Chemosphere 44 (2001) 31±36

Enhanced sludge granulation in up¯ow anaerobic sludge

blanket (UASB) reactors by aluminum chloride
a,*
H.Q. Yu , H.H.P. Fang a, J.H. Tay b

a
Centre for Environmental Engineering Research, Department of Civil Engineering, The University of Hong Kong, Hong Kong
b
School of Civil & Structural Engineering, Nanyang Technological University, Singapore

Abstract

Two up¯ow anaerobic sludge blanket (UASB) reactors were concurrently operated for 146 days to examine the
e€ects of aluminum chloride (AlCl3 ) on the sludge granulation process during start-up. Sludge granulation (de®ned as
that over 10% of granules were larger than 2.0 mm) was achieved in the control reactor (R1) in approximate three
months. Introduction of Al3‡ at a concentration of 300 mg/l reduced the sludge granulation time by approximate one
month. Throughout the experiment the AlCl3 -added reactor (R2) had a higher biomass concentration, e.g., 13.8 g-
MLVSS/l versus 12.8 g-MLVSS/l on Day 146. Granules became visible earlier in R2 compared with R1 (35 days versus
65 days). The average size of granules from R2 was larger than that from R1. The results demonstrated that AlCl3
enhanced the sludge granulation process in the UASB reactors. Ó 2001 Elsevier Science Ltd. All rights reserved.

Keywords: Aluminum chloride; Granular sludge; Start-up; UASB

1. Introduction Granulation may be initiated by bacterial adsorption

The up¯ow anaerobic sludge blanket (UASB) reactor
has been used increasingly in recent years to treat a
variety of industrial wastes and municipal wastes
(Lettinga et al., 1993; Fang et al., 1995). The UASB
process involves the anaerobic degradation of organic
wastes using a biomass which is not attached to a sup-
port medium but which aggregates, under favorable
conditions, to produce particles with good settlement
characteristics. These particles are known as granules,
and their formation, commonly termed granulation,
generally enhances the eciency of the process, pro-
ducing high biomass retention times. Microbial granu-
lation involves di€erent tropic bacterial groups, and
physico-chemical and microbiological interactions.
Many factors contribute, in one form or another, to the
granulation process (Schmidt and Ahring, 1996).
*
Corresponding author. Fax: +852-2559-5337.
E-mail address: hqyu@hkucc.hku.hk (H.Q. Yu).
and adhesion to inert matters, inorganic precipitates
and/or to each other through physico-chemical interac-
tions and syntrophic relationships (Fang et al., 1995;
Schmidt and Ahring, 1996). These substances serve as
initial precursors (carriers or nuclei) for new bacterial
growth. These loosely adhered bacterial aggregates are
strengthened by extracellular polymers secreted by bac-
teria and form ®rmly attached initial granules (Kosaric
and Blaszczyk, 1990; Schmidt and Ahring, 1996). It has
been shown that divalent metal ions, such as Ca2‡ and
Fe2‡ , enhance the granulation (Mahoney et al., 1987;
Schmidt and Ahring, 1993; Shen et al., 1993). Divalent
ions were reported to play an important role in micro-
bial aggregation. It was found that extracellular poly-
mers prefer to bind multi-valent metals due to the
formation of stable complexes (Rudd et al., 1984; Ma-
honey et al., 1987).
Aluminum chloride (AlCl3 ) is widely used as an in-
organic coagulant in water and wastewater treatment. It
is conceivable that this coagulant might also enhance the
granulation process through physico-chemical functions

0045-6535/01/$ - see front matter Ó 2001 Elsevier Science Ltd. All rights reserved.
PII: S 0 0 4 5 - 6 5 3 5 ( 0 0 ) 0 0 3 8 1 - 7
32 H.Q. Yu et al. / Chemosphere 44 (2001) 31±36
as do calcium and iron salts. However, no study has
been reported that Al3‡ could bene®t for granulation.
Therefore, the aim of this study was to investigate the
e€ects of AlCl3 on the sludge granulation process during
start-up.
2. Materials and methods
Experiments were performed in parallel in two
identical reactors for 146 days. The empty bed volume of
each reactor was 7.3 l with an internal diameter of 100
mm and a liquid height of 930 mm. The two reactors
were housed in a temperature control room maintained
at 35°C. One reactor (R1) without adding AlCl3 was
served as control, while AlCl3 was added to another
reactor (R2) to give an in¯uent 300 mg/l of Al3‡ .
Soluble synthetic wastewater was used as feed to the
reactors. The in¯uent chemical oxygen demand (COD)
was kept constant at 4000 mg/l throughout the study.
Table 1 shows the detailed composition of the synthetic
waste. The organic COD in wastewater was provided by
peptone, glucose and meat extract, whereas several nu-
trients and trace elements, such as nitrogen, phosphorus,
sulfur, calcium, iron and magnesium were also added.
The ratio of COD:N:P was approximately 200:4:1. The
bu€er capacity was maintained by the addition of so-
dium bicarbonate.
The seeding inoculum was the sludge taken from the
anaerobic digesters of a local sewage treatment plant.
The raw sludge was screened through a 0.2 mm sieve to
remove the big debris and ®bers before seeding. The
sieved sludge had 23.6 g/l of suspended solids (SS), 17.0
g/l of volatile suspended solids (VSS), and a sludge
volume index (SVI) of 43 ml/g. The speci®c methano-
genic activity (SMA) was 0:26 g CH4 COD=g
VSSd). Each reactor was seeded with 3.5 l of inoculum
with biomass content of 8.5 g-VSS/l.
Biomass were periodically taken from the sampling
points arranged along the columns to determine the
mixed liquid VSS concentrations. Gas production was
Table 1
Composition of the synthetic wastewater of 4000 mg-COD/l
Constituents mg/l
Bacteriological peptone 800
Glucose 2720
Meat extract (Lab-Lemco powder) 560
Sodium bicarbonate, NaHCO3 1500±2500
Calcium chloride, CaCl2
2H2 O 38
Magnesium sulfate, MgSO4
7H2 O 42
Ammonium chloride, NH4 Cl 160
Ferrous sulfate, FeSO4
7H2 O 32
Potassium dihydrogen ortho-phosphate, 80
KH2 PO4
recorded by wet gas meters (Shinagawa W-NK-0.5) and
gas composition was analyzed by Gas Chromatography
(HP 5890A) with a 2 m long and 3 mm-ID packed
column (Haye-Sep Q, 80/100 mesh) and thermal con-
ductivity detector with a temperature of 2000°C. Helium
was used as the carrier gas in the gas chromatography
operation with a ¯owrate of 30 ml/min.
For estimating the size distribution of the sludge
particles taken from the bottom sampling point, solid
samples were classi®ed into six fractions using labora-
tory sieves (Endecotts Ltd., London, England) with
various openings (0.2, 0.6, 1.0, 2.0, 4.0 mm). The sludge
particles were ®rst placed in the sieve with the biggest
opening (4.0 mm). The particles were gently submerged
in water and shaken to let the smaller particles pass
through this sieve. The particles passing through were
collected in a container. The smaller particles collected
in the above sieve were then placed in the next sieve
(opening of 2.0 mm) and the above procedures were
repeated until all of the ®ve sieves were used. The sludge,
taken from the bottom sampling point, was measured
for SMA. The SMA test was conducted in a 250 ml
Kimax ¯ask at 35  1°C under anaerobic conditions as
described previously (Fang et al., 1994). The analyses of
COD and mixed liquor VSS were performed according
to the standard methods (APHA, AWWA and WEF,
1992).
3. Results and discussion
The initial organic loading rate (OLR) for both re-
actors was set at 2.0 g-COD/l/d based on the total liquid
volume of 7.3 l and a hydraulic retention time (HRT) of
48 h. The loading rate was then increased in steps to 2.7,
4.0, 5.3 and 8.0 g-COD/l/d by reducing the HRT to 36,
24, 18, and 12 h, correspondingly.
The COD removal eciencies of the two reactors are
illustrated in Fig. 1. Initially the COD removal e-
ciencies were low, but the R2 had a slightly higher re-
moval than R1. With the progress of the experiment, the
COD removal eciencies of the two reactors generally
Fig. 1. COD removal eciency.
 H.Q. Yu et al. / Chemosphere 44 (2001) 31±36 33

kept increasing, but the eciency of R2 increased faster
than that of R1. However, after reaching an OLR of 5.3
g-COD/l/d, R1 and R2 achieved similar levels of COD
removal.
Table 2 shows the change in MLVSS concentration
in the two reactors. The overall biomass increase was
0.030 g-VSS/d for R1 and 0.034 g-VSS/d for R2. The
initial decrease in MLVSS concentrations was due to the
washout of biomass. The MLVSS concentration reached
a minimum in R1 on Day 30. Throughout the study, R1
had a lower MLVSS concentration compared with R2,
although the di€erence was insigni®cant at the ®nal
stages.
Because of inferior settleability, more biomass was
washed out from R1, resulting in a higher VSS con-
centration in the e‚uent of R1 than that of R2 at the
initial stages (Fig. 2). The lower e‚uent VSS concen-
tration from R2 was likely resulting from with the im-
provement of biomass settleability through addition of
AlCl3 .
The pro®les of biomass concentration at di€erent
stages are shown in Table 3. Initially, the biomass was
loose and expanded easily. With the proceeding of
granulation, the biomass was progressively strati®ed
with the granules settled in the lower part of the reaction
zone. When granules were formed increasingly in the
reaction zone, a dense sludge bed and a thin sludge
blanket were formed with a clear interface between
them.
As shown in Fig. 3, the biogas production increased
at a rapid rate initially in the two reactors (Days 1±25).
At the later stages (after Day 25), biogas production
increased with the increase in OLR. An interesting
phenomenon was observed regarding the methane per-
centages (Fig. 4). Initially, the methane percentages of
the two reactors were low and increased considerably in
a short period. On Day 50, the average methane per-
centages were approximately 72% for R1 and 73% for
R2 with no respect to earlier or later granule initiation

Table 2
Biomass MLVSS concentrations (g/l) in each reactor
Reactor Operatinal time (d)
0 20 30 60 90 120 146
R1 8.5 7.0 6.1 7.3 9.5 11.8 12.9
R2 8.5 6.7 7.3 8.9 11.1 12.6 13.4
Fig. 3. Biogas production rate.

Fig. 2. E‚uent VSS level. Fig. 4. Methane percentage.

Table 3
Pro®les of biomass concentration (g/l)
Height R1 R2
(cm) Day 30 Day 60 Day 90 Day 146 Day 30 Day 60 Day 90 Day 146
2 18.6 24.7 42.9 57.5 22.5 39.6 47.4 59.6
12 18.2 23.0 41.1 55.1 22.4 35.4 45.2 56.9
32 18.0 22.2 37.4 48.5 19.3 31.6 32.8 50.4
52 5.8 10.8 11.8 32.6 9.1 9.8 17.6 29.3
72 1.9 1.1 0.9 0.8 1.2 0.8 1.2 0.5
82 0.4 0.3 0.3 0.2 0.2 0.2 0.1 0.1
34 H.Q. Yu et al. / Chemosphere 44 (2001) 31±36
and values of OLR. This may suggest that the methane
percentages had no direct correlation with the formation
of granules.
After 30 days of operation, small granules with di-
ameters of 0.20±0.6 mm became visible at the bottom of
R2. These initial granules grew rapidly; after 30 more
days of operation large granules with diameters over 2.0
mm were formed. The rapid granulation was mainly due
to increasing biological growth under the high OLR (4.0
g-COD/l/d). Physical strati®cation to the sludge blanket
under high OLR and gas production also contributed to
the rapid granulation. After about 30 days of rapid
growth, the growth rate of granules was reduced, indi-
cating that a mature and stable granulation had been
developed. The reactor R1 had a slower granulation
process compared with R2.
The granule size distributions of the two reactors are
illustrated in Fig. 5. In R1, no granules were found until
Day 65. Sixty-®ve days into start-up, approximately 45%
of the sludge samples were in the range of 0.2±0.6 mm.
On Day 146, approximately 60% of the granules from
R1 were in the range of 1.0±4.0 mm and only 8% of the
sample measured above 4.0 mm. The granule size dis-
tribution for R1 was signi®cantly di€erent from that for
R2 on any given day. For R2, 12% of the samples
measured above 2.0 mm on Day 60. By Day 146, ap-
Fig. 5. Size distributions (by weight) of granules taken from the
bottom sampling points of (a) the control; (b) the reactor added
with Al3‡ A: d < 0:2 mm; B: 0:2 < d < 0:6 mm; C:
0:6 < d < 1:0 mm; D: 1:0 < d < 2:0; E: 2:0 < d < 4:0 mm; F:
d > 4:0 mm).
proximately 11% of granules in R2 had a diameter over
4.0 mm. Larger granule sizes in R2 may be due to the
presence of AlCl3 as a binding factor. This result implies
that the presence of AlCl3 had promoted granule for-
mation by allowing aggregates to form earlier and to
achieve a larger size.
For the purpose of comparison, the ``time needed to
accomplish granulation'' in this study is de®ned as when
over 10% of granules were larger than 2.0 mm in the
reactor. Accordingly, the granulation was achieved in
R1 within three months, while granulation was achieved
in R2 after two months of operation. R2 had a higher
biomass concentration and had visible granules earlier
compared with R1. The average size of granules in R2
was larger at any given stages (Fig. 5). These results
clearly indicate that 300 mg=l Al3‡ improved the bio-
mass retention and achieved a fast granulation process.
In R2, more bacteria were maintained (Table 2), and the
COD digestion rate was also higher at the initial stages
compared with R1. Therefore, AlCl3 showed a positive
e€ect on the COD digestion rate at the initial stages. The
positive e€ect of Al3‡ in granulation are likely to be
explained by the assumption that Al3‡ lowers the n
potential (Schmidt and Ahring, 1996). In addition, the
Al3‡ can bridge between negatively charged groups on
cell surfaces, which is important in adhesion phenomena
(Alibhai and Forster, 1986; Quarmby and Forster,
1995).
Table 4 shows the SMA of the granular sludges on
Days 30, 60, 90 and 146, using acetate as substrate. The
SMA of the seed sludge was 0:26 g CH4 COD=g
VSS=d. The methanogenic activity of granules increased
steadily with increasing OLR. After OLR was increased
to 5.3 g-COD/l/d, the SMA value of R2 dropped slightly,
while the SMA value of R1 continued to increase.
Compared with the biomass from R1, the lower SMA
values of the biomass in R2 might be attributed to the
presence of larger granules, in which mass transfer was
reduced. Mass transport inside the granules is consid-
ered to be solely by di€usion (Pavlostathis and Giraldo-
Gomez, 1991). The resistance to substrate di€usion in-
side granules increases proportionally with physical
granular size, making the substrate less available to the
granule core and eventually resulting in substrate de®-
ciency or depletion (Alphenaar et al., 1993).
The dry mass percentage of the granules from R2 was
21%, much higher than the corresponding value (10%)
Table 4
SMA (g CH4 COD=g VSS=d) using acetate as substrate
for the biomass in two reactors
Reactor Operating time (d)
0 30 60 90 146
R1 0.26 0.70 1.04 1.28 1.32
R2 0.26 1.02 1.24 1.13 1.10
 H.Q. Yu et al. / Chemosphere 44 (2001) 31±36
of the granules from R1. This indicates that the presence
of aluminum increased the dry mass of the granules
mainly by increasing the concentration of minerals in the
granules. The increased mineral content was very likely
the result of more aluminum precipitates trapped in the
granules. Aluminum accumulation within granules
mostly resulted from the interaction of aluminum ions
with exopolysaccharide polymers because aluminum
concentration in bacterial cells is very low. The addi-
tional inorganic portion in the granules from R2 might
also be a reason to lower down the SMA values as
shown in Table 4.
It should be noticed that the di€erence of sludge
granulation process among the two reactors was mainly
found only at the initial stages (Days 1±60). At the ®nal
stage (Days 130±146), the control reactor behaved very
similarly to the Al3‡ -added reactor; the former had a
little less biomass concentration but a slightly higher
speci®c methanogenic activity than the latter, resulting
in very similar COD removals and e‚uent VSS levels in
the two reactors. This suggests that the addition of AlCl3
played an important role only in the initial stage of
granulation and its e€ect was diminished after the de-
velopment of mature granules.
The scanning electric microscopic observation of
sludges revealed that the sludges from R1 had a very
similar composition and morphology to those from R2.
The granules from both reactors were predominantly
composed of Methanosarcina-like species.
Comparison between the present study and the
studies regarding the e€ect of Ca2‡ and Fe2‡ on sludge
granulation shows that, Al3‡ had a positive e€ect very
similar to Ca2‡ and Fe2‡ (Mahoney et al., 1987; Schmidt
and Ahring, 1993; Shen et al., 1993). All these ions
promoted granule formation by allowing aggregates to
form earlier and to achieve a larger size, and resulted in
a faster granulation process and a shortened start-up
period for UASB reactors. Besides, the addition of these
ions did not lead to in a di€erence in predominant mi-
croorganisms. These suggest that all the three ions en-
hanced the sludge granulation process through the same
physico-chemical functions. However, when Ca2‡ is
added, mineral precipitates such as CaCO3 and
Ca5 OH…PO4 †3 are formed and there is a risk of cemen-
tation of the UASB reactors (van Langerak et al., 1998).
The same case is true for the formation of FeS when
Fe2‡ is added to UASB reactors. On the other hand,
there is no such a risk when Al3‡ is used. Therefore, Al3‡
might be a better option than Ca2‡ and Fe2‡ for en-
hancing the sludge granulation in UASB reactors.
4. Conclusions
In this study sludge granulation was achieved in the
control reactor in approximately 3 months. Introduc-
35
tion of Al3‡ at a concentration of 300 mg/l reduced the
sludge granulation time by approximately 1 month.
The Al3‡ -added reactor had a higher biomass concen-
tration throughout experiment. It had visible granules
earlier (35 days versus 65 days) and larger granules
compared with R1. The results demonstrated that
AlCl3 enhanced sludge granulation process. However,
its e€ect diminished as granulation process became
mature.
Acknowledgements
The authors wish to thank the Hong Kong Research
Grants Council for the partial support of this study and
one author (HQY) also wishes to thank The University
of Hong Kong for the Research Fellowship.
References
Alibhai, K.R.K., Forster, C.F., 1986. An examination of the
granulation process in UASB reactors. Environ. Tech. 7,
193±200.
Alphenaar, P.A., Perez, M.C., Lettinga, G., 1993. The in¯uence
of substrate transport limitation on porosity and methano-
genic activity of anaerobic sludge granules. Appl. Micro-
biol. Biotechnol. 39, 276±280.
Standard Methods for the Examination of Water and Waste-
water, 18th ed., Washington, DC.
Fang, H.H.P., Chui, H.K., Li, Y.Y., Chen, T., 1994. Perfor-
mance and granule characteristics of UASB process treating
wastewater with hydrolyzed proteins. Water. Sci. Technol.
30, 55±63.
Fang, H.H.P., Chui, H.K., Li, Y.Y., 1995. E€ect of degradation
kinetics on the microstructure of anaerobic biogranules.
Water. Sci. Technol. 32, 165±172.
Kosaric, N., Blaszczyk, R., 1990. Microbial aggregates in
anaerobic wastewater treatment. Advan. Biochem. Eng. &
Biotechnol. 42, 27±62.
Lettinga, G., Man, A.D., ver der Last, A.R.M., Wiegant, W.,
Frijins, J., van Buuren, J.C.L., 1993. Anaerobic treatment of
domestic sewage and wastewater. Water. Sci. Technol. 27,
67±73.
Mahoney, E.M., Varangu, L.K., Cairns, W.L., Kosaric, N.,
Murray, R., 1987. The e€ect of calcium on microbial
aggregation during UASB reactor start-up. Water. Sci.
Technol. 19, 249±260.
Pavlostathis, S.G., Giraldo-Gomez, E., 1991. Kinetics of
anaerobic treatment: a critical review. Crit. Rev. Environ.
Conam. 21, 411±499.
Quarmby, J., Forster, C.F., 1995. A comparative study of the
structure of thermophilic and mesophilic anaerobic gran-
ules. Enzy. Microbiol. Technol. 17, 493±498.
Rudd, T., Sterritt, R., Lester, J., 1984. Complexation of heavy
metals by extracellular polymers in the activated sludge
process. J. Water. Poll. Contam. Fed. 56, 1260±1268.
Schmidt, J.E., Ahring, B.K., 1993. E€ects of magnesium on
thermophilic acetate-degrading granules in up¯ow anaero-
36 H.Q. Yu et al. / Chemosphere 44 (2001) 31±36

bic sludge blanket (UASB) reactors. Enzy. Microbiol.
Technol. 15, 304±310.
Schmidt, J.E., Ahring, B.K., 1996. Granular sludge formation
in up¯ow anaerobic sludge blanket (UASB) reactors.
Biotechnol. Bioeng. 49, 229±246.
Shen, C.F., Kosaric, N., Blaszczyk, R., 1993. The e€ect of
selected heavy metals (Ni, Co and Fe) on anaerobic granules
and their extracellular polymeric substance. Water. Res. 27,
25±33.
van Langerak, E.P.A., Gonzalez-Gil, G., van Aelst, A., van
Lier, J.B., Hamelers, H.V.M., Lettinga, G., 1998. E€ects of
high calcium concentrations on the development of meth-
anogenic sludge in up¯ow anaerobic sludge bed (UASB)
reactors. Water. Res. 32, 1255±1263.