To dehydrate alcohols, the pervaporation process has been used. In pervaporation process, there are various mass transfer steps such as diffusion in bulk, sorption into membrane, diffusion within membrane and evaporation. In this work, the tubular type module is manufactured, the modelling of the sorption phenomena is developed and the effect of the evaporation at downstream side on the mass transfer is investigated.
Instead of sheet type membrane generally used in pervaporation, the tubular type membrane has been prepared for the dehydration of ethanol and isopropanol. The membrane is formed either on inner or on outer surface of a porous ceramic support with cellulose acetate by the dip-coating and rotation-drying technique. Both of the ends of the tubular support are glazed to prevent the leak in longitudinal direction of the membrane and to be easily equippied into the pervaporation cell. The state of coating surface is investigated by SEM photography. The membrane consists of nonporous active layer, intermediate layer and nonprous support layer. The pore size of the ceramic support measured by mercury porosimeter is ca. 0.1㎛. The thickness of the active layer of CA membrane is ca. 30㎛. The intermediate layer exists between polymer and ceramic support and its thickness is estimated to be about 1㎛. It is difficult to strip the coated polymer from the ceramic support due to the existence of the intermediate layer, which plays a better role in membrane durability and strengthens membrane stability. The surface of the intermediate layer contacted with ceramic support is very rough. The selectivities of ethanol/water and IPA(isopropanol)/water system are in the range of 4~11 and 8~240 respectively. The overall mass transfer coefficient of water $k_{wo}$ is determined at three temperatures as a function of feed flow rate. In a membrane coated on inner surface, the dependence of $k_{wo}$ on the feed flow rate has a good agreement with the Leveque correlation, $k_{wo}$ ∝ $Re^{1/3}$. The activation energy of water in the ethanol dehydration is hardly affected by the feed flow rate. As the feed temperature increases, the mass transfer coefficients of water in ethanol and IPA dehydration are increased. It is found that the effect of temperature on flux is more considerable than that of feed flow rate.
The new model based on the possibility of contacting liquid molecule and the active site which has the numerical concept of free volume within thin polymer membrane. In this model, the amount of solvent sorbed in polymer is expressed by the relaxation rate and the sorption rate which are coupled with each other. Especially the new model can be expressed by $1-exp(-k_it^{1/2})$ in case of rapid relaxation. This model has been simulated by Runge-Kutta method for initial- value problem and compared with Fickian diffusion. The results show that this model is able to account well for Fickian and non-Fickian sorption behavior including sigmoid and two-stage sorption and satisfy the boundary condition at equilibrium. In addition, this model can explain the oversorption (overshoot of sorption).
In pervaporation process, the heat supply is required for the phase transition of the permeate and the temperature drop is resulted from the phase transition during the permeate evaporation. The temperature drop in small-scale pervaporation experiments is not easy to detect, because the pervaporation module is submerged in a constant- temperature water bath and its residence time is too short for the temperature drop to be conscious. This temperature drop affects the mass transfer rate, which cannot be observed in other membrane processes. Therefore if the correlations obtained from other membrane processes without phase transition were applied to the pervaporation process without modifications, incorrect results might be estimated. Thus the enhancement of the mass transfer rate by the increased flow rate can be misconceived as the boundary layer effect like other membrane processes. However it becomes generally known that the temperature drop affects the mass transfer in pilot pervaporation, where the constant temperature of the module is difficult to maintain due to its size and the high flux membrane. The pilot pervaporation is usually operated in low Reynolds number conditions unlike in laboratory pervaporation. In this study, the effect of the phase transition of the permeate on the temperature drop of the retentate has been investigated through the operation of the pilot pervaporator which is equipped with a plate-frame type module. The effect of the temperature drop on the mass transfer has been discussed in term of the heat supply ratio.
The heat supply ratio is defined as the relative amount of heat supplied from the retentate to the permeate for the phase transition and its value can be obtained from the heat and mass balances and from the measured temperature drop between inlet and outlet flow. The heat supply ratio has lain between 0.03~0.7 during the experiments and varies as operating condition. Its value decreases with the increase of the feed flow rate but the effect of the feed temperature on the heat supply ratio does not come out clear. As the heat supply ratio increases, the temperature drop between inlet and outlet flow increases and it makes the mass transfer rate lower. The temperature drop also increases with the increase of the feed temperature. The reason is that the heat of evaporation is more needed by the enhancement of the permeation flux. In other membrane processes without phase transition, the flow rate affects mainly the concentration polarization but in pervaporation it affects not only the concentration polarization but also the temperature profile within the module. So the reduction of the mass transfer rate is induced by the temperature drop, which becomes more explicit as the feed temperature and the heat supply ratio increase.
Therefore the changes of the permeation flux by the temperature drop as well as the resistance in the boundary layer by the concentration polarization can be affected by the flow rate. This phenomena need not to be considered in other membrane processes but in the pervaporation it is not negligible. In case of vapor feed near boiling point, however, the temperature drop is not so dominant and its effect is negligible, because of the energy supply by the phase transition of the retentate.
