Desalination Technology for Management Water Issue in California. 15 pages

Desalination 357 (2015) 140–149

Contents lists available at ScienceDirect

Desalination

journal homepage: www.elsevier.com/locate/desal

Comparison of pre-treatment technologies towards improving reverse osmosis desalination of cooling tower blow down

Jonas Löwenberg a, Jörn Ansgar Baum a, Yannick-Serge Zimmermann a,b, Cornelis Groot c, Wilbert van den Broek d, Thomas Wintgens a,⁎ a Institute for Ecopreneurship, School of Life Sciences, University of Applied Sciences and Arts Northwestern Switzerland, Gründenstrasse 40, CH-4132 Muttenz, Switzerland b Institute for Environmental Research (Biology V), RWTH Aachen University, 52074 Aachen, Germany c Dow Benelux BV, H.H. Dowweg 5, 4542 NM Hoek, The Netherlands d Evides Industriewater, Kanaalweg 1, 4337 PA, Middelburg, The Netherlands

H I G H L I G H T S G R A P H I C A L A B S T R A C T

• CTBD of high DOC shows to be chal- lenging for membrane treatment.

• Membrane based desalination of CTBD is affected by pre-treatment conditions.

• PAC dosage proves beneficial for mem- brane filtration performance on CTBD.

⁎ Corresponding author. E-mail address: [email protected] (T. Wintg

http://dx.doi.org/10.1016/j.desal.2014.11.018 0011-9164/© 2014 Elsevier B.V. All rights reserved.

a b s t r a c t

a r t i c l e i n f o

Article history: Received 16 August 2014 Received in revised form 18 November 2014 Accepted 20 November 2014 Available online 1 December 2014

Keywords: Pre-treatment Powdered activated carbon Ultrafiltration Reverse osmosis Wastewater reuse Cooling tower blow down

The suitability of three different pre-treatment technologies namely powdered activated carbon (PAC) adsorp- tion, coagulation with ferric chloride and ultrafiltration (UF) as pre-treatment to reverse osmosis (RO) desalina- tion of a cooling tower blow down (CTBD) was investigated. Special attention was paid to the capability of the pre-treatment options to remove dissolved organic carbon (DOC) from the CTBD by applying advanced DOC characterization. Furthermore, the direct effect on the performance of reverse osmosis (RO) desalination was investigated. Advanced DOC analysis showed clear differences between the pre-treatment technologies in removal of total DOC and in removal of certain DOC fractions. By investigation of RO normalized flux decline after different pre- treatments an improvement of 10–20% by UF treatment and an additional improvement of 10–20% by PAC/UF treatment were found in comparison to RO performance on not pre-treated CTBD. The achieved results indicate the negative influence of chemical additives in the CTBD matrix on RO fouling and performance as well as pre- treatment efficiency.

ens).

141J. Löwenberg et al. / Desalination 357 (2015) 140–149

With PAC/UF pre-treatment resulting in the least fouling development the normalized flux decline remained high with roughly 40%–50% over 5 days of RO operation indicating the challenging nature of high DOC CTBD treatment for re-use.

© 2014 Elsevier B.V. All rights reserved.

1. Introduction

With only 0.8% of all water on earth being fresh water, several re- gions suffer from severe water scarcity. Traditional fresh water sources are diminishing due to over-exploitation and become saline by salt water intrusion in coastal areas. Even more challenging is the fact that in such water depleted areas industrial processes often need a consider- able amount of the available fresh water. Among these, large water con- sumers are cooling towers (CT) which discharge large volumes of wastewater in the form of cooling tower blow down (CTBD). Wang et al. [24] reported that 60%–70% of industrial fresh water demand is used in CT. In 2005, 49% of fresh water withdrawal in the U.S. was main- ly caused by usage in CT for thermoelectric power generation. Water treatment technologies such as desalination for fresh water production or water reuse are considered a key factor for sustainable development in the future [10,12]. The large water demand and the large volume of discharged wastewater by CT allow for large water saving potentials in industrial processes through CTBD re-use [4,30].

For instance You et al. [28] reported that wastewater from CT is in- valuable among industrial wastewaters due to its high volumes and rel- atively little contamination. Nevertheless, wastewater of cooling towers contains considerable amounts of particles, colloids and salts. Chemicals added to the cooling water, such as ammonia and phosphate for corro- sion control, are typical constituents of a CTBD [19]. In particular high amounts of ions such as Ca2+, Mg2+, CO3

2− and HCO3 − hinder the direct

reuse and make treatments such as desalination necessary [24]. Among the available membrane desalination processes are reverse osmosis (RO), nanofiltration (NF) and electro dialysis (ED). However, the use of these technologies requires pre-treatment due to the sensitivity of the membranes towards fouling [10,24]. RO desalination has increasing- ly been applied and surpassed thermal desalination in 2001 when RO accounted for 51% of total new desalination capacity [12]. Furthermore, recent research activities on CTBD have mainly focused on membrane pre-treatment to RO such as microfiltration (MF) and ultrafiltration (UF) [30]. MF and UF as porous membrane processes for pre- treatment offer several advantages such as short treatment times, low space demand, complete particle removal and constant permeate quality [10]. The applicability of MF and UF processes for CTBD pre- treatment was shown with both processes being able to supply permeate of low turbidity and silt density index (SDI) values to the subsequent dense membrane desalination process at high water recovery rates of up to 95% [24,31]. However, membrane processes are associated with membrane fouling which reduces the process perfor- mance, and thereby, increases energy demand and maintenance re- quirements. Dissolved organic carbon (DOC) is often considered to be a crucial factor for fouling, causing pore clogging and gel formation [2,16]. Due to the addition of several chemicals such as anti-scalants, corrosion inhibitors and bio-dispersants to the cooling water [4,29], the chemical composition of CTBD is different from other typical water sources.

While several studies investigated the feasibility of membrane pro- cesses for the re-use of CTBD, little information on the direct comparison of different pre-treatment technologies on fouling development and process performance of an RO treatment of CTBD is available. Therefore, within this study three different pre-treatment technologies were applied to investigate the influence on RO fouling and process perfor- mance. The CTBD of a CT close to the Dow Benelux' premises (Zeeland, The Netherlands) was used for the experiments. Zeeland is considered a water scarce region due to a high water demand by industry, agriculture

and households, exceeding the available freshwater sources. This makes the transport of fresh water over long distances necessary, raising fresh water costs and leading to intensified efforts to increase water manage- ment efficiency in the local industry [8].

Since the investigated CTBD is high in DOC concentration special at- tention was paid to the capability of the pre-treatment technologies to remove DOC from the water. Namely, the pre-treatment technologies investigated are UF, powdered activated carbon (PAC) adsorption and coagulation with FeCl3. Following the investigation of DOC reduction by the different technologies, the influence of coagulation and PAC ad- sorption coupled to UF as hybrid membrane processes on RO perfor- mance and fouling development was studied and evaluated based on normalized flux development and fouling analysis.

The study was carried out to investigate economically feasible treat- ment schemes for re-use of high DOC CTBD by RO and study the contri- bution of DOC removal towards a stable RO filtration and reduced fouling development.

2. Material and methods

2.1. Water quality

The water used for the experiments was taken from the blow down of a CT next to the Dow premises in Terneuzen (Netherlands). The CT is operated as a natural draft counter flow CT. The CTBD contained various chemicals added in the CT, including sulfuric acid for pH adjustment, phosphate as corrosion inhibitor, anti-scalants and biocides to prevent microbiological growth. A rough process scheme is given in Fig. 1.

2.2. Coagulation

Coagulation with FeCl3 (Tri-fer 200, Aregger Chemie, Switzerland) was performed in jar tests to investigate the influence on DOC profile and removal. In-line coagulation was applied when coupled to UF as in- dicated in Fig. 2.

The batch experiments were carried out with the Flocculator SW6 (Stuart, United Kingdom) in duplicates allowing the investigation of three parameter variations in parallel. Jars were filled with 2 L CTBD and placed in the Flocculator SW6. Fe3+ at concentrations of 10, 20 and 50 mg/L was simultaneously added to all flasks in the beginning of a fast stirring period of 1 min at 250 rpm. For the next 30 min a re- duced stirring speed of 25 rpm was applied to support flock-building. Samples of the supernatant were taken 2 cm from the surface after 60 min of sedimentation.

When coagulation was applied in combination with UF, Fe3+ was dosed into a small contact reactor with a volume of about 100 mL. Dos- age was performed with a membrane pump FMM 20 KPDC-P (KNF Neuberger GmbH, Germany) from a stock solution of FeCl3 and de- ionized water with a concentration of 0.28 g Fe3+/L with a pH of 3 in the stock solution.

2.3. Adsorption

Adsorption tests were carried out with PAC SAE Super (Norit Acti- vated Carbon, The Netherlands). The chosen PAC has a specific surface area of 1300 m2/g with a median diameter d50 of 15 mm. The iodine number and the methylene blue adsorption are 10.50 and 0.28 g/g, re- spectively [34]. Jars were filled with 200 mL CTBD and placed on a

Fig. 1. Cooling tower process scheme (concentration factor: 4–5; volume 1 million m3/y, make-up water: = sand filtered surface water).

142 J. Löwenberg et al. / Desalination 357 (2015) 140–149

magnetic stirrer plate (200 rpm) to yield a homogeneous distribution. To reach concentrations of 20, 50, 100 and 200 mg/L, PAC was dosed to each of the stirred batch contactors from a stock suspension. Samples were taken after 60 min contact time. The samples were filtered through a 0.45 μm nylon membrane filter (Titan Filtration Systems, U.S.) to remove PAC and prevent further adsorption.

When PAC addition was coupled to UF, the dosage was performed similarly to the coagulation experiments (see previous chapter) with PAC instead of Fe3+ as in-line dosage. The targeted PAC concentration was 50 mg/L.

Fig. 2. Process sche

2.4. Membrane process set-up

UF and RO experiments were performed on two separate installa- tions. Permeate from the UF process was collected until 25 L of perme- ate were produced.

The UF pretreatment was performed with a self-assembled UF unit allowing automatic operation with periodic backwashing. During the UF experiments, an Inge Multibore® membrane potted as a single fiber module (Inge GmbH, Greifenberg, Germany) was operated in dead-end filtration mode. RO filtration was performed with the OSMO

me of UF/RO.

Table 1 Operating conditions of membrane process.

Ultrafiltration Reverse Osmosis

Commercial product Inge Multibore® FILMTEC™ BW30FR Material Polyethersulfone (PES) Polyamide (PA) Nominal pore size [nm] 20 Operating mode Dead-end Cross-flow Flux [L/(m2·h)] 60 15–40 Filtration time [h] 0.33 120 Pressure [bar] 0.1–0.6 15 Recovery [%] 86 Full recirculation of permeate Membrane area [cm2] 90 40

143J. Löwenberg et al. / Desalination 357 (2015) 140–149

Inspector (Convergence Beheer BV, Enschede, The Netherlands) and a FILMTEC™ membrane BW30FR of Dow FILMTEC. The process scheme for the membrane filtration is given in Fig. 2.

The RO process was operated in cross-flow mode with a cross-flow velocity of 0.16 m/s as used by Vrouwenvelder et al. [23] to simulate RO fouling. The RO filtration was run continuously for 5 days at a con- stant feed pressure of 15 bar and full recirculation of the permeate and concentrate into the feed tank. With the adjusted operating param- eters the first 7.9 cm of an industrial RO process is simulated where foul- ing typically occurs and scaling is negligible [20].

The experiments were conducted on untreated CTBD, pre-treated with UF, UF coupled with Fe3+ addition (Fe3+/UF) and the combination of PAC addition and UF (PAC/UF). The experiments were carried out in double or triple under the same operating conditions.

Most relevant operating conditions and characteristics of the mem- brane processes are given in Table 1.

UF experiments were performed over two days each to determine the fouling tendency of the CTBD on the UF membrane and the DOC re- moval by UF. Backwash was carried out periodically every 20 min and chemical cleaning with NaOH and NaOCl (200 ppm active chlorine,

Table 2 Water quality of investigated CTBD in comparison to other CTBD from literature.

Parameter of CTBD waters North China [32] New Mexico, USA [4]

Sum parameters Temperature [°C] 4–40 TSS [mg/L] 5–80 Conductivity [μS/cm] 1650–2100 pH 7.8–9.3 BOD [mg/L] COD [mg/L] 0–10 Turbidity [NTU] DOC [mg/L]

Inorganic constituents PO4

3− [mg/L] Fetot [μg/L] Calcium [mg/L] 115 Magnesium [mg/L] 22 Chloride [mg/L] 90 Bicarbonate [mg/L] 401 Silica [mg/L] 116 Nitrate [mg/L] Mn [μg/L] F [μg/L] Ba [mg/L] Sr [μg/L]

Table 3 DOC analysis of CTBD over 1 year time.

DOCtot Biopolymers Humic substan

Average [mg/L] 58.7 (±5.9) 4.1 (±0.5) 37.2 (±2.3)

Standard deviations from n = 7.

pH 12) was applied when a transmembrane pressure (TMP) of 600 mbar was reached. The UF was backwashed for 1 min with a flux of 200 L/(m2·h·bar). Pressure decay tests were performed before and after each UF experiment at 800 mbar feed pressure and observation of pressure loss over 60 min. Experimental results were only considered if pressure loss over 60 min was less than 50 mbar.

2.5. Gravimetric analysis

For the measurement of the fouling layer weight, several RO mem- brane pieces were isolated after the experiments and compared to the weight of a virgin membrane. The membranes were dried for 2 h at 105 °C and spent an additional 30 min in an exsiccator before weighing. The analytic balance XS205 Dual Range with a readability of 0.01 mg of Mettler Toledo (Greifensee, Switzerland) was used.

2.6. Liquid chromatography-organic carbon detection (LC-OCD)

For the characterization of the organic matrix in the various process streams as well as the determination of their DOC concentration chromatographic analysis was performed using the liquid chromatogra- phy-organic carbon detection (LC-OCD) device (DOC Labor Huber, Germany). Measurement was performed as described elsewhere [14, 17].

2.7. Inductively coupled plasma mass spectroscopy (ICP-MS)

The inductively coupled plasma mass spectrometry (ICP-MS) device 7500cx (Agilent Technologies Inc., U.S.) was used for the quantification of metals in the membrane fouling/scaling layer. RO membrane pieces were soaked in aqua regia (65% HNO3 and 32% HCl, 1:3 v/v) over night and the solutions analyzed for the isotopes 57Fe, 24Mg and 44Ca.

Hebei, China [39] Hebei, China [40] CTBD This study (n = 6)

4–28.5 5.0–21.0 19.1 (±1.2) 7.0–222 12 ± 4.6

1300–2000 3,620 ± 810 8.2–8.8 8.2–8.75 7.9 (±0.2)

1.4 (±0.6) 2.9–4.8 3–4.8 2–33 5.0–23 7.3 ± 1.0

59.1 (±1.5)

0.72 5.9 (±1.0) 4.7–19.7 50–280 b100

481.7 (±17.2) 63.0 (±1.4) 500; n = 1 n.a.

28–109 50–200 0.9 (±0.2) 86.7 (±17.7) b100 840 L; n = 1 145 (±21.2); n = 2 1500 (±0); n = 2

ces Building blocks Low molecular weight (LMW) substances

6.0 (±2.4) 4.5 (±1.3)

Fig. 3. Influence on DOC profile by different pre-treatment technologies; (a) untreated CTBD; (b) Fe3+ coagulation at pH 5.5; (c) PAC adsorption; (d) ultrafiltration.

144 J. Löwenberg et al. / Desalination 357 (2015) 140–149

85Rb was used as an internal standard and analysis was performed as described elsewhere [33] pressurizing the octopole with 5.0 mL/min helium.

3. Results and discussion

3.1. Water quality

Water quality parameters of the CTBD used for the experiments are given in Table 2 in comparison to quality parameters of CTBD found in literature. All values of the CTBD used in this study were determined over a period of one year prior to the experiments while data for ad- vanced DOC characterization and investigation of seasonal variations were gathered over the period of one year during which the experi- ments were carried out.

The CTBD investigated in this study showed higher contaminations for various constituents such as magnesium (40–80 mg/L), calcium (350–600 mg/L), phosphate (5–15 mg/L) and chloride (400– 600 mg/L). Furthermore, the TSS concentration and turbidity of the CTBD were comparably low (Tables 2 and 3). However, its DOC concen- tration was higher (59.1 ± 1.5 mg/L) than generally reported in literature [29–31] while it is known that DOC concentration in cooling processes may reach DOC values of 120 mg/L or higher [9]. Therefore, the investi- gated CTBD may be considered to pose a larger challenge for treatment techniques than previously reported.

The results of the advanced DOC analysis with LC-OCD from May 2012 till June 2013 are given as average values (± standard deviation) in Table 3.

The CTBD DOC profile was fairly stable over the analyzed period with high concentrations of humic substances and a considerable amount of biopolymers (Table 2). Even though the ratio of biopolymers to total DOC is fairly low, a concentration of 4.1 mg/L is still considerably high since biopolymer concentration in wastewater treatment plant effluent was shown to be in the range of 1.1 ± 0.5 mg/L [17]. However, the high biopolymer concentration is not surprising considering the source of the make-up water being sand filtered surface water and concentration fac- tor in the cooling tower.

3.2. DOC removal by different technologies

3.2.1. Coagulation and settling Coagulation with 20 mg Fe3+/L and pH-adjustment to 7.5

(unchanged), 6.5 and 5.5 prior to Fe3+ dosage showed increasing DOC removal with decreasing pH. DOC removal was determined to be 16, 22 and 32% for pH of 7.5, 6.5 and 5.5 respectively.

Furthermore, subsequent coagulation experiments with Fe3+ con- centrations of 10, 20 and 50 mg/L at pH 5.5 showed an increasing DOC removal with Fe3+ dosage amounts of 11, 40 and 50 mg/L (Fig. 3(b)). DOC profile analysis with LC-OCD showed complete biopolymer remov- al as well as partial humic substance and building block removal at con- centrations above 20 mg Fe3+/L and pH 5.5 (Fig. 3(b)).

The higher DOC removal at lower pH values is in line with findings of other studies on coagulation with FeCl3 which also showed preferable removal of higher molecular weight organics such as biopolymers [15, 18,25] and propose charge neutralization as the main coagulation mechanism. This is also supported by findings of Siéliéchi et al. [22] who suggest sweep coagulation and adsorption not to be the main mechanisms for coagulation of humic substances in the pH range of 6–8.

However, after the sedimentation phase the turbidity of the super- natant was determined to be higher than the raw water at all pH values and Fe3+ concentrations investigated (Supplementary data; Tables S 1 and S 2). The increase of turbidity is supposed to be caused by the aggre- gates of hydroxylated iron (such as Fe(OH)2+) and organic matter which are not able to settle in the given sedimentation time and stay in suspension. This may be caused by chemicals such as anti-scalants

145J. Löwenberg et al. / Desalination 357 (2015) 140–149

present in the CTBD which negatively affect the formation of settable flocks. This phenomenon of a reduced sediment volume is also men- tioned by Siéliéchi et al. [22] who reported a shrinking of humic/hydrox- ylated iron complexes at certain low mixing conditions resulting in a reduced sediment volume. Therefore, while coagulation with FeCl3 re- sulted in the highest DOC removal, separation by sedimentation seemed to demand for additional efforts in order to improve settable flock formation.

3.2.2. PAC adsorption PAC adsorption resulted in lower DOC removal in comparison to co-

agulation. With PAC adsorption, a small removal of all DOC fractions with a preference towards LMW substance adsorption (Fig. 3(c); Table 4) was achieved. The preferable adsorption of LMW substances has been previously shown [5,17]. The achieved DOC removal by PAC was 6%, 7%, 10% and 20% at concentrations of 20, 50, 100 and 200 mg PAC/L, respectively, and a contact time of 30 min.

The specific DOC removal was in the range of 0.06–0.17 mg DOC/mg PAC which was in the range of previously reported removals [17]. There, specific adsorption at a dosage of 20 mg PAC/L was shown to be 0.08 mg DOC/mg PAC in wastewater treatment plant effluent [17]. The highest specific adsorption of 0.17 mg DOC/mg PAC was achieved at low PAC concentrations of 20 mg/L. This may indicate the presence of very hy- drophobic substances in the CTBD which readily adsorb on the PAC even at low concentrations.

3.2.3. Ultrafiltration During UF, the permeate quality parameters showed a strong reduc-

tion in turbidity while the DOC concentration remained unchanged (Fig. 3(d)). However, the analysis of the DOC profile of the backwash water showed a slightly higher concentration of biopolymers concen- tration (data not shown) indicating a slight removal of these substances. The UF of the raw CTBD caused a strong transmembrane pressure (TMP) increase during filtration cycles and the necessity for a high frequency chemical cleaning every 6–7 h of operation. Chemical cleaning was performed when a TMP of 600 mbar was exceeded. The TMP increase (Supplementary data: Fig. S 1) was higher than reported for other CTBD containing less DOC at similar operating conditions [29].

Based on the data, it was assumed that most organic substances in the CTBD are smaller than the pore size of the membrane, i.e. b20 nm. The quick fouling development during UF may be caused by slight re- tention of biopolymers which have been previously shown to cause ir- reversible membrane fouling [26,32].

3.3. Reverse osmosis desalination

3.3.1. Flux development Experiments carried out over five days each, showed that UF and

PAC/UF pre-treatment improved the RO performance (by on average 15% and 32%, respectively) in comparison to filtration of untreated CTBD (Fig. 4(a) and (c)). On the other hand, the pre-treatment by Fe3+ with subsequent UF showed no improvement of RO performance (Fig. 4(b)).

UF pre-treatment reduced the total suspended solid concentration and turbidity but had a negligible effect on DOC removal (Fig. 3(d)). Therefore, it can be assumed that suspended solids and colloids are mainly responsible for the difference in normalized flux decline

Table 4 Comparison of DOC removal and turbidity of different pre-treatment technologies.

Coagulation (50 mg/L; pH 5.5)

DOC removal [%] 50 Spec. DOC removal 0.65 [mg DOC/mg Fe3+] Turbidity [NTU] (after treatment) 26.9

a Indication of slight biopolymer removal by increased concentration in backwash water.

between untreated and UF pre-treated RO filtration. Furthermore, the experiments indicate that UF pre-treatment as a stand-alone process is not sufficient since RO performance is still poor with a normalized flux decline of around 60% over 5 days (Table 5).

Experiments with PAC/UF pre-treated CTBD (50 mg PAC/L) showed a further improvement of the RO filtration measured in normalized flux decline of less than 50% (Fig. 4(c), Table 5). Since it was shown that PAC adsorption resulted in a minor decrease of all DOC fractions with a preference of LMW substance adsorption (Fig. 3(c)), the results show the DOC influence on RO fouling during the filtration of the specific water. Furthermore, the beneficial effect of PAC adsorption on RO filtra- tion in combination with the preferable adsorption of LMW substance indicates the influence of this substance group in this specific water. In- vestigation of the chemical additives present in the CTBD revealed a bio- dispersant to be in this molecular weight range (data not shown) which is believed to have a considerable influence on membrane fouling.

The process combination of coagulation with 20 mg Fe3+/L and UF on CTBD prior to RO desalination did not result in an improvement concerning the normalized RO flux development compared to untreat- ed CTBD. Even though coagulation removed higher amounts of DOC from the water with almost complete biopolymer removal, the normal- ized flux development was similar to that observed during RO filtration on untreated CTBD (Fig. 4(b); Table 5). Analysis of the fouling layer composition (Fig. 6) revealed the deposition of residual iron on the membrane, which introduced a fouling mechanism not found relevant in the other experiments, thereby, over-compensating the effect of DOC reduction during pre-treatment. While batch experiments already indicated the weak sedimentation behavior of flocks produced during coagulation (increased turbidity of supernatant even after filtration), a certain portion of the added Fe3+ appeared to have remained in solu- tion and passed the UF membrane.

The comparison showed that stand-alone UF pre-treatment as well as the pre-treatment with the process combination PAC/UF resulted in higher amounts of permeate produced (Table 5). Assuming that fouling on the RO membranes is partly depending on the amount of permeate volume produced, the flux decline per total permeate volume is a more relevant indicator for the pre-treatment efficiency than the nor- malized flux decline alone. Concerning this parameter the PAC/UF pre- treatment resulted in an improvement of the process by 60% compared to untreated CTBD.

However, the laboratory experiments documented within this work demonstrate the challenging treatment of CTBD with high DOC concen- tration and indicate the potentially negative influence of chemical addi- tives on the RO process performance and pre-treatment efficiency. While PAC/UF treatment was found most suitable among the investigat- ed pre-treatment technologies, the normalized RO flux decline over 5 days was still high (38% and 46%) indicating further optimization po- tential and the need for piloting to investigate long-term behavior of the RO performance.

3.3.2. Membrane autopsy & fouling analysis Membrane autopsy confirmed the findings of the evaluation of RO

performance by normalized flux development. Decreasing weight of the fouling developed on the RO membrane after different pre- treatment processes were observed in the order untreated N Fe3+/UF pre-treated N UF pre-treated N PAC/UF pre-treated (Fig. 5(e)). While the fouling layer weight after the filtration of untreated CTBD was

PAC adsorption (200 mg/L) Ultrafiltration

20.4 ~0a

0.06 [mg DOC/mg PAC] n.a. n.a. 0.8

Fig. 4. Normalized flux development during RO on pretreated waters.

146 J. Löwenberg et al. / Desalination 357 (2015) 140–149

considerably higher than that of Fe3+/UF pre-treated CTBD, the normal- ized flux decline during filtration was similar (Table 5), indicating that the depositions on the membrane after Fe3+/UF pre-treatment caused a lighter but denser (less permeable) fouling layer.

A comprehensive fouling layer after operation on untreated CTBD (Fig. 5(a)) confirmed the fouling previously indicated by the strongest flux decline during RO filtration. The fouling layer found after the filtration of CTBD pre-treated with Fe3+ coagulation was not

Table 5 Operating results of RO after different pretreatments.

Flux decline after 5 days [%] Permeate produced [kg] Flux decline per permeate [%/kg]

Raw CTBD 1 (Exp. 1) 78 9.9 7.7 Raw CTBD 2 (Exp. 2) 68 10.8 6.3 UF (Exp. 1) 59 12.0 4.9 UF (Exp. 2) 57 12.9 4.4 UF (Exp. 3) 59 11.6a 5.1 PAC/UF (Exp. 1) 46 14.4 3.3 PAC/UF (Exp. 2) 38 17.6 2.2 Fe3+/UF (Exp. 1) 63 10.7 5.9 Fe3+/UF (Exp. 2) 71 8.6a 8.3 Fe3+/UF (Exp. 3) 63 9.4b 6.7

a 3 days and 22.5 h of operation. b 3 days and 17 hour of operation.

147J. Löwenberg et al. / Desalination 357 (2015) 140–149

as comprehensive while causing similar normalized flux decline (Fig. 5(c); Table 5).

The pre-treatment with UF as well as the combination of PAC/UF showed the least visible fouling layer (Fig. 5(b); (d)). Nevertheless, there was still some discoloring visible on both membranes.

3.3.3. Inorganic deposition in fouling layer Pre-treatment of the CTBD with 20 mg Fe3+/L resulted in a consider-

able iron deposition on the RO membrane (Fig. 6) indicating that resid- ual iron was responsible for the flux decline during RO filtration, even though high amounts of DOC were removed by the coagulation. Resid- uals from coagulation by Al3+ have been previously reported to cause severe membrane fouling by colloidal aluminum even at low concentra- tion of 50 μg/L [11]. Ohno et al. [21] report NF fouling by aluminum re- siduals at concentrations as low as 18 μg/L while they observed no effect of residual iron on NF performance. However, the catalyzing effect of Fe3+ on silica fouling and the capability of forming iron hydroxides, bridging with organic and inorganic matter causing a fouling layer were previously shown [7,27] (Fig. 6).

The fouling deposition after UF as well as the combination of PAC/UF showed the lowest content of Mg, Ca and Fe indicating that some inor- ganic colloids are retained by the UF membrane [3,13] as well as by the PAC/UF process (Fig. 6). It is assumed that Ca and Mg formed cationic bridges with organic matter during UF and PAC/UF pre-treatment and partly remained in the UF fouling layer (not analyzed) as previously shown in UF processes [1,6].

4. Conclusion

• UF as a stand-alone process was able to reduce turbidity and remove suspended solids from the CTBD thereby improving RO performance.

• PAC adsorption was capable of reducing all DOC fractions to a

Fig. 5. Fouling layer after 5 days of RO filtration on (a) untreated CTBD, (b) UF pretreatment, (c) brane and fouling layer (n = 3).

minor extent with a clear preference towards lower molecular weight organic substances. PAC adsorption coupled to UF was found to be the most suitable process combination to improve RO performance under the experimental conditions applied.

• Coagulation with Fe3+ was best suitable to reduce the DOC con- centration in the CTBD with complete biopolymer removal at suffi- cient Fe3+ dosage (50 mg/L). However, settlement of the agglomerates did not occur within 1 h and coagulation resulted in an increase in turbidity. The hybrid membrane process Fe3+/ UF was not able to improve the RO process mostly due to residual Fe3+ in the UF permeate which caused RO fouling during 5 days of filtration. Optimization of the coagulation towards reduction of residual dissolved Fe3+ or the addition of anti-scalants to bind re- sidual iron during RO is assumed to improve the process combina- tion due to high DOC removal by coagulation.

• The presence of chemical additives in the CTBD is believed to have caused the challenging coagulation as well as the improved RO performance after removal of a small fraction of the DOC (LMW substances) by PAC. The identification and the specific removal of chemical additives in high DOC CTBD is believed to be the key to feasible membrane treatment.

Acknowledgments

The research leading to these results has received funding from the European Union Seventh Framework Programme (FP7/2007–2013) under grant agreement no. 608490. The authors are grateful for the sup- port of this research within the E4Water project.

Furthermore, the authors thank Dominique Ritty (Institute for Ecopreneurship, University of Applied Sciences and Arts Northwestern Switzerland) for his valuable technical support commitment.

Fe3+/UF pretreatment and (d) PAC/UF UF pretreatment; (e) gravimetric analysis of mem-

Fig. 6. Average (± standard deviation) of element concentrations in fouling layer of reverse osmosis membrane (n = 3).

148 J. Löwenberg et al. / Desalination 357 (2015) 140–149

Appendix A. Supplementary data

Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.desal.2014.11.018.

References

[1] A.J. Abrahamse, C. Lipreau, S. Li, S.G.J. Heijman, Removal of divalent cations reduces fouling of ultrafiltration membranes, J. Membr. Sci. 323 (2008) 153–158, http://dx. doi.org/10.1016/j.memsci.2008.06.018.

[2] K.O. Agenson, T. Urase, Change in membrane performance due to organic fouling in nanofiltration (NF)/reverse osmosis (RO) applications, Sep. Purif. Technol. 55 (2007) 147–156, http://dx.doi.org/10.1016/j.seppur.2006.11.010.

[3] W.-Y. Ahn, A.G. Kalinichev, M.M. Clark, Effects of background cations on the fouling of polyethersulfone membranes by natural organic matter: experimental and mo- lecular modeling study, J. Membr. Sci. 309 (2008) 128–140, http://dx.doi.org/10. 1016/j.memsci.2007.10.023.

[4] S.J. Altman, R.P. Jensen, M.A. Cappelle, A.L. Sanchez, R.L. Everett, H.L. Anderson Jr., L.K. McGrath, Membrane treatment of side-stream cooling tower water for reduc- tion of water usage, Desalination 285 (2012) 177–183, http://dx.doi.org/10.1016/j. desal.2011.09.052.

[5] J. Altmann, A.S. Ruhl, F. Zietzschmann, M. Jekel, Direct comparison of ozonation and adsorption onto powdered activated carbon for micropollutant removal in ad- vanced wastewater treatment, Water Res. 55 (2014) 185–193, http://dx.doi.org/ 10.1016/j.watres.2014.02.025.

[6] S. Arabi, G. Nakhla, Impact of calcium on the membrane fouling in membrane biore- actors, J. Membr. Sci. 314 (2008) 134–142, http://dx.doi.org/10.1016/j.memsci.2008. 01.037.

[7] F.H. Butt, F. Rahman, U. Baduruthamal, Characterization of foulants by autopsy of RO desalination membranes, Desalination 114 (1997) 51–64, http://dx.doi.org/10. 1016/S0011-9164(97)00154-9.

[8] Cefic, Annual Report Europe, Responsible Care–Building Trust, Cefic, 2007. (http:// cefic-staging.amaze.com/Documents/ResponsibleCare/Annual%20reports/ cefic2007.pdf., accessed: 01.06.2014).

[9] M.R. Choudhury, M.-K. Hsieh, R.D. Vidic, D.A. Dzombak, Corrosion management in power plant cooling systems using tertiary-treated municipal wastewater as make- up water, Corros. Sci. 61 (2012) 231–241, http://dx.doi.org/10.1016/j.corsci.2012. 04.042.

[10] C. Fritzmann, J. Löwenberg, T. Wintgens, T. Melin, State-of-the-art of reverse osmosis desalination, Desalination 216 (2007) 1–76, http://dx.doi.org/10.1016/j.desal.2006. 12.009.

[11] C.J. Gabelich, K.P. Ishida, F.W. Gerringer, R. Evangelista, M. Kalyan, I.H. “Mel” Suffet, Control of residual aluminum from conventional treatment to improve reverse os- mosis performance, Desalination 190 (2006) 147–160, http://dx.doi.org/10.1016/j. desal.2005.09.002.

[12] L.F. Greenlee, D.F. Lawler, B.D. Freeman, B. Marrot, P. Moulin, Reverse osmosis desa- lination: water sources, technology, and today's challenges, Water Res. 43 (2009) 2317–2348, http://dx.doi.org/10.1016/j.watres.2009.03.010.

[13] Y. Hao, A. Moriya, Y. Ohmukai, H. Matsuyama, T. Maruyama, Effect of metal ions on the protein fouling of hollow-fiber ultrafiltration membranes, Sep. Purif. Technol. 111 (2013) 137–144, http://dx.doi.org/10.1016/j.seppur.2013.03.037.

[14] S.A. Huber, A. Balz, M. Abert, W. Pronk, Characterisation of aquatic humic and non- humic matter with size-exclusion chromatography-organic carbon detection–

organic nitrogen detection (LC-OCD–OND), Water Res. 45 (2011) 879–885, http:// dx.doi.org/10.1016/j.watres.2010.09.023.

[15] G. Korshin, C.W.K. Chow, R. Fabris, M. Drikas, Absorbance spectroscopy-based exam- ination of effects of coagulation on the reactivity of fractions of natural organic mat- ter with varying apparent molecular weights, Water Res. 43 (2009) 1541–1548, http://dx.doi.org/10.1016/j.watres.2008.12.041.

[16] S. Lee, W.S. Ang, M. Elimelech, Fouling of reverse osmosis membranes by hydrophil- ic organic matter: implications for water reuse, Desalination 187 (2006) 313–321, http://dx.doi.org/10.1016/j.desal.2005.04.090 (Integrated Concepts in Water Recycling Integrated Concepts in Water Recycling).

[17] J. Löwenberg, A. Zenker, M. Baggenstos, G. Koch, C. Kazner, T. Wintgens, Comparison of two PAC/UF processes for the removal of micropollutants from wastewater treat- ment plant effluent: process performance and removal efficiency, Water Res. 56 (2014) 26–36, http://dx.doi.org/10.1016/j.watres.2014.02.038.

[18] A. Matilainen, M. Vepsäläinen, M. Sillanpää, Natural organic matter removal by co- agulation during drinking water treatment: a review, Adv. Colloid Interf. Sci. 159 (2010) 189–197, http://dx.doi.org/10.1016/j.cis.2010.06.007.

[19] M.S. Mohsen, Treatment and reuse of industrial effluents: case study of a thermal power plant, Desalination 167 (2004) 75–86, http://dx.doi.org/10.1016/j.desal. 2004.06.115 (Desalination Strategies in South Mediterranean Countries).

[20] T. Melin, R. Rautenbach, Membranverfahren - Grundlagen der Modul- und Anla- genauslegung. 3. Auflage, Springer-Verlag, Berlin/Heidelberg, 2007.

[21] K. Ohno, Y. Matsui, M. Itoh, Y. Oguchi, T. Kondo, Y. Konno, T. Matsushita, Y. Magara, NF membrane fouling by aluminum and iron coagulant residuals after coagulation– MF pretreatment, Desalination 254 (2010) 17–22, http://dx.doi.org/10.1016/j.desal. 2009.12.020.

[22] J.-M. Siéliéchi, B.S. Lartiges, G.J. Kayem, S. Hupont, C. Frochot, J. Thieme, J. Ghanbaja, J.B. d' Espinose de la Caillerie, O. Barrès, R. Kamga, P. Levitz, L.J. Michot, Changes in humic acid conformation during coagulation with ferric chloride: implications for drinking water treatment, Water Res. 42 (2008) 2111–2123, http://dx.doi.org/10. 1016/j.watres.2007.11.017.

[23] J.S. Vrouwenvelder, J.A.M. van Paassen, L.P. Wessels, A.F. van Dam, S.M. Bakker, The membrane fouling simulator: a practical tool for fouling prediction and con- trol, J. Membr. Sci. 281 (2006) 316–324, http://dx.doi.org/10.1016/j.memsci. 2006.03.046.

[24] Z. Wang, Z. Fan, L. Xie, S. Wang, Study of integrated membrane systems for the treatment of wastewater from cooling towers, Desalination 191 (2006) 117–124, http://dx.doi.org/10.1016/j.desal.2005.04.125 (International Congress on Mem- branes and Membrane Processes International Congress on Membranes and Mem- brane Processes).

[25] J. Xie, D. Wang, J. van Leeuwen, Y. Zhao, L. Xing, C.W.K. Chow, pH modeling for max- imum dissolved organic matter removal by enhanced coagulation, J. Environ. Sci. 24 (2012) 276–283, http://dx.doi.org/10.1016/S1001-0742(11)60717-1.

[26] H. Yamamura, K. Okimoto, K. Kimura, Y. Watanabe, Hydrophilic fraction of natural organic matter causing irreversible fouling of microfiltration and ultrafiltration membranes, Water Res. 54 (2014) 123–136, http://dx.doi.org/10.1016/j.watres. 2014.01.024.

[27] S.G. Yiantsios, D. Sioutopoulos, A.J. Karabelas, Colloidal fouling of RO mem- branes: an overview of key issues and efforts to develop improved prediction techniques, Desalination 183 (2005) 257–272, http://dx.doi.org/10.1016/j. desal.2005.02.052 (Desalination and the Environment Desalination and the Environment).

[28] S.-H. You, D.-H. Tseng, G.-L. Guo, J.-J. Yang, The potential for the recovery and reuse of cooling water in Taiwan, Resour. Conserv. Recycl. 26 (1999) 53–70, http://dx.doi. org/10.1016/S0921-3449(98)00075-5.

149J. Löwenberg et al. / Desalination 357 (2015) 140–149

[29] H. Zeng, J. Zhang, C. Ye, Comparison of an ultrafiltration membrane fed with raw seawater, coagulated seawater and cooling tower blowdown, Desalination 244 (2009) 199–207, http://dx.doi.org/10.1016/j.desal.2008.04.044.

[30] J. Zhang, L. Chen, H. Zeng, X. Yan, X. Song, H. Yang, C. Ye, Pilot testing of outside-in MF and UF modules used for cooling tower blowdown pretreatment of power plants, Desalination 214 (2007) 287–298, http://dx.doi.org/10.1016/j.desal.2006. 12.004.

[31] J. Zhang, H. Zeng, C. Ye, L. Chen, X. Yan, Pilot test of UF pretreatment prior to RO for cooling tower blowdown reuse of power plant, Desalination 222 (2008) 9–16, http://dx.doi.org/10.1016/j.desal.2007.01.123 (European Desalination Society and Center for Research and Technology Hellas (CERTH), Sani Resort 22–25 April 2007, Halkidiki, Greece European Desalination Society and Center for Research and Technology Hellas (CERTH), Sani Resort).

[32] X. Zheng, M. Ernst, P.M. Huck, M. Jekel, Biopolymer fouling in dead-end ultrafiltra- tion of treated domestic wastewater, Water Res. 44 (2010) 5212–5221, http://dx. doi.org/10.1016/j.watres.2010.06.039.

[33] Y.-S. Zimmermann, A. Schäffer, P.F.-X. Corvini, M. Lenz, Thin-film photovoltaic cells: long-term metal(loid) leaching at their end-of-life, Environ. Sci. Technol. 47 (2013) 13151–13159, http://dx.doi.org/10.1021/es402969c.

[34] B. Zwickenpflug, M. Böhler, B. Sterkele, A. Joss, H. Siegrist, J. Traber, W. Gujer, M. Behl, F. Dorusch, J. Hollender, T. Ternes, G. Fink, Einsatz von Pulveraktivkohle zur Elimination von Mikroverunreinigungen aus kommunalem Abwasser (Use of pow- dered activated carbon for the removal of micropollutants from municipal waste- water), Final Report of the MicroPoll Project, EAWAG on Behalf of the Swiss Federal Office for the Environment, Dübendorf, CH2010. (http://www.eawag.ch/ forschung/eng/schwerpunkte/).