Professional Documents
Culture Documents
Accepted Manuscript: Bioresource Technology
Accepted Manuscript: Bioresource Technology
Accepted Manuscript: Bioresource Technology
PII: S0960-8524(16)31391-8
DOI: http://dx.doi.org/10.1016/j.biortech.2016.09.124
Reference: BITE 17147
Please cite this article as: Praveen, P., Yun Ping Heng, J., Loh, K-C., Tertiary Wastewater Treatment in Membrane
Photobioreactor using Microalgae: Comparison of Forward Osmosis & Microfiltration, Bioresource Technology
(2016), doi: http://dx.doi.org/10.1016/j.biortech.2016.09.124
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Tertiary Wastewater Treatment in Membrane
Photobioreactor using Microalgae: Comparison of
Forward Osmosis & Microfiltration
*Corresponding author:
Associate Professor, Department of Chemical and Biomolecular Engineering,
National University of Singapore, 4 Engineering Drive 4, Singapore 117585.
Email: chelohkc@nus.edu.sg; Tel.: +65 6516 2174; Fax: +65 6779 1936
1
Abstract
Discharge of wastewater with high nitrogen and phosphorus content is a major cause of
with Chlorella vulgaris for continuous tertiary wastewater treatment. Both the bioreactors
exhibited good biomass accumulation (over 2 g/L), although the OMPBR achieved better
nutrients removal due to high rejection properties of the membranes. At 2 days HRT, the
OMPBR achieved nitrogen and phosphorus removal efficiencies of 86-99% and 100%,
respectively, whereas the corresponding values in the MPBR were 48-97% and 46%,
respectively. Based on the energy input, the total operating costs for OMPBR were 32-45 %
higher than that of the MPBR, and filtration cost for OMPBR was 3.5-4.5 folds higher than
that of the MPBR. These results indicate that the integration of membrane filtration with
Wastewater treatment
2
1 Introduction
membrane separation. These bioreactors offer the advantages of small footprint, high biomass
retention, large loading rates, process flexibility and enhanced effluent quality (Judd, 2008).
MBRs are typically associated with the activated sludge process, and these have commercial
uses in the treatment of both municipal and industrial wastewater (Huang & Lee, 2015).
Recently, MBRs have also found application in nitrogen (N) and phosphorus (P) removal
from tertiary wastewater using autotrophic microalgae (Choi, 2015; Marbelia et al., 2014).
such as microfiltration (MF) or ultrafiltration (UF). In the past decade however, filtration
based on forward osmosis (FO) has also been investigated as an economical and sustainable
alternative to conventional filtration (Huang & Lee, 2015). FO is akin to natural osmosis,
wherein a draw solution (DS) with high osmotic pressure is used to drive water transport
bioreactors (OMBR) have the advantages of high solute rejection, low energy costs and
reduced membrane fouling, which have been amply demonstrated in biological wastewater
treatment, as well as in nutrients recovery from wastewater (Achilli et al., 2009; Praveen &
Due to the absence of hydraulic pressure, FO is deemed to be more economical than the
corresponding conventional filtration processes (McGinnis & Elimelech, 2007; Zhao et al.,
2012). However, some recent studies have challenged this notion by demonstrating that FO
can be highly energy intensive if the high cost of DS recycle is added to the cost of filtration
(McGovern & Lienhard, 2014; Semiat, 2008). Furthermore, while the effects of external
3
investigated in detail (Luo et al., 2015; Qiu & Ting, 2013), the influence of these factors on
the energy input to the OMBR has not been examined. ECP mitigation requires increase in
DS flow rate, whereas salt accumulation demands high rate of DS supply, and both of these
factors may increase the operating costs (Praveen et al., 2015). As a result, there is ambiguity
on the claims pertaining to the energy efficiency of FO-based processes, and OMBRs may
not be more economical to operate than conventional MBRs. In order to better understand the
detailed and systematic comparison of these bioreactors is needed. In this research, we have
Tertiary wastewater with high N and P concentrations is a common cause for eutrophication
in natural water bodies (Sulzacova et al., 2015). Although many physical, chemical and
biological methods are being used in tertiary wastewater treatment, the use of microalgae is
advantageous for many reasons (Beuckels et al., 2015; Fernandez et al., 2012): high growth
rates, high N and P uptake rates, generation of microalgal biomass, and greenhouse gas
conventional high rate algae ponds which exhibit low efficiency due to poor light penetration,
low biomass concentration and high bioreactor footprint (Wang et al., 2012). Such high-
retention time (HRT), without risk of biomass washout (Marbelia et al., 2014).
The objective of this research was to design and operate a membrane photobioreactor
operated under identical conditions. The energy input to the MPBR and OMPBR were
4
estimated and the bioreactors were compared based on biomass accumulation, nutrients
removal kinetics, efficiency and energy demand. Chlorella vulgaris was used as the model
organism due to its high growth rate, large N and P demand and excellent tolerance to
All the chemicals used in this research were of analytical grade and purchased either from
Chlorella vulgaris ATCC 13482 used throughout this study was cultivated in Bold’s Basal
Medium (BBM) supplemented with 5% CO2 enriched air at 0.2 gas volumes per reactor
volume per minute (VVM), and provided with 2000 lux light intensity. All media, pipette
tips, and Erlenmeyer flasks fitted with cotton plugs were autoclaved before use.
2.2 OMPBR
2.2.1 Experimental Setup
Fig. 1 shows a schematic of the OMPBR setup. The bioreactor tank (20 cm length x 12.5 cm
module was prepared using commercial thin film composite (TFC) FO membranes (HTI,
USA) and the module was immersed in the bioreactor tank for osmotic filtration. Two pieces
effective filtration area of 0.036 m2. The membrane module was designed in such a way that
the active layer faced the wastewater and the support layer faced the DS. A continuous stream
of humidified 5% CO2-enriched air was sparged in the OMPBR at a rate of 0.4 VVM to
provide inorganic carbon to the microalgae. The OMPBR was illuminated from all four sides
and the top using fluorescent lights of 1500-2000 lux intensity. A 2 L reservoir designed with
5
an overflow outlet at 1.5 L provided the DS. The beaker was stirred on a magnetic stirrer and
the DS was recirculated in the membrane module using a peristaltic pump (Masterflex, USA).
DS concentration in the reservoir was maintained constant using a feedback control system
compensated through pumping of concentrated DS stock solution into the reservoir. Overflow
from the DS reservoir was collected as OMPBR effluent. The wastewater feed tank was
For MPBR operation, the setup in Fig. 1 was modified to remove the DS reservoir and the
feedback control. The FO membranes in the membrane module were replaced with PVDF
MF membranes (Newton & Stokes, Singapore), and the effluent was filtered through the
2.2.2 Operation
Synthetic wastewater was used for the bioreactor operation. The wastewater contained: 4
mg/L NH4+-N, 0.35 mg/L NO3--N and 1.8 mg/L PO43--P (Arbib et al., 2014). Other
prevent nutrient limitation on microalgae growth. The bioreactors were operated at HRTs
varying from 1-2 days with complete biomass retention. Only 10 mL liquid was removed
daily to monitor suspended biomass and nutrients concentrations. The bioreactor pH was
regularly monitored and maintained within 7-7.5. Although microalgae growth in tertiary
wastewater can be affected by medium composition, pH, temperature and light exposure (Shi
et al., 2014), these factors would not have elicited different response in the MPBR and the
OMPBR. Consequently, only the effects of HRTs on the bioreactor performance were
studied.
6
2.3 Analytical Methods
Microalgal biomass density was determined by measuring the optical density (OD) of the
Japan). The OD was used to compute the biomass concentration by the formula: dry cell
weight (mg/L) = 542*OD540 (Praveen & Loh, 2015). The concentrations of NH4 +-N, NO3--N
and PO43--P were measured using phenate method, spectrophotometric method and ascorbic
acid method of standard methods handbook, respectively (APHA et al., 2012). The energy
usage for the equipment was monitored using an Ecoplug energy meter (Solargy, Singapore).
The overall removal efficiency and the average removal rate were calculated based on the
concentration difference between the wastewater and the effluent at steady state as: Removal
wastewater - Conc. in effluent)/HRT. Since the MPBR and OMPBR were operated in
continuous mode, the operation was not repeated for reproducibility. Instead, the
uncertainties associated with the measurement of steady state values were estimated and
Fig. 2 shows the changes in nutrients concentrations over 43 days of MPBR operation. At 2
days HRT, the concentrations of NH4+-N, NO3--N and PO43--P decreased rapidly due to
assimilation by the microalgae, and stabilized after 5-8 days at 0.11 mg/L, 0.19 mg/L and
0.98 mg/L, respectively. The corresponding removal efficiencies were 97%, 46% and 51%,
whereas the average removal rates were 1.94 mg/L-day, 0.08 mg/L-day and 0.41 mg/L-day,
respectively. The removal efficiency of NH4 +-N was higher as compared to the other
7
nutrients, which indicated that it could be the limiting nutrient (Praveen & Loh, 2016a). On
the other hand, relatively low removal efficiency for NO3--N could be attributed to the
favorable uptake of NH4+-N over NO3--N by the microalgae (Di Termini et al., 2011). Effects
of increased nutrients loading rate was next investigated through a decrease in the HRT.
When the HRT was decreased to 1.5 days, the effluent NH4+-N concentration increased to a
maximum of 0.45 mg/L on day 18, before decreasing and stabilizing at 0.24 mg/L. The
overall removal efficiency was 94%, at a removal rate of 2.51 mg/L-day. In comparison, the
change in HRT affected NO3--N removal adversely and its concentration in the effluent
increased monotonically and stabilized at 0.25 mg/L, with a poor removal efficiency of 29%
and a removal rate of 0.06 mg/L-day. Similar changes were observed for PO43--P, for which
the effluent concentration increased to 1.11 mg/L, corresponding to removal efficiency and
removal rate of 38% and 0.46 mg/L-day, respectively. While the changes in NO3--N could be
attributed to a relatively higher accumulation of NH4 +-N in the MPBR, the decrease in PO43--
P removal indicated that the N/P ratio in the MPBR was not ideal (Di Termini et al., 2011).
Despite being the limiting substrate and under operating conditions characterized by a
relatively poor N/P ratio, NH4+-N was not completely exhausted in the MPBR. The MPBR
performance was not affected by nutrients concentrations alone, but there could have been
other factors influencing nutrients uptake by the microalgae. Since the microalgae
concentration was high at this stage, it was possible that light intensity could be a limitation
Similar trends in nutrients removal were obtained in the third and the last stage of MPBR
operation at an HRT of 1 day. After a temporary spike, NH4 +-N concentration stabilized at
0.65 mg/L, with a removal efficiency of 84%, and a removal rate of 3.35 mg/L-day. The
concentration of NO3--N in the MPBR effluent increased to 0.3 mg/L, with a removal
8
efficiency and removal rate of 14% and 0.05 mg/L-day, respectively. The changes in PO43--P
followed a similar trend and PO43--P concentration increased to 1.29 mg/L, which translated
to a low removal efficiency of 28% and removal rate of 0.51 mg/L-day. Although the
increase in nutrients loading rates in the MPBR enhanced the removal rates of NH4 +-N and
PO43--P, the increase was not proportional to the loading rates. These trends are typical during
continuous bioreactor operations, as the contact time between the microorganisms and the
nutrients was lowered at high loading rates. Moreover, several parameters such as light
intensity, microalgae activity and N/P ratio might change in the MPBR with changes in HRT,
During MPBR operation, even though the removal efficiencies for the nutrients decreased at
lower HRTs, the removal rates for NH4+-N and PO43--P remained high at the high loading
rates. These could be attributed to the higher nutrients availability and increasing biomass
concentration in the bioreactor. For NO3--N, on the other hand, both removal efficiency and
removal rates decreased with decreasing HRTs. This discrepancy in NO3--N removal
augments the hypothesis that the microalgae exhibited poor NO3--N uptake in the presence of
NH4 +-N. These results are consistent with other MPBRs where NO3--N removal was
significantly lower than that of NH4 +-N (Singh & Thomas, 2012).
Fig. 3 shows the changes in nutrients concentrations over 51 days of OMPBR operation. Due
to the low pore size and high rejection properties of the FO membranes, two different
concentrations profiles developed, on either side of the membranes, for each of the nutrients.
At 2 days HRT, NH4+-N concentration in the OMPBR (Fig. 3a) decreased gradually from 4
mg/L and stabilized at 0.15 mg/L after 12 days. NH4+-N concentration in the effluent
increased to reach a maximum of 0.82 mg/L on day 2, which then decreased gradually to
9
stabilize at 0.03 mg/L after 12 days. The average removal rate and removal efficiency were
concentration in the OMPBR (Fig. 3b) remained stable at about 0.3 mg/L during the first 10
days of operation, whereas its concentration in effluent stabilized at 0.04 mg/L after 12 days.
The average removal rate and removal efficiency were computed as 0.15 mg/L-day and 87%,
respectively. PO43--P removal in the OMPBR was nearly complete, and its concentration in
the effluent remained below 0.02 mg/L throughout the operation, resulting in the average
removal rate and the removal efficiency of 0.9 mg/L-day and close to 100%, respectively.
PO43--P concentration in the OMPBR (Fig. 3c) increased slowly to reach 2 mg/L after 12
days, but it gradually decreased thereafter to 1.54 mg/L at the end of the first stage of
OMBR in the first few days of operation is typical (Achilli et al., 2009; Praveen & Loh,
2016b). This accumulation is attributed to the low pore size of the active layer of the FO
membranes, which imparts the membranes high solute rejection properties. The accumulation
trend is reversed after a few weeks of operation, when there is large biomass accumulation in
the bioreactor. However, no such accumulation has been observed during the OMPBR
operation, even though the concentration profiles developed on either side of the FO
membranes were widely different and nutrients rejection by the membranes was excellent.
While NO3--N and PO43--P concentrations in the OMPBR remained flat during the first few
days of operation, there was no accumulation of NH4 +-N and its concentration in the OMPBR
decreased monotonically until it matched the concentration in the effluent. Two conclusions
can be drawn from these results: (1) the loading rates of the nutrients at 2 days HRT was not
very high, and these could be matched by the nutrients uptake rate exhibited by the
microalgae, and; (2) microalgal uptake rate was highest for NH4+-N, which indicated that
10
NH4 +-N was the limiting nutrient for microalgae growth and metabolism (Gao et al., 2015).
While these results are consistent with the findings obtained during the MPBR operation,
these also highlight the role of membranes in nutrients removal in the OMPBR, especially in
the removal of PO43--P. Since the removal efficiency for PO43--P was the highest even though
it was not the limiting nutrient and the wastewater had low N/P ratio, the removal of PO43--P
was mainly through membrane rejection. Such high rejection efficiency of PO43--P has been
reported in literature, and it has also been used to design novel OMBR configurations for
When nutrients concentrations in the effluent had stabilized after 16 days of OMPBR
operation at 2 days HRT, the HRT was reduced to 1.5 days and the corresponding nutrients
loading rates was increased. Under the new operating conditions, there was an immediate
maximum of 0.94 mg/L on day 20, whereas the concentration in the effluent increased to a
maximum of 0.19 mg/L in the same period. However, these spikes were temporary and
NH4 +-N levels in the OMPBR and in the effluent soon decreased, and stabilized at 0.44 mg/L
and 0.07 mg/L, respectively, on day 34. The removal efficiency and average removal rate
were determined as 98% and 2.62 mg/L-day, respectively. Contrary to NH4+-N, NO3--N
concentrations on both sides of the FO membranes continued to rise and stabilized at higher
levels on days 27-28. On day 34, NO3--N concentrations in the OMPBR and in the effluent
were 0.62 mg/L and 0.17 mg/L, respectively, with a removal efficiency of 63% and a
removal rate of 0.15 mg/L-h. PO43--P concentration in the OMPBR was also affected by the
3.7 mg/L, which was nearly double the concentration in the feed wastewater. On the other
hand, PO43--P concentration in the effluent remained unchanged throughout the operating
11
period. The consequent removal efficiency was close to 100% and the average removal rate
During the second stage of OMPBR operation (1.5 days HRT), the nutrients loading rates
were 25% higher and a spike in nutrients concentration in both the OMPBR and effluent
streams were expected. The increase in concentrations were expected to be temporary though,
as the increase in biomass concentration and the consequent increase in the nutrients uptake
rate could reverse the accumulation trends. While such a trend was observed in NH4+-N
throughout this operating period. The effluent PO43--P concentration remained undetectable
despite PO43--P accumulation in the OMPBR because of FO-based filtration rather than
biological assimilation. It was also observed that the removal rates for NH4 +-N and PO43--P
increased at higher loading rates, but there was no appreciable changes in the removal rate of
NO3--N and it remained at 0.15 mg/L-day. This again indicated the preferential uptake of
NH4 +-N over NO3--N by the microalgae, as was observed during MPBR operation.
Having achieved stable concentration profiles, nutrients loading rate to the OMPBR was
increased further on day 35 by adjusting the HRT to 1 day. Under this operating condition,
the concentrations of the nutrients adjusted quickly in the OMPBR. NH4+-N concentration in
the bioreactor increased to a maximum of 4.5 mg/L and then decreased to stabilize at 3.5
mg/L. On the other hand, NO3--N and PO43--P concentrations in the bioreactor increased
monotonically, and stabilized at 1.41 mg/L and 8.36 mg/L, respectively. After 51 days of
operation, NH4+-N level in the effluent had stabilized to 0.31 mg/L, with a removal efficiency
of 92% and a removal rate of 3.7 mg/L-day, whereas NO3--N concentration in the effluent
was 0.24 mg/L with corresponding removal efficiency and rate of 31% and 0.11 mg/L-day,
respectively. On the contrary, despite the large increase in the loading rate and high PO43--P
12
accumulation in the OMPBR, PO43--P concentration in the effluent remained below the
detection limits, again resulting in close to 100% removal, whereas the removal rate
At 1 day HRT, some deterioration in OMPBR performance was observed. Although the
NH4 +-N and PO43--P removal were still excellent, there was substantial increase in their
accumulation in the OMPBR, notwithstanding the presence of a large and growing amount of
biomass in the bioreactor. Even NH4+-N, which was the limiting nutrient, exhibited
temporary accumulation in the bioreactor. Therefore, it was possible that there were changes
in the metabolic activity of the microorganisms during OMPBR operation. Nevertheless, the
effluent concentrations for the nutrients were below the discharge limit under all the
operating conditions (Boelee et al., 2012; Ji et al., 2013). Moreover, effluent quality in the
OMPBR was better than that observed in the MPBR, which demonstrated the advantages of
Changes in the activity and diversity of the microorganisms with time are common
phenomena in OMBRs during municipal wastewater treatment using activated sludge. These
changes have been attributed to the gradual increase in the salinity of the OMBR, due to salt
(Huang & Lee, 2015; Qiu & Ting, 2013). Fig. S1 shows salinity profiles of the OMPBR over
time. It can be seen that salinity increased gradually and monotonically throughout the
operating period, and the conductivity of the OMPBR was 30 mS after 51 days of operation.
In contrast, the optimal growth medium (BBM) for microalgae had a conductivity of only
about 3 mS. Since freshwater microalgae, such as C. vulgaris, do not have high salt tolerance
(Shen et al., 2015), it was likely that the lower microalgae activity (hence relatively low
nutrients removal) during the last stage of OMPBR operation (1 day HRT) was a
13
consequence of the high salt levels in the bioreactor. It can also be seen from Fig. S1, that the
contribution of accumulated nutrients in the OMPBR to the overall salinity was insignificant.
It can therefore be concluded that the change in the salinity of the OMPBR was driven mainly
Salt accumulation is considered one of the biggest challenges to the sustainability of OMBRs.
Not only is the presence of high salt concentrations in the bioreactors harmful for biomass
growth and metabolism, salinity increase in the bioreactor also necessitates an increase in the
DS concentration to maintain the HRT (Huang & Lee, 2015). This would enhance the risk of
well as increase operating costs since it would be necessary to increase the DS strength to
maintain the constant HRT. Literature has indicated that the best strategy to alleviate this
challenge would be to improve membrane design for better salt rejection (Phillip et al., 2010),
Fig. 4 shows the changes in the suspended microalgal biomass concentration in the MPBR
and the OMPBR. In both bioreactors, biomass concentration profiles followed similar trends.
During the first 14-17 days of operation, the biomass concentration in suspension increased
monotonically and reached a maximum of 228 mg/L in the MPBR and 278 mg/L in the
OMPBR. The increase in biomass concentration was characterized by the dark green color of
the microalgal suspension. Although further increase in biomass concentration was expected
in the bioreactors, it did not happen. On the contrary, there was a gradual decline in the
suspended biomass concentration in the bioreactors and the intensity of the greenish
14
in both the bioreactors had stabilized at 10-20 mg/L, and the microalgae suspension looked
almost clear.
The reason for this change in the color of the cell suspension was mainly due to the gradual
sedimentation of the microalgal biomass to the bottom of the bioreactor tank due to the lack
of any mechanical mixing. Although attempts were made to improve mixing at the bottom of
the tank using a magnetic stirrer, the rectangular design of the bioreactor tank prevented
efficient mixing and the biomass settled mainly in the un-mixed areas. In addition, C.
vulgaris also exhibited a high tendency to attach to the acrylic surfaces of the bioreactors and
a visible amount of attached biomass was observed on the surfaces after 3 weeks of operation
in both bioreactor setups. Furthermore, the microalgae also exhibited a high tendency to self-
aggregate, which accelerated the rate of sedimentation. The biomass aggregation and
attachment were most visible when microalgae concentration had reached a relatively high
concentration in the bioreactors. The process of biomass attachment and sedimentation were
independent of the nutrients loading rates. It was also apparent that most of the biomass
produced during the later stages of operation remained in attached or aggregated form, and
played a major role in nutrients removal. Since the biomass attached to the walls of the
bioreactor tank were better illuminated as compared to those at the bottom surface, it was
likely that the contribution of the attached biomass in the bioremediation was much higher
Since the suspended microalgae concentration was representative of the actual biomass
concentration in the MPBR/OMPBR only in the first 3 weeks of operation, the biomass from
the MPBR and OMPBR were collected at the end of the operation to estimate the actual
amount of biomass in the bioreactors. The collected biomass included the settled and
suspended biomass, as well as the biomass attached to the surfaces, although not all the
attached biomass could be completely recovered. The measurements were performed using
15
both absorbance, as well as dry cell weight estimation. It was found that the total biomass
accumulated in the OMPBR was well over 2 g/L. These results indicate the potential of the
with other studies reporting microalgae cultivation in MBRs (Choi, 2015; Gao et al., 2015;
The high rate of biomass attachment to the surfaces and the high tendency to aggregate can
substances (EPS) by microalgae in the bioreactor, or changes in the zeta potential of the
microalgae (Gerde et al., 2014; Wang et al., 2014). The zeta potential of the microalgae had
decreased by nearly four times during the operating period, whereas there was significant
Fig. 5 summarizes the removal efficiencies of the nutrients in the MPBR and the OMPBR at
different HRTs. It is obvious that the removal of all the three nutrients was higher in the
OMPBR. While the removal efficiencies for NH4+-N and NO3--N in the two bioreactors
differed by 2-8% and 17-41%, respectively, the difference was 49-72% for PO43--P, under
different operating conditions. Some extent of performance deterioration was observed for
both the bioreactors when these were operated at lower HRTs, but the OMPBR effluent
The differences between the removal efficiencies of the MPBR and the OMPBR were also
reflected in the corresponding removal rates, as shown in Fig. 6. Once again, the removal
rates in the OMPBR were better for all the three nutrients under all the different operating
conditions. While the difference in the average removal rates between the OMPBR and the
MPBR for NH4+-N was only about 10%, for NO3--N, it was 46-60% and 54-72% for PO43--P.
16
The removal rates, in general, increased at lower HRTs, and PO43--P removal rate in the
OMPBR increased proportionately with the loading rates under different HRTs, resulting in
significant differences between the performances of the two bioreactors. It was also observed
that unlike NH4 +-N and PO43--P, NO3--N removal rates in both the bioreactors decreased
when the HRTs were lowered. This anomaly in the removal trend could again be due to the
increased accumulation of NH4+-N in the OMPBR, which was the preferred N-source for the
microalgae.
Since MPBR and OMPBR operations were carried out under identical conditions, and using
the removal efficiencies was not due to microbial metabolism; the difference between the
performances of the MPBR and OMPBR were due to the differences in the filtration
Table 1 shows comparison of the performance of the MPBR and the OMPBR with
and nutrients removal trends and the removal efficiencies in the MPBR were consistent with
the other studies. On the other hand, the nutrients removal kinetics in the OMPBR was
different from those reported in literature, and the removal efficiency was comparatively
better. A relatively higher biomass accumulation was observed in this study, which could be
attributed to a lower HRT and complete biomass retention maintained during the operation.
Fig. S2 (supplementary data) shows the changes in the permeate flux and the HRT during the
MPBR and the OMPBR operation. In the MPBR, the flux remained relatively constant during
the 43 days of operation without adjustment in the hydraulic pressure. Although SEM images
17
the membranes (images not shown), the biofouling potential of the microalgae was typically
lower than that of bacteria and activated sludge (Bilad et al., 2014). Besides, the permeate
flux of 3.18-6.37 LMH was relatively low for MF, and the extent of biofouling might not
have been significant to affect the relatively low flux. Compared to MPBR, the variation in
the permeate flux was higher in the OMPBR, which could have been caused by: biofouling,
salt accumulation and concentration polarization (Zhao et al., 2012). Since microbial growth
and other characteristics in the MPBR and OMPBR were similar, and the bioreactors were
operated under identical operating conditions, it was unlikely that the membrane biofouling
would have been more severe in the OMPBR, especially in the absence of any hydraulic
pressure in FO-based filtration. On the other hand, the increase in the salinity of the OMPBR
(Fig. 5) resulted in changes in the net osmotic pressure gradient across the FO membranes,
which could have resulted in a decrease in the permeate flux. Although the strength of the DS
was increased over time to neutralize the effects of the increased salinity in the OMPBR,
these changes were not instantaneous and might be responsible for the variations.
Furthermore, during long-term operation of the FO-based processes, the accumulation of salts
inside the membranes, also known as internal concentration polarization (ICP) might also
decrease the effective osmotic pressure gradient and lower the permeate flux (Praveen et al.,
2015). This could be further aggravated when high strength saline solutions were present on
both sides of the membranes, as observed during the last stage of OMPBR operation at 1 day
HRT.
It has been reported that one of the advantages of FO is the low energy requirement of the
process, as compared to other pressure driven filtration processes such as MF (McGinnis &
Elimelech, 2007). However, based on the laboratory-scale bioreactors setup used in this
study, it is obvious that the OMPBR setup was more complex and needed more equipment as
18
compared to that of the MPBR. Although the permeate flux was independent of the DS flow
rate in the OMPBR, the DS had to be circulated above a minimum flow rate (>50 rpm) to
prevent the adverse effects of ECP on the flux. Moreover, the increase in the salinity of the
OMPBR during the operation required gradual increase in DS strength, which would
contribute to an increase in the operating costs. On the contrary, the setup for the MPBR was
relatively simple, and the operating conditions were very straightforward. It was also
observed that there was no significant increase in the transmembrane pressure (TMP) during
the operation.
In order to investigate and compare the energy efficiency of the MPBR and OMPBR further,
the energy input to the two bioreactors were measured using an energy meter and the results
are summarized in Table 2. It can be seen that the energy input to the OMPBR was higher
due to the costs of operating the extra pump, magnetic stirrer and the feedback control
system. It should be noted that the cost of gas sparging was not taken into account (since gas
The fundamental difference between the MPBR and OMPBR was in the way the effluent was
drawn from the bioreactor tanks. During OMPBR operation, the peristaltic pump used to
recirculate DS in the membrane module was operated above 50 rpm, using 3.1 mm internal
diameter (ID) silicone tubing, to prevent the occurrence of ECP (data not shown). The pump
speed remained constant throughout the operating period as the DS flow rates did not
influence the HRT. However, the OMPBR setup also included a second peristaltic pump
which was used to add concentrated DS stock solution in the DS reservoir to maintain a
constant DS concentration in the reservoir. The second peristaltic pump was activated every
time the salinity of the DS reservoir dropped below the set value. The duration for which the
pump was operated depended on the HRT, bioreactor salinity and the DS concentration. For
19
example, the pump was operated for longer duration at lower HRT when the DS was diluted
On the contrary, the MPBR setup was simple with a single peristaltic pump used to draw
water directly through the MF membranes. Since ECP was not an issue, the pump speed was
regulated based only on the HRT. It was also observed that the MF membrane module could
draw a large volume of water even at a very low pump speed. For example, even at the lowest
speed of 3 rpm (using 3.1 mm ID tubing) and negligible TMP (pressure could not be detected
using pressure gauge), the water drawn by the peristaltic pump was > 12 mL/min, whereas
the actual requirement at all times was < 4 mL/min. Consequently, the peristaltic pump in the
MPBR setup had to be operated with smaller tubing of 0.8 mm ID to maintain the flux in the
required range. In order to make a fair comparison, the operating cost for the MPBR was
calculated assuming the pump speed 3 rpm and tubing size of 3.1 mm ID. Since water drawn
through MF in the MPBR at 3 rpm (using 3.1 mm ID tubing) was higher than that required in
the actual operation, the operating cost calculated for MPBR was higher than the actual cost
of operation.
Based on Fig. S3 (supplementary data) and Table 2, the cost of operating the MBR was < $
0.22 day-1 throughout the operating period under different HRTs. On the other hand, the costs
for OMPBR varied between s$ 0.29-0.32 day-1 depending on the HRTs. Consequently, the
cost of OMPBR operation at 2 days, 1.5 days and 1 day HRTs were 32%, 36% and 45%
higher than those for the MPBR, respectively. It was also observed that the cost of bioreactor
illumination was the highest among all, and it was more than the cost of operating all the
other equipment taken together. It was also observed that the cost of filtration in OMPBR was
3.5-4.6 folds higher than that in the MPBR. While the cost of DS recirculation was the
highest contributor to the filtration cost, the feedback control system too was energy
20
intensive, especially during the last stage of operation, when the HRT was lower and the
While these results are not in agreement with most of the early literature on FO-based
filtration, where FO has been described as more economical than conventional filtration
process (McGinnis & Elimelech, 2007; Zhao et al., 2012), these are consistent with recent
studies highlighting the high operating costs of FO-based processes (McGovern & Lienhard,
2014; Semiat et al., 2010). Some of these studies have demonstrated that the cost of recycling
and concentrating the DS from the diluted effluent is very high, and FO cannot energetically
outperform traditional filtration processes, including RO. This implies that the cost of
OMBRs with DS recycle will never be lower than the cost of conventional MBR. In this
research however, the costs of the DS and the costs of DS recycle were not considered in the
energy calculations. Despite that, the total operating cost of the OMPBR was higher than that
of the MPBR.
The large difference between the operating costs of the MPBR and OMPBR could be a result
of operating the bioreactor at a relatively low flux (< 4 mL/min). Under these operating
conditions, the TMP in the MPBR was very low particularly since the effective membrane
area was relatively high at 360 cm2. On the other hand, DS recirculation in the OMPBR was
independent of the permeate flux. Therefore, if the bioreactors had been operated at a lower
HRT of 6-8 h, it is possible that the operating costs of the MPBR could have risen faster than
that of the OMPBR. It is also possible that the OMPBR could have been more effective, if the
bioreactor operation had been longer and there was significant membrane biofouling, as the
cost of traditional MBR operation under fouling conditions is typically higher than that of the
OMBRs (Luo et al., 2015). The cost of OMPBR operation can also be lowered significantly if
seawater or wasted brine solution from desalination plants were to be used, which are readily
available and do not need any recycling (Van der Bruggen & Luis, 2015). In such a scenario,
21
continuous supply of the DS will obviate the need for DS recirculation, resulting in additional
cost savings. Nevertheless, based on this study, using the laboratory-scale bioreactors setup,
the operating costs of the OMPBR was significantly higher than that of the MPBR. Besides,
the OMPBR also resulted in various operating challenges, especially due to salt
operation.
4 Conclusions
The integration of membrane filtration with photobioreactors in both MPBR and OMPBR has
bioreactors allow high biomass retention independent of the HRT, and yield high removal
rates and efficiency. However, there are several challenges pertaining to filtration: biofouling,
concentration polarization and salt accumulation, which must be addressed to make this
technology more economical and sustainable. Since bioreactor illumination constituted the
dominated component of the operating costs, future studies should harness solar irradiation to
Acknowledgement
This research was funded by the Singapore National Research Foundation under its
Competitive Research Program for the project entitled, “Advanced FO Membranes and
Membrane Systems for Wastewater Treatment, Water Reuse and Seawater Desalination”
22
References
Achilli, A., Cath, T.Y., Marchand, E.A., Childress, A.E. 2009. The forward osmosis
239(1-3), 10-21.
APHA, AWWA, WEF. 2012. Standard Methods for the Examination of Water and
Arbib, Z., Ruiz, J., Álvarez-Díaz, P., Garrido-Pérez, C., Perales, J.A. 2014. Capability of
treatment, CO2 bio-fixation and low cost biofuels production. Water Res., 49(0), 465-
474.
Beuckels, A., Smolders, E., Muylaert, K. 2015. Nitrogen availability influences phosphorus
Bilad, M.R., Arafat, H.A., Vankelecom, I.F.J. 2014. Membrane technology in microalgae
Boelee, N.C., Temmink, H., Janssen, M., Buisman, C.J.N., Wijffels, R.H. 2012. Scenario
23
Boonchai, R., Seo, G. 2015. Microalgae membrane photobioreactor for further removal of
nitrogen and phosphorus from secondary sewage effluent. Korean J. Chem. Eng.,
32(10), 2047-2052.
Chekli, L., Phuntsho, S., Kim, J.E., Kim, J., Choi, J.Y., Choi, J.S., Kim, S., Kim, J.H., Hong,
S., Sohn, J., Shon, H.K. 2016. A comprehensive review of hybrid forward osmosis
systems: Performance, applications and future prospects. J. Membr. Sci., 497, 430-
449.
2195-2205.
Di Termini, I., Prassone, A., Cattaneo, C., Rovatti, M. 2011. On the nitrogen and phosphorus
Fernandez, F.G.A., Gonzalez-Lopez, C.V., Sevilla, J.M.F., Grima, E.M. 2012. Conversion of
Gao, F., Li, C., Yang, Z.-H., Zeng, G.-M., Feng, L.-J., Liu, J.-z., Liu, M., Cai, H.-w. 2016.
photobioreactor for biomass production and nutrients removal. Ecol. Eng., 92, 55-61.
Gao, F., Yang, Z.H., Li, C., Zeng, G.M., Ma, D.H., Zhou, L. 2015. A novel algal biofilm
Gerde, J.A., Yao, L.X., Lio, J.Y., Wen, Z.Y., Wang, T. 2014. Microalgae flocculation:
Impact of flocculant type, algae species and cell concentration. Algal Res., 3, 30-35.
24
Honda, R., Boonnorat, J., Chiemchaisri, C., Chiemchaisri, W., Yamamoto, K. 2012. Carbon
Huang, L.Y., Lee, D.J. 2015. Membrane bioreactor: A mini review on recent R&D works.
Ji, M.K., Abou-Shanab, R.A.I., Kim, S.H., Salama, E., Lee, S.H., Kabra, A.N., Lee, Y.S.,
Hong, S., Jeon, B.H. 2013. Cultivation of microalgae species in tertiary municipal
wastewater supplemented with CO2 for nutrient removal and biomass production.
Judd, S. 2008. The status of membrane bioreactor technology. Trends Biotechnol., 26(2),
109-116.
Luo, W.H., Hai, F.I., Kang, J.G., Price, W.E., Nghiem, L.D., Elimelech, M. 2015. The role of
Marbelia, L., Bilad, M.R., Passaris, I., Discart, V., Vandamme, D., Beuckels, A., Muylaert,
McGovern, R.K., Lienhard, J.H. 2014. On the potential of forward osmosis to energetically
Phillip, W.A., Yong, J.S., Elimelech, M. 2010. Reverse Draw Solute Permeation in Forward
25
Praveen, P., Loh, K.-C. 2016a. Nitrogen and phosphorus removal from tertiary wastewater in
Praveen, P., Loh, K.-C. 2016b. Osmotic membrane bioreactor for phenol biodegradation
Praveen, P., Loh, K.-C. 2015. Photosynthetic aeration in biological wastewater treatment
99(23), 10345-10354.
Praveen, P., Nguyen, D.T.T., Loh, K.-C. 2015. Biodegradation of phenol from saline
Qiu, G., Law, Y.M., Das, S., Ting, Y.P. 2015. Direct and complete phosphorus recovery from
bioreactor process with seawater brine as draw solution. Environ. Sci. Technol.,
49(10), 6156-6163.
Qiu, G., Ting, Y.-P. 2013. Osmotic membrane bioreactor for wastewater treatment and the
Ruiz-Marin, A., Mendoza-Espinosa, L.G., Stephenson, T. 2010. Growth and nutrient removal
in free and immobilized green algae in batch and semi-continuous cultures treating
Semiat, R. 2008. Energy Issues in Desalination Processes. Environ. Sci. Technol., 42(22),
8193-8201.
Semiat, R., Sapoznik, J., Hasson, D. 2010. Energy Aspects in Osmotic Processes.
26
Shen, Q.H., Gong, Y.P., Fang, W.Z., Bi, Z.C., Cheng, L.H., Xu, X.H., Chen, H.L. 2015.
Singh, G., Thomas, P.B. 2012. Nutrient removal from membrane bioreactor permeate using
85.
Sulzacova, K., Trtilek, M., Rataj, T. 2015. Phosphorus removal using a microalgal biofilm in
a new biofilm photobioreactor for tertiary wastewater treatment. Water Research, 71,
55-63.
Van der Bruggen, B., Luis, P. 2015. Forward osmosis: understanding the hype. Rev. Chem.
Wang, B., Lan, C.Q., Horsman, M. 2012. Closed photobioreactors for production of
Wang, M., Kuo-Dahab, W.C., Dolan, S., Park, C. 2014. Kinetics of nutrient removal and
Xu, M., Li, P., Tang, T.Y., Hu, Z.Q. 2015. Roles of SRT and HRT of an algal membrane
Zhao, S.F., Zou, L., Tang, C.Y.Y., Mulcahy, D. 2012. Recent developments in forward
27
Figure Captions
Figure 1 Schematic diagram of the OMPBR setup: (1) compressed air; (2) compressed CO2;
(3) flow meter; (4) humidification tank; (5) MBR tank; (6) membrane module; (7) air
diffuser; (8) feed tank; (9) weighing scale; (10) stirring plate; (11) DS; (12) concentrated DS
stock; (13) effluent; (14) peristaltic pump; (15) conductivity meter, and; (16) data logger
Figure 2 Kinetics of nutrients removal and removal efficiency in MPBR: (a) NH4+-N; (b)
Figure 3 Kinetics of nutrients removal and removal efficiency in OMPBR: (a) NH4+-N; (b)
treatment
Figure 6 Comparison of nutrients removal rate in MPBR and OMPBR at different HRTs: (a)
28
Figure 1
29
HRT = 2 days HRT = 1.5 days HRT = 1 day
(a)
(b)
2.0 100
(c)
Removal efficiency (%)
80
PO43--P conc. (mg/L)
1.5
60
1.0
40
0.5
20
0.0 0
0 5 10 15 20 25 30 35 40
Time (days)
Figure 2
30
HRT = 2 days HRT = 1.5 days HRT = 1 day
(a)
(b)
10 100
Removal efficiency (%)
(c) 8 80
PO43--P conc. (mg/L)
6 60
4 40
2 20
0 0
0 5 10 15 20 25 30 35 40 45 50
Time (days)
Figure 3
31
300
250
Biomass conc. (mg/L)
OMPBR
200
MPBR
150
100
50
0
0 5 10 15 20 25 30 35 40 45 50
Time (days)
Figure 4
32
105
(a) MPBR OMPBR
95
90
85
80
75
2 1.5 1
HRT (days)
100
(b) MPBR OMPBR
NO3--N removal efficiency (%)
80
60
40
20
0
2 1.5 1
HRT (days)
100
80
60
40
20
0
2 1.5 1
HRT (days)
Figure 5
33
4.0
2.0
1.0
0.0
2 1.5 1
HRT (days)
0.20
(b) MPBR OMPBR
NO3--N removal (mg/L-day)
0.16
0.12
0.08
0.04
0.00
2 1.5 1
HRT (days)
2.0
MPBR OMPBR
PO43--P removal (mg/L-day)
(c)
1.6
1.2
0.8
0.4
0.0
2 1.5 1
HRT (days)
Figure 6
34
Table 1 Comparison of MPBR and OMPBR performance with other studies
Max.
Wastewater
HRT cell %N %P
Microalgae composition Reference
(days) conc. removal removal
(mg/L)
(g/L)
C. vulgaris,
TN = 7.5 Honda et
B. braunii, 1-2 0.92 91 60
TP = 0.15 al., 2012
S. platensis
TN = 40.02 Choi,
C. vulgaris 3.4 - 96.4 92.8
TP = 9.24 2015
TN = 7-22 Marbelia
C. vulgaris 2 0.7 80 50-80 et al.,
TP = 1.6-2.2 2014
TN = 18.8 Boonchai
Chlorophyceae
2 1.3 66.5 94.5 & Seo,
sp. TP = 1.01 2015
TN = 9.51 Xu et al.,
C. vulgaris < 0.5 1.5 73.4 91.3
TP = 1.81 2015
35
92-99 100
36
Table 2 Energy input and operating costs of MPBR and OMPBR
OMPBR MPBR
OMPBR Power (kWh) Cost ($) MPBR Power (kWh) Cost ($)
2 days HRT 0.002 0.006 2 days HRT < 0.0096 < 0.028
1.5 days HRT 0.005 0.014 1.5 days HRT < 0.0096 < 0.028
1 day HRT 0.012 0.035 1 day HRT < 0.0096 < 0.028
Total Total
2 days HRT 0.101 0.292 2 days HRT < 0.0766 < 0.220
1.5 days HRT 0.104 0.300 1.5 days HRT < 0.0766 < 0.220
1 day HRT 0.111 0.321 1 day HRT < 0.0766 < 0.220
37
Highlights
38