See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/279312641

Evaluation of methods for reverse osmosis membrane integrity monitoring

for wastewater reuse

Article in Journal of Water Process Engineering · September 2015

DOI: 10.1016/j.jwpe.2015.05.001

CITATIONS READS

21 976

6 authors, including:

Eric (Xiao) Huang Sunny Jiang

California Institute of Technology University of California, Irvine
28 PUBLICATIONS 799 CITATIONS 111 PUBLICATIONS 7,253 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

National Science Foundation Partnerships for International Research and Education View project

All content following this page was uploaded by Eric (Xiao) Huang on 16 March 2018.

The user has requested enhancement of the downloaded file.

 Journal of Water Process Engineering 7 (2015) 161–168

Contents lists available at ScienceDirect

Journal of Water Process Engineering

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

Evaluation of methods for reverse osmosis membrane integrity

monitoring for wastewater reuse
Xiao Huang a , Joon H. Min b , Wenzhou Lu a,c , Kris Jaktar b , Christopher Yu d ,
Sunny C. Jiang a,∗
a
Civil and Environmental Engineering, University of California, Irvine, USA
b
BKT, 1225 N. Patt Street, Anaheim, CA, USA
c
College of Architectural, Civil Engineering and the Environment, Ningbo University, China
d
Montrose Environmental Group, Irvine, CA, USA

a r t i c l e i n f o a b s t r a c t

Article history: Wastewater reclamation with advanced membrane technology holds great potential to supplement
Received 18 December 2014 the diminishing drinking water supply for human consumption. High-pressure reverse osmosis (RO)
Received in revised form 29 April 2015 membrane processes offer a high level of pathogen removal capacity. However, the lack of recognized
Accepted 1 May 2015
membrane integrity monitoring methods has restricted the pathogen removal credits allocation to the
Available online 26 June 2015
processes. This research investigated the feasibility of using flow cytometry (FCM), dynamic light scat-
tering (DLS) analyzer and traditional water quality analyzers for RO membrane integrity monitoring.
Keywords:
BD AccuriTM C6 flow cytometer demonstrated good sensitivity and reproducibility for quantifying virus
Membrane integrity
Reverse osmosis
reduction rate along the treatment processes, which provide direct evidence for RO membrane integrity
Flow cytometry monitoring. DLS (Nanotrac Ultra) showed promise to be used as a qualitative membrane integrity moni-
Dynamic light scattering toring tool by characterizing particle size distributions in water. Traditional water quality analyzers were
TOC tested online in a pilot RO system with intentional introduced integrity breaches. Total organic carbon
(TOC) measurements showed the best sensitivity to reflect different levels of integrity breaches. Feasi-
bility analysis based on the instrument sensitivity, capital, maintenance and operating costs shows that
an integrated system including more than one monitoring tools would be more reliable and economical
for high-pressure membrane integrity monitoring.
© 2015 Elsevier Ltd. All rights reserved.

1. Introduction States, only 7–8% of municipal wastewater effluent is currently

being reclaimed and mainly for non-potable applications [1]. Large
The intensification of potable water scarcity has become a global scale indirect potable reuse projects are still restricted in a few
issue affecting the economic and social development in many coun- states (e.g., California, Florida and Texas) [2]. The primary concern
tries. With climate change, population growth and increasing water associated with wastewater reclamation is the public health risk
demand, the problem cannot be solved without the exploration caused by the potential exposure to microbial contaminants, such
of new water supplies. Municipal wastewater is one of the “new” as pathogenic protozoa, bacteria and viruses. Currently, reverse
water resources which has not been fully tapped. In the United osmosis (RO) membrane is the most widely used technology for
high quality reclaimed water production. Microfiltration (MF) or
ultrafiltration (UF) is usually coupled with RO to mitigate the
fouling problem of RO membranes [3,4]. Low-pressure membrane
Abbreviations: CDPH, California department of public health; DLS, dynamic light systems (MF, UF) can remove most of the protozoa and bacteria,
scattering; DOM, dissolved organic matter; EC, electrical conductivity; EEM, fluo- but they have limited virus removal capacity. High-pressure RO
rescence excitation-emission matrix; EFM, epifluorescence microscopy; FCM, flow
systems, in theory, completely remove all microorganisms includ-
cytometry; FL1, florescence channel 1; FL3, florescence channel 3; MF, microfiltra-
tion; RO, reverse osmosis; SSC, side scatter; TDS, total dissolved solids; TOC, total ing viruses, as RO systems are designed for nanosized particles and
organic carbon; UF, ultrafiltration; WRD, water replenishment district. ion removal. The combination of RO with MF or UF offers a high
∗ Corresponding author at: 716 E Engineering Tower, University of California, level of pathogen removal capacity, which can produce reclaimed
Irvine, The Henry Samueli School of Engineering, Irvine, CA 92697-7070 USA. water meeting drinking water standards. However, pathogens may
Fax: +1 949 824 3672.
pass through the membrane barriers when the system integrity is
E-mail address: [email protected] (S.C. Jiang).

http://dx.doi.org/10.1016/j.jwpe.2015.05.001
2214-7144/© 2015 Elsevier Ltd. All rights reserved.
162 X. Huang et al. / Journal of Water Process Engineering 7 (2015) 161–168

compromised. For example, oxidation damage to the active layer of seawater, wastewater contains more fluorescent interference sub-
RO membranes can happen when the feed water contains chlorine stances, such as organic matters, heavy metals and even some
residual or other oxidants [5]. The aging of O-rings used to isolate autofluorescent particles. Using FCM for direct virus detection in
the feed and permeate streams on the product core tube could also reclaimed water needs further investigation. A new nanoparticle
result in the loss of system integrity. Due to the lack of means to analyzer, dynamic light scattering (DLS) analyzer, was also used
accurately monitor RO membrane integrity, most states currently in this study to detect viral sized particles in treated wastewa-
do not allocate any specific pathogen removal credits to RO pro- ter. Unlike FCM, DLS targets all submicron particles in suspension
cesses [6]. For example, the California Department of Public Health including inorganic particles. Thus, it can be used as a surrogate
(CDPH) requires 12-log removal of viruses from raw wastewa- method for virus detection. DLS is based on the theory that par-
ter to final effluent for indirect potable reuse. While 6-log virus ticles in suspension are under Brownian motion and the speed of
removal credits were allocated to traditional wastewater treat- the particles is in reverse proportion to their sizes. When moving
ment processes with advanced disinfection treatments, no removal particles are illuminated with a laser (photons), the scattered light
credit was given to RO processes. The currently approved California fluctuating rates will be recorded and the particle sizes are deter-
potable reuse projects were required to store the highly purified mined by using the Stokes-Einstein equation [17]. As both FCM and
RO permeate in an environmental buffer system (i.e., groundwa- DLS are not able to process continuous water samples at the cur-
ter basin) for at least 6 months to acquire additional 6 logs virus rent stage, grab samples from a wastewater reclamation plant were
decay credits (one month per log) before it can be used as drink- used during the study. Considering this limitation, new models of
ing water sources. The rationality of this regulation is currently in water quality analyzers (TDS, turbidity and TOC) with improved
debate [2]. A monitoring system that sensitively detects RO mem- sensitivity and online capability were also tested in a pilot RO sys-
brane integrity failures will install confidence among regulatory tem set up in the same plant. A feasibility analysis was carried out
agencies for virus removal credits allocation. at the end of the study based on the instrument sensitivity, capital,
Due to their small sizes (20–220 nm), low infection dose and maintenance and operating costs.
high resistance to some commonly used disinfection processes,
viruses in wastewater are the primary microbiological regulatory
2. Materials and methods
targets for high-pressure RO systems [7,8]. The ultimate goal of
RO membrane integrity monitoring is to ensure no or only accept-
2.1. Description of the water reclamation plant
able level of viruses can pass through the membrane barriers.
Membrane integrity monitoring techniques are usually classified
Water samples were collected from the Leo J. Vander Lans
as direct or indirect methods. Direct methods refer to tests that are
Advanced Water Treatment Facility of Water Replenishment Dis-
applied to the membrane or the membrane module, such as pres-
trict (WRD) of Southern California. Sampling points (see Supporting
sure/vacuum hold, diffusive air flow and bubble point test. These
information Fig. S1, indicated by triangles) were selected based on
methods are mainly employed by membrane manufacturers before
the treatment processes of the plant, which include MF inflow, MF
membrane installation, as they can only be conducted off-line. In
outflow, RO feed and RO permeate. After purging the sampling valve
contrast, indirect methods are based on the feed and permeate
by running the water for 5 min, 1 l of water was collected in 1.5 l
water quality. A baseline is first established with intact membranes
sterile Whirl-Pak sampling bag and stored in an ice box during
for certain constituents in permeate. The membrane integrity prob-
transportation to the lab. Fresh samples were tested by FCM and
lems are reflected by the deviations from the established baseline.
DLS within 4 h after sample collection.
TOC, turbidity and electrical conductivity (EC) are indirect methods
based on constituents naturally present in the feed stream. But their
sensitivity has been questioned due to the high purity of the RO 2.2. Optimization of direct virus enumeration by FCM
permeate [9,10]. Recently, fluorescence excitation-emission matrix
(EEM) spectroscopy was proposed for RO membrane integrity mon- The BD Accuri C6 flow cytometer (Accuri Cytometers, Ann Arbor,
itoring by analyzing the dissolved organic matter (DOM) in RO feed MI, USA) was used in this study for direct virus enumeration. The
and permeate [11]. Although EEM provides a wealth of information sample pre-treatment method of Brussaard [18] was used as “the
about DOM, identifying fluorescent signatures for calculating DOM reference protocol” for trouble shooting issues associated with
rejection rate can be difficult due to the stochastic nature of DOM in the testing of reclaimed water. Briefly, 1 mL water samples were
wastewater [12]. To improve the sensitivity and specificity, seed- fixed with glutaraldehyde (0.5% final concentration); the samples
ing studies by adding bacteriophages and artificial fluorescent dyes were then flash frozen with liquid nitrogen and stored at –80 ◦ C;
(e.g., Rhodamine-WT) to the feed water are still used occasionally samples were thawed at 35 ◦ C and diluted with TE-buffer (10 mM
[13]. Considering the advantages and disadvantages of currently Tris–HCl,1 mM EDTA; pH 8.0) before testing; the staining was car-
available methods, an ideal RO membrane integrity monitoring ried out in the dark at 80 ◦ C water bath with 0.5X SYBR Green I
method should include the following features: (1) targeting viruses (final concentration) for 10 min. This initial protocol resulted in
naturally present in the feed water; (2) sensitive enough to demon- high background noise, which could not be clearly separated from
strate the required removal rate of viruses; (3) online capability to the virus signal. To determine the best sample pre-treatment pro-
provide real-time or near real-time water quality information of cedures, a MF inflow sample from WRD was tested under different
the permeate. sample fixation, staining and dilution conditions (Table 1). The
In this study, flow cytometry (FCM) was first time tested as a same sample was also filtered with 30 Kilodalton (KDa) molecular
RO membrane integrity monitoring tool to detect total virus par- mass cut-off Amicon Ultra-4Centrifugal Filters (Millipore, Bed-
ticles including bacteriophages in treated wastewater. FCM is an ford, MA) to remove all the viruses and used as a blank to assess
accurate and fast method for analyzing biological particles in sus- the reagent noise as well as determine the detection limit of the
pension. It has been employed for enumeration of total bacteria in method. The BD Accuri C6 flow cytometer threshold was set at
drinking water [14]. The viability of bacteria can also be assessed 450 nm in fluorescence channel 1 (FL1) to exclude the instrument
by staining the samples with different fluorescent dyes [15]. Using electronic noise. Green and red fluorescence were collected in the
FCM for virus detection is more challenging due to their small FL1 channel (533 ± 30 nm) and the FL3 channel (>670 nm) on a log-
sizes. Marie et al. [16] first reported the enumeration of viruses arithmic scale, respectively. Data analysis was carried out using
in seawater by staining samples with SYBR-Green I. Compared to the BD CFlow® software. The total virus counts were obtained by
 X. Huang et al. / Journal of Water Process Engineering 7 (2015) 161–168 163

Table 1
Optimization of sample pre-treatment protocol for enumeration of virus particles using flow cytometry. A microfiltration inflow sample from WRD was tested under different
conditions. Data of relative counts were normalized to the highest count obtained (=1).

Fixation Dye Liquid N2 Incubation Diluent Relative counts

frozen

Glutaraldehyde concentration SYBR-Green I SYBR-Gold

0 0.5% 1% 2% 5% 0.5x 1x 2x 0.5x 1x 2x − + RTa (15min) 80 ◦ C (10min) TE MQb

Best * * * * * 1
1 * * * * * 88.9%
2 * * * * * 83.1%
3 * * * * * 87.8%
4 * * * * * 102.7%
5 * * * * * 71.6%
6 * * * * * 71.7%
7 * * * * * 79.7%
8 * * * * * 85.9%
9 * * * * * 83.6%
10 * * * * * 44.8%
11 * * * * * 66.4%
12 * * * * * 80.2%
13 * * * * * 85.3%
14 * * * * * 92.9%
15 * * * * * 19.4%
16 * * * * * 30.5%
17 * * * * * 44.3%
18 * * * * * 42.2%
19 * * * * * 54.8%
20 * * * * * 94.9%
21 * * * * * 82.5%
22 * * * * * 94.4%
23 * * * * * 79.5%
24 * * * * * 69.7%
26 * * * * * 88.9%
26 * * * * * 80.6%
27 * * * * * 21.2%
a
RT, room temperature.
b
MQ, Milli-q water.

subtracting the counts in blank from the counts in the sample. To 2.4. RO pilot study
validate the accuracy of virus counts by FCM, pure cultures of bac-
teriophage T4 (ATCC 11,303-B4) and MS2 (ATCC 15,597-B1) were A pilot scale RO system was set up at WRD to evaluate the
seeded in autoclaved 1x phosphate-buffered saline (PBS buffer) as feasibility of using new models of water quality analyzers for
positive control samples. The results from FCM were compared online membrane integrity monitoring. The MF outflow of the
with those obtained by the double agar layer titration method plant was used as feed water for the system. The pilot system is
[19] and epifluorescence microscopy (EFM) direct counting method a two-stage RO system using a 3:1 array (Spiral wound, Polyamide
[20]. The optimized protocol was then used to monitor the virus membrane, CSM, Anaheim, CA), where the concentrate from the
removal efficiencies of membrane processes at WRD for a month first stage membranes is treated again by the second stage mem-
long period. brane. The pilot system includes built-in permeate and concentrate
flow meters, feed and permeate pressure gauges and a conductiv-
ity meter, which monitored TDS results continuously with data
recorded hourly. Fig. 1 shows the schematics of the RO pilot lay-
out. Valves were installed at the end of the permeate lines through
T-connectors to throttle the flow to the moitoring instruments. A
GE Sievers 900 Online TOC Analyzer (GE Analytical Instruments,
2.3. Detection of nanosized particles by DLS
Boulder, CO) was installed to record the TOC changes of RO per-
meate with an interval of every 4 min. The operating range was
Grab samples from WRD were tested with the Nanotrac Ultra
set from 1 ␮g/L to 1000 ␮g/L. A HACH FilterTrak 660sc (Hach Com-
(Microtrac, Montogomeryville, PA) DLS system. The instrument
pany, Loveland, CO) was employed to measure the turbidity of the
theoretically can detect particle size from 0.8 to 6500 nm and is
RO permeate flow at an interval of every 5 min. The operational
optimized to detect particles in low concentration suspensions. The
range of the analyzer is 0–5000 mNTU with a resolution as low as
manufacturer’s testing procedure was followed for the reclaimed
0.3 mNTU. After system start up, the RO pilot system was contin-
water samples. First, the sampling cell was thoroughly flushed
uously operated and the baseline conditions were established for
with distilled water to remove any residual particles before test-
different monitoring parameters in the first 24 h. The system oper-
ing. Analysis time of 90 s was used for all the samples and the
ating conditions are summarized in Supporting Information Table
cleaner samples (with low particle concentration) were always
S1. Later, the O-ring on the high pressure end (feed inlet) of the
tested first in multiple sample runs. The particle size distributions
second pass pressure vessel was artificially damaged (Fig. 1, the
were displayed based on % volume of total particles. To determine
star indicates the location of damaged O-ring) to simulate a typical
the concentration threshold of detection, RO feed samples from
failure of RO systems [9]. Five levels of damage were introduced
WRD were diluted in distilled water to the final concentrations of
to the O-ring (stretched, cut off, two notches, three notches and
2%, 4% 12% and 20% (volume). The threshold was determined when
four notches). The pilot was run under each testing condition for
a signal above the background noise was detected.
164 X. Huang et al. / Journal of Water Process Engineering 7 (2015) 161–168

Fig. 1. RO pilot layout. Star on the graph indicates the location of the compromised O-ring.

24 h. The changes of TDS, turbidity and TOC were recorded con- neous increase in background noise. Side scatter (SSC) vs. green
tinuously and the sensitivity of different monitoring parameters fluorescence (FL1) plots were used to separate viruses from back-
was compared with the baseline. The turbidity data for the last two ground noise in the reference protocol [18]. However, in this study,
days could not be retrieved from the turbidity meter due to a mal- we found that using green fluorescence (FL1, 533 ± 30 nm) vs. red
function of the HACH controller. To remedy the lost data, turbidity fluorescence (FL3, >630 nm) plots provided better discrimination
values measured from grab samples were entered. FCM and DLS when samples were stained with SYBR-Gold (Fig. 2). The emis-
were not tested during the pilot study due to the unavailability of sion spectrum of SYBR-Gold is 500–700 nm in comparison with
the equipment at the time. 500–625 nm for SYBR-Green I. This shift towards red fluorescence
by SYBR-Gold may explain the better discrimination when using
FL1 vs. FL3 plot [23]. Fig. 2 shows one of the MF inflow samples
3. Results and discussion
tested under the optimized sample pre-treatment protocol. The
Virus signal was clearly separated from background noise in the
3.1. Optimize FCM assay for reclaimed water
gated R1 region (Fig. 2b), while the virus-free blank had minimal
interference within the same gated area (Fig. 2a). Based on the back-
Although the sensitivity of FCM has been significantly improved
ground counts (background noise) in R1 region of the virus-free
by the utilization of high efficient fluorescent nucleic acid dyes
blank (Fig. 2a), the detection limit of FCM for viruses in reclaim
such as SYBR-Green and SYBR-Gold, the detection of viruses is
water was determined as ∼6 × 104 VLPs/mL. A previous study by
still approaching the detection limit of the instrument [14]. A total
Tomaru and Nagasaki [24] showed that FCM and EFM counts of
of 28 testing trials were carried out in an attempt to optimize
large DNA algal viruses (∼200 nm) were similar to each other. How-
the sample pre-treatment protocol for reclaimed water (Table 1).
ever, FCM underestimated smaller DNA and RNA viruses (∼40 nm)
The results showed that samples stained with SYBR-Gold generally
because the FCM counts were lower than the results from the
yielded higher particle counts than those stained with SYBR-Green
culture-based titration method (most-probable-number). In the
I (Table 1). This observation is in agreement with the report by Chen
current study, viruses counted by EFM and FCM were generally
et al. [21], who employed EFM for virus enumeration. Besides stain-
comparable and they were both higher than plaque assay counts
ing, the sample fixation procedure also significantly affected the
for double stranded DNA coliphage T4 and single stranded RNA col-
virus counts (Table 1). Glutaraldehyde is widely used as a preser-
iphage MS2 (Table 2). The standard deviations of FCM counts were
vative to prevent the degradation of viruses and to improve the
always less than 5%. Current results indicate that the BD Auccri
incorporation of dye into the viral DNA/RNA [22]. Based on current
C6 can effectively capture DNA (T4), RNA (MS2) viruses as well as
results, 2% glutaraldehyde fixation was necessary to separate the
viruses naturally presented in reclaimed water.
virus signal from the background noise. Although 5% glutaralde-
hyde fixation had slightly higher counts than 2% fixation (Table 1),
it also resulted in the increase of background noise in certain sam- 3.2. Monitoring virus removal efficiency by membrane processes
ples (data not shown). Thus, 2% glutaraldehyde fixation was chosen
and used for all samples in the later tests. Freezing samples with liq- A month long monitoring study was carried out to examine
uid nitrogen and heating samples at 80 ◦ C for 10 min were expected the virus removal efficiency by membrane processes using grab
to improve the combination of dye and viral DNA/RNA [18]. How- samples (40 total) from WRD. Total virus particles in MF inflow
ever, both procedures resulted in lower virus counts, and thus are ranged between 7 × 107 and 2 × 108 /ml over the one-month sam-
not suggested for reclaimed water samples (Table 1). A nearly 80% pling period (Fig. 3). The virus counts for MF outflow and the RO
decrease of virus counts was found in samples diluted with Milli- feed were nearly identical (Fig. 3). The MF removed one to two logs
q water compared to those diluted in TE buffer, which indicates of viruses, which agrees with previous studies using other count-
the importance of maintaining pH between 7.5–8.0 (preferably 8.0) ing methods [25,26]. No viruses were detected by FCM in the ten
as both SYBR-Green I and SYBR-Gold are pH sensitive (Table 1). RO permeate samples collected during the testing period (Fig. 3).
Among three different dye concentrations tested, 0.5X dye concen- The result was also confirmed by EFM (data not shown). Overall,
tration gave the highest counts. Beyond this value, increasing dye current results indicate that FCM can reliably quantify virus con-
concentrations resulted in lower virus counts due to the simulta- centration changes in water reclamation processes. Considering the
 X. Huang et al. / Journal of Water Process Engineering 7 (2015) 161–168 165

Fig. 2. Flow cytometric analysis of virus samples using FL1 vs. FL3 density plots. The samples were stained with 0.5X SYBR-gold (final concentration). FL1 is fluorescence
channel 1 that captures green fluorescence and FL3 is fluorescence channel 3 that captures red fluorescence. Gate R1 was used to separate viruses from other particles and
background noise. (a) 30 KDa membrane filtered microfiltration inflow sample from WRD was used as control to show regent background within the gated region; (b) a
microfiltration inflow sample from WRD demonstrates the separation of virus signal from background noise. The total virus count was obtained by subtracting the counts in
the blank (a) from the counts in the sample (b).

Table 2
Comparison of viral counts by different methods.

Plaque assay (PFU/mL) Epifluorescence microscopy (VLPs/mL)a Flow cytometry (VLPs/mL)

Phage MS2 (8.3 ± 1.3) x 109

(1.3 ± 0.3) x 10
10
(9.4 ± 0.06) x 1010
Phage T4 (7.7 ± 2.6) x 108 (2.6 ± 0.1) x 109 (3.0 ± 0.02) x 109
MF influent N.A.b (7.3 ± 1.2) x 107 (1.4 ± 0.02) x 108
MF effluent N.A. (7.7 ± 2.5) x 106 (7.3 ± 0.06) x 106
RO feed N.A. (1.3 ± 0.3) x 106 (6.2 ± 0.09) x 106
RO permeate N.A. B.D.c B.D.
Detection limit 1 1 × 105 6 × 104
a
VLPs, virus like particles.
b
N.A., data not available.
c
B.D., below detection limit.

has been developed for bacterial detection in drinking water [29].

With the flow injection and staining units, the sampling and stain-
ing steps are fully automated. Similar units may also be connected
with the Accuri C6 system to realize online detection of viruses.

3.3. Particle size distributions by DLS

The particle size and particle size distributions (% volume) ana-

lyzed by Nanotrac Ultra for samples from WRD are shown in
Fig. 4. MF inflow, which is the secondary effluent from the adja-
cent wastewater treatment plant, displayed a wide range of particle
peaks (Fig. 4a). The results demonstrate the heterogeneous nature
of particles in the secondary effluent. In a few samples, particles as
large as 6500 nm (the upper detection limit) were detected, which
could be unsettled activated sludge flocs. Commercial MF mem-
branes generally have nominal pore size range from 0.1–0.5 ␮m.
Theoretically, particles larger than the pore size (e.g., most bacte-
ria) are removed by MF through sieving. The buildup of a cake layer
Fig. 3. Variations in concentration of virus-like particles (VLPs/ml) at different sam- on the membrane surface during the course of filtration may further
pling points of WRD. Grab water samples were tested by FCM using optimized improve the removal of particles [30]. A well-defined separation for
protocol. The standard deviations of FCM counts were always less than 5% and are particles larger than 100 nm was observed for MF outflow (Fig. 4b),
not plotted on the graph.
which agrees with the size exclusion feature of MF. However, some
particles larger than 100 nm were detected in RO feed samples
virus concentration in MF inflow and the detection limit of FCM, at (Fig. 4c). Considering there is a storage tank between the sampling
least 3 logs removal of total virus particles for the MF/RO systems point of MF outflow and RO feed, those large particles detected in
can be demonstrated by FCM, although the actual virus removal rate the RO feed may indicate the regrowth of bacteria in the tank. No
could be higher. To further characterize the removal efficiency of RO distinct peaks larger than 2 nm were observed in the RO permeate
membranes, virus concentration methods using nano-fiber filters samples (Fig. 4d). Similar profiles were observed when testing with
[27] or a microfluidic system [28] could be employed to increase the distilled water. As those observed peaks (< 2 nm) were close
the detection limit of FCM. Recently, a lab-scale online FCM system to the theoretical detection limit (0.8 nm) of the instrument, they
166 X. Huang et al. / Journal of Water Process Engineering 7 (2015) 161–168

Fig. 4. Particle size distributions obtained from Nanotrac Ultra for samples from each sampling location: (a) MF inflow; (b) MF outflow; (c) RO feed; (d) RO permeate.

could be caused by vibrations or other background interferences. the turbidity in the permeate line was between 11.93 and 13.09
Since the majority of viruses are greater than 20 nm in diameter, mNTU, while the TDS was stable at 5 mg/L. The TOC baseline was
setting the size threshold at 10 nm should allow the distinction between 36.3 and 55.4 ␮g/L, indicating some low molecular weight
of viral size particles from the instrument noise. If particles larger organics are able to pass through the RO membranes. An intentional
than 10 nm are detected in RO permeate, it is likely that the system system breach by using a stretched O-ring was insufficient to com-
integrity is compromised. promise the system integrity (Fig. 5). However, a more aggressive
Besides particle size, the detection limit of DLS is also related to damage, by cutting the O-ring completely (open O-ring), showed
particle concentrations. According to DLS manufacturer’s research, that all three online water quality parameters significantly deviated
the concentration threshold of 30 nm diameter biological particles from the baseline (Fig. 5). Further experiments with notches cut on
is about 0.1 ppm (volume) [31]. However, the result was based on the O-ring also resulted in detectable system integrity breaches.
single size/property particles. It cannot be simply extrapolated to Current results showed that the new generations of water quality
environmental water samples, which normally contain particles of analyzers are sensitive enough to characterize the ultra-pure RO
varied sizes and different properties (organic/inorganic). As a par- permeate, but TOC is more sensitive in showing the subtle water
ticle’s ability to scatter light is proportional to the sixth power of its quality changes in the permeate line. The highest TOC level was
diameter, the presence of a few large particles in a sample can sig- observed at noon, while the lowest point was at mid-night. The “U”
nificantly lower the concentration threshold, thus detection limit of shape curve of each testing day is believed to follow the daily TOC
DLS is sample-specific. In current study, when the RO feed sample fluctuations of the feed water [26]. The temperature variation could
from the WRD was diluted in distilled water, 12% (volume) of the be another factor as TOC rejection is higher at lower temperature
RO feed sample was required to obtain signal above the background at night.
noise. Future development on lowering the concentration thresh- In addition to the lower limit of detection, the removal rate of
old of DLS would benefit the application of DLS as a RO integrity the targeted water quality parameter by intact RO membranes is
monitoring tool. another important factor that needs to be considered, because the
integrity breach is indicated by the deviation from the baseline. In
3.4. Online monitoring of membrane integrity this context, turbidity was not a viable option for integrity monitor-
ing due to the limited cross membrane changes (e.g., less than 30%
TDS, turbidity and TOC measurements collected from the RO at WRD). The turbidity increase caused by integrity problems could
pilot system are presented in Fig. 5. In the first 24 h of operation, be masked by the normal water quality fluctuations. The removal
 X. Huang et al. / Journal of Water Process Engineering 7 (2015) 161–168 167

Table 3
Overall performance and feasibility analysis of instruments for monitoring of membrane integrity.

Instrument Target Sensitivity Approx. capital cost Maintenance Operating cost On-line option
DLS Nano particles Moderate High (∼$40,000) Low Low Possible
Flow cytometer Viral particles High High (∼$40,000) Moderate Moderate Possible
TOC analyzer Organics High Moderate (∼$22,000) Moderate Moderate Yes
Turbidimeter Water clarity Low Low (∼$2,500) Low Low Yes
Conductivity meter (TDS) Ions Moderate Low (<$1,000) Low Low Yes

4. Feasibility analyses

All instruments tested in the current study were commercially

available. Based on the instrument sensitivity and costs, a feasibil-
ity analysis for each instrument in terms of full-scale installation is
presented in Table 3. FCM is the only method in this study which
can directly quantify virus particles. The log removal rate of virus
particles is the most convincing evidence to demonstrate the mem-
brane integrity and potential virus breakthrough. However, FCM
is also the most expensive method in maintenance and operating
costs. DLS has lower maintenance and operating costs but, cur-
rently, it can only be used as a qualitative membrane integrity
monitoring tool with limited sensitivity. Both FCM and DLS hold
the potential to be modified for online monitoring. The new Nan-
otrac Ultra is already equipped with a sampling probe for automatic
sample collection. Together with the minimal needs for sample
pretreatment, an on-line DSL system can be easily developed for
membrane integrity monitoring. The new generations of traditional
water quality analyzers also yield useful information on membrane
integrity. Among them, TOC analyzer provides the best sensitivity
among the tested online analyzers with a moderate cost. It should
be noted, however, TOC is a bulk measurement of organics, the rela-
tionship between TOC changes and potential virus breakthrough
needs to be further investigated.

5. Conclusions

The results and analyses of this study showed that using more
than one monitoring technique is more practical and reliable to
ensure the sensitive detection of integrity failure of the RO mem-
brane system. A monitoring system that includes TOC analyzer, DLS,
and FCM can provide a feasible way to realize the online and real-
time monitoring of RO membrane integrity. For example, when TOC
deviates from the baseline for a certain range, DLS and FCM can
be employed to determine if there are virus particles leaking into
the RO permeate. However, the integration of these methods as a
monitoring system would require further optimization.

Acknowledgements

This study was partially funded by the WateReuse Research

Foundation (WRF-09-06a). The authors thank David Draper (BD
Sciences) and Geoffrey Dimmer (Portland State University) for
their technical support on FCM. Special thanks to the WateReuse
Fig. 5. Online monitoring of pilot RO membrane integrity using TDS (a), Turbidity Research Foundation’s project manager, Caroline Sherony, the PAC
(b) and TOC (c). members Kevin Alexander, Bob Hultquist, Zia Bukhari, Michelle
Chapman, and the Long Beach Water Department, Dr. Robert Cheng.
Field work for this project was assisted by BKT United Staff, Dr.
Young J. Eum, Allen Chan, Jungwoo Lee, Daniel Kim and WRD staff,
Dr. Paul Fu, Richard Salas and Gary Glasser. Several companies pro-
rate of TOC and TDS (based on EC) are normally much higher (e.g., vided instruments including Microtrac (Neil Black), GE Analytical
95% and 80% at WRD, respectively), and therefore they are more Instruments (Erin England and Gary Evereklian) and Hach Company
sensitive to reflect small integrity breaches. It should be mentioned (Nayan Kapadian). Former and current water division managers
that, in this study, the integrity breaches were only introduced by with Psomas (Steve Winchester and Joe Boyle) provided support for
damaged O-rings, further study may be necessary to determine the the project. Partial support to Xiao Huang from U.S. National Science
instruments’ sensitivity to detect other types of integrity loss (e.g., Foundation Partnerships for International Research and Education
damage to membranes). (OISE-1243543) is acknowledged.
168 X. Huang et al. / Journal of Water Process Engineering 7 (2015) 161–168

Appendix A. Supplementary data [15] M.K. Ramseier, U. von Gunten, P. Freihofer, F. Hammes, Kinetics of membrane
damage to high (HNA) and low (LNA) nucleic acid bacterial clusters in
drinking water by ozone, chlorine, chlorine dioxide, monochloramine,
Supplementary data associated with this article can be found, ferrate(VI), and permanganate, Water Res. 45 (2011) 1490–1500.
in the online version, at http://dx.doi.org/10.1016/j.jwpe.2015.05. [16] D. Marie, C.P.D. Brussaard, R. Thyrhaug, G. Bratbak, D. Vaulot, Enumeration of
001 marine viruses in culture and natural samples by flow cytometry, Appl.
Environ. Microb. 65 (1999) 45–52.
[17] S.K. Brar, M. Verma, Measurement of nanoparticles by light-scattering
References techniques, Trac-trend Anal. Chem. 30 (2011) 4–17.
[18] C.P.D. Brussaard, Optimization of procedures for counting viruses by flow
[1] B. Matthews, L. Prieto, Update to the EPA guidelines for water reuse, Proc. cytometry, Appl. Environ. Microb. 70 (2004) 1506–1513.
Water Environ. Federation 2011 (2011) 2907–2912. [19] M.R. Clokie, Bacteriophages: Methods and Protocols: Isolation,
[2] NRC, Water Reuse Potential for Expanding the Nation’s Water Supply Through Characterization, and Interactions, volume 1, Humana PressInc, 2009, 2015.
Reuse of Municipal Wastewater, National Academies Press, Washington DC, [20] R.T. Noble, J.A. Fuhrman, Use of SYBR Green I for rapid epifluorescence counts
2012. of marine viruses and bacteria, Aquat. Microb. Ecol. 14 (1998)
[3] J.E. Drewes, M. Reinhard, P. Fox, Comparing microfiltration-reverse osmosis 113–118.
and soil-aquifer treatment for indirect potable reuse of water, Water Res. 37 [21] F. Chen, J.R. Lu, B.J. Binder, Y.C. Liu, R.E. Hodson, Application of digital image
(2003) 3612–3621. analysis and flow cytometry to enumerate marine viruses stained with SYBR
[4] L.S. Tam, T.W. Tang, G.N. Lau, K.R. Sharma, G.H. Chen, A pilot study for gold, Appl. Environ. Microb. 67 (2001) 539–545.
wastewater reclamation and reuse with MBR/RO and MF/RO systems, [22] K. Wen, A.C. Ortmann, C.A. Suttle, Accurate estimation of viral abundance by
Desalination 202 (2007) 106–113. epifluorescence microscopy, Appl. Environ. Microb. 70 (2004)
[5] J. Baker, L. Dudley, Biofouling in membrane systems – a review, Desalination 3862–3867.
118 (1998) 81–89. [23] F. Hammes, M. Berney, Y.Y. Wang, M. Vital, O. Koster, T. Egli, Flow-cytometric
[6] S. Allgeier, B. Alspach, J. Vickers, Membrane Filtration Guidance Manual, total bacterial cell counts as a descriptive microbiological parameter for
United States Environmental Protection Agency, US, 2005. drinking water treatment processes, Water Res. 42 (2008) 269–277.
[7] T. Asano, J.A. Cotruvo, Groundwater recharge with reclaimed municipal [24] Y. Tomaru, K. Nagasaki, Flow cytometric detection and enumeration of DNA
wastewater: health and regulatory considerations, Water Res. 38 (2004) and RNA viruses infecting marine eukaryotic microalgae, J. Oceanogr. 63
1941–1951. (2007) 215–221.
[8] S. Toze, Water reuse and health risks – real vs. perceived, Desalination 187 [25] H.O. Huang, T.A. Young, K.J. Schwab, J.G. Jacangelo, Mechanisms of virus
(2006) 41–51. removal from secondary wastewater effluent by low pressure membrane
[9] M. Kumar, S. Adham, J. DeCarolis, Reverse osmosis integrity monitoring, filtration, J. Membrane Sci. 409 (2012) 1–8.
Desalination 214 (2007) 138–149. [26] T. Asano, F. Burton, H. Leverenz, R. Tsuchihashi, G. Tchobanoglous, Water
[10] J.C. Kitis, M. Kitis, J.-H. Kim, B. Mi, B.J. Marinas, Microbial removal and Reuse: Issues, Technologies, and Applications, Metcalf & Eddy, Inc, New York,
integrity monitoring of high-pressure membranes, Am. Water Works Asso. 2007.
Res. Foundation (2003) 4–12. [27] D. Li, H.C. Shi, S.C. Jiang, Concentration of viruses from environmental waters
[11] S. Singh, R.K. Henderson, A. Baker, R.M. Stuetz, S.J. Khan, Characterisation of using nanoalumina fiber filters, J. Microbiol. Meth. 81 (2010) 33–38.
reverse osmosis permeates from municipal recycled water systems using [28] K.Y. Lien, J.L. Lin, C.Y. Liu, H.Y. Lei, G.B. Lee, Purification and enrichment of
fluorescence spectroscopy: implications for integrity monitoring, J. virus samples utilizing magnetic beads on a microfluidic system, Lab Chip 7
Membrane Sci. 421 (2012) 180–189. (2007) 868–875.
[12] M.-L. Pype, D. Patureau, N. Wery, Y. Poussade, W. Gernjak, Monitoring reverse [29] F. Hammes, T. Broger, H.U. Weilenmann, M. Vital, J. Helbing, U. Bosshart, P.
osmosis performance: Conductivity versus fluorescence excitation – emission Huber, R.P. Odermatt, B. Sonnleitner, Development and laboratory-scale
matrix (EEM), J. Membrane Sci. 428 (2013) 205–211. testing of a fully automated online flow cytometer for drinking water
[13] S. Adham, P. Gagliardo, D. Smith, D. Ross, K. Gramith, R. Trussell, Monitoring analysis, Cytom. Part A 81A (2012) 508–516.
the integrity of reverse osmosis membranes, Desalination 119 (1998) [30] USEPA, Low – Pressure Membrane Filtration for Pathogen Removal:
143–150. Application, Implementation and Regulatory Issues, Cincinatti, OH, 2001.
[14] Y. Wang, F. Hammes, K.D. Roy, W. Verstraete, N. Boon, Past, present and [31] P.E. Plantz, Nanotrac Ultra: Measurement of Low Concentration, Very Small
future applications of flow cytometry in aquatic microbiology, Trends Sized Particulate Systems, Application Note SL-AN-18 Rev A, Microtrac, Inc.,
Biotechnol. 28 (2010) 416–424. 2015.

View publication stats