Effect of powdered activated carbon on Chinese traditional medicine wastewater treatment in submerged membrane bioreactor with electronic control backwashing.

Abstract

Chinese traditional medicine wastewater, rich in macromolecule and easy to foam in aerobic biodegradation such as Glycosides, was treated by two identical bench-scale aerobic submerged membrane bioreactors (SMBRs) operated in parallel under the same feed, equipped with the same electronic control backwashing device. One was used as the control SMBR (CSMBR) while the other was dosed with powdered activated carbon (PAC) (PAC-amended SMBR, PSMBR). The backwashing interval was 5 min. One suction period was about 90 min by adjusting preestablished backwashing vacuum and pump frequency. The average flux of CSMBR during a steady periodic state of 24 d (576 h) was 5.87 L/h with average hydraulic residence time (HRT) of 5.97 h and that of PSMBR during a steady periodic state of 30 d (720 h) was 5.85 L/h with average HRT of 5.99 h. The average total chemical oxygen demand (COD) removal efficiency of CSMBR was 89.29% with average organic loading rate (OLR) at 4.16 kg COD/(m3 x d) while that of PSMBR was 89.79% with average OLR at 5.50 kg COD/(m3 x d). COD concentration in the effluent of both SMBRs achieved the second level of the general wastewater effluent standard GB8978-1996 for the raw medicine material industry (300 mg/L). Hence, SMBR with electronic control backwashing was a viable process for medium- strength Chinese traditional medicine wastewater treatment. Moreover, the increasing rates of preestablished backwashing vacuum, pump frequency, and vacuum and flux loss caused by mixed liquor in PSMBR all lagged compared to those in CSMBR; thus the actual operating time of the PSMBR system without membrane cleaning was extended by up to 1.25 times in contrast with the CSMBR system, and the average total COD removal efficiency of PSMBR was enhanced with higher average OLR.

Full text

Page 1

Journal of Environmental Sciences 19(2007) 1037–1042
Effect of powdered activated carbon on Chinese traditional medicine
wastewater treatment in submerged membrane bioreactor
with electronic control backwashing
LIU Xiao-lei1,2, REN Nan-qi1,∗, MA Fang1
1. School of Municipal and Environmental Engineering, Harbin Institute of Technology, Harbin 150090, China. E-mail: nanyuanx@126.com
2. School of Water Conservancy and Environment Engineering, Changchun Institute of Technology, Changchun 130012, China
Received 17 November 2006; revised 16 January 2007; accepted 25 January 2007
Abstract
Chinese traditional medicine wastewater, rich in macromolecule and easy to foam in aerobic biodegradation such as Glycosides,
was treated by two identical bench-scale aerobic submerged membrane bioreactors (SMBRs) operated in parallel under the same feed,
equipped with the same electronic control backwashingdevice.One wasused as the control SMBR(CSMBR) while the other wasdosed
with powdered activated carbon (PAC) (PAC-amended SMBR, PSMBR). The backwashing interval was 5 min. One suction period was
about 90 min by adjusting preestablished backwashing vacuum and pump frequency. The average flux of CSMBR during a steady
periodic state of 24 d (576 h) was 5.87 L/h with average hydraulic residence time (HRT) of 5.97 h and that of PSMBR during a steady
periodic state of 30 d (720 h) was 5.85 L/h with average HRT of 5.99 h. The average total chemical oxygen demand (COD) removal
efficiency of CSMBR was 89.29% with average organic loading rate (OLR) at 4.16 kg COD/(m3·d) while that of PSMBR was 89.79%
with average OLR at 5.50 kg COD/(m3·d). COD concentration in the effluent of both SMBRs achieved the second level of the general
wastewater effluent standard GB8978-1996 for the raw medicine material industry (300 mg/L). Hence, SMBR with electronic control
backwashing was a viable process for medium- strength Chinese traditional medicine wastewater treatment. Moreover, the increasing
rates of preestablished backwashing vacuum, pump frequency, and vacuum and flux loss caused by mixed liquor in PSMBR all lagged
compared to those in CSMBR; thus the actual operating time of the PSMBR system without membrane cleaning was extended by up
to 1.25 times in contrast with the CSMBR system, and the average total COD removal efficiency of PSMBR was enhanced with higher
average OLR.
Key words: electronic control backwashing; powdered activated carbon; membrane bioreactor; aerobic process; wastewater treatment
Introduction
Membrane bioreactor (MBR) is an improvement on
the activated sludge process in which solid separation is
achieved without the requirement of a secondary clarifier,
and it is now widely used for municipal and industrial
wastewater treatment (Benitez et al., 1995; Buisson et
al., 1998; Cote et al., 1998; Cornelissen et al., 2002;
Rosenberger et al., 2002). Although MBRs offer many
advantages over conventional processes, such as small
footprint and better effluent quality, membrane fouling
remains a major drawback. Membrane fouling is mainly
associated with the deposition of a filter cake or fouling
layer onto the membrane surface, thus limiting the perme-
ate flux. Fouling leads to frequent cleaning or replacement
of membranes, which in turn increases operating costs
(Gander et al., 2000).
Various methods have been adopted to control fouling
Project supported by the Hi-Tech Research and Development Program
(863) of China (No. 2002AA601310). *Corresponding author.
E-mail: rnq@hit.edu.cn.
during the operational cycle of the MBR process. Since the
bubbles generated by aeration are essential for suppressing
the build-up of the cake, most submerged MBRs (SMBRs)
adopt a configuration allowing the membrane surface to
contact intimately with the air bubbles, which then induce
a moderate shear stress (Benitez et al., 1995; Wen et al.,
1999; Hong et al., 2002; Chua et al., 2002). Periodic back-
washing improves membrane permeability and reduces
fouling, thus leading to optimal, stable hydraulic operating
conditions (Bouhabila et al., 2001; Albasi et al., 2002).
Adding powdered activated carbon (PAC) to a SMBR with
intermittent suction reduces membrane fouling effectively
in a long-term operation, and operating intervals can be
extended about 1.8 times compared to that without PAC
(Li et al., 2005).
This study focused on an understanding of the feasibility
of operating SMBR with an electronic control backwash-
ing on Chinese traditional medicine wastewater treatment,
which is difficult to be treated by aerobic biodegradation
because it is rich in the easy-to-foam medium that leads
to solid loss and because of the presence of molecules

Page 2

1038 LIU Xiao-lei et al.Vol. 19
with large molecular weight that are hard to biodegrade.
The effect of adding PAC was also explored by comparing
filtration characteristics of the SMBRs with and without
PAC, in bulk under the same feed.
1 Materials and methods
1.1 Membrane and bioreactor
Experimentswere carried out with twoidentical SMBRs
equipped with same electronic control backwashing device
operating in parallel. Both were set to the same effective
volume with the same feed tank and the same water-
level equilibrium box. One SMBR was used as the control
(control SMBR, CSMBR) while the other was dosed with
PAC (PAC-amended SMBR, PSMBR). Bench-scale rigs
(Fig.1) of each SMBR comprised a rectangular tank of
500 mm × 200 mm × 500 mm having an effective volume
of 35 L, with a vertical-mounted submerged hollow fiber
membrane module (Hangzhou Zheda Hyflux Hualu Mem-
brane Tec. Co., Ltd., China). The membrane was made of
polypropylenewithaporesizeof0.1µm,anditsmolecular
weight cut-off was approximately 80 kDa. Each membrane
module had a filtration area of 1 m2. Both membrane
modules had been used for several months prior to this
study, they were thoroughly washed and backwashed in
situ with a 1.5% (w/v) NaClO solution until results from
the tap water trials showed that the two SMBR systems
had almost the same initial filtration resistance before the
operation commenced.
1.2 Sewage
Settled sewage was a dilution of the Chinese tradi-
tional medicine industrial wastewater collected directly
from the product line in the Harbin No.2 Chinese Tra-
ditional Medicine Plant. The wastewater was rich in
Fig. 1 Schematic diagram of the experimental device. (1) highly placed
feed tank; (2) gate valve; (3) float-controlled valve; (4) water-level
equilibrium box; (5) rewound valve; (6) bioreactor; (7) hollow-fiber
membrane module; (8) triggered vacuum gauge; (9) normally open
solenoid valve; (10) pump; (11) backwashing tank; (12) normally closed
solenoid valve; (13) pressure gauge; (14) air compressor; (15) air flow
meter; (16) diffuser; (17) electric heater.
macromolecule that was difficult to be treated and easy to
foam in aerobic biodegradation, such as Glycosides, so it
was diluted with tap water before feeding to the reactors in
view of the fact that the aerobic SMBRs would be foaming
sharply and avoid overloading with strong feed. Chemical
oxygen demand (COD) concentration in the wastewater
ranged from 143682.2 to 586479.8 mg/L, and was diluted
to 574.8–3201.3 mg/L as required in feed. pH value of
the wastewater ranged from 6.75 to 7.54 and the dilution
showed a relatively stable neutral pH without regard to
dilution rates. BOD/COD fluctuated between 0.32 and
0.37 in both the wastewater and the dilution, while the
COD:TN:TP remained about 258:3:1.
1.3 Seed sludge
The reactors were seeded with the activated sludge
collected from the return-sludge line of the conventional
activated sludge process in the Harbin No. 2 Chinese
Traditional Medicine Plant. It was introduced evenly into
PSMBR and CSMBR. Each 35 L SMBR contained 12.5
L sludge and the remaining parts were filled with tap
water. An electric heater inside the reactor was opened,
to maintain the bioreactor temperature at 25±1°C. An air-
diffuser was placed under the membrane modules so that
an uplifting two-phase flow of bubbling air and mixed
liquor could remove the fouling layer formed on the
membrane. In order to abate solids loss caused by foam,
airflow was controlled using a flow meter (Model LZB,
Tianjin Wuhuan, China), with an airflow rate fixed at
approximately 12–15 L/min. PAC was sieved (100–120
mesh) and rinsed several times to remove impurities. After
seeding the reactors, 87.5 g PAC was added to PSMBR
immediately; hence the initial PAC concentration was
2500 mg/L. The reactors were then allowed to stabilize
for 6 h without further modification before starting the
experiments.
1.4 Suction and backwashing
The electric switch box used was composed of two dig-
ital display-timing relays, two microminiaturized counter-
type electronic timer totalizers, and eight switches.
Suction flow is indicated with solid arrowheads in
Fig.1. Constant-flux filtration was carried out using an
electromagnetic metering pump (ES-B30, Iwaki, Japan) on
the permeate stream. The reading of the triggered vacuum
gauge mounted up while membrane fouling got worse. A
connection signal was send out when the reading pointer
touched the upper limit, which pointed at the preestab-
lished backwashing vacuum, and then the normally open
solenoid valves closed while the normally closed solenoid
valves opened, hence the pipeline was switched from
suction flow to backwashing flow. Counter-type electronic
timer stopped at the same time and the reading showed the
time of one suction period.
Hollow arrowheads in Fig.1 show the flow of backwash-
ing. Backwashing was completed using effluent stored
in a backwashing tank. The backwashing interval was
controlled by the display-timing relay, which sent out a
connection signal at the end of backwashing. Then the

Page 3

No. 9Effect of powdered activated carbon on Chinese traditional medicine wastewater treatment······
normally open solenoid valves opened while the normally
closed solenoid valves closed. Thus the pipeline was
switched from backwashing flow to suction flow, and the
counter-type electronic timer restarted to count time once
again.
Bubbling is only partially efficient, as bubbles are
capable of limiting particle deposition and polarization
phenomena, but not internal fouling. For given conditions
of aeration, periodic backwashing gave an additional ef-
ficiency by decreasing internal fouling (Bouhabila et al.,
2001; Albasi et al., 2002). Pump frequency and aeration
rate was not increased but remained unchanged during
backwashing considering economy and facilitation. Thus
membrane fouling was lightened by the cooperation of
aeration and backwashing, which was an optimal way to
reduce fouling. The backwashing interval was 5 min.
1039
1.5 Sampling
Three samples were collected everyday, namely: in-
fluent, supernatant, and effluent. The supernatant was
obtained by filtering a sample of mixed liquor through a
filter paper of 1.6 µm. The role of the supernatant was
to distinguish between microbial removal that occurred
within the activated sludge itself, and removal that was at-
tributable solely to the membrane (which includes fouling
layer). Thus, the difference in microbial concentration be-
tween the influent and the supernatant represents microbial
removal by activated sludge, while the difference between
the supernatant and the effluent represents microbial re-
moval by membrane. Group data that comprised pump
frequency, flux, vacuum, preestablished backwashing vac-
uum and reading of counter-type electronic timer recorded
at 10 min after a suction period began, as the suction period
was going to end, and at 10 min after the next suction
period began, was collected twice a day at different time.
The data collection interval between CSMBR and PSMBR
was less than 30 min.
No wasting of biomass took place except for sampling,
thereby giving an infinite solid retention time (SRT). Dis-
solved oxygen (DO) concentration in the both SMBRs was
maintained at 2.54–3.90 mg/L. Microscopic observation
confirmed that there was no bulking sludge rudiment
caused by filamentous organisms in both SMBRs through-
out the trial.
1.6 Analytical items and methods
All items on the quality of the influent, supernatant,
and effluent, together with the mixed liquor suspend-
ed solids (MLSS) and mixed liquor volatile suspended
solids (MLVSS) were measured according to the standard
methods (American Public Health Association, 1995).
Bacteria in the bioreactors were observed with a light-level
microscope (CX31PTSF, Olympus, Philippines). DO and
the temperature in the bioreactors were measured with a
DO meter (Oxi 330i, WTW, Germany). Scanning electron
microphotographs of the fouled membrane fiber were
taken by a scanning electron microscope (JSM-840, JEOL,
Japan) after preparation following the standard procedure.
2 Results and discussion
2.1 Flux, pump frequency, and preestablished back-
washing vacuum
Membrane was fouled in bioreactor by the accumulation
of mixed liquor components on the internal and external
structures of the membrane. By means of optimizing
the control of backwashing interval, preestablished back-
washing vacuum and pump frequency, flux and hydraulic
residence time (HRT) could be fluctuated within a range,
and thus fouling could be remedied by the cooperation of
backwashing and aeration. As time went by, the greater
the membrane fouling, the greater the readings of the
triggered vacuum gauge, and the smaller the flux. Hence,
suction period became shorter and backwashing frequent,
as preestablished backwashing vacuum and pump frequen-
cy remained stable. Thus the total effluent quantity fell
quickly, which was definitely uneconomical. Therefore it
was necessary for preestablished backwashing vacuum and
pump frequency to be adjusted in time according to the
variation of flux and vacuum. Owing to the wastewater
characteristics, long HRT was required to keep low organic
loading rate (OLR). One suction period was about 90
min by adjusting preestablished backwashing vacuum and
pump frequency. The average flux of CSMBR during a
steady periodic state of 24 d (576 h) was 5.87 L/h and
that of PSMBR during a steady periodic state of 30 d (720
h) was 5.85 L/h (Fig.2a). Therefore, the average HRT of
CSMBR was 5.97 h, while the average HRT of PSMBR
was 5.99 h.
Fig. 2 Flux (a), pump frequency (b) and preestablished backwashing vacuum (c) in PSMBR and CSMBR.

Page 4

1040LIU Xiao-lei et al.Vol. 19
Because of long-term contact, with almost zero wastage
except for sampling, the combination of deposited foulants
and membrane material became more consolidated, which
led to a progressive blocking of membrane pores, thus
reducing the effective filtration area. Moreover, a loss of
effective filtration area by fouling would increase local
flux in some regions of the membrane, which might reach
critical local filtration conditions. A rapid deposition of
particles and colloidal aggregation on the membrane led
to the formation of a cake layer and a sharp increase
in preestablished backwashing vacuum. Finally, the cake
layer became thicker and more compact with long-term
operation, leading to a rapid increase in hydraulic filtration
resistance. In this case, the cake is the real filtering barrier,
providing a hydraulic filtration resistance, which is much
higher than that of the membrane. On day 25, flux of
CSMBR descended sharply even as pump frequency rose
to 100%, which meant the membrane inside needed to
be regenerated, so CSMBR stopped with preestablished
backwashing vacuum at –48 kPa. However, PSMBR con-
tinued operating for the next 6 d and stopped on day 31
with preestablished backwashing vacuum at –50 kPa for
the same reason. The actual operating time of the PSMBR
system without membrane cleaning was extended by up
to 1.25 times in contrast with the CSMBR system under
long-term operation conditions. And both preestablished
backwashing vacuum and pump frequency of CSMBR
were higher than those of PSMBR (Figs.2b and 2c)) during
the first 25 d. This indicated that adding PAC into the
SMBR could reduce the accumulation of foulants on the
membrane surface and prevent the reduction of permeate
flux.
2.2 COD removal effect
COD removal effect both in CSMBR and in PSMBR
are shown in Fig.3. After the stopping of CSMBR on day
25, COD concentration in influent of PSMBR kept on
increasing till the end. The average total COD removal
efficiency of CSMBR was 89.29% with average OLR at
4.16 kg COD/(m3·d) while that of PSMBR was 89.79%
with average OLR at 5.50 kg COD/(m3·d). COD concen-
trationineffluentofbothSMBRsachievedthesecondlevel
of the general wastewater effluent standard GB8978-1996
for the raw medicine material industry (300 mg/L). Hence,
SMBR with electronic control backwashing was a viable
process for medium-strength Chinese traditional medicine
wastewater treatment. Moreover, addition of PAC could
prolong the operation time and enhance the average total
COD removal efficiency. However, to reduce COD level
in effluent, combined anaerobic treatment processes before
aerobic SMBR with electronic control backwashing was
needed in the case of high-strength Chinese traditional
medicine wastewater treatment.
Fouling through the formation of a dynamic layer at
the membrane surface might be expected to reach equi-
librium once the adhesive forces between the layer and
the membrane substrate are balanced by the shear forces
at the layer-solution interface. When the dynamic fouling
layer was thick enough to format a biofilm, the COD
removal efficiency of membrane was strengthened, but
membrane fouling became worse. Then, the thicker the
layer became, the higher the COD removal efficiency of
membrane achieved. In CSMBR, COD removal efficiency
of bioreactor was lower than that in PSMBR, while COD
removal efficiency of membrane was higher. This indicated
that PAC served as a filter to reduce foulants in the bulk
solution by adsorption and flocculation in PSMBR, which
decreased the foulants loading the membrane surface and
alleviated membrane fouling by reducing the thickness of
the cake layer.
2.3 MLSS, MLVSS, and VSS/SS
As there was no wastage of biomass except in sampling,
similar rising trends in MLSS and MLVSS concentration
were observed both in CSMBR and in PSMBR (Fig.4).
Although both MLSS and MLVSS grew more and more
slowly, MLSS grew faster than MLVSS, while VSS/SS
in CSMBR descended on the whole during the operation.
The addition of PAC made VSS/SS in PSMBR much
lower than that in CSMBR. MLVSS concentrations of both
CSMBR and PSMBR were almost the same during the
period of the first 25 d. Because the PAC distribution in
Fig. 3 COD variation (a) and removal efficiency (b) in PSMBR and CSMBR.

Page 5

No. 9Effect of powdered activated carbon on Chinese traditional medicine wastewater treatment······
1041
Fig. 4 MLSS, MLVSS, and VSS/SS variations in PSMBR and CSMBR.
PSMBR was so uneven that some PAC might deposit in/at
the corners in the bioreactor even after aeration, MLSS
concentration in PSMBR was higher than that in CSMBR,
and ranged from 1135 to 1655 mg/L during operation,
although the initial addition concentration of PAC was
2500 mg/L. However, this did not disturb the research to
contrast PSMBR with CSMBR. For a more stable long-
term operation, PAC would be added to the bioreactor
intermittently and not once at the very beginning. Sludge
discharge was proposed, and fresh PAC with an amount
equal to that lost from the discharged sludge would be
supplied sporadically to keep a stable PAC concentration
in the bioreactor.
2.4 Vacuum and flux loss caused by mixed liquor
Results from the tap water trials showed that both
the relationship between tap water vacuum (z) and pump
frequency (x) and the relationship between tap water flux
(y) and pump frequency (x) were linear with the adjusted
R-square statistic obtained on the fittings, 0.99 in both
PSMBR (z = –13.286x – 0.6238, y = 24.974x + 0.4762)
and CSMBR (z = –11.686x – 0.1571, y = 24.004x +
0.5217). To a certain extent, tap water vacuum and flux
were both in direct ratio with pump frequency. Slight
differences between them illustrated the dissimilarity of
head loss from pipeline in both PSMBR and CSMBR
system during a suction period.
Permeability of SMBR in long-term operation was al-
ways found to be lower than that attained for tap water,
because of the existence of biomass. Therefore the follow-
ing terms were defined in this paper. For a certain pump
frequency, the readings of the triggered vacuum gauge
minus the tap water vacuum gained from the relationship
between tap water vacuum (z) and pump frequency (x) was
the vacuum loss caused by mixed liquor, which was caused
by the mixed liquor without regard to effects of pipeline.
Similarly, for a certain pump frequency, the flux minus the
tap water flux gained from the relationship between tap
water flux (y) and pump frequency (x) was the flux loss
caused by mixed liquor. Vacuum and flux loss caused by
mixed liquor increased during operation (Fig.5) in both
PSMBR and CSMBR, and vacuum and flux loss caused
by mixed liquor in CSMBR was higher than those in
PSMBR during the first 25 d. Since both SMBRs had
same feed and similar MLSS and MLVSS concentration,
the difference in vacuum and flux loss was possibly due
to other factors such as EPS concentration, colloid sizes,
molecular forces, etc. among these substances. In general,
activated sludge in CSMBR consists of weak flocs, which
break easily in a turbulent environment. The destruction
of flocs could promote the release of colloidal and some
soluble components such as EPS and soluble microbial
products, from the inner side of microflocs to the bulk
solution, causing an increase in the viscosity of the mixed
liquor. PAC can serve as an adsorbent and coagulant,
leading to continuous depletion of fine colloids and EPS
in the bulk phase by adsorption and flocculation. Li et al.
(2005) has reported that adding of PAC could lower the
mean apparent viscosity of the SMBR system by nearly
45%. This result in a decrease in growth of vacuum and
flux loss caused by mixed liquor in PSMBR.
Fig. 5 Flux loss (a) and vacuum (b) caused by mixed liquor in bioreactor of PSMBR and CSMBR.