ELSEVIER

Desalination 179 (2005) 315-322

DESALINATION
www.elsevier.com/locate/desal
Us e o f nanofi l trati on for partial sof t eni ng o f very hard water
Sa mi r a Ghi z e l l a oui a*, Ai s s a Chi ba ni a, Souhe i l a Ghi z e l l a oui b
aD@artement de chimie, bD@artement de chimie Industrielle, Universit~ de Mentouri de Constantine,
Route de Ain E1 Bey, 25000 Constantine, Alg&ie
TeL~Fax: +213 (31) 61 43 44; email: gsamira@yahoo.eom
Received 30 September 2004; accepted 22 November 2004
Abstract
Nanofiltration, a recent membrane technique, was used for the softening of very hard water. Our study was based
on the elimination of one part of the temporary hardness of Hamma water which provides Constantine City with
drinking water. Two techniques were used in order to obtain a partial softening. The first is related to weak pressures
(0.5, 1,2 bar) whereas the second is based on relatively high pressures (4-16 bar) and which is applied on feed water.
However, we should mention that at the end of this experience, 50% of filtered water was mixed with 50% of hard
water. The results show that this membrane technique is of great interest in the process of eliminating temporary
hardness of water, which depends mainly on the applied pressure and the tangential flow rate.
Keywords: Drinking water; Hardness; Calcium carbonate; Partial softening; Nanofiltration
1. Introducti on
Waters f r om Hamma wi t h a hardness of
60 French degrees (600 ppm in CaCO3) must be
treated bef or e any use due to the large quantities
of cal ci um carbonat e (scaling). Under these con-
ditions, the consequences o f t hese deposits are of
t hree types: hydraul i c, t hermal and mechani cal .
Accor di ng to Khalil et al. [ 1 ], to fi ght against
scaling, it is necessar y to use chemi cal or physi cal
*Corresponding author.
processes. Chemi cal met hods, based on the use of
scaling inhibitors such as car boxyl at e or phos-
phonat es, are effect i ve but not possi bl e in the case
of wat er o f distribution. Several physi cal pro-
cesses such as mi cr o- el ect r ol ysi s of water, the
generat i on o f el ect romagnet i c waves or the drain-
age duct t hr ough per manent magnet s have been
gi ven, for exampl e, by Rosset et al. [2], Rosset
[3], Maydat chevs ky [4], and Cl eret and L6di on
[5]. All of these studies as wel l as others led to a
search for good met hods of i nhi bi t i on of the
Presented at the conference on Membranes in Drinking and Industrial Water Production, L "Aquila, Italy, 15-17 November
2004. Organized by the European Desalination Society.
0011-9164/05/$- See front matter 2005 Elsevier B.V. All rights reserved
doi: 10.1016/j.desal.2004.11.077
316 S. Ghizellaoui et al. / Desalination 179 (2005) 315-322
f ur r i ng capaci t y o f wat er wh i c h r emai ns appl i -
cabl e for dr i nki ng wat er.
The us e o f t he nanof i l t r at i on (NF) me mb r a n e
t echni que has r ecent l y b e c o me an i mpor t a nt
t e c hnol ogy f or separ at i on. It f unc t i ons wi t hout
t he addi t i on o f chemi cal s , wi t h a r el at i vel y l ow
c ons umpt i on o f ener gy. The me mb r a n e s are us ed
mor e of t en f or t he t r e a t me nt o f subsoi l wat er s
[6,7], f or wa s t e wa t e r and f i nal l y f or sever al
i ndust ri al sect or s (t ext i l es, phar maceut i cal s ,
chemi cal , etc. ) [8,9].
NF was us ed her e, on one ha nd, t o ens ur e
we a k r et ent i on o f mo n o v a l e n t i ons (HCO3, C1-)
but good r et ent i on o f t he di va l e nt i ons ( Ca 2,
Mg 2+, SO42-); and on t he ot her ha nd, f or t he
el i mi nat i on o f par t o f t he t e mpor a r y har dnes s o f
Ha mma wat er, wh i c h pr ovi de s t he Ci t y o f Con-
st ant i ne wi t h dr i nki ng wat er. We wo r k e d wi t h
l ow pr es s ur es (0.5, 1, 2 bar). I n addi t i on, we
exer t ed r el at i vel y s t r ong pr es s ur es ( 4 - 1 6 bar) on
wat er wh i c h was mi x e d at t he e nd o f t he
t r eat ment wi t h 50% ha r d wat er.
Table 1
Harnma water analysis
Parameter Value
Temperature, C 31-33
pH 6.95-7.94
EC, mS/cm 0.9-1.30
TOC, mg/L 3.84-7.5
CO2, mg/L 16.21-37.05
HCO~, mg/L 320-432
el -, mg/L 128-150
SO~-, mg/L 120-212
PO 3-, mg/L 0.006-0.02
NO3, mg/L 3.90-7
NO~, mg/L
+
NH4, mg/L
TH, mg/L CaCO 3 475-600
Ca 2+, mg/L 124-131
Mg 2+, mg/L 30.48-43.2
Na +, mg/L 84-115
K +, mg/L 2.22-13
2. Physi cochemi cai characteri sti cs of Hamma
water
The anal ysi s gi ve n i n Tabl e 1 s hows t hat
Ha mma wat er s have a l ow or gani c ma t t e r c ont e nt
( TOC: 3. 84- 7. 5 mg/ L) , a r at her hi gh sal t c ont e nt
(EC: 0. 9- 1. 30 mS/ c m) and an appr eci abl e har d-
ness (TH: 4 7 5 - 6 0 0 mg / L CaCO3).
3. Analysis of deposits
3.1. X-ray analysis
The depos i t s wer e anal yzed by X- r ay di f f r ac-
t i on. The scal e was f o u n d t o cr yst al l i ze i n t he
f or m o f a mi xt ur e o f cal ci t e a nd ar agoni t e.
3.2. IR spectrum
The I R spect r u, (Fi g. 1) o f t he depos i t i s s ued
f r om t he Ha mma wat er pr es ent s at 25 C sever al
bands o f abs or pt i ons :
3417. 8 ~ O H o f w a t e r
2514. 7 ~ or gani c i mpur i t i es r el at ed t o usual
pol l ut i on ( e qui pme nt , bui l di ngs , oper at or s)
1796. 9 ~ char act er i st i c o f cal ci t e
1429. 9 -~ CO 3 + s houl der , whi c h r epr es ent s a
mi xt ur e o f cal ci t e and ar agoni t e
500 and 1000 ~ a l u mi n o si l i cat es, pr obabl y
c o mi n g f r om t r aces o f cl ay
875. 9 and 848 ~ char act er i st i c o f cal ci t e
321. 8 ~ char act er i st i c o f cal ci t e
4. Experi ment al
4.1. Apparat us
The pi l ot s ys t e m us e d wa s a c omme r c i a l 12 L
Mi l l i por e Pr os cal e modul e . Fi g. 2 s hows a sche-
mat i c di a gr a m o f t he e xpe r i me nt a l s ys t e m t hat
was e qui ppe d wi t h a Na n o ma x 50 c ompos i t e
spi ral p o l y a mi d e me mb r a n e . The me mb r a n e
S. Ghi zel l aoui et al. / Desal i nati on 179 (2005) 315-322 317
2 , 4 0
2 . 2 _
2 . 0
1 . 8 . .
1 , 6 ....
1 . 4 :
L 2 J
1. 0
0 . 8
0 . 6
0 . 4 2 5 1 4 . 7
i796.9
0.2. . . . . . . . . . . . . . . . . . . . . . . . /
0 . 0 4 i ...........................................................................................................................................................................................
4 0 0 0 . 0 3 0 0 0 2 0 0 0
1 4 Y , $ . 6
' i
I
i
i
/
~ 7 5 . 9
J
[ 7 1 _ ' : 1 . 7
? 6 n . o
:i
3 2 ] . 8
i i
e
' l l 6 I . Q
" , ,
..................... . . . . . . . . . . . . . . J . . . . . . . . . . . . . . . . . . . . . . . .
.............................................................................................. 7 ...........................................................................................
1 5 0 0 1090 500
c m - I
200.0
Fig. 1. IR spectrum o f the deposit obtained at 25C.
4 7 8
c>r<
j I
Fig. 2. Schematic diagram o f the nanofiltration pilot. 1 tank, 2 feed pump, 3 and 4 pressure regulation, valves, 5 nanofiltra-
tion module, 6 permeate recirculation, 7 retentate recirculation, 8 heat exchanger. D, f l ow meter; P, pressure sensor;
T, temperature sensor.
MWCO wa s 3 5 0 Da l t o n and t he f i l t er s urf ace
area wa s 0. 37 m 2. Al l t he e x pe r i me nt s we r e
carri ed out at a c ons t ant f e e d c o mp o s i t i o n by
r e c y c l i n g bot h t he ret ent at e and t he pe r me at e t o
t he f e e d tank.
318 S. Ghizellaoui et al. / Desalination 179 (2005) 315-322
4.2. Procedure
A 4-L vol ume of Hamma raw water was used
in every NF test. Samples of the retentate and the
permeate were taken for analysis after 60 min of
filtration. Following each experiment, the mem-
brane was washed with a solution of hydrochloric
acid at a pH 2 and then rinsed with distilled water
until a conductivity of about 1 #S/ cm of the
flushing water was reached.
4.3. Analytical determinations
Determination of the anions (Cl-, SO~-) was
carried out by chromatographic analysis (Waters
501) at a flow rate of 1.3 mL/ mi n using a con-
ductimetric detector (Waters 431) and an anion
column (Shodex 1-524A). One liter ofel uant con-
tained 0.207 g of para-hydroxybenzoic acid,
5 mL of methanol and sufficient ethanolamine to
adjust the solution pH to 8.
The bicarbonates were analyzed with a Dohr-
mann-Xertex DC80 apparatus. The concentration
of the cations (Ca 2+, Mg 2+) was determined by
atomic absorption using a Varian 1275 apparatus
with an air-acetylene flame.
5. Results
The NF treated part at low pressures has been
recently presented [10] and is used here only for
an eventual comparison with the behavior of NF
at high pressure.
5.1. Influence of low transmembrane pressure on
the retention of the ions
Runs were carried out at operating pressures
of 0.5, 1 and 2 bar. These pressures are suitable
for achieving partial softening, leaving sufficient
minerals in the drinking water. Water temperature
was maintained at 20C t hroughout the run.
Table 2 summarizes mi ni mum and maxi mum
values of all chemical species analyzed in the
Table 2
Composition of feed and permeate water
F e e d Pe r me a t e Retention
water (%)
Conductivity 1010-1032 725-813 18-26
(#S/cm at 20C)
CI-, mg/L 129-130 121- 126 0.75-2
SO 2-, mg/L 1 3 1 - 1 3 2 7.9-9.2 93-94
HCO3, mg/L 408-466 298- 365 10-30
Ca >, mg/L 1 2 0 - 1 2 5 78-89 24-34
Mg 2, mg/L 36-37 22-27 23-40
pH 7.30-7.58 7.72-8.19 - -
feed, retentate and permeate respectively after
60 min of filtration Figs. 3 and 4 show the effect
of pressure on retention of the Ca > and Mg 2+
cations and of the HCO; , C1- and SO4 > anions.
The retention of both cations and anions increases
with the applied pressure.
The retention of Ca > ions was found to be
lower than that of Mg >. This may be explained
by the smaller ionic radius of Mg >, the value of
which ranges from 0.74A. to 1.04/~. The higher
solvation energy of Mg 2+ ions acts to augment
their retention.
In the case of anions, retention of the sulfate
ion was very high, but total elimination (100%)
was not obtained because of the relatively high
sulfate concentration in the water. Bicarbonate
ions were better retained than chloride ions.
5.2. Influence transmembrane pressure on f l ow
The tests were carried out by maintaining the
rate of feed constant, temperature at 20C and by
varying the pressure from 4 to 16 bar. The
evolution of flow according to the transmembrane
pressure (TMP) is given in Fig. 5. It should be
noted that when the pressure increases, variation
is observed between the line relating to pure
water and that corresponding to Hamma water.
This variation is due to the accumulation of the
S. Ghizellaoui et al. / Desalination 179 (2005) 315-322 319
30
~ 2 0
O
'Z=
e-
ca
~ 10
Ca 2~
--o-- Mg 2~ - . . . . . . j l ,
0
0 . 4 o ; s 1 .'2
P (bar)
Fig. 3. Percent retention of cations.
i
116 2.0
110
O
c a
100
90
80
70
60
50
40
30
20
10
0
0.4
A - - -A-
. . . . . . . . . . . . . . . . . . . . . . . i .
....... o.. CI-
-i -- HCO~
.... .. SO~"
018 1.2 ' 1.'6
P (bar)
Fig. 4. Percent retention of anions.
210
ions on the membrane at the origin o f concen-
tration polarization. Thi s gradient o f concentra-
tion i nvol ves additional resistance through the
membrane, whi ch is in addition due to the
osmotic pressure.
Th i s o b s e r v a t i o n wa s a l s o ma d e b y Xu a n d
Spencer [ 11 ] and Boucar d in Paugam [ 12]. Whi l e
wor ki ng wi t h concent r at ed salts, t hese aut hors
not ed a st rong di f f er ence bet ween the sl ope of t he
ri ght -hand si de o f f l ow t o wat er and t hat o f the
f l ow of per meat e due to the phenomenon o f
mai nt ai ni ng concent r at i on pol ari zat i on.
5.3. Influence a high transmembrane pressure on
the retention of the ions
Part o f the wat er was sof t ened by exert i ng
pressures r angi ng bet ween 4 and 16 bar and by
4O
1
30
2 5
20
Pure water
Hamma water
15-
i0-
5 2
0 ' , , , ,
0
y
t l
i
4 6 8 10 12 14 16 18
P (bar)
20
Fig. 5. Solvent flow and Hamma flow as a function of
pressure.
Table 3
Composition of feed and permeate water
Feed Permeate Retention
water (%)
Conductivity 998-1033 572-905 39.35-47.05
( #S/ c at 20C)
CI-, mg/L 120-125 80-100 12.5-17.8
SO42-, mg/L 116-158 1.30-10 98.0-98.6
HCO3-, mg/L 414-475 238-247 32.00-39.77
Ca >, mg/L 128-132 54-65 45.30-49.45
Mg 2, mg/L 36-39 12.5-13.2 55.77-63.40
pH 7.14-7.72 7.61-8.38 - -
Table 4
Composition of mixed water
Mixture
pH 7.16-8.07
Conductivity, #S/cm at 20C 799-828
CF, mg/L 105-110
SO4 z-, mg/L 66.66-91.20
HCO~, mg/L 342-385
Ca >, mg/L 89-I00
Mg >, mg/L 23.4-25.0
320 S. Ghizellaoui et al. / Desalination 179 (2005) 315-322
43
42
41
40
39
= 38
@
= 3 7
3 6 : . , . a
35
34
33
32
31
HCO 3
aj g t4 l b ' 1 2 '1 5 18
P ( bar )
Fig. 6. Percent retention of HCO 3,
50
--- 49
= 48
0
o = 47
46
45
Fig. 7.
Ca 2+
6 8 1'0 1'2 " 14
P (bar)
Percent retention of Ca 2+.
1'6 18
65
64
63
=~ 62
61
"= 60
=
59
58
57
56
55
Mg 2+
P (bar)
Fig. 8. Percent retention of Mg 2.
ma i nt a i ni ng t he s a me t e mpe r a t ur e us e d bef or e
20C. It is i mpor t a nt t o not e t hat at t he end o f
fi l t rat i on t he s of t ened wat er was mi x e d wi t h 50%
r aw wat er. Al l r esul t s are r e por t e d i n Tabl es 3
and 4.
It s houl d be n o t e d t hat dur i ng t he fi l t rat i on o f
t he Ha mma wat er (Fi gs. 6 - 8 ) b y t he me mbr a ne ,
t he r et ent i on o f t he i ons i ncr eas ed wi t h pr essur e,
t he n e vol ve d sl i ght l y, and t he n st abi l i zed in t he
shape o f a pl at eau. Th e s a me obs er vat i on was
ma de by Ra na t a ms kul [13].
The i ncr ease i n r et ent i on can be expl ai ned as
i ndi cat ed by Me hi gu~ne [14] by t he i ncr ease i n
t he s ol vent f l ow, and as t he t r ans f er o f sol ut e does
not i ncr ease wel l as s ol ve nt does, t he sol ut e is
di vi de d i nt o a v o l u me o f mo r e s i gni f i cant sol vent .
The pe r me a t e wi l l be c onc e nt r a t e d l ess and t he
rat e o f r et ent i on wi l l be mo r e si gni f i cant . Anot he r
expl anat i on can al so be gi ven: u n d e r t he ef f ect o f
pr essur e, t her e is a pr ogr e s s i ve a c c umul a t i on o f
i ons on t he sur f ace o f t he me mb r a n e i nvol vi ng
concent r at i on pol ar i zat i on, wh i c h t her ef or e estab-
l i shes a gr adi ent o f c onc e nt r a t i on be t we e n t he
sur f ace me mb r a n e and t he f eed sol ut i on. Thi s
di f f er ence i n c onc e nt r a t i on i nduc e s a di f f usi onal
f l ow i n t he oppos i t e di r ect i on o f c onve c t i ve f l ow
i nvol vi ng a r e duc t i on i n t r ansf er o f t he i ons t o t he
l evel o f t he me mb r a n e and t hus mo r e si gni f i cant
r et ent i on.
It is not i c e d t hat f or l ow pr es s ur es (Fig. 9) t he
rat e o f r et ent i on o f c hl or i de s is weak, but it
i ncr eases r e g u l a r l y - - a l t h o u g h s l i g h t l y - - wi t h t he
i ncr ease in pr es s ur e. In cont r ast , t he sul phat es
s h o we d t he hi ghe s t r at e o f r et ent i on, whi c h is
i nde pe nde nt o f TMP (Fi g. 10) . These i ons, ha vi ng
t he s ame negat i ve char ge as t he me mbr a ne , are
p u s h e d ba c k b y t hi s me mb r a n e . Ther ef or e, t her e
wi l l is an a c c umul a t i on o f ne ga t i ve char ges cl ose
t o t he me mbr a ne .
For l ow pr es s ur es , t he mo n o v a l e n t Co- i ons
(C1-) cr oss t he me mb r a n e mo r e easi l y. The
r et ent i on o f bi car bonat es i ncr eas es sl i ght l y wi t h
t he i ncr ease o f pr es s ur e, but t he y are r et ai ned
bet t er t han chl or i des due t o t he st eri c ef f ect whi c h
S. Ghizellaoui et al. / Desalination 179 (2005) 315-322 321
18
17
",--16
~ 15 ,t,,,a
CO
14
13
12
1() 1'2
P (bar)
Fig. 9. Percent retention of C1-.
C I "
' 1 4 ' 1 6 18
70-
65 ~
,-, 6O;
S
v
= 5 5 .
50-
45 ~
c~ 4O ~
35 ~ Ca 2'
3C 260 ' 2:20 '240" 260 ' 280' 300 '3:20 ' 340 ' 360
v (I/h)
Fig. 11. Percent retention of Ca > as a function of the rate
of tangential flow.
99
98
e- .
@
~, 97
9 6
4 6 B 10
P ( b a r )
Fig. 10. Percent tetention of SO~-.
1 '2 ~ 8
S 0 4 "2
i
114 16
carries it. Moreover, their concentration being
raised, their retention remai ns weak (Fig. 6).
5.4. Influence rate o f t angent i al f l o w on the
retention
Retention increases si mul t aneousl y with the
rate of circulation. Thus the increase of the tan-
gential flow rate mi ght be linked to a decrease of
circulation. Consequent l y, this coul d generat e the
increase of the retention.
We not ed that the increase speed led to a
slight increase in the rate of retention of divalent
ions (Ca >, Mg >, SO42-) and HCO~. Figs. 11-14
75
7O
= 65
o
" ~
60-
~ D
55" Mg 2~
50
200 ' 220' 2,~0' 260' 280' 300' 30' 340' 360
V0Aa)
Fig. 12. Percent retention of Mg 2+ as a function of the rate
of tangential flow.
show that at a fi xed pressure the drive of divalent
to a hi gher speed towards the retentate is more
significant compared to t he membrane. The influ-
ence of the sweepi ng forces towards the permeat e
is smaller, and the forces on the surface are more
effective in this case.
In contrast, i f the rate of tangential flow is
low, the quantity of the ions on the level of the
membrane is high (quantity of solute in the pores
is high). I f the sweepi ng forces towards the
permeat e become stronger than the forces of
surface, there will be a decrease of retention of
ions such as Ca >, Mg 2+, SO 2- and HCO 3.
322 S. Ghizellaoui et al. / Desalination 179 (2005) 315-322
98.4-
98.2
.~" 98.1
~=9s.0
97. 9
97. 8
97.7.
s 0 } -
97.6 260'20 '2~i0 '2do 2~0 '360 '3:}0 '34~0 '360
V ( l / h)
Fig. 13. Percent retention of SO~- as a function of the rate
of tangential flow.
50-
~, 45-

= 4@
Q
35-
30-
25-
HCO 3"
260 ' 220 '240 '260 '28'0 '300 '320 '34'0 '360
v (l / h)
Fig. 14. Percent retention of HCO{ as a function of the
rate of tangential flow.
6. Co n c l u s i o n s
The principal objective of this work was to
study the process of water softening, which is
necessary for the protection of the distribution
network of drinking waters and industrial facili-
ties against scaling. With this intention, two
techniques on the laboratory scale were tested
with NF. The first study used low pressures (0.5,
1 and 2 bar), and the second one used relatively
strong pressures (4-16 bar) on raw water. There-
after, the latter (50%) was mixed with filtered
water (50%).
The results reveal that the last process
produced partially softened water of good quality
containing minerals which are necessary for its
consumption. Moreover, the interest of NF for the
elimination of temporary hardness was con-
firmed. Rates of retention reached 50% for Ca 2+
and 40% for HCO3 at relatively strong pressures
and 34% for Ca 2+ and 30% for HCO3 at low
pressures. These rates of retention mainly depend
on the tangential flow rate. However, in the case
of pressure, we noted a certain evolution at the
beginning which is then stabilized in the shape of
a plate due to the concentration polarization
phenomenon.
Re f e r e n c e s
[1] A. Khalil, P. Sassiat, C. Colin, C. Meignen, C.
Gamier, C. Gabrielli, M. Keddam and R. Rosset,
TSM, 1992, pp. 259-263.
[2] R. Rosset, D. Mercier and S. Douville, Ann. Fais,
Exp. Chim., 90(938) (1997) 41-65.
[3] R. Rosset, L'actualit4 chimique, (1992) 125-148.
[4] G. Maydatchevsky, TSM, 6 (1995) 1-29.
[5] D. Cleret and J. L6dion, TSM, 2 (1998) 55-67.
[6] A. Khalil and V.S. Praptowidodo, Desalination, 132
(2000) 287-292.
[7] C.A.C. Van de Lisdonk, J.A.M. Van Paassen and
J.C. Schippers, Desalination, 132 (2000) 101-108.
[8] Y. Garba, S. Taha, N. Gondrexon, J. Gabon and
G. Dorange, J. Membr. Sci., 168 (2000) 135-141.
[9] K. Mehiguene, Y. Garba, S. Taha, N. Gondrexon and
G. Dorange, Sep. Pur. Technol., 15 (1999) 181-187.
[10] S. Ghizellaoui, S. Taha, G. Dorange, A. Chibani and
J. Gabon, Desalination, 7 (2004) 133-138.
[11] X. Xu and H.G. Spencer, Desalination, 113 (1997)
85-93.
[12] L. Paugam, Th6se de I'ENSC, Rennes, 2002.
[13] C. Ranatamskul, T. Urase and K. Yamamoto, Water
Sci. Technol., 38(4-5) (1998) 453-462.
[14] K. Mehigu6ne, Th6se de I'ENSC, Rennes, 1999.