influence of ph on treatment of dairy wastewater by
TRANSCRIPT
General rights Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.
Users may download and print one copy of any publication from the public portal for the purpose of private study or research.
You may not further distribute the material or use it for any profit-making activity or commercial gain
You may freely distribute the URL identifying the publication in the public portal If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.
Downloaded from orbit.dtu.dk on: Jun 23, 2022
Influence of pH on treatment of dairy wastewater by nanofiltration using shear-enhanced filtration system
Luo, Jianquan; Ding, LuHui
Published in:Desalination
Publication date:2011
Document VersionPublisher's PDF, also known as Version of record
Link back to DTU Orbit
Citation (APA):Luo, J., & Ding, L. (2011). Influence of pH on treatment of dairy wastewater by nanofiltration using shear-enhanced filtration system. Desalination, 278(3-3), 150-156.
Desalination 278 (2011) 150–156
Contents lists available at ScienceDirect
Desalination
j ourna l homepage: www.e lsev ie r.com/ locate /desa l
Influence of pH on treatment of dairy wastewater by nanofiltration usingshear-enhanced filtration system
Jianquan Luo, LuHui Ding ⁎EA 4297 TIMR, University of Technology of Compiegne, 60205 Compiegne Cedex, France
⁎ Corresponding author. Tel.: +33 3 4423 4634; fax:E-mail address: [email protected] (L.H. Ding).
0011-9164/$ – see front matter © 2011 Elsevier B.V. Adoi:10.1016/j.desal.2011.05.025
a b s t r a c t
a r t i c l e i n f oArticle history:Received 16 March 2011Received in revised form 4 May 2011Accepted 5 May 2011Available online 28 May 2011
Keyword:MilkFoulingConcentration polarizationRotating disk module
In the dairy industry, the use of acid, alkaline cleaners and sanitizers affects wastewater characteristics andtypically results in a highly variable pH, which will bring an impact on the wastewater on-line treatment. Inthis study, the effect of pH on treatment of diary wastewater by nanofiltration was investigated using arotating disk membrane module. With increase of pH, NF membrane permeability was improved butconcentration polarization increased. Due to a stronger electrostatic repulsion between salt ions andmembrane at higher pH, permeate conductivity decreased. However, because of membrane swelling effect atalkaline pH, the neutral organic solutes passed through membrane more easily, leading to an increase ofChemical Oxygen Demand (COD) in permeate. Membrane fouling could be greatly alleviated at pH of 10, andthe fouling layer was mainly caused by the adsorption of acid radicals-protein aggregates, which also resultedin pH reduction and conductivity increase in retentate. Real dairy wastewaters with different pH were alsotreated and the results were very similar to those of model solution, while membrane fouling was verysensitive to feed pHwhen pHwas larger than 8. These results from laboratory-scale tests can serve as valuableguide for process control in industrial applications.
+33 3 4423 7942.
ll rights reserved.
© 2011 Elsevier B.V. All rights reserved.
1. Introduction
The dairy industry, like most other food industries, produces alarge volume of wastewater, essentially composed of diluted milk,which are responsible for a 1–3% loss inmilk components (lactose andproteins) [1]. Moreover, dairy wastewater also contains muchchemicals (acids, alkalis and detergents), most of which comingfrom clean in place (CIP) systems [2]. This effluent results in watereutrophication and is hazardous to aquatic life and soils, causingsignificant environmental problems when it is discarded withouttreatment. There are many technologies used to treat dairy waste-water, such as coagulation [3], ecological treatment system [4],anaerobic or/and aerobic reactors [5], membrane separation [6].Membrane technologies have been often considered a promisingmethod to produce reusable water from dairy wastewater [2,7,8], andthey can be combined with other technologies to improve treatmentefficiency [9,10].
Several investigations have shown that nanofiltration (NF) wasadequate for concentration of dairy effluents and production ofreusable water [2,6,7,11]. However, with increase of organic contentin retentate during a concentration process, concentration polariza-tion phenomenon becomes more important, due to a decrease ofshear rate in crossflow modules, leading to flux decline and permeate
quality deterioration [12]. Increasing fluid velocity can enhance theshear rate at membrane surface, but also result in a large pressuredrop between inlet and outlet of membrane module, which leads to asignificant increase of local fouling at the inlet and nonuniformtransmembrane pressure (TMP). Shear-enhanced filtration systems,for example, a rotating disk membrane (RDM) module, can solve thisproblem because it generates a very high shear rate at membranesurface (1–5×105 s−1) without pressure drop, reducing concentra-tion polarization effectively [13,14]. It has been shown recently by Luoet al. [15] that, under extreme hydraulic conditions of highest TMPwith high shear rate, the NF-RDM filtration system could produce abetter permeate quality and save energy, because of its very highpermeate flux (100–500 Lm−2h−1).
Another problem for on-line treatment of dairy wastewater by NFis the pH variation of effluents, since the use of acid, alkaline cleanersand sanitizers in the dairy industry influences wastewater character-istics and typically results in a highly variable pH (3–11) [5,16]. As weknow, amembrane process is greatly affected by feed pH because bothmembrane performances and solute properties change with pHvariation [17-21]. Rabiller-Baudry et al. [22] reported that, withincrease of pH, calcium ions and phosphate radicals linked to caseinswere increased, and the size of casein aggregates was minimum at pHof about 9, thus causing the lowest critical and limiting flux forcrossflow filtration of skim milk by NF. They also found thatmembrane fouling could be alleviated by elevating pH in feed.However, there are few papers dealing with the role of pH intreatment of dairy wastewater, and the influence of pH on permeate
Table 1Properties of NF270 membrane [17,21].
Index NF270
Membrane material PolyamideMolecular weight cutoff (Da) 150–200Lpa (Lm−2 h−1 bar−1) 25 °C 11.3±0.3Max. temperature (°C) 45Max. pressure (bar) 41Contact angle (sessile drop) (°) 30Zeta potential (mV) −62.5 (pH=6)
−70.5 (pH=8)−75.0 (pH=10)
Isoelectric point (pH) ~3.2
a Pure water permeability.
151J. Luo, L. Ding / Desalination 278 (2011) 150–156
quality for NF of dairy wastewater is not very clear yet and needs to beinvestigated systematically. Furthermore, because of special fluxbehavior and fouling mechanisms for shear-enhanced filtrationsystem [15], the effect of pH on treatment of dairy wastewater byshear-enhanced filtration system may be different from that incrossflow filtration.
In this study, a RDMmodule was used to treat model and real dairywastewaters by a NF270 membrane at different pH. The focus of thiswork was to discuss the effect of pH on permeate flux and soluterejection during short-term full recycle tests, considering concentra-tion polarization but not membrane fouling. Long-term experimentsat different pH were conducted to examine the variation of flux andsolute permeate during concentration processes. Membrane foulingwas studied in long-term tests and the formation mechanism wasdiscussed. All the conclusions were used to explain the differentphenomena for real effluents with different pH that found in previoustests. The present work should be very useful for understanding theeffect of pH on NF process of dairy wastewater and serve as valuableguides for process control in industrial production.
2. Materials and methods
2.1. Experimental set-up and membranes
The RDMmodule has been designed and built in our laboratory. Asshown in Fig. 1, thismodule is fed from a thermostatic and stirred tankcontaining 12 L of fluid by a volumetric diaphragm pump (Hydra-cell,Wanner, USA). The peripheral pressure (pc) was adjusted by a valveon outlet tubing and measured at the top of the cylindrical housing bya pressure sensor (DP 15–40, Validyne, USA), and the data wascollected automatically by a computer. The permeate was collected ina beaker placed on an electronic scale (B3100 P, Sartorius, Germany)connected to a computer in order to determine the permeate flux. ANF membrane NF270 (Dow-Filmtec) was chosen to use according toprevious study [15], and based on the manufacturer's data sheet and
Fig. 1. Schematic diagram of the experiment se
literature [17,21], the properties of this NF membrane are shown inTable 1.
2.2. Test fluid
A model effluent (pH=6.84) was prepared from commercial UHTskim milk (Lait de Montagne, Carrefour, France), diluted 1:2 to one-third of normal concentration with deionized water (Aquadem E300,Veolia Water, France). Real effluents were collected from a local dairyfactory (Lactalis, Clermont, France) in two different batches, with pHof 8.72 and 9.56, respectively. Real effluents were pretreated by twosieves with pore size of 0.25 mm and 0.10 mm (Prolabo, Paris,France).
2.3. Experimental procedure
A new membrane was used for each series of experiments toensure the same initial membrane conditions for the entire study. Themembranes were soaked in deionized water for at least 48 h prior touse, and pre-pressured with deionized water for 30 min under a
t-up for rotating disk membrane module.
10 20 30 40
40
60
80
100
120 6 7 8 9 10
Flu
x (L
m-2
h-1
)
TMP (bar)
pH
Fig. 2. Variation of permeate flux as a function of TMP at different pH values. Rotatingspeed=2000 rpm; T=35 °C; conductivity in feed=2.2 ms·cm-1.
152 J. Luo, L. Ding / Desalination 278 (2011) 150–156
pressure of 40 bar at 25 °C. After stabilization, the pure water flux ofmembranes was measured at five pressures of 20, 16, 12, 8, and 4 barto calculate water permeability (Lp). Before the experiments started,the feed was heated to 35 °C, and pH was well adjusted with 1 MNaOH and 1 M HCl to specific values as required, then conductivitywas adjusted to the same value by adding NaCl for each series tests,during which the feed was fully recycled in the system at zero meantransmembrane pressure (TMP), and this process lasted about 10 minfor each test. Then experiments were performed at a feed flow rate of180 L h−1 in two series: short-term full recycle tests and long-termconcentration tests.
2.3.1. Series 1: short-term tests with full recyclingIn order to investigate the effect of pH on permeate flux, solute
rejection and concentration polarization at a constant feed concen-tration, this series was performed with permeate and retentaterecycling (3 L). The pre-filtration was carried out for 30 min at TMP of10 bar and rotating speed of 2000 rpm, to ensure that membranebecame stable after pH changing. Then TMP increased in steps from 10to 40 bar, and at the endwas returned to 20 bar to see if concentrationpolarization has increased. Samples were all collected in permeate10 min after the beginning of each TMP increment in order to havestabilized flux and transmission conditions. After each test (at samepH), used feed was evacuated from the system and the membranewas flushed with deionized water, and then new feed was added. Bycomparing the pure water permeability of membrane before and afterfiltration, it was found that no visible irreversible fouling was found inthis series of experiments.
2.3.2. Series 2: long-term concentration testsIn this series, the permeate was not returned to feed tank and a
12 L feed was concentrated to 1.5 L (2 L for pH of 9–10) at a constantrotating speed of 2000 rpm and a TMP of 40 bar. Samples werecollected in permeate every half of volume reduction ratio (VRR). Inorder to obtain data of membrane fouling at the same permeatevolume for all pH, there was no pre-filtration process under pressurefor this series. When a concentration test ended, the filtration systemwas flushed with deionized water until the rinsing water came outclear, and Lp wasmeasured again to evaluate the degree of irreversiblefouling. Subsequently, a cleaning agent solution (P3 Ultrasil 10,Ecolab, USA) at 0.05% concentration was used to clean themembranesat room temperature (20–25 °C) for 30 min. Then the system wasrinsed by deionized water again until the pH of the rinse water wasneutral. Finally, the membranes were soaked in deionized water for30 min to eliminate the effect of swelling by cleaning and then Lp wasmeasured again. The membrane used was taken out and a newmembrane was fitted for the next pH test.
2.4. Analytical methods
Turbidities of retentate and permeate were measured with a RatioTurbidimeter (Hach, USA). Chemical oxygen demand (COD) wasmeasured using Nanocolor Kits (Machery-Nagel, Hoerdl, France) inorder to quantify organic matter concentration. Conductivity wasmeasured with Multi-Range Conductivity Meter (HI 9033, Hanna,Italy) and pH was measured with pH Meter (MP 125, Mettler Toledo,Switzerland).
2.5. Calculated parameters
The volume reduction ratio (VRR) is defined as:
VRR =Vo
VR; ð1Þ
where Vo is initial feed volume and VR is retentate volume.
The mean TMP is obtained by integrating the local pressure overthe membrane area as follows:
TMP = pc−14ρk2ω2R2
; ð2Þ
where pc is the measured peripheral pressure, ρ is the density of thefluid at 35 °C and k is the velocity factor (0.89 for this system), ω thedisk angular velocity and R the housing inner diameter.
The irreversible fouling index (IF) can be expressed as apercentage of pure water permeability decrease after the experiment.
IF %ð Þ = Lpi−LpfLpi
× 100 ð3Þ
where Lpi and Lpf are the initial and final pure water permeabilityrespectively.
3. Results and discussion
3.1. Effect of pH on permeate flux and solutes rejection (series 1)
Fig. 2 shows the variation of permeate flux with TMP at differentpH. Permeate flux increased almost linearly with TMP and the fluxvalues at different pH were very close, except for the flux profile at pH10, which was a little higher at low TMP and then becomes lower thanthe others at 40 bar. As we know, at the same TMP, permeate flux isgoverned by membrane permeability and concentration polarizationof solutes according to the resistance model [23]. Mänttäri et al. [17]found that when pH increased from 8 to 11, the hydraulicpermeability of NF270 membrane went up, due to an increase ofsurface hydrophilicity. As seen in Fig. 2, permeate flux does notincrease with pH from 8 to 9, meaning that concentration polarizationincreases with pH, and this negative effect on permeate flux cancelsthe positive effect of membrane permeability rise. When pH rose to10, presumably, at low TMP, concentration polarization increased less,while the permeability was much increased, leading to a fluxelevation; but at high TMP, concentration polarization phenomenonwas more important, thus decreasing the flux.
In order to verify the above hypothesis, an in-depth analysis ofconcentration polarization was carried out. In this case, concentrationpolarization layer was mainly composed of two parts, salt ions andprotein molecules. First, due to the high diffusion coefficient of saltions, concentration polarization by salt ions was mainly controlled bytheir rejections according to the Film model [24]; as shown in Fig. 3,the increase of pH reduces permeate conductivity, meaning that moresalt ions accumulates at membrane surface. Secondly, protein
10 20 30 40
20
40
60
80
100
120
6 7 8 9 10
Per
mea
te C
OD
(m
gO
2L-1
)
TMP (bar)
pH
Fig. 5. Variation of permeate COD as a function of TMP at different pH values. Rotatingspeed=2000 rpm; T=35 °C; conductivity in feed=2.2 ms·cm-1.
10 20 30 40
250
500
750
1000
1250
6 7 8 9 10
TMP (bar)
pHP
erm
eate
Co
nd
uct
ivit
y(µs
.cm
-1)
Fig. 3. Variation of permeate conductivity as a function of TMP at different pH values.Rotating speed=2000 rpm; T=35 °C; conductivity in feed=2.2 ms·cm-1.
153J. Luo, L. Ding / Desalination 278 (2011) 150–156
molecules were almost completely rejected by NF270 and theirconcentration polarization was governed by back transfer coefficientof solutes. Because of the relatively low mobility, the dispersion ofprotein polarization layer showed a little hysteresis when TMPdecreased [25], and a series of experiments was designed to evaluatethe effect of pH on concentration polarization of protein. In the firsttest, TMP was increased from 10 to 20 bar, permeate flux wasmeasured at 20 bar (first test), and then TMP increased in steps from30 to 40 bar for 30 min. In the second test, TMP was directly loweredto 20 bar from 40 bar and the fluxwas recorded at 20 bar during 5 min(second test). It can be seen in Fig. 4 that, after operating at high TMPof 30–40 bar for about 30 min, permeate flux declines at the sameTMP of 20 bar. The decline was caused by the hysteresis ofconcentration polarization dispersion and it was larger at pH≥8,especially at pH=10. Therefore, it could be concluded that, higher pHresulted in more severe concentration polarization, leading to a loss indriving force via the enhanced osmotic pressure (salt ions layer) andan increase in filtration resistance (protein molecules layer), and thisnegative effect was much stronger when exerting higher pressure.When operating at low TMP, the positive effect of membranepermeability increase was dominant, and thus permeate flux at pHof 10 was highest. While TMP increased to 40 bar, the negative effectdue to concentration polarization aggravation surpassed the positiveone, and as a result, permeate flux at pH of 10 was lowest.
6 7 8 9 100
20
40
60
80
100
Flu
x (L
m-2
h-1
)
pH
first test second test
-2.59% -2.90% -7.11%
Fig. 4. Flux decline by concentration polarization of protein at different pH (no visibleirreversible fouling was found). TMP =20 bar; Rotating speed=2000 rpm; T=35 °C;conductivity in feed=2.2 ms·cm-1.
Solution pH affected the charged characteristic of membrane[17,21], and it was also reported that the NF membrane skin couldswell with increase of pH, and the swelling could be much moresusceptible to pH in concentrated salt solutions [26]. Therefore, soluterejection by NF would be greatly influenced by pH in dairywastewater. As seen in Fig. 3 and Fig. 5, as pH rises from 6 to 10,permeate conductivity decreases but permeate COD shows a reversetrend. This could be explained as follows: first, the main separationmechanism for salt ions was electrostatic repulsion, and with increaseof pH, membrane was more negatively charged [17,21], inducing astronger electrostatic effect between salt ions and membrane, thusleading to a decrease of conductivity in permeate, but this effectwould be weakened by flux decline due to concentration polarizationincrease at maximum TMP; secondly, steric exclusion was thedominant mechanism for rejecting neutral organic matters by NF,and permeate CODwasmainly composed of lactose for dairy effluents,which would pass through membrane more easily at high pH becauseof membrane pore swelling effect [15,17,21,27]. In addition, for a pHof 10, with increase of TMP, solute concentration in permeategradually stopped decreasing or went up a little, which also resultedfrom the severe concentration polarization at high pH. Actually,solutes passed through membrane by two transport mechanisms—convection and diffusion [28]. The first was controlled by permeateflux, and solute concentration in permeate decreased at a higher
1 2 3 4 5 6 7 8 90
100
200
300
400
500
VRR
6.36.87.48.49.310.1
Initial pH
Flu
x (L
m-2
h-1
)
Fig. 6. Variation of permeate flux during concentration process at different initial pH.TMP=40 bar; Rotating speed=2000 rpm; T=35 °C; conductivity in feed=2.8 ms·cm-1.
1 2 3 4 5 6 7 8 90
800
1600
2400
3200
4000
6.3 7.4 8.4 9.3 10.1
VRR
Initial pH
Per
mea
te c
on
du
ctiv
ity
(µs
cm-1
)
Fig. 7. Effect of pH on themean permeate conductivity in permeate accumulated at eachhalf value of VRR during concentration process. TMP=40 bar; Rotating speed=2000rpm; T=35 °C; conductivity in feed=2.8 ms·cm-1.
0
20
40
60
80
100
7.4
Rec
ove
ry (
%)
Initial pH6.3 8.4 9.3 10.1
Fig. 8. Membrane fouling removal by chemical cleaning at different operating pH.
154 J. Luo, L. Ding / Desalination 278 (2011) 150–156
water flux due to the “dilution effect”; the second was governed bysolute concentration at membrane surface, and solute concentrationin permeate increased when concentration polarization rose. Withincrease of TMP, convection transport was dominant in the earlystage, and as permeate flux increased, concentration polarizationbecamemore andmore important [29], and finally diffusion transportprevailed over convection, causing increase of solutes in permeatewith TMP.
3.2. Effect of pH on concentration process (series 2)
Concentration tests were performed under extreme hydraulicconditions (TMP=40 bar, rotating speed =2000 rpm) to examinethe effect of pH in a long-term run. In Fig. 6, as VRR increases,permeate flux steadily declines except for the cases of pH of 9.3 and10.1, where flux increases at first before decreasing. This was causedby the absence of pre-filtration, as membrane performance did notbecome stable before the concentration tests started, and the rise ofmembrane permeability at alkaline pH occurred at the beginning ofconcentration processes. Moreover, as seen in Fig. 6, permeate flux atinitial pH above 8 is much lower than those at pH 6–7. The mainreason was the severe concentration polarization for alkaline pH asdiscussed in Section 3.1, and at the same time, the total salt ionscontent in retentate affected the permeate flux during concentrationprocess. As shown in Fig. 7, when VRR is less than 3, permeateconductivities at initial pH of 8.4 and 9.3 are lower than that at a pH of7.4, and more salt ions accumulate at membrane surface at higher pH,partially inducing a lower permeate flux. For initial pH of 6.3, althoughthe flux was very high, the permeate conductivity was so large that itexceed the allowed limit (1000 μs cm−1) [13]; for neutral pH, profilesof flux and permeate conductivity were the most suitable; for initialpH of 10.1, because of the largest concentration polarization, the flux
Table 2Effect of pH on permeate quality⁎ and membrane fouling.
Initial feed pH Turbidity(NTU)
COD(mgO2 L−1)
Conductivity(μs cm−1)
IF(%)
6.3 0.50 46 1441 28.797.4 0.52 36 660 23.568.4 0.52 53 650 20.419.3 0.58 73 515 23.2610.1 0.63 152 795 14.05
⁎ Mean turbidity, COD and conductivity in accumulated permeate at the end offiltration.
was lowest, resulting in a drop of salt ion rejection. It was reportedby Rabiller-Baudry et al. [22] that, for alkaline pH, calcium ions andphosphate radicals were more likely to link to caseins, and thisresulted in the formation of cross-linked organic polarization layerson the membrane surface through active organic-calcium complex-ation [30]. This couldwell explainwhy concentration polarizationwasthe largest at highest pH.
Table 2 shows the mean turbidity, COD, and conductivity inaccumulated permeate and irreversible fouling at the end ofconcentration. Due to the low permeate flux and some membraneswelling for pH of 9.3 and 10.1, turbidity and COD in permeateincreased. The lowest COD in permeate for initial pH of 7.4 was causedby its highest permeate flux, and the lowest conductivity in permeatefor initial pH of 9.3 resulted from the strong electrostatic repulsionbetween membrane and salt ions. Thereby, the mean COD andconductivity in accumulated permeate at the end of filtration at pH 7–9 are well below the “authorized limit” (COD~125 mgO2 L−1,conductivity~1000 μs cm−1) [13]. The irreversible fouling wassmallest for initial pH of 10.1, and this result was in agreement withthat in Rabiller-Baudry et al. [22]. On the one hand, NF270 membranewas more charged at pH of 10 and thus its anti-fouling performancewas enhanced; on the other hand, due to a strong inter-chainelectrostatic repulsion at pH of 10.1 [31], the cross-linked proteinmacromolecules become less compact and made more difficult toform a fouling layer [32]. However, the initial pH of 9.3 did not reducemembrane fouling (see Table 2), and even decreased the cleaningefficiency, as shown in Fig. 8. A possible reason was that, the size ofcasein aggregates was minimum at pH of about 9 [22], thus inducingthe most compact fouling layer at the membrane surface. However, asseen in Fig. 8, chemical cleaning can effectively remove most ofmembrane fouling in NF of dairy wastewater.
3.3. pH variation in retentate during treatment
pH in retentate decayed with increase of VRR for all concentrationexperiments, as shown in Fig. 9. Luo et al. [15] suggested that, the acidradicals (e.g. phosphate) could precipitate with caseins as well as withcalcium, and thus more hydrogen ions were produced by electrolysisand pH decreased. If this is true, pH reduction would occur withfouling layer formation, and conductivity in retentate will beenhanced because hydrogen ions have the highestmolar conductivity.To verify this assumption, full recycle tests for a cleaned membraneand a fouled membrane were carried out, and both pH andconductivity in retentate were monitored. The full recycle operationcan ensure that solute concentration in retentate is constant andthe variations of pH and conductivity in retentate are just causedby fouling formation. The difference between these two series
1 2 3 4 5 6 7 8 9
6
7
8
9
10p
H in
ret
enta
te
VRR
Fig. 9. Variation of pH in retentate with VRR during concentration process. TMP=40bar; Rotating speed=2000 rpm; T=35 °C; conductivity in feed=2.8 ms·cm-1.
Table 3Comparison of treatment results between model solution and industrial effluent withdifferent pH.
Index Model solution Batch 1 Batch 2
Feed/permeate
Turbidity (NTU) NA/0.57 101/0.56 100/0.57COD (mgO2 L−1) 36000/54 297/b15 580/b15Conductivity (μs cm−1) 2170/685 1084/525 1516/317pH 6.8/6.6 8.7/7.9 9.6/9.00IF (%) 21.88 26.87 13.40
155J. Luo, L. Ding / Desalination 278 (2011) 150–156
experiments was that, membrane fouling was negligible in 2 h in thefirst case with frequent chemical cleaning, but for the second case,membrane fouling has been aggravating due to adsorption of caseinaggregates to membrane surface. It can be seen in Fig. 10 that, when amembrane fouling is not visible (permeate flux is almost constant),pH and conductivity in retentate keep almost constant, whilemembrane is used again and again without any chemical cleaning,permeate flux declines with time, and pH in retentate reduces andconductivity goes up as expected. Accordingly, the membrane foulingmechanism in NF of dairy wastewater by RDM module could bediscussed as follows: the phosphates, lactate and citrate ions in dairyeffluents precipitated with caseins, and these aggregates adsorbed tomembrane surface, leading to a decline of permeate flux; at the sametime, more hydrogen ions were generated by electrolysis and pH inretentate dropped, whereas conductivity increased.
0
100
200
300
400
cleaned membrane fouled membrane
0 30 60 90 1203
4
5
6
7
pH
in r
eten
tate
cleaned membrane fouled membrane
solid: pH blank: conductivity
Time (min)
1000
1500
2000
2500
3000 Reten
tate con
du
ctivity (µs cm-1)
Flu
x (L
m-2
h-1
)
Fig. 10. Variation profiles of flux, pH and conductivity in retentate with time for the fullrecycle process. TMP=40 bar; Rotating speed=2000 rpm; T=35 °C; conductivity infeed=2.2 ms·cm-1.
3.4. Treatment of real effluents
As seen in Table 3, real effluents have much less organic mattersand inorganic salt than model solution, but their pH was much higherdue to the use of alkaline solution as cleaning agent. Moreover, thecomponents and concentration of effluents in milk factory changedwith time. Tests were performed in two batches of real effluents withdifferent pH in order to find the effect of pH on treatmentperformance of NF. Fig. 11 shows the variation of permeate fluxwith VRR for different feeds. Unexpectedly, the flux of real effluentswas of the same level as that of model solution despite differentcompositions (see Table 3). This result was well discussed in aprevious study [15]. Here, the point that needs to be clarified, is thereason why the flux for real effluents was so different at pH of 8.72and 9.56. First, when VRR was less than 2.4, permeate flux for Batch 1was lower than the other one, which resulted from the highermembrane permeability for Batch 2 with higher pH (see Section 3.1)and greater membrane fouling for Batch 1 (see Table 3). Then, due tothe severe fouling for Batch 1, its pH in retentate reduced a lot andthus concentration polarization aswell as salt ions rejection decreased(see Section 3.1). These effects were positive for permeate flux, asthey counteracted the negative effect of an increase of soluteconcentration in retentate with increasing VRR, therefore, inducinga stable flux profile for VRRN2. However, for Batch 2, membranefouling was low and pH in retentate decreased little, and consequent-ly, salt ions rejection remained very high (see Table 3). With increaseof VRR, solute concentration increased, and permeate flux decreaseddue to the rise of osmotic pressure atmembrane surface. Furthermore,the results in Table 3 were consistent with those of model feed inTable 2, confirming the effect of pH on NF of dairy wastewater by RDMmodule. However, by comparing the model solution at an initial pH of9.3 with the real effluent of Batch 2 at a pH of 9.6, it could be foundthat, although their pH was only a little different, their flux profilesand membrane fouling were quite different. For the model feed, pH in
1 2 3 4 5 6 7 8 9 100
100
200
300
400
500
VRR
Model solution: pH 6.84Real effluent 1: pH 8.72Real effluent 2: pH 9.56
Flu
x (L
m-2
h-1
)
Fig. 11. Comparison of permeate flux between model solution and real effluents withdifferent pH. TMP=40 bar; rotating speed=2000 rpm; T=35 °C.
156 J. Luo, L. Ding / Desalination 278 (2011) 150–156
retrentate decreased from 9.3 to 8.3 and irreversible foulingwas about23%. For the real feed with pH of 9.6, the fouling was just about 13%.This means that membrane fouling is very sensitive to pH of dairyeffluent in a range of 8–10.
4. Conclusions
The pH variation of dairy wastewater had a large impact on theseparation performance of NF. Operating at acid pH resulted in highpermeate flux, but low salt removal, and for alkaline pH, membranefouling could be alleviated, but permeate flux decreased due to asevere concentration polarization. A dairy wastewater with pH of 7–8 was most suitable to be treated by the NF-RDM module underextreme hydraulic conditions because of a good compromise betweenpermeate flux and membrane fouling as well as a satisfactorypermeate quality.
With increase of pH, the electrical charge of NF270membrane roseand the stronger electrostatic repulsion between salt ions andmembrane, which lowered permeate conductivity, while COD inpermeate increased because of membrane pore swelling at alkalinepH. The acid radicals (e.g. phosphate) in dairy effluents couldprecipitate with caseins to form aggregates, which adsorbed tomembrane surface and induced permeate flux decline. As the foulinglayer formed at membrane surface, more hydrogen ions wereproduced by electrolysis, and thus pH in retentate decreased, whereasconductivity in retentate increased.When pHwas about 9, membranefouling was more compact and chemical cleaning efficiency fell.
Real dairy wastewater with different pH was processed underextreme hydraulic conditions and the results were similar to thoseof model solution. Moreover, the peculiar flux profiles for the realwastewater were mainly caused by pH variation during concentrationprocess, and membrane fouling was very sensitive to feed pH in arange of 8–10.
Acknowledgments
The financial support of China Scholarship Council (CSC) forJianquan Luo's thesis fellowship is acknowledged. The authors thankFilmTec Corporation, USA for providing NF270 membranes.
References
[1] M. Vourch, B. Balannec, B. Chaufer, G. Dorange, Treatment of dairy industrywastewater by reverse osmosis for water reuse, Desalination. 219 (2008)190–202.
[2] P. Fernandez, F.A. Riera, R. Alvarez, S. Alvarez, Nanofiltration regeneration ofcontaminated single-phase detergents used in the dairy industry, J. Food Eng. 97(2010) 319–328.
[3] I.A. Sengil, M. Ozacar, Treatment of dairy wastewaters by electrocoagulation usingmild steel electrodes, J. Hazard. Mater. 137 (2006) 1197–1205.
[4] S.L. Lansing, J.F. Martin, Use of an ecological treatment system (ETS) for removal ofnutrients from dairy wastewater, Ecol. Eng. 28 (2006) 235–245.
[5] B. Demirel, O. Yenigun, T.T. Onay, Anaerobic treatment of dairy wastewaters: areview, process, Biochem. 40 (2005) 2583–2595.
[6] B. Balannec, M. Vourch, M. Rabiller-Baudry, B. Chaufer, Comparative study ofdifferent nanofiltration and reverse osmosis membranes for dairy effluenttreatment by dead-end filtration, Sep. Purif. Technol. 42 (2005) 195–200.
[7] M. Vourch, B. Balannec, B. Chaufer, G. Dorange, Nanofiltration and reverse osmosisof model process waters from the dairy industry to produce water for reuse,Desalination. 172 (2005) 245–256.
[8] O. Akoum, M.Y. Jaffrin, L.H. Ding, M. Frappart, Treatment of dairy process watersusing a vibrating filtration system and NF and RO membranes, J. Membr. Sci. 235(2004) 111–122.
[9] J.A. Scott, K.L. Smith, A bioreactor coupled to a membrane to provide aeration andfiltration in ice-cream factory wastewater remediation, Water Res. 31 (1997)69–74.
[10] B. Sarkar, P.P. Chakrabarti, A. Vijaykumar, V. Kale, Wastewater treatment in dairyindustries — possibility of reuse, Desalination. 195 (2006) 141–152.
[11] I. Koyuncu, M. Turan, D. Topacik, A. Ates, Application of low pressurenanofiltration membranes for the recovery and reuse of dairy industry effluents,Water Sci. Technol. 41 (2000) 213–221.
[12] M. Turan, Influence of filtration conditions on the performance of nanofiltrationand reverse osmosis membranes in dairy wastewater treatment, Desalination.170 (2004) 83–90.
[13] M. Frappart, O. Akoum, L.H. Ding, M.Y. Jaffrin, Treatment of dairy process watersmodelled by diluted milk using dynamic nanofiltration with a rotating diskmodule, J. Membr. Sci. 282 (2006) 465–472.
[14] N. Moulai-Mostefa, M. Frappart, O. Akoum, L.H. Ding, M.Y. Jaffrin, Separation ofwater from metal working emulsions by ultrafiltration using vibratory mem-branes, J. Hazard. Mater. 177 (2010) 978–982.
[15] J.Q. Luo, L.H. Ding, Y.H. Wan, P. Paullier, M.Y. Jaffrin, Application of NF-RDM(Nanofiltration Rotating Disk Membrane) module under extreme hydraulicconditions for the treatment of dairy wastewater, Chem. Eng. J. 163 (2010)307–316.
[16] J.R. Danalewich, T.G. Papagiannis, R.L. Belyea, M.E. Tumbleson, L. Raskin,Characterization of dairy waste streams, current treatment practices, andpotential for biological nutrient removal, Water Res. 32 (1998) 3555–3568.
[17] M. Manttari, A. Pihlajamaki, M. Nystrom, Effect of pH on hydrophilicity and chargeand their effect on the filtration efficiency of NF membranes at different pH, J.Membr. Sci. 280 (2006) 311–320.
[18] A.L. Ahmad, L.S. Tan, S.R.A. Shukor, The role of pH in nanofiltration of atrazine anddimethoate from aqueous solution, J. Hazard. Mater. 154 (2008) 633–638.
[19] M. Nilsson, G. Tragardh, K. Ostergren, The influence of pH, salt and temperature onnanofiltration performance, J. Membr. Sci. 312 (2008) 97–106.
[20] J.Q. Luo, S.P. Wei, Y. Su, X.R. Chen, Y.H. Wan, Desalination and recovery ofiminodiacetic acid (IDA) from its sodium chloride mixtures by nanofiltration, J.Membr. Sci. 342 (2009) 35–41.
[21] M. Dalwani, N.E. Benes, G. Bargeman, D. Stamatialis, M. Wessling, A method forcharacterizing membrane during nanofiltration at extreme pH, J. Membr. Sci. 363(2010) 188–194.
[22] M. Rabiller-Baudry, H. Bouzid, B. Chaufer, L. Paugam, D. Delaunay, O. Mekmene, S.Ahmad, F. Gaucheron, On the origin of flux dependence in pH-modified skimmilkfiltration, Dairy. Sci. Technol. 89 (2009) 363–385.
[23] J.G. Wijmans, S. Nakao, J.W.A. Van Den Berg, F.R. Troelstra, C.A. Smolders,Hydrodynamic resistance of concentration polarization boundary layers inultrafiltration, J. Membr. Sci. 22 (1985) 117–135.
[24] W.R. Bowen, F. Jenner, Theoretical descriptions of membrane filtration of colloidsand fine particles — an assessment and review, Adv. Colloid. Interfac. 56 (1995)141–200.
[25] J.Q. Luo, L.H. Ding, Y. Su, S.P. Wei, Y.H. Wan, Concentration polarization inconcentrated saline solution during desalination of iron dextran by nanofiltration,J. Membr. Sci. 363 (2010) 170–179.
[26] V. Freger, T.C. Arnot, J.A. Howell, Separation of concentrated organic/inorganic saltmixtures by nanofiltration, J. Membr. Sci. 178 (2000) 185–193.
[27] V. Freger, Swelling and morphology of the skin layer of polyamide compositemembranes: an atomic force microscopy study, Environ. Sci. Technol. 38 (2004)3168–3175.
[28] J.G. Wijmans, R.W. Baker, The solution-diffusion model: a review, J. Membr. Sci.107 (1995) 1–21.
[29] H.C. van der horst, J.M.K. Timmer, T. Robbertsen, J. Leenders, Use of nanofiltrationfor concentration and demineralization in the dairy industry: model for masstransport, J. Membr. Sci. 104 (1995) 205–218.
[30] Y. Yu, S. Lee, S. Hong, Effect of solution chemistry on organic fouling of reverseosmosis membranes in seawater desalination, J. Membr. Sci. 351 (2010) 205–213.
[31] H. Bouzid, M. Rabiller-Baudry, L. Paugam, F. Rousseau, Z. Derriche, N.E. Bettahar,Impact of zeta potential and size of caseins as precursors of fouling deposit onlimiting and critical fluxes in spiral ultrafiltration of modified skim milks, J.Membr. Sci. 314 (2008) 67–75.
[32] A.S. Al-Amoudi, Factors affecting natural organic matter (NOM) and scalingfouling in NF membranes: a review, Desalination. 259 (2010) 1–10.