- Email: [email protected]
Bacteriophage MS-2 removal by submerged membrane bioreactor
ARTICLE IN PRESS
Water Research 39 (2005) 4211–4219 www.elsevier.com/locate/watres
Bacteriophage MS-2 removal by submerged membrane bioreactor Chii Shang, Hiu Man Wong, Guanghao Chen Department of Civil Engineering, Hong Kong University of Science and Technology, Clear Water Bay, Kowloon, Hong Kong Received 5 November 2004; received in revised form 10 June 2005; accepted 10 August 2005
Abstract A membrane bioreactor (MBR) may serve as a pre-disinfection or disinfection unit, in addition to its solid/liquid separation and biological conversion functions, to produce sewage effluent of high quality. This bench-scale pilot study focuses on investigating the performance of a submerged MBR in pathogen removal and the factors affecting the removal, using a 0.4-mm hollow-fiber membrane module submerged in an aeration tank and bacteriophage MS-2 as the indicator organism. Removal of the MS-2 phage was found to be contributed by physical filtration by the membrane itself, biomass activity in the aeration tank and bio-filtration achieved by the biofilm developed on the membrane surface. The membrane alone gave poor virus removal (0.470.1 log) but the overall removal increased substantially with the presence of biomass and the membrane-surface-attached biofilm. The contributions of the suspended biomass and attached biofilm to the phage removal are dependent on the inter-related parameters including the concentration of mixed liquor suspended solids (MLSS), the sludge retention time (SRT) and the food to mass (F/M) ratio. The correlations between effluent flux/trans-membrane pressure and virus removal give evidence that phage removal in the MBR is most likely susceptible to both biological and physical factors including the quantity and property of the biomass and the biofilm and the membrane pore size reduction. r 2005 Elsevier Ltd. All rights reserved. Keywords: Bacteriophage; Biofilm; Bioreactor; Disinfection; MBR; Membrane
1. Introduction Membrane separation has long been considered as a ‘‘safe and clean’’ physical means to yield some disinfection credits for water purification and reuse purposes. It reduces the problem associated with the generation of harmful disinfection by-products (DBPs) from chemical disinfection processes (e.g., chlorination or chloramination) by cutting down the chemical dosages. The membrane bioreactor (MBR) technology is similar to Corresponding author. Tel.: +852 2358 7885; fax: +852 2358 1534. E-mail address: [email protected] (C. Shang).
conventional activated sludge processes except that membranes (mostly microfiltration membranes) are used to extract effluent. Therefore, much higher concentrations of mixed liquor suspended solids (MLSS) (normally 45000 mg/L) and long sludge ages (commonly 420 days) can be kept in the aeration tank since the settlability of the flocs is not a concern. MBRs have demonstrated outstanding performance with respect to its biological conversion and solid removal (common values of effluent: TSSo5 mg/L, CODo40 mg/L, TNo12 mg/L and TPo2.2 mg/L) (Churchouse and Brindle, 2002; van der Roest et al., 2002; Wong et al., 2003). MBRs also provide advantages in sludge reduction (Metcalf & Eddy, Inc., 2003) and hence the sludge
0043-1354/$ - see front matter r 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2005.08.003
ARTICLE IN PRESS 4212
C. Shang et al. / Water Research 39 (2005) 4211–4219
treatment cost can be reduced. In addition to these profound features, MBRs are now being considered as an effective, non-hazardous alternative disinfection means to achieve pathogen control in wastewater effluents. Numerous studies have reported outstanding microbial removal achieved by MBRs when Escherichia coli, fecal coliform bacteria, and/or bacteriophages were used as the indicators (e.g., Churchouse and Brindle, 2002; van der Roest et al., 2002; Wong et al., 2003; Ueda and Horan, 2000). Churchouse and Brindle (2002) have demonstrated that a full-scale MBR plant produced exceptionally good-quality effluent over years including 3–6 log removal of fecal coliform bacteria and 2–5 log removal of F+ coliphage. A bench-scale pilot study by Ueda and Horan (2000) showed that an MBR could achieve 2–6 log removal of indigenous T-even-like bacteriophage (with a mean size of 200 nm) and almost complete removal of fecal coliform bacteria (up to 7 log). Only 0.91 log removal of the same type of phages could be achieved by the conventional activated sludge processes (Ueda and Horan, 2000). Therefore, with such high-quality effluent of MBR processes, dosages of disinfectants applied afterwards can be much reduced or the following disinfection processes may be completely eliminated. To assess the disinfection credit achieved by MBRs or to guarantee the effluent microbiological quality, viruses are expected to be more suitable indicator organisms than bacteria since viruses are much smaller and harderto-straining than bacteria and considered to be more resistant to common disinfectants (Leong, 1983). Meanwhile, the importance of removing viruses from wastewater is increasingly being recognized because of the epidemiological significance of viral pathogens (Melnick, 1984). Due to the difficulty in assaying animal viruses, bacteriophages have been suggested as viral indicators because they closely resemble enteric viruses in terms of the structure, morphology, size and behavior (Maier et al., 2000). Bacteriophage MS-2 (hereafter referred to as ‘‘MS-2’’) was chosen to be the indicator in this study in consideration of its smallest size (0.02–0.025 mm) among viruses and hence it is representative to address the ability of pathogen removal by an MBR. The mechanisms of virus rejection by MBRs are still ambiguous and the extent of virus removal is likely subject to the operational conditions. It has been reported that virus removal by an MBR became upgraded with the presence of biomass and the visible sticky biofilm developed on the membrane surface (Ueda and Horan, 2000), though the normal pore size of the microfiltration membrane (0.4 mm) was larger than the size of the tested phages (0.2 mm). However, the performance of removing viruses far smaller than the membrane pore size of an MBR and the affecting
operational parameters are not clear and the deficient knowledge in this area has prompted this study. This paper describes an experimental study on evaluating the performance of a submerged MBR in MS-2 removal and the factors affecting the removal efficiency. Bench-scale pilot MBR units were constructed using microfiltration hollow-fiber membrane modules with a nominal pore size of 0.4 mm. The overall removal efficiency and the contributions of solely the membrane, the suspended biomass, and the membrane-attached biofilm were examined. The affecting factors in evaluation included the concentration of MLSS, the sludge retention time (SRT), the food to mass (F/M) ratio, the operating flux and the concentration of MS-2.
2. Materials and methods 2.1. Bench-scale MBR Membrane modules used in this study were provided by the Mitsubishi Rayon Corporation (MRC). Table 1 shows the characteristics of the membrane module. The membrane module was submerged in a reactor and an air diffuser was equipped at the base of the reactor to provide both aeration and membrane surface cleaning (airflow rate of 3 L/min). The operating volume of the aeration tank was 19 L with a footprint area of 26 cm 25 cm. The effluent pump was operated at a constant suction force but in an intermittent suction mode (13 min ON/2 min OFF). The initial flow rate during the suction in a clear water tank with a clean membrane module was 50 L/day. The corresponding flux and hydraulic retention time (HRT) were 0.25 m/ day and 9 h, respectively. The intermittent ceasing of the suction and the scraping effect generated by the rising air bubbles prevented solid deposition on the membrane surface and extended the service period between membrane cleanings. Pressure gauges were installed in the effluent side for monitoring the changes in suction pressure (or trans-membrane pressure (TMP)). The effluent flux was recorded by measuring the effluent volume flowed in a definite time.
Table 1 Characteristics of the microfiltration membrane module Membrane type Material Nominal pore size (mm) Maximum allowable flux (m3/m2/ day) Total surface area (m2)
Microporous hollow fiber Polyethylene 0.4 0.4 0.2
ARTICLE IN PRESS C. Shang et al. / Water Research 39 (2005) 4211–4219 Table 2 Compositions of the organic nutrient stock and mineral water Organic nutrient stock (per 1 L of deionised water)
Mineral water (per 100 L of deionised water)
7.5 g Peptone 15 g Yeast extract
100 g NaCl 2.75 g CaCl2 2.25 g MgSO4 7H2O 0.025 g FeCl3 6H2O
The reactor was first fed with returned sludge taken from the Shatin Sewage Treatment Works, Hong Kong and cultivated in the laboratory. The reactor was fed continuously with synthetic water containing organic nutrients and mineral water with compositions shown in Table 2. The flows of the two waters were adjusted so that, before a trial test began, the biomass could be stabilized with a specific level of MLSS and a sludge age with a specific F/M ratio and at a designated flow rate. It was presumed that the system reached stabilization when both the MLSS level and the sludge age (or the amount of sludge wasted every day) kept constant for more than a week. The bioreactors were operated under ambient temperature (18–22 1C), pH 5–8 and dissolved oxygen of 2–3 mg/L. 2.2. MS-2 culture preparation and assay Bacteriophage MS-2 stocks (ATCC 15597-B1) were prepared in suspension using their host cells E. coli (ATCC 15597) solutions. Active cultures of the E. coli host were first cultivated for 18–24 h at 37 1C in the medium suggested by the American Type Culture Collection. The MS-2 stock was then added into the active host cultures and the infected bacteria were incubated for 18–24 h to allow the propagation of the phage. Calcium chloride was added to promote the adsorption of MS-2 to the bacterial cells (Adams, 1959). The cultured MS-2 solutions were used directly without further treatment. The concentrations of the free-flowing MS-2 in samples were enumerated using the Double Agar Layer Method (Adams, 1959). An immobilized phage would develop a semi-transparent circle, or a plaque, to indicate the lysis of the host cells and represent the progeny of a single immobilized phage. The number of plaques (plaque forming unit or PFU) can represent the number of phages present in a suspension of a given volume. 2.3. Experimental procedures A test run was always first conducted in a ‘‘clean system’’ by using a cleaned membrane in a tank
4213
containing only synthetic water without any suspended solids. The MS-2 removal obtained in the clean system was recorded as the ‘‘baseline removal’’, which allowed differentiation of the intrinsic removal capacity of the membrane from the additional removal induced by biomass and/or biofilm in the following test runs. Membrane cleaning to remove biofilm deposition from the previous run to restore the original TMP was conducted by backwashing the membrane module with a 600 mg/L hypochlorite solution for 15 min. The cleaned membrane was then submerged in a reactor containing suspended biomass stabilized at specific values of MLSS and SRT. To study the effects of MLSS, the biomass was stabilized at 6000, 8000 and 10,000 mg/L with the same SRT of 200 days. The corresponding F/M ratios were 0.05, 0.04 and 0.03/day, respectively. For examining the effects of SRT, the biomass was stabilized at a SRT of 50 and 200 days with the same MLSS of 6000 mg/L. The corresponding F/M ratios were 0.06 and 0.05/day, respectively. The MBR units were operated continuously for a period of time to allow the development of membrane-attached biofilm. Whichever biomass was present in the tank, the units were fed with a dosage of the MS-2 solution at concentrations of around 107 PFU/mL. Samples were periodically collected from the aeration tanks (the bulk solution) and the effluent outlet of the MBR units, after initiating test runs with rapid mixing. If a test run lasted for more than a few hours, the tanks were periodically fed with the MS-2 solution at same dosages. The samples taken from the bulk solution were filtrated with 0.45 mm filter papers before the viral assay was made so that only free-flowing (not floc associated) MS-2 was measured. The purpose of sample filtration was to distinguish the removal by the membrane (including the biofilm) and that by the suspended biomass, where the latter was examined in separated tests without the membrane and effluent discharge. At different time intervals, a sample was taken from the mixed liquor, filtered with a 0.45 mm filter paper and subjected to MS-2 numeration. At least three samples were tested at each time and some tests were repeated under the same conditions for quality control. The biofilm formation in the repeated tests was generally reproducible. Removal efficiency can be expressed as log removal value (LRV): LRV ¼ log
C , C0
(1)
where C0 and C refer to the free-flowing MS-2 concentrations before and after treatment, respectively. When membranes were involved in tests, the C0 was taken from the mixed liquor and the C was taken from the effluent outlet. Under such cases, LRVm, was reported, which refers to the log removal by the membrane (or biofilmed membrane) and has excluded
ARTICLE IN PRESS C. Shang et al. / Water Research 39 (2005) 4211–4219
those phages associated with the suspended solids. In addition, a normalized log removal value (NLRVm) is sometimes reported in this paper, which has subtracted the ‘‘baseline removal’’ obtained in the ‘‘clean system’’. This normalization allows differentiation of the additional removal induced by biofilm from the intrinsic removal capacity of the membrane (Mysore et al., 2003). Last, it should be noted that the term ‘‘biofilm’’ used here and in the following discussion refers to the phageresistant layer gradually developed on the outer and inner membrane surface during MBR operation, though the roles of the deposited solids and the attached biomass cannot be distinguished and the enhanced virus removal efficiency is likely attributable to both. 2.4. Other measurements The concentration of MLSS in the MBR was measured in accordance with Greenberg et al. (Standard Methods) (1998). Sludge bioactivities for both organic decomposition and nitrification was measured and characterized by the specific oxygen uptake rate (SOUR), which is also known as the oxygen consumption or respiration rate. It is defined as the milligram of oxygen consumed per gram of volatile suspended solids (VSS) per hour. The detailed procedure can be referred to that stated in Greenberg et al. (Standard Methods) (1998).
3.0 Log Removal Value, LRVm
4214
2.5 2.0 1.5 1.0 0.5 0.0 0
5
10 15 Time (day)
20
25
Fig. 1. Phage removal by the biofilmed membrane at MLSS ¼ 6000 mg/L and SRT of 50 days.
membrane cleaning to control the effluent microbiological quality by an additional disinfection process. For comparison, using a hollow-fiber membrane with a same pore size of 0.4 mm but T-even-like phage of 0.2 mm (ten times larger than MS-2) at a much higher initial flux (1.2 m/d, which was 4–5 times higher than that used in the current study), Ueda and Horan (2000) demonstrated a negligible phage removal (about 0.1 to 0.3-log) in a ‘‘clean system’’.
3.2. Phage removal by biofilmed membrane 3. Results and discussion 3.1. Phage removal by un-biofilmed membrane Phage removal in the MBR was proposed to be primarily contributed to the biofilm developed on the membrane surface, which physically reduced the membrane pore size, chemically adsorbed phages, and biologically allowed the predation of phages by other microorganisms (Ueda and Horan, 2000). Here, we examined the relative contributions of the membrane only, the biomass only, and the membrane-attached biofilm to the phage removal. The membrane itself provides the first obstruction for pathogen removal because of the sieving effect. The contribution of solely the membrane to phage removal (referred to as the ‘‘baseline removal’’ in Fig. 1) stayed at about 0.3–0.4 log throughout the examination period. Leakage of phages in such a ‘‘clean system’’ in the absence of both biomass and biofilm is expected since the average pore size of the membrane fibers (0.4 mm) is much larger than the size of the bacteriophage MS-2 (0.02 mm). It should be noted (data not shown) that the baseline removal consistently restored to similar values (70.1 log) after 15 min backwash with a 600 mg/L hypochlorite solution. Therefore, in practice, attention should be paid right after
By submerging the membrane into an aeration tank containing stabilized biomass under an MLSS of 6000 mg/L and a SRT of 50 days, as shown in Fig. 1, phage removal improved over time as a result of the development of a biofilm layer on the membrane surface. It can be explained that the virus transport is hindered due to the pore reduction and phage adsorption onto the biofilm (Ueda and Horan, 2000). The statement was supported by visual and scan electron microscopy (SEM) inspections of the membrane surface. After submerging the membrane module in the reactor for a few weeks, a slime or gel layer was visually observed. SEM inspection of the external surface (on the feed side) of the biofilmed membrane showed that the surface was fully covered with biofilm layers, while internal blockage and partial coverage presumably by extracellular polymeric substances (EPS) were observed in the membrane pores by taking SEM images from the internal surface on the permeate side. Fig. 1 also shows that, in the first 2 days, fastest improvement of phage removal was recorded. A ‘‘primary biofilm’’ (Ridgway and Flemming, 1996) was said to have formed. Thereafter, the phage removal improved gradually, which can be attributed to the formation of the so-called ‘‘secondary biofilm’’ (Ueda and Horan, 2000). The error bars as
ARTICLE IN PRESS C. Shang et al. / Water Research 39 (2005) 4211–4219
shown in Fig. 1 indicate that more than 0.5 log removal can be considered as significant.
4215
3.0
3.3. Phage removal by suspended biomass Phages are also expected to associate with biomass flocs and then get removed by flocculation or cell adsorption in the aeration tank. Fig. 2 shows the phage removal in the absence of the membrane in two tanks with biomass stabilized at two specific conditions: (A) MLSS of 6000 mg/L, SRT of 50 days, and an F/M ratio of 0.06/day, and (B) MLSS of 10,000 mg/L, SRT of 200 days, and an F/M ratio of 0.03/day. The phage removal under the two conditions increased over the sampling time and reached 0.85 and 0.6 log for F/M ratios of 0.06 and 0.03/day, respectively, after 9 h (one HRT of the MBR system). Although a higher F/M ratio seems to give higher phage removal, such small difference is not statistically significant. The phage removal achieved under both conditions in the current study was slightly less than that (0.91 log) of a conventional activated sludge plant that was operated at HRT of 9.7 h, MLSS of 1420 mg/L, and an F/M ratio of 0.62/day (Ueda and Horan, 2000). Combining the data obtained from the two studies, the bioadsorbility of the suspended biomass is unlikely to be considerably affected by the HRT and the biological operational factors. 3.4. Contributions of the MBR components Fig. 3 summarizes the individual contribution of the MBR components to the phage removal under MLSS of 6000 mg/L and SRT of 50 days. The contribution of the membrane alone is indicated by ‘‘Membrane’’ while ‘‘9-h-Biomass’’ refers to the removal by the suspended biomass in the absence of membrane after one HRT (9 h). The ‘‘9-h-Biofilm’’ and ‘‘21-day-Biofilm’’ refer to the NLRVms by the biofilm after submerging a clean 3.0 MLSS=6,000mg/L, SRT=50day, F/M=0.06/d
Log Removal Value
2.5
MLSS=10,000mg/L, SRT=200day, F/M=0.03/d
2.0 1.5
2.0 1.5 1.0 0.5 0.0 Membrane
9-hr-Biomass
9-hr-Biofilm 21-day-Biofilm
Fig. 3. Contributions of the MBR components to phage removal at MLSS of 6000 mg/L and SRT of 50 days.
membrane in the aeration tank for one HRT and 21 days, respectively. The NLRVms have excluded the contribution from biomass and the ‘‘baseline removal’’ by the sole membrane. As shown, the contributions of the membrane, the suspended biomass, the 9-h-biofilm and the 21-day-biofilm to the phage removal in the MBR are 0.4 log, 0.8 log, 0.3 log and 2.1 log, respectively. The phage removal by the suspended biomass contributed a good portion of the phage removal, especially when a clean or cleaned membrane was submerged in the aeration tank up to 9 h. Nevertheless, given enough time to allow the biofilm to develop, the removal by the biofilm improved significantly and it was the biofilm developed on the membrane surface playing the most important role in removing the MS-2 phage. The development of a biofilm on the membrane surface was proposed to make the major contribution to pathogen removal (Ueda and Horan, 2000) and biofilms were suggested to trap and accumulate virus-sized particles (Flood and Ashbolt, 2000; Sutherland et al., 2004). The above findings indicate that a real MBR system may achieve higher viral removal since, in realworld operations, chemical cleaning of membranes is commonly conducted every few months so that the contribution from the biofilms can be more significant. In addition, the suspended solids in the sewage may also adsorb phases or contribute to membrane pore blockage to enhance phage removal. 3.5. Effects of MLSS levels on phage removal
1.0 0.5 0.0
Log Removal Value
2.5
0
2
4 6 Time (hr)
8
10
Fig. 2. Phage removal by suspended biomass in the absence of membrane under the specific conditions.
To investigate how the characteristics of biomass affected the biofilm formation and hence the phage removal in the MBR, MS-2 phage removal by the biofilm developed on the membrane surface at three different MLSS levels (6000, 8000 and 10,000 mg/L), which were commonly applied in full-scale MBR systems, was examined. The SRT was 200 days in all cases. Fig. 4 presents the results. The LRVs are the
ARTICLE IN PRESS C. Shang et al. / Water Research 39 (2005) 4211–4219
4216 3.0
1
Flux / Initial Flux
2.5
NLRVm
2.0 1.5 1.0
0.95
0.9
0.85
0.5 MLSS=6,000
0.0
MLSS=8,000
MLSS=10,000
MLSS=6,000
MLSS=8,000
MLSS=10,000
0.8 0
2
4
6 8 Time (Day)
10
12
0
14
5
10 Time (day)
(a)
Fig. 4. Phage removal by the biofilm at different MLSS levels (baseline removal excluded).
15
20
3
3.6. Effects of sludge age on phage removal Fig. 6 shows phage removal (NLRVm) in the MBR stabilized at an MLSS of 6000 mg/L under two SRTs (50 and 200 days). Although the concentrations of MLSS
TMP - TMPo (kPa)
2.5 2 1.5 1 0.5 MLSS=6,000
MLSS=8,000
MLSS=10,000
0 0
2
4
6 8 Time (day)
(b)
10
12
14
Fig. 5. Changes of (a) flux/initial flux ratios and (b) TMP under different MLSS levels at the same SRT of 200 days.
3.0 2.5 2.0 NLRVm
NLRVms, which refer to only those achieved by the biofilm formed on the membrane surface. The phage removal at MLSS of 6000–10,000 mg/L showed rapid improvement during the initial stage of filtration (with 6000 mg/L gave slightly higher removal values) and followed by gradual enhancement of removal to reach similar removal values. It was originally postulated that more suspended solids would deposit on the membrane surface if the concentration of biomass in the reactor was higher; however, the fact that the similar or lower removal with increasing MLSS during the tested period was observed. The finding indicates that the nature of the biomass (e.g., the amount/composition of EPS), which shall change with the change of MLSS or the F/M ratios, is more important than the concentration of that in influencing the phage removal. As discussed previously, the biofilm layer developed on the membrane surface (internally or externally) does control pathogen removal the most in the MBR. The extent of membrane fouling is commonly quantified by the changes of the effluent flux or the TMP. In the cases of MLSS of 6000 and 8000 mg/L, the flux dropped (Fig. 5a) and the TMP increased (Fig. 5b), in accordance with the increase in phage removal (Fig. 4b). However, larger fluctuations in the flux and the TMP were observed at 10,000 mg/L MLSS, implying that the biofilm accumulated cannot smooth out the flow instability. Lee et al. (2001) reported that higher MLSS concentrations (in the range of 2000 and 5000 mg/L) alleviated an increase in TMP. The data shown in Fig. 5 support the statement and extend its validity up to 8000 mg/L MLSS; however, it is not the case at 10,000 mg/L MLSS.
1.5 1.0 0.5
MLSS=6000mg/L, SRT=50d MLSS=6000mg/L, SRT=200d
0.0
0
2
4
6
8 10 Time (day)
12
14
16
Fig. 6. Phage removal by the biofilm developed under the same MLSS level but different solid retention time (baseline removal subtracted).
were the same for both cases, the system with longer SRT yielded at least 1 log higher MS-2 removal. The system with the SRT of 200 days also gave faster improvement in phage removal than that of 50 days. Together with the discussion in Section 3.5, phage
ARTICLE IN PRESS C. Shang et al. / Water Research 39 (2005) 4211–4219
removal is susceptible to the changes in operational conditions such as the MLSS level and the SRT which in turn influence the characteristics of the membraneattached biofilm. Recall that a higher F/M ratio (or lower MLSS at fixed SRT) yielded faster and larger improvement in phage removal in Section 3.5, while a lower F/M ratio (or longer SRT at fixed MLSS) yielded better phage removal here. This contradiction suggests that the phage removal is controlled by the inter-related parameters including MLSS, SRT and F/M ratios. 3.7. Correlations between phage removal and membrane fouling Fig. 7 shows the correlations between the normalized phage removal by the biofilm developed under the specific conditions and the corresponding TMP. As shown, at an F/M ratio of 0.06, TMP can be used to estimate the phage removal in the MBR. However, TMP cannot reflect phage removal when comparing results obtained under different conditions. In fact, as F/M ratios decreased, the correlation between TMP and phage removal diminished. Hence, membrane fouling is not always correlated well with phage removal. Since TMP is only a measurement of physical blockage of the membrane, the results also indicate that blockage or membrane fouling is not the only factor in affecting phage rejection. Fig. 7 shows that TMP increases to larger extent at a higher F/M ratio while giving lower NLRVm, compared to those obtained at lower F/M ratios. In order to interpret the above phenomena, the relationship between operational conditions and sludge characteristics could be first explored. It has been reported in the literature that the overall EPS production is low at a long SRT (or perhaps at a low F/M ratio) (Chang and Lee, 1998). This explains the TMP increase at a SRT of 200 days was lower (because of lower total EPS
4217
production) than the data obtained at a SRT of 50 days, although the MLSS levels were the same. Lee et al. (2003) suggested that the ratio of protein to carbohydrate in EPS is more important than the quantity of EPS in controlling filtration resistance or membrane fouling. At high SRT or lower F/M ratios, the carbohydrate in microbial flocs declined while the amount of protein on the cell surface increased (Lee et al., 2003). This may explain how the operational conditions affect the characteristics of the suspended biomass and attached biofilm and hence the phage removal by the two phases. Lee et al. (2003) also suggested that a longer SRT led to decreases in microbial activity in MBRs at a SRT of 20–60 days. The current study, however, showed that the microbial activity, measured as the SOUR, of the biomass with SRTs of 50 and 200 days was similar and at about 50 mg O2/g VSS/h, indicating that, at higher SRTs, the microbial activities cannot be used to explain the difference in biomass characteristics. 3.8. Effects of phage concentration and effluent flux Fig. 8 illustrates the relationship between the influent and effluent phage counts for three cases at various influent phage concentrations and effluent fluxes. For the case of a clean system (clean membrane in clear water), the LRVs are independent of the influent phage concentration; implying that the water quality of the effluent depends on that of the influent. This is obvious because the pore size of the membrane is much larger than the size of the phage; more phages pass through the membrane as more phages are present in the reactor. The other two cases are log removal of phages by the biofilmed membrane submerged in clear water at different effluent flow rates. With the presence of biofilm, the LRVm increases with increasing the influent phage concentration. These results are in agreement with Ueda and Horan (2000) who concluded that phage removal is independent of phage concentration in the
3.0
3 2.5
2.5 2 1.5
LRVm
NLRVm
2.0
Clean System Biofilm @30 L/d Biofilm @50 L/d
1.0
0.0
1
MLSS=6000, SRT=50, F/M=0.06 MLSS=6000, SRT=200, F/M=0.05 MLSS=8000, SRT=200, F/M=0.04 MLSS=10000, SRT=200, F/M=0.03
0.5 0
1
3 4 2 TMP-TMPo (kPa)
5
1.5
0.5 6
Fig. 7. Correlations between phage removal (by the biofilm only) and the changes of trans-membrane pressure (TMP) (with initial TMP subtracted) under different operational conditions.
0 103
104 105 106 107 108 109 Phage Concentration in Bulk Solution (PFU/ml)
Fig. 8. Relationship between phage removal and phage concentrations in the bulk solution at different operating fluxes.
ARTICLE IN PRESS 4218
C. Shang et al. / Water Research 39 (2005) 4211–4219
absence of biomass or biofilm; while it becomes dependent on the concentration with the presence of biomass or biofilm. The biofilm present in the reactor helps reduce the pore size of the membrane and hence the passage of phages becomes limited. Therefore, the phage concentration in the effluent may be relatively stable even with fluctuation in the influent phage concentration. Fig. 8 shows that phage removal decreases with increasing effluent flux but the effect appears to highly depend on the dosing concentration of phages. However, the effect of operating flux on phage removal is expected to be negligible in real-world applications since the concentration of phages found in sewage is about 102–105 PFU/mL, which is much lower than the concentration used in this study. Besides, it should be noted that the study of the effect of the effluent flux was conducted after the biofilm had developed. It is likely that the operating flux can affect the development of the biofilm.
4. Conclusions Phage removal in the MBR can be attributed to three components: the membrane, the biomass and the membrane-attached biofilm. Only 0.470.1 log reduction of bacteriophage MS-2 could be achieved by the clean membrane in the absence of biomass. At 6000-mg/L MLSS, 200-day SRT, and 9-h HRT, the suspended biomass contributed to 0.8 log phage removal. When immersing the membrane in the stabilized biomass, the overall removal increased substantially after sufficient time (over weeks) was given to allow the development of biofilm on the membrane surface. Therefore, phage removal in the MBR improves by prolonging the operational period between membrane cleanings. However, after each cleaning, special attention should be paid to control the effluent pathogen concentrations by an additional disinfection process. The characteristics of the biomass and the biofilm and their contributions to phage removal are influenced by the inter-related parameters including the concentration of MLSS, the SRT and the F/M ratio. Under the tested conditions, operating at lower MLSS (among 6000, 8000 and 10,000 mg/L with same SRT) or longer SRT (among 50 and 200 days with same MLSS) yields faster and larger improvement in phage removal. Lowering the F/M ratios diminishes the correlation between TMP and phage removal. In addition, the correlations between phage removal and trans-membrane pressure (TMP) conclude that membrane fouling is merely one of the several factors that influence phage removal in the MBR. Phage removal in the MBR is most likely susceptible to biological factors such as the quantity and property of the biofilm.
Last, phage removal improves as the biofilm is allowed to develop, but this decreases the effluent flux or increases the TMP to reduce the capacity of the MBR system or to increase the pumping cost. Balancing this dilemma is of great importance if the effluent pathogen control relies fully or partially on the disinfection credit granted to an MBR. It should be noted that although bacteriophage MS-2 was used to indicate the likelihood pathogen removal by MBR in this study, it is possible that the surface characteristics of viruses and phages are of great variety and their adsorption to the complex biofilm/biomass may also vary. Acknowledgments The insightful comments of three anonymous reviewers are gratefully acknowledged. This study was supported in part by the Hong Kong Research Grants Council under grant DAG03/04.EG25. The bench-scale membrane modules were provided by the Mitsubishi Rayon Corporation. Certain commercial materials are identified in this paper to specify the experimental procedures. Such identification does not imply that the materials are necessarily the best available for the purpose.
References Adams, M.H., 1959. Bacteriophages. New York, NY, Interscience. Chang, I.S., Lee, C.H., 1998. Membrane filtration characteristics in membrane coupled activated sludge system—the effect of physiological states of activated sludge on membrane fouling. Desalination 120, 221–233. Churchouse, S., Brindle, K., 2002. Operational experience of full scale membrane bioreactor sewage treatment plants. In: Proceedings of the IWA Third World Water Congress, International Water Association, Melbourne, Australia. Flood, J.A., Ashbolt, N.J., 2000. Virus-sized particles can be entrapped and concentrated one hundred fold within wetland biofilms. Adv. Environ. Res. 3, 403–411. Greenberg, A.E., Clesceri, L.S., Eaton, A.D. (Eds.), 1998. Standard Methods for the Examination of Water and Wastewater, 20th ed. APHA-AWWA-WEF, Washington, DC. Lee, J., Ahn, W.Y., Lee, C.H., 2001. Comparison of the filtration characteristics between attached and suspended growth microorganisms in submerged membrane bioreactor. Water Res. 35 (10), 2435–2445. Lee, W., Kang, S., Shin, H., 2003. Sludge characteristics and their contribution to microfiltration in submerged membrane bioreactor. J. Membrane Sci. 216, 217–227. Leong, L.Y.C., 1983. Removal and inactivation of viruses by treatment processes for potable water and wastewater: a review. Water Sci. Technol. 15, 91–114. Maier, R.M., Pepper, I.L., Gerba, C.P., 2000. Environmental Microbiology. Academic Press, San Diego, CA.
ARTICLE IN PRESS C. Shang et al. / Water Research 39 (2005) 4211–4219 Melnick, J.L., 1984. Enteric Viruses in Water: Monographs in Virology, vol. 15. Karger, Basel. Metcalf & Eddy, Inc., 2003. Wastewater Engineering: Treatment and Reuse, fourth ed. McGraw-Hill, Boston (revised by Tchobanoglous, G., Burton, F.L., Stensel, H.D.). Mysore, C.S., Leparc, J., Poussade, Y., Jacangelo, J.G., Huffman, D.E., 2003. A pilot evaluation of low-pressure membranes for the removal of bacteria and viruses in municipal waters. In: Proceedings of the AWWA Annual Conference, Anaheim, CA. American Water Works Association, Denver, CO. Ridgway, H.F., Flemming, H.C., 1996. Membrane biofouling. In: Mallevialle, J., Odendaal, P.E., Wiesner, M.R. (Eds.), Water Treatment Membrane Processes. McGraw-Hill, Boston, pp. 6.1–6.62.
4219
van der Roest, H.F., Lawrence, D.P., van Bentem, A.G.N., 2002. Membrane Bioreactors for Municipal Wastewater Treatment, Water and Wastewater Practitioner Series: STOWA Report. IWA Publishing, London. Sutherland, I.W., Hughes, K.A., Skillman, L.C., Tait, K., 2004. The interaction of phage and biofilms. FEMS Microbiol. Lett. 232 (1), 1–6. Ueda, T., Horan, N.J., 2000. Fate of indigenous bacteriophage in a membrane bioreactor. Water Res. 34 (7), 2151–2159. Wong, H.M., Tam, L.S., Tang, T.W., Shang, C., Ho, K.M., Chen, G.H., 2003. Rejection of bacteriophage by membrane bioreactor. In: Proceedings of the International Conference on Pollution in the Metropolitan and Urban Environment, Hong Kong.
Water Research 39 (2005) 4211–4219 www.elsevier.com/locate/watres
Bacteriophage MS-2 removal by submerged membrane bioreactor Chii Shang, Hiu Man Wong, Guanghao Chen Department of Civil Engineering, Hong Kong University of Science and Technology, Clear Water Bay, Kowloon, Hong Kong Received 5 November 2004; received in revised form 10 June 2005; accepted 10 August 2005
Abstract A membrane bioreactor (MBR) may serve as a pre-disinfection or disinfection unit, in addition to its solid/liquid separation and biological conversion functions, to produce sewage effluent of high quality. This bench-scale pilot study focuses on investigating the performance of a submerged MBR in pathogen removal and the factors affecting the removal, using a 0.4-mm hollow-fiber membrane module submerged in an aeration tank and bacteriophage MS-2 as the indicator organism. Removal of the MS-2 phage was found to be contributed by physical filtration by the membrane itself, biomass activity in the aeration tank and bio-filtration achieved by the biofilm developed on the membrane surface. The membrane alone gave poor virus removal (0.470.1 log) but the overall removal increased substantially with the presence of biomass and the membrane-surface-attached biofilm. The contributions of the suspended biomass and attached biofilm to the phage removal are dependent on the inter-related parameters including the concentration of mixed liquor suspended solids (MLSS), the sludge retention time (SRT) and the food to mass (F/M) ratio. The correlations between effluent flux/trans-membrane pressure and virus removal give evidence that phage removal in the MBR is most likely susceptible to both biological and physical factors including the quantity and property of the biomass and the biofilm and the membrane pore size reduction. r 2005 Elsevier Ltd. All rights reserved. Keywords: Bacteriophage; Biofilm; Bioreactor; Disinfection; MBR; Membrane
1. Introduction Membrane separation has long been considered as a ‘‘safe and clean’’ physical means to yield some disinfection credits for water purification and reuse purposes. It reduces the problem associated with the generation of harmful disinfection by-products (DBPs) from chemical disinfection processes (e.g., chlorination or chloramination) by cutting down the chemical dosages. The membrane bioreactor (MBR) technology is similar to Corresponding author. Tel.: +852 2358 7885; fax: +852 2358 1534. E-mail address: [email protected] (C. Shang).
conventional activated sludge processes except that membranes (mostly microfiltration membranes) are used to extract effluent. Therefore, much higher concentrations of mixed liquor suspended solids (MLSS) (normally 45000 mg/L) and long sludge ages (commonly 420 days) can be kept in the aeration tank since the settlability of the flocs is not a concern. MBRs have demonstrated outstanding performance with respect to its biological conversion and solid removal (common values of effluent: TSSo5 mg/L, CODo40 mg/L, TNo12 mg/L and TPo2.2 mg/L) (Churchouse and Brindle, 2002; van der Roest et al., 2002; Wong et al., 2003). MBRs also provide advantages in sludge reduction (Metcalf & Eddy, Inc., 2003) and hence the sludge
0043-1354/$ - see front matter r 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2005.08.003
ARTICLE IN PRESS 4212
C. Shang et al. / Water Research 39 (2005) 4211–4219
treatment cost can be reduced. In addition to these profound features, MBRs are now being considered as an effective, non-hazardous alternative disinfection means to achieve pathogen control in wastewater effluents. Numerous studies have reported outstanding microbial removal achieved by MBRs when Escherichia coli, fecal coliform bacteria, and/or bacteriophages were used as the indicators (e.g., Churchouse and Brindle, 2002; van der Roest et al., 2002; Wong et al., 2003; Ueda and Horan, 2000). Churchouse and Brindle (2002) have demonstrated that a full-scale MBR plant produced exceptionally good-quality effluent over years including 3–6 log removal of fecal coliform bacteria and 2–5 log removal of F+ coliphage. A bench-scale pilot study by Ueda and Horan (2000) showed that an MBR could achieve 2–6 log removal of indigenous T-even-like bacteriophage (with a mean size of 200 nm) and almost complete removal of fecal coliform bacteria (up to 7 log). Only 0.91 log removal of the same type of phages could be achieved by the conventional activated sludge processes (Ueda and Horan, 2000). Therefore, with such high-quality effluent of MBR processes, dosages of disinfectants applied afterwards can be much reduced or the following disinfection processes may be completely eliminated. To assess the disinfection credit achieved by MBRs or to guarantee the effluent microbiological quality, viruses are expected to be more suitable indicator organisms than bacteria since viruses are much smaller and harderto-straining than bacteria and considered to be more resistant to common disinfectants (Leong, 1983). Meanwhile, the importance of removing viruses from wastewater is increasingly being recognized because of the epidemiological significance of viral pathogens (Melnick, 1984). Due to the difficulty in assaying animal viruses, bacteriophages have been suggested as viral indicators because they closely resemble enteric viruses in terms of the structure, morphology, size and behavior (Maier et al., 2000). Bacteriophage MS-2 (hereafter referred to as ‘‘MS-2’’) was chosen to be the indicator in this study in consideration of its smallest size (0.02–0.025 mm) among viruses and hence it is representative to address the ability of pathogen removal by an MBR. The mechanisms of virus rejection by MBRs are still ambiguous and the extent of virus removal is likely subject to the operational conditions. It has been reported that virus removal by an MBR became upgraded with the presence of biomass and the visible sticky biofilm developed on the membrane surface (Ueda and Horan, 2000), though the normal pore size of the microfiltration membrane (0.4 mm) was larger than the size of the tested phages (0.2 mm). However, the performance of removing viruses far smaller than the membrane pore size of an MBR and the affecting
operational parameters are not clear and the deficient knowledge in this area has prompted this study. This paper describes an experimental study on evaluating the performance of a submerged MBR in MS-2 removal and the factors affecting the removal efficiency. Bench-scale pilot MBR units were constructed using microfiltration hollow-fiber membrane modules with a nominal pore size of 0.4 mm. The overall removal efficiency and the contributions of solely the membrane, the suspended biomass, and the membrane-attached biofilm were examined. The affecting factors in evaluation included the concentration of MLSS, the sludge retention time (SRT), the food to mass (F/M) ratio, the operating flux and the concentration of MS-2.
2. Materials and methods 2.1. Bench-scale MBR Membrane modules used in this study were provided by the Mitsubishi Rayon Corporation (MRC). Table 1 shows the characteristics of the membrane module. The membrane module was submerged in a reactor and an air diffuser was equipped at the base of the reactor to provide both aeration and membrane surface cleaning (airflow rate of 3 L/min). The operating volume of the aeration tank was 19 L with a footprint area of 26 cm 25 cm. The effluent pump was operated at a constant suction force but in an intermittent suction mode (13 min ON/2 min OFF). The initial flow rate during the suction in a clear water tank with a clean membrane module was 50 L/day. The corresponding flux and hydraulic retention time (HRT) were 0.25 m/ day and 9 h, respectively. The intermittent ceasing of the suction and the scraping effect generated by the rising air bubbles prevented solid deposition on the membrane surface and extended the service period between membrane cleanings. Pressure gauges were installed in the effluent side for monitoring the changes in suction pressure (or trans-membrane pressure (TMP)). The effluent flux was recorded by measuring the effluent volume flowed in a definite time.
Table 1 Characteristics of the microfiltration membrane module Membrane type Material Nominal pore size (mm) Maximum allowable flux (m3/m2/ day) Total surface area (m2)
Microporous hollow fiber Polyethylene 0.4 0.4 0.2
ARTICLE IN PRESS C. Shang et al. / Water Research 39 (2005) 4211–4219 Table 2 Compositions of the organic nutrient stock and mineral water Organic nutrient stock (per 1 L of deionised water)
Mineral water (per 100 L of deionised water)
7.5 g Peptone 15 g Yeast extract
100 g NaCl 2.75 g CaCl2 2.25 g MgSO4 7H2O 0.025 g FeCl3 6H2O
The reactor was first fed with returned sludge taken from the Shatin Sewage Treatment Works, Hong Kong and cultivated in the laboratory. The reactor was fed continuously with synthetic water containing organic nutrients and mineral water with compositions shown in Table 2. The flows of the two waters were adjusted so that, before a trial test began, the biomass could be stabilized with a specific level of MLSS and a sludge age with a specific F/M ratio and at a designated flow rate. It was presumed that the system reached stabilization when both the MLSS level and the sludge age (or the amount of sludge wasted every day) kept constant for more than a week. The bioreactors were operated under ambient temperature (18–22 1C), pH 5–8 and dissolved oxygen of 2–3 mg/L. 2.2. MS-2 culture preparation and assay Bacteriophage MS-2 stocks (ATCC 15597-B1) were prepared in suspension using their host cells E. coli (ATCC 15597) solutions. Active cultures of the E. coli host were first cultivated for 18–24 h at 37 1C in the medium suggested by the American Type Culture Collection. The MS-2 stock was then added into the active host cultures and the infected bacteria were incubated for 18–24 h to allow the propagation of the phage. Calcium chloride was added to promote the adsorption of MS-2 to the bacterial cells (Adams, 1959). The cultured MS-2 solutions were used directly without further treatment. The concentrations of the free-flowing MS-2 in samples were enumerated using the Double Agar Layer Method (Adams, 1959). An immobilized phage would develop a semi-transparent circle, or a plaque, to indicate the lysis of the host cells and represent the progeny of a single immobilized phage. The number of plaques (plaque forming unit or PFU) can represent the number of phages present in a suspension of a given volume. 2.3. Experimental procedures A test run was always first conducted in a ‘‘clean system’’ by using a cleaned membrane in a tank
4213
containing only synthetic water without any suspended solids. The MS-2 removal obtained in the clean system was recorded as the ‘‘baseline removal’’, which allowed differentiation of the intrinsic removal capacity of the membrane from the additional removal induced by biomass and/or biofilm in the following test runs. Membrane cleaning to remove biofilm deposition from the previous run to restore the original TMP was conducted by backwashing the membrane module with a 600 mg/L hypochlorite solution for 15 min. The cleaned membrane was then submerged in a reactor containing suspended biomass stabilized at specific values of MLSS and SRT. To study the effects of MLSS, the biomass was stabilized at 6000, 8000 and 10,000 mg/L with the same SRT of 200 days. The corresponding F/M ratios were 0.05, 0.04 and 0.03/day, respectively. For examining the effects of SRT, the biomass was stabilized at a SRT of 50 and 200 days with the same MLSS of 6000 mg/L. The corresponding F/M ratios were 0.06 and 0.05/day, respectively. The MBR units were operated continuously for a period of time to allow the development of membrane-attached biofilm. Whichever biomass was present in the tank, the units were fed with a dosage of the MS-2 solution at concentrations of around 107 PFU/mL. Samples were periodically collected from the aeration tanks (the bulk solution) and the effluent outlet of the MBR units, after initiating test runs with rapid mixing. If a test run lasted for more than a few hours, the tanks were periodically fed with the MS-2 solution at same dosages. The samples taken from the bulk solution were filtrated with 0.45 mm filter papers before the viral assay was made so that only free-flowing (not floc associated) MS-2 was measured. The purpose of sample filtration was to distinguish the removal by the membrane (including the biofilm) and that by the suspended biomass, where the latter was examined in separated tests without the membrane and effluent discharge. At different time intervals, a sample was taken from the mixed liquor, filtered with a 0.45 mm filter paper and subjected to MS-2 numeration. At least three samples were tested at each time and some tests were repeated under the same conditions for quality control. The biofilm formation in the repeated tests was generally reproducible. Removal efficiency can be expressed as log removal value (LRV): LRV ¼ log
C , C0
(1)
where C0 and C refer to the free-flowing MS-2 concentrations before and after treatment, respectively. When membranes were involved in tests, the C0 was taken from the mixed liquor and the C was taken from the effluent outlet. Under such cases, LRVm, was reported, which refers to the log removal by the membrane (or biofilmed membrane) and has excluded
ARTICLE IN PRESS C. Shang et al. / Water Research 39 (2005) 4211–4219
those phages associated with the suspended solids. In addition, a normalized log removal value (NLRVm) is sometimes reported in this paper, which has subtracted the ‘‘baseline removal’’ obtained in the ‘‘clean system’’. This normalization allows differentiation of the additional removal induced by biofilm from the intrinsic removal capacity of the membrane (Mysore et al., 2003). Last, it should be noted that the term ‘‘biofilm’’ used here and in the following discussion refers to the phageresistant layer gradually developed on the outer and inner membrane surface during MBR operation, though the roles of the deposited solids and the attached biomass cannot be distinguished and the enhanced virus removal efficiency is likely attributable to both. 2.4. Other measurements The concentration of MLSS in the MBR was measured in accordance with Greenberg et al. (Standard Methods) (1998). Sludge bioactivities for both organic decomposition and nitrification was measured and characterized by the specific oxygen uptake rate (SOUR), which is also known as the oxygen consumption or respiration rate. It is defined as the milligram of oxygen consumed per gram of volatile suspended solids (VSS) per hour. The detailed procedure can be referred to that stated in Greenberg et al. (Standard Methods) (1998).
3.0 Log Removal Value, LRVm
4214
2.5 2.0 1.5 1.0 0.5 0.0 0
5
10 15 Time (day)
20
25
Fig. 1. Phage removal by the biofilmed membrane at MLSS ¼ 6000 mg/L and SRT of 50 days.
membrane cleaning to control the effluent microbiological quality by an additional disinfection process. For comparison, using a hollow-fiber membrane with a same pore size of 0.4 mm but T-even-like phage of 0.2 mm (ten times larger than MS-2) at a much higher initial flux (1.2 m/d, which was 4–5 times higher than that used in the current study), Ueda and Horan (2000) demonstrated a negligible phage removal (about 0.1 to 0.3-log) in a ‘‘clean system’’.
3.2. Phage removal by biofilmed membrane 3. Results and discussion 3.1. Phage removal by un-biofilmed membrane Phage removal in the MBR was proposed to be primarily contributed to the biofilm developed on the membrane surface, which physically reduced the membrane pore size, chemically adsorbed phages, and biologically allowed the predation of phages by other microorganisms (Ueda and Horan, 2000). Here, we examined the relative contributions of the membrane only, the biomass only, and the membrane-attached biofilm to the phage removal. The membrane itself provides the first obstruction for pathogen removal because of the sieving effect. The contribution of solely the membrane to phage removal (referred to as the ‘‘baseline removal’’ in Fig. 1) stayed at about 0.3–0.4 log throughout the examination period. Leakage of phages in such a ‘‘clean system’’ in the absence of both biomass and biofilm is expected since the average pore size of the membrane fibers (0.4 mm) is much larger than the size of the bacteriophage MS-2 (0.02 mm). It should be noted (data not shown) that the baseline removal consistently restored to similar values (70.1 log) after 15 min backwash with a 600 mg/L hypochlorite solution. Therefore, in practice, attention should be paid right after
By submerging the membrane into an aeration tank containing stabilized biomass under an MLSS of 6000 mg/L and a SRT of 50 days, as shown in Fig. 1, phage removal improved over time as a result of the development of a biofilm layer on the membrane surface. It can be explained that the virus transport is hindered due to the pore reduction and phage adsorption onto the biofilm (Ueda and Horan, 2000). The statement was supported by visual and scan electron microscopy (SEM) inspections of the membrane surface. After submerging the membrane module in the reactor for a few weeks, a slime or gel layer was visually observed. SEM inspection of the external surface (on the feed side) of the biofilmed membrane showed that the surface was fully covered with biofilm layers, while internal blockage and partial coverage presumably by extracellular polymeric substances (EPS) were observed in the membrane pores by taking SEM images from the internal surface on the permeate side. Fig. 1 also shows that, in the first 2 days, fastest improvement of phage removal was recorded. A ‘‘primary biofilm’’ (Ridgway and Flemming, 1996) was said to have formed. Thereafter, the phage removal improved gradually, which can be attributed to the formation of the so-called ‘‘secondary biofilm’’ (Ueda and Horan, 2000). The error bars as
ARTICLE IN PRESS C. Shang et al. / Water Research 39 (2005) 4211–4219
shown in Fig. 1 indicate that more than 0.5 log removal can be considered as significant.
4215
3.0
3.3. Phage removal by suspended biomass Phages are also expected to associate with biomass flocs and then get removed by flocculation or cell adsorption in the aeration tank. Fig. 2 shows the phage removal in the absence of the membrane in two tanks with biomass stabilized at two specific conditions: (A) MLSS of 6000 mg/L, SRT of 50 days, and an F/M ratio of 0.06/day, and (B) MLSS of 10,000 mg/L, SRT of 200 days, and an F/M ratio of 0.03/day. The phage removal under the two conditions increased over the sampling time and reached 0.85 and 0.6 log for F/M ratios of 0.06 and 0.03/day, respectively, after 9 h (one HRT of the MBR system). Although a higher F/M ratio seems to give higher phage removal, such small difference is not statistically significant. The phage removal achieved under both conditions in the current study was slightly less than that (0.91 log) of a conventional activated sludge plant that was operated at HRT of 9.7 h, MLSS of 1420 mg/L, and an F/M ratio of 0.62/day (Ueda and Horan, 2000). Combining the data obtained from the two studies, the bioadsorbility of the suspended biomass is unlikely to be considerably affected by the HRT and the biological operational factors. 3.4. Contributions of the MBR components Fig. 3 summarizes the individual contribution of the MBR components to the phage removal under MLSS of 6000 mg/L and SRT of 50 days. The contribution of the membrane alone is indicated by ‘‘Membrane’’ while ‘‘9-h-Biomass’’ refers to the removal by the suspended biomass in the absence of membrane after one HRT (9 h). The ‘‘9-h-Biofilm’’ and ‘‘21-day-Biofilm’’ refer to the NLRVms by the biofilm after submerging a clean 3.0 MLSS=6,000mg/L, SRT=50day, F/M=0.06/d
Log Removal Value
2.5
MLSS=10,000mg/L, SRT=200day, F/M=0.03/d
2.0 1.5
2.0 1.5 1.0 0.5 0.0 Membrane
9-hr-Biomass
9-hr-Biofilm 21-day-Biofilm
Fig. 3. Contributions of the MBR components to phage removal at MLSS of 6000 mg/L and SRT of 50 days.
membrane in the aeration tank for one HRT and 21 days, respectively. The NLRVms have excluded the contribution from biomass and the ‘‘baseline removal’’ by the sole membrane. As shown, the contributions of the membrane, the suspended biomass, the 9-h-biofilm and the 21-day-biofilm to the phage removal in the MBR are 0.4 log, 0.8 log, 0.3 log and 2.1 log, respectively. The phage removal by the suspended biomass contributed a good portion of the phage removal, especially when a clean or cleaned membrane was submerged in the aeration tank up to 9 h. Nevertheless, given enough time to allow the biofilm to develop, the removal by the biofilm improved significantly and it was the biofilm developed on the membrane surface playing the most important role in removing the MS-2 phage. The development of a biofilm on the membrane surface was proposed to make the major contribution to pathogen removal (Ueda and Horan, 2000) and biofilms were suggested to trap and accumulate virus-sized particles (Flood and Ashbolt, 2000; Sutherland et al., 2004). The above findings indicate that a real MBR system may achieve higher viral removal since, in realworld operations, chemical cleaning of membranes is commonly conducted every few months so that the contribution from the biofilms can be more significant. In addition, the suspended solids in the sewage may also adsorb phases or contribute to membrane pore blockage to enhance phage removal. 3.5. Effects of MLSS levels on phage removal
1.0 0.5 0.0
Log Removal Value
2.5
0
2
4 6 Time (hr)
8
10
Fig. 2. Phage removal by suspended biomass in the absence of membrane under the specific conditions.
To investigate how the characteristics of biomass affected the biofilm formation and hence the phage removal in the MBR, MS-2 phage removal by the biofilm developed on the membrane surface at three different MLSS levels (6000, 8000 and 10,000 mg/L), which were commonly applied in full-scale MBR systems, was examined. The SRT was 200 days in all cases. Fig. 4 presents the results. The LRVs are the
ARTICLE IN PRESS C. Shang et al. / Water Research 39 (2005) 4211–4219
4216 3.0
1
Flux / Initial Flux
2.5
NLRVm
2.0 1.5 1.0
0.95
0.9
0.85
0.5 MLSS=6,000
0.0
MLSS=8,000
MLSS=10,000
MLSS=6,000
MLSS=8,000
MLSS=10,000
0.8 0
2
4
6 8 Time (Day)
10
12
0
14
5
10 Time (day)
(a)
Fig. 4. Phage removal by the biofilm at different MLSS levels (baseline removal excluded).
15
20
3
3.6. Effects of sludge age on phage removal Fig. 6 shows phage removal (NLRVm) in the MBR stabilized at an MLSS of 6000 mg/L under two SRTs (50 and 200 days). Although the concentrations of MLSS
TMP - TMPo (kPa)
2.5 2 1.5 1 0.5 MLSS=6,000
MLSS=8,000
MLSS=10,000
0 0
2
4
6 8 Time (day)
(b)
10
12
14
Fig. 5. Changes of (a) flux/initial flux ratios and (b) TMP under different MLSS levels at the same SRT of 200 days.
3.0 2.5 2.0 NLRVm
NLRVms, which refer to only those achieved by the biofilm formed on the membrane surface. The phage removal at MLSS of 6000–10,000 mg/L showed rapid improvement during the initial stage of filtration (with 6000 mg/L gave slightly higher removal values) and followed by gradual enhancement of removal to reach similar removal values. It was originally postulated that more suspended solids would deposit on the membrane surface if the concentration of biomass in the reactor was higher; however, the fact that the similar or lower removal with increasing MLSS during the tested period was observed. The finding indicates that the nature of the biomass (e.g., the amount/composition of EPS), which shall change with the change of MLSS or the F/M ratios, is more important than the concentration of that in influencing the phage removal. As discussed previously, the biofilm layer developed on the membrane surface (internally or externally) does control pathogen removal the most in the MBR. The extent of membrane fouling is commonly quantified by the changes of the effluent flux or the TMP. In the cases of MLSS of 6000 and 8000 mg/L, the flux dropped (Fig. 5a) and the TMP increased (Fig. 5b), in accordance with the increase in phage removal (Fig. 4b). However, larger fluctuations in the flux and the TMP were observed at 10,000 mg/L MLSS, implying that the biofilm accumulated cannot smooth out the flow instability. Lee et al. (2001) reported that higher MLSS concentrations (in the range of 2000 and 5000 mg/L) alleviated an increase in TMP. The data shown in Fig. 5 support the statement and extend its validity up to 8000 mg/L MLSS; however, it is not the case at 10,000 mg/L MLSS.
1.5 1.0 0.5
MLSS=6000mg/L, SRT=50d MLSS=6000mg/L, SRT=200d
0.0
0
2
4
6
8 10 Time (day)
12
14
16
Fig. 6. Phage removal by the biofilm developed under the same MLSS level but different solid retention time (baseline removal subtracted).
were the same for both cases, the system with longer SRT yielded at least 1 log higher MS-2 removal. The system with the SRT of 200 days also gave faster improvement in phage removal than that of 50 days. Together with the discussion in Section 3.5, phage
ARTICLE IN PRESS C. Shang et al. / Water Research 39 (2005) 4211–4219
removal is susceptible to the changes in operational conditions such as the MLSS level and the SRT which in turn influence the characteristics of the membraneattached biofilm. Recall that a higher F/M ratio (or lower MLSS at fixed SRT) yielded faster and larger improvement in phage removal in Section 3.5, while a lower F/M ratio (or longer SRT at fixed MLSS) yielded better phage removal here. This contradiction suggests that the phage removal is controlled by the inter-related parameters including MLSS, SRT and F/M ratios. 3.7. Correlations between phage removal and membrane fouling Fig. 7 shows the correlations between the normalized phage removal by the biofilm developed under the specific conditions and the corresponding TMP. As shown, at an F/M ratio of 0.06, TMP can be used to estimate the phage removal in the MBR. However, TMP cannot reflect phage removal when comparing results obtained under different conditions. In fact, as F/M ratios decreased, the correlation between TMP and phage removal diminished. Hence, membrane fouling is not always correlated well with phage removal. Since TMP is only a measurement of physical blockage of the membrane, the results also indicate that blockage or membrane fouling is not the only factor in affecting phage rejection. Fig. 7 shows that TMP increases to larger extent at a higher F/M ratio while giving lower NLRVm, compared to those obtained at lower F/M ratios. In order to interpret the above phenomena, the relationship between operational conditions and sludge characteristics could be first explored. It has been reported in the literature that the overall EPS production is low at a long SRT (or perhaps at a low F/M ratio) (Chang and Lee, 1998). This explains the TMP increase at a SRT of 200 days was lower (because of lower total EPS
4217
production) than the data obtained at a SRT of 50 days, although the MLSS levels were the same. Lee et al. (2003) suggested that the ratio of protein to carbohydrate in EPS is more important than the quantity of EPS in controlling filtration resistance or membrane fouling. At high SRT or lower F/M ratios, the carbohydrate in microbial flocs declined while the amount of protein on the cell surface increased (Lee et al., 2003). This may explain how the operational conditions affect the characteristics of the suspended biomass and attached biofilm and hence the phage removal by the two phases. Lee et al. (2003) also suggested that a longer SRT led to decreases in microbial activity in MBRs at a SRT of 20–60 days. The current study, however, showed that the microbial activity, measured as the SOUR, of the biomass with SRTs of 50 and 200 days was similar and at about 50 mg O2/g VSS/h, indicating that, at higher SRTs, the microbial activities cannot be used to explain the difference in biomass characteristics. 3.8. Effects of phage concentration and effluent flux Fig. 8 illustrates the relationship between the influent and effluent phage counts for three cases at various influent phage concentrations and effluent fluxes. For the case of a clean system (clean membrane in clear water), the LRVs are independent of the influent phage concentration; implying that the water quality of the effluent depends on that of the influent. This is obvious because the pore size of the membrane is much larger than the size of the phage; more phages pass through the membrane as more phages are present in the reactor. The other two cases are log removal of phages by the biofilmed membrane submerged in clear water at different effluent flow rates. With the presence of biofilm, the LRVm increases with increasing the influent phage concentration. These results are in agreement with Ueda and Horan (2000) who concluded that phage removal is independent of phage concentration in the
3.0
3 2.5
2.5 2 1.5
LRVm
NLRVm
2.0
Clean System Biofilm @30 L/d Biofilm @50 L/d
1.0
0.0
1
MLSS=6000, SRT=50, F/M=0.06 MLSS=6000, SRT=200, F/M=0.05 MLSS=8000, SRT=200, F/M=0.04 MLSS=10000, SRT=200, F/M=0.03
0.5 0
1
3 4 2 TMP-TMPo (kPa)
5
1.5
0.5 6
Fig. 7. Correlations between phage removal (by the biofilm only) and the changes of trans-membrane pressure (TMP) (with initial TMP subtracted) under different operational conditions.
0 103
104 105 106 107 108 109 Phage Concentration in Bulk Solution (PFU/ml)
Fig. 8. Relationship between phage removal and phage concentrations in the bulk solution at different operating fluxes.
ARTICLE IN PRESS 4218
C. Shang et al. / Water Research 39 (2005) 4211–4219
absence of biomass or biofilm; while it becomes dependent on the concentration with the presence of biomass or biofilm. The biofilm present in the reactor helps reduce the pore size of the membrane and hence the passage of phages becomes limited. Therefore, the phage concentration in the effluent may be relatively stable even with fluctuation in the influent phage concentration. Fig. 8 shows that phage removal decreases with increasing effluent flux but the effect appears to highly depend on the dosing concentration of phages. However, the effect of operating flux on phage removal is expected to be negligible in real-world applications since the concentration of phages found in sewage is about 102–105 PFU/mL, which is much lower than the concentration used in this study. Besides, it should be noted that the study of the effect of the effluent flux was conducted after the biofilm had developed. It is likely that the operating flux can affect the development of the biofilm.
4. Conclusions Phage removal in the MBR can be attributed to three components: the membrane, the biomass and the membrane-attached biofilm. Only 0.470.1 log reduction of bacteriophage MS-2 could be achieved by the clean membrane in the absence of biomass. At 6000-mg/L MLSS, 200-day SRT, and 9-h HRT, the suspended biomass contributed to 0.8 log phage removal. When immersing the membrane in the stabilized biomass, the overall removal increased substantially after sufficient time (over weeks) was given to allow the development of biofilm on the membrane surface. Therefore, phage removal in the MBR improves by prolonging the operational period between membrane cleanings. However, after each cleaning, special attention should be paid to control the effluent pathogen concentrations by an additional disinfection process. The characteristics of the biomass and the biofilm and their contributions to phage removal are influenced by the inter-related parameters including the concentration of MLSS, the SRT and the F/M ratio. Under the tested conditions, operating at lower MLSS (among 6000, 8000 and 10,000 mg/L with same SRT) or longer SRT (among 50 and 200 days with same MLSS) yields faster and larger improvement in phage removal. Lowering the F/M ratios diminishes the correlation between TMP and phage removal. In addition, the correlations between phage removal and trans-membrane pressure (TMP) conclude that membrane fouling is merely one of the several factors that influence phage removal in the MBR. Phage removal in the MBR is most likely susceptible to biological factors such as the quantity and property of the biofilm.
Last, phage removal improves as the biofilm is allowed to develop, but this decreases the effluent flux or increases the TMP to reduce the capacity of the MBR system or to increase the pumping cost. Balancing this dilemma is of great importance if the effluent pathogen control relies fully or partially on the disinfection credit granted to an MBR. It should be noted that although bacteriophage MS-2 was used to indicate the likelihood pathogen removal by MBR in this study, it is possible that the surface characteristics of viruses and phages are of great variety and their adsorption to the complex biofilm/biomass may also vary. Acknowledgments The insightful comments of three anonymous reviewers are gratefully acknowledged. This study was supported in part by the Hong Kong Research Grants Council under grant DAG03/04.EG25. The bench-scale membrane modules were provided by the Mitsubishi Rayon Corporation. Certain commercial materials are identified in this paper to specify the experimental procedures. Such identification does not imply that the materials are necessarily the best available for the purpose.
References Adams, M.H., 1959. Bacteriophages. New York, NY, Interscience. Chang, I.S., Lee, C.H., 1998. Membrane filtration characteristics in membrane coupled activated sludge system—the effect of physiological states of activated sludge on membrane fouling. Desalination 120, 221–233. Churchouse, S., Brindle, K., 2002. Operational experience of full scale membrane bioreactor sewage treatment plants. In: Proceedings of the IWA Third World Water Congress, International Water Association, Melbourne, Australia. Flood, J.A., Ashbolt, N.J., 2000. Virus-sized particles can be entrapped and concentrated one hundred fold within wetland biofilms. Adv. Environ. Res. 3, 403–411. Greenberg, A.E., Clesceri, L.S., Eaton, A.D. (Eds.), 1998. Standard Methods for the Examination of Water and Wastewater, 20th ed. APHA-AWWA-WEF, Washington, DC. Lee, J., Ahn, W.Y., Lee, C.H., 2001. Comparison of the filtration characteristics between attached and suspended growth microorganisms in submerged membrane bioreactor. Water Res. 35 (10), 2435–2445. Lee, W., Kang, S., Shin, H., 2003. Sludge characteristics and their contribution to microfiltration in submerged membrane bioreactor. J. Membrane Sci. 216, 217–227. Leong, L.Y.C., 1983. Removal and inactivation of viruses by treatment processes for potable water and wastewater: a review. Water Sci. Technol. 15, 91–114. Maier, R.M., Pepper, I.L., Gerba, C.P., 2000. Environmental Microbiology. Academic Press, San Diego, CA.
ARTICLE IN PRESS C. Shang et al. / Water Research 39 (2005) 4211–4219 Melnick, J.L., 1984. Enteric Viruses in Water: Monographs in Virology, vol. 15. Karger, Basel. Metcalf & Eddy, Inc., 2003. Wastewater Engineering: Treatment and Reuse, fourth ed. McGraw-Hill, Boston (revised by Tchobanoglous, G., Burton, F.L., Stensel, H.D.). Mysore, C.S., Leparc, J., Poussade, Y., Jacangelo, J.G., Huffman, D.E., 2003. A pilot evaluation of low-pressure membranes for the removal of bacteria and viruses in municipal waters. In: Proceedings of the AWWA Annual Conference, Anaheim, CA. American Water Works Association, Denver, CO. Ridgway, H.F., Flemming, H.C., 1996. Membrane biofouling. In: Mallevialle, J., Odendaal, P.E., Wiesner, M.R. (Eds.), Water Treatment Membrane Processes. McGraw-Hill, Boston, pp. 6.1–6.62.
4219
van der Roest, H.F., Lawrence, D.P., van Bentem, A.G.N., 2002. Membrane Bioreactors for Municipal Wastewater Treatment, Water and Wastewater Practitioner Series: STOWA Report. IWA Publishing, London. Sutherland, I.W., Hughes, K.A., Skillman, L.C., Tait, K., 2004. The interaction of phage and biofilms. FEMS Microbiol. Lett. 232 (1), 1–6. Ueda, T., Horan, N.J., 2000. Fate of indigenous bacteriophage in a membrane bioreactor. Water Res. 34 (7), 2151–2159. Wong, H.M., Tam, L.S., Tang, T.W., Shang, C., Ho, K.M., Chen, G.H., 2003. Rejection of bacteriophage by membrane bioreactor. In: Proceedings of the International Conference on Pollution in the Metropolitan and Urban Environment, Hong Kong.