Volatile fatty acids concentration in real wastewater by forward osmosis

Author’s Accepted Manuscript Volatile fatty acids concentration in real wastewater by forward osmosis Gaetan Blandin, Bàrbara Rosselló, Victor M. Monsalvo, Pau Batlle-Vilanova, Jose M. Viñas, Frank Rogalla, Joaquim Comas www.elsevier.com/locate/memsci

PII: DOI: Reference:

S0376-7388(18)31662-4 https://doi.org/10.1016/j.memsci.2019.01.006 MEMSCI16776

To appear in: Journal of Membrane Science Received date: 17 June 2018 Revised date: 13 November 2018 Accepted date: 5 January 2019 Cite this article as: Gaetan Blandin, Bàrbara Rosselló, Victor M. Monsalvo, Pau Batlle-Vilanova, Jose M. Viñas, Frank Rogalla and Joaquim Comas, Volatile fatty acids concentration in real wastewater by forward osmosis, Journal of Membrane Science, https://doi.org/10.1016/j.memsci.2019.01.006 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Volatile fatty acids concentration in real wastewater by forward osmosis 1,*

1

3

3

3

3

Gaetan Blandin , Bàrbara Rosselló , Victor M. Monsalvo , Pau Batlle-Vilanova , Jose Mª Viñas , Frank Rogalla , Joaquim Comas

1, 2

1

LEQUIA, Institute of the Environment, University of Girona, Spain

2

ICRA, Catalan Institute for Water Research, Girona, Spain

3

FCC Aqualia, Innovation and Technology Department, Av. del Camino de Santiago, 40, edificio 3, 4ª planta,

Madrid, Spain *Corresponding author: [email protected]

Abstract Forward osmosis (FO) was studied as a concentration step for volatile fatty acids (VFA, especially acetic acid) to optimise downstream microbial desalination cell (MDC). First, it was demonstrated that water concentration factor (WCF) of wastewater (WW) above 10 (15-30) is achievable with seawater or brine as draw solution and similar flow for feed and draw solution. It was also observed that VFA rejection by FO membrane is highly connected to pH. At pH= 7.5 rejection rates above 80% are achievable working with domestic WW and therefore concentration of VFA by FO process is realistic. Nevertheless, concentration of VFA present in pre-treated real domestic WW proved to be more challenging with regards to fouling /biofilm formation which favours the biodegradation of VFA. Thanks to fouling mitigation (WW pre-filtration with microfiltration membrane and interbatch osmotic backwashing) and biodegradation strategies implemented (by applying N2 sparging and avoiding air contact with the WW), VFA concentration (and especially acetic acid) from 60-80 up to 300-400 mg·L-1 is possible. Maintaining high permeation flux, high VFA recovery and concentration factor of VFA during 20 batches of operation was achieved, being more difficult the stable operation of the FO concentration process at high WCF and VFA concentration.

Keywords: forward osmosis; water reuse; desalination; volatile fatty acids; acetic acid; concentration

1 Introduction In the overall growing context of water scarcity, alternative methods to safely produce water at the lowest costs from seawater desalination or wastewater (WW) reuse are highly studied [1-4]. 1

Moreover, integrating WW treatment and desalination in one plant can also result in potential economic benefits by synergistically lowering water intake costs and optimising energy efficiency of water treatment [5, 6]. In that context, salinity gradient technologies such as forward osmosis (FO) have been studied to take advantage of the osmotic difference of both streams and proved to be beneficial both from the energetic side but also to produce high quality water and to lower operational costs [7-9]. Similar concepts of salinity gradient based processes were applied to the colocation of desalination and WW treatment streams using pressure retarded osmosis (PRO) or reverse electro-dialysis (RED) towards energy and water harvesting [6, 10, 11]. In the meantime, another innovative process called microbial desalination cell (MDC), has been developed by modifying microbial electrosynthesis cell (MEC) to take advantage of the energy content from WW, where electricity is produced thanks to the degradation of organic matter by bioelectrogenic bacteria in the anode chamber [12-14]. The electric current obtained is used to promote ions migration through ion exchange membranes in a self-sustained desalination system. In MDC, in comparison with MEC, a third channel fed with the saline stream is coupled in-between cationic and anionic compartments separated by ionic exchange membranes. During MDC process, ions are removed from seawater and the resulting low salinity effluent can be further polished in a downstream RO system operated at low pressure and energy consumption. Nowadays, upscaling of MDC is one of the technical challenges that are under evaluation in water desalination. Using MDC as pre-treatment of RO, the EU funded MIDES project aims at producing safe drinking water at energy consumption below 0.5 kWh·m-3 [15], and demonstrate the feasibility of this technology in different demonstration sites after process optimisation, design and development of electrodes and membranes. Another technical barrier of MDC implementation is to provide sufficient biodegradable organic matter from the WW stream to the bacteria growth in the anodic chamber. In order to optimise the availability of easily utilizable organic matter by bioelectrogenic bacteria, WW can be pre-treated in hydrolytic anaerobic bioreactor, where the hydrolysis of organic matter takes place to break it down to highly biodegradable volatile fatty acids (VFA), such as acetic acid, to be used as feed in the MDC. However, the VFA concentration achieved depends on the WW source, which is insufficient for an optimal operation of the MDC (> 300 mg/L VFA, acetic acid). Forward osmosis has proven to be of interest for the treatment of complicated streams [16] and could be adapted to such environment where a saline stream is locally available [17]. Thanks to its low operating pressure and high rejection, FO exhibits low fouling propensity while allowing the 2

production of high water permeate quality and concentration of most other feed components (or contaminants) [18-20]. Potential application in many WW streams such as raw WW with the implementation for example of osmotic membrane bioreactor, concentration of activated sludge, treatment of landfill leachate and concentrate of anaerobic digester or for potable water reuse are envisioned [7, 16, 19, 21-27]. Still, recent works proved that higher fouling rate is expected when operating at increasing flux [28]. Therefore further work is requested regarding fouling and its mitigation especially using real WW streams. High rejection of organic compounds is commonly accepted with FO membranes and therefore would make it an adapted technology to VFA concentration. However, most studies were dedicated to emerging organic contaminants [29] while only few information is related to VFA recovery and concentration. Among the few studies dedicated to FO and VFA, one demonstrated that sludge preconcentration (10 times) by FO allowed for 4.4 times more production of VFA during downstream anaerobic digestion but significant amount of COD was lost during aeration [30] and, in that case, FO concentration was upstream VFA production and therefore VFA rejection by FO was not assessed. Another more specific study evaluated VFA rejection by FO membranes [31]; rejection above 90% was achieved but only for pH above 5 (above pKa of dissociation of tested VFA). Whilst the results are promising, a mixed solution of VFA was used and therefore specific rejection of each VFA has not been defined so far. To act as an efficient process to concentrate VFA in WW for downstream MDC process optimisation, FO should achieve (1) high water concentration factor (WCF), (2) high rejection of VFA and (3) sustainable operation with real WW to avoid fouling and VFA biodegradation. The aim of this study was to assess all those aspects through a step by step methodology. In a first step, FO potential to extract water from WW (WCF) was assessed using different feeds and draw salinities. VFA rejection by FO membrane was then evaluated using synthetic VFA solutions and for different pH. Concentration factor, VFA rejection, fouling propensity as well as fouling mitigation and adapted cleaning procedures were assessed then using real domestic WW. Finally, long term operation test (20 batches) was performed using real domestic WW and applying optimised operating conditions from previous steps.

3

2 Materials and method 2.1 FO concentration setup It has been decided to operate the system using similar initial feed and draw volume (flow) as a first option as it is a realistic hypothesis in the context of combining water reuse and desalination and in the specific context of the study. Also, the system was operated in batch mode (i.e. draw dilution with time; not constant draw concentration as in other studies) as it is more representative of the tested application and allows for an estimation of the achievable concentration rate. Similar setups and operating conditions were used as in our former studies [28, 32]. New generation of thin film composite (TFC) commercially available FO membranes obtained from Porifera Inc. (Hayward, CA, USA) was used with active layer facing the feed solution [33]. The filtration cell was provided by Metacrilats Futura, (Banyoles, Spain) following our requested design and featured a membrane surface area of 82 cm2 Peristaltic pump Model LabM3 from Baoding Schenchen precision pump (Baoding city, China) with double pump head that could be operated in the range of 100-1000 mL.min-1 was used for both feed and draw circulation. 1.2 mm thickness diamond-type polypropylene mesh spacer were used both in feed and draw channels, pump flowrate of 0.3 L.min-1 (i.e. cross flow velocity of 0.1 m.s-1) for both feed and draw channels operated in co-current flow. Due to the specificity of the concentration and to avoid dead volumes, a 1 L decantation cone was used as feed container (Figure 1). 1 L of feed and draw solution were prepared and the system was operated in batch until no permeation flux was observed (osmotic equilibrium). The water flux (Jw) crossing the membrane from the feed to the draw was determined by measuring the increase of mass of the draw solution over time, using a Kern PCB 6000-1 balance (Balingen, Germany). Salt content of the feed solution as a function of time was determined from the feed solution conductivity measured with a Crison conductimeter probe obtained from Hach lange Spain (L’Hospitalet de Llobregat, Spain) using a NaCl-conductivity calibration curve and was used to calculate reverse salt diffusion (RSD, Js) [32]. All data were recorded using a Bluetooth based system provided by Instrument works (Waterloo, Australia); the conductivity sensor was connected to a Bluetooth Arduino Controller and the balance to a Bluetooth-RS232 Adaptor. Visualization, acquisition and storage of all data was realised thanks to the Dataworks data app installed on an Apple Ipod (Cupertino, CA, USA).

4

Figure 1: Schematic of the FO setup

2.2 Water concentration factor Given the specific need of this study, dedicated indicators were used with regards to achievable concentration and associated filtration time. The first indicator is the WCF that represents the concentration of the feed solution over time and is calculated using the following equation: WCF

Equation 1

With V0 the feed initial volume and V(t) the feed volume at any moment (t) during the filtration. WCF at the end of the filtration (when osmotic equilibrium was reached) was especially of interest. Also, given the fact that the initial goal of this study was to concentrate VFA to a factor of 5, it is a requirement to concentrate water by a factor of 5, therefore the time to achieve a WCF of 5 (WCF=5) was also used as a comparative filtration indicator. The impact of feed and draw solution salinities on permeation flux, achievable WCF and on time to reach WCF=5 were first assessed using clean feed and draw solutions. Achieved WCF were experimentally obtained when osmotic equilibrium was reached; i.e. no more permeation flux. DI 5

water, tap water, 0.5 g.L-1 and 1 g.L-1 salinity (0 to 1500 µS.cm-1 conductivity range) were tested as feed solution (with 35 g.L-1 SW as draw) and then 5, 20, 35 and 70 g.L-1 as draw solution and tap water as feed. All saline solutions were prepared using sea salts (>99.4% NaCl) provided by Vicens i Batllori S.L. (Banyoles, Spain).

2.3 Wastewater All filtration tests with WW were carried out using real domestic WW pre-treated in an hydrolytic anaerobic reactor working in a domestic WW treatment plant operated by FCC Aqualia in Spain. These samples were received and tested through the duration of the study and will be named as (real) WW from now on. On the received samples, a pre-treatment with microfiltration was also tested; Microfiltration was achieved using submerged plates obtained from Kubota (MF cartridges 203). All samples were stored in a fridge after reception and until further use for FO filtration.

2.4 Fouling tests Draw solution of 70 g.L-1 synthetic seawater, mimicking seawater brine, was used following recommendations of concentration tests and given that such stream is available in seawater RO desalination plant. Fouling behaviour was assessed following a methodology described in former study by operating the system for 4 consecutive batches [28, 34]. Each time, 1 L of feed and draw solution were prepared and the system was operated until no permeation flux was observed (osmotic equilibrium). Then new feed and draw solution were prepared. Longer filtration time and initial flux decrease over batches attest for potential fouling behaviour. In addition, cleaning procedure was evaluated; it consisted in osmotic backwashing [34, 35] (DI water as draw, 35 g.L-1 seawater as feed during 1 hour, co-current flow, cross flow velocity= 0,1 m.s-1 ) followed by a flushing of both feed and draw channel at high cross flow velocity (0.3 m.s-1) during 1 minute with DI water. Osmotic backwashing was evaluated either as final treatment (after 4 batches) or implemented in between batches (intermediate backwashing). Flushing allows for removal of particles that were dissociated from the membrane during osmotic backwashing as well as cleaning of both channels where salinity was reversed [28]. The efficiency of the cleaning procedure was evaluated by measuring the flux before and after cleaning using clean solutions (i.e., DI water as feed and 70 g.L-1 SW as draw) and comparing it with the initial flux of the new membrane before fouling. As alternative to cleaning in

6

place, manual cleaning was also tested and consisted in removing the membrane from the filtration cell and rinsing it with a spray of water and slight scrub by hand wearing laboratory gloves.

2.5 VFA analysis VFA (and especially acetic acid) concentration is the objective of the FO pre-concentration step of this study. Typically, it is expected to increase the VFA concentration of the treated domestic WW from 60-80 to > 300 mg.L-1. For all tests, samples were first filtered at 0.2 µm; then 85 µL of crotonic acid (standard) and 100 µL phosphoric acid (stabilisation) were added to 1.5 mL of filtered sample and stored in the fridge. VFA analyses were performed by the analytical service of the university of Girona (STR) using an Agilent 7890A (Agilent Technologies, US) gas chromatograph (GC) equipped with a DB-FFAP column and a flame ionisation detector (FID). The concentration of VFA in the liquid phase is given in mg.L-1. Even if no VFA biodegradation has been noticed in the prepared samples even after several weeks, all samples were typically analysed within 3 days. At first, concentration tests were run using DI water spiked with 80 mg.L-1 of 5 selected VFA (acetic acid, propionic acid, butyric acid, isobutyric acid and valeric acid) and 35 g.L-1 draw solution. Rejection and concentration tests were realised first without pH adjustment (pH=4) and then after pH adjustment (pH=7.5, which corresponds to the typical pH range of the real WW). 4 samples were collected through the duration of the rejection/concentration tests; rejection average values and standard deviation are reported. Permeation flux was maintained at 15±2 L.m-2.h-1 during the test. VFA analyses performed in all received WW samples were highly variable and generally lower than expected (Acetic acid concentration from 15 to 95 mg.L-1) possibly as a result of initial concentration variation and/or (different degree of) biodegradation during transport of WW samples from the full scale plant to the lab. Therefore, it has been decided to spike systematically all the samples with a mix of each VFA at a concentration of 80 mg.L-1 to allow for normalised operation and to study the behaviour of 5 VFA (not just acetic acid). VFA concentration of all samples was analysed after spiking and before each filtration tests. In parallel, we observed that VFA were also partly consumed even after spiking in WW and storage in the fridge leading probably also to some bacterial development. VFA (and acetic acids especially) are highly appreciated by microorganisms and therefore a great care has to be put to limit conditions of bacterial growth that can on one side lead to degradation of VFA and on the other side to the 7

development of microorganisms during the storage that can then after consume VFA during the filtration. Thus for further tests with VFA in WW the following procedure has been implemented: (1) MF filtration of WW samples after reception of the samples to remove s before each filtration test solids, (2) sparging with nitrogen in the filtered WW to remove oxygen, (3) storage of the samples in the fridge, (4) spiking with VFA just before filtration tests, (5) coverage of the feed tank to limit contact with air. VFA rejection (RVFA) has been calculated based on feed concentration and as described in Equation 2. RVFA

Equation 2

With CVFA(t) the VFA concentration and WCF (t) the WCF at the moment of sampling time t, and CVFA(i) the initial VFA concentration in the feed solution. Also, similarly to WCF the VFA concentration factor (VFA CF) was calculated using Equation 3: VFA CF

Equation 3

2.6 Long term operation Long term evaluation was then conducted through 20 consecutive batches. All batches were performed to reach a minimum of WCF=5 which is the minimum requirement to increase the VFA concentration of the treated domestic WW from 60-80 to > 300 mg.L-1. Based on observations collected during initial tests (WCF, fouling and VFA analysis and rejection), the optimised following operating conditions were chosen: draw solution of 70 g.L-1 to allow for higher concentration factor/shorter filtration time, MF pre-filtration and interbatch osmotic backwashing to limit fouling and potential VFA biodegradation. Samples storage and VFA spiking following previous section recommendations was applied to avoid potential VFA biodegradation. VFA concentration was systematically analysed at the beginning and the end of the filtration batches both on the feed and draw solutions; VFA concentration over time during some selected batches was also assessed. Water permeation fluxes, time to reach WCF=5, final WCF were monitored as before. After batch 8, manual (hand) cleaning of the membrane was performed; even if no damage was visible, membrane salt selectivity dropped during batch 9, maybe as consequence of damages during cleaning, and therefore the membrane was replaced before starting batch 10. Two batches per day 8

were generally performed but with slightly different filtration time for practical reasons: one during the day (8-11h) and one overnight (11-14h). Fouling development over batches also impacted filtration time to reach minimum WCF=5. As such, the final concentration factor may vary significantly and be much higher during overnight tests when filtration time was longer. For each test, the feed sample was taken out of the fridge few hours before filtration so to start the filtration at constant temperature (20°C).

3 Results and discussion 3.1 Initial evaluation with pure water 3.1.1 Flux and water concentration factor The impact of draw concentration and feed water salinities on permeation flux, achievable WCF and time to reach WCF=5 were systematically studied (Figure 2). Achievable WCF were experimentally obtained when osmotic equilibrium was reached (i.e. no more permeation flux) as a consequence of draw dilution over time, salt concentration in the feed and RSD passage [36].

(a) 2000

30

1500 20 1000 10

500

0

0 0

50

Draw concentration (g.L-1)

100

Jw (L.m-2.h.-1) & Achieved WCF

30

800

(b) 25

700 600

20

500

15

400 300

10

200 5

Time WCF=5 (min)

2500

Time WCF=5 (min)

Jw (L.m-2.h-1) & Achievable WCF

40

100

0 0

500

1000

0 1500

Initial feed conductivity (µS.cm-1)

Figure 2: Impact of (a) draw concentration (tap water as feed) and (b) initial feed conductivity (35 g.L1 SW as draw) on initial permeation flux (Jw), achieved WCF and filtration time required to reach WCF=5. At first and as expected, draw solution salinity proved to be a key parameter (Figure 2a); the highest the draw solution salinity the highest the flux [28, 37] and the achieved WCF. Both 35 and 70 g.L-1 draw concentration allowed for a high permeation flux and minimised filtration time. On one hand, 9

high WCF was achieved, i.e. 13.4 and 33 using 35 and 70 g.L-1 draw synthetic SW respectively and tap water as feed solution (Figure 2a) confirming the feasibility to use FO to reach high concentration rate. On the other hand, it was not possible to reach WCF=5 when using 5 g.L-1 draw solution. Even when using 20 g.L-1 draw solution, achieved WCF remains close to the minimum objective of WCF=5 and therefore operation with higher salinity draw (above 35 g.L-1) to ensure process feasibility is preferred (Figure 2a). In the range of salinity tested, increasing feed water conductivity did not have a major impact on the initial permeation flux (Figure 2b) but significantly reduced the achieved WCF (Figure 2b). Such behaviour was expected since salts present in the feed solution concentrate, increasing the feed salinity and consequently the osmotic equilibrium is reached more rapidly. Typically, when moving from 0 (DI water) to 1500 µS.cm-1 (1 g.L-1), the achieved WCF was reduced by more than 2 units but still remained around 8 for 1500 µS.cm-1 feed conductivity which is important in the context of this study. Importantly also, even when using the highest draw concentration (70 g.L-1) and reaching the highest WCF (33), final feed conductivity remained at 20mS. In practice, such salinity level would probably not be reached since such high WCF is not required but is not incompatible with MDC operation where microorganisms can tolerates very high salinity streams [38-40]. The use of brine produced in RO SW desalination plants as draw solution (70 g.L-1 draw) is especially interesting in the context of this study as it represents a waste from the most widely implemented desalination process and therefore its usage does not alter the productivity of the desalination plant and furthermore, dilution of that RO brine will also be in favour of reducing the impact of its discharge in the environment [41, 42]. In practice and on large scale operation, apart from feed and draw solution concentrations, achievable WCF will be dependent from the ratio of feed and draw solution streams that will be site specific. Also, thanks to counter-current cross flow operation and by optimising module arrangement higher recovery can be reached [36, 43].

10

3.1.2 Volatile fatty acids (VFA) / acetic acid rejection 100%

120

(b)

(a) 300

pH=4 Feed pH=4 Draw pH=7.5 Feed pH=7.5 Draw

200

80 60% 60 40% pH=4 20%

100 0%

pH=7.5

40

MW (g.M-1)

100

80% Rejection

Acetic acid concentration (mg.L-1)

400

20 0

0 0

2

4

6 8 10 WCF Figure 3: (a) Acetic acid concentration as function of WCF and pH in the feed and draw solution and (b) VFA rejection as function of feed pH (pH =4 and 7.5, DI water as feed, 35g.L-1 SW as draw solution)

It was first confirmed that the final concentration of acetic acid greatly depends on the pH of the feed solution. Acetic acid was poorly rejected at pH 4, which resulted in a poor VFA concentration in the resulting feed solution and passage to the draw solution (Figure 3a). A linear and sharp increase of acetic acid concentration was observed at pH 7.5, as well as WCF increased up to 4.4, reaching a concentration of acetic acid concentration in the range of the desired final VFA threshold concentration (300 mg.L-1). Thus, pH dependent VFA rejection was confirmed (Figure 3b) [31, 44]. All tested VFA have a pKa of around 4.8 [45], thus at pH=4 all are under R-COOH form, in that case the rejection is mainly steric dependent (bigger molecules are better rejected, (Figure 3b). In fact, acetic acid has the smallest molecular weight of the tested compounds and consequently is poorly rejected at pH=4 (33%). Higher rejection was observed for n-butyric acid than for iso-butyric and valeric acids despite a similar to lower MW; such behaviour indicates the complexity of rejection mechanisms for small organics. As already reported for nanofiltration membranes, steric rejection can also be affected by molecule geometry and can explain higher rejection of (longer carbon chain) n-butyric in comparison versus iso-butyric acid [46]. Rejection can also be largely dependent from electrostatic interactions, hydrophilicity, membrane solutes interactions and therefore further work is required to fully understand VFA rejection mechanisms by FO membranes [46-49]. On the other hand, around 90% rejection of all VFA was achieved at pH=7.5 in line with previous work [31]. Above pKa, the degree of ionisation of VFA increases and become negatively charged [31]; 11

therefore the rejection is enhanced by electrostatic repulsion with negatively charged FO membrane surface [50-52]. This point is important and reassuring since typical pH of the WW to be concentrated is around pH of 7-8 value where VFA and especially acetic acid are highly rejected by FO membranes. Even if not significantly observed here, at higher WCF with VFA concentration increasing in feed solution and water permeation flux decreasing with osmotic dilution, apparent rejection may decrease.

3.2 Evaluation tests with real domestic WW 3.2.1 Fouling propensity and mitigation strategies Fouling tests were performed with received pre-treated in a hydrolytic anaerobic reactor domestic WW (1) as is, (2) on MF pre-filtered WW and (3) MF pre-filtered WW and implementation of osmotic backwashing in between each batches as explained in section 2.6. Normalised initial flux over batches and after fouling and cleaning procedure are presented in Figure 4. WW Filtered WW Filtered WW +interbatch cleaning

30

(a) 25

WW Filtered WW Filtered WW +interbatch cleaning

30

(b)

25 Jw (L.m-2.h-1)

20

Jw (L.m-2.h-1)

20

15

15

10

10

5

5

0

before fouling

0 1

2

3

4

after fouling

after osmotic backwash

after manual cleaning

Batches

Figure 4: Comparison of initial flux during (a) filtration batches (real WW as feed, 70 g.L-1 SW as draw) and (b) recovery with cleaning (DI water as feed, 70 g.L-1 SW as draw) for WW, pre- filtered WW and pre-filtered WW combined with interbatch cleaning. Non pre-filtered WW caused a severe membrane fouling during the first batch which led to a reduction of the initial flux by 50% on the 2nd batch (Figure 4a). Important fouling behaviour was confirmed through the 4 successive fouling batches (Figure 4a), leading to a severe 80% overall loss

12

of initial flux. Cleaning procedure and flux evaluation with clean solutions allowed to compare flux before and after four batches of operation and confirmed the severe flux decline observed during fouling batches (Figure 4b) but to a lower extend (only 60% flux loss). This indicates that flux decline during batches with tested WW is not only due to fouling; most likely salts accumulation in the fouling layer also occurs, leading to enhance concentration polarization phenomenon and affecting the actual osmotic pressure driving force [53, 54]. Also, osmotic backwashing allowed a limited partial recovery of the initial flux. It attests of the high fouling propensity of the tested WW used in this study and that the applied cleaning strategy did not show sufficient efficiency for stable process operation. Loss of initial flux was limited to 36% after the fourth batch for the WW pre-treated by MF, attesting for the efficiency of MF to mitigate fouling (Figure 4a). Implementing interbatch backwashing further helped to maintain a high initial flux with a loss below 30% after the fourth batch. Positive impact of fouling mitigation techniques was confirmed during the cleaning evaluation procedure (Figure 4b). However, maximum flux recovery was limited to about 80% of the initial values in all cases even after manual cleaning (tested only for pre- filtered WW and pre-filtered WW combined with interbatch cleaning). Additional in place cleaning techniques such as chemical cleaning with different agents: i) NaOCl (to remove organic fouling), ii) citric acid (to remove scaling) and, iii) ethanol (in case of membrane drying) were performed but no beneficial effect was observed with neither of those. The membrane appeared slightly coloured but no fouling layer formed on the membrane surface was observed. This fact suggests that this incomplete flux recovery can be caused by some irreversible fouling within the membrane or a modification of the membrane properties. This phenomenon should be further studied as it can affect long term /full scale operation of the process. Also, NaOCl was tested but is not recommended since, as for RO membranes, the polyamide active layer of FO membrane is not highly tolerant to chlorine; cleaning in place solutions such as alkaline treatment, chelating agents, surfactants used in RO and tested for osmotic processes will be preferred in the future for organic fouling removal [55-57]. As an alternative to pre-treatments or more intensive cleaning, operating at lower flux has proven to be beneficial for fouling mitigation in FO [28]. In the tested application, achievable WCF and initial flux are both dependent from the available draw concentration and limit the flexibility to operate at lower flux. However, as mentioned earlier and considering large scale operation, counter flow can allow not only to reach higher WCF but also lower (but more constant) flux [43] and consequently will limit fouling issues. 13

3.2.2 VFA concentration and recovery In order to standardize the initial VFA concentration in the feed, samples were spiked with a mix of all VFA at a concentration of 80 mg.L-1 as in section 3.1.2 (followed by pH adjustment at pH=7.5 to balance the acidification that occurred with VFA addition). High rejection of all VFA by FO (around 90% at pH=7.5) was confirmed for real domestic WW (Figure 5a) [31]. VFA concentration controlled on the draw solution at the end of the filtration batches remained below the detection limit (<5mg.L1

). VFA recovery was calculated as the ratio of the VFA concentration at the beginning and the end of

each batch. 100%

100%

(a)

(b) 80% % VFA recovery

Rejection

80%

60%

60%

40%

40% DI water

20%

20%

WW

0%

0%

5

5 Acetic acid Propionic acid Isobutyric acid Butyric acid Valeric acid

3

Acetic acid Propionic acid Isobutyric acid Butyric acid Valeric acid

(c) 4

VFA CF

4

VFA CF

Batch 1 Batch 2 Batch 3 Batch 4

2

3

(d)

2 1

1

0

0 0

1

2

WCF

3

4

5

1

2

3 WCF

4

5

Figure 5: (a) VFA rejection in WW and DI water and VFA concentration: (b) % recovery of VFA at the end of the batches with WW spiked with VFA, VFA CF in function of WCF during (c) batch 1 and (d) batch 4 spiked samples (real WW as feed, 70 g.L-1 SW as draw) 14

Then, VFA were spiked to WW sample and concentration of VFA was followed during the 4 consecutive batches of fouling tests (filtered WW, no intermediate backwashing, Figure 5). High VFA recovery values were observed during batch 1 for all VFA (from 85% for acetic acid and up to 92% for valeric acid). However, VFA recovery sharply decreased over batches; acetic and propionic acid were no longer found already on the second batch, butyric and valeric acid disappeared fully on the third batch as shown by the poor recovery (Figure 5b). Only isobutyric acid was found at the end of the concentration step during the four batches but showing also a decrease of recovery over batches. VFA concentration kinetic during filtration was then analysed for batch 1 (clean membrane) and batch 4 to get further insight of concentration process; VFA CF as function of WCF is reported in Figure 5c&d. In batch 1 with the clean membrane (Figure 5c), a linear increase of VFA CF with WCF was observed, indicating high rejection and concentration rate. This confirms that VFA concentration by FO is a feasible solution, even with real WW. Totally different behaviour was observed in batch 4 with a membrane partly fouled. Increase of VFA CF was initially observed up to WCF of about 1.5 (i.e. during the 3 first hours of filtration) but then CF of butyric, propionic and acetic decreased, lowering down to 0 when WCF reaches 4.5. Valeric acid followed the same trend but delayed in time (higher WCF). Only isobutyric acid CF kept increasing with time until WCF of 4.5. However, as shown in (Figure 5b) at the end of the batch 4 when WCF reached 10, isobutyric acid also disappeared. VFA concentration was controlled in the draw solution but no VFA were found (below detection limit) demonstrating that the loss of VFA in the feed solution was not due to a lack of membrane selectivity. Loss of VFA in the feed can only be explained by biodegradation. The fact that the phenomenon occurred in the batch 4 and not in batch 1 indicates that biodegradation can be connected to the development of biofilm on the membrane (and even in the pipes). Additionally, the fact that isobutyric acid was still present at WCF=4.5 (Figure 5d) but disappeared at the end of batch 4 (Figure 5b) can be explained by longer residence time. VFA biodegradation was demonstrated in a former study and kinetic proved to be dependent on the substrate content [58]. In line with former studies, acetic acid proved to be the most biodegradable compounds being consumed first [58, 59]. Slower degradation of iso-butyric

was also observed in several works [58-60]; the slower

degradation of isomeric form versus normal form was explained as consequence of structural differences of components and possibly as an isomerisation from the normal to the isomeric form [60]. Our results are also in line with observation from where n- form butyric, propionic proved to be consumed before isobutyric acid (when glucose pre-grown inoculum was used) [58]. Faster degradation of butyrate than proprionate was observed in [61] as a consequence of low

15

concentration of propionate degrading bacteria. However, most of those studies were realised in the late 90’s, with consideration of anaerobic digestion conditions and pointed out various degradation effect depending on the inoculum and interactions in-between VFA. Further studies are required with regards to the specific context of VFA degradation during concentration and storage after anaerobic digestion to fully explain VFA degradation mechanisms and should consider the influence of other organics in the WW to be concentrated. More generally, the presence of concentrated VFA favours the development of biofilm that further enhances VFA consumption in repeated batches. It confirms that fouling mitigation has a crucial role to play not only to mitigate water flux losses but even more to avoid VFA biodegradation by the biofilm developed.

3.3 Long term evaluation with real wastewater Long term evaluation was finally conducted through 20 consecutive batches and considering all former observations so to limit fouling (and biofilm development) and to optimise VFA concentration (i.e. MF pre-filtration, interbatch osmotic backwashing, sparging with nitrogen in the filtered WW to remove oxygen, spiking with VFA just before filtration tests and coverage of the feed tank to limit contact with air).

3.3.1 Flux and water concentration factor Water permeation fluxes, time to reach WCF=5 and permeation flux during osmotic backwashing were monitored (Figure 6). 1000

30

(b)

(a) 20

Time (min)

Jw (L.m-2.h-1)

800

10

600 400 200

Initial flux 0

Time to reach WCF=5

0 0

5

10

15 Batch

20

25

0

5

10

15

20

25

Batch

16

0

After batch 8

(c)

(d)

Jw (L.m-2.h-1)

Backwash flux -10

After batch 20

-20

-30 0

5

10 15 Batch

20

25

Figure 6: Monitoring of (a) initial flux, (b) time to reach WCF=5, (c) backwash flux and (d) membrane surface pictures during long term evaluation experiments (real WW as feed, 70 g.L-1 SW as draw) A slight decrease of initial flux with batches was observed but flux remained high (above 14 L.m-2.h-1) even for batch 20 (Figure 6a). It demonstrates the efficiency of the MF pre-treatment and cleaning strategy (interbatch osmotic cleaning and flushing) to mitigate fouling. When looking at the time to reach WCF=5, a higher sensitivity was observed, with filtration time multiplied by 2 (from 400 min to around 800 min) when initial flux is in the lowest range (Figure 6b). It can be explained by a generally lower permeation flux leading to extended filtration time especially when WCF gets high. The loss of performance over batches is also seen on osmotic backwash flux (Figure 6c, values are negative since the flux is reversed). In fact it was observed that flux during the osmotic backwash (after the initial stabilisation of 10 min) remained constant over time (i.e. no flux improvement) and therefore it may indicate that osmotic backwash in this case is efficient only during the first minutes. Consequently reducing osmotic backwashing time below 1 hour may be possible in order to reduce downtime operation [28, 35]. In general, monitoring backwash flush is a potentially interesting control tool that should be further developed so to (1) monitor loss of performances in normal operation and (2) to adjust backwash duration time [35]. Fouling was expected to be responsible of flux reduction but in fact minimal fouling was observed after batch 8 and 20 (Figure 6d) and as already mentioned in section 3.2.1, no further flux recovery was obtained even after manual cleaning, chemical cleaning, ethanol treatment. Also, the change of membrane (after batch 8, following damage during manual cleaning test) demonstrated significant increase of flux, back to initial level. Thus it reinforces the hypotheses of irrecoverable fouling, change in membrane properties or damages during cleaning that should be further evaluated.

17

3.3.2 VFA concentration and recovery Follow up of acetic acid concentration over batches demonstrated that the objective of concentration from 80 up to 300 mg.L-1 can be reached; up to 500 mg.L-1 was even obtained (Figure 7a). Very interestingly also, a trend of decrease of concentration (biodegradation) over batches resulting from biofilm development as observed in section 3.2.2 (Figure 5) was not detected during long term experiments (Figure 7a). As such, it confirms that the pre-treatment and cleaning procedure implemented were succesfull to avoid biofilm formation on the membrane so to avoid biodegradation of acetic acid. However, very high variability in acetic acid final concentration was experienced from one batch to the other (1-7 and 14-19). Low final concentration were typically observed during batches operated overnight (2,4, 6, 15, 17, 19, 20) when the system was operated with higher filtration time (12-15 h) and higher WCF (7-18). Due to such variability and despite very low recovery in some batches, an overall average final concentration of 265 mg.L -1 was reached for acetic acid with average recovery of 45% (Figure 7b). Important variation of final concentration of other VFA were also observed but overall final concentration was less sensible to operation conditions during the FO concentration process, reaching final VFA concentration above 400 mg.L-1 and recovery rates of 70-80% (Figure 7b&c).

18

Conc Acetic acid (mg.L-1)

600

Final

Initial

(a)

500 400 300 200 100 0 0

2

4

6

8

10 Batch

1,000

VFA Conc (mg.L-1)

16

18

20

Final

(c)

(b)

80% VFA Recovery (%)

Initial

400

14

100%

800 600

12

60% 40%

200

20%

00

0%

Figure 7: (a) Initial and final concentration of acetic acid over the 20 batches of FO concentration runs, (b) initial and final VFA average concentration and (c) VFA average recovery rate over the 20 filtration batches To get further insight of the high variability of acetic acid (and other VFA) recovery and final concentration, monitoring of VFA concentration during filtration and as function of WCF was carried out (Figure 8).

19

100%

500 400 300

(b)

(a)

Acetic acid propionic acid isobutyric acid butyric acid valeric acid

% recovery VFA (mg.L-1)

Concentration VFA (mg.L-1)

600

200 100

80% 60% Acetic acid propionic acid isobutyric acid butyric acid valeric acid

40% 20% 0%

0 0

2

4

WCF

6

8

0

10

5 WCF

10

100%

Batches 10-20

(c)

% recovery VFA

80% 60% Acetic acid Propionic acid isobutyric acid butyric acid Valeric acid

40% 20% 0% 0

5

10

WCF

15

20

25

Figure 8: (a) VFA concentration and (b) recovery profile during FO filtration, and (c) acetic acid recovery as function of WCF. A linear increase of all VFA concentration and recovery rates were achieved at increasing WCF values up to 4.5. Higher WCF values led to a severe decrease of recovery, which was particularly significant for acetic acid whose recovery rate dropped down to 34% for WCF of 10 (Figure 8a and b). Analysis of % recovery of acetic acid for all batches (from 10-20) which featured different WCF completely confirm this observation (Figure 8c). On one hand, with lower WCF (4-5) which corresponds to shorter filtration time (7-10 h, day operation), 70-80% of acetic acid recovery is achieved. On the other hand very low recovery is obtained when WCF is in-between 10-20 (12-15 h, overnight operation). In line with results from section 3.2.2, recovery of other VFA remains in all cases higher than that of acetic acid, confirming the highest biodegradability of the latter while isobutyric acid proved to be less sensitive in our conditions and with the WW used. Longer residence time certainly favours VFA biodegradation. Additionally, at high recovery, (1) feed tank level is very low (and

20

possibly pump cavitation can happen) bringing air to the feed solution, (2) intense recirculation is occurring exposing the feed intensively to the biofilm that potentially starts to develop and finally (3) higher WCF also means higher concentration of VFA that will promote biological development and may decrease apparent rejection as mentioned in 3.1.2. Towards full scale implementation as well as for further laboratory scale studies, the design of process limiting residence time and avoiding contact with air even at high WCF will be a key challenge. Alternatively, operating at WCF=5 avoids producing WW with very high VFA concentration allowing for more flexibility in the design and limit biofilm formation and associated biodegradability issues.

4 Conclusions WCF above 10 (15-30) is largely achievable with pre-treated real domestic WW using seawater or brine as draw solution. VFA concentration can be largely affected by pH (poor rejection at acidic pH) and biodegradation. As such, VFA concentration using real WW proved to be more challenging with regards to fouling mitigation and VFA biodegradation. It is possible to reach VFA concentration (and especially acetic acid) up to 300-400 mg.L-1 thanks to the application of operation strategies to mitigate fouling and biodegradation but still remains very sensitive to biological development that is potentially very intense in such environment. Further studies should be conducted to optimise this process especially with regards to industrial application at full scale; full scale operation using FO modules in counter flow will also be in favour of reaching higher WCF, can allow for operating with lower salinity draw solutions, lower (but more constant) flux and consequently will limit fouling issues. High permeation flux above 14 L.m-2.h-1 was maintained during long term operation tests (20 batches) but loss of flux despite intense cleaning of the membrane was noticed. Longer operation, dedicated tests, membrane characterization and autopsy may help improving the understanding of mechanisms involved.

Acknowledgment The research leading to these results has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 685793, the People Programme (Marie Curie Actions) of the Seventh Framework Programme of the European Union (FP7/2007-2013) under REA grant agreement n° 600388 (TECNIOspring programme), and from the Agency for Business Competitiveness of the Government of Catalonia, ACCIO. LEQUIA and ICRA were recognized as

21

consolidated research groups by the Catalan Government with codes 2017-SGR-1552 and 2017 SGR 1318, respectively.

References 1.

2. 3. 4. 5.

6. 7. 8. 9. 10. 11. 12. 13. 14.

15. 16. 17.

18. 19.

The Long Read: Do we have the technology for potable water reuse? 2017; Available from: https://www.desalination.biz/news/3/The-Long-Read-Do-we-have-the-technology-forpotable-water-reuse/8660/. Angelakis, A.N. and B. Durham, Water recycling and reuse in EUREAU countries: Trends and challenges. Desalination, 2008. 218(1–3): p. 3-12. Ansari, A.J., et al., Forward osmosis as a platform for resource recovery from municipal wastewater - A critical assessment of the literature. Journal of Membrane Science. Hochstrat, R., et al., Options for water scarcity and drought management—the role of desalination. Desalination and Water Treatment, 2010. 18(1-3): p. 96-102. Pankratz, T. Desalination technology trends. Biennial Report on Seawater Desalination 2004 28/06/2016]; Available from: http://texaswater.tamu.edu/readings/desal/desaltechtrends.pdf. Vanoppen, M., et al., Salinity gradient power and desalination, in Sustainable energy from salinity gradients. 2016, Elsevier. Woodhead Publishing. p. 281-313. Blandin, G., et al., Efficiently Combining Water Reuse and Desalination through Forward Osmosis—Reverse Osmosis (FO-RO) Hybrids: A Critical Review. Membranes, 2016. 6(3). Cath, T.Y., et al., A multi-barrier osmotic dilution process for simultaneous desalination and purification of impaired water. Journal of Membrane Science, 2010. 362(1–2): p. 417-426. Attarde, D., et al., Energy-efficient seawater desalination and wastewater treatment using osmotically driven membrane processes. Desalination, 2017. 413: p. 86-100. Sim, V.S., et al., Strategic co-location in a hybrid process involving desalination and pressure retarded osmosis (PRO). Membranes, 2013. 3(3): p. 98-125. Mei, Y. and C.Y. Tang, Co-locating reverse electrodialysis with reverse osmosis desalination: Synergies and implications. Journal of Membrane Science, 2017. 539: p. 305-312. Cao, X., et al., A New Method for Water Desalination Using Microbial Desalination Cells. Environmental Science & Technology, 2009. 43(18): p. 7148-7152. Mehanna, M., et al., Using microbial desalination cells to reduce water salinity prior to reverse osmosis. Energy & Environmental Science, 2010. 3(8): p. 1114-1120. Borjas, Z., A. Esteve-Núñez, and J.M. Ortiz, Strategies for merging microbial fuel cell technologies in water desalination processes: Start-up protocol and desalination efficiency assessment. Journal of Power Sources, 2017. 356: p. 519-528. MIDES: microbial desalination for low energy drinking water. Available from: http://midesh2020.eu/. Coday, B.D., et al., The sweet spot of forward osmosis: treatment of produced water, drilling wastewater, and other complex and difficult liquid streams. Desalination 2014. 333: p. 23-35. Ansari, A.J., et al., Factors governing the pre-concentration of wastewater using forward osmosis for subsequent resource recovery. Science of The Total Environment, 2016. 566–567: p. 559-566. Chekli, L., et al., A comprehensive review of hybrid forward osmosis systems: Performance, applications and future prospects. Journal of Membrane Science, 2016. 497: p. 430-449. Lutchmiah, K., et al., Forward osmosis for application in wastewater treatment: A review. Water Research, 2014. 58: p. 179-197.

22

20.

21. 22. 23. 24. 25. 26. 27. 28. 29. 30.

31. 32. 33.

34.

35. 36.

37.

38. 39. 40.

Qin, J.J., W.C.L. Lay, and K.A. Kekre, Recent developments and future challenges of forward osmosis for desalination: A review. Desalination and Water Treatment, 2012. 39(1-3): p. 123136. Cornelissen, E.R., et al., Membrane fouling and process performance of forward osmosis membranes on activated sludge. Journal of Membrane Science, 2008. 319(1–2): p. 158-168. Gao, Y., et al., Direct concentration of municipal sewage by forward osmosis and membrane fouling behavior. Bioresource Technology. Hickenbottom, K.L., et al., Forward osmosis treatment of drilling mud and fracturing wastewater from oil and gas operations. Desalination, (0). Holloway, R.W., A. Achilli, and T.Y. Cath, The osmotic membrane bioreactor: a critical review. Environmental Science: Water Research & Technology, 2015. 1(5): p. 581-605. Holloway, R.W., et al., Forward osmosis for concentration of anaerobic digester centrate. Water Research, 2007. 41(17): p. 4005-4014. Blandin, G., et al., Retrofitting membrane bioreactor (MBR) into osmotic membrane bioreactor (OMBR): A pilot scale study. Chemical Engineering Journal, 2018. 339: p. 268-277. Blandin, G., et al., Can osmotic membrane bioreactor be a realistic solution for water reuse? npj cleanwater, (2018, In press). Blandin, G., et al., Fouling and cleaning of high permeability forward osmosis membranes. Journal of Water Process Engineering, 2016. 9: p. 161-169. Coday, B.D., et al., Rejection of trace organic compounds by forward osmosis membranes: A literature review. Environmental Science and Technology, 2014. 48(7): p. 3612-3624. Cagnetta, C., et al., Increased carboxylate production in high-rate activated A-sludge by forward osmosis thickening. Chemical Engineering Journal, 2017. 312(Supplement C): p. 6878. Jung, K., et al., Permeation characteristics of volatile fatty acids solution by forward osmosis. Process Biochemistry, 2015. 50(4): p. 669-677. Blandin, G., et al., Validation of assisted forward osmosis (AFO) process: Impact of hydraulic pressure. Journal of Membrane Science, 2013. 447: p. 1-11. Kim, J., et al., Practical considerations for operability of an 8″ spiral wound forward osmosis module: Hydrodynamics, fouling behaviour and cleaning strategy. Desalination, 2017. 404: p. 249-258. Blandin, G., A.R.D. Verliefde, and P. Le-Clech, Pressure enhanced fouling and adapted antifouling strategy in pressure assisted osmosis (PAO). Journal of Membrane Science, 2015. 493: p. 557-567. Motsa, M.M., et al., Osmotic backwash of fouled FO membranes: Cleaning mechanisms and membrane surface properties after cleaning. Desalination, 2017. 402: p. 62-71. Phuntsho, S., et al., Osmotic equilibrium in the forward osmosis process: Modelling, experiments and implications for process performance. Journal of Membrane Science, 2014. 453(0): p. 240-252. McCutcheon, J.R., R.L. McGinnis, and M. Elimelech, Desalination by ammonia–carbon dioxide forward osmosis: influence of draw and feed solution concentrations on process performance. Journal of membrane science, 2006. 278(1-2): p. 114-123. Sevda, S., et al., Microbial desalination cells as a versatile technology: functions, optimization and prospective. Desalination, 2015. 371: p. 9-17. Kim, Y. and B.E. Logan, Simultaneous removal of organic matter and salt ions from saline wastewater in bioelectrochemical systems. Desalination, 2013. 308: p. 115-121. Luján-Facundo, M.J., et al., A study of the osmotic membrane bioreactor process using a sodium chloride solution and an industrial effluent as draw solutions. Chemical Engineering Journal, 2017. 322: p. 603-610. 23

41.

42. 43.

44. 45. 46. 47.

48.

49. 50.

51.

52.

53. 54.

55.

56.

57. 58. 59.

Roberts, D.A., E.L. Johnston, and N.A. Knott, Impacts of desalination plant discharges on the marine environment: A critical review of published studies. Water Research, 2010. 44(18): p. 5117-5128. Sadhwani, J.J., J.M. Veza, and C. Santana, Case studies on environmental impact of seawater desalination. Desalination, 2005. 185(1): p. 1-8. Desormeaux, E. Porifera: creating a world with less waste, water reuse instead of discarded, concentrated products. erik-desormeaux.squarespace.com/s/Porifera-Technology-OverviewMarch2017.pdf 2017 01/03/2017]; Available from: erikdesormeaux.squarespace.com/s/Porifera-Technology-Overview-March2017.pdf. Ozaki, H. and H. Li, Rejection of organic compounds by ultra-low pressure reverse osmosis membrane. Water Research, 2002. 36(1): p. 123-130. Weast, R.C. and M. Astle, CRC handbook of chemical and physics. D-142, 1980. Bellona, C., et al., Factors affecting the rejection of organic solutes during NF/RO treatment— a literature review. Water research, 2004. 38(12): p. 2795-2809. Verliefde, A.R.D., et al., The role of electrostatic interactions on the rejection of organic solutes in aqueous solutions with nanofiltration. Journal of Membrane Science, 2008. 322(1): p. 52-66. Verliefde, A.R.D., et al., Influence of solute-membrane affinity on rejection of uncharged organic solutes by nanofiltration membranes. Environmental Science and Technology, 2009. 43(7): p. 2400-2406. D'Haese, A.K.H., et al., On negative rejection of uncharged organic solutes in forward osmosis. Journal of Membrane Science, 2018. 548: p. 22-31. Blandin, G., et al., Impact of hydraulic pressure on membrane deformation and trace organic contaminants rejection in pressure assisted osmosis (PAO). Process Safety and Environmental Protection, 2016. 102: p. 316-327. D'Haese, A., et al., Trace organic solutes in closed-loop forward osmosis applications: influence of membrane fouling and modeling of solute build-up. Water Res, 2013. 47(14): p. 5232-44. Xie, M., et al., Relating rejection of trace organic contaminants to membrane properties in forward osmosis: Measurements, modelling and implications. Water Research, 2014. 49: p. 265-274. Boo, C., et al., Colloidal fouling in forward osmosis: Role of reverse salt diffusion. Journal of Membrane Science, 2012. 390–391(0): p. 277-284. Hoek, E.M.V. and M. Elimelech, Cake-Enhanced Concentration Polarization: A New Fouling Mechanism for Salt-Rejecting Membranes. Environmental Science and Technology, 2003. 37(24): p. 5581-5588. Ang, W.S., et al., Fouling and cleaning of RO membranes fouled by mixtures of organic foulants simulating wastewater effluent. Journal of Membrane Science, 2011. 376(1): p. 196206. Chen, S.C., G.L. Amy, and T.-S. Chung, Membrane fouling and anti-fouling strategies using RO retentate from a municipal water recycling plant as the feed for osmotic power generation. Water research, 2016. 88: p. 144-155. Li, X., et al., Cleaning strategies and membrane flux recovery on anti-fouling membranes for pressure retarded osmosis. Journal of Membrane Science, 2017. 522: p. 116-123. Aguilar, A., C. Casas, and J.M. Lema, Degradation of volatile fatty acids by differently enriched methanogenic cultures: Kinetics and inhibition. Water Research, 1995. 29(2): p. 505-509. Vavilin, V.A. and L.Y. Lokshina, Modeling of volatile fatty acids degradation kinetics and evaluation of microorganism activity. Bioresource Technology, 1996. 57(1): p. 69-80.

24

60. 61.

Wang, Q., et al., Degradation of volatile fatty acids in highly efficient anaerobic digestion. Biomass and Bioenergy, 1999. 16(6): p. 407-416. Griffin, M.E., et al., Methanogenic population dynamics during start‐up of anaerobic digesters treating municipal solid waste and biosolids. Biotechnology and bioengineering, 1998. 57(3): p. 342-355.

25

Highlights: 

30 times concentration factor of WW was achieved using RO brine as draw solution



High rejection of VFA by FO membranes is achieved at pH=7.5



Concentration of VFA up to 400 mg.L-1 was achieved with real WW



Intense VFA biodegradation can happen due to organic fouling and biomass development

26