You are on page 1of 7

Journal of Membrane Science 265 (2005) 153159

Long-term performance of membrane distillation process

Marek Gryta
Technical University of Szczecin, Institute of Chemical and Environment Engineering, Department of
Water Technology and Environment Engineering, ul. Puaskiego 10, 70-322 Szczecin, Poland

Received 25 January 2005; received in revised form 5 April 2005; accepted 27 April 2005
Available online 11 July 2005


The results of the over 3 years time research on the direct contact membrane distillation applied for production of the demineralised
water have been presented. The hydrophobic capillary polypropylene membranes (Accurel PP S6/2) were used in these studies. The inlet
temperature of the feed and the distillate was 353 and 293 K, respectively. The membranes were found to be thermally stable, and good
separation characteristics was maintained throughout the whole period of the investigations. The SEM observations of the capillary membrane
confirmed that their morphology remained unchanged. It was found that water (permeate from reverse osmosis process) did not cause the
wetting of the used membranes despite a long-term membrane module exploitation. The precipitation of CaCO3 on the membrane surface
was observed when tap water was used directly as a feed. A partial wetting of the membrane was found in this case. The wettability resulted
in the increased electrical conductivity of the distillate obtained from 0.9 to 2.5 S/cm.
2005 Elsevier B.V. All rights reserved.

Keywords: Membrane distillation; Hydrophobic membrane; Pilot plant; Membrane wetting

1. Introduction taneous crystallization, are reported [16]. The MD process

was also successfully applied for the separation of volatile
In the process of membrane distillation (MD) the gas phase metabolites (e.g. ethanol) in the membrane bioreactors [7].
(non-wetted porous membrane) is in the contact with the liq- Membrane distillation has been known for 40 years, how-
uid phase [1]. During the process only the water vapour and ever, it is still used on a laboratory scale. The tests, which
other volatile components present in the feed are transported were conducted in the 1980s with a view to implement this
across the membrane. One of the ways of putting the process process have resulted in establishing only a few pilot plants
into practice is the direct contact MD (DCMD), in which [1,2,8]. In recent years, the process has been tested under the
the membrane separates the hot feed from the cold distil- industrial conditions for the purpose of fruit juices concentra-
late [13]. In this case, the vapour pressure gradient, which tion (osmotic MD) [9]. The difficulties involved are basically
results from the different temperatures and the solution com- associated with a phenomenon of membrane wetting and the
positions in the boundary layers adjacent to the membrane, is formation of deposits on the membrane surface (fouling and
the driving force of the mass transport across the membrane scaling). The latter phenomena are responsible not only for a
[1,2,4]. decline of the plant efficiency, but also may cause a damage
MD process demonstrates the excellent properties to retain of the MD module over a short time [4,5,7].
the non-volatile solutes, therefore, it is often proposed for Currently, the membranes for the MD process have not
the purpose of water desalination and wastewater treatment yet been produced, therefore, the hydrophobic membranes
[15]. A number of other applications, such as HCl recovery manufactured for the microfiltration process are commonly
and the concentration of the salt solutions with their simul- used in the MD experiments [1,2,8]. Moreover, the litera-
ture reports on the MD studies usually describe a short-time
Fax: +48 91 449 46 86. membrane usage. Only a few papers deal with the studies per-
E-mail address: formed over a period of several weeks [4,5,7,8,10]. Taking

0376-7388/$ see front matter 2005 Elsevier B.V. All rights reserved.
154 M. Gryta / Journal of Membrane Science 265 (2005) 153159

into consideration the above factors, the question how long due to a change of the membranes pores size. During the first
the membranes proposed for MD process could be operate 72 h of the polypropylene membranes exploitation (Accurel
still remains unanswered. PP, Celgard), an increase of the diameter of the large pores
In many cases the polypropylene Accurel PP capillary by up to 25% together with a simultaneous collapse of the
membranes proved to be successful in the MD process small ones, was observed [11].
[25,7,8,11,12]. The results of the MD studies with the A partial wetting of the membranes, associated with the
Accurel PP S6/2 membranes, performed during the 3-year phenomena proceeding on their surface or resulting from
(including the shutdowns periods) module exploitation, were the transformation of the membrane material characteristics,
presented in this paper. The MD process was employed to poses problems which are difficult to eliminate [4,10]. The
produce the demineralised water. deposit formed on the membrane surface is a factor impeding
the possibility of the long-term MD modules exploitation.
This deposit causes clogging of the evaporation surfaces
2. Theory which leads to a decline of modules efficiency [8,12,17].
Furthermore, the membrane surface is wetted in the place
The retention of the gas phase in membrane pores during where the deposit was formed, hence, the liquid may pene-
the MD is an essential condition for the process function- trate the adjacent pores [7]. This phenomenon will certainly
ing. This condition is met when the membranes made of be accelerated if the formed deposit comprises the salt crys-
highly hydrophobic polymers, such as polyvinylidenefluo- tals growing into the pores [4,5,12].
ride (PVDF), polypropylene (PP) or polytetrafluoroethylene
(PTFE) are employed [113]. Apart from the hydrophobic
character of the membrane material, the liquid surface pres- 3. Experimental
sure, pores diameter and the hydraulic pressure influence
on the possibility of the liquid penetration into the pores. The investigations of the DCMD process were carried
This relation is described by the LaplaceYoung (Kelvin law) out in a continuous mode using a fully automated experi-
equation [1,2,8,14]. mental set-up shown in Fig. 1. The MD installation consists
4B cos of two thermostatic cycles, the feed and the distillate one,
P = PF PD = (1) which were connected to the membrane module. A filter net
(80 mesh) was mounted at the module inlet to prevent clog-
where B is the pore geometry coefficient (B = 1 for cylin- ging of the capillary membranes. The RE4 microprocessor
drical pores), is the surface tension of the liquid, is regulators compatible with the PT100 thermometers were
the contact angle, dP is the pore diameter, PF and PD are employed for to control the operation of the MD installa-
the hydraulic pressure on the feed and distillate side, respec- tion. The membrane module with the diameter of 0.025 m
tively. According to Eq. (1), each membrane is characterized and the effective length of 0.53 m was installed in a verti-
by a critical pressure above which the liquid will penetrate cal position. The module was equipped with 30 hydrophobic
the membrane pores. This pressure is known as the liquid capillary polypropylene membranes (Accurel PP S6/2, Mem-
entry pressure of the water (LEPW) [1,2,15]. For example,
if the contact angle value is 130 (PTFE membrane), the
water ( = 72 103 N/m) flows into the cylindrical pores
having the diameter of 1 m at pressure of 185 kPa [1]. Tak-
ing into consideration the possibility of membrane wetting,
it is recommended that for MD process the maximum diam-
eter of the membrane pores does not exceed 0.5 m [2].
Moreover, the usefulness of Eq. (1) may be limited by the
uncertainty of angle value. A number of LEPW values
determined experimentally, for the pores diameter within the
range 0.22 m, were presented in the work [8]. For these
presented LEPW results the value calculated from Eq. (1)
is equal to 93.5 0.2 for the polypropylene Accurel PP
At the initial stage of a new MD module working, a small
increase of the permeate flux caused by the minor changes
in the membrane morphology was observed [1012]. The
hydraulic pressure of the liquid may lead to a slight com- Fig. 1. The experimental set-up for the DCMD process: 1MD capillary
module, 2heat exchanger, 3distillate tank, 4feed tank, 5impeller
paction of membrane, which shortens the vapour diffusion
pump, 6heater, 7flowmeter, 8filter, 9RE4 controller, 10RP7 cur-
path and as a consequence the permeate flux is enhanced rent relay, Tthermometer, Pmanometer, conductometer, V1, V2,
[16]. Moreover, an increase of the process yield may be also V3valves.
M. Gryta / Journal of Membrane Science 265 (2005) 153159 155

brana GmbH, Germany), with the outside/inside diameter

dout /din = 2.6 mm/1.8 mm. The capillary membranes have the
pore size with the maximum and nominal diameter of 0.55
and 0.22 m, respectively, and the porosity of 72% (the man-
ufacturers data). The membranes were arranged as a parallel
bundle of braided capillaries (three membranes in the braid).
The total active surface area of membranes (A) for the mass
transfer was calculated for the internal capillary diameter and
amounted to 0.0889 m2 . The values of the permeate flux were
calculated from the equation:

Ji = 24 (L/(m2 h)) (2)

where A = 0.0889 m2 , VD (L) is the permeate volume col-

lected over a period of time ti (h) of the MD process Fig. 2. Time-events during water demineralisation in MD process.
The feed flows inside the capillaries, whereas the distillate feed was found. This mode was the most frequently used
flows through the inter-tubular space. The impeller pumps variant of the installation work.
were used to provide the flow of liquids in the MD installa- Mode II. The obtained permeate was returned to the feed
tion. The volumetric flow of both the permeate and the feed tank (valve V1, open valve; valve V2, closed). In this case,
was measured by the flowmeters. The calculated feed flow the feed concentration was constant over the entire course
rate alongside the membrane surface was 1.35 m/s, whereas of the MD investigations. This mode of work was found its
that of the distillate was 0.25 m/s. The initial hydraulic pres- application at weekends.
sure of both streams (on the feed and distillate side) at the The electrical conductivity and the total dissolved solids
module inlet was equal (regulation by throttling valve V3), (TDS) of the used waters were measured with a 6P Ultram-
and amounted to 52 kPa. Only in the S1 series (the initial eter (Myron L Company). A Yobin Yvon Ultrace 238 JY
400 h of the experiments), the pressure on the distillate side inductively coupled plasma atomic emission spectrometer
was lower, reaching 50 kPa. The inlet temperature of the feed (ICP-AES) was used to determine the content of cations in
and distillate was 353 and 293 K, respectively, and was main- the samples. The content of the inorganic carbon (IC) and the
tained constant during the MD process. total organic carbon (TOC), both in the feed and the permeat,
The permeate from the reverse osmosis process (RO), the was determined using a TOC-Analyzer multi N/C (Analytic
installation of which was supplied with tap water, was used as Jena).
a feed in the MD studies. For the RO process the module with The membrane morphology and the composition of the
the BW3040 element (Film Tec) was applied. Furthermore, fouling layer were studied using a Jeol JSM 6100 scanning
at the final stage of the experiments, the MD installation was electron microscopy (SEM) coupled with the energy disper-
supplied with the tap water. The MD installation was oper- sion spectrometry (EDS). The accelerating voltage used was
ated in a continuous mode, in a few measurement series. The 20 kV. The samples were sputter coated with gold and palla-
installation was shutdown between the measurement series dium.
for a period ranging from a few weeks to several months.
The scheme of the time-events occurring during the opera-
tion of the MD module is presented in Fig. 2. 4. Results and discussion
While the installation was shut down, the MD module
was filled with the retentate and the distillate obtained from The operation of the presented MD installation (Fig. 1) in
the completed series. Before the next series was started, the a continuous mode allowed the simulation of the conditions
installation was emptied and rinsed with the 2% HCl solution, of the membrane module work similar to those of the indus-
followed by the distilled water. The installation was then sub- trial installation. This is of a significant importance in the
sequently filled with the RO permeate and the MD process case of the investigations associated with the evaluation of
was initiated. The studies were conducted in the two modes the fouling and wetting phenomena. In traditional laboratory
of the installation operation: installations, the membrane temperature decreases during the
Mode I. The MD permeate was collected in a continuous night shutdown. This may cause the salt precipitation from
way (valve V1, closed; valve V2, open) and a level of the the feed, which accelerates membranes wetting [4,5,12], and
liquid in the feed tank was kept constant by the continuous therefore, the results of the laboratory studies may substan-
dosing of the new portions of the feed water. As a result of tially differ from those of the industrial plants.
the permeation of water vapour through the membranes, a The substances dissolved in the feed water have a sig-
gradual increase of the solute concentration in the recycled nificant influence on the membranes wetting. In order to
156 M. Gryta / Journal of Membrane Science 265 (2005) 153159

remove the solutes, the water treated in the process of the

reverse osmosis was used in the MD experiments. The RO
pilot installation was supplied with the tap water having the
electrical conductivity of 605650 S/cm. The correspond-
ing solute content (TDS) in the tap water was found to be
in the range 409430 mg/L. The TOC analysis revealed that
the tap water contained 2931 mg/L of inorganic carbon (IC)
and 6.88.5 mg/L of the total organic carbon (TOC). The
average concentrations (in mg/L) of the major ions were
as follows: 29.2 Na, 59.2 Ca, 16.6 Mg, 6.1 K and 2.8 Si.
The composition of the permeate obtained from the RO pro-
cess was dependent on the degree of water recovery, which
was in the range 6075%. On an average, the RO perme-
ate contained (mg/L): 0.98 Na, 0.78 Ca, 0.46 Mg, 0.3 K
and 0.043 Si. The amount of the compounds containing car-
Fig. 4. SEM micrograph of the inner surface of the Accurel PP S6/2 mem-
bon (TOC analysis) was reduced to a value of 0.60.9 mg/L brane.
(TOC) and 1.22.2 mg/L (IC). The RO permeate has the elec-
trical conductivity of 7.111.9 S/cm and the TDS value of
4.57.6 mg/L. on the membrane surface are larger. The existing differences
The MD process, in which the RO permeate was employed occur in the layer adjacent to the surface up to the depth of
as the feed water, was carried out in a few series and the about 2040 m. The pore size distribution is more uniform
experimental results are shown in Fig. 3. At the initial stage in the deeper layers of the membranes wall [12]. Some of the
of the investigations, an increase of the module efficiency pores located on the membrane surface are very large, some-
from 670 to 715 L/(m2 d) was observed. One of the reasons times having the diameter of 510 m (Fig. 4). It was found,
causing a change of the permeate flux may be associated that the pore size on the outer surface (distillate side) was sig-
with a change of the membrane morphology, the occurrence nificantly larger than that on the surface inside the capillaries
of which was indicated in other works [10,11]. The SEM (feed side). According to Eq. (1) for the angle value equal
observations performed with the brand-new membrane and 93.5 , the pores with the diameter equal 5 m are filling with
the samples of the membranes collected from the module water at the pressure of about 35 kPa. Such a small LEPW
after the MD tests did not reveal any significant differences value means that the surface large pores could be filled with
in the morphology. Thus, it can be concluded that the polymer water (especially on the distillate side) at the initial stage of
material employed for membranes production exhibit a good the MD tests, which caused a decrease of thickness of the gas
thermal stability. However, even a slight increase of the pore layer entrapped in the membrane. This shortens the vapour
size, difficult to assess visually, can lead to the reduction of diffusion path and consequently the MD process efficiency
the resistances of vapour diffusion across the membrane and was increased. Unfortunately, wetting of the large surface
to increase the process efficiency [1,2,10]. pores initiates the same process for the smaller pores adja-
The observed initial permeate flux increase could also cent to them. Finally, the liquid filled the pores throughout
result from the asymmetrical structure of the used mem- the wall and the MD process stops in a given part of the mem-
branes. The SEM observation revealed that the Accurel PP brane surface, which results in the decline of the membrane
membranes display a sponge structure and the pores located module efficiency.
After the initial changes, the course of the MD process was
stabilized and its efficiency amounted to about 700 L/(m2 d)
over the next 200 h (Fig. 3). However, at the end of the S1
series (period 230400 h), the module efficiency decreased to
about 640 L/(m2 d). The quality of the obtained distillate was
not deteriorate and the electrical conductivity remained sta-
ble at a level of 1.6 S/cm. This indicated that the observed
decline of the module efficiency was not the result of the
membrane wettability. In the case of the membrane wetting,
the solutes diffuse through the pores filled with the liquid,
which resulted in a significant increase of the permeate elec-
trical conductivity [4,14]. Moreover, in the S1 series, the
hydraulic pressure on the feed side was slightly higher than
that of the distillate side, which additionally enhanced the
Fig. 3. Variation of the permeat flux and feed concentration (TDS) as a possibility of leakage and consequently the distillate con-
function of operating time of the MD process. tamination.
M. Gryta / Journal of Membrane Science 265 (2005) 153159 157

Fig. 6. The result of SEM-EDS analysis of deposit formed on the filter net
(Fig. 5).
Fig. 5. SEM micrograph of the filter net with the deposit formed during the
MD process. used exclusively as a feed water. Several shutdown periods
occurred between the measurement series. The module was
It was found that the major reason of the observed perme- filled with the solutions from the accomplished measure-
ate flux decline was the formation of the deposits layer in the ment series over these periods. The S2 series was started
filter mounted at the inlet of the module (Fig. 1). The SEM after a standstill lasting 1 year and the permeate flux about
image of the filter surface covered by deposit after complet- 690 L/(m2 d) was obtained. That efficiency was similar to that
ing the S1 series is shown in Fig. 5. The formation of deposit obtained for a new module. Thus, it can be concluded that 1-
was also observed in other measurement series. The deposit year storage of the module with the channels filled with the
caused the clogging of the flow channels through the filter processing water will not result in the membranes wettabil-
net, which led to an increase of the flow resistance, thus the ity. However, the formation of biofilm on the walls of the
feed hydraulic pressure increased from 50 to 54 kPa. The feed and the distillate tanks was observed during the installa-
feed was pumped with the use of the impeller pump, there- tion shutdown. The SEM image of the deposit collected from
fore an increase of the feed flow resistance caused a decline the tank walls and subsequently separated with the use of the
of the flow rate. As a consequence, the feed flow rate in the MILLEX-HV filter is presented in Fig. 7. The microbial tests,
module decreased from 1.35 to 1.23 m/s. A decrease of the demonstrated in previous works [17], indicated that non-
flow rate caused an unfavourable increase of the temperature fermenting gram-negative rods were present in the collected
polarization and comprises a possible reason of the observed deposit. Therefore, taking into consideration the possibility
reduction of the MD process efficiency [1,2,4,8]. of biofouling occurrence it is recommended to drain the liq-
Furthermore, the trace amounts of deposit were accumu- uids from the MD installation before a prolonged standstill.
lated at the entrances of some capillary membranes and it A period of time of the standstill between the measurement
caused a disturbance of the feed flow inside the capillaries. series was presented in Fig. 2. Irrespective of their duration,
The mechanisms of this phenomenon was presented in work the MD installation efficiency was found to be similar to the
[18]. The removal of the deposit both from the filter and the initial permeate flux (Fig. 8). The stability of the electrical
membrane inlets enables the recovery of the initial process conductivity of obtained distillate confirms that the mem-
efficiency each time (Fig. 3). This result confirmed the pre- branes wettability was not observed in the course of feeding
vious results, that the formation of deposits at the inlet of the
capillary membranes comprises one of the major problems
of the MD process operation over an extended period [18].
The SEM-EDS analysis demonstrated that the deposit
accumulated on the filter net contained a large amounts of
Ca and Si (Fig. 6). During the performed MD tests the instal-
lation worked mostly according to Mode I. Hence, a constant
dosing of the feed water to the feed tank led to a gradual
increase of the solutes in the feed (Fig. 3, TDS). However,
the obtained concentration of the solutes was relatively low,
thereby, one can hardly explain the formation of deposit at
this concentration. This problem will be the subject of the
further works.
The experimental results presented in Fig. 3 were obtained
during almost the 2-year studies of the MD process per- Fig. 7. SEM micrograph of the micro-organisms in the deposit collected
formance. In these investigations the RO permeate was from the distillate tank walls.
158 M. Gryta / Journal of Membrane Science 265 (2005) 153159

Fig. 8. The influence of the operating time of the MD process and the nature
of the feed water on the changes of the permeate flux and the electrical
conductivity of distillate. Fig. 10. SEM micrograph of CaCO3 deposit on the membrane surface, feed:
tap water.

the MD installation with the RO permeate. The higher val- late was similar to the analytic zero. This confirms a fact that
ues of the electrical conductivity (Fig. 8, at the beginning regardless of the time of the process duration, membranes
of the measurement series), resulted from the quality of the demonstrated a high retention of inorganic solutes. Only a
distilled water, which was used for filling the distillate tank trace amount of TOC, probably associated with the volatile
at the start of the installation. A rapid decline of the electri- compounds, was detected in the distillate. It is characteristic
cal conductivity of the water in the distillate tank, observed feature of the MD process that volatile organic compounds,
after consecutive hours of process operation, confirmed that similarly to water vapour, are not retained [1,2].
the obtained permeate was pure water. The average value In the final stage of the MD studies, tap water was applied
of the distillate electrical conductivity amounted to 1 S/cm. as the feed water, which led to a rapid decline of the process
However, a lower value of the electrical conductivity equal to efficiency. As a result of heating the feed, the HCO3 ions
0.4 S/cm was achieved in other MD investigations with the present in the water undergo the decomposition and a con-
same membranes, as reported in work [8]. In the experimen- siderable amount of CaCO3 precipitates on the membrane
tal set-up (Fig. 1), the container lids were not airtight and the surface (Fig. 10), which reduces the surface of water vapour-
ambient air could penetrate the distillate. The installation was ization (pores) [5]. The formed deposit was removed every
located in the chemical laboratory so the effect of the gaseous 4080 h by rinsing the module with a 25 wt.% HCl solution
emissions on the quality of the produced distillate was notice- which enabled the recovery of the initial process efficiency.
able. Gases (such as NH3 or CO2 ) while dissolving in the However, the results demonstrated in Fig. 8 show that a mul-
distillate, increased the value of the electrical conductivity. tiple repetition of this operation resulted in a gradual decline
The TOC analysis (Fig. 9) was found to be a more appro- of the maximum flux of the permeate (feeddistilled water).
priate method to evaluate the performance of membrane At the same time, an increase of the electrical conductivity
separation over the discussed period. Despite the increasing of produced distillate was observed. This confirms that the
value of the feed concentration, the IC content in the distil- observed decrease of the maximum permeate flux was caused
by a partial wetting of the membranes. This conclusion can
be confirmed by the fact that the module efficiency increased
from 555 to 650 L/(m2 d) as a result of the membranes dry-
ing (Fig. 8, 2800 h). Unfortunately, the permeate flux was
found to decline after restarting the MD process with tap
water. After 3300 h of running the MD process investiga-
tions, the installation was finally rinsed with a HCl solution.
Subsequently, the MD process was operated for almost 200 h,
feeding the installation with the distilled water. The permeat
flux, which was obtained at that time (period 33003500 h),
was found to be stable. That confirms that deposits formed
on the membranes surface (Fig. 10) were responsible for the
decline of the efficiency in the previous stage of the mod-
ule operation (feedtap water). As it has been demonstrated
in previous works, the formation of deposits on the mem-
Fig. 9. The changes of TOC and IC concentration in the feed and the distillate
brane surface has accelerated wetting of the membrane pores
as a function of the operating time of the MD, feed: RO permeate. [4,12].
M. Gryta / Journal of Membrane Science 265 (2005) 153159 159

5. Conclusions [2] A. Burgoyne, M.M. Vahdati, Direct contact membrane distillation,

Sep. Sci. Technol. 35 (2000) 1257.
[3] M. Tomaszewska, Membrane distillation examples of applications
The experimental results have demonstrated that the wet-
in technology and environmental protection, Pol. J. Environ. Stud. 9
tability of polypropylene Accurel PP S6/2 membranes by (2000) 27.
pure water does not occur in the MD process. These enable [4] M. Gryta, Concentration of NaCl solution by membrane distillation
a long-term operation of the MD modules made from these integrated with crystallization, Sep. Sci. Technol. 37 (15) (2002)
membranes for the separation of water solutions without the 3535.
[5] K. Karakulski, M. Gryta, A. Morawski, Membrane processes used
suspended solids. Therefore, the Accurel PP membranes can
for potable water quality improvement, Desalination 145 (2002)
be recommended for the application in the membrane distil- 315.
lation process. [6] V. Calabro, B.L. Jiao, E. Drioli, Theoretical and experimental study
It was found that the MD modules filled with processing on membrane distillation in the concentration of orange juice, Ind.
water could be storage for a few months without a deterio- Eng. Chem. Res. 33 (1994) 1803.
[7] M. Gryta, The fermentation process integrated with membrane dis-
ration of its efficiency. However, it is recommended that to
tillation, Sep. Purif. Technol. 24 (2001) 283.
drain the installation before standstill, due to the possibility [8] K. Schneider, W. Holz, R. Wollbeck, Membranes and modules for
of occurrence of biofouling phenomenon. ransmembrane distillation, J. Membrane Sci. 39 (1988) 25.
In order to prevent a clogging of the capillary inlets by the [9] F. Vaillant, E. Jeanton, M. Dornier, G.M. OBrien, M. Reynes,
suspensions present in the feed, it is important to locate the M. Decloux, Concentration of passion fruit juice on an indus-
trial pilot scale using osmotic evaporation, J. Food Eng. 47 (2001)
filtering systems close to the MD modules.
The deposits formed on the hydrophobic surface of mem- [10] F.A. Banat, J. Simandl, Theoretical and experimental study in mem-
brane cause that the pores adjacent to the deposit will be brane distillation, Desalination 95 (1994) 39.
filled with liquid. As a result of the partial wetting of the [11] A.M. Barbe, P.A. Hogan, R.A. Johnson, Surface morphology changes
membranes, the MD process efficiency decreases from 700 during initial usage of hydrophobic, microporous polypropylene
membranes, J. Membr. Sci. 172 (2000) 149.
to 550 L/(m2 d). Moreover, the electrical conductivity of the
[12] M. Gryta, Direct contact membrane distillation with crystallization
obtained distillate was increased from 0.9 to 2.5 S/cm. applied to NaCl solutions, Chem. Pap. 56 (2002) 14.
The removal of deposits from the membranes surfaces [13] M. Tomaszewska, Preparation and properties of flat-sheet membranes
(e.g. CaCO3 by washing them with a HCl solution) and rins- from poly(vinylidene fluoride) for membrane distillation, Desalina-
ing the membranes with pure water, followed by drying, tion 104 (1996) 1.
[14] H. Udriot, U. van Stockar, New method for the determination of
facilitates the recovery of the initial efficiency of the mem-
wettability in a membrane distillation process, in: Advances in sepa-
brane module. ration processes, Hemisphere Publishing Corporation, London, 1990,
pp. 135145.
[15] K. Byoung-Sik, P. Harriott, Critical entry pressure for liquids in
Acknowledgements hydrophobic membranes, J. Colloid Interface Sci. 115 (1987) 1.
[16] K.W. Lawson, M.S. Hall, D.R. Lloyd, Compaction of microporous
membranes used in membrane distillation. I. Effect on gas perme-
This work was supported by a grant from Polish State ability, J. Membr. Sci. 101 (1995) 99.
Committee for Scientific Research. [17] M. Gryta, The assessment of microorganism growth in the membrane
distillation system, Desalination 142 (2002) 79.
[18] K. Karakulski, M. Gryta, Water demineralisation by NF/MD inte-
grated processes, Desalination 177 (2005) 109.

[1] K.W. Lawson, D.R. Lloyd, Membrane distillation, J. Membr. Sci.

124 (1997) 1.