You are on page 1of 12

Journal of Membrane Science 246 (2005) 145–156

Osmotic MD and other membrane distillation variants


Marek Gryta∗
Institute of Chemical and Environment Engineering, Technical University of Szczecin, ul. Pulaskiego 10, 70-322 Szczecin, Poland

Received 12 February 2004; received in revised form 20 July 2004; accepted 30 July 2004
Available online 24 November 2004

Abstract

A literature survey concerning the osmotic membrane distillation was presented. This process is discussed and compared with other types
of membrane distillation, particularly with direct-contact membrane distillation. A critical evaluation of the process principles and parameters
was included in this literature review. Process nomenclature, membrane properties, transport phenomena and module design are discussed in
detail. The mass driving force in osmotic membrane distillation is considered. Various options of process application are evaluated.
© 2004 Elsevier B.V. All rights reserved.

Keywords: Membrane distillation; Osmotic membrane distillation; Hydrophobic membrane

1. Introduction but they do not alter the MD process principles. The gradient
of partial pressure across the membrane may be formed not
Osmotic membrane distillation (OMD) is one of the mem- only by temperature difference but also by the concentration
brane distillation (MD) variants, operated at low temperature. difference and by the properties of solutions separated by the
The MD comprises a relatively novel membrane process, membrane [1,3,11–13]. The application of vacuum on the
which can be applied for the separation of various aqueous distillate side or the flow of dry gas also allows to obtain the
solutions [1]. The hydrophobic membranes, with the pores desired effect [1,14].
filled by the gas phase, are used in this process [1,2]. The The membrane distillation is carried out in various modes
hydrophobic nature of the membrane prevents penetration of differing in a way of permeate collection, the mass transfer
an aqueous solution into the pores. Therefore, only volatile mechanism through the membrane, and the reason for driv-
components of the feed may be transported through the mem- ing force formation [1,7–9]. These differences were taken
brane in the MD process. The different content of the particu- into consideration in the nomenclature by the addition to the
lar components in the gas phase at both ends of the membrane term “membrane distillation” the words, which emphasised a
pores (concentration gradient) causes their transport across feature of a given variant. Various types of MD are known for
the membrane [1,3]. The composition of the gas phase above several years (Fig. 1): direct-contact MD (DCMD), air gap
the liquid surface is often expressed by partial pressure, and MD (AGMD), sweeping gas MD (SGMD) and vacuum MD
the partial pressure difference was therefore accepted as a (VMD). The OMD process, which has been developed dy-
driving force of MD process [1–9]. The value of this driving namically in recent years also, can be included in this group.
force depends on the solution temperature and composition The addition of word “osmotic” to “MD” is consistent
in the layers adjacent to the membrane surface. with historical development of MD process nomenclature.
The definitions of MD process do not consider the reasons This word indicates that OMD is a variant of the membrane
for formation of driving force [1,7–10]. These reasons may distillation the course of which is significantly influenced by
only affect the value of driving force and installation design, the solution concentration. Additionally, from the OMD term
ensue (via analogy to the creation of terms VMD and SGMD)
∗ Fax: +48 91 449 4686.
that a reason for the driving force formation is associated with
E-mail address: margryta@mailbox.tuniv.szczecin.pl. the osmotic pressure (water activity).

0376-7388/$ – see front matter © 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.memsci.2004.07.029
146 M. Gryta / Journal of Membrane Science 246 (2005) 145–156

Fig. 1. Types of MD process: (A) DCMD; (B) DCMD with liquid gap (Gore’s design); (C) AGMD; (D) VMD; (E) SGMD.

Several authors consider the OMD process as distinctive the earlier works [14,27,58–61]. In this variant, the feeding
from membrane distillation. Therefore, in the literature be- solution temperature is low and closed to the temperature
sides the term OMD [1,9,15–20] and isothermal membrane of the solution flowing on the other side of the membrane.
distillation (IMD) [21–23] are used the following terms: os- The vapour pressure difference across the membrane was
motic distillation (OD) [6,24–36], osmotic evaporation (OE) obtained by using a solution with a low water vapour pres-
[26,37–43], gas membrane extraction [44–48] and membrane sure (extraction solution) on the distillate side. In order to
osmotic distillation (MOD) [30,49]. increase the partial pressure difference, some authors pro-
The osmotic membrane distillation is similar to direct con- posed to maintain the feed temperature a few degrees higher
tact MD. Therefore, the equations derived for DCMD variant than the temperature of extraction solution [16,25,27,30].
are successfully applied also for the description of the vapour The extraction solutions comprise the concen-
transport in OMD process [6,22,25,27,37,40,50]. However, trated solutions of salt (NaCl, CaCl2 , MgCl2 , MgSO4 )
several differences between DCMD and OMD variants exist. [6,23,27,28,37–40,51,62] and some organic liquids (glyc-
Indicated difference between these processes resulted from a erol, polyglycols) [29,41,44–48]. A choice of extraction
fact that the driving force in DCMD process is formed by the solution plays a significant role, but information concerning
temperature gradient, whereas, in OMD by the concentration this subject is limited. The different permeate fluxes were
gradient. obtained for different extracting solutions under similar
Identification of the MD process mainly with the operational conditions (equal flow rates, temperature and
DCMD variant [22,24,27,28,30,32,37,40,47,51,52] is the ba- concentration) [41,42]. However, if the dependence of flux
sic source of misunderstanding in the determination of OMD on the Re number and on the difference of the water activity
classification. The authors involved in membrane distillation will be investigated, then it appeared that no difference
investigations often abbreviate DCMD term to MD, which is between the used solutions was found.
consistent with the terminology recommended for this pro- The extraction solution should possess the appropriate
cess [7,8]. However, this abbreviation of the process name properties: high water solubility (low water activity), low
is probably a reason for misconceptions. In this situation, a volatility, low viscosity, and high surface tension [47]. Several
note given by Smolders and Franken [8]: “To avoid complica- salts solutions exhibits these properties, e.g. lithium, sodium,
tion, it is better to use the full term direct-contact membrane calcium and magnesium chlorides, and other salts currently
distillation” seems to be adequate. used as desiccant agents. However, the problems associated
with toxicity should be taken into consideration in the case
of their use in food processing. Glycols and glycerol are re-
2. Fundamentals of OMD process garded as potential organic extractants since they are consid-
ered as non-toxic compounds. Several glycols were success-
The OMD process was patented at the end of the last cen- fully tested for ethanol removal from diluted solutions, such
tury [26,38,53–57], however, its beginning can be found in as fermentation broth [16,45,48]. Nevertheless, only propy-
M. Gryta / Journal of Membrane Science 246 (2005) 145–156 147

lene glycol (PG) can be used in contact with food. Unfortu- the membrane [11,25,28,37]. A value of the chemical poten-
nately, the properties of PG are not so favourable as those tial depends on the temperature, pressure, and the solution
offered by other glycols. A significant concentration of PG composition. Depending on the MD variant, the influence of
in the concentrated juice was found after the OMD process. these parameters on the µ value is different and some of
This fact can be explained by a relatively high volatility of them can be omitted. The difference of the chemical potential
PG, which causes that the PG diffuses in some extent through for ith component in two different states of the system (1 and
the gas membrane. In addition, the penetration pressure of PG 2) is given by [66,67]:
solutions into the membrane pores is only one bar (low sur-
face tension). Due to these properties, the use of PG in OMD api
2
ai2
µi = µ2i − µ1i = RT ln = RT ln (3)
was rejected [44,48]. Glycerol is a good extractant since its api
1 ai1
volatility is practically zero and the surface tension is con-
siderably large. The regeneration of this extractant is easy, and
concentration of glycerol by evaporation is a well known op- api
2
ai2
eration without problems associated with corrosion and scal- = (4)
api
1 ai1
ing which makes the operation with brines difficult [44,48].
The OMD process is most often used to remove where api and ai are the activity of ith component in the
water from liquid foods such as fruit and vegetable vapour and liquid phases, respectively, R is the gas constant
juices, milk, instant coffee and tea, and various non- and T is the temperature. When we assume the state (1) as a
food aqueous solutions being thermally non-resistant state of pure substance (0), then api
1 = a0 and a1 = a0 = 1,
pi i i
[9,21,22,24,26–32,39–44,47,48,51,62]. Low operation tem- and Eq. (4) can be transformed into:
peratures eliminate many problems associated with the con- api
ventional evaporation processes (reaction of non-enzymatic ai = 0 or api = api 0
ai (5)
api
browning—Maillard’s reaction), inhibit the degradation of
flavour and colour, and a loss of volatile aromas (volatility of where api0 is the pressure activity (fugacity) of the saturated
which rapidly increases with feed temperature). vapour of a pure component in case when a liquid is only
at pressure of its saturated vapour (p0i ). Using the pressure
activity coefficient (φ), the Eq. (5) can be rewritten in the
3. Mass transfer form:

The classical models of gas diffusion in porous media, pi φi = p0i φi0 ai (6)
such as molecular or Knudsen diffusion, have been used to
If the pressure of the saturated vapour above both the pure
describe the mass transfer across the membrane in MD pro-
component and the solution is not too high (does not exceed
cess [6,39–41,52,63]. The obtained permeate flux (N) is de-
0.1 MPa), then a deviation of the gaseous phase from the ideal
pendent on the membrane permeability (Lm ) and a value of
state can be neglected, which gives φi = φi0 (Lewis–Randall
the driving force (Π) which causes the mass transfer across
rule) [66], thus we obtain a frequently used equation [25]:
the membrane pores. This relationship can be described by
equation [11,64,65]: pi (x, T ) = p0i (T )ai (x) (7)

N = Lm Π (1) Substituting Eq. (7) to Eq. (2), we obtain the equation for
determination of the driving force in MD:
The driving force in MD is a partial pressure gradient in
the vapour phase given by [1,6,14,22,27,28,50]: Π = p = p0F aF − p0D aD = p0F xF γF − p0D xD γD (8)
where xi is the molar fraction of water and γ i is the water
Π = p = pF − pD (2)
activity coefficient in the solution.
where pF and pD are vapour pressures at the feed/membrane The form of Eq. (8) can be simplified depending on the
and membrane/distillate interface, respectively. Frequently, type of application and MD variant used. In the case of
the driving force is expressed only by the parameters which DCMD, when the obtained distillate is pure water (aD = 1),
in a given MD variant have a large effect on the p value, e.g. and the feed is a diluted solution (γ F = 1), the partial pressure
temperature difference in DCMD [27,45,47] or the activity can be determined using the Raoult law, and:
difference in OMD [42,47]. However, the omission of the Π = p0F xF − p0D (9)
remaining parameters which have a small effect on the p
value causes that the dependence of the permeate flux on the However, one should remember that Eq. (9) can be used in
used driving force Eq. (1) is non-linear [39,47,51]. practice only for diluted solutions.
Similarly, like in other membrane processes, the mass When the feed temperature in DCMD is similar to that of
transfer through the membrane in the MD process is caused the distillate (i.e.p0F ≈ p0D ), then due to a large concentration
by the chemical potentials difference (µ) on both sides of of solutes in the feed (aF < aD thus pF < pD ) we have reversed
148 M. Gryta / Journal of Membrane Science 246 (2005) 145–156

flux, that is, the flow of water vapour from the distillate to instead of p [25,41,42]:
the feed [14,27,58–61]. In order to maintain the flow direc-
tion of vapour from the feed to distillate, the water activity N = K(aFb − aD
b
) (12)
in the distillate should be reduced (under isothermal condi- This relationships is linear in the case when pure water is
tion). In practice, it can be realised by incorporation of the used as a feed and the resistance of the mass transfer on the
concentrated solutions of salt or a liquid strongly absorbing brine side is negligible, and only the membrane resistance
water (glycerol, glycols) into the distillate side. This variant contributes to mass transport [41]. In reality, even for the
is realised as the osmotic membrane distillation. Moreover, water-brine system the value of K changes along with the
the isothermal conditions characteristic for OMD are not un- brine concentration. When a was decreased by 34% then N
usual particularly for MD, since VMD and SGMD also can decreased by 46% [51]. A decrease of water activity in solu-
be operated under the temperature gradient approximately tion is associated with the increase of solute concentration.
equal to zero [68,69]. As a solute concentration increases the density and viscosity
The OMD process is frequently considered as distinctive of the brine also increases, and as a consequence the diffu-
from MD since the driving force causing the removal of water sion coefficient decreases. In this situation, the mass transfer
from feed is a, but not p. However, the driving force resistance in liquid phase was increased, and K coefficient
in OMD is also p, as can be obtained from Eq. (8) for was decreased. When water on the feed side was replaced
assumption that p0F = p0D : by solution, an instability of K coefficient was increased. A
decrease of the activity of sugar solution by 13.2% caused
Π = p = p0 (aF − aD ) (10) a decrease of the efficiency from 10.3 to 1.1 kg/m2 h. This
is caused by an exponential increase of viscosity with in-
Hence, the driving force in OMD depends not only on a, creasing content of sugar in solution, whereas the diffusion
but also on the temperature, the function of which is p0 . coefficient strongly decreases under these conditions [51].
An increase of the permeate flux along with the increase of
membrane temperature (Tm ) is frequently discussed in OMD
[16,24,28], but it is seldom incorporated into the expression 4. Polarization phenomena
of the mass driving force [25,41]. Differences in the form
of Eqs. (9) and (10) result only from the simplification and 4.1. Temperature polarization
mathematical transformations applied in Eq. (8). Moreover,
the situation described by Eq. (10) concerns a particular case. The temperature polarization causes that the temperatures
In real OMD systems, the assumption that p0F = p0D is not in the layers adjacent to the membrane (T1 and T2 ) differ from
true with regard to occurring temperature difference [6,50], those measured in the bulk of the feed (TF ) and distillate/brine
thereby, the value of the driving force of OMD process should (TD ) [4–6]. However, the temperature profiles formed in the
be calculated from Eq. (8). DCMD and OMD are different—Fig. 2. In DCMD, we have
The application of the driving force in the form represented T1 > T2 , whereas in OMD is inversely, T1 < T2 . The temper-
by Eq. (2) is inconvenient because it requires the knowledge ature gradient is obtained due to the evaporation at the feed
of both temperature and solution composition (activity) in side and the condensation at the distillate side, even if the
the membrane/solution interface. The interfacial conditions bulk temperatures of two liquids are equal, like in OMD
are not always accessible, therefore the water transport in the [29,37,50]. Conduction of heat from the brine to feed in-
system is often given by Eq. (11) which refers to the bulk duces a decrease of the polarization effect in OMD. In the
conditions of the liquid streams [6,25,28,38,41]: case of DCMD, the situation is reversed, heat associated with
 −1 the mass transfer as well as conducted through the membrane
1 1 1 flows in the same direction, i.e. from the feed to distillate side,
N= K(pbF − pbD ) = + + pb (11)
KF Lm KD therefore, the temperature polarization effect is increases.
where the global coefficient K is given by the series of wa-
ter transport resistances from the feed to the distillate side,
namely from the bulk of the dilute solution towards the evap-
oration surface (1/KF ), water vapour transport through the
membrane (1/Lm ) and from the condensation surface to the
bulk of the brine (1/KD ). The coefficient K depends on the
membrane morphology, solution concentration, module de-
sign, hydrodynamic conditions and temperature. Therefore,
the obtained values of K are different and depend on the sys-
tem in which they were determined [6,25,27,28,41,42].
The permeate flux is often given by Eq. (12) in which the
a (estimated for the bulk conditions of the liquid) is used Fig. 2. Temperature profiles during DCMD (A) and OMD (B) processes.
M. Gryta / Journal of Membrane Science 246 (2005) 145–156 149

In the case where the bulk temperature is controlled and


maintained at the same level in both liquids (“well stirred”
cell), the transmembrane temperature difference is given by
[6,29,50]:
N H
T = (13)
hm + (1/ hF + 1/ hD )−1
The temperature difference (T = T2 − T1 ) in OMD pro-
cess should be minimized because the vapour pressure gra-
dient induced by this temperature difference is opposite to
the pressure gradient created by the concentration gradient
[6,27,28,37,44,47,50]. Therefore, the occurring T causes Fig. 3. Temperature profiles during OMD process carried out in a large-area
that the driving force for vapour transport decreases. This module: (A) initial temperature profile; (B) final temperature profile.
dependence in DCMD is reverse, because the permeate flux
increases for higher T. Mengual et al. [28] estimated the and the driving force was obtained in the form similar to Eq.
temperature difference in OMD system on the basis of the (15):
equation: L
p = p0 (Tm ) a + p0 (Tm )am T (19)
N H = hA T (14) RTm2
Gostoli proposed to use the coefficient Θ which represents
where A is membrane area. They obtained the value of T the fraction of the driving force which is effective for the mass
within the range from 0.5 to 0.8 K (on each membrane side) transport through the membrane [6,29,50]:
for h = 1900 W/m2 K (hD = hF ) and N = 5–8 × 10−8 kg/m2 s.
Therefore, they concluded that such values would lead to N = Lm p = ΘLm pb (20)
negligible decrease of the vapour flux. On the contrary, Vah-  
dati and Priestman [52] have found that a small temperature MW L2 p0 (Tm )am Lm
Θ= 1+ (21)
difference can significantly affects the driving force induced RTm 2
hm + (1/ hF + 1/ hD )−1
by even very high concentration gradients. The thermal effect
due to evaporation and condensation at both liquid-membrane The value Θ = 0.85 was obtained for TF200 membrane
interface increases with increasing permeate flux. A high used to evaporate pure water at 25 ◦ C with NaCl solution in
vapour flux of 12 kg/m2 h causes the transmembrane tem- a stirred cell module. In the case of co-current flow system,
perature difference of approximately 2◦ C, which induces a with the same liquids, a 31% decrease of flux was obtained
30% reduction of the driving force [6]. as a result of the temperature difference formed across the
The temperature effect on the driving force can be also membrane [29].
estimated by the Clausius-Clapeyron equation [50]: The temperature profiles presented in Fig. 2B may be ob-
tained only in a stirred cell of small laboratory apparatus.
p0 (Tm )ML However, in the OMD pilot plant modules the existing tem-
p = pb − am T (15) perature gradients causes the heat transfer between the mem-
RTm2
brane surfaces and the bulk of solutions. As a consequence,
where Tm is the membrane temperature, L is the heat of water the feed temperature decreases and the brine temperature in-
vaporisation and pb is the water vapour pressure difference creases, what is presented in Fig. 3A. The formation of the
calculated under the bulk conditions, i.e. at the temperature temperature difference (Tb = TD – TF ) causes a further de-
TF = TD , and am is the average water activity: (aF + aD )/2. crease of the driving force formed by the concentration dif-
The dependence of the vapour pressure on temperature ference. In the extreme case, the mass transfer can be even
can be given in the form [25,28,70]: stopped. For this case, considering that T is proportional
  to Tb , it is possible to find the relationship between the ac-
L tivity difference corresponding to an arbitrarily chosen con-
p (T ) ∝ exp −
0
(16)
RT centration difference and its “equivalent” bulk temperature
difference, given by the following expression [25]:
For the symmetric system, the authors [25,70] assumed
that: RTm a
Tb ∝ (22)
L am
T
T1 = T2 = Tm − (17)
2 The interesting results was fact, that a bulk tempera-
ture difference (Tb ) equals to 5 K equivalent the driv-
T LT
1 and 1 (18) ing force of OMD created by a concentration differ-
2Tm 2RTm2 ence about 3 M NaCl/dm3 [37]. For the system: distilled
150 M. Gryta / Journal of Membrane Science 246 (2005) 145–156

water–membrane–5 M NaCl/dm3 (stirred cells) at 40 ◦ C the


OMD flux was stopped when water was cooled to 34 ◦ C [25].
The temperature profile also changes along the OMD mod-
ule [47,50]. Near the entry section, the profile is similar to that
represented in Fig. 2B, obtained for symmetric case (equal
inlet temperatures and heat capacities). As a result, the heat
transfer between the membrane surfaces and the bulk of so-
lutions the temperature gradient presented in Fig. 3A is cre-
ated. Moreover, in this system the temperature difference T
across the membrane increases. An asymptotic value T∞
is finally reached for which the convective heat flux (QV )
through the membrane (mass transfer) is exactly balanced by
the conductive heat back-flux (QC ) [29,47,50]:
Fig. 4. Concentration polarization profile in OMD process.
N H
T∞ = (23)
hm Some authors [52] have assumed that the brine concentra-
tion in the layer adjacent to the membrane (c2 ) can be reduced
The temperature profile presented in Fig. 3B will be ob- to zero when temperature differences are applied across the
tained if the OMD system works without the heat losses. membrane (TF > TD —such as in DCMD) in order to increase
the OMD flux [30]. Along with the increase of permeate flux,
4.2. Concentration polarization the diffusion resistance in the salt solution was much higher
than the heat resistance. Therefore, a high flux established due
In OMD process (similar as in other MD variants), only to the temperature difference results in the complete concen-
water passes through the membrane, therefore the solute con- tration polarization, i.e. the vapour pressure difference due
centration near the membrane on the feed side is larger than to the concentration gradient was equal to zero, with almost
that in the bulk, whereas the solute concentration on the ex- pure water on both membrane surfaces. This result was con-
tract side is lower than that in the bulk. This phenomenon is firmed by repeating the experiment with pure water on both
termed as the concentration polarization and is resulting in a sides, i.e. the same flux was achieved as with the salt system
driving force reduction across the membrane [1,6,52]. [52].
The concentration polarization in DCMD process is usu- In accordance with the film theory, the salt concentrations
ally considered in the context of a solute build-up on the in the bulk phase (cD ), and at the solution–membrane inter-
feed side, with the distillate being pure liquid, as in typi- face (c2 ) are related to the volume flux by equation [6,28]:
cal desalination applications [27]. This effect has a smaller  
JV
influence on the performance of DCMD. Therefore, several cD = c2 exp (25)
kM
authors assume that the temperature polarization is essential
in the case of DCMD, whereas the concentration polarization However, when the flux in OMD process is much lower
in the case of OMD [6]. However, the polarization effect in than the mass transfer coefficient (kM ) we can assume that
DCMD variant becomes important in the case of concentra- cD and c2 have similar values. In this case, the flux is given
tion of solutions with solute content close to the saturated by [28,37]:
state [71]. Lm kM
The concentration polarization in the OMD process is im- N= pb (26)
portant, particularly on the side of extracting solution. The kM + [dp/dc]b Lm cD
presence of the layers of concentrated feed and diluted brine where kM is the mass transfer coefficient for solute in the
along the surface of the membrane can significantly decrease layer adjacent to the membrane.
the difference of water activity (Fig. 4). In this situation, the Gostoli described the relationship between the flux and
driving force of the process decreases in accordance with Eq. solute concentration on the feed and brine side by equation
(8): [50]:

Π = p1 − p2 = p0F a1 − p0D a2 x1 aF − a 1
(24) N = kM
F
ρF ln ≈ kM
F
ρF (27)
xF γF xF,ln
A significant concentration polarization effect was ob- xD a2 − a D
served during OMD with glycol [47]. As a result, the per- N = kM
D
ρD ln ≈ kM
D
ρD (28)
x2 γD xD,ln
meate flux varies in the non-linear way with the driving force
evaluated under the conditions prevailing in the bulk phase. The value of kM coefficient depends on the physical prop-
The influence of concentration polarization increases with erties of the solution, as well as on the hydrodynamic prop-
the increase of the permeate flux. erties of the system. This coefficient can be obtained either
M. Gryta / Journal of Membrane Science 246 (2005) 145–156 151

Table 1
Characteristics of membranes used in OMD process
Materials Name Manufacturer Configuration/Din s (␮m) ε (%) dp (␮m) dmax (␮m) Literature
(␮m)
PTFE FHLP Millipore Flat 175 70–80 0.25 25, 28
PP Accurel PP Q3/2 Enka A.G. Capillary/1000 200 70 0.2 – 47, 50
Accurel PP S6/2 Enka A.G. Capillary/2600 400 70 0.2 0.6 43, 47, 50
Celgard 2500 Hoechst Celanese Co. Flat 28 45 0.05 0.07 23, 42
Metricel Gelman USA Flat 90 55 0.1 42
PVDF Durapore HVHP Millipore Co. Flat 125 75 0.45 41
Durapore GVSP Millipore Co. Flat 108 80 0.22 23
Durapore GVHP Millipore Co. Flat 125 70 0.20 28, 41
125 75 0.22
PTFE supported by Gore-Tex 10387 Gore & Associates Flat 8.5 78 0.2 – 31, 39
PP net TF450 Pall-Gelman Flat 30–70 80 0.45 – 28, 33, 41, 51
178* 60*
TF200 30–70 80 0.2
165–178* 60*
TF200 Gelman Instruments Co. Flat 60* , 165* 60 0.2 – 38, 45, 47
TF1000 Gelman Sciences Flat 178* 80 1 – 28
* Polypropylene net with di > 10 ␮m [6]. Din : inner diameter.

experimentally or estimated with the use of empirical corre-


lations [6].

5. Membranes for OMD

A fundamental condition required to carry out the OMD


process is the maintenance of the gas phase in the mem-
brane pores. The application in this process of the mem-
branes manufactured from polymers such as polypropylene
(PP), polytetrafluoroethylene (PTFE) and poly(vinylidene
fluoride) (PVDF), with a pore diameter ranging from 0.1 to
1 ␮m in this process fulfils the conditions of being non-wetted
by various aqueous solutions in the OMD process. The prop-
erties of membranes used in OMD studies are summarised in Fig. 5. The influence of the thickness of used membrane on the permeate
Table 1. flux during OMD and DCMD processes.
The MD process is carried out with the membranes having
a thickness in the range from 8 ␮m to about 500 ␮m. Since the strength of the membrane. Therefore, the thin membranes are
permeate flux in OMD process is proportional to the recip- supported by net, e.g. Gelman PTFE membranes (Table 1).
rocal of the pore length, the membranes should be as thin as Moreover, one should not consider that thick membranes
possible. Moreover, the heat resistance of the membrane de- are recommended for DCMD. In fact, the heat resistance
creases with decreasing thickness. This is advantage of OMD increases along with the increase of membrane thickness,
because the membranes used in this process must exhibit the which causes an increase of both the temperature and vapour
heat conductive as high as possible [27], which results from pressure differences. However, similarly to OMD the perme-
Eq. (23). The membranes with porosity of 60–80% are used ate flux in the DCMD increases with decreasing membrane
in this process. Taking into consideration a high temperature thickness [72]. It was found that along with an increase of
resistance of gas (mixture of air and water vapour) as well as the membrane thickness the Lm coefficient decreased faster
used polymers, a decrease of membrane thermal resistance than the Π increases and consequently the permeate flux
can be only achieved by a reduction of the membrane thick- decreases. The effect of the membrane thickness on the per-
ness. A decrease of membrane thermal resistance facilitates meate flux in the both processes is shown in Fig. 5.
the heat transfer from the brine to feed, which causes a reduc- One of the limitations of OMD process is the possibility of
tion of T and the increase of Π (for the case TF ≈ TD ). wetting of the hydrophobic membrane resulting in a decline
This dependence is opposite to that in DCMD where a high of both the permeate flux and separation performance. Dur-
membrane heat resistance helps to maintain the temperature ing the MD operation, the membrane pores become gradually
gradient. The thickness is usually limited by the mechanical filled with liquid (water logging [73]). The application of thin
152 M. Gryta / Journal of Membrane Science 246 (2005) 145–156

membranes will facilitate their wetting, which would require its operating conditions. This effect is probably a reason for
the application of the regeneration procedure (cleaning and large discrepancies in the results published in the literature
drying) more frequently. However, the membrane regenera- [6,37,38]. The major application of OMD process includes
tion in the OMD process [31] and DCMD process [71] is dif- the concentration of fruit juices and other temperature non-
ficult, and the obtained results are not promising. Generally, it resistance liquids. Process allows the production of concen-
is recommended to avoid the fouling and membrane wetting. trated fruit juices with quality and composition very close to
In the case of hydrophobic membranes, this can be achieved fresh one, without added flavours and with a high vitamin
by the modification of the membrane surfaces through their content at a cost comparable with the conventional product
coating by a thin layer of a hydrophilic polymer [23,73]. obtained by evaporation [44]. The limiting concentration of
The membrane coating is necessary for concentration of oily juices in OMD process is associated with very high viscos-
feeds, because the uncoated membranes were promptly wet- ity of juices (>0.2 N/m) at concentrations of sucrose exceed-
ted even for low concentrations of oil (limonene) dispersed ing 68◦ Brix [21,23,26,42]. For the feed concentration below
in water [23]. The coated membranes were found to maintain 40 wt% of sugar, the permeate flux varies in accordance with
their properties during the concentration of oil emulsions for the membrane permeability, which indicates that the process
periods up to 24 h. These observations confirm the efficiency is controlled by the membrane resistance. At higher concen-
of coated membranes for OMD of oily feeds. tration (50–68 wt%), the mass transfer resistance on the feed
Flat membranes, such as Gelman series TFxxx (Table 1), side is dominant, and there is no advantage in using a more
are often used in OMD process. The permeate flux obtained permeable membrane [23].
for TF450 is slightly lower than that for TF200 [28], although Different configurations of modules were tested
it should be reversed by taking into consideration the pore di- for OMD applications. The membranes may be flat
ameter. However, these membranes were developed for mi- [6,23,24,27–29,38–42,44–48] or in a form of hollow fibers
crofiltration (MF) applications and the manufactured data (re- [21,22,26,27,30,43,44,47,50,62]. Considering a high vis-
jection and hydraulic permeability) are characteristic for MF. cosity of the liquids flowing through a OMD module the
Measurements of the hydraulic permeability are not relevant application of plate-and-frame [24,29,38,44] or spiral-wound
for membranes used in MD, because they involve other mass [27,37] configurations is recommended. Moreover, the plate
transfer mechanisms such as a viscous flow instead of the dif- and frame module (also spiral-wound) have membrane
fusion transport of gas. For the TF200 membrane, a greater with lower thickness, thus the obtained flux is larger in
value of Lm was obtained than that for TF450 membrane as comparison to capillary modules [39,44,47,48].
reveals the analysis of the mass transfer in DCMD on the ba- The performance of OMD process was also evaluated on
sis of the pore diameters determined by the liquid expulsion a commercial scale with modules having the area above 1 m2
permoporometry method [74]. [22,29,31,36,37,43,48,75]. In the pilot plant or industrial in-
The role of TF200 and TF450 in OMD process was dis- stallation, the two solutions are pumped along the opposite
cussed in several work [6,63,75]. The TFxxx membranes sides of the membrane, e.g. in a counter-current flow. How-
have a heterogeneous structure: the thin PTFE layer is sup- ever, the hydrodynamic conditions are slightly different from
ported by PP net. The pore diameters determined in these those created by simple agitation often used at laboratory
works for a PTFE layer of both TF200 and TF450 were scale [25,28,37,39,41,45]. Under unfavourably conditions of
similar. An observation performed after OMD experiments fluid mixing, the OMD process can be strongly limited by the
demonstrated that PP layer was wetted by the aqueous solu- concentration polarisation [6,40,48,51,62]. A relatively high
tions. Thus, only the Teflon layer constitutes the resistance flux (8–12 dm3 /m2 h) was obtained with the new plate-and-
for water vapour transport through the membrane, while the frame lab-scale OMD module [6,38,51]. In this design, the
PP net contributes to the mass transfer resistance in the liquid fluids were circulated into square channels evenly distributed
phase. The mass transfer resistance in the vapour phase com- so that the velocity of the fluids is identical throughout entire
prises 40–70% of the total resistance. Under favourable hy- membrane surface. The most advantageous hydrodynamic
drodynamic conditions in the circulating brine, the resistance conditions were obtained for Re > 1000.
of liquid film entrapped in PP support net represents up to Several kinds of fruit juice (orange, apple, and grape)
30% of the total resistance. On the contrary, the contribution were subjected to the concentration using a batch mode
of the support layer becomes negligible under unfavourable operation with Syrinx plate-and-frame configuration SR-72
hydrodynamic conditions and 56% of the total mass transfer [24]. A plate-and-frame module with working area of 1.5 m2
resistance comes from the diluted brine boundary layer at the equipped with ten PTFE flat sheet membranes was used for
condensation interface of the membrane [6]. concentration of orange juice above 60◦ Brix [76]. In other
work, the grape juice was concentrated to above 68 Brix [62].
In this case, the OMD experiments were carried out with the
6. OMD modules and process applications use of a Hoechst-Celanese Liqui-Cell hollow fibre contac-
tor containing macroporous, hydrophobic Celgard membrane
The magnitude of the permeate flux obtained in the OMD (membrane area 1 m2 ). The same membrane area but in the
process is significantly affected by the module design and plate-and-frame configuration was used for concentration of
M. Gryta / Journal of Membrane Science 246 (2005) 145–156 153

orange, apple and grape juices [75]. The fluxes vary from 7. Summary
0.02 to 2.8 dm3 /m2 h depending on the nature and obtained
concentration of the fruit juices. Johnson et al., reports that Several significant operational differences exist among
the flux rates ranging from 5 to 10 dm3 /m2 h for the instal- particular variants of membrane distillation, however the ba-
lation with spiral wound membranes and juice concentration sic principle of MD (evaporation of feed through the pores
with the solid content exceeding up to 75%. In their instal- of non-wetted membrane) remains the same, also in OMD.
lation, the diluted salt solution was separately regenerated Therefore, certain differences between DCMD and OMD
by concentration in pervaporation installation equipped with cannot be a reason of a statement that the OMD process is
hollow fiber membranes [37]. not a type of membrane distillation.
The concentration of whole juices creates several prob- In OMD process, the membrane separates the solutions
lems. For this type of juices, a pilot plant with plate-and- with the same temperatures, hence, the vapour pressure gra-
frame modules having membrane area of 25 m2 was devel- dient results from a high solute concentration in the extracting
oped [44,48]. The modules were especially designed to han- solution flowing on the distillate side. However, the heat is
dle the whole juices with a high content of pulp. A net shaped transferred from feed to the membrane and from membrane
spacer was used on the extract side, while the juice side path to brine, and as a results the feed cools down and the brine
was smooth. Although this configuration is not optimal con- warms up. Therefore, the temperature difference across the
sidering the mass transfer efficiency, it allows processing of membrane existed, and the mass transfer in OMD is also as-
non clarified juice. sociated with the heat transfer.
Helically-wound hollow fibre modules offer a significant Taking into consideration a small value of the driving
improvement in the hydrodynamic conditions on the shell- force, the OMD process is strongly affected by the concen-
side in comparison with the axial flow modules. In the case tration polarization, occurring particularly on the brine side.
of processing of viscous feeds in OMD process with the use Therefore, the module design, which provides appropriate
of helical module, the higher concentrations of solutes and hydrodynamic conditions, is very important.
two times larger flux were obtained than those obtained for The results obtained from the OMD process evaluation
module design with axial flow [21,22]. demonstrate several advantages, which promote its industrial
The ultrafiltration (UF) is proposed as a pretreatment pro- application. However, the implementation of OMD process
cess for juice concentration by DCMD and OMD processes would require the performance of the long-term tests. Based
[25,62,77]. In the case of OMD, the performance of this pro- on these tests, it should be possible to determine whether
cess was affected by pore diameters of the membranes used the most important disadvantages of DCMD (e.g. fouling
in the UF process [62]. The UF process of single strength and wetting of membranes) also concern the OMD variant.
Gordo grape juice uses membrane with a nominal pore di- This knowledge should be acquired within the next years,
ameter of 0.5 ␮m did not result in the increase of the OMD taking into consideration a fact that the OMD experiment are
flux. An increase of the flux during subsequent concentra- currently carried out in the several pilot plants.
tion of permeate by OMD was observed for UF membranes
with nominal pore diameters of 0.1 ␮m or less. It has been
also demonstrated that UF results in a small increase in juice Nomenclature
surface tension, which in turn reduces the tendency for mem-
brane wetting. Symbols
During the concentration of tomato puree, the adhesion of A membrane area
fatty substances including tomato pigments to the membrane ai activity of i component in the liquid phase
surface was indicated, and as a consequence the decline of api activity of i component in the vapour phase
permeate flux was observed [31]. The fouling comprises a (fugacity)
major limitation in the industrial implementation of process, c solute concentration (kg/m3 )
thus the determination of OMD performance during long- dp pore diameter (m)
term run is necessary. However, this subject has received little H vapour enthalpy (J/kg)
attention in papers published so far. h convective heat transfer coefficient (W/m2 K)
A combination of DCMD with OMD for concentration hm thermal conductivity coefficient of the mem-
of fluosilicic acid was studied [17]. In this case, saturated brane (W/m2 K)
NaCl brine was used as a cold solution. The presence of brine JV volume permeate flux (m3 /m2 s)
not only increases the vapour pressure difference, but also K overall (apparent) mass transport coefficient
creates the possibilities of sodium fluosilicate production. A kM mass transfer coefficient (m/s)
part of the volatile fluoride compounds transported through L latent heat (J/kg)
the membrane would react with NaCl and the precipitation of Lm membrane permeability coefficient (kg/m2 s)
Na2 SiF6 occurred. However, the precipitation of this salt also M molecular weight (kg/kmol)
proceeded on the membrane surface, therefore, the permeate
flux decreased as a result of scaling.
154 M. Gryta / Journal of Membrane Science 246 (2005) 145–156

[10] W.J. Koros, Y.H. Ma, T. Shimidzu, Terminology for membranes and
p partial pressure of saturated vapour over the membrane processes (IUPAC Recommendation 1996), J. Membr.
solution (N/m2 ) Sci. 120 (1996) 149–159.
p0 partial pressure of saturated vapour over the [11] J.G.A. Bitter, Transport Mechanisms in Membrane Separation Pro-
pure solvent (N/m2 ) cesses, Plenum Press, New York, 1991.
[12] G. Vasaru, G. Müller, G. Reinhold, T. Fodor, The Thermal Diffusion
Q heat (W)
Column, The Deutscher Verlag Der Wissenschaften, Berlin, 1969.
R gas constant (J/kmol K) [13] G.L. Hassler, Means and method for mass and heat transfer, United
Re Reynolds number States Patent Office, No. 3129145 (1964).
s membrane thickness (m) [14] M.T. Bryk, R.R. Nigmatullin, Membrane distillation, Russ. Chem.
T temperature (K) Rev. 63 (1994) 1047–1062.
[15] M.P. Godino, L. Peña, C. Rincón, J.I. Mengual, Water produc-
x mole fraction in liquid phase
tion from brines by membrane distillation, Desalination 108 (1996)
91–97.
Greek letters [16] M.P. Godino, L. Peña, J.I. Mengual, Evaluation of the membrane
Π driving force distillation coefficient, in: Proceedings of Euromembrane’97, June,

˜ temperature polarization coefficient in OMD University of Twente, 1997, p. 346.
[17] M. Tomaszewska, Concentration and purification of fluosilicic
ε membrane porosity
acid by membrane distillation, Ind. Eng. Chem. Res. 39 (2000)
φ pressure activity coefficient 3038–3041.
γ activity coefficient [18] J.I. Mengual, L. Peña, Membrane distillation, Colloid Interface Sci.
µ chemical potential difference 1 (1997) 17–29.
ρ density (kg/m3 ) [19] M.P. Godino, L. Peña, J.I. Mengual, Membrane distillation: theory
and experiments, J. Membr. Sci. 121 (1996) 83–93.
[20] L. Peña, M.P. Godino, C. Rincón, J.I. Mengual, A method to evaluate
Subscripts the net membrane distillation coefficient, J. Membr. Sci. 143 (1998)
A air 219–233.
b bulk [21] A.J. Costello, P.A. Hogan, A.G. Fane, Performance of helically-
C conducted wound hollow fibre modules and their application to isothermal
membrane distillation, in: Euromembrane’97, 23–27 June, Univer-
D distillate/brine (extraction solution)
sity of Twente, The Netherlands, 1997, pp. 403–405 (book of
F feed abstracts).
m membrane [22] A.G. Fane, M. Costello, P.A. Hogan, R.W. Schofield, Mem-
V vapour brane distillation and osmotic (isothermal membrane) distillation:
0 pure component factor for enhanced performance, in: Proceedings of the Work-
shop on “Membrane Distillation, Osmotic Distillation and Mem-
1 boundary layer on the feed side
brane Contactors”, CNR-IRMERC, 2–4 July, Cetraro, Italy, 1998,
2 boundary layer on the distillate side pp. 1–6.
[23] J. Mansouri, A.G. Fane, Membrane development for processing of
oily feeds in IMD, osmotic distillation: developments in technol-
ogy and modelling, in: Proceedings of the Workshop on “Membrane
Distillation, Osmotic Distillation and Membrane Contactors”, CNR-
References IRMERC, 2–4 July, Cetraro, Italy, 1998, pp. 43–46.
[24] J. Sheng, R.A. Johhnssson, M.S. Lefebvre, Mass and heat trans-
[1] K.W. Lawson, D.R. Lloyd, Membrane distillation, J. Membr. Sci. fer mechanisms in the osmotic distillation process, Desalination 80
124 (1997) 1–25. (1991) 113–121.
[2] A.C.M. Franken, J.A.M. Nolten, M.H.V. Mulder, D. Bargeman, C.A. [25] M.P. Godino, L. Peña, J.M. Ortiz de Zárate, J.I. Mengual, Coupled
Smolders, Wetting criteria for the applicability of membrane distil- phenomena membrane distillation and osmotic distillation through
lation, J. Membr. Sci. 33 (1987) 285–298. a porous hydrophobic membrane, Sep. Sci. Technol. 30 (1995)
[3] R. Taylor, R. Krishna, Multicomponent Mass Transfer, John Wiley, 993–1011.
New York, 1993. [26] M. Courel, M. Dornier, G.M. Rios, M. Reynes, P. Deblay, Osmotic
[4] R.W. Schofield, A.G. Fane, C.J.D. Fell, R. Macoun, Factor af- evaporation: a new technique for fruit juice concentration, in: Eu-
fecting flux in membrane distillation, Desalination 77 (1990) 279– romembrane’97, 23–27 June, University of Twente, The Netherlands,
294. 1997, p. 432 (book of abstracts).
[5] M. Gryta, M. Tomaszewska, Heat transport in the membrane distil- [27] M.M. Vahdati, G.H. Priestman, Developments in membrane and
lation process, J. Membr. Sci. 144 (1998) 211–222. osmotic distillation, Meeting “New Direction in Distillation and
[6] M. Courel, M. Dornier, G.M. Rios, M. Reynes, Modelling of wa- Absorption”, The Institution of Chemical Engineers—Fluid Sep-
ter transport in osmotic distillation using asymmetric membrane, J. aration Process Group, 3 November, Sheffield University, UK,
Membr. Sci. 173 (2000) 107–122. 1993.
[7] A.C.M. Franken, S. Ripperger, Terminology for Membrane Distilla- [28] J.I. Mengual, J.M. Ortiz de Zárate, L. Peña, A. Velázquez, Osmotic
tion, European Society of Membrane Science and Technology, 1988. distillation through porous hydrophobic membranes, J. Membr. Sci.
[8] K. Smolders, A.C.M. Franken, Terminology for membrane distilla- 82 (1993) 129–140.
tion, Desalination 72 (1989) 249–262. [29] A. Cervellati, G. Zardi, C. Gostoli, Osmotic distillation: develop-
[9] G. Jonsson, A.I. Gill, Mass transfer limitations in membrane con- ments in technology and modelling, in: Proceedings of the Work-
tactors and various forms of membrane distillation, in: Proceed- shop on “Membrane Distillation, Osmotic Distillation and Mem-
ings of “Engineering with Membranes”, 3–6 June, Granada, Spain, brane Contactors”, CNR-IRMERC, 2–4 July, Cetraro, Italy, 1998,
2001. pp. 39–42.
M. Gryta / Journal of Membrane Science 246 (2005) 145–156 155

[30] Z. Wang, F. Zheng, S. Wang, Experimental study of membrane dis- [50] C. Gostoli, Thermal effects in osmotic distillation, J. Membr. Sci.
tillation with brine circulated in the cold side, J. Membr. Sci. 183 163 (1999) 75–91.
(2001) 171–179. [51] M. Courel, M. Dornier, J.M. Herry, G.M. Rios, M. Reynes, Effect of
[31] R.J. Durham, M.H. Nguyen, Hydrophobic membrane evaluation and operating conditions on water transport during the concentration of
cleaning for osmotic distillation of tomato puree, J. Membr. Sci. 87 sucrose solutions by osmotic distillation, J. Membr. Sci. 170 (2000)
(1994) 181–189. 281–289.
[32] B.L. Jiao, R. Molinari, V. Calabrò, E. Drioli, Application of mem- [52] M.M. Vahdati, G.H. Priestman, Reducing boundary layer effects in
brane operations in concentrated citrus juice processing, Agro-Ind. membrane osmotic distillation, ICHEME Research Event, London,
Hi-Tech. (1991) 19–27. 1994, pp. 177–179.
[33] M. Courel, E. Tronel-Peyroz, G.M. Rios, M. Dornier, M. Reynes, [53] M.S.M. Lefebvre, Method of performing osmotic distillation, Patent
The problem of membrane characterization for the process of os- No. U.S. 4781837 (1988).
motic distillation, Desalination 140 (2001) 15–25. [54] P. Deblay, Process for at least partial dehydration of an aqueous
[34] A.M. Barbe, J.P. Bartley, A.L. Jacobs, R.A. Johnson, Retention of composition and devices for implementing the process, FR Patent
volatile organic flavour/fragrance components in the concentration No 91/13013 (1991).
of liquid foods by osmotic distillation, J. Membr. Sci. 145 (1998) [55] COGIA Société Anonyme, Procédé de déshydratation au moins par-
67–75. tielle d’une composition aqueuse et dispositifs pour mettre en oeuvre
[35] J. Mansouri, A.G. Fane, Osmotic distillation of oily feeds, J. Membr. le procédé, Eur. Patent No. 0539280 A1930428 (1993).
Sci. 153 (1999) 103–120. [56] E.G. Beaudry, J.R. Herron, C.E. Jochums, L.E. Medina (OSMO-
[36] F. Ali, M. Dornier, A. Duquenoy, M. Reynes, Evaluating transfer of Osmotec Inc.), Apparatus for direct osmosis concentration of fruit
aroma compounds during the concentration of sucrose solutions by juices, U.S. Patent No. 5281430 (1994).
osmotic distillation in batch-type pilot plant, J. Food. Eng. 60 (2003) [57] C. Gostoli, A. Cervellati, Procedimento di concentrazione di succhi
1–8. di frutta ed altre miscele acquose mediante estrazione a membrana
[37] W. Kunz, A. Benhabiles, R. Ben-Aim, Osmotic evaporation through gassosa e relativo dispositivo (Process for concentrating fruit juices
macroporous hydrophobic membranes: a survey of current research and other aqueous mixtures by gas membrane extraction and related
and applications, J. Membr. Sci. 121 (1996) 25–36. device), Ital. Patent No. RM97A000678 (1997).
[38] M. Courel, G.M. Rios, M. Reynes, Influence of hydrodynamics [58] P.K. Weyl, Recovery of demineralized water from saline waters,
on osmotic evaporation performance, in: Proceedings of the Work- United States Patent Office, No. 3340186 (1967).
shop on “Membrane Distillation, Osmotic Distillation and Mem- [59] Z. Honda, H. Komada, K. Okamoto, M. Kai, Nonisothermal mass
brane Contactors”, CNR-IRMERC, 2–4 July, Cetraro, Italy, 1998, transport of organic aqueous solution in hydrophobic porous mem-
pp. 35–38. brane, Proc. Eur.–Jpn. Congr. Membr. Membr. Process. (1984)
[39] A. Benhabiles, R. Ben Aim, Osmotic evaporation: a new membrane 587–594.
process for concentration of thermosensitive aqueous solutions, in: [60] G.C. Sarti, C. Gostoli, Use of hydrophobic membranes in thermal
Proceedings of the Workshop on “Membrane Distillation, Osmotic separation of liquid mixtures: theory and experiments, in: E. Drioli,
Distillation and Membrane Contactors”, CNR-IRMERC, 2–4 July, M. Nakagaki (Eds.), Membranes and Membrane Processes, Plenum
Cetraro, Italy, 1998, pp. 81–84. Publ. Corp., New York, 1986, pp. 349–360.
[40] J. Romero, G.M. Rios, J. Sanchez, A. Saavedra, Modelling and [61] E. Drioli, V. Calabrò, Y. Wu, Microporous membranes in membrane
simulation of osmotic evaporation process, in: Proceedings of “En- distillation, Pure Appl. Chem. 58 (1986) 1657–1662.
gineering with Membranes”, 3–6 June, Granada, Spain, 2001, [62] A.F.G. Bailey, A.M. Barbe, P.A. Hogan, R.A. Johnson, J. Sheng,
pp. 297–301. The effect of ultrafiltration on the subsequent concentration of
[41] V.D. Alves, I.M. Coeihoso, Mass transfer in osmotic evaporation: grape juice by osmotic distillation, J. Membr. Sci. 164 (2000) 195–
effect of process parameters, J. Membr. Sci. 208 (2002) 171–179. 204.
[42] V. Alves, I. Coelhoso, Osmotic evaporation: an alternative process to [63] C. Gostoli, S. Bandini, S. Di Francesca, G. Zardi, Analysis of a
concentrate fruit juice, in: A. Noworyta, A. Trusek-Hołownia (Eds.), reverse osmosis process for concentrating solutions of high osmotic
Using Membranes to Assist of Cleaner Processes, ARGI, Wrocław, pressure, Trans. IChemE 74 (C) (1996) 101–109.
2001, pp. 243–248 (Poland). [64] M.C. Porter (Ed.), Handbook of Industrial Membrane Technology,
[43] F. Vaillant, E. Jeanton, M. Dornier, G.M. O’Brien, M. Rey nes, Noyes Publication, New Jersey, U.S., 1990.
M. Decloux, Concentration of passion fruit juice on an industrial [65] M. Cheryan, Ultrafiltration and Microfiltration Handbook, Tech-
pilot scale using osmotic evaporation, J. Food Eng. 47 (2001) 195– nomic Publishing Company, Lancaster, U.S., 1998.
202. [66] J. Szarawara, Termodynamika Chemiczna Stosowana, WNT,
[44] A. Cervellati, G. Zardi, C. Gostoli, Gas membrane extraction: a new Warszawa, 1997.
technique for the production of high quality juices, Fruit Process. 10 [67] P. Atkins, Physical Chemistry, Oxford University Press, Oxford,
(1998) 417–421. 1998.
[45] C. Gostoli, S. Bandini, Gas membrane extraction of ethanol by [68] H. Mahmud, A. Kumar, R.M. Narbaitz, T. Matsuura, A study of
glycols: experimental and modelling, J. Membr. Sci. 98 (1995) mass transfer in the membrane air-stripping process using microp-
1–12. orous polypropylene hollow fibers, J. Membr. Sci. 179 (2000) 29–
[46] S. Bandini, C. Gostoli, Ethanol removal from fermentation broth by 41.
gas membrane extraction, J. Membr. Sci. 70 (1992) 119–127. [69] G. Rajalo, O. Tereping, T. Petrovskaya, Thermal forced membrane
[47] S. Bandini, C. Gostoli, Concentrating aqueous solutions by gas desorption–absorption of ammonia, J. Membr. Sci. 89 (1994) 93–99.
membrane extraction, in: Proceedings of “EUROMEMBRANE’95”, [70] P. Godino, L. Peña, J.I. Mengual, Membrane distillation: theory and
18–20 September, University of Bath, UK, 1995, pp. II 19–II 24. experiments, J. Membr. Sci. 121 (1996) 83–93.
[48] M. Bregoli, A. Cervelati, R. Ferrarini, C. Leoni, A. Zani, G. Zardi, [71] M. Gryta, Concentration of NaCl solution by membrane distilla-
C. Gostoli, Development and modeling of a plate & frame mem- tion integrated with crystallization, Sep. Sci. Technol. 37 (2002)
brane contactor for juice and wine processing, in: Proceedings of 3535–3558.
“Engineering with Membranes”, 3–6 June, Granada, Spain, 2001, [72] K. Schneider, W. Hölz, R. Wollbeck, Membranes and modules for
pp. 259–263. transmembrane distillation, J. Membr. Sci. 39 (1988) 25–42.
[49] Z. Wang, F. Zheng, Y. Wu, S. Wang, Membrane osmotic distillation [73] D.Y. Cheng, S.J. Wiersma, Composite membrane for membrane dis-
and its mathematical simulation, Desalination 139 (2001) 423–428. tillation system, USA Patent 4419242 (1983).
156 M. Gryta / Journal of Membrane Science 246 (2005) 145–156

[74] L. Martinez, F.J. Florido-Diaz, P. Pradanos, A. Hernandez, [76] M.M. Vahdati, G.H. Priestman, Enhancement of the performance
in: Proceedings of the 41st Microsymposium “Polymer Mem- characteristic of membrane in osmotic distillation., in: Proceedings
branes”, 16–19 July, Prague, Characterization of hydrophobic of the 2nd Conference on Advances in Biochemical Engineering,
microporous membranes used in membrane distillation (2001) ICHEME, 1994, pp. 94–96.
41. [77] V. Calabro, B.L. Jiao, E. Drioli, Theoretical and experimental study
[75] J. Sheng, Osmotic distillation technology and its applications, Aus- on membrane distillation in the concentration of orange juice, Ind.
tral. Chem. Eng. Conf. 3 (1993) 429–432. Eng. Chem. Res. 33 (1994) 1803–1808.

You might also like