You are on page 1of 10

Biochemical Engineering Journal 133 (2018) 130–139

Contents lists available at ScienceDirect

Biochemical Engineering Journal
journal homepage:

Regular article

A new Sponge-GAC-Sponge membrane module for submerged
membrane bioreactor use in hospital wastewater treatment
Qusay F. Alsalhy a,∗ , Faris H. Al-Ani b , Arshed E. Al-Najar b
Membrane Technology Research Unit, Chemical Engineering Department, University of Technology, Alsinaa Street 52, Baghdad, Iraq
Building and Construction Engineering Department, University of Technology, Alsinaa Street 52, Baghdad, Iraq

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

Article history: A new Sponge-GAC-Sponge membrane module design for use in a membrane bioreactor (SGSMBR)
Received 9 January 2018 is presented in this study. This work highlights an alternative MBR design in which a composite
Accepted 11 February 2018 Sponge-Granular Activated Carbon-Sponge (SGS) layer is covered around the membrane module. The
Available online 12 February 2018
performance and membrane fouling of both the SGSMBR and a University of Cape Town with membrane
(UCT-MBR) system are investigated for use in hospital wastewater treatment. It has been found that
decreasing the hydraulic retention time (HRT) from 8 to 4 h resulted in higher COD, NH3 , and P removal
Membrane bioreactor
efficiency in the SGSMBR process when compared with the UCT-MBR process. Membrane fouling is con-
Sponge layer
PVC ultrafiltration
trolled in the SGSMBR by decreasing the cake layer thickness on the membrane surface by about 96%. The
Cake layer flux recovery efficiency (FRE%) of the membranes was highly improved in the new SGSMBR design. The
Membrane fouling COD, NH3 , and P removal efficiency was improved significantly from 73.6, 84.9, and 58% by using UCT-
MBR to 85, 96, and 71%, respectively by using UCT-SGSMBR and SGSMBR. Finally, the SGSMBR showed
biomass retention superior to that measured in the UCT-MBR. This work reveals, for the first time, that a
composite layer covering the membrane module is a viable alternative to anoxic and anaerobic conditions
in MBR systems.
© 2018 Elsevier B.V. All rights reserved.

1. Introduction However, fouling, and bio-fouling in particular, acutely affects
MBR processes. Generally, bio-fouling occurs via the accumulation
In recent decades, membrane bioreactor (MBR) systems have of biomass at a membrane surface due to inorganic precipitation
emerged as a suitable alternative to traditional biological recla- and adhesion of bacterial cells to the surface of the membrane [4,5].
mation processes for carbon source as well as nutrient removal One of the most serious operational problems encountered in MBR
[1]. MBRs have several advantages, including high removal effi- systems, bio-fouling causes rapid declines in the permeate flux,
ciency [1], low space requirements, perfect total solids removal, reduced membrane productivity, excess membrane substitution,
low sludge generation, and perfect retention of biomass inside the and increased operational costs.
system [2,3]. Various methods have been used in MBRs to decrease fouling
of the membrane. Ngo and Guo [6] reported that the distributed
injection of air through submerged MBR (SMBR) processes, along
with the addition of low-dosage green bioflocculant (GBF), reduced
layer fouling to nearly zero after 70 d of operation; this method
Abbreviations: SGSMBR, sponge – granular activated carbon – sponge – mem- also reduced backwash recurrence. Wei et al., [7] was explored a
brane bioreactor; SGS, sponge – granular activated carbon – sponge; UCT-MBR, chemical cleaning-in-place (CIP) method for treatment of domestic
University of Cape Town with membrane; HRT, hydraulic retention time; SRT, solids wastewater in a long-term, pilot-scale operation using MBRs. The
retention time; COD, chemical oxygen demand; FRE, flux recovery efficiency; GAC,
use of a biomass support medium such as powder activated car-
granular activated carbon; MBR, membrane bioreactor; GBF, green bioflocculant;
WWTP, wastewater treatment plant; CIP, chemical cleaning-in-place; PAC, powder bon (PAC), plastic media, and sponges in MBRs is a powerful and
activated carbon; EPSs, extracellular polymeric substances; SMBR, sponge mem- promising strategy for membrane fouling control [8,9]. Combina-
brane bioreactor; PVC, polyvinyl chloride; DMAc, dimethylacetamide; PWP, pure tions of a MBR and PAC, in which PAC is used as a pre-treatment
water permeability; AS, activated sludge; MLSS, mixed liquor suspended solid; BAG, before membrane technology, have proven promising in wastew-
biomass attached growth; ZnO, zinc oxide; DO, sissolved oxygen.
∗ Corresponding author. ater pre-treatment.
E-mail address: (Q.F. Alsalhy).
1369-703X/© 2018 Elsevier B.V. All rights reserved.
Q.F. Alsalhy et al. / Biochemical Engineering Journal 133 (2018) 130–139 131

layer around the membrane module functions as an alternative to
Nomenclature the aerobic, anoxic and anaerobic conditions in typical MBR sys-
tems; specifically, the first sponge sub-layer (at the outer surface
of the composite layer) works as an alternative to aerobic condi-
V Volume (m3 )
tioning, the second sub-layer (i.e., the internal porous surfaces of
t Time (h)
the sponge and granular activated carbon) works as an alterna-
A Area (m2 )
tive to anoxic conditioning, and the third sponge sub-layer works
P Transmembrane pressure (bar)
as an alternative to anaerobic conditioning, as the previous two
R Removal efficiency (%)
sub-layers prevent dissolved oxygen transport to this final sub-
Cp Pollutant concentrations of the influent (mg/L)
layer. Other benefits of this MBR design include low cost due to the
Cf Pollutant concentrations of the effluent (mg/L)
possibility of reactivating sub-layers through periodic cleaning as
J Flux ((l/m2 h)
well as reduced membrane fouling, enhanced permeation flux and
M Mass of the permeated water (l)
improved phosphorus and nitrogen removal. Moreover, this new
A Effective membrane area (m2 )
method facilitates long-term maintenance in SGSMBR processes.
The objectives of this study are investigating the effects of a
new sponge-granular activated carbon composite layer covered the
Actually, higher concentrations of PAC in the MBR resulted in membrane on the performance of a SGSMBR for actual hospital
better adsorption, deterioration, and biodegradation and reduced wastewater treatment, examining the effects of the SGS composite
Mixed Liquor Suspended Solid (MLSS) fouling components such layer on biomass growth on sponge surfaces and interior sub-layers
as Extracellular Polymeric Substances (EPSs) and fine colloids as well as on the cake layer thickness on the membrane surface
[10]. Sponges are considered to be ideal attached-growth mate- are evaluated, proving that the SGS composite layer covered the
rial as they can be used as a transportation medium for active membrane module works as an alternative to the anoxic and anaer-
biomass, decrease cake layer accumulation on the membrane sur- obic conditions in typical MBR systems; and studying effects of
face and retain biomass by encouraging a hybrid growth system hydraulic retention time (HRT) on the COD, NH3 , and P removal effi-
(that combines suspended and attached growth) [10,11]. Sponges ciency in the SGSMBR process when compared with the UCT-MBR
have performed well in biological treatments because of several process. In addition, a comparison was conducted in an effort to
advantages, for example; high interior porosity and specific sur- evaluate the performance of UCT-MBR, UCT-SGSMBR and SGSMBR
face area, high stability to hydrolyses, light weight and economy systems.
The impact on nitrogen removal of using a sponge cube
as a biomass supported medium was studied by Deguchi and
Kashiwaya [13]. They found that the rates of nitrification and den-
2. Material and methods
itrification in the sponge system were higher than those measured
in a suspended growth reactor by 1.5 and 1.6 times, respectively.
2.1. Materials
Guo et al., [14] studied the impacts on fouling and permeate flux
of the membrane of suspending a cubic sponge in the tank of the
PVC resins of 65 kg/mol were used as membrane materials and
SMBR. They reported that the suspended-sponge membrane pro-
were obtained from the Georgia Gulf Company (Georgia, USA). N,
cess, in which the sponge filled 10% of the tank volume, decreased
N-dimethylacetamide (DMAc) solvents were supplied by Sigma-
membrane fouling and enhanced the sustainable flux by 100% over
Aldrich. ZnO nanoparticles (99%; 10–30 nm in diameter; Product
that of the SMBR alone. Nguyen et al., [3] demonstrated that the
No. 8411DL) were purchased from SSNano, USA.
suspended sponge membrane bioreactor (SSMBR) succeeded in
decreasing membrane fouling and had high organic carbon and
nutrient removal efficiencies.
Moreover, the influence of sponge size and type on the perfor-
mance of a conventional biological system was reported by Nguyen 2.2. Membrane preparation
et al., [15]. They used sponges of types S28-30/45R, S28-30/60R,
S28-30/80R and S28-30/90R with dimensions of 1 cm3 , 2 cm3 and The PVC polymer material was dried at 60 ◦ C in an oven for 4 h to
3 cm3 . The results indicated similar removal efficiencies for nutri- remove moisture. The casting solution was prepared through the
ents and organic substances among the various types of sponge. addition of dried 10 wt.% PVC to 90 wt.% N,N-dimethylacetamide
The studies mentioned above explored the effects of introduc- (DMAc) solvent. The PVC was mixed with the DMAc using a mag-
ing cubic sponges and granular activated carbon (GAC) to the MBR netic stirrer for 2 d at 200 rpm and 40 ◦ C until homogeneity was
tank, finding that these additions improved treatment performance achieved, after which the inorganic ZnO nanoparticles were added.
and decreased membrane fouling. Indeed, the direct addition of the Then, the final casting solution with ZnO nanoparticles was placed
cubic sponge to the bioreactor causes mixing with the activated in an Ultrasonic water bath for 15 min to prevent ZnO nanoparticle
sludge, making it impossible to separate the sponges and the sludge aggregation in the final casting solution.
at the end of the process. Also while the addition of GAC causes a The polymer solutions were cast with 200 ␮m thicknesses uti-
significant reduction in membrane fouling, it is not possible to reuse lizing a motorized film applicator (CX4 mtvmesstechnik, Germany)
the GAC again. We therefore focus on, and solve, these problems in under atmospheric conditions. The prepared membranes were sub-
this study. merged in a non-solvent coagulation bath (i.e., deionized water) at
This study presents a new Sponge-GAC-Sponge membrane room temperature for deposition. The nascent membrane was kept
module in a membrane bioreactor (SGSMBR) design. This work in tap water for 48 h to remove the remaining DMAc. Finally, the
highlights an alternative MBR design in which a composite layer, membrane was moved and kept in a 30 wt% glycerol solution for
composed of three sub-layers made of Sponge-Granular Activated 48 h to protect the membrane structure from collapse and crack-
Carbon-Sponge (SGS), is covered around the membrane module ing. Three identical membrane sheets were selected for membrane
instead of distributing the cubic sponges and granular activated characterization and used in ultrafiltration UF tests to procure mean
carbon throughout the MBR tank. In this design, the composite flux and pollutant removal values.
132 Q.F. Alsalhy et al. / Biochemical Engineering Journal 133 (2018) 130–139

2.3. Characterization of the membrane surface where Jwater,1 and Jwater,2 are the flux before and after the MBR test,
2.3.1. Scanning electron microscopy
SEM is a common technique used to analyse the structure 2.6. Sponges
of a prepared membrane. SEM images were made with a TES-
CAN VEGA3 SB instrument (EO Elektronen-Optik-Service GmbH, In the SGSMBR system, two kinds of polyester-urethane sponges
Germany) with an accelerating voltage of 30 KV. Cross-sections of (PUSs) were used: coarse pore structure (S 28–30/45R) (density
the flat sheet membranes were obtained via freeze fracturing in of 28–30 kg/m3 with 45 cells per 25 mm) and fine pore structure
liquid nitrogen to obtain a clean, brittle fracture. (S 16–18/80R) (density of 16–18 kg/m3 with 80 cells per 25 mm).
The dimensions of both the coarse and fine pore structure sponge
2.3.2. Atomic force microscope sheets were 9.5 × 15.5 × 1.5 cm3 . The two types of sponge were pur-
The flat-sheet membrane was subjected to broad surface exami- chased from Al-Hillal Company for sponge industry, Baghdad, Iraq,
nation using an AFM made by Angstrom Advanced Inc. (USA; model whereas the GAC were purchased from ROMIL PURE CHEMISTRY,
AA3000) in contact mode with a reusable silicon tip. The observa- Cambridge GB-CB259QT.
tions obtained include appraisal of the topography (the rise and fall
of the sample surface), the lateral force, and the deviation. Measur-
2.7. Sponge-GAC-Sponge membrane bioreactor (SGSMBR)
able pore estimate dissemination was set up using the IMAGER 4.31
program for the external surfaces of each flat-sheet membrane.
The PVC/ZnO flat sheet membrane module had an average pore
2.4. Permeation flux and the removal of pollutants size of 211 nm, as evaluated by AFM, and a surface area of 18 cm2 .
The membrane cell was covered in three sub layers above the
The water flux and pollutants removal studies for the mem- PVC/ZnO flat sheet membrane cell as shown in Fig. 2. The top layer
branes were conducted using cross-flow filtration experiments. All is formed from coarse-pore-structure sponge, the middle layer con-
experiments were performed in cross flow mode with a membrane tains GAC, and the bottom layer is comprised of fine-pore-structure
cell under a −0.4 bar vacuum using a feed solution temperature of sponge, as shown in Fig. 2.
25 ◦ C. The effective membrane area was 18 cm2 . The volume of the
solution was 5 L. Permeate was collected in a graduated cylinder
until it reached steady state. Pure water permeability (PWP) was 2.8. Experimental setup
calculated as follows:
The experiment was conducted using three pilot plants, which
V were fed hospital wastewater from a storage tank via a feed suc-
PWP = (1)
t.A.P tion pump. This actual wastewater was collected downstream of
where PWP is the PWP of the membrane (l/(m2 h bar)), V is the a wastewater treatment plant (WWTP) located in Hilla hospital
volume collected (L), t is permeate collection time (h), P is the (Iraq).
transmembrane pressure, and A is the membrane surface area (m2 ). The pilot plant composed of a UCT-MBR as shown in Fig. 1a;
The removal R (%) of the dissolved pollutant was calculated as the UCT-MBR composed of three tanks: an anaerobic tank (8.0 L),
follows: an anoxic tank (16 L), and the MBR (32 L). The PVC/ZnO flat-sheet
 membrane modules were installed in the MBR. Air was injected
R(%) = 1− × 100 (2) at the base of the aerobic tank to supply oxygen for microorgan-
isms and reduce fouling on the membrane. The anaerobic tank and
where Cf and Cp are the pollutant concentrations of the influent the anoxic tank each contained one stirrer. Mixed liquors from the
and effluent, respectively. MBR tank were returned to the anoxic tank (Recirculation 1) and
then, from anoxic tank to the anaerobic tank (Recirculation 2) by
2.5. Membrane fouling analysis two pumps. Hospital wastewater was pumped into the anaerobic
tank from the storage tank. COD was consumed partially and phos-
Antifouling and permeation experiments of PVC/ZnO flat-sheet phorus was released in the chamber of anaerobic process. Then,
membrane were carried out under constant vacuum of −0.4 bar for the effluent moved to the anoxic tank and the MBR. Organic matter
pure water and AS with an MLSS of 10000 mg/L to prevent com- oxidation, nitrification and phosphorus accumulation occurred in
paction effects in the membrane. The membrane pure water flux the aerobic zones. Recirculation 1 consisted of NO3 recirculation
was measured by applying a −0.4 bar vacuum for 30 min before the from the membrane tank to the bioreactor anoxic chamber, which
main flux test. The permeation flux was measured for 90 min and permitted nitrogen removal and reduced the impact of nitrate in
calculated as follows: wastewater come in the anaerobic zone. Recirculation 2 is refers
to the returned wastewater to anaerobic tank from anoxic cham-
J= (3) ber in order to increase the organic matter utilization and provided
the optimal conditions for fermentation of organic material and
where J, M, t and A are the permeation flux, mass of the permeated phosphorus uptake in the anaerobic tank.
water, permeation time and effective membrane area, respectively. The UCT-SGSMBR is shown in Fig. 1a; the UCT-SGSMBR is the
Then, sludge filtration was performed in the UCT-MBR process same as the UCT-MBR in construction and operation except for the
shown in Fig. 1 for 90 min. Next, the flat sheet membranes were addition of the multilayer sponge covering the membrane cell. A
ejected from the MBR and cleaned gently with distilled water, after schematic of the SGSMBR is presented in Fig. 1b; the effective vol-
which the pure water flux was measured to check the flux recovery ume of the bioreactor was 32 L. Actual wastewater was pumped
efficiency (FRE) of the flat sheet membranes, which was calculated into the reactor using a feeding pump to control the feed rate while
as follows: the effluent flow rate was controlled by a suction pump. Level sen-
  sor was used to control the wastewater volume in the reactor. A
FRE (%) × 100 (4) pressure gauge was used to measure the TMP and air was injected
at the base of the aerobic tank to supply oxygen for microorgan-
Q.F. Alsalhy et al. / Biochemical Engineering Journal 133 (2018) 130–139 133

Fig. 1. Schematic diagram of: A) the lab-scale of UCT-MBR and UCT-SGSMBR B) SGSMBR system.

Fig. 2. SGSMBR membrane module.
134 Q.F. Alsalhy et al. / Biochemical Engineering Journal 133 (2018) 130–139

Table 1
Operating conditions and working concentration of MLSS in the steady state.

Parameter Run 1 Run 2 Run 3 2.10. Assessment of water quality

ZnO (wt.%) 0.1 0.1 0.1 Influents and membrane effluents were gathered from the UCT-
PVC (wt.%) 10 10 10
MBR for estimation of conventional parameters. COD, MLSS, NH3 ,
Volume (L) anaerobic 8 8 0
anoxic 16 16 0
and P were determined in accordance with common Standard
aerobic 32 32 32 Methods (APHA, 2005).
HRT (h) anaerobic 2 2 0
anoxic 4 4 0
aerobic 8 8 4
3. Results and discussion
SRT (d) 25 25 25
MLSS (g/L) 10 10 10 3.1. Membrane properties
Vacuum (bar) 0.4 0.4 0.4
Operation time (d) 21 21 21
Fig. 3 shows three AFM images of the top surface of the PVC
membranes and SEM images of the cross-sectional structure of the
membranes prepared with 0.1 g of ZnO nanoparticles in casting
isms. The SGSMBR was filled with sludge from the local Wastewater
solution. In these AFM images, the dark regions are valleys (i.e.,
Treatment Plant and acclimatized to actual wastewater.
pores) and the bright regions are high points or nodules. It can be
seen that the nodular aggregates have merged together to form a
2.9. Process operating conditions number of string-like structures with ZnO nanoparticles.
The AFM technique reveals that the average roughness and pore
The bioreactor was initially filled with seed sludge obtained diameter of the PVC/ZnO flat sheet membrane are 21.5 (±2.2) nm
from Al-Rustumia Treatment Plant, Baghdad. The stabilization and 211.32 (±5.64) nm, respectively. Regarding the structure of the
period for each MBR configuration prior to the 21 days of operation cross-section of the membrane, it can be noticed from Fig. 3B that
was 45 days. The initial MLSS was 1.5 (±0.09) g/L, and during the the cross-sectional structure of the membrane is composed of two
stabilization period, MLSS become 8.8 (±1.1) g/L. In the UCT-MBR micro-void structure.
the concentration of the dissolved oxygen (DO) was kept within
1–4 mg/L in the aerobic zone, while DO in the anoxic and anaero-
3.2. COD removal
bic zones was limited to <0.2 mg/L and <0.1 mg/L, respectively. To
maintain a constant temperature of 25 ± 2 ◦ C in the system, a tem-
Fig. 4 presents the changes in influent and effluent COD and
perature controller was used. The rates of Recirculations 1 and 2
the COD removal efficiency as function of time for UCT-MBR, UCT-
were respectively, 300% and 100% from the influent. An interrupted
SGS-MBR, and SGSMBR experiments. For UCT-MBR experiment at
filtration method, which consisted of 10 min of suction after 1 min
HRT = 8 h, the influent COD concentration fluctuates extensively,
of repose, was adopted in this effort. The MBR was operated under
ranging between 0.99 (±0.05) and 10.3 (±0.08) mg/L. However,
a constant vacuum (−0.4 bar). The hydraulic retention time (HRT)
the removal efficiency is approximately 73.6% (±0.8) in the UCT-
was maintained at about 2 h for anaerobic processing, 4 h for anoxic
MBR as shown in Fig. 4a. The influent COD during the UCT-SGSMBR
processing and 8 h for aerobic processing, while the solids reten-
experiment at HRT = 8 h varies between 0.825 (±0.06) and 0.88
tion time (SRT) was fixed at 25 d per the optimum experimental
(±0.11) mg/L as shown in Fig. 4b. It can be observed that the
conditions found in Brown et al., and Metcalf and Eddy [16,17]. The
COD removal efficiency measures 85.1% (±2.6). Regarding SGSMBR
operating conditions and working concentrations of the MLSS in
experiment, the influent COD concentration fluctuates extensively,
this study are shown in Table 1.
ranging between 700 and 860 mg/L in Fig. 4c. However, the removal
Moreover, in Al-Hilla hospital, south of Baghdad, Iraq, where
efficiency is approximately 85% in the SGSMBR when the HRT is
surgical operations are carried out within 7 days each month;
decreased from 8 to 4 h (i.e., HRT decreased by 50%).
hence, throughout this period of time the characteristics of the
The lack of a significant reduction in COD removal with decreas-
actual wastewater that collected downstream by wastewater treat-
ing HRT may be explained by the use of the sponge layer as a porous
ment plant (WWTP) is significantly changed due to the blood
biomass support increasing the possibility of contact between
disposal procedures in the sanitary network, which may lead to
microorganisms and the organic substrate. The current study shows
an highly increase in the concentration of organic compounds and
that HRT has a negligible impact on COD removal. The SGSMBR sys-
nutrients. For that reason, it is prefer to conduct the wastewater
tem has succeeded in removing COD with a decrease in the HRT
treatment experiments only 21 days each month.
from 8 to 4 h. This suggests that the new SGSMBR can be oper-
Regarding the measurement of the MLSS in each layer, the
ated with a short HRT because of much higher concentrations of
sponge was squeezed and rinsed with distilled water in order to
microorganisms developed in the solid than in the suspension;
remove all of biomass out of the sponge, and the following equation
therefore the rate of degrading of the substrate is higher.
was used to calculate the amount of MLSS:

MLSS = (dry weight of residue and filter 3.3. Nitrification and denitrification
− dry weight of filter alone, in gm)/mL of sample (5)
Fig. 5 shows NH3 removal in the UCT-MBR, UCT-SGS-MBR, and
SGSMBR systems. In UCT-MBR experimental configuration (Fig. 5a),
While the amount of MLSS remaining in the sponge was calcu- the HRT is 8 h, the average influent NH3 concentration is 166 (±4.6)
lated as follows: mg/L and the average removal efficiency is 84.9% (±3.1), with efflu-
ent concentrations of 24.9 (±1.7) mg/L. These results indicate good
MLSS = (dry weight of residue and filter (sponge) nitrification in the system. In the UCT-SGSMBR experimental con-
figuration, the HRT is 8 h, the average influent NH3 concentration
− dry weight of filter (sponge) alone, in gm)/mL of sample
is 156.4 (±3.1) mg/L, and the average removal efficiency is 96.7%
(6) (±2.1), with effluent concentrations of 5 (±0.5) mg/L as shown
Q.F. Alsalhy et al. / Biochemical Engineering Journal 133 (2018) 130–139 135

Fig. 3. AFM and SEM images of the PVC/ZnO flat sheet membrane.

in Fig. 5b. These values indicate that nitrification and denitrifica- bacteria on the inner surface. In the SGSMBR experimental con-
tion are entirely complete in this system. This increment may be figuration, the HRT is decreased to 4 h, the average influent NH3
explained by additional denitrification in the sponge layer. The concentration is 158 (±2.4) mg/L and the average removal effi-
surface of the coarse sponge, which consists of large pore sizes, ciency is 96.6% (±0.8), with effluent concentrations of 5.2 (±0.7)
can host nitrifying and denitrifying bacterial growth, where the mg/L as shown in Fig. 5c. These values indicate that nitrification
top surface of the sponge represents an aerobic microenvironment and denitrification are entirely complete in this system, which has
and the inner surface represents an anoxic microenvironment; this a HRT of 4 h and does not use either anaerobic or anoxic tanks.
results in more aerobic bacteria on the top surface and more anoxic
136 Q.F. Alsalhy et al. / Biochemical Engineering Journal 133 (2018) 130–139

Fig. 4. COD concentrations and removal efficiency for a: UCT-MBR, b: UCT-SGS-
MBR, and c: SGSMBR experiments.
Fig. 6. NO3 influent and effluent concentration for UCT-MBR, UCT-SGSMBR, and
SGSMBR experiments.

Fig. 5. NH3 concentrations and removal efficiency for a: UCT-MBR, b: UCT-SGS-MBR,
and c: SGSMBR experiments. Fig. 7. P concentrations and removal efficiency for a: UCT-MBR, b: UCT-SGSMBR,
and c: SGSMBR experiments.

SGSMBR is the best configuration for the treatment of nutrients
due to the sponge layer, which is a perfect connected development (when the HRT = 4 h). These results indicate complete denitrifica-
medium because it can act as a biomass support medium. Moreover, tion in the SGSMBR systems.
the enhancement of the nutrient removal efficiency is attributed to
the trapped of the large amount of biomass in the sponges, much 3.4. Phosphorus removal
higher concentrations of microorganisms developed in the solid
than in the suspension, therefore the rate of degrading the substrate Fig. 7 shows variations in influent and effluent P and P removal
is higher and accordingly increases the nitrification and denitrifi- efficiency as a function of time in the UCT-MBR, UCT-SGS-MBR,
cation. Deguchi and Kashiwaya [13] found that the nitrification and and SGSMBR systems. The performance of the UCT-MBR system
denitrification rates in the sponge system were 1.5 and 1.6 times has been investigated for 21 d using a submerged PVC/ZnO flat
higher, respectively, than those measured in a suspended growth sheet MBR. The influent P concentration fluctuates extensively
reactor. from 16 to 18 mg/L. The phosphorus removal efficiency is stable
Cho et al., [18] found that the nutrient removal efficiency was at approximately 58% as shown in Fig. 7a. A simple bio-P removal
higher under short HRTs than under long HRTs; increasing the HRT process must include at least two stages, including a non-aerated
by 30% resulted in a 10% decrease in the nutrient removal effi- stage and an aerated stage. In the first stage, influent wastewater
ciency. However, the short HRT and high flux resulted in increased and returned sludge are combined under non-aerated conditions.
membrane fouling. It is clear from the SGSMBR results that deni- Phosphate release from the biomass and soluble chemical oxygen
trification is completed even with a short HRT and that the system demand (COD) removal are observed under these anaerobic condi-
is not affected by membrane fouling. tions in the absence of both oxygen and nitrate. In the subsequent
SGSMBR located in the aerobic tank is the best nutrient treat- aerated stage, phosphate removal, COD removal and nitrification
ment configuration because nitrification and denitrification occur occur. Phosphorus-rich sludge from this aerobic reactor is usually
in the same (aerobic) tank, eliminating the need for anaerobic discarded. Many processes also include an intermediate denitrifica-
and anoxic tanks. Neither membrane fouling nor treatment per- tion stage, often referred to as an “anoxic” stage, in which nitrified
formance is affected by operation under low HRT and high SRT. mixed liquor is recycled from the aerated reactor. Barnard [19,20]
Higher NH3 removal in the SGSMBR may be caused by the pro- showed that phosphate release under non-aerated conditions is a
motion of biomass-oxidizing bacteria on the sponge layer, which prerequisite to subsequent aerobic phosphate uptake.
provides anoxic conditions in the inner pores and on the bottom In the UCT-SGSMBR experimental configuration (Fig. 7b), the
layer of the sponge. HRT is 8 h, the average influent P concentration is 17.5 (±0.9) mg/L
NO3 removal in the UCT-MBR, UCT-SGSMBR and SGSMBR exper- and the average removal efficiency is 71.2% (±0.9), with effluent
iments is summarized in Fig. 6. NO3 concentrations are 45 (±1.6) concentrations of 5.04 mg/L and the performance of this system
mg/L in the influent and 30.2 (±2.1) mg/L in the effluent for the has been investigated for 21 d. In the SGSMBR experimental con-
UCT-MBR configuration; 45 (±1.9) and 0 mg/L in the influent and figuration (Fig. 7c), the performance has been investigated for 21 d,
effluent, respectively, for the UCT-SGSMBR configuration (when the the HRT is 4 h, the average influent P concentration is 17.4 (±1.2)
HRT = 8 h and the SRT = 25 d); and 45.8 (±2.3) and 0 mg/L in the mg/L and the average removal efficiency is 71.1% (±1.2), with efflu-
influent and effluent, respectively, for the SGSMBR configuration ent concentrations of 5 (±0.65) mg/L. The results also indicate that
Q.F. Alsalhy et al. / Biochemical Engineering Journal 133 (2018) 130–139 137

Fig. 9. Flux recovery efficiency (FRE) of the UCT-MBR and SGSMBR.

Fig. 8. Flux of pure water and filtration of activated sludge in MBR versus time for

P removal decreases after 5–12 d. The results show that the sponge
layer is an ideal biomass support medium for the SGSMBR system.
Sponge with fine pore structure has an effect on the removal of P
due to the small size of the floc, where the small size of the organic
matter particles affects the rate of the hydrolysis. This is because
the small particles size has a larger surface area and thus degrades
rapidly with reduction of the stabilization time compared to large
The P removal (effluent P < 5.0 mg/L) also indicates that
increased biological phosphorus removal, as well as excess phos-
phorus uptake, can be achieved by the accumulation of phosphate
organisms on the sponge layer. Furthermore, microorganisms
attached to the sponge and in the MLSS remove a portion of the
phosphorus biologically, as P is an essential nutrient for biomass Fig. 10. Bio-layer of the UCT-MBR and SGSMBR.
growth [21]. According to the COD, NH3 , and P removal data, the
SGSMBR system succeeds due to the composite layer around the
membrane module, which provides an alternative to anoxic and
anaerobic processing in the MBR system. SGSMBR is an alterna-
tive technique for treatment of wastewater to traditional UCT-MBR
technique. It shows many advantages such as capable of high nitro-
gen and phosphorus removal, simple operation, low cost, small area
required, membrane fouling controlled, and alternative to anoxic
and anaerobic condition.

3.5. Membrane antifouling performance and long-term operation

Membranes that are used for MBR processes shown in Fig. 1
should be able to recover, in terms of water flux, after being in
Fig. 11. Permeate flux versus time for long-term experiment for UCT-MBR and
contact with sludge. Therefore, it is necessary to compare the mem-
brane water flux before and after activated sludge filtration. After
the activated sludge ultrafiltration test, the membranes were gen-
tly cleaned and the pure water fluxes were measured again; the in traditional MBRs. Yang et al., [22] reported that a Sponge-MBR
results are shown in Fig. 8. Flux through the membranes reaches a system was proficient at controlling membrane fouling, and espe-
nearly steady value over time, and the water flux is much higher cially the cake layer on the membrane. The result was an 86%
than the sludge flux. The membrane fouling resistance is usually reduction in cake resistance and a 20% flux increase compared to
evaluated using the FRE value. Higher FRE values indicate better the MBR alone.
membrane antifouling ability during the ultrafiltration process. As Regarding the long-term operation, permeate flux against
can be seen in Fig. 9, the FRE value is 60% (±2) for the UCT-MBR and elapsed time for the long-term (21 day) experiment for UCT-MBR
95.3% (±3.5) for the SGSMBR. and SGSMBR is presented in Fig. 11. It can be seen that the permeate
In Fig. 10 the decrease in the cake layer with the addition of a flux in the UCT-MBR configuration has started with 33.4 (±0.88)
sponge layer is clear; because the MLSS is not attached directly to L/m2 .h and after long-term operation of 21 d, the permeate flux
the membrane, and the multilayer sponge prevented the activated decline to 24 L/m2 .h, whereas in the SGSMBR configuration the per-
sludge from moving to the MBR. The cake layer is 36.87 ␮m thick meate flux has started with 63.8 L/m2 .h and after 21 d of long-term
in the UCT-MBR (difference between bio-layer thickness after and operation the permeate flux slightly decreased to 63.3 L/m2 .h. In
before the process) as shown in Fig. 10, where the addition of the fact, the sponge/GAC acts as a pre-filter and prevents accumula-
multilayer sponge is decreases the cake to 1.58 ␮m in the SGSMBR. tion of too much biomass on the membrane during the long-term
In fact, SGS module acts like the “real” membrane, the PVC/ZnO operation.
membrane having, merely, the role of finishing the treatment and It is worth to mention here for the first time that the concept
preventing some microorganisms to leave with the treated water. of the sponge/GAC pre-filtration is new and very interesting. It is
The SGS module can solve the issue of cake formation so prevalent clear that this pre-filtration will long-term in fact be beneficial and
138 Q.F. Alsalhy et al. / Biochemical Engineering Journal 133 (2018) 130–139

Table 2
The types of bacteria isolated from the sponges.

Condition Name of bacteria

Aerobic Escherichia coli
Enterobacter sp.
Bacillus subtilis
Klebsiella pneumoniae
Proteus mirabilis
Anoxic Streptococcus intermedius
Anaerobic Actinomyces sp.

The bacteria were isolated and investigated using different bio-
chemical tests and the Gram stain procedure [25]. Table 2 shows the
Fig. 12. MLSS variation of the activated sludge of the experiments. types of bacteria isolated from the two types of sponges. The types
of bacteria isolated from the two types of sponges confirmed our
hypothesis that the composite layer covered the membrane module
this pre-filter will accumulate too much biomass and protecting
working as an alternative to aerobic, anoxic and anaerobic pro-
the polymeric membrane.
cessing in the MBR system; for example, the first sponge sub-layer
at the outer surface working as an alternative to aerobic process-
3.6. MLSS in activated sludge
ing, the second sub-layer (i.e., the internal porous surfaces of the
sponge and granular activated carbon) functions as an alternative
MLSS levels in the activated sludge in the UCT-MBR and biomass
to anoxic processing, and the third sub-layer of sponge functions
attached growth (BAG) in the SGSMBR experiments are presented
as an alternative to anaerobic processing.
in Fig. 12. Every day, MLSS was tested in order to monitor the phys-
ical properties of the activated sludge. The MLSS is 10 g/L in the
UCT-MBR. In the SGSMBR, the BAG can be divided into two types of 4. Conclusions
microorganisms: the BAG in the upper sponge layer aerobic zone
(which has a coarse pore structure), and that in the lower-layer A new membrane module design involving a SGSMBR is consid-
anaerobic zone (which has a fine pore structure). The activated ered in this work. This is the first work to investigate the effects of
sludge content was measured at the beginning of the experi- using a sponge-granular activated carbon composite layer covered
ment. The MLSS development rates are 26.7(±0.03) (g biomass/g around the membrane on the performance and operating parame-
sponge) and 1.7 (±0.01) (g biomass/g sponge) in the upper and ters of a SGSMBR for actual hospital wastewater treatment.
lower layers, respectively. The results indicate that the BAG in the The essential conclusions of this work are:
SGSMBR was higher than that in the UCT-MBR. The coarse layer
has more attached biomass than does the fine layer, which leads to • Using the new SGS composite layer module design, the COD, NH3
increased denitrification due to the anoxic environment provided and P removal efficiencies are improved from 73.6%, 84.9% and
and increases the flux to the second layer. The fine-pore layer has 58%, with an 8-h HRT in the UCT-MBR to 85%, 96.6% and 71.1%,
better P removal because the previous two sub-layers prevent the respectively, with a shorter HRT of 4 h. These removal efficien-
transport of dissolved oxygen (DO) to the final layer, creating an cies increase because large amount of biomass are trapped in the
anaerobic environment. Additionally, the small pore size prevents sponges, which increases nitrification and denitrification.
large floc and microorganisms from attaching and/or adhering to • Membrane fouling is controlled by reducing the cake layer thick-
the membrane surface, reducing membrane fouling. ness on the membrane surface in the SGSMBR by approximately
In general, a high MLSS concentration is not favorable in the 96%.
bioreactor because it increases the sludge viscosity as well as it • The membrane flux recovery efficiency (FRE%) is substantially
does membrane fouling. Meng et al., [23] found that high MLSS improved in the new SGSMBR design because microorganisms
concentrations were the main cause of high viscosity due to the are not directly attached to the membrane; the multilayer sponge
presence of EPS. Meng et al., [24] was found similar observations, prevents the transport of the activated sludge to the MBR.
which suggested that sludge viscosity was affected by the MLSS • The composite layer covered the membrane module works as
concentration, EPS and filamentous bacteria. In this study, high an alternative to anoxic and anaerobic conditioning in the MBR
MLSS concentrations do not affect the viscosity or the fouling rate. system.
In general, the SGSMBR shows better biomass retention than does
the UCT-MBR because of biomass entrapment in the sponge layer.
3.7. Identification of bacteria
[1] H.H. Ngo, W.S. Guo, W. Xing, Evaluation of a novel sponge-submerged
membrane bioreactor (SSMBR) for sustainable water reclamation, Bioresour.
Three samples were taken from the sponge. Sample No. 1 was Technol. 99 (2008) 2429–2435.
taken from the top sponge layer (1 st coarse sponge layer from the [2] W.S. Guo, H.H. Ngo, C.G. Palmer, W. Xing, A.Y.J. Hu, A. Listowski, Roles of
sponge sizes and membrane types in a single stage sponge-submerged
surface facing the bioreactor) and Sample No. 2 was taken from membrane bioreactor for improving nutrient removal from wastewater for
the bottom sponge layer (1 st coarse sponge layer from the sur- reuse, Desalination 249 (2009) 672–676.
face facing the GAC layer) in order to identify the types of bacteria [3] T.T. Nguyen, H.H. Ngo, W.S. Guo, J.X. Li, A. Listowski, Effects of sludge
concentrations and different sponge configurations on the performance of a
present and to improve the aerobic and anoxic environments in sponge-submerged membrane bioreactor, Appl. Biochem. Biotechnol. 167
the same layers. Sample No. 3 was taken from the second sponge (2012) 1678–1687.
layer (i.e., the surface facing the membrane) to improve the anaer- [4] W.S. Guo, H.H. Ngo, J.X. Li, A mini-review on membrane fouling, Bioresour.
Technol. 122 (2012) 27–34.
obic environment. The samples were cultured on different media, [5] J. Kim, J. Shin, H. Kim, J.Y. Lee, M. Yoon, S. Won, B.C. Lee, K.G. Song, Membrane
including MacConkey’s agar, Blood agar, Eosin methylene blue agar, fouling control using a rotary disk in a submerged anaerobic membrane
and Nutrient agar, under aerobic, anoxic, and anaerobic conditions. sponge bioreactor, Bioresour. Technol. 172 (2014) 321–327.
Q.F. Alsalhy et al. / Biochemical Engineering Journal 133 (2018) 130–139 139

[6] H.H. Ngo, W.S. Guo, Membrane fouling control and enhanced phosphorus [15] T.T. Nguyen, H.H. Ngo, W. Guo, A. Johnston, A. Listowski, Effects of sponge size
removal in an aerated submerged membrane bioreactor using modified green and type on the performance of an up-flow sponge bioreactor in primary
bioflocculant, Bioresour. Technol. 100 (2009) 4289–4291. treated sewage effluent treatment, Bioresour. Technol. 101 (2010) 1416–1420.
[7] C.H. Wei, X. Huang, R. Ben Aim, K. Yamamoto, G. Amy, Critical flux and [16] P. Brown, S.K. Ong, Y.W. Lee, Influence of anoxic and anaerobic hydraulic
chemical cleaning-in-place during the long-term operation of a pilot-scale retention time on biological nitrogen and phosphorus removal in a
submerged membrane bioreactor for municipal wastewater treatment, Water membrane bioreactor, Desalination 270 (2011) 227–232.
Res. 45 (2011) 863–871. [17] Metcalf, Eddy, Wastewater Engineering: Treatment and Reuse, fourth edition,
[8] S.Z. Ahmmad, A. Zealand, J. Dolfing, C. Mota, Low-energy treatment of McGraw Hill Inc., New York, NY, 2003.
colourant wastes using sponge bio-filters for the personal care product [18] J. Cho, K. Guen Song, S.H. Lee, K.H. Ahn, Sequencing anoxic/anaerobic
industry, Bioresour. Technol. 129 (2013) 634–638. membrane bioreactor (SAM) pilot plant for advanced wastewater treatment,
[9] H.H. Ngo, M.C. Nguyen, N.G. Sangvikar, T.T.L. Hoang, W.S. Guo, Simple Desalination 178 (2005) 219–225.
approaches towards a design of an attached-growth sponge bioreactor [19] J.L. Barnard, Biological nutrient removal without the addition of chemicals,
(AGSB) for wastewater treatment and reuse, Water Sci. Technol. 54 (11–12) Water Res. 9 (1975) 485.
(2006) 191–197. [20] J.L. Barnard, A review of biological phosphorus removal in the activated
[10] C.A. Ng, D. Sun, M.J.K. Bashir, S.H. Wai, L.Y. Wong, H. Nisar, B. Wu, A.G. Fane, sludge process, Water S. A. 2 (1976) 136.
Optimization of membrane bioreactors by the addition of powdered activated [21] L.M. Yuan, C.Y. Zhang, Y.Q. Zhang, Y. Ding, D.L. Xi, Biological nutrient removal
carbon, Bioresour. Technol. 138 (2013) 38–47. using an alternating of anoxic and anaerobic membrane bioreactor (AAAM)
[11] C. Psoch, S. Schiewer, Direct filtration of natural and simulated river water process, Desalination 221 (2008) 566–575.
with air sparging and sponge ball application for fouling control, Desalination [22] Q. Yang, J. Chen, F. Zhang, Membrane fouling control in a submerged
197 (2006) 190–204. membrane bioreactor with porous, flexible suspended carriers, Desalination
[12] J.A. Guillén Sánchez, P.R. Cuéllar Guardado, C.M. Lopez Vazquez, L.M. de 189 (2006) 292–302.
Oliveira Cruz, D. Brdjanovic, J.B. van Lier, Anammox cultivation in a closed [23] F. Meng, H. Zhang, F. Yang, Y. Li, J. Xiao, X. Zhang, Effect of filamentous
sponge-bed trickling filter, Bioresour. Technol. 186 (2015) 252–260. bacteria on membrane fouling in submerged membrane bioreactor, J. Membr.
[13] H. Deguchi, M. Kashiwaya, Study on nitrified liquor recycling process Sci. 272 (2006) 161–168.
operations using polyurethane foam sponge cubes as a biomass support [24] F. Meng, B. Shi, F. Yang, H. Zhang, Effect of hydraulic retention time on
medium, Water Sci. Technol. 30 (6) (1994) 143–149. membrane fouling and biomass characteristics in submerged membrane
[14] W.S. Guo, H.H. Ngo, S. Vigneswaran, W. Xing, P. Goteti, A novel sponge bioreactors, Bioprocess Biosyst. Eng. 30 (2007) 359–367.
submerged membrane bioreactor (SSMBR) for wastewater treatment and [25] R.H. Harvey, P.C. Champe, B.D. Fisher, Microbiology, 2nd edition, 2007, pp.
reuse, Sep. Sci. Technol. 43 (2008) 273–285. 50–65.