Pervaporation process and apparatus for carrying out same

The invention is directed to pervaporation processes as well as to apparatuses for carrying out such processes. More in particular, the present invention is directed to improvements in pervaporation processes by improving mass transfer. Pervaporation hεl|s emerged as a promising separation technique, for instance for separating aqueous and organic solutions. The heat that is required for the evaporation of the compounds that are to be removed is generally supplied from the sensible heat carried by the liquid feed. However, the resulting temperature drop has been found to reduce the flux through the membrane. As a consequence, the membrane area required for a given separation is much larger as compared to the hypothetical isothermal case. In the art, several methods have been suggested to diminish the temperature drop in pervaporation systems, so as to avoid flux reduction. It has inter alia been suggested to use a series of alternating pervaporation modules and heat exchangers (see e.g. R. Rautenbach et al., J. Membr. Sci. 25 (1985) 25-54). However, such pervaporation systems increase the investment and maintenance costs considerably as compared to a single unit.

It has also been suggested to supply heat directly into the pervaporation module in order to maintain a constant retentate temperature (J. Bausa et al., Ind. Eng. Chem. Res. 39 (2000) 1658-1672). Various designs have been presented to supply energy directly into the pervaporation module, for instance using electrical resistances, see e.g. EP-A-O 118 760; or by using external heating with steam or another fluid (see M. Schleger et al., Desalination 163 (2004) 281-286). Although these methods can improve the pervaporation efficiency, which may result in a reduced membrane area, these known methods have drawbacks; because of safety reasons electrical resistances cannot be applied with flammable media, while external heating in multi-tubular modules requires a system of concentric tubes where the heating fluid is fed to the shell

side and the retentate is fed to the annulus of each set of concentric tubes. In general such pervaporation modules have a complex construction increasing the maintenance and total cost.

EP-A-O 294 827 and US-A-5 512 179 describe processes wherein a feed that is at least partly in the form of a vapor is separated by pervaporation.

It is an object of the present invention to provide a pervaporation process and apparatus that solves at least in part the above-mentioned disadvantages. It was found that this object can be met by carrying out the pervaporation process using a feed stream that is partly vaporized. Thus in a first aspect the present invention is directed to a pervaporation process wherein a feed stream is at least partly vaporized prior to separating it in a pervaporation unit. Preferably the vapour is produced by means of a distillation column, in particular the reboiler thereof. The pervaporation unit may be placed inside the distillation column or connected to it, such that the vapour/liquid mixture is contacted with the pervaporation unit. In accordance with the present invention it is also possible to retrofit the pervaporation unit in existing distillation equipment. Placing the pervaporation unit inside the distillation column is especially advantageous when one component has to be removed from a multicomponent mixture and this component can not be obtained from the distillation column as pure distillate or bottom product. Connecting the pervaporation unit to the distillation column is particularly convenient when binary mixtures have to be separated, as would for instance be the case when an azeotropic mixture is separated.

US-A-5 174 900 describes a process and an apparatus for treatment of a fluid feed using a wafer structure, which may comprise hollow fibres. Feeding a partially vaporized feed is not disclosed, nor suggested in US-A-5 174 900, nor are improvements with respect to mass transfer that are obtained with the present invention and which improvements are related to feeding a partially vaporized feed. US-A-5 232 593 also describes the use of a

certain cartridge, which may comprise hollow fibres and is also silent with respect to feeding a partially vaporized feed.

In a first embodiment the present invention is directed to a pervaporation process wherein a feed stream is separated into a permeate stream and a retentate stream by a membrane, wherein said feed stream contains vapour and liquid. It was found that vapour-liquid two-phase operation offers an effective way to control the retentate temperature drop in pervaporation units. Two-phase operation may be created by vaporizing a fraction of the feed or retentate. The vapour flow is then combined with the liquid feed stream and the combined streams are fed to the pervaporation unit. The term "pervaporation" as used herein refers to a separation process which is based on difference in interaction with the separating membrane. Typically, the components are separated by the membrane by applying a lower pressure on the filtrate side than on the feed side of the membrane. A typical configuration comprises a vacuum pump that is situated on the filtrate side of the membrane, as schematically depicted in Figure 1. By applying a pressure that is below the feed pressure, one of the components will preferentially pass through the membrane. Subsequently it may be condensed on the filtrate side, while the retentate is decreased in its concentration of this component. Instead of, or in addition to, applying a so-called "sweep"-gas or

-vapour may be useful in operating the pervaporation unit. The permeate is diluted with the sweep-stream and consequently the partial pressure of the permeating species is low.

The invention is furthermore directed to a separation unit that comprises a pervaporation unit that can be fed with a mixture of vapour and liquid. Such a separation unit (1) is schematically depicted in Figure 1. It comprises at least one membrane (2). A feed stream comprising liquid (3) and vapour (4) is fed to unit (1). The vapour in the feed stream may be obtained e.g. by vaporizing a small part of the liquid feed stream. Vacuum pump (6) is used to decrease the pressure on the permeate side of the membrane, e.g. below 0.1

bara, for instance from 0.01 to 0.05 bara. Condenser (5) is used to cool and condensate the permeate, which may then be collected. A purified retentate stream (7) is obtained.

The vapour phase that is present in the feed in accordance with the present invention has a multifunctional role, especially in low liquid flow rate applications. It reduces temperature and concentration polarization near the membrane surface by promoting turbulence, thus increasing the pervaporation efficiency. Surprisingly it was found that also the selectivity of the pervaporation may increase in accordance with the present invention. The vapour may also supply latent heat to the liquid by condensation. Two-phase operation in multi-tubular systems can be applied for flow inside tubes or for transversal flow externally to tubes in co-current or counter-current operation. Pervaporation units with counter-current vapour- liquid operation could be placed directly inside a distillation column replacing a section of packing or trays.

Pervaporation processes may be characterized by the two quantities permeance and selectivity. For dehydration separations, the water permeance F _w is defined as the ratio of the water flux (J _w ) through the membrane and the water partial vapour pressure difference over the membrane.

J,., ⁽ D

/; = ^χ _* r _* pl -y _* p ^v

where pi is the vapour pressure of pure water, p ^p is the total pressure on the permeate side and χ _w is the water activity coefficient in the retentate.

The intrinsic selectivity and the process selectivity are defined according to equation 2. x and y are the molar fractions in the liquid and permeate sides, respectively, for water w and the other compound i .

λ (2) x..

CC =

For the intrinsic selectivity, the local retentate molar fractions on the membrane surface can be used. The feed liquid molar fraction can be used for the process selectivity.

The concentration polarization index (CPI) can be defined as the ratio of the local water concentration on the membrane surface and the local water concentration in the liquid bulk. Temperature polarization can be defined as the local difference between the bulk temperature in the retentate and the membrane surface temperature.

The process of the present invention may be carried out in different flow conditions. Several two-phase flow regimes exist, see for instance Y. Taitel et al., AIChE Journal 26 (1980) 345-354. One typical regime is the slug flow regime, a schematic representation of a unit that is operated under slug flow conditions is shown in Figure 2. In the slug flow regime a Taylor bubble zone can be distinguished with a falling liquid film, and a liquid slug zone (see also R.C. Fernandes, et al., AIChE Journal 29 (1983) 981-989). The liquid slug consists of a wake zone and a remaining liquid slug. Mass and heat transfer to the membrane surface in the slug flow regime are mainly determined by the liquid and gas velocities and the length of each zone.

The falling film is formed when liquid from the liquid slug zone is pushed to the tube wall by the Taylor bubble forming a thin film that travels downwards around the bubble. The velocity field in the liquid slug ahead of the Taylor bubble then changes. The effect of the rising bubble on the liquid velocity field ahead of it is restricted to distances of about ID from the bubble tip, where D is the diameter of the tube. The falling film enters into the liquid slug zone generating a zone of high turbulence called wake zone. The size of the wake zone is typically between 5D and QD (A.M.F.R. Pinto et al., Experiments in Fluids 31 (2001) 643-652). The high velocities in the wake zone

decrease, in a sufficiently long liquid slug (>6D-7D), to low values causing a relatively calm liquid behind the wake zone.

In one embodiment the feed stream is at least partly vaporized prior to feeding it into the pervaporation unit containing the membrane. The vaporization can be done using known means, such as a boiler or heat exchanger.

It is also possible to produce the vapour by means of a distillation column, in particular the reboiler thereof. In this embodiment, the pervaporation unit is placed inside the distillation column or connected to it such that the vapour/liquid mixture is contacted with the pervaporation unit. In accordance with the present invention it is also possible to retrofit the pervaporation unit in existing distillation equipment.

The amount of vapour in the feed stream usually depends on the amount of component to be removed and is typically from 0.1 - 10, preferably from 1-5 wt.%, more preferably from 2-4 wt.%, based on the total weight of the feed stream. In principle the heat that is needed for the vaporization of the component to be removed (the permeate), may be supplied by condensation of the vapour present in the feed stream. So, at equal heat of vaporization, the amount of vapour in the feed stream is in general equal to, or larger than, the amount of component to be removed.

Vapour flow rates that are too low are generally not desirable because they do not provide for the desired effect with respect to mixing and improved mass transfer. Values that are too high on the other hand, are generally not cost-effective because the heat that is required to vaporize a high fraction does not generally lead to corresponding improvements in permeativity and/or selectivity. Vapour flow rates can be adjusted during operation so that they are within a suitable range. Suitable values may depend on the type of configuration used and other process parameters. For instance, in case the selective membrane layer is on the outside of tubular membranes gas phase Reynolds numbers (Re) are preferably from 5 to 200, e.g. between 9

to 102. Typically for that situation, the unit is operated at Re = 102. When Re drops to 9, additional vapour is injected. (Re = p v -D/r\; wherein p is the gas density; υ is the superficial gas velocity; D is the inside hydraulic diameter of the empty module housing; and η is the gas viscosity). Typical gas velocities range from 0.001 to 1 m/s, e.g. from 0.05 to 0.2 m/s in case the selective membrane layer is on the inside of tubular membranes.

The membranes as applied in the present invention may have any suitable size and shape, such as tubular membranes and flat sheets, or sheets that are configured in a certain fashion, e.g. spirally wound up. Furthermore, so-called modules can be used. Generally, tubular membrane structures are preferred because they result in a more compact design of the apparatus. For tubular membranes, the selective membrane layer may be present on the inside of the tubes or on the outside. When the selective membrane layer is on the inside of a tubular membrane, the vapour and liquid flow through the tube enhances mass transfer to the membrane surface. The permeate flows on the outside. This embodiment is schematically illustrated in Figures 4 and 5. Several kind of two-phase regimes could occur in the process of the present invention depending on the liquid and vapour flow rates, from slug flow to churn and annular flow.

If the selective membrane layer is on the outside of a tubular membrane, the vapour and liquid can be lead over the outside of the tubular membranes in co-current or counter-current mode. The permeate will then flow on the inside of the membrane tubes.

Different terms may be employed to describe tubular membranes (viz. membranes having the shape of a tube), depending in particular on their diameter (see for instance "Nomenclature and symbols in membrane science and technology", G.H. Koops, Sept. (1995)"). Hollow fibre membranes are typically tubular membranes having diameter < 1.5 mm. Capillary membranes

are typically tubular membranes having a diameter of 1.5 to 5 mm. Tubular membranes having a diameter > 5 mm are often simply referred to as membrane tubes. All of these tubular membranes can be used in accordance with the present invention. In one specific embodiment of the present invention the pervaporation step is carried out using hollow tube-shaped membranes because they allow for compact and flexible design and good separation. If hollow tube-shaped membranes are used it is preferred to apply the separation membrane (typically in the form of a coating) on the outside surface thereof, because this provides a membrane surface area that is considerably larger when compared with hollow tube-shaped membranes coated on the inside. Hollow tube-shaped membranes when used in accordance with the present invention preferably have an outer diameter of 0.5 -50 mm, typically a wall thickness 1-2 mm and variable length (typically 0.1 to 1 m or more). Hollow tube-shaped membranes may for instance prepared by dipcoating a hollow tube-shaped membrane support in a suitable solution.

In general, typical dimensions for hollow tube-shaped membranes when used in accordance with the present invention are 0.5-50 mm outer diameter, preferably 1-25 mm, more preferably 3-15 mm, having a wall thickness of typically 0.5 - 5 mm, preferably 1 - 2 mm, e.g. about 1.5 mm. The length of the hollow tube-shaped membranes may vary widely depending on the application, e.g. from several centimeters, e.g. 10 cm to one meter or more.

When hollow tube-shaped membranes with the separation membrane on the outside are used, it is preferred to apply the vacuum for the pervaporation on the inside of the tube, so that the filtrate is collected in the hollow tube-shaped membranes. From there it can be collected in a suitable duct that is in connection with the lumen of the hollow tube-shaped membranes.

It is also possible to combine the pervaporation process of the present invention with chemical reaction. To this end it may be advantageous

to apply a catalyst in the membrane or on its surface, such as zeolites, oxides, ion exchange resins or combinations thereof. In this way reactive pervaporation processes, wherein for instance the desired product is separated from the reaction mixture so that the reaction equilibrium shifts to the favourable side. Also solid acids and enclosed homogeneous acids can be applied in or on the surface of the membrane, for instance to bring about a certain chemical conversion.

The pervaporation process according to the invention may be employed for the separation of all kinds of mixtures. An important advantage is the high selectivity and also the high energy efficiency. Preferred applications include those wherein the feed stream is selected from azeotropic mixture, in particular water/ethanol; methanol/methyl-tert-butylether; and isopropyl alcohol/water. The invention may furthermore be employed in acetic acid dehydration; benzene-cyclohexane separation; production of esters; dimethyl carbonate production; dehydration of alcohols; MTBE production; methylisobutylketone production; hexadecyl eruciate production; DMF (dimethylformamide) dehydration or propyl propionate production.

Other examples of processes in which the present invention may find use are separation or purification of alcohols (methanol, ethanol, propanol (both isomers), butanol (all isomers), pentanol (all isomers), cyclohexanol, benzyl alcohol, and the like), separation or purification of ketones (acetone, butanone (MEK), methyl isobutyl ketone (MIBK), and the like), separation or purification of aromatics (benzene, toluene, phenol, and the like), separation or purification of amines (triethylamine, pyridine, aniline, and the like), separation or purification of esters (methyl acetate, ethyl acetate, butyl acetate, and the like), separation or purification of ethers ethyl tert-butyl ether (ETBE), di-isopropyl ether (DIPE), tetrahydrofuran (THF), dioxane, and the like), separation or purification of nitriles, such as acetonitrile, the separation or purification of aliphatics, in particular of C3 - Cs aliphatics, and the

separation or purification of chlorinated hydrocarbons (dichloromethane, perchloroethylene, and the like).

An apparatus according to the present invention comprises a boiler for evaporating a fraction of a liquid stream so that a vapour stream can be obtained, a suitable duct for combining the remaining liquid stream and said vapour to a mixture and for feeding said mixture to a pervaporation unit, which pervaporation unit comprises a membrane and a vacuum pump which is connected to the room that is on the permeate side of the membrane.

In a preferred embodiment, the apparatus according to the present invention comprises a multitude of the hollow, tube-shaped membranes, which are arranged in a layer, which layer comprises two or more parallel tube- shaped membranes. This embodiment is schematically depicted in Figure 3a.

It is preferred that the apparatus comprises two or more of the layers arranged in a vertical manner, as schematically indicated in Figure 3b. In this embodiment the longest axis of the hollow tube-shaped membranes of each layer is perpendicular to the longest axis of the hollow tube-shaped membranes of the neighbouring layer or layers. The vacuum pump is preferably connected to the room that is formed by the inside of said hollow tube-shaped membranes. Different tube inclinations can be used (horizontal, vertical or in between). As compared with the known way of operating, single-phase flow, two-phase flow in accordance will increase the performance of tubular membranes regardless the inclination.

Examples

Three kinds of experiments were performed: using air in a lab scale setup and using vapour and air in a bench scale pervaporation unit. The influence of the secondary phase on temperature and concentration polarization is demonstrated by measuring flux and selectivity in experiments with air. For the lab scale experimental setup, the extent of concentration and

temperature polarization have been also calculated, based on measurements of bubble rise velocity and bubble and liquid sizes. Using vapour, the additional effect of heat supply by condensation is shown in the bench scale pervaporation unit. Dehydration of 1,4-dioxane has been performed in the lab scale setup and dehydration of isopropyl alcohol (IPA) has been carried out in the bench scale setup using silica membranes.

The influence of the addition of a vapour phase on concentration and temperature polarization in a pervaporation process was determined experimentally in a single membrane tube based on the dehydration of 1,4- dioxane and in a bench scale pervaporation unit for dehydration of IPA. The additional effect of heat supply by a condensing vapour was measured for dehydration of IPA in the bench scale pervaporation unit, where vapour was obtained by partially vaporizing the IPA-water feed mixture.

Lab scale setup; air as secondary phase

The setup for the air-liquid two-phase experiments is shown in Figure 4. The ceramic pervaporation membrane supplied by Pervatech (The Netherlands) consisted of an alumina support tube with 7 mm internal diameter and 50 cm length, internally coated with a silica separation layer. Heating oil was supplied to the jacket of the storage vessel from an external heating bath to control the feed temperature to the pervaporation membrane at 70 °C. The feed liquid, an aqueous solution of 1,4-dioxane with 9.0 wt. % water, was pumped to a chamber at the bottom of the vertically placed membrane. The liquid superficial velocity in the tube was 0.048 m/s (110 mL/min), for both laminar flow and two phase experiments. For two-phase experiments air was injected in the same chamber using a solenoid valve. Liquid and air were recycled to the storage vessel after passing the pervaporation membrane and air was released in the storage vessel.

Prior to the pervaporation experiments the pervaporation tube was replaced by an identical transparent glass tube. A digital camera was

positioned at a height of 40 cm from the tube bottom to measure bubble size, liquid slug size and bubble rise velocity at air flow rates between 40 and 250 mL/min and several solenoid valve frequencies. Under the experimental conditions used in this study slug flow was created. The experimental results in the glass tube were used to obtain a desired size of the Taylor bubbles and liquid slugs in the membrane tube by adjusting the vapour flow rate, and the opening and closure frequency of the solenoid valve.

In order to characterize the membrane tube, additional single phase pervaporation experiments were performed in the turbulent regime using high liquid superficial velocities of above 3 m/s. At these superficial velocities the concentration and temperature drop in the liquid and the concentration and temperature polarization on the membrane are negligible.

The permeate side of the experimental setup consisted of a system of cold traps and a vacuum pump, shown in Figure 4. A metering needle valve was used to adjust the vacuum pressure at 10 mbar at the permeate side outside the membrane tube. The vapour at the permeate side was condensed using liquid nitrogen and was weighted as a function of time to determine the total flux. The water concentration in the retentate was measured by Karl- Fischer titration. The water concentration in the permeate stream was measured by gas chromatography using a TCD detector. The feed water concentration was virtually constant during one experiment and was readjusted when necessary.

Bench - scale setup; vapour as secondary phase The bench scale unit, built by GTI (The Netherlands) is sketched in

Figure 5. Two pervaporation modules identified as Pervl and Perv2, supplied by Pervatech (The Netherlands), were placed vertically on top of the reboiler. Each membrane module contained 7 pervaporation membrane tubes, which are the same tubes as used in the lab scale setup, in a shell and tube configuration. At the shell side the vacuum was adjusted to 85 mbara. At this

pressure cooling water can be used to condense the permeate stream that is collected in independent cold trap systems for each pervaporation module. Single phase experiments and two-phase experiments using air or vapour were performed. An IP A- water solution of 17 wt % water was fed at 7 L/hr to the reboiler and the membrane module with a feed pump, as shown Figure 5. The heating coil duty in the reboiler was used to control either the liquid feed temperature to the pervaporation modules for single-phase experiments and experiments with air or the vapour feed flow rate for vapor-liquid two-phase experiments using a PID controller. For gas-liquid two-phase pervaporation experiments, air was used with flow rates between 60 and 700 mL/s. The hydrodynamic regimes achieved were different from the lab scale experiments. Calculations according to Taitel et al. indicate that slug flow was obtained at low vapour and air flow rates while for higher flow rates churn flow and annular flow were reached. The inlet and outlet temperatures, the permeate pressure in each module and the liquid levels in the reboiler and storage vessel were monitored using the software InTouch®. The liquid-vapor mixture remaining at the top of the pervaporation modules was condensed using cooling water and was recycled to the storage tank. The retentate and permeate water concentrations were measured with Karl-Fisher titration and gas chromatography, respectively. The water content in the liquid feed was monitored and was kept constant by recycling the permeate stream to the storage tank.

Results

Dioxane - water system; gas - liquid pervaporation

Experimental values of the total flux through the membrane as a function of the air flow rate for two-phase pervaporation in slug flow of 1,4 dioxane at 343 K and 9 wt. % water are presented in Figure 6 for two different liquid slug sizes at a liquid flow rate of 110 mL/niin. The average total flux for

laminar flow at zero gas flow is shown for comparison. The total flux through the membrane strongly increases when a small amount of gas is injected in the feed stream. Low air flow rates, as low as 0.7 mL/s, are sufficient to increase the total flux twofold as compared to single-phase pervaporation in laminar flow. The retentate temperature drop in the slug flow experiments is around 9 K. Experimental values of the total flux are generally higher for a liquid slug of 5.7 cm than for 4.6 cm. The total flux decreases slightly at increasing gas flow rate, in particular for a liquid slug length of 4.6 cm because of the higher ratio of the Taylor bubble length to the slug unit length. For liquid slugs of 5.7 cm this reduction is less important since the mass transfer coefficient in the wake zone is higher for a liquid slug of 5.7 cm than for 4.6 cm slugs. The velocities in the wake zone of a liquid slug shorter than 7D, i.e. 4.6 cm, are reduced by the next ascending Taylor bubble.

The process selectivity as a function of air flow rate using air-liquid two phase is presented in Figure 7. It shows a low process selectivity for single phase in laminar flow. However, the process selectivity for water increases when air is injected and increases slightly at increasing gas flow rates.

Using a small amount of air, as low as 0.5 mL/s, increases the flux at least twofold and the selectivity 50% compared with laminar single phase experiments.

Experimental values on slug flow and theoretical prediction of polarization.

As mentioned above, the mass and heat transfer are calculated based on experimental data of Taylor bubble rise velocity and Taylor bubble and liquid slug sizes. The mass and heat transfer coefficients are used to calculate the concentration and temperature polarization for dehydration of 1,4-dioxane at the experimental conditions of the measurements.

The bubble sizes measured for two different liquid slug sizes as a function of the air superficial velocity for 1,4-dioxane are presented in Figure

8. The bubble size increases with the gas superficial velocity. For the two liquid slug lengths the bubbles sizes were similar at higher superficial velocities. Figure 9 shows the bubble rise velocity as a function of the total superficial velocity for 1,4-dioxane at 343 K and 9 wt. % water. The bubble rise velocity increases at a rate of 2.9 times the total superficial velocity as compared to the theoretical value of 2.3. This higher rate suggests bubble- bubble interactions due to the short liquid slug length used in the experiments.

Experimental values of the total flux and selectivity measured in single phase at turbulent and laminar conditions in the lab-scale setup are presented in Table I. The experimental values in turbulent conditions have been used as intrinsic parameters of the membrane. The intrinsic parameters of the membrane coupled with the mass and heat transfer coefficients in laminar flow have been used to calculate the total flux and selectivity. These calculated results are also included in Table I. It can be concluded that with the intrinsic parameters of the membrane a good prediction of the total flux and process selectivity are obtained. The same intrinsic parameters have been included for the calculation of flux in slug flow regime.

Table L Experimental and calculated total flux and selectivity for dewatering of 1,4-dioxane at 70 °C and 9.0 wt. % water in single phase.

* Feed flow rate is 110 ml/min

A comparison between calculated (predicted) and experimental values of the total flux for air-liquid two-phase pervaporation in slug flow for dewatering of 1,4-dioxane at 343 K is presented in Figure 10 (9 wt. % water at a liquid slug length of 5.7 cm) for slug flow together with experimental values

of the total flux for single phase in the turbulent and laminar flow regime and the total fluxes in slug flow regime. Calculated values of total flux for slug flow with isothermal conditions are also included. All calculated and experimental values for slug flow are for a liquid slug length of 5.7 cm. Good predictions of the total flux for slug flow have been obtained with a maximum difference of 6 %. For liquid slug lengths of 4.6 cm similarly good prediction, have been obtained. However, the predicted process selectivities are 40 % higher than the experimental values.

The calculated concentration polarization index and temperature polarization are 0.39 and 5.8 K, respectively, for laminar flow experiments.

Under slug flow conditions, the concentration polarization index increases to at least 0.83 and the temperature polarization reduces at most 1.8 K. The addition of gas has reduced concentration polarization more than 50% and temperature polarization more than 70% for dehydration of 1,4-dioxane. The total flux in the slug regime using air, with a liquid temperature drop of 8.5 K, is 20 % lower compared with experimental values in the turbulent regime. However, Figure 10 shows that an isothermal operation, e.g. using vapour instead air, in slug flow increases the total flux resulting in a performance that is only 10 % lower than under turbulent conditions.

IPA - water system; vapor-liquid pervaporation The total flux as a function of the vapour feed flow rate is presented in Figure 11 (dewatering of IPA at 353 K and 15 wt. % water) for both pervaporation modules connected in series in the bench scale pervaporation setup with a liquid flow rate of 117 niL/min. In single phase experiments total fluxes of 0.85 and 0.60 kg/m2 h have been measured for Pervl and Perv2, respectively. In absence of vapor, the 30 % lower performance of Perv2 is mainly due to the liquid temperature drop in the axial direction. The liquid feed temperature to Perv2 is around 9 K lower than to Pervl reducing the driving force for mass transfer.

Figure 11 shows that when vapour is added to the liquid feed higher fluxes are achieved and the differences in total flux between Pervl and Perv2 are eliminated. The total flux through the membrane increases at least twofold using relatively low vapour flow rates, as low as 12 mL/s, as compared to single phase laminar flow experiments. The water transport through silica membranes for 1,4-dioxane-water mixtures is higher than for IPA-water mixtures as shown in Figures 6 and 9+3, respectively. The total flux through the membranes increases with increasing vapour flow rates and in general the total fluxes in Perv2 are higher or similar to the fluxes in Pervl. The process selectivity is shown in Figure 12 (de watering of IPA at

353 K and 15 wt. % water) as a function of vapour flow rate. Vapour flow rates higher than 86 mL/s increase the selectivity twofold as compared to single- phase laminar flow experiments. The process selectivity increases only slightly at higher vapour flow rates.

Water - IPA system; experimental comparison of two-phase pervaporation using air and vapor.

Experimental results of the total flux for the dewatering of IPA using air-liquid two-phase pervaporation in the bench scale setup are shown in Figure 13, which shows a comparison of the measured total flux as a function of the air and vapour flow rates for dewatering of IPA at 353.5 K and 15 wt. % water. The lines are drawn to guide the eye. Also experimental results of the total flux using vapour have been included. When air is used Perv2 has a lower performance than Pervl due to the retentate temperature drop. The total flux increases around 20% for Pervl and 40% for Perv2 when a condensing vapour instead air is used (Figure 11+13). Also, using vapour, the differences in flux between the pervaporation modules are eliminated. The maximum obtained total flux through the pervaporation membrane in Pervl with air is 2.2 kg/ m ² h, but this flux increases to 3.1 kg/ m ² h when vapour is used.

From the experiments performed on the dehydration of 1,4-dioxane and isopropyl alcohol it follows that small amounts of air or vapour are sufficient to increase the total flux and selectivity in pervaporation modules at least twofold as compared to the performance in single phase laminar flow. For slug flow experiments, the total flux, selectivity, concentration and temperature polarization have been calculated based on experimental data of bubble size, liquid slug size and bubble rise velocity. Good predictions of the total flux and selectivity have been found for single phase and two-phase flow experiments. Compared with laminar flow experiments for the dehydration of 1,4-dioxane, calculations show that concentration polarization and temperature polarization are reduced by 50% and 70%, respectively. Water flux increases by 20% and 40% when vapour instead of air is used for dehydration of isopropyl alcohol in two pervaporation modules connected in series due to the reduction of the liquid temperature drop. Vapor-liquid two- phase pervaporation proves to increase fluxes through internally coated membranes by reducing the liquid temperature drop in pervaporation systems and for relatively long residence times by enhancing mass and heat transfer towards the membrane surface