Arabian Journal for Science and Engineering

https://doi.org/10.1007/s13369-018-3096-3

REVIEW ARTICLE - MECHANICAL ENGINEERING

Nanofiltration as a Pretreatment Step in Seawater Desalination:

A Review
Bassel A. Abdelkader1 · Mohamed A. Antar1 · Zafarullah Khan1

Received: 8 June 2017 / Accepted: 1 February 2018

© King Fahd University of Petroleum & Minerals 2018

Abstract
One of the main applications of nanofiltration (NF) is in the pretreatment stage of seawater desalination. NF has high rejection
rates for divalent ions and could eliminate the scaling species that pose serious fouling problems in seawater desalination. This
review comprehensively examines recent advances in NF membrane research in seawater desalination. Significant progress
has been made in understanding the mechanism of solute transport through NF membranes and has resulted in the development
of predictive models based on the Spiegler–Kedem model and the modified Nernst–Planck equation. The contributions of each
type of transport mechanism through NF membranes, i.e., convection, diffusion, and electro-migration, have been reported.
A review of recent progress made in the development of integrated NF membrane and seawater desalination processes is
included. Work related to membrane fouling, which is a key problem in NF, is also discussed.

Keywords Nanofiltration · NF · Desalination · Pretreatment

1 Introduction the removal of scale-forming agents (mainly sulfates) that

adversely affect desalting processes.
Due to continuous increases in the population, higher liv- NF membranes typically achieve divalent ion rejection
ing standards and needs, and the expansion of industrial and of 75–99%, whereas monovalent ion rejection is typically
agricultural activities, there has been a continual and rapid in the range of 30–50% [1]. For this reason, NF alone is
increase in the global demand for clean water. The scarcity of insufficient in reducing the total dissolved salts (TDS) con-
clean water is recognized as a global threat to domestic, agri- tent to the permissible limit of < 500 ppm set by the World
cultural and industrial activities and thus to the very existence Health Organization [2]. To optimize ion rejection while tak-
of society. Urgent measures are needed to avoid this threat. To ing energy efficiency into account, NF should be integrated
meet these increased demands for clean water, there has been with other major desalination processes, such as pretreatment
extensive research on desalination technologies that convert in seawater reverse osmosis (NF–SWRO), multistage flash
vast, available yet unusable water resources, such as sea and distillation (NF–MSF), multi-effect distillation (NF–MED),
brackish water, to usable forms. and NF–SWRO-reject -MSF desalting technologies.
One major impediment to the widespread application of As shown in Fig. 1, brine temperature limitations in ther-
seawater desalination technology (thermal-based or mal seawater desalination techniques (e.g., MSF and MED)
membrane-based) is its relatively high production costs. are dependent on the presence and concentrations of diva-
However, many recent advancements in seawater desalina- lent ions. Scaling occurs under supersaturation conditions
tion technology have led to appreciable declines in water and proceeds with nucleation, crystal growth, and precip-
production costs. One such advancement is the applica- itation [3]. Temperature, pH, TDS, and ion concentrations
tion of nanofiltration (NF) as a pretreatment technique for influence the rate of scale formation. NF could eliminate
both thermal and membrane-based processes. NF targets the scaling species in thermal seawater desalination plants,
which helps to increase the top brine temperature (TBT) and
B Mohamed A. Antar the number of stages in MSF or MED technologies, thereby
[email protected] increasing the gain output ratio (GOR) [4]. At present, scaling
is controlled by the addition of anti-scaling chemicals. There
1 Mechanical Engineering Department, King Fahd University are two types of scalants in MSF and MED technologies:
of Petroleum and Minerals, Dhahran, Saudi Arabia

123
 Arabian Journal for Science and Engineering

Fig. 1 Chemical relations of scale deposition in thermal seawater desalination [6]

alkaline scales caused by calcium carbonate and magnesium such as membrane swelling, charge variation and electro-
hydroxide and non-alkaline scales caused by calcium sul- static effect.
fate [5]. The TBT is generally less than approximately 112 ◦ C This review discusses the potential of using nanofiltra-
in MSF and 66 ◦ C in MED. Studies have considered using tion as a pretreatment step for seawater desalination. It starts
NF membrane softening to remove divalent ions as a possi- by considering the performance of NF membrane, which
ble pretreatment for distillation processes. This step would includes ion rejection, permeate flux and membrane char-
allow operation under a higher TBT of 170 ◦ C for thermal acterization. Then the performance of NF as a pretreatment
desalination units. process for seawater desalination is highlighted and com-
Several literature reviews [7–11] on NF membranes have pared with other pretreatment methods. Then, the theory
been published. The use of NF membranes for pretreatment of ion transport through the NF membranes is discussed
in different desalination processes, membrane fouling, mea- followed by membrane fouling that represents the main oper-
surements of membrane pore size and roughness, membrane ation challenge. The wide scope of this review shows the
performance, and modeling have been discussed. Hilal et potential of using NF for future investigation.
al. [7] discussed the use of NF membranes as a pretreatment
step in seawater desalination. The authors reported that NF
membranes were able to remove microorganisms, turbidity
and a fraction of dissolved salts. However, the NF membranes
2 Performance of NF Membranes
encounter a major problem of fouling, which resulted in high
Llenas et al. [12,13] experimentally investigated six com-
energy demand and reduced lifetimes.
mercial NF membranes using synthetic and real seawater to
Mohammad et al. [8] reported on the fabrication of NF
identify the optimum membranes and operating conditions
membranes by interfacial polymerization and discussed the
for the removal of various ionic species that result in the
transport of ionic species through NF membranes based on
buildup of scale in SWRO, such as CaSO4 , Mg(OH)2 , and
the Nernst–Planck equation. The authors also addressed the
CaCO3 . The authors measured the membrane surface rough-
prevention and mitigation of NF membrane fouling. Bruggen
ness and pore sizes. Membrane pore sizes were obtained
et al. [9] reviewed the drawbacks of NF, discussed the prob-
based on the rejection of neutral solutes, i.e., glycerin, glu-
lems and mitigation of membrane fouling, and examined
cose, and 2-propanol. Membrane roughness was measured
solute transport models and simulations. Luo et al. [10]
using atomic force microscopy (AFM), as shown in Fig. 2.
discussed the effects of pH and salt concentration on NF
Ion chromatography, total carbon analysis and inductively
membrane flux, rejection and fouling and reported that the
coupled plasma mass spectrometry were used to measure
presence of salt and high pH resulted in increased permeabil-
anion (Cl− , SO2− − + +
4 , and Br ), cation (Na , K , Ca , and
2+
ity. This is because high pH can increase effective membrane 2+
Mg ), inorganic carbon, boron and strontium concentra-
pore size. Furthermore, the pH effect on the performance
tions. The operating pressure varied between 2 and 20 bars.
of the membrane can be explained by multi-mechanisms,
Results showed that the rejection of the scale-forming diva-

123
Arabian Journal for Science and Engineering

Fig. 2 AFM pictures of the studied NF membranes: a NF270; b NF200; c NF90; d ESNA 1-LF2; e K-SR2; f NF99HF [12]

lent ions was high in all tested membranes. Furthermore, the roughness characterizes the membrane effective surface area.
rejection of scale-forming ions was nearly the same for all Water permeability of membranes depends on the surface
membranes, whereas the rejection of monovalent ions was morphology such as surface area and thickness of the protu-
lower for real seawater than for synthetic seawater. berances. Through a comparison of morphologies between
Surface roughness is a measure of the texture of a surface membranes having different water permeability, a membrane
or the deviation of the actual membrane surface topogra- with a larger surface area or smaller thickness shows higher
phy from an ideal atomically smooth surface. Thus, surface water permeability [14].

123
 Arabian Journal for Science and Engineering

Fig. 3 SEM images of an NF270 membrane [17]

Hilal et al. [15] studied NF membrane performance for González et al. [17] investigated the rejection of sul-
both seawater and brackish water. They used three com- fates and chlorides in highly concentrated saline water using
mercial NF membranes (NF270, NF90, NF30) to treat salt NF270 membranes. The thickness of the membrane’s effec-
solution (NaCl) at concentrations up to 25,000 ppm. They tive layer was 0.7 ± 0.27 µm, as determined by scanning
determined the pore size and pore size distribution of the electron microscopy (SEM) as shown in Fig. 3. Sulfate rejec-
membranes using AFM. NF90 has the lowest pore size and tion reached 96%, whereas chloride rejection reached only
the highest roughness and porosity among the studied mem- 11%. The authors reported that sulfate rejection decreased
branes. It was found that NF30 has a pore size ranging from with increasing chloride concentration.
0.43 to 0.90 nm with a mean pore size of 0.61 nm and a stan- Izadpanah et al. [18] studied the ability of NE4040-90 NF
dard deviation of 0.12 nm. The roughness (RMS) is 1.45 nm membranes to remove ions from diluted seawater at operat-
and the calculated porosity is 12%. NF30 has a higher pore ing pressures of 4–10 bars. Three diluted seawater samples
size and the lowest roughness and porosity. Finally, NF270 were tested with TDS 9750, 7220, 4970 mg/L. The rejections
has the highest pore size with medium roughness and poros- of Ca2+ , Mg2+ , and total hardness were 96–98%, whereas
ity. The applied pressure varied between 2 and 9 bars in the TDS was 79–89%. However, the rejection decreased
the membrane tests. For all tested membranes, salt rejection slightly at considerably higher pressures. The salt rejection
and permeation flux increased with increasing pressure and and permeate flux for the NF membranes increased with
decreasing salt concentration. NF90 had the highest rejec- increasing pressure, as shown in Figs. 4 and 5. Transport
tion of up to 95%, this is due to the fact that NF90 membrane through NF membranes can be explained in terms of diffusion
has the smallest pore size. Nevertheless, NF30 had the lowest and convection [7]: as the pressure increases, the contribu-
rejection of 6% at a salinity of 5000 ppm. At a higher salinity tion of convection overcomes diffusion due to high water
of 25,000 ppm, salt rejection of NF90 membrane dropped to flux; therefore, the rejection increases. However, at extremely
41%, whereas that of the NF30 membrane dropped to 3%. high pressures [19], the rejection decreases with increas-
The salt rejection of the NF270 membrane was 11–29%. ing pressure for certain salts. Both the flux and polarization
NF270 membrane showed the highest flux, whereas the NF30 modulus increased with increasing pressure. At extremely
membrane had the lowest flux because NF30 membrane had high pressures (e.g., 34 bar), an increased polarization mod-
the lowest porosity. ulus overcomes the increased flux effect and thus results in
Ryabchikov et al. [16] investigated the performance of decreased rejection.
three different types of NF membranes (NF90, ERN, and The effects of ion compositions on NF rejection were
SR90) in seawater desalination using a cross-flow arrange- investigated [21,22]. Increasing salt concentration increases
ment. The operation pressures were 8–18 atm. The concen- anions rejection and decreases the cations rejection. The
trations of Cl− , SO2− 2+ 2+
4 , Ca , and Mg , the temperature overall ion rejection is high at high cation concentrations.
and pH of the solution were measured. Flux and salt rejec- However, the flux decreased as the cation concentrations
tion increased with increasing applied pressure. The total increased. Similarly, for increasing anion concentrations,
salt rejection was 30–40%, whereas the rejection of the salts the flux decreased (except for HCO− 3 for which the flux
influencing water hardness (Ca2+ and Mg2+ ) was 90–95%. increased). Chloride ions had stronger effects on the rejec-
tion of magnesium ions than on the rejection of calcium ions.

123
Arabian Journal for Science and Engineering

Fig. 4 Effect of feed pressure on ion rejection [20] Fig. 6 Effect of water temperature on ion rejection [20]

ate flux increased with increasing pressure. Temperature and

feed flow rate had weaker effects than pressure. Scaling was
likely in the first stage, whereas the scaling index was lower
in the second stage because the majority of the scaling ions
were removed in the first stage.
Harrison et al. [19] investigated the performance of dual-
stage NF membranes for seawater desalination. Three types
of membranes (NF90, NE-90, and TS-80) were used to inves-
tigate water flux, ion rejection, and fouling. The operating
pressure varied from 10 to 40 bar. The results showed that
dual-stage NF membranes desalinated the seawater. Vuong
et al. [27] studied a two-stage NF system for seawater desali-
nation. The first-stage permeate served as the feed for the
second stage. The operating pressures in the first and sec-
Fig. 5 Effect of feed pressure on water flux [20] ond stages were 36 and 20 bar, respectively. The rejection of
divalent ions was greater than that of monovalent ions.
A portion of the data available for NF membranes is
The electrostatic repulsion between sulfate and bicarbonate
presented in Table 1. The rejection percentages of the
ions and the NF membrane was strong at low concentrations.
NF membranes were highly dependent on the operating
Temperature impacted the membrane properties by alter-
conditions, ion compositions and membrane characteristics
ing diffusion of water and ions and the polymer separation
(Tables 1, 2).
layer [20,23,24]. At high temperatures, the polymer chain
in the separation layer increases and becomes more effec-
tive, resulting in increased membrane pore sizes. Diffusion of
water and ions also improves at high temperatures, as shown 3 Performance of NF Membranes in
in Fig. 6. Thus, the permeate flux increases with increas- Integrated Systems
ing temperature. Meanwhile, rejection of most ions increases
slightly, whereas that for Na and Cl ions decreases slightly. 3.1 UF–NF
This trend occurs because Na and Cl ions have relatively
smaller sizes than other ions, which enables them to easily Song et al. [29] investigated the performance of NF mem-
penetrate the NF membrane. branes in seawater desalination pretreatment by operating an
Many studies have investigated the performance of dual- integrated membrane system (UF–NF) for 500 h. There were
stage NF–NF membranes [19,20,25–27]. Liu et al. [20] two cleaning periods (after 280 and 440 h). The operating
examined the performance of a dual-stage NF system, in pressure was 20.3 bar at the start and it was then increased
which the first stage was operated at 35 bar and the second to 20.5 bar in the second period (after the first cleaning).
stage was operated at 20 bar. TDS, ion rejection and perme- The pressure increased to 20.6 bar in the third period (after

123
 Table 1 Performance of commercial NF membranes
Description Pressure Permeate Rejection (%) References
(bar) flow
(L/m2 h)
Ca2+ HCO−
3 Mg2+ SO2−
4 Na+ K+ C O32− Sr2+ Cl− Total TDS
hardness

123
Synthetic seawa-NF270 From 2 to 20 bar 15–120 37–63 37–60 65–82 97–98 10–22 18–28 – 48–74 17–27 – – [12]
ter
NF200 15–65 32–63 37–68 62–85 98–99 8–25 4–22 44–79 4–22
NF90 1–15 95–98 55–88 94–99 95–98 37–70 22–47 94–99 60–88
NF99HF 15–115 43–72 22–72 69–82 99 8–20 0–18 58–80 25–42
ESNA 1-LF2 25–140 25–40 15–38 38–48 65–75 38–42 15–25 23–40 17–23
K-SR2 30–130 45–62 15–42 72–82 98 3–10 8–12 60–78 10–15
Real seawater NF270 From 2 to 20 bar 20–120 43–70 35–58 65–85 95–98 5–30 12–28 – – 5–15 – – [13]
NF200 15–80 38–70 32–67 67–90 98–99.5 5–38 10–28 10–25
NF90 10–25 83–95 65–87 87–95 90–96 25–60 18–53 30–65
NF99HF 18–120 35–68 28–58 63–86 95–98 0–15 5–20 8–25
ESNA 1-LF2 50–280 15–17 10–38 22–23 47–67 8–12 5–15 5–12
K-SR2 30–190 38–60 22–40 55–75 97–98 5–8 4–10 0–12
(NaCl) salt solu-NF90 c= From 2 to 10 bar 2–40 – – – – – – – – – – 40–95 [15]
tion up to 25,000 5000 ppm
ppm
c= 2–25 27–75
10,000
c= 0.5–8 10–70
15,000
c= 0.5–4.2 7–52
20,000
c= 0.3–3 2–40
25,000
NF270 c= 20–110 11–29
5000 ppm
c= 15–90 6–21
10,000
c= 15–78 5–16
15,000
Arabian Journal for Science and Engineering
 Table 1 continued
Description Pressure Permeate Rejection (%) References
(bar) flow
(L/m2 h)
Ca2+ HCO−
3 Mg2+ SO2−
4 Na+ K+ C O32− Sr2+ Cl− Total TDS
hardness

c= 15–75 3–14
20,000
c= 10–60 3–11
25,000
NF30 c= 5–23 2–6
5000 ppm
Arabian Journal for Science and Engineering

c= 4–21 0.5–5
10,000
c= 3.5–19 0.5–4.2
15,000
c= 3–17 0–3.7
20,000
c= 2.5–15 0–3
25,000
CaCl2 500 mg/L NF90–4040 From 8 to 29.05 82.5–84.9 – 94.8–96 99.6–99.8 – – – – 20–29 – – [16]
MgSO4 2000 18 bar 38.38
mg/L
NaCl 2000 mg/L
SR90-4040 60.70 64–71 87.2–89.8 97.7–99.1 17–25 25–29
51.88
ERN KP-100-1016 64.29 94.2–96.8 97.8–98.5 99–99.3 – 45–59
Ca = 254 mg/L NE 4040-90 4 – 97.0588 – 96.3333 – – – – – – 96.5957 82.3541 [18]
Mg = 345 mg/L
6 98.2353 98.3333 98.2979 89.3964
8 97.6471 98 97.8723 89.2555
9 97.0833 97.9798 97.6871 82.1128
10 97.9167 97.1717 97.415 82.2359
IMS design pro- ESNA1- R = 60% 32.8 bar – 70.5 62.5 82.4 87.7 29 3.3 91.7 76 33.6 – 38 [28]
gram LF2
UF–NF ESNA3 4040 20.3 bar 34.71883 65.48263 98.96104 17.17797 16.61161 60.44234 24.46682 [29]

123
 Table 1 continued
Description Pressure Permeate Rejection (%) References
(bar) flow
(L/m2 h)
Ca2+ HCO−
3 Mg2+ SO2−
4 Na+ K+ C O32− Sr2+ Cl− Total TDS
hardness

123
Real seawater NF90-2540 30 bar 19 97.5 54.4 98.8 98.8 15.4 17.4 21.9 19.5 [30]
NF270-2540 65.8 95.5 76 96.9 99.2 50 48.9 56.3 51.7
ESNA3 R= 10 From 7 to 20 bar From 14.2 to 44.320 – 38 93 – – 75 – – 42 – [31]
Q/Q f
15 25 45 93 76 41.5
20 26 50 94 77 41
20 30 52 95 77 40.5
30 34 55 96 78 40
35 35 57 97 79 40
4–8 NF elements SR90 400 From 2 to 14 bar – 23 86 76 98 10 14 – – 4 – – [32]
5 NF modules 18 – 80.9 63.3 88 93.3 26.7 – – – 26.7 87 37.3 [33–39]
22 89.6 76.6 94 97.8 46.3 46.3 93.3 54.1
31 89.2 81.3 91.1 92.8 57.7 57.7 90.8 62.8
Arabian Journal for Science and Engineering
Arabian Journal for Science and Engineering

Reference
the second cleaning). The increased operating pressure indi-

[12,40]

[12,40]
[12,40]

[12,41]
[28,42]

[31,42]
[12,43]

[18,44]
cated that fouling occurred. Membrane fouling is difficult to

[40]

[15]
chemically clean. Salt rejection and permeate flux decreased
with time before the chemical cleaning and decreased more
Maximum feed

sharply after the chemical cleaning. Su et al. [45] studied the

performance of a UF-dual-stage, NF-integrated system. The
flow (m3 /h)

UF filtrate turbidity was nearly zero. For the dual-stage NF,

the permeate flux decreased with decreasing feed tempera-

53.5

4.09
1.4

1.4

17

17
–

–
ture. Salt rejection and energy consumption increased with
increasing operating temperature. The rejection of divalent
Maximum oper-
pressure

ions exceeded 95%.

3.2 NF and Ion-Exchange

ating
(bar)

41.6

41.6
41

40
55

41
-

-
Al-Rawajfeh et al. [46,47] studied the effects of NF and salt
precipitation as pretreatment processes on scale formation in
ating temperature
Maximum oper-

MSF and RO desalination processes. They used four differ-

ent configurations. Results showed that sulfate scaling was
significant in the MSF process, whereas carbonate scaling
(◦ C)

was significant in the RO process. Furthermore, the authors

45

45

45

45
50

45
–

showed that NF efficiently removed sulfate ions. The perfor-

Pore size (nm)

mance of the hybrid ion-exchange (IX) and NF process in

desalinating brackish water and seawater was also investi-
gated [48–53]. The ion-exchange-treated water served as the
0.34

0.46

0.69
0.49

0.46

0.61

feed to the NF membrane. IX converted more than 95% of the

0.5

1
-

chloride ions to sulfate ions. The NF process exhibited high

Roughness (nm)

sulfate rejection (greater than 99%). The NF retentate was

rich in sulfate and could be used to regenerate the exhausted
IX resin.
103.3

49.07

12.29
5.35
7.39

0.76

1.45

3.3 FO–NF
–

–
Polyamide thin-

Polyamide thin-

Polyethersulfone
Membrane type

Tan et al. [54] studied a hybrid forward osmosis and NF (FO–

film composite

film composite

NF) system for seawater desalination. The FO draw solution

Composite

Polyamide
polyamide

was circulated into a two-stage NF membrane as shown in

Fig. 7 at pressures up to 43 bar. Results showed that the
–

second-stage NF was necessary to desalinate seawater. A

hybrid FO–NF system for brackish water was also studied.
MICRODYN- NADIR Gmbh

This system required a lower hydraulic pressure than RO

Koch membrane systems
Dow chemical company
Table 2 Properties of commercial NF membranes

and had less flux decline caused by membrane fouling and

no pretreatment requirements [55,56]. Altaee and Hilal. [57]
NITTO DENKO

studied a multistage NF–FO–RO system for brackish water.

CSM Company
Manufacturer

Hydranautics

The system recovery rate exceeded 90%.

Alfa Laval
Company

3.4 NF–SWRO

Kaya et al. [30] investigated the use of NF as a pretreat-

Membrane name

ment stage for reverse osmosis in seawater desalination. They

ESNA3 4040
NF270-2540

ESNA1-LF2

NE4040-90
NF90-2540

examined the performance of a single NF (NF90 and NF270)

NF99HF

in SWRO and an integrated system (NF + RO). The NF

NF200

K-SR2

NF30
SR90

membranes were individually employed in closed loops at

30 bars. For the integrated system study, the NF permeate
10

served as the feed for the SWRO membrane. The perme-

1

2
3
4
5
6

7
8

123
 Arabian Journal for Science and Engineering

Fig. 7 Schematic diagram of the hybrid FO–NF system for seawater desalination [54]

ate recovery of the NF90 membrane was less than that of 3.6 NF–MSF
the NF270 membrane. Salt rejection of the NF90 membrane
was greater than that of the NF270 membrane. Al-Hajouri Mabrouk et al. [62] compared the results of numerical
et al. [58] investigated the performance of NF membranes simulations and experiments using a NF membrane as a pre-
in pretreatment processes for SWRO with respect to con- treatment step for an MSF pilot unit driven by solar energy.
ductivity and feed pressure over a period of 8 years. Feed Four solar concentrator modules with tracking systems were
pressure was varied from 18 to 38 bars. The initial perme- connected in series. Oil was thermally heated by solar energy
ability was high. However, a decrease in NF salt rejection as it passed through a shell and tube heat exchanger oppo-
was observed due to fouling. Chemical cleaning was con- site the NF permeate. The system was tested under a TBT of
ducted every 9 months. Song et al. [31] investigated the 100 ◦ C. The GOR was 15. The predicted GOR was 18 at a
performance of NF membranes to evaluate scaling poten- TBT of 130 ◦ C.
tial of NF–SWRO. NF membranes rejected sulfate ions at Awerbuch [32] used NF as a pretreatment for MSF to
10–15 bars and produced permeate with a lower scaling increase the TBT. The performance of the SR90 membrane
potential in thermal desalination. AlTaee et al. [26] simulated was studied. He reported that the use of NF reduced foul-
dual-stage NF–NF, NF–RO and single-stage RO membrane ing and increased the percentage of NF permeate used in the
processes using ROSA software [59] to compare perfor- MSF, which thereby increased the TBT. The performance of
mances, energy requirements, specific power consumption, the NF membrane as a pretreatment for MSF and RO desali-
and overall costs. Results showed that the NF–NF process nation processes was also studied by another research group
had the lowest cost and lowest specific power consump- from SWCC [33–39]. Three integrated systems were pre-
tion, whereas the RO process had the highest specific power pared and tested (NF–RO, NF–MSF, and NF–ROreject -MSF)
consumption. as shown in Figs. 9 and 10. NF permeate was used in the
MSF. The feed pressure was varied from 20 to 40 bar. For
the first 34 days, the MSF was operated at a TBT of 120 ◦ C
3.5 NF–SWRO–MD without using anti-scalants. Results indicated that no scale
formation occurred. However, the fouling factor increased
Drioli et al. [60] investigated the performance of integrated with increases in the recovery ratio. The MSF was then oper-
membrane systems and found that the integrated systems had ated for 50 days at a TBT of 130 ◦ C with 35 ppm acid. No
higher recovery factors than the RO unit alone. The produc- scaling issues were observed. In the trihybrid system, the NF
tion was nearly double that of the RO unit. El-Zanati et al. [61] permeate was used as the SWRO feed, and the SWRO reten-
investigated the performance of an integrated NF–SWRO– tate was used as the MSF feed. Due to the low concentrations
MD membrane system. NF was used in a pretreatment step of calcium and sulfate ions in the NF permeate, the MSF was
for the RO system. The rejected brine from the SWRO and operated at a TBT of 130 ◦ C. The NF feed temperature was
NF systems was used as feed for the MD process as shown in varied from 24 to 34 ◦ C, and the pressure varied from 24 to
Fig. 8. The NF and RO feed pressures were 31 and 34.2 bar, 28 kg/cm2 . The NF recovery rate increased with increased
respectively. The permeate flow rate was 42 m3 /d. The brine temperature and pressure.
flow rates for NF and RO were 30 and 28 m3 /d, respectively. Al-Rawajfeh [28,63] studied the influence of NF pretreat-
The MD feed rate was 58 m3 /d at 35 ◦ C. ment on scale formation in (MSF) desalination systems. The
NF rejection and permeate flow rates were simulated by the

123
Arabian Journal for Science and Engineering

Fig. 8 Schematic diagram of the hybrid NF–RO–MD seawater desalination process [61]

Fig. 9 Schematic diagram of

the dihybrid NF/MSF
desalination system [36]

Fig. 10 Schematic diagram of

the trihybrid NF/RO/MSF
desalination system [39]

123
 Arabian Journal for Science and Engineering

RO and MED systems. Curcio et al. [65] investigated the

performance of an integrated NF crystallization (NF–MC)
system. A NF90 membrane was operated at 34.5 bar. The
sulfate rejection exceeded 99%, and the recovery factor was
50%. Drioli et al. [66] studied the performance of a MF–
NF–RO–MD/crystallization system. Water recovery factor
was 92.8%. The addition of MC to the membrane-integrated
system resulted in reduced freshwater costs and brine con-
centrations.

Fig. 11 Effect of NF on sulfate scale potential in a BR–MSF plant [28]

4 Other Methods Used as Pretreatments for
the Thermal Desalination Process

4.1 FO–MSF

Altaee et al. [6,67,68] developed a simple mathematical

model to estimate the performance of a FO membrane inte-
grated with MSF or MED. The FO system used rejected
brine from MSF or MED as a draw solution and seawater as
the donor solution as shown in Fig. 13. The results showed
that the concentrations of multivalent ions were reduced
and the water flux across the FO membrane increased with
increasing seawater salinity. The concentration of divalent
ions decreased as the operating temperature in thermal plant
Fig. 12 Shifting the maximum TBT by NF [28] decreased, which in turn increased the recovery rate of the FO
system. Darwish et al. [69] studied the use of FO membranes
as a pretreatment for an existing MSF. With a FO recovery
Hydranautics Nitto Denko Company (IMS software). NF of 40%, the MSF could operate at 135 ◦ C without signif-
permeate was used as feed to the MSF. Scaling potential icant calcium sulfate scaling. The osmotic pressure across
decreased with increases in the percentage of NF-treated the membrane decreased with increases in the FO recovery
feed. Calcium carbonate and calcium sulfate deposition rates ratio.
sharply decreased through successive stages. The Skillman
index represents the ratio between the actual concentrations
of calcium carbonate and calcium sulfate and their solubility
at certain temperatures. Scale does not form when the solubil- 4.2 Salt Precipitation and MSF
ity is greater than the actual concentration (i.e., the Skillman
index is less than one). The effect of NF-treated feed per- The use of salt precipitation as a pretreatment for seawa-
centage on the Skillman index is shown in Fig. 11. The TBT ter desalination was investigated [70–72]. Al-Rawajfeh [70]
can be increased up to 175 ◦ C at 100% NF pretreatment, as investigated the performance of salt precipitation as a pre-
shown in Fig. 12. treatment for MSF at different make-up percentages (0–
100%). Calcium and magnesium ions were reduced by
3.7 NF and Salt Production 98–99%, whereas sulfate ions were reduced by approxi-
mately 56%. The Skillman index was used to estimate the
Turek et al. [64] investigated three types of NF membranes sulfate scale potential. The TBT reached 150 ◦ C at a 70%
(TFC-SR2 KOCH, NF270, and NF200) to evaluate their abil- make-up in the MSF and reached 170 ◦ C at 100%. Ayoub
ity in an UF–NF–RO–MED-salt crystallization-integrated et al. [71] showed that salt precipitation eliminated calcium
system. They reported that salt crystallization resulted in ions and resulted in a high rejection of magnesium ions of
decreased total costs and reduced waste products. This inte- approximately 99%. Table 3 presents a comparison between
grated system offered high water recovery (78.2%) and a different methods used as pretreatment for thermal desalina-
water unit cost of $0.5/m3 , which was less than those of tion processes.

123
Arabian Journal for Science and Engineering

Fig. 13 Schematic flow diagram of the FO–MSF hybrid system [69]

Table 3 Comparison between different methods used as pretreatment for the thermal desalination process
Method Maximum TBT (◦ C) Advantages Drawbacks References

FO 135 Simple no pressure needed Requires higher surface area [6,67–69]

Salt precipitation 170 Can eliminate 99% of scaling species Large area addition of alkalizing agents [70–72]
NF 170 Can eliminate 99% of scaling species High pressure [28,33–39]

5 Modeling of the NF Membrane first term on the right-hand side of the equation represents
the transport due to diffusion, whereas the second term rep-
NF modeling is critical for predicting flux, rejection, and resents the transport due to convection. P is the permeability
separations. Modeling also reduces the number of experi- of pure water, Ps is the permeability of the solute, x is the
ments and thus saves time and money. Two main approaches normal distance to the membrane, C is the concentration, and
have been used to model the transport of ions and dissolved σ is the rejection coefficient. The volume flux is given by Eq.
species. The first approach is based on the Spiegler–Kedem (2), where Jv is the volume flux, P is the pressure differ-
model [73–75], whereas the second approach is based on the ence across the membrane and π is the osmotic pressure
Nernst–Planck equation [75,76]. difference.
The solute rejection is calculated by integrating Eq. (1):
5.1 Spiegler–Kedem Model
1−σ Cp
R = 1− =1− (3)
In the Spiegler–Kedem model, the NF membrane is consid- 1−σF Cm
 
ered a black box, and the porosity of the membrane is ignored. Jv (1 − σ )
F = exp 1 − (4)
Therefore, complete details on the transport mechanisms Ps
cannot be determined [75]. This approach characterizes the
membrane in terms of salt permeability and rejection coef- where R is the rejection value, Cp is the permeate concen-
ficients. This model is applicable to binary salt systems. tration and Cm is the feed concentration.
For multicomponent mixtures, this model cannot accurately
quantify the effects of concentration polarization on mem- 5.2 Extended Nernst–Planck Equation
brane surfaces [73,74].
The extended Nernst–Planck equation [75–77] describes ion
dCs transport in terms of the porosity ratio, membrane thickness,
Js = −Ps + (1 − σ ) Cs Jv (1)
dX and membrane charge density. Rejection of mixed salts is
Jv = P (P − σ π ) (2) accurately predicted by this model. The model uses struc-
tural parameters, i.e., thickness, pore size, and tortuosity, to
The governing equation of the Spiegler–Kedem model is illustrate the mechanisms of transport of ions and dissolved
given by Eq. (1), where Js represents the solute flux. The species. The transport of solutes through pores in the mem-

123
 Arabian Journal for Science and Engineering

brane is based on hydrodynamic principles.

dCi,pore
Ji,pore = −Di,pore + K i,c Ci,pore jw (5)
dX
Equation (5) is the extended Nernst–Planck equation, where
Ji is the flux. The first term on the right represents the trans-
port due to diffusion, whereas the second term represents the
transport due to convection. D is the solute diffusivity, and
K is the mass transfer coefficient.
The Donnan–steric pore model (DSPM) [78,79] is based
on the extended Nernst–Planck equation. This model consid-
ers the effects of convection and diffusion on the transport
of ions through membrane pores. This model characterizes
the membrane in terms of the porosity ratio, pore radius, and
Fig. 14 Effect of normalized membrane charge density on the contri-
membrane charge density. butions of different transport mechanisms [82]
The Donnan–steric pore and dielectric exclusion model
(DSPM–DE) [80,81] considers the dielectric exclusion phe-
nomenon as an additional effect. The dielectric exclusion
(DE) occurs due to the interactions of ions with polarized
interfaces between media of different dielectric constants. The contributions of each mechanism have been inves-
The DSPM–DE model is based on the extended Nernst– tigated [82–84]. Convection is the dominant mechanism
Planck equation, as shown in Eq. (6), and accounts for ionic at high membrane thickness-to-porosity ratios and high
diffusion, convection through membrane pores, and electro- permeate flux. Diffusion is the dominant mechanism at
migration effects inside the membrane. Ionic separation low membrane thickness-to-porosity ratios, low permeate
through the membrane is characterized by three separation volume flux and high membrane charge density. Electro-
mechanisms, i.e., steric hindrance, dielectric exclusion, and migration dominates solute transport for moderate membrane
Donnan equilibrium. charge densities [82].

dCi,pore Z i Ci,pore Di,pore dψ

Ji,pore = −Di,pore − F
dX RT dX
+ K i,c Ci,pore jv (6)
6.1 Influence of the Membrane Charge Density
where Ji represents the flux. The first term on the right rep-
resents transport due to diffusion, the second term represents The contributions of convection increase with decreasing
transport due to potential gradient, and the third term repre- membrane charge density because the concentration of co-
sents transport due to convection. Z is the valence of the ionic ions inside the membrane is proportional to the convective
species, R is the universal gas constant, T is temperature, F flux of co-ions within the membrane. This flux increases
is the Faraday constant and ψ is the membrane potential. as the membrane charge density decreases according to the
Furthermore, Donnan exclusion model. Diffusion is the dominant mech-
n   anism involved in electrolyte transport. The contributions
z i Jv
dϕ i=1 Di, p K i,c Ci − Ci,pore of diffusion increase with the membrane charge density.
= Fn (7) However, both diffusion and convective flux decrease as the
dx RT i=1 z i Ci
membrane charge density increases. The decrease in convec-
dCi Jv   z i Ci dϕ
= K i,c Ci − Ci,pore − F (8) tive co-ion flux is more significant than that in the diffusive
dx Di,pore RT dx co-ion flux [82]. Therefore, the contributions of diffusion
 
Ci  2 zi F increase and the contributions of convection decrease with
= 1 − rs /rp exp − ϕ D (9)
Ci,pore RT membrane charge despite the decreased diffusive co-ion flux.
As shown in Fig. 14, the contributions of the three mech-
anisms level off at high membrane charge densities. The
6 NF Transport Mechanisms contributions of both convection and electro-migration tend
to be equal due to the asymptotic behavior of the electrical
The mechanisms responsible for solute transport through NF potential gradient and co-ion concentrations in the mem-
membranes are convection, diffusion, and electro-migration. brane [82,83].

123
Arabian Journal for Science and Engineering

Fig. 16 Effect of flux on the contributions of different transport mech-

Fig. 15 Effect of membrane pore size on the contributions of different anisms [82]
transport mechanisms [82]

tion of convection decreases and the contribution of diffusion

6.2 Influence of the Effective Pore Radius increases as the membrane charge density increases. Electro-
migration is the dominant mechanism for modest values of
The diffusion contributions increase and the contributions membrane charge density. This phenomenon occurs only
of convection and migration decrease with decreases in the when the co-ion diffusion coefficients surpass those of the
pore radius (Fig. 15). Although the mean co-ion convective counter ions. Electro-migration remains the dominant mech-
and migration flux decreases with decreasing pore size, the anism, even at high membrane charge density values, if the
decrease is less significant than that of the mean co-ion dif- permeate flux is increased [82].
fusion flux. At a low permeate flux, diffusion is considered
the dominant mechanism in electrolyte transport [82,83]. 6.5 Influence of the Thickness-to-Porosity Ratio

Figure 17 shows that decreasing the thickness-to-porosity

6.3 Influence of the Permeate Flux ratio results in an increase in the contribution of diffusion to
solute transport, whereas the contributions of convection and
As the permeate volume flux decreases, the contributions of electro-migration decrease due to the decrease in the Peclet
convection and migration decrease, as shown in Fig. 16. Con- number in the membrane. As the thickness-to-porosity ratio
vection becomes the dominant mechanism at high permeate increases, these contributions vary even though the retention
volume flux, whereas diffusion is the dominant mechanism remains constant. The variations in convection, diffusion and
at low permeate volume flux. The contribution of diffusion to electro-migration balance each other such that the amount
solute transport increases with decreasing permeate volume of solute transport through the membrane to the permeate
flux because diffusion flux reaches a critical value at high per- remains constant [82].
meate flux, whereas convection flux and electro-migration Ahmad et al. [73] developed a model for multiple solute
increase. The contributions of electro-migration and convec- filtration by extending the Spiegler–Kedem model. The
tion tend to be equal at any permeate flux at high membrane model parameters included the permeability of pure water,
charge densities [82,84]. a solute permeability constant, a rejection coefficient and
a mass transfer coefficient. The parameters were estimated
6.4 Influence of the Diffusion Coefficient using the Levenberg–Marquardt method and the Gauss–
Newton algorithm. The simulation results showed that the
The diffusion coefficient represents the ratio between the flux and permeate concentrations were in good agreement
molar flux due to molecular diffusion and the species concen- with experimental data. Radcliffe et al. [76] discussed the
tration gradient. For unequal diffusion coefficients between use of the extended Nernst–Planck equation for NF and
the co-ions and counter ions, the contributions of electro- reverse osmosis modeling. The role of dielectric exclusion
migration and convection tend to reach a constant value, and from the NF membrane was discussed. However, the inabil-
the electro-migration contribution has a value at a membrane ity to explain the physical properties and physicochemical
charge density of zero due to the diffusive potential formed phenomena inside nanopores was a major issue when apply-
by the presence of more transportable ions. The contribu- ing this theory to seawater desalination.

123
 Arabian Journal for Science and Engineering

Matin et al. [94] discussed the causes of biofouling,

biofilm formation, the role of the EPS and the events that lead
to biofouling. They found that membrane surface modifica-
tion was more promising than biocide and feed pretreatment
because biocide and feed pretreatment aim to limit microbial
concentrations in the feed stream, whereas membrane surface
modification reduces the adhesion of bacteria or inactivates
bacteria if adsorbed [95–98].
A mathematical model for a cross-flow NF system exam-
ined membrane resistance to estimate the scaling potential of
RO systems using NF membranes [99,100]. The concentra-
tions of the feed water were estimated, and the concentrations
of the sulfate and carbonate ions were calculated. The results
showed that surface foulant contents were primarily due to
Fig. 17 Effect of the membrane thickness-to-porosity ratio on the con- calcium carbonate and calcium sulfate. Lin et al. [101–103]
tributions of different transport mechanisms [82] reported that the hydrophilicity of NF membranes had a
strong dependence on the surface roughness, NF membranes
have a reasonably linear relationship between salt perme-
Aleman et al. [78] developed a mathematical model to ation and pure water permeability. The high hydrophilicity
predict the performance of NF membranes based on the of NF membranes plays a vital role for preventing organic
Nernst–Planck equation, Donnan equilibrium model, and fouling, which can reduce the cleaning frequency and extend
Gouy–Chapman theory. The model parameters were the per- the lifespan of the membrane. Hydrophilic membranes corre-
meability of pure water, pore size, and surface electrical spond to lower membrane fouling potential than hydrophobic
potential and were independent of the concentration and ones [104,105].
operating pressure. Roy et al. [80] modeled different NF
configurations using the Donnan–steric pore and dielectric 7.1 Effect of Pressure
exclusion model. Flat sheet and spiral wound membrane
configurations performance is similar at similar geometric Lin et al. [106] examined the effects of a constant flow
properties and under similar operating conditions. Nev- velocity of 2000 ppm for CaSO4 deionized water solution
ertheless, the computational time and complexity reduces at membrane feeds of 345–966 kPa. The feed water recov-
significantly. ery increased with the operating pressure. Because of this
relationship, the osmotic pressure and concentration polar-
ization at the surface increased; thus, the flux reduction was
7 Membrane Fouling sharp. The nucleation stage was considerably shorter at high
operating temperatures due to the high supersaturation. Lee
Fouling is one of the main problems in NF that shortens et al. [107,108] found that concentration polarization and
membrane lifetime. Fouling also requires membrane chem- surface crystallization increased as the operating pressure
ical cleaning, particularly if high concentrations of organic increased. Heterogeneous crystallization was favored at high
and inorganic components are filtered from the feed [85–88]. pressures and low velocities.
Fouling is caused by the adherence of inorganic and organic
materials to membrane surfaces and pores. Inorganic fouling 7.2 Effect of Feed Flow Rate
is due to the buildup of salt scalants on the membrane surface,
whereas organic fouling is caused by natural organic mat- The feed flow rate has a significant impact on crystal
ter, such as derivatives of humic acid [89–91]. Al-Amoudi et growth and scale precipitation, particularly in homogenous
al. [92,93] discussed organic and inorganic fouling, strategies crystallization. Vrijenhoek et al. [109] reported that foul-
to minimize fouling, membrane cleaning methods and the ing significantly decreased with increases in the cross-flow
effects of pressure, temperature, surface roughness, perme- velocity. Lee et al. [107,108] showed that at high feed flow
ate flux and pH. Organic fouling involves interrelationships rates, the dominant mechanism of membrane fouling was
between chemical and physical interaction. Inorganic fouling bulk crystallization rather than surface crystallization. At
is a result of scale formation. To avoid scaling, NF systems lower feed flow rates, the dominant mechanism was surface
should be operated at conditions lower than critical solubility crystallization. The relationship between deposited mass and
limits. Fouling can be reduced by membrane treatment, mem- flux decline was linear, which indicated that the flux drop was
brane modification, feed pretreatment and chemical cleaning. due to NF fouling. Therefore, the feed concentration at the

123
Arabian Journal for Science and Engineering

membrane wall is higher than the bulk concentration because is considerably shorter due to the greater degree of supersatu-
of concentration polarization at lower flow rates. ration. For some salts, the rejection decreases with increasing
pressure due to increased membrane polarization. Thus, for
7.3 Effect of Temperature these salts, rejection decreases at high pressures.
Temperature has a small effect on ion rejection. Increas-
Temperature has an influence on scale formation on NF mem- ing the temperature increases the permeate flux and ion
brane surfaces. For specific precipitates, the thermodynamic rejection (however, the Na and Cl ion rejection slightly
solubility product is a function of temperature [110]. Her et decrease). Temperature has an influence on scale formation
al. [111] concluded that calcium carbonate particles were on membrane surfaces. For a specific precipitate, the ther-
detected on the NF membrane surface at 20 ◦ C. Increas- modynamic solubility product is a function of temperature.
ing temperatures decrease the solubility product constants of Increasing the temperature decreases the solubility product
CaCO3 and CaSO4 , producing a homogenous crystallization. constants of CaCO3 and CaSO4 , resulting in a homogenous
In the bulk concentration, calcium carbonate precipitation crystallization. In the bulk concentration, calcium carbon-
occurs on the membrane surface due to high supersatura- ate precipitation on membrane surfaces occurs due to high
tion and high temperatures [107,108]. Andirstsos et al. [112] supersaturation at high temperatures.
investigated temperature effect due to the crystal structure of Salt rejection and permeate flux decrease with time before
CaCO3 . They reported that the deposits were calcite at 25 ◦ C chemical cleaning and decrease more sharply after chemical
and aragonite at 45 ◦ C. cleaning. At high operating pressures, the duration of the
nucleation stage is considerably shorter due to the greater
7.4 Effect of Surface Roughness degree of supersaturation, which increases membrane foul-
ing. At lower feed flow rates, the dominant mechanism is
Hobbs et al. [113] investigated the effect of surface roughness surface crystallization. At higher feed flow rates, the domi-
on fouling. A membrane was characterized using streaming nant mechanism of membrane fouling is bulk crystallization.
potential analysis (SPA) to measure the surface charge, AFM The relationship between deposited mass and flux decline is
to measure the surface roughness, and contact angle measure- linear, which indicates that flux decreases are due to mem-
ments to determine the hydrophobicity. Results showed that brane fouling. The surface foulant contents are primarily
membrane fouling increased with increasing surface rough- attributed to calcium sulfate and calcium carbonate. Increas-
ness. Vrijenhoek et al. [109] used AFM to show that more ing the surface roughness increases membrane fouling.
particles were deposited as the surface roughness increased; Future work should focus on developing more robust
this phenomenon resulted in a greater flux decline. membranes or techniques to control and reduce fouling.
However, more intensive and critical investigations of NF
applications are required to improve and optimize the effec-
8 Concluding Remarks tiveness of hybrid desalination plants. Additionally, contin-
ued NF membrane research would play a vital role in the
NF can be used to reject calcium sulfate and calcium carbon- development of seawater desalination technology. NF-based
ate. The process has the potential to eliminate scaling species pretreatments can provide the optimum feed water quality,
in seawater desalination plants. This elimination allows which makes the desalting process more energy efficient and
increasing TBT in thermal desalination processes, which in result in reduced costs of freshwater.
turn enables higher numbers of stages in the MSF or MED
processes and thus increases the GOR. The performances of Acknowledgements The authors acknowledge the support provided by
King Fahd University of Petroleum and Minerals to carry put this work.
NF membranes integrated with various types of desalination
technologies, including multistage flash, reverse osmosis,
forward osmosis, membrane distillation and ion-exchange
processes, have been reported. The operating pressures var-
ied from 2 to 20 bar, which is lower than those used in References
many typical RO processes. The major problem encountered
1. Wang, L.K.; Chen, J.P.; Hung, Y.-T.; Shammas, N.K.: Membrane
by these systems was fouling, which results in high energy and Desalination Technologies, vol. 13(9) (2011)
demand, high operation costs and reduced membrane life- 2. WHO: Total dissolved solids in drinking-water background doc-
times. ument for development of. Heal. criteria other Support. Inf., vol.
Increasing the operating pressure increases salt rejection 2, p. 8 (1996)
3. Voutchkov, N.: Desalination Engineering Planning and Design,
and permeate flux because of the contribution of convec- vol. 9 (2013)
tion at high pressures due to high water flux. However, at 4. El-Dessouky, H.T.; Ettouney, H.M.: Fundamentals of Salt Water
extremely high pressures (e.g., 34 bar), the nucleation stage Desalination, vol. 1 (2002)

123
 Arabian Journal for Science and Engineering

5. Cotruvo, J.; Voutchkov, N.; Fawell, J.; Payment, P.; Cunliffe, D.; 25. Liu, J.; Yuan, J.; Xie, L.; Ji, Z.: Exergy analysis of dual-stage
Lattemann, S.: Desalination Technology Health and Environmen- nanofiltration seawater desalination. Energy 62, 248–254 (2013)
tal Impacts (2010) 26. AlTaee, A.; Sharif, A.O.: Alternative design to dual stage NF
6. AlTaee, A.; Mabrouk, A.; Bourouni, K.: A novel forward osmo- seawater desalination using high rejection brackish water mem-
sis membrane pretreatment of seawater for thermal desalination branes. Desalination 273(2–3), 391–397 (2011)
processes. Desalination 326, 19–29 (2013) 27. Vuong, D.X.: Two stage nanofiltration seawater desalination sys-
7. Hilal, N.; Al-Zoubi, H.; Darwish, N.A.; Mohammad, A.W.; Abu tem. United State Patent 7,144,511 B2 (December 2006)
Arabi, M.: A comprehensive review of nanofiltration membranes: 28. Al-Rawajfeh, A.: Influence of nanofiltration pretreatment on scale
treatment, pretreatment, modelling, and atomic force microscopy. deposition in multi-stage flash thermal desalination plants. Therm.
Desalination 170, 281–308 (2004) Sci. 15(1), 55–65 (2011)
8. Mohammad, A.W.; Teow, Y.H.; Ang, W.L.; Chung, Y.T.; Oatley- 29. Song, Y.; Su, B.; Gao, X.; Gao, C.: The performance of polyamide
Radcliffe, D.L.; Hilal, N.: Nanofiltration membranes review: nano fi ltration membrane for long-term operation in an integrated
recent advances and future prospects. Desalination 356, 226–254 membrane seawater pretreatment system. Desalination 296, 30–
(2015) 36 (2012)
9. Van der Bruggen, B.; Mänttäri, M.; Nyström, M.: Drawbacks of 30. Kaya, C.; Sert, G.; Kabay, N.; Arda, M.; Yüksel, M.; Ege-
applying nanofiltration and how to avoid them: a review. Sep. men, Ö.: Pre-treatment with nanofiltration (NF) in seawater
Purif. Technol. 63, 251–263 (2008) desalination—preliminary integrated membrane tests in Urla,
10. Luo, J.; Wan, Y.: Effects of pH and salt on nanofiltration—a critical Turkey. Desalination 369, 10–17 (2015)
review. J. Memb. Sci. 438, 18–28 (2013) 31. Song, Y.; Gao, X.; Gao, C.: Evaluation of scaling potential in a
11. Zhou, D.; Zhu, L.; Fu, Y.; Zhu, M.; Xue, L.: Development of pilot-scale NF-SWRO integrated seawater desalination system. J.
lower cost seawater desalination processes using nano filtration Memb. Sci. 443, 201–209 (2013)
technologies—a review. Desalination 376(1219), 109–116 (2015) 32. Awerbuch, L.: Water desalination process using ion selective
12. Llenas, L.; Martínez-Lladó, X.; Yaroshchuk, A.; Rovira, M.; membranes. United States Patent, 6,998,053 B2 (February 2006)
de Pablo, J.: Nanofiltration as pretreatment for scale prevention 33. Al-Sofi, M.K.; Hassan, A.M.; Mustafa, G.M.; Dalvi, G.I.; Kither,
in seawater reverse osmosis desalination. Desalin. Water Treat. M.N.M.: Nanofiltration as a means of achieving higher TBT of
36(September), 310–318 (2011) >120◦ C in MSF. Desalination 118(September 1998), 123–129
13. Llenas, L.; Ribera, G.; Martínez-Lladó, X.; Rovira, M.; de Pablo, (1998)
J.: Selection of nanofiltration membranes as pretreatment for scal- 34. Hassan, A.M.; Al-Sofi, M.K.; Farooque, A.M.; Dalvi, A.G.I.;
ing prevention in SWRO using real seawater. Desalin. Water Treat. Jamaluddin, A.T.M.; Kither, N.M.; Al-Amoudi, A.S.; Al-Tisan,
51(September), 930–935 (2013) I.R.: A nano filtration (Nf) membrane pretreatment of swro feed
14. Kurihara, M.; Sasaki, T.; Nakatsuji, K.; Kimura, M.; Henmi, M.: and Msf make-up. Desalination 118, 35–51 (1998)
Low pressure SWRO membrane for desalination in the mega-ton 35. Hassan, A.M.; Al-Sofi, M.K.; Al-Amoudi, A.S.; Jamaluddin,
water system. Desalination 368, 135–139 (2015) A.T.M.; Farooque, A.M.; Rowaili, A.; Dalvi, A.G.I.; Kither, N.M.;
15. Hilal, N.; Al-Zoubi, H.; Mohammad, A.W.; Darwish, N.A.: Mustafa, G.M.; Al-Tisan, I.R.: A new approach to membrane
Nanofiltration of highly concentrated salt solutions up to seawater and thermal seawater desalination processes using nanofiltration
salinity. Desalination 184(May), 315–326 (2005) membranes (Part 1). Desalination 118(1998), 35–51 (1998)
16. Ryabchikov, B.E.; Panteleev, A.A.; Gladush, M.G.: Performance 36. Hamed, O.A.; Mustafa, K.B.; Ghulam, M.; Khalid Al-Shail, A.;
testing of seawater desalination by nanofiltration. Pet. Chem. Mardouf, A.-R.; Al-Washmi, H.; Hassan, A.M.; Al-Sulami, S.:
52(7), 465–474 (2012) A nanofiltration membrane pretreatment of SWRO feed & MSF
17. Pérez-González, A.; Ibáñez, R.; Gómez, P.; Urtiaga, A.M.; Ortiz, make-up. Saline Water Conversion Corporation (2002)
I.; Irabien, J.A.: Nanofiltration separation of polyvalent and mono- 37. Hamed, O.A.; Hassan, A.M.; Al-Shail, K.; Farooque, M.A.: Per-
valent anions in desalination brines. J. Memb. Sci. 473, 16–27 formance analysis of a trihybrid NF/RO/MSF desalination plant.
(2015) Desalin. Water Treat. 1(October), 215–222 (2009)
18. Izadpanah, A.A.; Javidnia, A.: The ability of a nanofiltration mem- 38. Hassan, A.M.: Fully integrated NF-thermal seawater desali-
brane to remove hardness and ions from diluted seawater. Water nation process and equipment. European Patent Application,
4, 283–294 (2012) 1,614,660,A1 (2006)
19. Harrison, C.J.; Le Gouellec, Y.A.; Cheng, R.C.; Childress, A.E.: 39. Hamed, O.A.: Overview of hybrid desalination systems—current
Bench-scale testing of nanofiltration for seawater desalination. J. status and future prospects. Desalination 186(1–3), 207–214
Environ. Eng. 133(November), 1004–1014 (2007) (2005)
20. Liu, J.; Xie, L.; Wang, Z.; Yuan, J.: Dual-stage nanofiltration 40. http://www.dow.com/en-us/water-and-process-solutions/
seawater desalination: water quality, scaling and energy consump- products#q=%20. Dow Chemicals Company
tion. Desalin. Water Treat. 52(1–3), 134–144 (2014) 41. http://www.kochmembrane.com/. Koch Membrane
21. Kaewsuk, J.; Lee, D.Y.; Lee, T.S.; Seo, G.T.: Effect of ion com- 42. Hydranautics NITTO DENKO Company. http://www.
position on nanofiltration rejection for desalination pretreatment. membranes.com/
Desalin. Water Treat. 43(1–3), 260–266 (2012) 43. http://www.alfalaval.com/. Alfa Laval company
22. Bowen, W.R.; Mohammad, A.W.: Diafiltration by nanofiltration: 44. Company, C.: Product Specification Sheet/Model NE4040-90
prediction and optimization RID A-3578-2011. AIChE J. 44(8), 45. Su, B.; Wu, T.; Li, Z.; Cong, X.; Gao, X.; Gao, C.: Pilot study of
1799–1812 (1998) seawater nanofiltration softening technology based on integrated
23. Mänttäri, M.; Pihlajamäki, A.; Kaipainen, E.; Nyström, M.: Effect membrane system. Desalination 368, 193–201 (2015)
of temperature and membrane pre-treatment by pressure on the 46. Al-Rawajfeh, A.E.: Hybrid salts precipitation-nanofiltration pre-
filtration properties of nanofiltration membranes. Desalination treatment of MSF and RO seawater desalination feed. Membr.
145(1–3), 81–86 (2002) Water Treat. 3(4), 253–266 (2012)
24. Nilsson, M.; Trägårdh, G.; Östergren, K.: The influence of pH, 47. Al-Rawajfeh, A.E.; Fath, H.E.S.: Integrated salts precipitation
salt and temperature on nanofiltration performance. J. Memb. Sci. and nano-filtration as pretreatment of multistage flash desalina-
312(1–2), 97–106 (2008) tion system and nano-filtration as pretreatment of multistage flash
desalination. Heat Transf. Eng. 33, 272–279 (2012)

123
Arabian Journal for Science and Engineering

48. Hilal, N.; Kochkodan, V.; Al, H.; Mandale, S.; Al-Jlil, S.A.: A 67. Altaee, A.; Zaragoza, G.: A conceptual design of low fouling and
combined ion exchange-nano fi ltration process for water desali- high recovery FO–MSF desalination plant. Desalination 343, 2–7
nation: I. Sulphate-chloride ion-exchange in saline solutions. (2014)
Desalination 363, 44–50 (2015) 68. Altaee, A.; Mabrouk, A.; Bourouni, K.; Palenzuela, P.: Forward
49. Hilal, N.; Kochkodan, V.; Al, H.; Johnson, D.: A combined ion osmosis pretreatment of seawater to thermal desalination: High
exchange-nano fi ltration process for water desalination: II. Mem- temperature FO–MSF/MED hybrid system. Desalination 339(1),
brane selection. Desalination 363, 51–57 (2015) 18–25 (2014)
50. Hilal, N.; Kochkodan, V.; Al, H.; Mandale, S.; Al-Jlil, S.A.: A 69. Darwish, M.; Hassan, A.; Mabrouk, A.N.; Abdulrahim, H.; Sharif,
combined ion exchange-nano fi ltration process for water desali- A.: Viability of integrating forward osmosis (FO) as pretreat-
nation: III. Pilot scale studies. Desalination 363, 58–63 (2015) ment for existing MSF desalting unit. Desalin. Water Treat.
51. Sarkar, S.; SenGupta, A.K.: A new hybrid ion exchange- 3994(November), 1–11 (2015)
nanofiltration (HIX-NF) separation process for energy-efficient 70. Al-Rawajfeh, A.E.: Prevention of scale deposits in desalination:
desalination: process concept and laboratory evaluation. J. Memb. pre-treatment of feed seawater by salts precipitation. Tafila Tech-
Sci. 324(1–2), 76–84 (2008) nical University (TTU), pp. 1–4 (2011)
52. Sarkar, S.; SenGupta, A.K.: A hybrid ion exchange-nanofiltration 71. Ayoub, G.M.; Zayyat, R.M.; Al-hindi, M.: Precipitation soften-
(HIX-NF) process for energy efficient desalination of brack- ing? a pretreatment process for seawater desalination. Env. Sci.
ish/seawater. Water Sci. Technol. Water Supply 9(4), 369–377 Pollut. Res. 21, 2876–2887 (2014)
(2009) 72. Constantz, B.R.; Farsad, K.; Fernandez, M.: Desalination meth-
53. SenGupta, A.K.; Sarkar, S.: United States Patent, US 7,901,577 ods and systems that include carbonate compound precipitation.
B2 (2011) United States Patent, 7,931,809 B2 (April 2011)
54. Tan, C.H.; Ng, H.Y.: A novel hybrid forward osmosis- 73. Ahmad, A.L.; Chong, M.F.; Bhatia, S.: Mathematical modeling
nanofiltration (FO–NF) process for seawater desalination: draw and simulation of the multiple solutes system for nanofiltration
solution selection and system configuration. Desalin. Water Treat. process. J. Memb. Sci. 253(1–2), 103–115 (2005)
13(1–3), 356–361 (2010) 74. Murthy, Z.V.P.; Gupta, S.K.: Estimation of mass transfer coeffi-
55. Zhao, S.; Zou, L.; Mulcahy, D.: Brackish water desalination by a cient using a combined nonlinear membrane transport and film
hybrid forward osmosis-nanofiltration system using divalent draw theory model. Desalination 109(1), 39–49 (1997)
solute. Desalination 284, 175–181 (2012) 75. Nora’aini, A.; Wahab, M.A.: The development of predictive mod-
56. Phuntsho, S.; Hong, S.; Elimelech, M.; Shon, H.K.: Forward eling of nanofiltration membrane properties? a review. Reg. Symp.
osmosis desalination of brackish groundwater: meeting water Membr. Sci. Technol., 15 (2004)
quality requirements for fertigation by integrating nanofiltration. 76. Oatley-Radcliffe, D.L.; Williams, S.R.; Barrow, M.S.; Williams,
J. Memb. Sci. 436, 1–15 (2013) P.M.: Critical appraisal of current nanofiltration modelling strate-
57. Altaee, A.; Hilal, N.: High recovery rate NF–FO–RO hybrid sys- gies for seawater desalination and further insights on dielectric
tem for inland brackish water treatment. Desalination 363, 19–25 exclusion. Desalination 343, 154–161 (2014)
(2015) 77. Geraldes, V.; Brites Alves, A.M.: Computer program for simula-
58. Al-hajouri, A.A.; Al-Amoudi, A.S.; Farooque, A.M.; Al-Amoudi, tion of mass transport in nanofiltration membranes. J. Memb. Sci.
A.S.; Farooque, A.M.: Long term experience in the operation 321(2), 172–182 (2008)
of nanofiltration pretreatment unit for seawater desalination at 78. Garcia-Aleman, J.; Dickson, J.M.: Mathematical modeling of
SWCC SWRO plant. Desalin. Water Treat. 3994, 1861–1873 nanofiltration membranes with mixed electrolyte solutions. J.
(2015) Memb. Sci. 235(1–2), 1–13 (2004)
59. ROSA Software | Dow Water & Process Solutions (Online). http:// 79. Mohammad, A.W.; Pei, L.Y.; Kadhum, A.A.H.: Characteriza-
www.dow.com/en-us/water-and-process-solutions/resources/ tion and identification of rejection mechanisms in nanofiltration
design-software/rosa-software. Accessed: 24 Feb 2017 membranes using extended Nernst–Planck model. Clean Technol.
60. Drioli, E.; Laganà, F.; Criscuoli, A.; Barbieri, G.: Integrated Environ. Policy 4, 151–156 (2002)
membrane operations in desalination processes. Desalination 122, 80. Roy, Y.; Sharqawy, M.H.; Lienhard, J.H.: Modeling of flat-sheet
141–145 (1999) and spiral-wound nanofiltration configurations and its application
61. El-Zanati, E.; El-Khatib, K.M.: Integrated membrane-based in seawater nanofiltration. J. Memb. Sci. 493, 360–372 (2015)
desalination system. Desalination 205(May 2006), 15–25 (2007) 81. Bandini, S.; Vezzani, D.: Nanofiltration modeling: the role of
62. Mabrouk, A.N.A.; Fath, H.E.S.: Experimental study of high- dielectric exclusion in membrane characterization. Chem. Eng.
performance hybrid NF–MSF desalination pilot test unit driven by Sci. 58(15), 3303–3326 (2003)
renewable energy. Desalin. Water Treat. 51(37–39), 6895–6904 82. Szymczyk, A.; Labbez, C.; Fievet, P.; Vidonne, A.; Foissy, A.;
(2013) Pagetti, J.: Contribution of convection, diffusion and migration
63. Al-Rawajfeh, A.E.: Nanofiltration pretreatment as CO2 deaera- to electrolyte transport through nanofiltration membranes. Adv.
tor of desalination feed: CO2 release reduction in MSF distillers. Colloid Interface Sci. 103(1), 77–94 (2003)
Desalination 380, 12–17 (2016) 83. Sidek, N.M.; Fauziah, S.; Draman, S.; Abdullah, I.: Prediction of
64. Turek, M.; Chorazewska, M.: Nanofiltration process for seawa- nanofiltration membrane performance? analysis of ion transport
ter desalination-salt production integrated system. Desalin. Water mechanisms. Asian Res. Publ. Netw. 10(1), 337–342 (2015)
Treat. 7(November), 178–181 (2009). ST—Nanofiltration process 84. Fang, J.; Deng, B.: Rejection and modeling of arsenate by
for seawater nanofiltration: contributions of convection, diffusion and electro-
65. Curcio, E.; Ji, X.; Quazi, A.M.; Barghi, S.; Di Profio, G.; migration to arsenic transport. J. Memb. Sci. 453, 42–51 (2014)
Fontananova, E.; Macleod, T.; Drioli, E.: Hybrid nanofiltration- 85. Lapointe, J.F.; Gauthier, S.F.; Pouliot, Y.; Bouchard, C.: Foul-
membrane crystallization system for the treatment of sulfate ing of a nanofiltration membrane by a Beta-lactoglobulin tryptic
wastes. J. Memb. Sci. 360(1–2), 493–498 (2010) hydrolysate: impact on the membrane sieving and electrostatic
66. Drioli, E.; Curcio, E.; Di Profio, G.; Macedonio, F.; Criscuoli, A.: properties. J. Memb. Sci. 253(1–2), 89–102 (2005)
Integrating membrane contactors technology and pressure-driven 86. Thorsen, T.: Concentration polarisation by natural organic matter
membrane operations for seawater desalination. Chem. Eng. Res. (NOM) in NF and UF. J. Memb. Sci. 233(1–2), 79–91 (2004)
Des. 84(3), 209–220 (2006)

123
 Arabian Journal for Science and Engineering

87. Long, F.; Zhu, A.; Wang, X.L.; Zhu, W.P.: Membrane flux and 101. Lin, J.; Tang, C.Y.; Huang, C.; Tang, Y.P.; Ye, W.; Li, J.; Shen, J.;
CaCO3 crystallization in the unstirred dead-end nanofiltration of Van den Broeck, R.; Van Impe, J.; Volodin, A.; Van Haesendonck,
magnetic solution. Desalination 186(1–3), 243–254 (2005) C.; Sotto, A.; Luis, P.; Van der Bruggen, B.: A comprehen-
88. Beyer, F.; Rietman, B.M.; Zwijnenburg, A.; van den Brink, P.; sive physico-chemical characterization of superhydrophilic loose
Vrouwenvelder, J.S.; Jarzembowska, M.; Laurinonyte, J.; Stams, nanofiltration membranes. J. Memb. Sci. 501, 1–14 (2016)
A.J.M.; Plugge, C.M.: Long-term performance and fouling anal- 102. Lin, J.; Tang, C.Y.; Ye, W.; Sun, S.P.; Hamdan, S.H.; Volodin, A.;
ysis of full-scale direct nanofiltration (NF) installations treating Van Haesendonck, C.; Sotto, A.; Luis, P.; Van der Bruggen, B.:
anoxic groundwater. J. Memb. Sci. 468, 339–348 (2014) Unraveling flux behavior of superhydrophilic loose nanofiltration
89. Farooque, A.M.; Al-amoudi, A.S.; Hassan, A.M.: Chemical clean- membranes during textile wastewater treatment. J. Memb. Sci.
ing experiments for performance restoration of NF membranes 493, 690–702 (2015)
operated on seawater feed 1. IDA Conference March (2002) 103. Lin, J.; Ye, W.; Zeng, H.; Yang, H.; Shen, J.; Darvishmanesh, S.;
90. Flora, J.R.V.: Stochastic approach to modeling surface fouling of Luis, P.; Sotto, A.; Van der Bruggen, B.: Fractionation of direct
ultrafiltration membranes. J. Memb. Sci. 76, 85–88 (1993) dyes and salts in aqueous solution using loose nanofiltration mem-
91. Al-Amoudi, A.S.; Farooque, A.M.: Performance restoration and branes. J. Memb. Sci. 477, 183–193 (2015)
autopsy of NF membranes used in seawater pretreatment. Desali- 104. Zhang, M.; Qiang Liao, B.; Zhou, X.; He, Y.; Hong, H.; Lin,
nation 178(1–3), 261–271 (2005). SPEC. ISS H.; Chen, J.: Effects of hydrophilicity/hydrophobicity of mem-
92. Al-Amoudi, A.; Lovitt, R.W.: Fouling strategies and the cleaning brane on membrane fouling in a submerged membrane bioreactor.
system of NF membranes and factors affecting cleaning efficiency. Bioresour. Technol. 175, 59–67 (2015)
J. Memb. Sci. 303, 4–28 (2007) 105. Kim, D.S.; Kang, J.S.; Lee, Y.M.: The influence of membrane
93. Al-Amoudi, A.S.: Factors affecting natural organic matter (NOM) surface properties on fouling in a membrane bioreactor for
and scaling fouling in NF membranes: a review. Desalination wastewater treatment. Sep. Sci. Technol. 39(4), 833–854 (2005)
259(1–3), 1–10 (2010) 106. Lin, C.-J.; Shirazi, S.; Rao, P.; Agarwal, S.: Effects of operational
94. Matin, A.; Khan, Z.; Zaidi, S.M.J.; Boyce, M.C.: Biofouling in parameters on cake formation of CaSO4 in nanofiltration. Water
reverse osmosis membranes for seawater desalination: phenom- Res. 40(4), 806–816 (2006)
ena and prevention. Desalination 281(1), 1–16 (2011) 107. Lee, S.; Kim, J.; Lee, C.H.: Analysis of CaSO4 scale forma-
95. Vrouwenvelder, J.S.; Manolarakis, S.A.; van der Hoek, J.P.; van tion mechanism in various nanofiltration modules. J. Memb. Sci.
Paassen, J.A.M.; van der Meer, W.G.J.; van Agtmaal, J.M.C.; 163(1), 63–74 (1999)
Prummel, H.D.M.; Kruithof, J.C.; van Loosdrecht, M.C.M.: 108. Lee, S.; Lee, C.H.: Effect of operating conditions on CaSO4 scale
Quantitative biofouling diagnosis in full scale nanofiltration and formation mechanism in nanofiltration for water softening. Water
reverse osmosis installations. Water Res. 42(19), 4856–4868 Res. 34(15), 3854–3866 (2000)
(2008) 109. Vrijenhoek, E.M.; Hong, S.; Elimelech, M.: Influence of mem-
96. Kang, S.T.; Subramani, A.; Hoek, E.M.V.; Deshusses, M.A.; Mat- brane surface properties on initial rate of colloidal fouling of
sumoto, M.R.: Direct observation of biofouling in cross-flow reverse osmosis and nanofiltration membranes. J. Memb. Sci.
microfiltration: mechanisms of deposition and release. J. Memb. 188(1), 115–128 (2001)
Sci. 244(1–2), 151–165 (2004) 110. Sheikholeslami, R.: Mixed salts—scaling limits and propensity.
97. Schneider, R.P.: Conditioning film-induced modification of sub- Desalination 154(2), 117–127 (2003)
stratum physicochemistry—analysis by contact angles. J. Colloid 111. Her, N.; Amy, G.; Jarusutthirak, C.: Seasonal variations of nanofil-
Interface Sci. 182(1), 204–213 (1996) tration (NF) foulants: identification and control. Desalination
98. Sadr Ghayeni, S.B.; Beatson, P.J.; Schneider, R.P.; Fane, A.G.: 132(1–3), 143–160 (2000)
Adhesion of waste water bacteria to reverse osmosis membranes. 112. Andritsos, N.; Kontopoulou, M.; Karabelas, A.J.; Koutsoukos,
J. Memb. Sci. 138(1), 29–42 (1998) P.G.: Calcium carbonate deposit formation under isothermal con-
99. Huang, Q.; Ma, W.: A model of estimating scaling potential in ditions. Can. J. Chem. Eng. 74(6), 911–919 (1996)
reverse osmosis and nanofiltration systems. Desalination 288, 40– 113. Hobbs, C.; Hong, S.; Taylor, J.: Effect of surface roughness on
46 (2012) fouling of RO and NF membranes during filtration of a high
100. Faridirad, F.; Zourmand, Z.; Kasiri, N.; Kazemi Moghaddam, M.; organic surficial groundwater. J. Water Supply Res. Technol. 55,
Mohammadi, T.: Modeling of suspension fouling in nanofiltration. 559–570 (2006)
Desalination 346, 80–90 (2014)

123