ARTICLE IN PRESS

Water Research 39 (2005) 42994308

www.elsevier.com/locate/watres

Conventional and thermophilic aerobic treatability

of high strength oily pet food wastewater using
membrane-coupled bioreactors
R. Kuriana, C. Acharyab, G. Nakhlab,, A. Bassib
a
Department of Civil & Environmental Engineering, University of Western Ontario, Canada
b
Department of Chemical & Biochemical Engineering, University of Western Ontario London, On, N6A 5B9, Canada
Received 24 March 2005; received in revised form 21 July 2005; accepted 16 August 2005

Abstract

Although thermophilic treatment systems have recently gained considerable interest, limited information exists on the
comparative performances of membrane-coupled bioreactors (MBR) at thermophilic and conventional conditions. In
this study aerobic MBRs operating at room temperature (20 1C) and at lower thermophilic range (45 1C) were
investigated for the treatment of dissolved air otation (DAF) pretreated pet food wastewater. The particular
wastewater is characterized by oil and grease (O & G) concentrations as high as 6 g/L, COD of 51 g/L, BOD of 16 g/L
and volatile fatty acid (VFA) of 8.3 g/L. The performances of the two systems in terms of COD, BOD and O & G
removal at varying hydraulic retention time (HRT) are compared. COD removal efciencies in the thermophilic MBR
varied from 75% to 98% and remained constant at 94% in the conventional MBR. The O & G removal efciencies
were 6686% and 98% in the thermophilic and conventional MBR, respectively. Interestingly, high concentrations of
VFA were recorded, equivalent to 5073% of total COD, in the thermophilic MBR efuent. The observed yield in the
thermophilic MBR was 40% of that observed in the conventional MBR.
r 2005 Elsevier Ltd. All rights reserved.

Keywords: Thermophilic; MBR; Removal efciency; High strength wastewater; Oily wastes

1. Introduction proven UASB and EGSB treating wastewaters rich in

Biological treatment systems have been developed to
treat high strength wastewater successfully (Driessen
and Speert, 1999; Zhang et al., 2003; Rajbhandari and
Annachhatre, 2004). Economic benets related to
anaerobic treatment processes have made it a prefer-
ential treatment option for high strength wastewaters
(Tchobanoglous et al., 2003; Kumar and Bachman,
2000). However, Hwu (2001) has documented that
washout/otation-related failures occur in the well-
Corresponding author. Fax: 1 519 850 2921.
E-mail address: gnakhla@eng.uwo.ca (G. Nakhla).
long-chain fatty acids (LCFA) and lipids. Moreover, the
anaerobic digester designed to treat this particular
wastewater failed to produce acceptable efuent quality
(Acharya et al., 2003). It must be asserted that the high
O & G in this rendering waste follows efcient two-stage
oil cooking and recovery operations and a DAF unit.
Thermophilic treatment systems have gained consid-
erable interest in recent years as an alternative biotech-
nological process for the treatment of high-strength and
high-temperature industrial wastewaters (LaPara et al.,
1999) due to high biodegradation rates and reduced
sludge generation (Rozich and Bordacs, 2002). Limita-
tions in its practicality exist due to the reduced sludge

0043-1354/$ - see front matter r 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.watres.2005.08.030
 ARTICLE IN PRESS
4300 R. Kurian et al. / Water Research 39 (2005) 42994308
settleability as documented by C - etin and Surucu (1990),
Barr et al. (1996), Tripathi and Grant Allen (1999) and
Lapara and Alleman (1999) with increasing tempera-
ture. At the low biomass yields observed under
thermophilic conditions, the washout of biomass due
to poor settlability is a critical issue, affecting process
stability. MBR technology becomes particularly attrac-
tive when the retention of solids for extended periods of
time is a critical issue (Chen et al., 2003). MBR
technology has been widely used to treat industrial
wastewaters (Bouhabila et al., 2001; Scott and Smith,
1997) at mesophilic or room temperatures. But few
researchers have explored the potential of a thermophilic
MBR (Lopetegui and Sancho, 2003; Joore et al., 2002;
Lapara et al., 2001).
Limited information exists on MBR systems treating
the same high strength wastewater at thermophilic
conditions and at typical MBR conditions or with no
temperature control. The novelty of this research project
lies primarily in the comparative side-by-side perfor-
mance assessment of the two MBR and its application in
treating animal fat-contaminated wastewaters. The
project also deals with a highly complex wastewater
and the results presented here contradict existing
literature available on thermophilic treatment systems
with respect to the yield and impact of food to
microorganism ratio (F/M). The failure of the conven-
tional anaerobic digester to treat this wastewater makes
this research an interesting and potential solution to the
problem. In this paper, the authors compare the
performance of two MBR systems operated at thermo-
philic range (45 1C) and room temperature (1820 1C) to
treat high strength rendering wastewater.
2. Experimental conditions
Laboratory-scale MBRs were operated under varying
hydraulic retention time (HRT) and temperature condi-
tions. The wastewater treated in the experiment was
Table 1
Raw and DAF efuent characteristics
Parameters (mg/L) Raw wastewater
Range
TSS 17,31161,700
VSS 15,18059,800
TCOD 74,9251,54,100
SCOD 13,12518,450
TBOD5 80,000
VFA
Ammonia-N 7971400
Phosphate 500830
O&G 38,800
generated from a rendering facility in St. Marys,
Ontario, producing intermediates for pet food from
byproducts of meat processing industry. The wastewater
is characterized by particularly high concentrations of O
& G, which would be detrimental to any biological
treatment system. The wastewater generated at this
facility undergoes physical separation of solids and O &
G using a DAF. The high variability in the wastewater
generated is depicted in Table 1. In the experiments
conducted in this study the DAF efuent was employed.
2.1. Experimental setup
2.1.1. Thermophilic MBR
The system setup, as illustrated in Fig. 1(a), consisted
of a 0.5 L intermediate storage tank followed by a 3.5 L
glass reactor. Thermophilic seed sludge was obtained
from an existing thermophilic treatment facility operat-
ing at 40 1C. The reactor temperature was maintained at
4573 1C using a Corning hotplate stirrer. The reactor
was aerated continuously with compressed air through
air diffusers. A peristaltic pump (Minipuls3, Gilson Inc.,
Canada) was used to pump inuent continuously into
the reactor and mixed liquor through an external cross
ow membrane, with the retentate recycled back to the
reactor and the permeate collected. The membrane
employed was a UMP-153, Pall Biopharmaceuticals,
with a pore size of 0.2 mm and surface area of 0.08 m2.
The particular membrane was selected owing to its
ability to withstand high temperatures (maximum of
80 1C) as compared to that used in the conventional
MBR (maximum of 40 1C).
The system was operated for 158 days: at a 7-day
HRT and SRT of 35 days for 75 days (days 075) and at
5-day HRT and SRT of 30 days for 83days (days
76158). The SRT was calculated based on wastage of
sludge incurred due to the washing of the membrane
rather than deliberate wastage, as this was deemed
unnecessary. The membrane was disconnected and
cleaned by backwashing using tap water, reversing the
DAF efuent
Average Range Average
36,857 115012,866 4651
34,383 110010,966 4571
96,660 10,02550,500 21,000
16,757 686019,980 12,300
80,000 800016,000 14,100
46098390 6471
1060 6401440 1030
665 3401059 626
38,800 6006000 1230
 ARTICLE IN PRESS
R. Kurian et al. / Water Research 39 (2005) 42994308
Fig. 1. Experimental setup (a) thermophilic MBR; (b) conventional MBR.
inuent and efuent ports and nally washing with 1%
nitric acid solution as recommended by vendor. The
cleaning procedure took 1520 min per washing, there-
fore, did not signicantly interrupt the continuous
operation of the MBR.
2.1.2. Conventional MBR
The experimental setup, as illustrated in Fig. 1(b),
consisting of a 150 L storage tank followed by a 27 L
activated sludge reactor, was started with sludge from a
local municipal treatment plant. The reactor, operating
at room temperature of 1872 1C, was aerated with
compressed air through air diffusers. The ZW-1
membrane module from Zenon Environmental, Oak-
ville, Ont, with a pore size of 0.04 mm and surface area of
0.047 m2 was employed to retain reactor solids. A
peristaltic pump (Minipuls3, Gilson Inc., Canada) was
used to draw permeate through the immersed membrane
unit and pump inuent into the reactor. The storage
tank was replenished with DAF efuent every 57 days.
The system operation continued over 105 days: at 6.3-
day HRT for 85 days (day 085) and at 5-day HRT for
20 days (days 86105). The SRT was calculated as 290
days based on sludge wastage due to sampling and
membrane cleaning.
2.2. Analytical methods
The two systems were closely monitored throughout
the operating period. Inuent, reactor-mixed liquor and
membranes permeate (nal efuent) were collected 23
times per week for analytical purposes. Parameters
analyzed included total suspended solids (TSS), volatile
suspended solids (VSS), COD, BOD5, ammonia-N,
nitrate-N, nitrite-N and O & G. Inuent, reactor-mixed
liquor and nal efuent samples were analyzed for
BOD5 using Standard methods (APHA, 1998). Particle
size distribution in the thermophilic MBR, determined
using Masterizer 2000 (Malvern Instruments Ltd., UK),
4301
revealed that less than 1.5% of the suspended solids
were ner than 1.2 mm, accordingly TSS and VSS were
measured using the Standard method 1.2 mm lter paper
(APHA, 1998). Samples were dried at 105 1C till
constant weight was attained, weighed, and then heated
to 550 1C for 15 min and weighed again. Total and
soluble COD were determined using the HACH Odyssey
Analyzer and the heating reactor with standard HACH
testing kits. Ammonia, nitrate and nitrite were deter-
mined using a high-performance liquid chromatograph
(HPLC, Dionex Canada Ltd., Oakville, Ont). Soluble
samples were extracted by ltering through a 0.45 mm
lter paper (Wheaton). Dissolved oxygen (DO) was
measured with a portable YSI Dissolved Oxygen Meter
Model 50. The VFA concentration was analyzed by a
gas chromatograph (Varian CP 3800) with ame
ionization detector (FID) equipped with a fused silica
column (30 m
0.32 mm
5 mm). Nitrogen was used as
carrier gas at 5 mL/min. detector and injector tempera-
ture were set at 200 1C while the oven was programmed
to 110165 1C at a ramp of 20 1C/min for 9.25 min.
3. Results and discussion
3.1. Contaminant removal
The two systems operated as completely stirred
bioreactors with membrane modules serving to com-
pletely retain biomass and are compared on the basis of
HRT, though the reactors were of different volumes.
The inuent and efuent quality at steady state for each
HRT for thermophilic conditions and conventional
MBR conditions is summarized in Table 2. It is
noteworthy that despite the short 20-day operation of
the conventional MBR at 5-day HRT steady state was
attained as reected by the low-standard deviations
(Table 2), particularly considering the inuent varia-
bility.
 ARTICLE IN PRESS
4302 R. Kurian et al. / Water Research 39 (2005) 42994308

3.1.1. Suspended solids

94.2712.5 (7)

62.3710.5 (6)
97.7715.3 (5)
The nal efuent contained only soluble contaminants

99.773.5 (7)
% removal
(Table 2) in both systems. Membranes facilitated
biomass retention and sludge age control as reected

100
by biomass concentrations as high as 15 g VSS/L in the

conventional MBR system and as high as 2.8 g VSS/L in
the thermophilic MBR. The steady-state biomass con-
11437360 (7)

451730 (6)
58780 (5)
centrations in the two MBRs (Table 3) reveal that in the
2575(7) conventional MBR, reactor biomass actually increased
5 days

with the reduction in HRT as expected (given the

Based on average inuent O & G value of 608372480 (6) mg/l during 5-day HRT run (numbers within parenthesis reect number of samples.).

extremely long SRT for both the runs), but decreased in
0

the thermophilic reactor. There are primarily two

reasons for this drop in the thermophilic reactor biomass
Conventional MBR nal efuent

59.4710.2(8)
99.773.2 (8)
94.4712 (8)

with the increase in inuent wastewater ows and

% removal

loadings. The COD removal efciency in fact plum-

meted to 65%, with an efuent concentration of 6133 mg
100

COD/L. On average, at steady-state 7477 mg/L COD

was removed at the 5-day HRT compared with
11177440 (8)

4877150 (8)

18411 mg/L at the 7-day HRT. Additionally due to

2976 (8)

frequent membrane fouling, the estimated loss of

6.3 days

biomass through cleaning may have been higher than

actual, i.e. the reduction in SRT from 35 days is much


0

more than estimated.

8676.2a (6)
6573.7 (5)
7874.6 (5)
5171.2 (6)

71.6712 (6)
% removal

3.1.2. Overall COD removal

3.1.2.1. Thermophilic MBR at 7-day HRT; conventional
MBR at 6.3-day HRT. Although the HRTs in the
100

thermophilic and conventional MBR considered are not

exactly the same, they are deemed close enough to
363744.7 (6)
61337126 (5)

warrant fair comparative assessment of the two systems

21957345(5)

685778 (6)
596769 (6)

from a practical standpoint. A remarkably high-COD

removal efciency of 96% (Table 2) was achieved in the
5 days

thermophilic MBR system at 7-day HRT. The efuent

0

characteristics showed minimal changes considering the

extreme variations in the inuent characteristics, reect-
Steady-state performance of thermophilic and conventional MBR

95.571.7 (11)
97.770.8 (11)
69.371.0 (10)
Thermophilic MBR nal efuent

78.7710 (10)
93.871.0 (8)

ing excellent system robustness. The COD removal

% removal

efciency in the conventional MBR was 94% corre-

sponding to efuent of 11177114 mg COD/L (Table 2)
100

more than twice the thermophilic efuent of 516740 mg

COD/L.
516740 (11)
148720 (11)
197760 (10)

296717 (10)
6675 (8)

3.1.2.2. Thermophilic and conventional MBR at 5-day

HRT. While efuent COD was not affected by the
7 days

reduction of HRT from 6.3 to 5 days in the conventional

0

MBR it increased from 516 to 6133 mg/L in the

340375490 (39)
1991079183 (39)
971972725 (39)

289373000 (10)
11997351 (39)

18177548 (28)

thermophilic MBR. The deterioration in efuent quality

in the thermophilic MBR may be due, in part, to the
deterioration in membrane ux during this period,
Average
Inuent

requiring increased cleaning which resulted in an

estimated SRT reduction from 35 to 30 days. The
deterioration in efuent quality can also be attributed to
the insufcient contact time in the system. The higher
Parameters

solubility led to escape of contaminants through the

Table 2

NH4N
TBOD5

O&G
(mg/L)

TCOD

membrane in the thermophilic MBR, as apparent from

TKN
TSS

the efuent O & G data (Table 2). The data given in

 ARTICLE IN PRESS
R. Kurian et al. / Water Research 39 (2005) 42994308 4303

Table 2 further corroborates the kinetic limitation and substantiates that increased solubility in and by itself
short contact time alluded to above. would not result in better O & G removal, as evidenced
Thermophilic treatment is reported to enhance O & G by the O & G removal in the conventional and
or lipid removal efciencies (Becker et al., 1997; Li thermophilic MBR at 20 1C and 45 1C, respectively.
et al., 2002). However, the data of this study clearly
3.1.3. Contaminant removal across membrane
It is clear from Fig. 2, depicting the relation between
Table 3 permeate and reactor sCOD, that 28% and 32% of
TSS and VSS maintained in the MBRs sCOD removal occurred by ltration, in thermophilic
TSS (mg/L) VSS (mg/L)
and conventional MBRs, respectively. Given the varia-
bility of the data, the contribution of physical ltration
Thermophilic MBR 31457106 (12) 27977100 (12) to sCOD removal in both reactors is deemed to be the
HRT 7day same. Since the overall COD removals were over 90%
Conventional MBR 85277814 (10) 79757589 (10) (except in thermophilic at 5-day HRT) the contribution
HRT 6.3day of membrane ltration to overall removal of sCOD is of
Thermophilic MBR 17667443 (7) 15687423 (7) the order 2%. The reduced physical COD removal
HRT 5day
observed in the thermophilic MBR can be attributed to
Conventional MBR 20,74274282 (6) 14,72573742 (6)
the coarser membrane (pore size 0.2 mm) employed in
HRT 5day
comparison to the membrane (pore size 0.04 mm) used in
(Numbers within parenthesis reect number of samples.). the conventional MBR. Furthermore, the higher solu-
bility of contaminants at 45 1C contributed to the escape
of contaminants through the membrane pores in the
14000 Conventional MBR thermophilic MBR.
y = 0.72x Thermophilic MBR
Membrane permeate COD (mg/L)

12000 3.1.4. BOD5 removal

R2 Thermophilic = 0.92
The lower BOD5 removal efciency of 98% (Table 2)
10000 observed in the thermophilic MBR at 7-day HRT
relative to the conventional MBR may be attributed to
8000
the high fraction of VFA (Table 4) contributing to the
6000 COD in the efuent. At an HRT of 7 days, VFAs
accounted for almost 50% of the total efuent COD
4000
y = 0.68x
with acetic acid contributing to 34% of VFAs. At the 5-
day HRT the deterioration in efuent quality of the
2000 R2conventional = 0.89
thermophilic MBR is clearly reected by a sharp
increase in efuent acetic, propionic, butyric, valeric
0
and isovaleric acid concentrations. VFAs contributed to
0 5000 10000 15000 20000
Reactor sCOD (mg/L)
73% of the efuent COD, corresponding to a concen-
tration of 4461 mg COD/L, with acetic acid alone
Fig. 2. COD removal across membrane. contributing 2388 mg COD/L. The extremely high

Table 4
Data on VFA in efuent and inuent of thermophilic MBR

Inuent HRT 7 days HRT 5 days

Steady-state data

Acetic acid (mg/L) 19897944 (9) 5974 (4) 2388791 (5)

Propionic acid (mg/L) 21767311 (9) 129713 (4) 417739 (5)
Isobutyric acid (mg/L) 83785 (9) 070 (4) 573 (5)
Butyric acid (mg/L) 8917360 (9) 070 (4) 231799 (5)
Valeric acid (mg/L) 9747488 (9) 070 (4) 315722 (5)
Iso valeric acid (mg/L) 4987323 (9) 070 (4) 135719 (5)
COD equivalent of VFA (mg/L) 12,11878569 (9) 221722 (4) 44617246 (5)
Total COD (mg/L) 17,75277839 (9) 516741 (4) 61327126 (5)
Percentage COD equivalent of VFA of total COD (%) 49.970.04 (4) 73.170.03 (5)

(Numbers within parenthesis reect number of samples.).

 ARTICLE IN PRESS
4304 R. Kurian et al. / Water Research 39 (2005) 42994308

concentration of the readily biodegradable acetic acid in ammonia-N (20C)

the efuent indicates a lack of sufcient contact time for ammonia-N (45C)
proper degradation of contaminants at the 5-day HRT.
nitrate-N (20C)
The lower concentration of VFA measured in the
efuent during the 7-day HRT run corroborates this nitrate-N(45C)
conclusion. The high concentration of VFA in the 240 2.5
efuent is consistent with the literature. Chu (1995) has
demonstrated the increased accumulation of VFA in
autothermal thermophilic aerobic digesters (ATAD) 2
230
with a reduction in digestion time from 6 to 3 days

Ammonia-N (mg/L)

Nitrate-N (mg/L)
during the treatment of municipal sludges. Fothergill
and Mavinic (2000) have demonstrated the accumula- 1.5
tion of VFA in ATADs reaching as high as 757 mg/L at 220
digestion time of 3 days, clearly reecting incomplete
1
biodegradation.
In the conventional MBR 499% BOD removal
efciency was attained with VFAs contributing 21% of 210
0.5
the efuent COD during both runs. Acetic acid
contributed o5% of the efuent VFA in the conven-
tional MBR. The breakdown of VFAs in the conven- 200 0
tional MBR efuent is as follows: acetic acid: 572 mg/ 0 1 2 3 4
L, propionic acid: 2577 mg/L, isobutyric acid: Time (hr)
3279 mg/L, butyric acid: 2678 mg/L, valeric acid:
Fig. 3. Variation of NH4-N and NO3-N measured in batch
13 mg/L and isovaleric acid: 1773 mg/L. It is interesting experiments.
to contrast the persistence of higher VFA (propionic
through valeric acid) relative to acetic acid in the
conventional and thermophilic systems. Higher VFAs the reactors were operating at the same pH of 8.3, the
are known to be more slowly biodegradable aerobically different operational temperatures would affect different
than acetic acid, and thus their persistence in the ammonia stripping rates. Based on the reactor tempera-
conventional MBR is expected. Paradoxically, under tures and pH, the unionized ammonia would be 30%
thermophilic aerobic conditions, the higher VFAs are and 7% of the total ammonium nitrogen (TAN) in the
degraded while acetic acid and propionic acid are not. thermophilic and conventional MBR, respectively,
calculated based on Eq. (1) (Wyffels et al., 2003).
3.1.5. Ammonia removal 

In the thermophilic MBR ammonia removal efcien-
cies of 83% and 62% were observed during the HRT
runs of 7 and 5 days, respectively, corresponding to
average efuent concentrations of 197 and 363 mg/L
with minimal variation (Table 2). At the same time, the
removal efciencies in the conventional MBR for the 6.3
and 5-day HRT runs were 59% and 62%, respectively
(Table 2).
Batch nitrifying activity tests conducted on the
thermophilic MBR sludge, both at 45 1C and 20 1C
revealed no increase in nitrates or nitrites (Fig. 3). The
nitrites were measured to be zero throughout both
experiments. It must be asserted that during both tests at
20 1C and 45 1C, despite a drop of 10 mg/L of NH4-N
over the 4 h experimental period, the lack of increase in
nitrates indicates the minimal observed differences
between initial and nal concentrations are either due
to analytical error or due to abiotic losses. The lack of
nitrates and nitrites in both lab-scale continuous MBRs
conrms the absence of nitrication. Therefore, ammo-
nia removal mechanisms in the systems are cell synthesis
and stripping of free ammonia. Considering that both
NH3 -N=TAN 10pH =e6344=273 C
10pH . (1)
Taking into consideration the low yield observed in
the thermophilic MBR, it is understandable that a
considerable amount of the ammonia removal is due to
stripping of free ammonia. A nitrogen balance across
the thermophilic MBR was calculated. The ammonia
stripped was calculated using Eqs. (2) and (3) (Tchoba-
noglous et al., 2003).
ye H
C 0o =PT , (2)
log10 H A=T B, (3)
where ye is the moles of ammonia gas per mole of air; H
is the Henrys constant at 45 1C (atm); C 0o is the
concentration of unionized ammonia in the reactor
(mol NH4/mol liquid); PT is 1 atm; A, B are constants
for temperature correction and T is temperature (1K).
At the 7-day HRT steady-state period, ammonia-N
stripped constituted 66%, N removed as efuent 21%
and N removed in sludges was 13% of the inuent TKN
concentration of 1389 mg/L. These percentages corre-
spond to losses totaling to 459 mg N/d, 148 mg TKN/d
 ARTICLE IN PRESS
R. Kurian et al. / Water Research 39 (2005) 42994308 4305

and 78 mg TKN/d, respectively. Similarly, at the 5-day A sustained increase in the F/M, beyond 1.1 g COD/g
HRT the percentages were: ammonia-N stripped, 75%; VSS day, in the thermophilic MBR was followed by a
efuent N, 25%; wasted sludge, 1.4% of inuent TKN dip in the COD removal efciency after a lag period of
of 2200 mg/L. These percentages for the 5-day HRT one HRT. This was observed twice. From day 88 to day
correspond to mass rates of 843 mg N/d, 298 mg TKN/d 95 the F/M was sustained over 1.1 g COD/g VSS, the
and 31 mg TKN/d, respectively. It is obvious from these COD removal efciency dropped immediately after the
results that the major mechanism for nitrogen removal 88th day and fell below 80% by day 95. The trend was
in the thermophilic was stripping. The difference in repeated when F/M was sustained over 1.1 g COD/g
ammonia stripped at the two HRTs is due to the VSS from day 127158, during which COD removal
difference in the efuent ammonia concentrations (Table dropped from over 90% to less than 80%. Similar trends
2), given the airow rate was same, 3 Lpm, throughout were observed in the conventional MBR when the F/M
the study. Since the unionized ammonia in the conven- increased above 0.5 g COD/g VSS day as apparent from
tional MBR was only 7% of TAN (Table 2) compared Fig. 4(b) with COD removal efciencies dropping to less
to the 30% in the thermophilic MBR, at the than 85%.
400450 mg/L of TAN, ammonia stripping would
account for less than 2% of the inuent TKN loss and 3.2.2. Yield comparison
can therefore be neglected. There is a wide variability in thermophilic yields
documented. Research studies conducted on slaughter-
house efuent at 45 1C recorded yield of 0.35 g SS/g
3.2. Operational performance parameters CODremoved (Gariepy et al., 1989; Couillard and Zhu,
1993). Others have reported yields of 0.022 g SS/g
3.2.1. Foodto-microorganism ratio (F/M) CODremoved (Johnson and Hall, 1996) at 45 1C for
F/M was calculated based on inuent total COD, synthetic whitewater and 0.16 g SS/g CODremoved
HRT and VSS in the reactor. The variability in F/M (Berube and Hall, 2000) at 55 1C for synthetic kraft
during the study was due to variations in inuent pulp mill condensate. In this experiment, the yield was
contaminant and not due to changes in biomass calculated from a plot of cumulative VSS produced
concentrations in the reactors. Interestingly, at similar versus the cumulative COD consumed as shown in Figs.
contaminant loading, the thermophilic system showed 5(a) and (b) for the thermophilic and conventional
2.2 times higher biomass-specic COD removal ef- system, respectively. The yield calculations for the
ciency than the conventional MBR as illustrated in Figs. thermophilic biomass were based on the sludge wastage
4(a) and (b), possibly due to enhanced solubility of long incurred by membrane cleaning and were found to be
chain organics at higher temperature. This indicates a 0.08 g VSS/g COD. The TSS to VSS ratio in the
biokinetic advantage for thermophilic treatment sys- thermophilic MBR was 1.1470.1 throughout
tems, in this case, contrary to the ndings of Lapara et the experimental period. The yield of 0.08 g VSS/g
al. (2000). COD is in agreement with the 0.0570.03 g VSS/g COD

Removal Efficiency F/M Ratio COD removal efficiency F/M ratio

HRT = 7 d HRT = 5d
120 4
110 3.50
HRT 6.3 HRT 5 3.5
100
COD removal efficiency (%)

100 d d 3.00
COD removal efficiency (%)

3
F/M (gCOD/gVSS-d)
F/M (gCOD/gVSS-day)

2.50 80
90 2.5
2.00
60 2
80
1.50 1.5
70 40
1.00 1
60 20
0.50 0.5

50 0.00 0 0
0 50 100 0 50 100 150
(a) Time (days) (b) Time (days)

Fig. 4. Effect of F: M ratio on COD removal efciency (a) conventional MBR; (b) thermophilic MBR.
 ARTICLE IN PRESS
4306 R. Kurian et al. / Water Research 39 (2005) 42994308

y = 0.076x R2 = 0.9963
2
R = 0.9762 y = 0.186x+11.14
3 180
160
2.5
140
Cumulative VSS

Cumulative VSS
produced (g)

produced (g)
2 120
1.5 100
80
1 60
0.5 40
20
0 0
0 5 10 15 20 25 30 35 0 200 400 600 800 1000
(a) Cumulative
( COD consumed (g) b) Cumulative COD consumed (g)

Fig. 5. Yield calculation for biomass in (a) thermophilic MBR; (b) conventional MBR (Acharya et al., 2003).

documented by Suvilampi et al. (2003), in the treatment 0.9

of molasses from a sugar beet factory. The observed 0.8
yield in the conventional MBR was calculated to be
0.19 g VSS/g COD (Acharya et al., 2003). The thermo- Membrance flux (L/m2-hr) 0.7
philic yield was only 40% of that observed in the 0.6
conventional MBR, conrming that thermophilic sys-
tems are indeed a promising solution to the problem of 0.5
high sludge generation in conventional systems. The 0.4
reduced thermophilic yield observed here contradicts 0.3
results published by Lapara et al. (2000), in which
experiments were conducted in the temperature range 0.2
2565 1C with a gelatin and a-lactose growth medium, 0.1
and temperature was observed to affect minimal changes
0.0
on the observed yield.
0 50 100 150

3.2.3. Membrane flux Time (day)

The thermophilic MBR started operation at a
membrane ux of 0.8 L/m2 h. This is much lower than
the vendor recommended membrane ux of 100 L/m2 h
for the UMP-153 membrane module. Cleaning the
encapsulated membrane (UMP-153) was difcult, re-
sulting in a gradual decrease in ux with time (Fig. 6).
The membrane uxes reported in both cases were the
maximum attainable uxes immediately after washing
and these decreased from 0.8 to 0.36 L/m2 h by the 160th
day of operation. The trans-membrane pressure (TMP)
between washings increased from 5 to 25 psi. Initially,
cleaning frequency was every 5 days and as the ux
deteriorated to 0.6 L/m2 h it increased to every 2 days.
The HRT was maintained by gradually increasing the
pump ow rate between membrane washings which
resulted in a change in retentate ow rate but not the
permeate ow rate.
The conventional MBR operated at a clean mem-
brane ux of 6.2 L/m2 h, which is much lower than the
vendor recommended value of 20 L/m2 h for the ZW-1
membrane module. The reduced operational uxes in
both cases attest to the low lterability of the wastewater
treated in this study. The ZW-1 membrane continued to
Fig. 6. Deterioration of membrane ux in thermophilic MBR.
function at the same ux throughout the experiment at
TMP starting from 0 psi immediately after cleaning and
rising to 10 psi before cleaning, with a cleaning
frequency of 23 days.
4. Conclusions
The study compares the performance of a thermo-
philic and conventional MBR at hydraulic retention
times (HRTs) of 7 and 6.3 days, respectively, and 5 days.
It is concluded that the thermophilic treatment system
exhibited much higher sensitivity to the reduction in
HRT manifested as efuent quality deterioration,
compared to the conventional system. The thermophilic
MBR efuent consisted of a signicant fraction of
biodegradable contaminants, unlike the conventional
MBR. The seepage of oil and grease through the
membrane at higher temperatures also contributed to
the efuent deterioration at a lower HRT. The F/M had
 ARTICLE IN PRESS
R. Kurian et al. / Water Research 39 (2005) 42994308
a signicant impact on COD removal efciencies in both
the systems, with performances deteriorating beyond 1.1
and 0.5 g COD/g VSS d in the thermophilic and
conventional MBR, respectively.
Acknowledgments
The authors express their gratitude to CRESTech and
Finnie Distributors, St. Marys, Ont, for nancial
support.
Reference
Acharya, C., Nakhla, G., Bassi, A., Kurian, R., 2003.
Treatment of high strength pet food wastewater using two
stage membrane bioreactors. In: Proceedings of 77th
Annual Conference of Water Environmental Federation,
2004, New Orleans, USA.
APHAAWWAWPCF, 1998. Standard Method for Exam-
ination of Water and Wastewater, twentieth ed. American
Public Health Association, Washington, D.C, USA.
Barr, T.A., Taylor, J.M., Duff, S.J.B., 1996. Effect of HRT,
SRT and temperature on the performance of activated
sludge reactors treating bleached kraft mill efuent. Water
Res. 30 (4), 799810.
Becker, P., Koster, D., Popov, M.N., Markossian, S.,
Antranikian, G., Markl, H., 1997. The biodegradation of
olive oil and the treatment of lipid-rich wool scouring
wastewater under aerobic thermophilic conditions. Water
Res. 33 (3), 653660.
Berube, P., Hall, E., 2000. Effects of elevated operating
temperatures on methanol removal kinetics from synthetic
kraft pulp mill condensate using a membrane bioreactor.
Water Res. 34, 43594366.
Bouhabila, E.H., Ben Aim, R., Buisson, H., 2001. Fouling
characterization in membrane bioreactors. Sep.Purif. Tech-
nol. 2223, 123132.
- etin, F.D., Surucu, G., 1990. Effects of temperature and pH
C Technol. 45 (12), 143150.
on the settlability of activated sludge ocs. Water Sci.
Technol. 22 (9), 249254.
Chen, T.K., Chen, J.N., Ni, C.H., Lin, G.T., Chang, C.Y.,
2003. Application of a membrane bioreactor system for
opto-electronic industrial wastewater treatment- a pilot
study. Water Sci. Technol. 48 (8), 195202.
Chu, A., 1995. Volatile fatty acid metabolism in thermophilic
aerobic digestion of sludge. PhD. Thesis, Dept. of Civil
Engineering, University of British Columbia, Vancouver,
Canada.
Couillard, D., Zhu, S., 1993. Thermophilic aerobic process for
the treatment of slaughterhouse efuents with protein
recovery. Environ. Pollut. 79, 121126.
Driessen, W., Speert, P.Y., 1999. Anaerobic treatment of low,
medium and high strength efuent in the agro-industry.
Water Sci. Technol. 40 (8), 221228.
Fothergill, S., Mavinic, D.S., 2000. VFA production in
thermophilic aerobic digestion of municipal sludges. J.
Environ. Eng. 126 (5), 389396.
4307
Gariepy, S., Tyagi, R., Couillard, D., Tran, F., 1989.
Thermophilic process for protein recovery and an alter-
native to slaughterhouse wastewater treatment. Biol. Wastes
29, 93105.
Hwu, C.-S., 2001. Enhanced anaerobic biological treatment of
lipid-containing wastewaters. Zhongguo Huanjing Gong-
cheng Xuekan 11 (2), 151156.
Johnson, R., Hall, E., 1996. The Biological Treatment of
Recirculated Whitewater at High Temperatures. CCPA
Tech Sect. Pacic Coast and Western Branches Joint Spring
Conf., Jasper, Alberta, Canada, pp. 17.
Joore, L., Wortel, N., Bronold, N., 2002. Some perspectives of
thermophilic membrane bioreactor technologies (MBR) in
water loop closure concepts in Dutch recycled paper mills:
from pilot towards full-scale installations. Wochenblatt
fuer Papierfabrikation 130 (7), 432439.
Kumar, K.C., Bachman, T.C., 2000. Reduction in efuent SS
from UASB/EGSB reactors treating beverage wastewater.
Water Environment Federation and Purdue University
Industrial Wastes Technical Conference, St. Louis, MO,
USA, May 2124, 2000, pp. 431445.
Lapara, T.M., Alleman, J.E., 1999. Thermophilic aerobic
wastewater treatment. Water Res. 33 (4), 895908.
LaPara, T.M., Nakatsu, C.H., Alleman, J.E., Konopka, A.,
1999. Thermophilic aerobic biological wastewater
treatment: process performance and community
analysis. In: Water Environment Federation and Purdue
Industrial Wastes Conference, June 2730, 1999. IN, USA,
pp. 7992.
Lapara, T.M., Konopka, A., Nakatsu, C.H., Alleman, J.E.,
2000. Effects of elevated temperature on bacterial commu-
nity structure and function in bioreactors treating a
synthetic wastewater. J. Ind. Microbiol. Biotechnol. 24,
140145.
Lapara, T.M., Konopka, A., Nakatsu, C.H., Alleman, J.E.,
2001. Thermophilic aerobic treatment of a synthetic waste-
water in a membrane-coupled bioreactor. J. Ind. Microbiol.
Biotechnol. 26 (4), 203209.
Li, Y.Y., Sasaki, H., Yamashita, K., Seki, K., Kamigochi, I.,
2002. High-rate methane fermentation of lipid-rich food
wastes by a high-solids co-digestion process. Water Sci.
Lopetegui, J., Sancho, L., 2003. Aerated thermophilic biologi-
cal treatment with membrane ultralteration: alternative to
conventional technologies treating paper mill efuents.
Water Sci. Technol.: water Supply 3 (56), 245252.
Rajbhandari, B.K., Annachhatre, A.P., 2004. Anaerobic ponds
treatment of starch wastewater: case study in Thailand.
Environ. Eng. Manage., Asian Institute of Technology,
Pathumthani, Thailand. Bioresource Technol. 95 (2),
135143.
Rozich, A.F., Bordacs, K., 2002. Use of thermophilic biological
aerobic technology for industrial waste treatment. In: PMC
Technologies, Exton, PA, USA. Water Science and Tech-
nology, 46(45, 2nd World Water Congress: Wastewater
Treatment and Sludge Management, 2001), pp. 8389.
Scott, J.A., Smith, K.L., 1997. A bioreactor coupled to a
membrane to provide aeration and ltration in ice-cream
factory wastewater remediation. Water Res. 31 (1), 6974.
Suvilampi, J., Lehtomaki, A., Rintala, J., 2003. Comparison of
laboratory-scale thermophilic biolm and activated sludge
 ARTICLE IN PRESS
4308 R. Kurian et al. / Water Research 39 (2005) 42994308
processes integrated with a mesophilic activated sludge
process. Bioresource Technol. 88, 207214.
Tchobanoglous, G., Burton, F.L., Stensel, D.H., 2003. Waste-
water Engineering treatment and Reuse, fourth ed.. The
McGraw-Hill Companies, Inc., New York, USA.
Tripathi, C.S., Grant Allen, D., 1999. Comparison of meso-
philic and thermophilic aerobic biological treatment in
sequencing batch reactors treating bleached kraft pulp mill
efuent. Water Res. 33 (3), 836846.
Wyffels, S., Boeckx, P., Pynaert, K., Verstrate, W., Cleemput,
O.V., 2003. Sustained nitrite accumulation in a membrane-
assisted bioreactor (MBR) for the treatment of ammonium-
rich wastewater. J. Chem. Technol. Biotechnol. 78 (4),
412419.
Zhang, Z., Li, K., Zhang, Y., Wang, T., Xu, Z., Yu, S., 2003.
Treatment of high strength wastewater from denatured
starch production. Zhongguo Jishui Paishui 19 (3),
8687.