Professional Documents
Culture Documents
52020-2OBrienandXagoraraki2020JEE-MBRvirusesinvitedpaper9pages (1)
52020-2OBrienandXagoraraki2020JEE-MBRvirusesinvitedpaper9pages (1)
net/publication/342603304
CITATIONS READS
25 241
2 authors:
All content following this page was uploaded by Irene Xagoraraki on 11 May 2021.
Abstract: This review seeks to summarize the current literature regarding the removal of viruses from wastewater using membrane
bioreactors (MBRs). Membrane bioreactors are an effective technology for the removal of viruses from wastewater and do so with greater
efficiency than conventional activated sludge treatment plants. However, much is unclear about the capabilities and mechanisms of this pro-
cess. Membrane pore size is commonly thought to be an important factor impacting virus removal, but this effect is primarily observed when
the pore size is smaller than the virus diameter, and the effects vary by virus. A variety of factors can impact the effectiveness of MBRs in virus
Downloaded from ascelibrary.org by Evan O'Brien on 04/25/20. Copyright ASCE. For personal use only; all rights reserved.
removal, with the presence of a biofilm foremost among them. Balancing the presence of a biofilm with regular membrane backwashing and
cleaning to maintain consistent flux remains a critical consideration in MBR operation. Differences in reported removal efficiencies of human
viruses compared to model viruses (such as bacteriophages) call into question the reliability of these model viruses in assessing the effective-
ness of virus removal in MBRs. More investigation, particularly with full-scale MBR systems, must be performed to further understand the
potential of MBRs in virus removal. DOI: 10.1061/(ASCE)EE.1943-7870.0001743. © 2020 American Society of Civil Engineers.
however, it will quantify viruses that are both infective and non- Virus Removal in Full-Scale MBRs
infective. Plaque assay, meanwhile, will only quantify infective
Many studies have investigated the removal of viruses by full-
Downloaded from ascelibrary.org by Evan O'Brien on 04/25/20. Copyright ASCE. For personal use only; all rights reserved.
Table 2. Summary of prior studies investigating virus removal in full-scale membrane bioreactors
Membrane Removal
pore Capacity Detection efficiency
size (um) Location (m3 =d) MBR type Virus methods (logs) Reference
0.4 Northwest France 390 DNS Norovirus I qPCR 0–5.3 da Silva et al. (2007)
Norovirus II 0–5.5
0.1 Traverse City, MI 32,000 Submerged Human adenovirus qPCR 4.1–5.6 Kuo et al. (2010)
0.4 Bologna, Italy 1,935.36 Submerged F-specific coliphage Plaque assay 6.0 Zanetti et al. (2010)
Somatic coliphage 4.0
0.45 Northwest France 270 Submerged Norovirus qPCR 3.3–6.8 Sima et al. (2011)
Sapovirus 1.8–4.1
0.1 Traverse City, MI 32,000 Submerged Human adenovirus qPCR 3.4–4.5 Simmons and
Enterovirus 2.9–4.6 Xagoraraki (2011)
0.1 Traverse City, MI 32,000 Submerged Human adenovirus qPCR 4.1–6.3 Simmons et al. (2011)
Enterovirus 4.1–6.8
Norovirus II 3.5–4.8
0.4 Northern OH 12,900 DNS F-specific coliphage Plaque assay 4.58–6 Francy et al. (2012)
Somatic coliphage 2.67–4.04
Human adenovirus qPCR 2.38–4.86
Enterovirus 2.2–4.74
Norovirus I 1.51–3.32
0.4 Bologna, Italy 1,935.36 Submerged F-specific coliphage Plaque assay 5.77 De Luca et al. (2013)
Somatic coliphage 4.35
0.04 American Canyon, CA 5,700 Submerged Human adenovirus qPCR 3.9–5.5 Chaudhry et al. (2015b)
Norovirus II 4.6–5.7
F-specific coliphage 5.4–7.1
0.04 London, UK 574 Submerged Somatic coliphage Plaque assay 5.3 Purnell et al. (2015)
F-specific coliphage 3.5
Human-specific coliphage 3.8
DNS Southern CA 105,992 Side-stream Total viruses Flow cytometry 4.0 Huang et al. (2016)
0.4 Thuwal, Saudi Arabia 1,600 Submerged Human adenovirus dPCR 3.7 Jumat et al. (2017)
Enterovirus 1.7
0.4 Northwest France 1,800 Submerged Sapovirus qPCR 3.0 Miura et al. (2018)
Rotavirus 2.0
Norovirus 3.0
0.05 Sao Paulo, Brazil 164,160 Submerged Norovirus I qPCR 1.1 Prado et al. (2019)
Norovirus II 1.2
Note: DNS = did not specify.
Despite the effectiveness of MBR systems and disinfection tech- membrane pore size on removal efficiency of viruses. One study
nologies such as chlorine and ultraviolet light, human viruses have measured removal of bacteriophage MS2 (size of 0.025 μm) in
still been detected in wastewater effluent after these treatment steps membranes with pore sizes of 0.1, 0.08, 0.05, 0.04, and 0.03 μm.
by many of these studies (Francy et al. 2012; Huang et al. 2016; No significant differences in removal efficiency of indigenous MS2
Jumat et al. 2017; Simmons and Xagoraraki 2011). Other studies were observed among the different membranes (Hirani et al. 2010).
investigating treated effluent of MBR plants have reached the same Other studies also showed no significant difference in the removal
conclusion, detecting human viruses in effluent with both molecu- of bacteriophage T4 (size of 0.11 μm) between two membranes
lar and metagenomic methods (O’Brien et al. 2017a). with pore sizes of 0.1 and 0.22 μm (Lv et al. 2006; Zheng et al.
2005). Another study, however, produced different results, with
removal efficiency of bacteriophage T4 reported as 2.1 logs for
Parameters That Affect Virus Removal a membrane with 0.22 μm pore size and 5.8 logs for a membrane
with 0.1 μm pore size. Furthermore, removal efficiency of bacte-
In addition to the study of full-scale utilities, bench- and pilot-scale
riophage f2 (size of 0.025 μm) was reported as 0.3 and 0.5 logs for
MBR systems can assist in the evaluation of different factors on the
the 0.22 μm and 0.1 μm membranes, respectively (Zheng and
removal efficiency of viruses from water and wastewater. Numer-
Liu 2007).
ous studies have investigated a variety of factors that can impact
removal efficiency of viruses, and these factors are summarized
in the subsequent section. Membrane Biofilm
The role of the membrane biofilm in virus removal by MBRs has
Membrane Pore Size been studied in several bench-scale experiments. One such experi-
One point of differentiation among the various full-scale studies is ment (Wu et al. 2010) found that a clean membrane with a pore size
the pore size of the membrane used in the MBR facilities. A chart of 0.4 μm contributed only approximately 0.5 logs removal of so-
summarizing removal of viruses in relation to pore size is presented matic coliphages. However, when covered with a biofilm, the same
in Fig. 1. While the studies are limited in number, for some viruses, membrane could remove 1.8 to 2.6 logs of the virus. Similarly,
studies utilizing smaller pore sizes led to higher log removal values another study (Shang et al. 2005) observed that an MBR with
than studies with larger pore sizes. Log removal of adenovirus has the nominal pore size of 0.4 μm could initially (i.e., prior to mem-
been reported in the range of 2.38–4.86 with a membrane pore size brane fouling) remove just 0.4 logs of MS-2 coliphage. After
of 0.4 μm (Francy et al. 2012; Jumat et al. 2017), whereas ranges 21 days of operation, the removal efficiency increased to 2.3 logs;
from 4.1–6.3 have been reported with a pore size of 0.1 μm (Kuo the study concluded that membrane biofilm played an important
et al. 2010; Simmons et al. 2011). These studies find the same for role in removing the virus. However, another study indicated that
enterovirus; log removal of 1.7–4.74 is reported for membranes the presence of a biofilm was not necessarily a reliable indicator in
with a pore size of 0.4 μm (Francy et al. 2012; Jumat et al. removal efficiency of adenovirus (Yin et al. 2015).
2017), while another study reports log removal of 4.1–6.8 with Different elements of the MBR biofilm have been found to have
a pore size of 0.1 μm (Simmons et al. 2011). Norovirus, mean- different contributions to the removal of viruses from wastewater.
while, does not follow this trend; for example, one study using One bench-scale study (Lv et al. 2006) determined that the cake
a membrane pore size of 0.45 μm reported log removal from layer and gel layer of the membrane that develops during operation
3.3–6.8 (Sima et al. 2011), with another study utilizing a membrane make major contributions to virus removal. In the study, the oper-
pore size of 0.05 μm reporting log removal of 1.1 (Prado et al. ation of the membrane with developed gel and cake layers was per-
2019). To confirm this observation, reported log removal values formed, followed by destruction of the cake layer and operation of
of norovirus by a membrane pore size of 0.4 μm or greater were the membrane with the gel layer, and finally removal of the gel
compared with those by a pore size of 0.1 μm or less using a t-test. layer and operation of the membrane itself. The experiment resulted
No significant difference was found between the two groups in 1.7 log, 3.1 log, and 6.3 log removal of bacteriophage T4 for the
(p-value ¼ 0.62). 0.22 μm membrane itself, the membrane with the gel layer, and the
Other full-scale studies report similarly inconsistent log removal membrane with the gel and cake layers, respectively. This study
values for model bacteriophages with regards to pore size. For in- also found that for the 0.1 μm membrane investigated, the mem-
stance, one study investigating somatic coliphages reported a log brane alone was responsible for complete virus removal, implying
removal from 2.67–4.04 using a membrane with a 0.4-μm pore size that pore size can also impact which membrane components
(Francy et al. 2012), and two other studies reported log removal become responsible for removal. Another study observed that the
results of 4.0 (Zanetti et al. 2010) and 4.35 (De Luca et al. 2013). permeability of the cake layer can determine whether the removal
(a) (b)
(c) (d)
(e)
Fig. 1. Summary of studies reporting log removal of viruses in full-scale MBR systems in relation to pore size for (a) adenovirus; (b) enterovirus;
(c) norovirus; (d) F-specific coliphage; and (e) somatic coliphage. Values are represented as the maximum, central tendency, and minimum as reported
by the references. *Reference only reported an average value. ^Reference only reported minimum and maximum values.
efficiency of viruses increases or decreases due to cake formation membrane and therefore lowers flow rate and efficiency of the
(Yin et al. 2015). treatment plant. Because of this, cleaning of the membrane via
backwashing and chemicals is important to treat sufficient quan-
tities of water. Numerous bench-scale studies have investigated
Membrane Cleaning the effects of membrane cleaning on virus treatment, often leading
While membrane fouling leads to development of a biofilm that to lower levels of removal of bacteriophages (Hirani et al. 2010;
aids in virus removal, fouling also decreases flux through the Lv et al. 2006; Tam et al. 2007; Zheng et al. 2005). Another study
uted to increased fouling levels at lower sparging rates. els within minutes.
Membrane Flux
Virus Type and Concentration
One aforementioned bench-scale study (Wu et al. 2010) also sub-
jected a clean membrane to different operational fluxes and mea- Certain human viruses appear to be more easily removed by full-
sured the efficiency of the membrane in the removal of somatic scale MBRs than others. Studies measuring the removal of both
coliphage. The 0.4-μm clean membrane was subjected to fluxes human adenovirus and enterovirus reported log removal values that
of 12.5, 10.0, and 7.5 L m−2 h−1 and yielded log removal values were higher for adenovirus compared to enterovirus within the
of 0.7, 0.5, and 0.6 logs, respectively. Therefore, operational flux same study, sometimes to a large degree (Francy et al. 2012;
was not found to have a significant impact on removal efficiency of Jumat et al. 2017; Simmons and Xagoraraki 2011), although one
somatic coliphage. However, when applying different fluxes to a study reported the two viruses as having approximately equivalent
continuously operating membrane with developed cake and gel log removal values (Simmons et al. 2011). Other studies that did
layers, the experiment found that the removal efficiency of the not investigate both viruses together reported consistent log re-
gel layer increased with increased flux. The study claims that moval values, on the order of 4 log removal. Norovirus, meanwhile,
higher flux leads to higher rate of fouling, enhancing the develop- was sometimes reported to have lower log removal values com-
ment of the gel layer and its ability to entrap viral particles. Mem- pared to adenovirus and enterovirus (Francy et al. 2012; Simmons
brane flux may also impact virus removal in other ways. One study et al. 2011), while one study reported log removal values equal or
(Hirani et al. 2010) found that changes in flux can affect the MLSS higher than adenovirus (Chaudhry et al. 2015b). Other studies in-
concentration in an MBR, and another study found that MLSS vestigating norovirus report conflicting values as well, with one
concentration is related to removal efficiency of certain viruses reporting 1.1–1.2 log removal (Prado et al. 2019) and others rang-
(Miura et al. 2015). ing from 3.3 to 6.8 log removal (Sima et al. 2011). Another study
surveying an MBR plant over the course of several months found a
Membrane Coatings relatively constant effluent concentration of norovirus regardless of
influent concentration (da Silva et al. 2007).
Photocatalytic coatings have been found to improve removal effi- Bench-scale studies also show that certain viruses are often
ciency of viruses when paired with ultraviolet (UV) disinfection. found to undergo higher removal efficiencies in MBR systems than
One bench-scale study (Guo et al. 2015) investigated the removal others. One study (Ottoson et al. 2006) compared the removal ef-
of bacteriophage P22 using a membrane with a pore size of 0.8 μm ficiency of four viruses (enterovirus, norovirus, somatic coliphage,
in which the membrane was coated with a photocatalyst (TiO2 ). and F-specific coliphage) with a 0.4-μm membrane. Somatic col-
The coated membrane was coupled with UV to make a hybrid iphages and F-specific coliphages were found to have mean log
MF-UV treatment process. This process was considerably more ef-
removal values of 3.08 and 3.78, respectively, while enterovirus
fective at inactivating P22 (log removal of 5.0) than the membrane
and norovirus were determined to have mean log removal of
without the photocatalytic coating (log removal of 2.4) or micro-
1.79 and 1.14, respectively. Not only does this difference show
filtration and UV disinfection applied in series (log removal of 2.0).
a distinction between removal of different human viruses, but it also
Similar results were observed when measuring the inactivation of
brings into question the use of bacteriophages as models for human
human adenovirus (HAdV) in a subsequent study again utilizing a
viruses. Another study (Miura et al. 2015) using a 0.4-μm MBR
TiO2 -coated 0.8-μm membrane (Guo et al. 2018), with removal of
system also investigated removal of enterovirus and norovirus,
infectious HAdV being a full log greater for the hybrid photoca-
talytic process than the membrane alone or the hybrid nonhotoca- as well as sapovirus, via monthly sampling over a period of
talytic process. 16 months. Similar results were found, with enterovirus and nor-
ovirus having log removal values in the range of 0.3–3.2 (mean of
1.6) and 0.2–3.4 (mean of 1.3), respectively. The two viruses were
Mechanical Failure nearer in log removal for this study, but removal of enterovirus still
Various mechanical failures can impact MBR operation, including outpaced that of norovirus. Sapovirus, meanwhile, had higher re-
aeration failure, washout of biomass, compromised membranes via moval efficiency than both of the other viruses, with log removal
fiber abrasion, and breakage of membrane fibers. One bench-scale values in the range of 1.3–4.1, again demonstrating that virus
study investigated the effects of these events (Branch et al. 2016). type can impact removal efficiency in MBRs. The study suggested
A control experiment was first performed with normal operation that the differences observed among the three viruses could be
of the 0.04-μm membrane to determine log removal of FRNA attributed to how well the virus associates with MLSS, with
4e8 copies=mL corresponded to log removal values of 0.5, 1.1, the same size yet exhibit different trends of log removal with re-
and 1.8 respectively. This effect was attributed to a more rapid de- gards to membrane pore size. The same is true when investigating
velopment of an irreversible fouling layer on the membrane. It is the removal of bacteriophages in full-scale systems; Fig. 1 shows
noteworthy, however, that this trend implies that the effluent con- that while the removal of somatic coliphage may be impacted by
centration was independent of the influent concentration of HAdV pore size, removal of F-specific coliphages does not appear to have
in this experiment. any relationship with pore size. This difference is also observed in
studies investigating the effects of pore size utilizing bench-scale
Wastewater Characteristics systems. Removal of bacteriophages in bench-scale MBR systems
often do not exhibit any relationship with pore size, although some
Characteristics of the wastewater influent itself have also been studies indicate that pore size may matter when it is smaller than
found to impact removal efficiency of viruses in MBRs. Another virus size. A previous paper summarized removal of bacteriophages
bench-scale study (Branch et al. 2016) investigated the impact of in relation to pore size and reached similar conclusions (Xagoraraki
high concentrations of salt (NaCl), ammonia (NH3 ), and COD on et al. 2014). Ultimately, though, the available data is limited, and
the removal efficiency of FRNA bacteriophages by a 0.04-μm more studies are needed to better establish the relationship between
membrane. To establish baseline log removal, normal operation pore size and removal of different viruses.
of the membrane with no chemical additives yielded log removal Further complicating the use of bench-scale systems to study
values in the range of 4.1–7.3. The addition of 20 g=L NaCl low- virus removal is the fact that bacteriophages used to model enteric
ered the removal efficiency from 5.4 logs before NaCl addition to human viruses often exhibit log removal values considerably higher
3.9 logs 1 day following NaCl addition; this reduction was theo- than the human viruses, even within the same study. For example,
rized to be attributed to a reduction of virus adsorption to activated one study (Ottoson et al. 2006) reported much higher log removal
sludge flocs due to increased ionic strength. A concentration of for bacteriophages (3.08 for somatic coliphage, 3.78 for F-specific
0.7 g=L NH3 also depressed removal efficiency to from 5.6 logs
coliphage) compared to human viruses (1.79 for enterovirus, 1.14
to 4.4 logs, and this change was attributed to the increase in pH
for norovirus). Based on the available data from the reference, a
limiting FRNA aggregation. A 5 g=L COD concentration, mean-
two-tailed t-test determined that the phage removal values were sig-
while, significantly increased removal of FRNA to >6.9 logs, sig-
nificantly different from the human virus removal values to >99%
nifying complete removal of the virus. This change was attributed
confidence (p-value ¼ 0.002). The same is true in studies that have
to the reduction in pH inducing higher degrees of phage adsorption
quantified both bacteriophages and human viruses in full-scale
to other phages and flocs.
MBR systems. Fig. 1 shows that somatic coliphage may appropri-
The addition of coagulants has been found to improve the re-
ately model human viruses in full-scale systems, but this is not the
moval of viruses. One bench-scale study (Fiksdal and Leiknes
case for F-specific coliphage; one full-scale study (Chaudhry
2006) investigated removal of bacteriophage MS2 with a 0.2-μm
et al. 2015b) reported removal of F-specific coliphage at higher
membrane and found little to no removal with no coagulants added
levels than that of adenovirus and norovirus. Both of the studies
to the influent stream. Addition of aluminum-based coagulants
mentioned in this paragraph noted that the observed disparities
improved removal efficiency to log removal ranges of 6.7–7.5, a
could arise from differences in quantification methods between
dramatic increase compared to that with no coagulation.
phages and human viruses. Nonetheless, this discrepancy compli-
cates comparisons between removal of phages and human viruses.
Discussion Until there is adequate reconciliation of these quantification meth-
ods, the use of phages as model organisms may overestimate the
It is clear that membrane bioreactors are highly efficient at remov- actual effectiveness of MBR systems to remove human viruses.
ing viruses from wastewater. What remain unclear in the current Moreover, the variation in removal efficiency observed among dif-
literature are the precise mechanisms and factors that contribute ferent virus types makes it a challenge to determine the use of a
to this removal. The review of studies investigating full-scale MBR single model virus.
systems shows conflicting information about which viruses are Nonetheless, model bacteriophages are still useful on the
most efficiently removed in MBRs. Some studies that compared bench-scale as predictors of the efficiency of full-scale operation.
multiple human viruses within the same experiment report higher While direct comparisons are difficult to perform due to the variety
log removal values for adenovirus compared to others, such as of confounding variables, comparisons of studies utilizing the
enterovirus, while others show no marked difference between the same virus and membrane pore size tend to show similar results
two. Moreover, removal efficiency for norovirus appears to vary between bench- and full-scale operation of MBRs. For example,
from study to study, with log removal ranging from 1.1 to 6.8 a bench-scale study reported 4.1–7.3 log removal of F-specific
MF/RO systems.” Desalination 202 (1): 106–113. https://doi.org/10 reuse in the USA.” In Advanced treatment technologies for urban
.1016/j.desal.2005.12.045. wastewater reuse, edited by D. Fatta-Kassinos, D. D. Dionysiou, and
USEPA. 2015. Review of coliphages as possible indicators of fecal con- K. Kümmerer, 223–245. Cham, Switzerland: Springer.
tamination for ambient water quality. 820-R-15–098. Washington, Zanetti, F., G. De Luca, and R. Sacchetti. 2010. “Performance of a full-scale
DC: USEPA. membrane bioreactor system in treating municipal wastewater for reuse
Wong, K., T.-T. Fong, K. Bibby, and M. Molina. 2012. “Application of purposes.” Bioresour. Technol. 101 (10): 3768–3771. https://doi.org/10
enteric viruses for fecal pollution source tracking in environmental .1016/j.biortech.2009.12.091.
waters.” Environ. Int. 45 (Sep): 151–164. https://doi.org/10.1016/j Zheng, X., and J. Liu. 2007. “Virus rejection with two model human enteric
.envint.2012.02.009. viruses in membrane bioreactor system.” Sci. China, Ser. B Chem.
Wu, J., H. Li, and X. Huang. 2010. “Indigenous somatic coliphage removal 50 (3): 397–404. https://doi.org/10.1007/s11426-007-0047-3.
from a real municipal wastewater by a submerged membrane bio- Zheng, X., W. Lü, M. Yang, and J. Liu. 2005. “Evaluation of virus removal
reactor.” Water Res. 44 (6): 1853–1862. https://doi.org/10.1016/j in MBR using coliphages T4.” Chin. Sci. Bull. 50 (9): 862–867. https://
.watres.2009.12.013. doi.org/10.1360/04wb0087.