Investigation of environmental influences on membrane biofouling in a Southern California desalination pilot plant

Desalination 319 (2013) 1–9

Contents lists available at SciVerse ScienceDirect

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

Investigation of environmental influences on membrane biofouling in a Southern California desalination pilot plant Siqian Huang a, Nikolay Voutchkov b, Sunny C. Jiang a,⁎ a b

Department of Civil and Environmental Engineering, University of California, Irvine, CA 92697, USA Water Globe Consulting, LLC, 200 Broad Street, Stamford, CT 06901, USA

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

• This study develops a seawater membrane desalination biofouling indicator. • Membrane performance was related to environmental and water quality parameters. • Turbidity and SDI are inadequate to indicate membrane performance decline. • We identify chlorophyll as a sensitive indicator for membrane biofouling. • We show membrane fouling monitor is a useful tool for biofouling analysis.

a r t i c l e

i n f o

Article history: Received 6 June 2012 Received in revised form 29 December 2012 Accepted 11 March 2013 Available online 26 April 2013 Keywords: Seawater desalination RO membrane Biofouling indicator Biofilm Chlorophyll Total organic carbon

a b s t r a c t One of the challenges the seawater desalination industry faces today is reverse osmosis (RO) membrane biofouling. Traditional water quality parameters such as SDI and the RO feed water turbidity are inadequate at protecting the membrane from biofouling. This research investigated the environmental and water quality parameters in a Southern California desalination plant in order to develop a set of seawater desalination RO membrane biofouling indicators. Statistical analysis was performed on data collected onsite over two years. The relationships between operation parameters, rain precipitations, TOC, UV254, chlorophyll fluorescence in raw seawater and the performance loss of the RO desalination process are presented. The environmental triggers for accelerated RO membrane biofouling was further investigated by developing membrane fouling simulators at the desalination pilot plant. Biofouling was confirmed by confocal laser scanning microscopy investigation of membrane biofilm and live and dead bacterial cell counts. The results of this study indicated that biofouling was significantly correlated with water quality changes. Thus, chlorophyll fluorescence measurements can be used as a precursor for desalination membrane biofouling. © 2013 Elsevier B.V. All rights reserved.

Abbreviations: CLSM, Confocal Laser Scanning Microscope; DOC, dissolved organic carbon; EPS, extracellular polymeric substances; MF, microfiltration; NDP, net driving pressure; NTU, Nephelometric turbidity unit; RO, reverse osmosis; SDI, silt density index; SWRO, seawater reverse osmosis; TEP, Transparent extracellular polysaccharides; TOC, total organic carbon; UF, ultrafiltration. ⁎ Corresponding author. Tel.: +1 949 824 5527; fax: +1 949 824 3672. E-mail address: [email protected] (S.C. Jiang). 0011-9164/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.desal.2013.03.016

2

S. Huang et al. / Desalination 319 (2013) 1–9

1. Introduction One of the challenges the seawater desalination industry faces today is RO membrane fouling. Depending on its severity, it may have a measurable impact on the economics and reliability of freshwater production by desalination. The advances in pretreatment technologies, i.e. microfiltration and ultrafiltration, although have significantly reduced the inorganic fouling, organic and biofouling continue to plague desalination industry. Biofouling is caused by biofilm formation on the RO membrane surface by bacteria and their metabolites that naturally occur in the feed seawater. Biofilm is an organic film composed of live and dead microorganisms embedded in a polymer matrix, consisting of extracellular polymeric substances (EPS) such as polysaccharides, proteins and lipids [1]. The biofilm formation on the membrane surface increases the pressure needed for maintaining steady production of freshwater by the membrane elements [2]. In order to compensate for loss of productivity due to biofouling, the feed pressure of the seawater RO (SWRO) membrane system would need to be increased, which in turn would result in elevated energy requirement to produce the same volume of freshwater. Currently, seawater pretreatment methods such as oxidant-based disinfection, ultraviolet irradiation, and coagulation followed by granular media or membrane filtration can reduce the number of bacteria in the RO feed seawater significantly, but would typically not eliminate biofilm formation on the RO membranes [3]. Empirical experience from desalination plants' operators reveal that SWRO membrane fouling follows a seasonal cycle, which implies that environmental and water quality factors have strong influences on the membrane fouling potential. Previous studies on water reuse RO membrane biofouling suggest that the growth of biofilm-forming bacteria is dependent on the concentration of dissolved organic carbon (DOC) as energy and carbon sources in feed water [4]. However, typical seawater has a much lower concentration of DOC than natural freshwater and treated wastewater for reuse, and the effect of low concentration of organic nutrient on the establishing biofilm and the growth of biofilmforming microorganisms in seawater has not been well documented. Recent studies in biofilm-forming microorganisms on the SWRO membrane [5–7] revealed that the seawater biofouling microorganisms were very different than those in the wastewater and freshwater environment, and that these bacteria may respond to different triggers in the environment. So far, there were few attempts to investigate the correlation between the biofilm build up rate and the SWRO membrane operating parameters and the feed water quality index. Yet understanding such relationships may hold the key to predict and control the biofouling event. A readily available biofouling indicator can trigger preventive cleaning and regeneration of membrane productivity. This study collected environmental and water quality parameters in a Southern California desalination plant and presented statistical analysis of relationships between rain precipitations, total organic carbon (TOC), chlorophyll fluorescence in raw seawater and the performance loss of RO desalination process. The environmental triggers for accelerated RO membrane biofouling was further investigated by developing and deploying membrane fouling simulators at the desalination pilot plant. The results of this study indicated that biofouling was significantly correlated with water quality changes. Chlorophyll measurements can be used as a precursor for desalination membrane fouling. 2. Materials and methods 2.1. Desalination plant operation data The study was conducted at Carlsbad desalination pilot plant, a co-generation plant in north San Diego County, California. The facility uses the discharging seawater from Encinitas Power Plant's cooling

system as the intake water. As a result, the temperature of the intake water fluctuates significantly depending on the operation of the power plant. The pilot plant uses HYDRAcap Capillary Ultrafiltration Modules (Hydranautics, Oceanside, California) as the pretreatment of the SWRO process. The UF membrane surface polymer is hydrophilic polyethersulfone with nominal molecular weight cutoffs (MWCO) of 150 kDa to produce permeate turbidity of b 0.07 NTU. The SWRO system has two stages. Each stage has 2 pressure vessels containing 3 spiral-wound SWC5 RO membrane elements (Hydranautics). The typical recovery of the plant was 50%, and the permeate flow rate was between 2.3 and 4.6 m 3∙h−1 during normal operation. No chemical cleaning was performed during the test period. The pilot plant only conducted RO system flushing with RO permeate on a regular basis. To indicate the loss of RO membrane performance due to fouling, we compared the membrane performance of the aged membrane with that of the new membrane, termed fouling indicator (FI). FI is computed as: FI = R/Rnew; where R is the resistance of the membrane that has been in operation for greater than 1 week; Rnew is the average resistance of a new membrane that has been in operation for less than 1 week. The membrane resistance R is computed using: R = (NDP × TCF)/Flux, where NDP is net driving pressure, TCF is temperature correction factor and Flux is permeate flux. Based on FILMTEC Elements Technical Manual [8], NDP is calculated using the following equation:

NDP ¼ P f −

" # ΔP fc C fc −P p −πf pf −ð1−RÞ 2 Cf

Where Pf is feed pressure, ΔPfc is concentrate-side pressure drop, Pp is permeate pressure, πf is feed osmotic pressure, Cfc is average concentrate-side concentration, Cf is feed concentration, pf is concentration polarization factor, and R is salt rejection fraction. The empirical TCF for T ≤ 25 °C as defined by FILMTEC [8] 1 1 TCF ¼ e2640ð298−Tþ273Þ is used in the study because water temperature never exceed 25 °C at the study site. Rnew was found to be 0.059 kPa ∙ m − 1∙ s through calculation of the first week operation data of new membranes. FI is unit-less because it is the ratio of two R in kPa∙m−1∙s. Plant operation parameters including RO feed pressure, temperature, conductivity and flux were collected from daily data logger. The study was carried out in two phases. Phase I study analyzed pre-recorded plant operation data provided by the operator, encompassing the period between January 2008 and April 2009. In addition to SWRO performance indicator (FI), UF silt density index (SDI) and raw water turbidity (NTU) from the plant operation record were obtained and used in the data analysis. Water pH during the study period were constant and they were not included in the final analyses. Water conductivity was included in the computation of NDP as the component of feed osmotic pressure. The phase II study was carried out between July 2010 and July 2011, and all operation and water quality parameters were collected in real time. In addition to the UF SDI and raw water NTU, UF filtrate turbidity, raw water UV254 absorbance (Hach, Loveland, Colorado) were also taken onsite during the phase II study. 2.2. Environmental and water quality data To better characterize the relationship between the SWRO performance and environmental factors, additional environmental data including daily rain precipitation, chlorophyll fluorescence, and TOC were collected. Daily precipitation data were retrieved from a local weather station in Carlsbad (KCACARLS5) from www.wunderground. com for both phase I & II studies. Chlorophyll fluorescence from Scripps Institute of Oceanography (SIO) Pier were used as the approximation of the concentrations at the pilot plant for phase I study since no real time measurement was conducted onsite. SIO pier is located approximately

S. Huang et al. / Desalination 319 (2013) 1–9

3

slide and observed under a CLSM (Zeiss LSM 510 META). Two excitation/ emission wavelengths were used for the two florescent stains: 488 nm/ 500 nm for SYTO 9 and 510 nm/635 nm for propidium iodide. Images were captured at each wavelength and composited into one final image. The Z sectioning method was used to determine the thickness of the biofilm. Bacterial colonization was evaluated by visually counting the number of cells attached to the membranes surface, and was determined by the average counts from 10 images for each sample.

40 km south of the Carlsbad desalination pilot plant. The chlorophyll fluorescence at the SIO pier were collected by SIO as part of the Southern California Coastal Observing System (SCCOOS) using automated shore stations with continuous fluorescence data logging. The data were retrieved from the SCCOOS website at http://www.sccoos. org/data/habs. For phase II study, chlorophyll fluorescence were determined fluorometrically using the Turner Designs AquaFluor handheld fluorometer (Turner Designs, Sunnyvale, California) using EPA method 445.0 [9]. For TOC analysis, both raw water and UF filtrates were collected bi-weekly from the pilot plant and were stored frozen in a − 80 °C freezer until analysis. In the lab, 10× dilutions were made before testing by mixing 2 ml of seawater sample into 18 ml DI water. The GE Sievers 5310 C On-Line Total Organic Carbon (TOC) Analyzer (GE Instruments, Boulder, Colorado) which has a detection range of 4 parts per billion (ppb) to 50 parts per million (ppm) TOC, was used for the TOC analysis.

Multivariate regression statistical analysis was performed using STATA 10 (StataCorp, College Station, Texas). In the regression model, FI was used as the dependent variable and chlorophyll, SDI, UV254, precipitation and turbidity etc. were used as independent variables.

2.3. Bio-monitor system set up

3. Results

To directly monitor the biofilm production, a flat sheet RO membrane bio-monitor system was set up to run in parallel with the spiral wound SWRO system in the pilot plant using a side-stream of the UF-pretreated RO feed (Fig. 1) during the phase II study. The bio-monitor consists of a 5 × 20 cm flat sheet SWC5 membrane (Hydranautics) placed directly on top of a plastic support, the feed channel is separated by a single piece of 0.87 mm thick nylon membrane spacer. The cross section area of the feed channel is 44 mm 2. The system was maintained at 42 kPa feed pressure without permeation with a cross flow rate at 0.023 m 3∙h −1, which was designed to simulate the cross flow velocity of the spiral wound RO element in the pilot plant. The flat sheet membrane was removed from the monitor for analysis every two weeks and replaced with a new membrane each time. After each removal, the system was cleaned and rinsed with tap water. The bio-monitor was first set up in July, 2010. A total of 24 membranes were collected and analyzed during the study period.

3.1. Relationship between RO performance and water quality

2.4. Biofilm thickness and bacterial cell density by confocal laser scanning microscopy (CLSM) The retrieved flat sheet membranes from the bio-monitor were examined for biofilm thickness and total bacterial cell counts on the membrane surface by CLSM. Briefly, membranes were cut into 1 cm × 1 cm squares and stained using SYTO 9 green-fluorescent nucleic acid stain and propidium iodide red-fluorescent nucleic acid stain (FilmTracer™ LIVE/DEAD Biofilm Viability Kit, Invitrogen, Carlsbad, CA) for 30 min. The stained membranes were then mounted onto a glass

2.5. Statistical analysis

3.1.1. Phase I study Fig. 2 shows the 16-month data plot of Carlsbad pilot plant performance and water quality parameters between January 2008 and April 2009. The SWRO elements in the pilot plant were not new at the beginning of the data collection. Thus FI values were always greater than 1 during the period. FI showed fluctuation from the lowest 1.1 in Sep. 2008 to the highest 1.8 in Feb. 2008 and Mar. 2009 (Fig. 2A). An increase in raw water turbidity (Fig. 2B) and an elevation of UF SDI (Fig. 2C) were observed in May 2008 but did not trigger any observable change in RO performance. Increases of chlorophyll fluorescence were observed in early spring, around April 2009 (Fig. 2D). Precipitation records (Fig. 2E) showed a significant rainfall event in February 2009. However, the pilot plant was shut down for maintenance during that period. Multivariate regression analysis of FI against all water quality and environmental variables (excluding the pilot shutdown period) showed that the FI was significantly positively correlated with chlorophyll measurements in the coastal water (Table 1). However, the FI did not have a significant correlation with the turbidity of intake water nor the RO feed water SDI. Direct correlation analysis with rainfall was also performed but no significant correlation was detected (data not shown). 3.1.2. Phase II study The temporal plots of FI and environmental parameters between July 2010 and July 2011 are shown in Fig. 3. Phase II study was initiated

Fig. 1. Schematic of membrane bio-monitor system set up in Carlsbad desalination pilot plant.

4

S. Huang et al. / Desalination 319 (2013) 1–9

A

2.5

FI

2 1.5

1 0.5 0 3-Jan

B

3-Mar

2-May

1-Jul

30-Aug

29-Oct

28-Dec

26-Feb

27-Apr

26-Jun

8

Raw water turbidity NTU

6 4 2 0 3-Jan

C

3-Mar

2-May

1-Jul

30-Aug

29-Oct

28-Dec

26-Feb

27-Apr

26-Jun

2-May

1-Jul

30-Aug

29-Oct

28-Dec

26-Feb

27-Apr

26-Jun

4

UF SDI 3 2 1 0 3-Jan

D

3-Mar

8

RFU

6

Chlorophyll a

4 2 0 3-Jan

3-Mar

2-May

1-Jul

30-Aug

29-Oct

28-Dec

26-Feb

27-Apr

26-Jun

1-Jul

30-Aug

29-Oct

28-Dec

26-Feb

27-Apr

26-Jun

E 80.0

mm

60.0

Percipitation

40.0 20.0 0.0 3-Jan

3-Mar

2-May

Fig. 2. Temporal plots of environmental, water quality and operation parameters collected daily between January 2008 and April 2009 at Carlsbad desalination pilot plant.

with a set of new SWRO elements, thus, the FI increased slowly from 1.0 at the beginning of the RO membrane operation in July 2010 to around 2.0 in July 2011, indicating a graduate decline of the RO performance over the period (Fig. 3A). The average FI was at 1.4 for 10 of the 12 months study period, which was similar to the phase I study when the membrane resistance has reached equilibrium. A sudden jump in FI occurred in January 2011, approximately 10 days after a series of heavy rain events between late December 2010 and early January 2011 (Fig. 3B). On January 19th, the FI increased to 2.7, which was more than doubled the membrane resistance in the previous month (1.4), and reached as high as 3.0 later in February 2011. The significant

Table 1 Multivariate regression model outputs for phase I study between January 2008 and April 2009. In the regression model RO performance (FI) was used as the dependent variable and chlorophyll, SDI and NTU were used as independent variables. FI

Coef.

Std. Err.

t

P>|t|

[95% Conf. Interval]

Chlorophyll UF SDI NTU Constant

2.010141 0.178005 −0.39462 40.625

0.982526 1.804918 0.875764 3.858434

2.05 0.1 −0.45 10.53

0.043 0.922 0.653 0

0.066875 −3.39181 −2.12673 32.99369

3.953407 3.747818 1.337493 48.25631

S. Huang et al. / Desalination 319 (2013) 1–9

increase in membrane resistance indicated that the RO membrane element was fouled rapidly during the short period. The pilot plant changed out the unrecoverable fouled RO membrane elements, on

A

5

February 21, 2011, and the FI dropped back to normal operational value afterwards (Fig. 3A). Several gaps in the RO fouling index graph were due to the maintenance shutdowns of the pilot plant.

4

FI Membrane replacement

3

2

1

0 15-Jul

B

80

13-Sep

12-Nov

11-Jan

12-Mar

11-May

10-jul

13-Sep

12-Nov

11-Jan

12-Mar

11-May

10-jul

13-Sep

12-Nov

11-Jan

12-Mar

11-May

10-jul

Precipitation

mm

60 40 20 0 15-Jul

C

0.6

UV254 0.4

0.2

0 15-Jul

D

5.0 4.0

UF SDI

3.0 2.0 1.0 0.0 15-Jul

13-Sep

12-Nov

11-Jan

12-Mar

11-May

10-jul

11-Jan

12-Mar

11-May

10-jul

E 3000 Raw water TOC ppb

2000

1000

0 15-Jul

13-Sep

12-Nov

Fig. 3. Temporal plots of environmental, water quality and operation parameters collected daily (TOC data collected every other week) between July 15, 2010 and July 30, 2011 at Carlsbad desalination pilot plant.

6

S. Huang et al. / Desalination 319 (2013) 1–9

F 30.0 Turbidity NTU

20.0

10.0

0.0 15-Jul

13-Sep

12-Nov

11-Jan

12-Mar

11-May

10-jul

12-Nov

11-Jan

12-Mar

11-May

10-jul

12-Nov

11-Jan

12-Mar

11-May

10-jul

G 1.50 Filtrate Turbidity NTU

1.00

0.50

0.00 15-Jul

H

13-Sep

6

Chlorophyll a RFU

4

2

0 15-Jul

13-Sep

Fig. 3 (continued).

The UV254 readings also increased dramatically during the December rain events (Fig. s3C), indicating that rainfall brought in additional organic contents from land runoffs to the lagoon used for pilot plant intake. The elevated TOC in the feed water was confirmed by the grab sample TOC analyses (Fig. 3E). The accelerated membrane fouling that lead to membrane replacement was recorded after the raw water TOC exceeded 2000 ppb. In addition, the raw water turbidity (Fig. 3F) and chlorophyll fluorescence (Fig. 3H) also indicated the influence of rainfall and land runoffs on the feed water quality. Yet, UF SDI readings (Fig. 3D) and UF filtrate turbidity readings (Fig. 3G) did not reflect the influence of rainfalls during the study period. Multivariate regression analysis (Table 2) indicated that there was a significant correlation between the FI and the parameters of raw water turbidity (NTU), UV254 and chlorophyll fluorescence. Grab sample TOC concentrations were not included in the multivariate analysis due to differences in the time point measurements collected. Grab samples were taken every two weeks while the operational data were continuous. Although rain was the trigger for the environmental and water quality parameter changes, there was no correlation between the volume of rainfall and RO FI due to the delayed effect of rain and the cascade of events following rainfalls. The small volume or regional scale rainfalls that were observed in October 2010 and March 2011 did not trigger dramatic changes in

water quality as indicated by UV254 and raw water turbidity (Fig. 3C, F). However, raw water TOC from the grab samples and chlorophyll measurements were more sensitive to the rain events. Elevations of TOC concentration were detected following September 2010 and March 2011 rainfall as well as the less than 20 mm of rainfall in May 2011 (Fig. 3E). However, TOC concentrations were less than 1400 ppb for all samples except one. Chlorophyll measurements were elevated in early spring of 2011 reflecting low level of algal blooms (Fig. 3H). Although no dramatic fouling event was detected during this period, the decline of RO performance was observed throughout the spring and the FI reached 2.0 at the beginning of the summer (Fig. 3A). 3.2. Biofilm thickness and bacterial density Examination of flat sheet RO membranes retrieved from the bio-monitor using CLSM revealed significant variability in biofilm density and cell counts over the study period (Fig. 4). To link the RO performance deterioration with membrane biofouling, biofilm thickness and total bacterial cell counts on membranes retrieved every two weeks from the bio-monitor were plotted on the temporal scale (Fig. 5). The results showed the elevations of biofilm thickness and total bacterial cell counts on the bio-monitor membranes post major rain events (Fig. 5). Significantly higher numbers of cells were observed

S. Huang et al. / Desalination 319 (2013) 1–9

7

Table 2 Multivariate regression model outputs for phase II study between July 2010 and July 2011. In the regression model RO performance (FI) was used as the dependent variable and the rest of parameters were used as independent variables. FI

Coef.

Std. Err.

t

P>|t|

[95% Conf. Interval]

NTU UV254 Chlorophyll UF SDI Rain Constant

1.868747 −143.961 2.199339 −0.41695 −1.19568 16.26946

0.680119 48.52036 0.989939 0.544669 1.831825 1.04439

2.75 −2.97 2.22 −0.77 −0.65 15.58

0.007 0.004 0.029 0.446 0.516 0

0.515524 −240.501 0.229671 −1.50067 −4.84044 14.19146

on the membrane collected on October 24, 2010, between December 23, 2010 and January 26, 2011 and on March 2, 2011, indicating biofouling on the membrane over the two-week exposure. Regression analysis also showed that the total cell counts were significantly correlated (P b 0.1) with the TOC readings (Fig. 3E) in the feed seawater taken in the sampling period. 4. Discussion 4.1. Biofouling indicators SDI and turbidity are the two parameters that are widely used in desalination plants [3]. Both of them can be obtained by the onsite operator easily and quickly. In the membrane water treatment industry, an SDI of less than 5 and filtrate turbidity of less than 1 NTU is considered acceptable for the operation of most RO systems [10]. In

A

B

C

Fig. 4. Confocal laser scanning microscopy images of flat sheet membranes collected from the bio-monitor. A: 2-D image of membrane collected on Sept. 2, 2010; B: 2-D image of membrane collected on Jan. 6, 2011. C: 3-D image constructed by Z stacking for membrane collected on Mar. 2, 2011. Live cells are stained green and dead cells are stained red.

3.22197 −47.4208 4.169007 0.666774 2.449076 18.34747

this study, the Carlsbad desalination pilot plant had SDI readings between 1.8 and 4, and the filtrate NTU readings around 0.05 in both Phase I and Phase II studies, expanding over 28 months. Neither of the parameters were good indicators of the SWRO performance in the Carlsbad pilot plant as revealed by the lack of statistical correlation with the performance decline in the RO process. This result confirmed the practical experience by the operators on unpredictability of membrane fouling [11,12]. The failure of SDI and filtrate turbidity to indicate membrane performance is because biofouling is the main cause of the RO performance decline in UF pretreated feed water rather than the particulate fouling on the RO membrane [13]. Multiple investigators have suggested diverse types of biological fouling indicators including ATP analysis, total cell counts, lead element pressure drop measurement, and membrane fouling simulators, these methods either require a well-trained technical personnel for analysis or expensive equipment or set up that is not easily available to the desalination plant operation [14–16]. So far, there has not been a widely available, fast, simple, and real-time monitoring method that can serve as an indicator for membrane biofouling. The independent analysis of two sets of data at Carlsbad pilot plant showed that raw water turbidity (NTU) had a significant correlation with FI in one set of data but not the other. This result indicates that the raw water turbidity is not a reliable indicator because raw water turbidity increase can be caused by multiple factors. Following storm events, the particulates in the high turbidity water can be removed by UF pretreatment, while increase in DOC, which is often a confounding factor of turbidity elevation in coastal water due to urban runoff, cannot be removed by the pretreatment. Thus, depending on the cause of raw water turbidity elevation, it may or may not be responsible for membrane biofouling. UV254, a surrogate for organic carbon concentration, although showed a good correlation with RO FI in the phase II study, it was insensitive to the change of organic carbon under 1000 ppb. Thus, it may miss the condition that causes the progressive fouling due to organic accumulation and subsequent bacterial colonization on membrane surfaces. Grab sample TOC analysis using a high sensitivity TOC analyzer provided the most accurate information on membrane biofouling potential as indicted by the significant positive correlation with bacterial cell density on the membrane retrieved every two weeks. TOC results also indicated the switch from progressive fouling at TOC less than 1500 ppb to the accelerated fouling at TOC greater than 2000 ppb in California coastal water. The relationship between TOC concentration and the accelerated biofouling rate may be location dependent. Tampa Bay Desalination plant intake water has much higher concentration of baseline TOC without significant fouling until the heavy rainy season [17]. The baseline TOC in Tampa Bay water may contain a large portion of recalcitrant organic carbon not available for biological growth. Experimental studies have also reported that severe biofouling was observed only when assimilable organic carbon (AOC, dosed as acetate) exceeded 80 μg Ac-C/l [18]. Thus, it is important to establish a local threshold for TOC and fouling rate to protect the membrane operation. Although the importance of a routine TOC analysis for fouling protection is recognized by membrane fouling studies [19–21], high

8

S. Huang et al. / Desalination 319 (2013) 1–9

A

8.00E+04

Live Cell counts

6.00E+04

4.00E+04

2.00E+04

0.00E+00 15-Jul

29-Aug 13-Oct 27-Nov 11-Jan 25-Feb 11-Apr 26-May 10-Jul

B

6.00E+04

Dead Cell counts 4.00E+04

2.00E+04

0.00E+00 15-Jul

29-Aug 13-Oct 27-Nov 11-Jan 25-Feb 11-Apr 26-May 10-Jul

C

1.50E+05

Total Cell counts 1.00E+05

5.00E+04

0.00E+00 15-Jul 25

um

20

29-Aug 13-Oct 27-Nov 11-Jan 25-Feb 11-Apr 26-May 10-Jul

D Biofilm Thickness

15 10 5 0 15-Jul

29-Aug 13-Oct 27-Nov 11-Jan 25-Feb 11-Apr 26-May 10-Jul

Fig. 5. Biofilm thickness and total/live/dead bacterial cells on the surface of the RO membrane collected from the flat sheet bio-monitor. Error bars represent replication of ten different imaging fields.

sensitivity online TOC analyzers are expensive and are not available for most of the desalination plants. It also requires technical training for sample analysis and system maintenance. In comparison, chlorophyll measurements based on relative fluorescence provides instantaneous readings that can be obtained by either an in situ online monitoring system or a handhold fluorometer. As demonstrated in this study, chlorophyll fluorescence is sensitive to water quality changes and is positively correlated with RO membrane FI, suggesting the potential of a new sensitive indicator for biofouling prevention. Chlorophyll is an indirect measurement of phytoplankton (algae) population in the water. The relationship between phytoplankton blooms and SWRO performance has been implied in early studies [22,23]. A recent review by Caron et al. [24] emphasized the impact of algal toxin on desalinated water, while the fouling potential triggered by algal excreted transparent extracellular polysaccharides (TEP) is the main concern of Berman [25]. Marine algae were recognized as the main producer of TEP, several folds higher in concentration than

in the freshwater system [25,26]. In addition to TEP, cell lyses in the decline phase of algal bloom also contribute to the elevation of DOC in feedstock. Pressurized pretreatments such as MF and UF, escalate DOC in the filtrate due to breaking of fragile phytoplankton cells and releasing cellular content directly into the filtrate [27]. Furthermore, the phytoplankton bloom in the coastal water is often followed by heterotrophic bacterial bloom, which leads to an increase of bacterial loading in the feed water (Jiang et al. unpublished results). Regardless of the mechanism of fouling caused by the increase of phytoplankton density, chlorophyll can serve as a precursor to indicate the membrane biofouling potential. Vardon et al. [28] concerned the variability of bulk chlorophyll fluorescence and proposed to use flow cytometry, an expensive instrument, for direct measurement of algal cells in the coastal water. The result of our study is a good field demonstration of the application of chlorophyll fluorescence to predict membrane biofouling. Since the direct measurement of chlorophyll fluorescence is the easiest and the least expensive online approach for detecting the changes in water

S. Huang et al. / Desalination 319 (2013) 1–9

quality, it is the best option for a readily available biofouling indicator for operating desalination plants. 4.2. Biofouling monitor The results of this study further demonstrated that the membrane fouling monitor is a useful tool for biofouling analysis and detection, as it is cost effective comparing to RO membrane element autopsy. Several studies have used similar devices to simulate the RO membrane element performance and demonstrated high reliability and a good representation of the fouling simulator to the real RO membrane [14–16]. The result of the biofilm cell density and thickness clearly showed the accelerated biofilm formation on the simulator RO membrane after the major rain events. Significant correlation between the TOC concentration and the bacteria cell counts were observed during the test period, which leads to the conclusion that the membrane biofouling monitor is sensitive in detecting accelerated biofilm formation. The monitor system responded about one week earlier than the change of FI in spiral wound membrane element, suggesting that the fouling monitor can be deployed as an early warning system of biofouling events in the RO membrane process due to their small size and fast response rate. However, the monitor will not detect the graduate RO performance decline and cumulative biofouling effect. Fouling monitors should be operated in conjunction with readily available fouling indicator such as chlorophyll fluorescence. 5. Conclusions • Statistical analysis of relationships between operation parameters, precipitations, TOC, UV254, chlorophyll in raw seawater and the performance loss of RO desalination process indicated that biofouling was significantly correlated with water quality changes. • Filtrate turbidity and SDI are inadequate to indicate performance decline in the SWRO process. Chlorophyll measurements can be used as a rapid and sensitive precursor for desalination membrane fouling. • The results of this study further demonstrated that the membrane fouling monitor is a useful tool for biofouling analysis and detection, as it is cost effective comparing to RO membrane element autopsy. Acknowledgments The following groups and individuals are acknowledged for their support and contributions to this project: Daniel Marler and Steve LePage at the Carlsbad Desalination Pilot Plant; Steven Peck at Hydranautics. We thank Leda Katebian for proof reading the manuscript. Financial support for this work was provided by WateReuse Research Foundation award WRF-08-19. References [1] H.C. Flemming, T.R. Neu, D.J. Wozniak, The EPS matrix: The “house of biofilm cells”, J. Bacteriol. 189 (2007) 7945–7947. [2] H. Flemming, G. Schaule, T. Griebe, J. Schmitt, A. Tamachkiarowa, Biofouling the Achilles heel of membrane processes, Desalination 113 (1997) 215–225. [3] AWWA Research Foundation, Lyonnaise des eaux-Dumez (Firm), South Africa, Water Research Commission., Water treatment membrane processes, McGraw-Hill, New York, 1996.

9

[4] A. Rochex, J.M. Lebeault, Effects of nutrients on biofilm formation and detachment of a Pseudomonas putida strain isolated from a paper machine, Water Res. 41 (2007) 2885–2892. [5] M.L. Zhang, S. Jiang, D. Tanuwidjaja, N. Voutchkov, E.M.V. Hoek, B.L. Cai, Composition and variability of biofouling organisms in seawater reverse osmosis desalination plants, Appl. Environ. Microbiol. 77 (2011) 4390–4398. [6] H. Ivnitsky, I. Katz, D. Minz, G. Volvovic, E. Shimoni, E. Kesselman, R. Semiat, C.G. Dosoretz, Bacterial community composition and structure of biofilms developing on nanofiltration membranes applied to wastewater treatment, Water Res. 41 (2007) 3924–3935. [7] C.M. Pang, W.-T. Liu, Community structure analysis of reverse osmosis membrane biofilms and the significance of Rhizobiales bacteria in biofouling, Environ. Sci. Technol. 41 (2007) 4728–4734. [8] FILMTEC RO Elements Technical Manual, Dow Water and Process Solutions, USA, http://www.dowwaterandprocess.com/support_training/literature_manuals/ filmtec_manual.htm2009. [9] Method 445.0 In Vitro Determination of Chlorophyll a and Pheophytin a in Marine and Freshwater Algae by Fluorescence, United States Environmental Protection Agency, 1997. [10] Hydranautics RO membrane element SWC5 specification, HydranauticsOceanside, CA, USA, 2007, http://www.membranes.com/docs/8inch/SWC5-LD.pdf. [11] A. Alhadidi, A.J.B. Kemperman, R. Schurer, J.C. Schippers, M. Wessling, W.G.J. van der Meer, Using SDI, SDI+ and MFI to evaluate fouling in a UF/RO desalination pilot plant, Desalination 285 (2012) 153–162. [12] J.S. Choi, T.M. Hwang, S. Lee, S. Hong, A systematic approach to determine the fouling index for a RO/NF membrane process, Desalination 238 (2009) 117–127. [13] J. Leparc, S. Rapenne, C. Courties, P. Lebaron, J.P. Croué, V. Jacquemet, G. Turner, Water quality and performance evaluation at seawater reverse osmosis plants through the use of advanced analytical tools, Desalination 203 (2007) 243–255. [14] J.S. Vrouwenvelder, S.a. Manolarakis, J.P. van der Hoek, J.a.M. van Paassen, W.G.J. van der Meer, J.M.C. van Agtmaal, H.D.M. Prummel, J.C. Kruithof, M.C.M. van Loosdrecht, Quantitative biofouling diagnosis in full scale nanofiltration and reverse osmosis installations, Water Res. 42 (2008) 4856–4868. [15] J.S. Vrouwenvelder, M.C.M. van Loosdrecht, J.C. Kruithof, Early warning of biofouling in spiral wound nanofiltration and reverse osmosis membranes, Desalination 265 (2011) 206–212. [16] J.S. Vrouwenvelder, J.a.M. van Paassen, J.S. Vrouwenvelder, J.a.M. van Paassen, J.C. Kruithof, M.C.M. van Loosdrecht, Sensitive pressure drop measurements of individual lead membrane elements for accurate early biofouling detection, J. Membr. Sci. 338 (2009) 92–99. [17] L.A. Weinrich, E. Giraldo, M.W. LeChevallier, Development and application of a bioluminescence-based test for assimilable organic carbon in reclaimed waters, Appl. Environ. Microbiol. 75 (2009) 7385–7390. [18] J.S. Vrouwenvelder, D.V.D. Kooij, Diagnosis of fouling problems of NF and RO membrane installations by a quick scan, Desalination 153 (2002) 121–124. [19] A. Matin, Z. Khan, S.M.J. Zaidi, M.C. Boyce, Biofouling in reverse osmosis membranes for seawater desalination: phenomena and prevention, Desalination 281 (2011) 1–16. [20] M. Herzberg, M. Elimelech, Biofouling of reverse osmosis membranes: role of biofilm-enhanced osmotic pressure, J. Membr. Sci. 295 (2007) 11–20. [21] R. Schneider, L. Ferreira, P. Binder, E. Bejarano, K. Goes, E. Slongo, C. Machado, G. Rosa, Dynamics of organic carbon and of bacterial populations in a conventional pretreatment train of a reverse osmosis unit experiencing severe biofouling, J. Membr. Sci. 266 (2005) 18–29. [22] B. Kwon, N. Park, J. Cho, Effect of algae on fouling and efficiency of UF membranes, Desalination 179 (2005) 203–214. [23] D.A. Ladner, D.R. Vardon, M.M. Clark, Effects of shear on microfiltration and ultrafiltration fouling by marine bloom-forming algae, J. Membr. Sci. 356 (2010) 33–43. [24] D.A. Caron, M.-E. Garneau, E. Seubert, M.D.A. Howard, L. Darjany, A. Schnetzer, I. Cetinić, G. Filteau, P. Lauri, B. Jones, S. Trussell, Harmful algae and their potential impacts on desalination operations off Southern California, Water Res. 44 (2010) 385–416. [25] T. Berman, Biofouling: TEP — a major challenge for water filtration, Filtr. Sep. 47 (2010) 20–22. [26] E. Bar-Zeev, I. Berman-Frank, B. Liberman, E. Rahav, U. Passow, T. Berman, Transparent exopolymer particles: potential agents for organic fouling and biofilm formation in desalination and water treatment plants, Desalin. Water Treat. 3 (2009) 136–142. [27] N. Voutchkov, Considerations for selection of seawater filtration pretreatment system, Desalination 261 (2010) 354–364. [28] D.R. Vardon, M.M. Clark, D.A. Ladner, The potential of laser scanning cytometry for early warning of algal blooms in desalination plant feedwater, Desalination 277 (2011) 193–200.