with this process. This desorption process is facilitated by passing a hot inert carrier gas through the adsorbent bed. In this technique, condensable adsorbates can easily be recovered by passing the effluent from the adsorber through a condenser. However, if the adsorbate is noncondensable, then additional separation unit shall naturally be required to remove the adsorbate from carrier gas. Besides, heating of carrier gas consumes a large amount of energy and only a small fraction of it is utilized in desorption. Therefore, in order to recover contamination free noncondensable adsorbate without any significant loss and also to reduce the energy consumption, it is necessary to heat the bed without carrier gas. Jacketted heating is possible but it loses its applicability to large diameter beds used commercially, because of the relatively poor thermal conductivity of molecular sieves. Thus, the packed bed desorption in which bed is heated by internally imbedded electrical heaters, appears to be a potential alternative. To the best of our knowledge potentials of this technique h.ave not yet been fully explored. In our Laboratory, preliminary studies on this desorption process have proved its viability [Prasad (1988)]. 13X molecular sieves manufactured indigenously, were used as adsorbent. The bed saturated with water vapour was desorbed using an electrical heater, imbedded in the bed. Main emphasis in this research work was on dry bed conduction studies and so the effective thermal conductivity/diffusivity of bed was determined. It has been reported in the literature that the thermal conductivity of a packed bed depends upon temperature, but the dependence is scarcely estimated. Furthermore, these studies include few experimental runs on desorption. A mathematical model has also been proposed for this desorption process, which assumed desorption to be completely conduction controlled and ignored the intraparticle diffusional effects. Due to this reason, model predictions were quite far from the experimental ones; predicted and experimental desorption times are about 2.5 hour and 8-9 hour respectively. Besides, 13X molecular sieves used were indigenously manufactured, for which no systematic characterization was available and done by Prasad (1988). In view of the viability of proposed desorption process, the present research project has been undertaken. In formulating the objectives, it has been kept in mind that the results of this research work may further enhance our knowledge of this desorption process, so that it may be designed and used commercially. In the present study too,water vapour-13X molecular sieves has been considered as adsorbate-adsorbent system. The main objectives of the thesis are confined to three aspects of this desorption process, namely Characterization of 13X molecular sieves, Estimation of k(T) for the packed bed, and Development of Mathematical model. Accordingly, the conclusions have been summarized below. Characterization of 13X molecular sieves has been done by determining various properties, e.g. surface area and total micropore volume, by using adsorption measurements. For this purpose various available methods, viz. BET, Langmuir, t-Plot, Dubinin- Radushkevich (DR) equation and Dubinin-Astakhov (DA) equation have been used. Adsorption measurements have been done using two gaseous adsorbates - nitrogen and argon which are having a symmetric molecule and are nonpolar in nature. Computational algorithm and software have been developed to use adsorption measurement data in DR and DA equations. A software for estimation of two parameters in DA equation has also been developed; one of them, i.e. exponent n, is of direct use. Besides, statistical analysis of estimated parameters has been carried out by plotting 99.95% confidence limits of predicted values and a contour for 95% joint confidence region. The analysis shows that the estimated parameters are correlated, and their values are correct. Value of exponent, n, in DA equation has been found to be 1.95 and 1.78 for nitrogen and argon adsorption on 13X molecular sieves, respectively. * Critical analyses of the methods used and the results obtained have been carried out. The analysis indicates that the exponent, 2, used in DR equation can be safely used for the determination of micropore volume and surface area of 13X molecular sieves. This enables one to avoid the use of DA equation which is relatively cumbersome and time consuming as it requires estimation of the value of exponent, n. Besides, surface area obtained from BET method is approximately 17-25% less than that of Langmuir's method, DR equation, and DA equation with the two adsorbates used in the present study. This observation is also in agreement with the observations made in literature. The values of surface area and micropore volume of 13X molecular sieves have been found to be 755.8 m2/g and 0.2686 cm3/g respectively. The results obtained are closely in agreement with the values reported in literature. In order to obtain transient radial temperature profiles, an experimental unit has been designed and fabricated. It consists mainly of a cylindrical column (I.D. = 94 mm, Height = 600 mm), packed with 13X molecular sieves. An electrical heater imbedded into the bed coaxially, was used to provide a source of thermal energy. The bed was heated under dry conditions using two heater wattages, e.g. 0.1 and 0.13333 kW/m. A PC based on-line data acquisition system has been developed on an existing data logger. A software has been developed in C language to display and store the measured temperatures at desired locations. The software is a 6400 lines code. It is interactive, user friendly, and possesses several advanced features such as context sensitive help and security of data against power failure. The data acquisition software is general in application as it can also be used to acquire three types of other commonly measured variables, viz. DC voltage, AC voltage, and resistance. Dependency of effective thermal conductivity of a bed of 13X molecular sieves on temperature has been estimated by using transient radial temperature profiles obtained in the dry packed bed. Effective thermal conductivity, k, has been estimated to vary linearly with temperature T. The correlation obtained for k is as follows : k (T) = 8.17635 x 10'5 + 10.915427 x 10"7 (T - T0) where, T0 = 303 K Using the above correlation, effective thermal conductivity of a packed bed of 13X molecular sieves varies from 8.18 x 10'5 to 22.15 x 10"5 kW/m K for a temperature range of 303-431 K. These values are in agreement with the values reported in literature for the catalytic and other similar materials. Statistical analysis of the estimated parameters of correlation has been carried out by plotting the joint confidence contour for an approximate 95% confidence level. Although, the dependence of k on T is linear but its use in the model partial differential equations meant for estimation of parameters results into a nonlinear estimation problem. Therefore, the joint confidence contour plotted is of deflated banana shape in a two parameter space. The elongated shape of the contour indicates that the estimated parameters are highly correlated. A mathematical model for this thermal desorption process has also been developed in the present research work. This model also includes the transport of adsorbate within the molecular sieve pellet by diffusion. For simplicity, linear isotherm is assumed to govern the equilibrium relationship. This model consists of partial differential equations for bed as well as for the pellet and is to be solved by a suitable numerical method with appropriate initial and boundary conditions. As the solution of this class of differential equations is Hi complex and requires large computer time, therefore, it is desirable to simplify it without sacrificing accuracy of its predictions. To do so, adsorbed phase concentration profile within the molecular sieves pellet is assumed to be parabolic in shape with respect to its radius throughout the desorption. It is reported in the literature that this assumption, at one hand, is true in almost all the cases and at all the times, significantly simplifies the model and saves the computation time substantially, at the other. Simplified model equations of the bed have been solved numerically by using control volume finite difference method. This method has the advantage of satisfying the conservation equations in the discretized form even if the discretization grid is coarse. This method forms the basis of the computational algorithm developed in the thesis. The predictions of the model proposed here, adequately match the experimental observations. It has been found in the present study that the thermal desorption of water vapour from molecular sieves 13X is mainly controlled by intraparticle diffusion of adsorbate, and the effective diffusivity of water vapour in molecular sieves is strongly concentration dependent. Therefore, in order to further improve the predictions of the model, concentration dependence of diffusivity has been investigated. It has been found that a relationship of following type adequately represents the desorption behaviour. D(q>- * ( 1 - a2 A f Where a2 = 0.87753 and D0 = 1.5 x 10" m2/s . It is our view that the research work carried out in the thesis has a wide scope for applications in industry. The technique of desorption and the proposed model will be useful in the applications where recovery of adsorbate is a must to improve the economics of the process. This type of system is particularly suitable for smaller units which are used for mobile purposes/vehicles. The developed model can also be used for modelling the desorption from a large size adsorption column used in the process industries. In such a case, multiple heaters imbedded into the bed on a specific pattern, e.g. triangular, shall be used as a source of thermal energy. The computational algorithm developed for a single heater can also be used for this system with little modification, but it may require a large computer time. Fortunately, with the development of fast computing machines, solution of model equations for desorption with diffusional transport considerations for a large grid size is now feasible.