You are on page 1of 31

Biotechnology Advances 22 (2004) 533 – 563

www.elsevier.com/locate/biotechadv

Research review paper

State of the art of biogranulation technology for


wastewater treatment
Yu Liu *, Joo-Hwa Tay
Division of Environmental and Water Resources Engineering, School of Civil and Environmental Engineering,
Nanyang Technological University, 50 Nanyang Avenue, Singapore 639798, Singapore
Accepted 6 May 2004
Available online

Abstract

Biogranulation technology developed for wastewater treatment includes anaerobic and aerobic
granulation processes. Anaerobic granulation is relatively well known, but research on aerobic
granulation commenced only recently. Many full-scale anaerobic granular sludge units have been
operated worldwide, but no report exists of similar units for aerobic granulation. This paper reviews
the fundamentals and applications of biogranulation technology in wastewater treatment. Aspects
discussed include the models of biogranulation, major factors influencing biogranulation,
characteristics of biogranules, and their industrial applications. This review hopes to provide a
platform for developing novel granules-based bioreactors and devising a unified interpretation of the
formation of anaerobic and aerobic granules under various operation conditions.
D 2004 Elsevier Inc. All rights reserved.

Keywords: Biogranulation; Aerobic granules; Anaerobic granules; Operating parameters; Microbial structure;
Diversity; Mechanism of granulation

1. Introduction

Biogranulation involves cell-to-cell interactions that include biological, physical and


chemical phenomena. Biogranulation can be classified as aerobic and anaerobic granulation.
Biogranules form through self-immobilization of microorganisms. These granules are dense
microbial consortia packed with different bacterial species and typically contain millions of
organisms per gram of biomass. These bacteria perform different roles in degrading the

* Corresponding author.
E-mail address: cyliu@ntu.edu.sg (Y. Liu).

0734-9750/$ - see front matter D 2004 Elsevier Inc. All rights reserved.
doi:10.1016/j.biotechadv.2004.05.001
534 Y. Liu, J.-H. Tay / Biotechnology Advances 22 (2004) 533–563

complex industrial wastes. Compared to the conventional activated sludge, biogranules have
a regular, dense, and strong structure and good settling properties. They enable a high
biomass retention and withstand high-strength wastewater and shock loadings.
Formation of anaerobic granules has been extensively studied and is probably best
recognized in the upflow anaerobic sludge blanket (UASB) reactor. Many wastewater treatment
plants already apply anaerobic granulation technology (Alves et al., 2000). The feasibility and
efficiency of UASB reactors and their various modifications (e.g., the internal circulation (IC)
reactor) for removing biodegradable organic matter from municipal and industrial wastewater
have been successfully demonstrated (Lettinga et al., 1980; Fang and Chui, 1993; Schmidt and
Ahring, 1996). Anaerobic granular sludge is a dense microbial community that typically
includes millions of organisms per gram of biomass. None of the individual species in these
microecosystems is capable of completely degrading the influent wastes. Complete degradation
of industrial waste involves complex interactions between the resident species. Thus, granular
sludge reactors are desirable in wastewater biological treatment processes because a very high
number of organisms can be maintained in the bioreactor. This in turn implies that contaminant
transformation is rapid and highly concentrated; therefore, large volumes of waste can be
treated in compact bioreactors. In granular sludge reactors, the large size and relatively high
density of individual granules causes them to settle rapidly, which simplifies the separation of
treated effluent from the biomass. Anaerobic granular sludge has proved capable of treating
high-strength wastewater contaminated with soluble organic pollutants.
The anaerobic granulation technology has some drawbacks. These include the need for a
long start-up period, a relatively high operation temperature and unsuitability for low-
strength organic wastewater. In addition, anaerobic granulation technology is not suitable for
the removal of nutrients (N and P) from wastewater. In order to overcome those weaknesses,
research has been devoted to the development of aerobic granulation technology. The
development of aerobic granules was first reported by Mishima and Nakamura (1991) in a
continuous aerobic upflow sludge blanket reactor. Aerobic granules with diameters of 2 to
8 mm were developed, with good settling properties. Aerobic granulation has since been
reported in sequencing batch reactors (SBRs) by many researchers (Morgenroth et al., 1997;
Beun et al., 1999; Peng et al., 1999; Etterer and Wilderer, 2001; Tay et al., 2001a; Liu and
Tay, 2002) and has been used in treating high-strength wastewaters containing organics,
nitrogen and phosphorus, and toxic substances (Jiang et al., 2002; Moy et al., 2002; Tay et
al., 2002e; Lin et al., 2003; Yang et al., in press). Development of biogranules requires
aggregation of microorganisms. For bacteria in a culture to aggregate, a number of
conditions have to be met. The formation of anaerobic granules is a multiple-step process
that involves physicochemical and biological forces. This review is focused on progresses in
biogranulation technology developed for wastewater treatment.

2. Aerobic granulation technology

2.1. The formation of aerobic granules

Sludge is the microbial biomass that utilizes nutrient substrates present in wastewater.
Microbial granules can be regarded as compact and dense microbial aggregates with a
Y. Liu, J.-H. Tay / Biotechnology Advances 22 (2004) 533–563 535

spherical outer shape. The growth of aerobic granules is sometimes regarded as a special
case of biofilm development (Liu and Tay, 2002; Yang et al., 2004a). In fact, microbial
granulation is quite fundamental in biology and cell aggregation can be defined as the
gathering together of cells to form a fairly stable, contiguous, multicellular association
under physiological conditions (Calleja, 1984). Each aerobic granule is an enormous
metropolis of microbes containing millions of individual bacteria. Almost all aerobic
granules have been cultivated in sequencing batch reactors (SBRs). The SBR system is a
modified design of the conventional activated sludge process and has been widely used in
municipal and industrial wastewater treatment. Aerobic granulation may be initiated by
microbial self-adhesion. Bacteria are not likely to aggregate naturally because of the
repulsive electrostatic forces and hydration interactions among them.
Tay et al. (2001a) used different microscopic techniques to investigate how an aerobic
granule formed from seed sludge. For comparison, granules were cultivated in two reactors
fed with glucose in one case and acetate in the other case, as sole carbon sources. The
results showed that the seed sludge had a very loose and irregular structure, dominated by
filamentous bacteria. After operation in SBR for 1 week, compact aggregates appeared.
The filamentous bacteria gradually disappeared in the acetate-fed reactor; however, in the
glucose-fed reactor, filamentous bacteria still prevailed. Two weeks after the start-up, the
granular sludge with clear round outer shape was formed in both reactors. Although the
filamentous bacteria disappeared completely in acetate-fed reactor, they were still
predominant in glucose-fed reactor. This may imply that a high-carbohydrate feed
composed of glucose supports the growth of filamentous bacteria as reported in activated
sludge process previously (Chudoba, 1985). After operation for 3 weeks, aerobic granules
matured in both reactors. At this stage, both glucose- and acetate-fed granules had a very
regular round-shaped outer surface. The average aspect ratio of glucose-fed granules was
0.79 and 0.73 for acetate-fed granules. (Aspect ratio of a particle is the ratio of the lengths
of minor axis and major axis of an ellipse that is equivalent to the particle.) Compared to
acetate-fed granules, glucose-fed granules had a fluffy outer surface because of the
predominance of filamentous bacteria (Fig. 1). Scanning electron microscope (SEM)
observations further revealed that the glucose-fed mature aerobic granules indeed had a
filamentous dominant outer surface, while the acetate-fed aerobic granules had a very

Fig. 1. Macrostructures of glucose-fed (a) and acetate-fed (b) aerobic granules (Tay et al., 2001a).
536 Y. Liu, J.-H. Tay / Biotechnology Advances 22 (2004) 533–563

compact microstructure in which cells were tightly linked together and rodlike bacteria
were predominant (Fig. 2). It seems certain that aerobic granulation is a gradual process
involving the progression from seed sludge to compact aggregates, further to granular
sludge and finally to mature granules.

2.2. Factors affecting aerobic granulation

For cells in a culture to aggregate, a number of conditions have to be satisfied. This


section attempts to discuss the factors that may influence aerobic granulation.

2.2.1. Substrate composition


Aerobic granules have been successfully cultivated with a wide variety of substrates
including glucose, acetate, ethanol, phenol, and synthetic wastewater (Beun et al., 1999;
Peng et al., 1999; Tay et al., 2001a, 2003b; Moy et al., 2002; Jiang et al., 2002; Yang et al.,
in press; Schwarzenbeck et al., 2003). However, granule microstructure and species
diversity appear to be related to the type of carbon source. The glucose-fed aerobic
granules have exhibited a filamentous structure, while acetate-fed aerobic granules have
had a nonfilamentous and very compact bacterial structure in which a rodlike species
predominated. Aerobic granules have been also cultivated with nitrifying bacteria and an
inorganic carbon source (Tay et al., 2002b; Tsuneda et al., 2003). These nitrifying granules
showed an excellent nitrification ability. More recently, aerobic granules were also
successfully developed in laboratory-scale SBR for treating particulate organic matter-
rich wastewater (Schwarzenbeck et al., 2003).

2.2.2. Organic loading rate


The essential role of organic loading rate (OLR) in the formation of anaerobic granules
has been recognized. Relatively high organic loading rates facilitate the formation of
anaerobic granules in UASB systems. In contrast to anaerobic granulation, the accumu-
lated evidence suggests that aerobic granules can form across a very wide range of organic
loading rates from 2.5 to 15 kg chemical oxygen demand (COD)/m3 day (Moy et al., 2002;
Liu et al., 2003a). It seems that aerobic granulation is not sensitive to the organic loading
rate. Although the effect of organic loading rate on the formation of aerobic granules is

Fig. 2. Microstructures of glucose-fed (a) and acetate-fed (b) aerobic granules (Tay et al., 2001a).
Y. Liu, J.-H. Tay / Biotechnology Advances 22 (2004) 533–563 537

insignificant, the physical characteristics of aerobic granules depend on the organic


loading rate. The mean size of aerobic granules increased from 1.6 to 1.9 mm with the
increase of the organic loading from 3 to 9 kg COD/m3 day (Liu et al., 2003a). A similar
trend was also observed in anaerobic granulation (Grotenhuis et al., 1991). The effect of
organic loading rate on the morphology of aerobic granules in terms of roundness was
found to be insignificant, while the aerobic granules developed at different organic loading
rates exhibited comparable dry biomass density, specific gravity, and sludge volume index
(SVI), the physical strength of aerobic granules decreased with the increase of organic
loading rate (Liu et al., 2003a; Tay et al., in press). Similarly, in anaerobic granulation, a
high organic loading rate has been found to reduce strength of anaerobic granules; that is,
partial loss of structural integrity and disintegration can occur at high organic loading rates
(Morvai et al., 1992; Quarmby and Forster, 1995). It should be stressed that an increased
organic loading rate can raise the biomass growth rate and this in turn reduces the strength
of the three-dimensional structure of the microbial community (Liu et al., 2003c).

2.2.3. Hydrodynamic shear force


Evidence shows that a high shear force favors the formation of aerobic granules and
granule stability (Shin et al., 1992; Tay et al., 2001a). It was found that aerobic granules
could be formed only above a threshold shear force value in terms of superficial upflow air
velocity above 1.2 cm/s in a column SBR, and more regular, rounder, and compact aerobic
granules were developed at high hydrodynamic shear force (Tay et al., 2001a). The
granule density and strength were also proportionally related to the shear force applied
(Tay et al., 2003c). These observations may imply that the structure of aerobic granules is
mainly determined by the hydrodynamic shear force present in a bioreactor. However, it is
well known that extracellular polysaccharides can mediate both cohesion and adhesion of
cells and play a crucial role in maintaining the structural integrity in a community of
immobilized cells. Tay et al. (2001a) reported that the production of extracellular
polysaccharides was closely associated with the shear force and the stability of aerobic
granules was found to be related to the production of extracellular polysaccharides (Tay et
al., 2001c). The extracellular polysaccharides content normalized to protein content,
increased with the shear force estimated in terms of superficial upflow air velocity. Thus,
a high shear force stimulated bacteria to secrete more extracellular polysaccharides. In fact,
shear force-induced production of extracellular polysaccharides has been observed in
biofilms (Ohashi and Harada, 1994). Consequently, the enhanced production of extracel-
lular polysaccharides at high shear can contribute to the compact and stronger structure of
aerobic granules. Effects of shear on microorganisms and aggregates have been discussed
further elsewhere (Chisti, 1999a).

2.2.4. Settling time


In a SBR, wastewater is treated in successive cycles each lasting a few hours. At
the end of every cycle, the biomass is settled before the effluent is withdrawn. The
settling time acts as a major hydraulic selection pressure on microbial community. A
short settling time preferentially selects for the growth of fast settling bacteria and the
sludge with a poor settleability is washed out. Qin et al. (2004) reported that aerobic
granules were successfully cultivated and became dominant only in the SBR operated
538 Y. Liu, J.-H. Tay / Biotechnology Advances 22 (2004) 533–563

at a settling time of 5 min. Mixtures of aerobic granules and suspended sludge were
observed in the SBRs run at settling times of 20, 15, and 10 min. The production of
extracellular polysaccharides was stimulated and the cell surface hydrophobicity
improved significantly at short settling times. These findings illustrate the fact that
aerobic granulation is driven by selection pressure and the formation and characteristics
of the granules may be controlled by manipulating the selection pressure. Therefore,
choice of an optimal settling time is very important in aerobic granulation. Generally,
the mature aerobic granules tend to settle within 1 min, leaving a clear supernatant in
the reactor (Tay et al., 2001a). The easily retainable biomass in the reactor ensures a
faster and more efficient removal of organic pollutants in wastewater. Granules with
excellent settling properties are essential for the effective functioning of biological
systems treating wastewater.

2.2.5. Hydraulic retention time


In aerobic granulation, the light and dispersed sludge is washed out and the relatively
heavy granules are retained in the reactor. The SBR cycle time represents the frequency of
solids discharge through effluent withdrawal, or the so-called washout frequency, and it is
related to the hydraulic retention time (HRT) at a given exchange ratio. The latter is
defined as the volume of effluent discharged divided by the working volume of the SBR.
A short cycle time would suppress the growth of suspended solids because of frequent
washout of the suspended material. However, if the SBRs are run at an extremely short
cycle time, sludge loss has been observed through hydraulic washout because bacterial
growth has been unable to compensate. As a result, a complete washout of sludge blanket
occurs and leads to a failure of microbial granulation. Thus, the HRT should be short
enough to suppress the suspended growth, but long enough for microbial growth and
accumulation.
By its nature, a SBR is cyclic in operation. The SBR cycle time can serve as a main
hydraulic selection pressure on the microbial community in the system. Tay et al.
(2002b) investigated the effect of hydraulic selection pressure on the development of
nitrifying granules in column-type sequencing batch reactors. No nitrifying granulation
was observed in the SBR operated at the longest cycle time of 24 h because of a weak
hydraulic selection pressure while the nitrifying sludge was washed out in the SBR run
at the shortest cycle time of 3 h, which also prevented the development of nitrifying
granules. Excellent nitrifying granules were successfully developed in the SBR operated
at cycle times of 6 and 12 h. A short cycle time stimulates microbial activity and
production of cell polysaccharides and also improves the cell hydrophobicity. These
hydraulic selection pressure-induced microbial changes favor the formation of nitrifying
granules.

2.2.6. Aerobic starvation


The SBR operation is a sequencing cycle of feeding, aeration, settling, and
discharging of supernatant fluid. As a result, microorganisms growing in the SBR are
subject to periodic fluctuations in the environmental conditions. During operation cycles,
an important period of aerobic substrate starvation has been identified (Tay et al.,
2001a). The waste degradation time required tends to reduce with the increase in the
Y. Liu, J.-H. Tay / Biotechnology Advances 22 (2004) 533–563 539

number of operation cycles. The aeration period of the operation actually consists of two
phases: a degradation phase in which the substrate is depleted to a minimum, followed
by an aerobic starvation phase in which the external substrate is no longer available.
Under starvation conditions, bacteria became more hydrophobic which facilitates
microbial adhesion (Tay et al., 2001a). It is likely that aggregation is a strategy of
cells against starvation. It appears that the microorganisms are able to change their
surface characteristics when they face starvation (Tay et al., 2001a). Bossier and
Verstraete (1996) reported that under starvation conditions, bacteria become more
hydrophobic which likely facilitates adhesion or aggregation. Such changes contribute
to microbial ability to aggregate. Thus, starvation plays a role in the microbial
aggregation process and leads to stronger and denser granules. Although the periodical
starvation in SBR is important for microbial aggregation, the contribution of other
operation conditions should not be neglected.

2.2.7. Presence of calcium ion in feed


Jiang et al. (2003) reported that addition of Ca2 + accelerated the aerobic granulation
process. With addition of 100 mg Ca2 +/l, the formation of aerobic granules took 16 days
compared to 32 days in the culture without Ca2 + added. The Ca2 +-augmented aerobic
granules also showed better settling and strength characteristics and had higher poly-
saccharides contents. It has been proposed that Ca2 + binds to negatively charged groups
present on bacterial surfaces and extracellular polysaccharides molecules and thus acts as a
bridge to promote bacterial aggregation. Polysaccharides play an important role in
maintaining the structural integrity of biofilms and microbial aggregates, such as aerobic
granules, as they are known to form a strong and sticky nondeformable polymeric gel-like
matrix.

2.2.8. Intermittent feeding strategy


A periodic starvation occurs during the course of SBR operation (Tay et al., 2001a).
This periodic starvation has been shown to have a profound effect on cell hydrophobicity,
which is a key factor that affects aerobic granulation (Liu et al., 2003b). Tay et al. (2001a)
found that cell surface hydrophobicity was proportionally related to the starvation time in
SBR. McSwain et al. (2003) recently developed an operation strategy to enhance aerobic
granulation by intermittent feeding; that is, different filling times were applied to SBR
reactors to vary the feast-fast cycle. A feast-fast cycle or pulse feeding of the SBR favored
the formation of compact and dense aerobic granules. Under starvation conditions, bacteria
became more hydrophobic and this in turn facilitated microbial adhesion and aggregation
(Bossier and Verstraete, 1996).

2.2.9. Dissolved oxygen, pH and temperature


Dissolved oxygen (DO) concentration is an important variable that influences the
operation of aerobic wastewater treatment systems. Aerobic granules have formed at DO
concentration as low as 0.7 to 1.0 mg/l in a SBR (Peng et al., 1999). In addition, aerobic
granules have been successfully developed at DO concentrations of >2 mg/l (Tay et al.,
2002c; Yang et al., in press). It appears, therefore, that DO concentration is not a decisive
variable in the formation of aerobic granules. Concerning the roles of the reactor pH and
540 Y. Liu, J.-H. Tay / Biotechnology Advances 22 (2004) 533–563

temperature on aerobic granulation, detailed studies are lacking. Our unpublished work
suggests that these effects are not as important in aerobic granulation as they are in
anaerobic granulation.

2.2.10. Seed sludge


Aerobic granular sludge SBRs have been seeded with conventional activated sludge.
In anaerobic granulation, there is evidence that the characteristics of the seed sludge
profoundly influence the formation and properties of anaerobic granules. The important
factors that determine the quality of seed sludge for aerobic granulation appear to
include the macroscopic characteristics, settleability, surface properties (a high surface
hydrophobicity and low surface charge density are preferred), and microbial activity.
Little information is available on the role of seed sludge in aerobic granulation.

2.2.11. Reactor configuration


In almost all cases reported, aerobic granules were produced in column-type upflow
reactors. Reactor configuration has an impact on the flow pattern of liquid and microbial
aggregates in the reactor (Beun et al., 1999; Liu and Tay, 2002). Column-type upflow
reactor and completely mixed tank reactor (CMTR) have very different hydrodynamic
behaviors in terms of the interaction between flow and microbial aggregates. The air or
liquid upflow in column reactors can create a relatively homogenous circular flow and
localized vortexing along the reactor’s axis and microbial aggregates are constantly subject
to a hydraulic attrition. The circular flow apparently forces the microbial aggregates to
adapt a regular granular shape that has a minimum surface free energy. In a column-type
upflow reactor a high ratio of reactor height to diameter (H/D) can ensure a longer circular
flow trajectory which in turn provides a more effective hydraulic attrition to microbial
aggregates. However, in CMTRs microbial aggregates stochastically move with dispersed
flow in all directions. Thus, microbial aggregates are subject to varying localized
hydrodynamic shear force, upflow trajectories and random collisions. Under such circum-
stances, only flocs with irregular shape and size instead of regular granules occasionally
form (Liu and Tay, 2002). For practical applications, the SBR should have a high H/D ratio
to improve selection of granules by the difference in settling velocity (Beun et al., 1999).
A high H/D ratio and the absence of an external settler result in a reactor with a small
footprint.

2.2.12. Inhibition to aerobic granulation


Yang et al. (2004a,b) investigated the inhibitory effect of free ammonia on aerobic
granulation in a SBR fed with acetate as the sole carbon source. Aerobic granules formed
only when the free ammonia concentration was less than 23.5 mg/l and nitrification was
completely inhibited at a free ammonia concentration of >10 mg/l. The specific oxygen
utilization rates (SOURs) of heterotrophic and nitrifying bacteria were reduced by a factor
2.5 and 5.0 as the free ammonia concentration increased from 2.5 to 39.6 mg N/l. The high
free ammonia concentration resulted in a significant decrease of cell hydrophobicity and
also repressed the production of extracellular polysaccharides. Changes in hydrophobicity
and polysaccharide production were likely responsible for the failure of aerobic granula-
tion at high free ammonia concentrations. Yang et al. (2004b) demonstrated that free
Y. Liu, J.-H. Tay / Biotechnology Advances 22 (2004) 533–563 541

ammonia could hinder the formation of aerobic granules by inhibiting the energy
metabolism of microorganisms.

2.3. Characteristics of aerobic granules

Compared to the loose, fluffy, and irregular conventional activated sludge flocs, the
aerobic granular sludge is known to: (i) have denser and stronger microbial structure; (ii)
have regular, smooth round shape, and a clear outer surface; (iii) be visible as separate
entities in the mixed liquor during both the mixing and the settling phases; (iv) have a high
biomass retention and excellent settleability; (v) be capable to withstanding high flow
rates; (vi) be able to withstand high organic loading rates; (vii) be less vulnerable than the
suspended sludge to the toxicity of organic chemicals and heavy metals in wastewater. The
excellent settleability of aerobic granules simplifies the separation of treated effluent from
the granular sludge.

2.3.1. Morphology
Microscopic examination shows that the morphology of the aerobic granular sludge is
completely different from the floclike sludge. The shape of the granules is nearly
spherical with a very clear outline (Peng et al., 1999; Tay et al., 2001a,c; Zhu and
Wilderer, 2003). The granule size is an important parameter in the characterization of
aerobic granulation. The average diameter of aerobic granules varies in the range of 0.2
to 5 mm. This is mainly due to a balance between growth and abrasive detachment due
to the relatively strong hydrodynamic shear force in aerobic reactors (Liu and Tay, 2002;
Liu et al., 2003g). Hydrodynamic shear forces are known to control the prevailing size
of the suspended biosolids in many situations (Chisti, 1999a). Methods of estimating the
magnitudes of these forces under various conditions of operation have been discussed by
Chisti (1999a).

2.3.2. Settleability
The settling properties of aerobic granules determine the efficiency of solid – liquid
separation that is essential for the proper functioning of wastewater treatment systems.
The sludge volume index (SVI) of aerobic granules can be lower than 50 ml/g, which is
much lower than that of conventional bioflocs (Liu et al., 2003f; Qin et al., 2004). This
implies that from an engineering perspective, the settleability of sludge can be improved
significantly through the formation of aerobic granules so that it can be settled in a
more compact clarifier. The settling velocity of aerobic granules is associated with
granule size and structure and is as high as 30 to 70 m/h. This is comparable with that
of the UASB granules, but is at least three times higher than that of activated sludge
flocs (typical settling velocity of around 8 to 10 m/h). The high settling velocities of
aerobic granules allow the use of relatively high hydraulic loads to the reactors without
having to worry about washout of biomass (Beun et al., 2000; Tay et al., 2001b). Thus,
aerobic granulation can lead to more biomass retention in the reactor and this can
enhance the performance and stability of the reactor. A high concentration of the
retained biomass ensures a faster degradation of pollutants and relatively compact
reactors.
542 Y. Liu, J.-H. Tay / Biotechnology Advances 22 (2004) 533–563

2.3.3. Density and strength


The specific gravity of aerobic granules typically ranges from 1.004 to 1.065 (Etterer
and Wilderer, 2001; Tay et al., 2001a; Yang et al., in press). The granules with a high
physical strength withstand high abrasion and shear. The physical strength, expressed as
integrity coefficient (i.e., the ratio of residual granules to the total weight of the granular
sludge after 5 min of shaking at 200 rpm on a platform shaker, expressed as percent), is
higher than 95% for the aerobic granules grown on glucose and acetate (Tay et al., 2002c).
The physical strengths of aerobic and anaerobic granules are comparable. Aerobic
granules with smaller sizes tend to be more compact compared to larger aerobic granules
(Toh et al., 2003; Yang et al., 2004a).

2.3.4. Cell surface hydrophobicity


Cell surface hydrophobicity is an important affinity force in cell self-immobilization
and attachment processes (Pringle and Fletcher, 1983; Kos et al., 2003; Liu et al., 2003b).
The role of cell surface hydrophobicity in the formation of aerobic granules has not been
clear. Liu et al. (2003b) have linked the formation of heterotrophic and nitrifying granules
to the cell surface hydrophobicity. The hydrophobicity of granular sludge was nearly
twofold higher than that of conventional bioflocs. A high shear force or hydraulic selection
pressure imposed on microorganisms resulted in a significant increase in the cell surface
hydrophobicity, while the cell surface hydrophobicity seemed not to be sensitive to the
changes in the organic concentrations or loading rates in the range of 500 to 3000 mg
COD/l.

2.3.5. Specific oxygen utilization rate


In the environmental engineering field, microbial activity of microorganisms is
characterized by the specific oxygen utilization rate (SOUR). A very wide range of SOUR
values for aerobic granules have been reported (Morgenroth et al., 1997; Tay et al., 2001b;
Yang et al., 2003b; Zhu and Wilderer, 2003). The SOUR has been found to increase with
the increase in shear force specified in terms of superficial air velocity. It is obvious that
the increased shear force can stimulate the respiration activities of microorganisms in a
very significant manner (Tay et al., 2001b). This may have to do with the effect of
hydrodynamic shear on increasing the rate of oxygen transfer at the granule– liquid
interface (Chisti, 1999b). The biochemical reactions associated with bacterial metabolism
show in an approximately linear relationship between oxygen utilization and carbon
dioxide production: that is, relatively less cell mass is produced at high oxygen utilization
as the metabolism is faster and more of the substrate is converted to carbon dioxide. The
microbial activity represented by SOUR is inversely related to the hydraulic selection
pressure in terms of the settling time (Qin et al., 2004). The shorter settling time tends to
significantly stimulate the respiratory activity of these bacteria. This implies that the
microorganisms attempt to regulate their energy metabolism in response to changes in
hydraulic selection pressure.
During long storage of aerobic granules under anaerobic conditions, SOUR is used as
an indicator for evaluating the metabolic activity of the granular sludge. A decreased
SOUR indicates a reduced metabolic activity of granules after storage (Tay et al., 2002c;
Zhu and Wilderer, 2003). Once a reactor restarts, the activity of the granules returns to
Y. Liu, J.-H. Tay / Biotechnology Advances 22 (2004) 533–563 543

normal within a few days. When treating the wastewater containing toxic chemicals such
as phenol, SVI and SOUR values are necessary for monitoring the physiology changes of
granule. Biomass growth and substrate removal are linked to batch measurement of
SOUR. From the SOUR values, we can estimate the highest permissible loading rate.
Therefore, SOUR is an important characteristic for assessing the ability of aerobic granules
to handle high-strength industrial wastewaters.

2.3.6. Storage stability


The loss of granule stability and activity during an extended idling period is related to
the storage temperature. A high storage temperature accompanied with the absence of
external substrate can lead to endogenous respiration and a rapid disintegration of the
granules. Tay et al. (2002c) examined the effect of storage period of glucose- and acetate-
fed aerobic granules on granule activity and structure and found that the aerobic granules
grown on glucose only lost about 60% of initial metabolic activity in terms of specific
oxygen utilization rate (SOUR). In comparison, the activity of the acetate-fed granules was
reduced by about 90%. Broth granules were stored for 4 months in tap water at 4 jC. Zhu
and Wilderer (2003) found that after 7 weeks of storage of aerobic granules in ambient
environment, the granules could regain microbial activity within a week.
Ng (2002) studied the effect of different storage solutions, e.g., tap water, physiological
and nutrient solutions, on the stability and activity of aerobic granules stored for a period
of 8 weeks at 4 jC. The granules stored in all storage media became more irregular and
smaller at the end of week 8 as compared to fresh granules at week 0 and released soluble
organic material with an accompanying drop in the solution pH due to cell hydrolysis. A
partial desegregation of granules was observed. The decay rate of aerobic granules during
the anaerobic storage appears to correlate with the content of volatile solids (VS) in the
granules. In summary, the loss of granule activity and structural integrity during storage
depend on the storage temperature, duration, the storage medium, and the characteristics of
the granules.
Compared to fresh granules, the strength of the stored granules has been observed to
decrease by 7 –8% for glucose- and acetate-fed aerobic granules after 4 months storage
at 4 jC (Tay et al., 2002c). However, the size of aerobic granules stored in tap water and
physiological solution decreased by 34%, and 22%, respectively, at the end of 8-week
storage (Ng, 2002). In addition, the apparent color of aerobic granules stored in tap
water and physiological solution turned from brownish-yellowish (fresh granules) to
gray and dark black at the end of storage. The granules that were stored in the phosphate
buffered saline (PBS) experienced the least change in color. It is reasonable to expect
that the change of apparent color during storage results from anaerobic metabolism of
the granules.
Zhu and Wilderer (2003) reported that aerobic granular sludge could be stored for up to
7 weeks without loss of integrity and metabolic potential of the granules; that is, even
without the substrate and oxygen supply, the granular sludge remained intact. The
metabolic activity of the stored aerobic granules recovered progressively as soon as the
reactor was resupplied with substrate and oxygen. This implies that aerobic granular
sludge can be kept on standby in case the wastewater flow and concentration fall below the
normal levels or when toxic effects occur occasionally. Clearly, the use of aerobic
544 Y. Liu, J.-H. Tay / Biotechnology Advances 22 (2004) 533–563

granulation is applicable to situations where substrate availability varies greatly with time.
This is a very common scenario.

2.4. Microbial structure and diversity

In wastewater treatment system, studying the microbial structure and diversity is


essential for a comprehensive understanding of the microbial community. However, such
a study is quite complicated. This is especially so in the microbial ecology of aerobic
granules.

2.4.1. Microbial structure


Confocal laser-scanning microscopy (CLSM) has been used with different oligonucle-
otide probes, specific fluorochromes, and fluorescent microspheres for studying the
microstructure of aerobic granules (Tay et al., 2002d, 2003a; Toh et al., 2003; Jang et
al., 2003; Meyer et al., 2003). The obligate aerobic ammonium-oxidizing bacterium
Nitrosomonas spp. was found mainly at a depth of 70 to 100 Am from the granule surface,
and aerobic granules contained channels and pores that penetrated to a depth of 900 Am
below the granule surface. The porosity peaked at depths of 300 to 500 Am from the
granule surface (Tay et al., 2002d, 2003a). These channels and pores would facilitate the
transport of oxygen and nutrients into and metabolites out of the granules. Polysaccharide
formation peaked at a depth of 400 Am below the granule surface. The anaerobic
bacterium Bacteroides spp. also detected at a depth of 800 to 900 Am from the granule
surface (Tay et al., 2002e), while a layer of dead microbial cells was located at a depth of
800 to 1000 Am (Toh et al., 2003). In order to fully utilize the aerobic microorganisms in
the granules, the optimal diameter for aerobic granules should be less than 1600 Am, which
is twice the distance from the granule surface to the anaerobic layer (Tay et al., 2002d).
Consequently, smaller granules will be more effective for aerobic wastewater treatment as
these granules have more live cells within a given volume of granules.
More recently, Liu et al. (2004) observed the mushroomlike structure of aerobic
granules developed at high substrate N/COD ratios. The nitrifying population was mainly
located at a depth of 70 to 100 Am from the surface of the granule (Tay et al., 2002d).
Research has shown that biofilms of mixed bacterial communities can form thick layers
consisting of differentiated mushroomlike structures (Costerton et al., 1981) that are
similar to structures observed in aerobic granules. There is strong evidence that bacteria
sense and move towards nutrients (Prescott et al., 1999). It has been demonstrated that
biofilms can form the mushroomlike structures by simply changing the diffusion rate;
that is, the biofilm structure is largely determined by nutrient concentration (Wimpenny
and Colasanti, 1997). Since nitrifying bacteria are slow-growing, the mushroomlike
structures seem to result from the demand of nitrifying population for nutrients. These
structures improve the access of the nitrifying population to nutrients. As Watnick and
Kolter (2000) noted, in mixed biofilms, bacteria distribute themselves according to who
can survive best in the particular microenvironment and the high complexity of the
resulting microbial community appears to be beneficial to its stability. Consequently, the
distribution of different microbial populations in a granule may have an effect on its
stability.
Y. Liu, J.-H. Tay / Biotechnology Advances 22 (2004) 533–563 545

2.4.2. Microbial diversity


Microbial diversity of aerobic granules has been studied by molecular biotechnology
techniques (Yi et al., 2003; Tay et al., 2002d; Jang et al., 2003; Meyer et al., 2003; Tsuneda
et al., 2003). Heterotrophic, nitrifying, denitrifying, P-accumulating bacteria, and glyco-
gen-accumulating bacteria have been identified in aerobic granules developed under
different conditions (Jang et al., 2003; Meyer et al., 2003; Tsuneda et al., 2003; Lin et
al., 2003; Liu et al., 2003f; Yang et al., 2003b). The microbial diversity of aerobic granules
is closely related to the composition of culture media, in which they are developed and
structure of aerobic granules. Anaerobiosis and dead cells have been documented at the
centers of aerobic granules (Tay et al., 2002e). The presence of anaerobic bacteria in
aerobic granules is likely to result in the production of organic acids and gases within the
granules. These end products of anaerobic metabolism can destroy the granules or at least
diminish their long-term stability.

2.5. Mechanisms of aerobic granulation

For bacteria to form aerobic granules a number of conditions need to be met and the
physical, chemical, and biological forces contributing to granulation need to be viewed in
combination. Liu and Tay (2002) proposed a model for the aerobic granulation as
consisting of the following steps. Step 1. Physical movement to initiate bacterium-to-
bacterium contact. The factors involved in this step are hydrodynamics, diffusion mass
transfer, gravity, thermodynamic effects, and cell mobility. Step 2. Stabilization of the
multicell contacts resulting from the initial attractive forces. These attractive forces are
physical forces (e.g., Van der Waals forces, opposite charge attraction, thermodynamically
driven reduction of the surface free energy, surface tension, hydrophobicity, filamentous
bacteria that can bridge individual cells), chemical forces, and biochemical forces
including cell surface dehydration, cell membrane fusion, signaling, and collective action
in bacterial community. Step 3. Maturation of cell aggregation through production of
extracellular polymer, growth of cellular clusters, metabolic change, environment-induced
genetic effects that facilitate the cell – cell interaction and result in a highly organized
microbial structure. Step 4. Shaping of the steady state three-dimensional structure of
microbial aggregate by hydrodynamic shear forces (Chisti, 1999a).
Cell surface hydrophobicity may play a crucial role in the initiation of aerobic
granulation (Liu et al., 2003b). According to the thermodynamic theory, increasing cell
surface hydrophobicity would cause a corresponding decrease in the excess Gibbs energy
of the surface and promote cell – cell interaction to further drive the self-aggregation of
bacteria out of suspending liquid (hydrophilic phase). Hydrophobic binding is considered
of prime importance for cell – cell attachment (Pringle and Fletcher, 1983; Bos et al., 1999;
Liu et al., 2003h). A high cell surface hydrophobicity would result in a stronger cell – cell
interaction and the formation of a denser structure. Extracellular polysaccharides can
mediate both cohesion and adhesion of cells and play a crucial role in maintaining the
structural integrity in a community of immobilized cells. The polysaccharide contents of
aerobic granules tend to be much higher than that of sludge flocs (Tay et al., 2001c). Cell
polysaccharides also contribute greatly to aerobic granulation. Qin et al. (2004) observed
that aerobic granules were successfully cultivated in the SBRs operated at a settling time
546 Y. Liu, J.-H. Tay / Biotechnology Advances 22 (2004) 533–563

of < 15 min, while only bioflocs appeared in the reactor run at the a settling time of 20 min.
The shorter settling time was seen to significantly improve the production of cell
polysaccharide. A feature of the SBR is cyclic operation and the settling time acts as
hydraulic selection pressure on the microorganisms. Selection pressure can be used to
induce microbial changes that favor the formation of aerobic granules.
Although mechanisms and models for aerobic granulation have been described, they do
not provide a complete picture of the granulation process. Intercellular communication and
multicell coordination are known to contribute to the organization of bacteria into spatial
structures. Quorum sensing has been shown to be one example of social behavior in
bacteria, as signal exchange among individual cells allows the entire population to choose
an optimal way of interacting with the environment.
The cellular automaton model shows that biofilm structure is determined by localized
substrate concentration (Wimpenny and Colasanti, 1997). A cell can determine its position
in a concentration gradient of an extracellular signal factor and uses this to modify its
development (Gurdon and Bourillot, 2001). Research on cell –cell communication (Davies
et al., 1998; Pratt and Kolter, 1999; Ben-Jacob et al., 2000) confirms that cell – cell
signaling is effective in developing aerobic granules and organizing the spatial distribution
of the bacteria in the granules. Quorum-sensing effects in aerobic granules need to be
further examined.

2.6. Applications of aerobic granulation technology

The performance of a biological system for wastewater treatment depends significantly


on the active biomass concentration, the overall biodegradation rates, the reactor configu-
ration, and the feeding rates of the pollutants and oxygen. Process efficiency of large-scale
treatment plants can be improved by using aerobic granular sludge in ways that allow high
conversion rates and efficient biomass separation to minimize the reactor volume. Treatment
capacities can be easily varied to accommodate varying loading rates, wastewater compo-
sition, and treatment goals by bioaugmentation with specifically developed granules.

2.6.1. High-strength organic wastewater treatment


Granulation of the sludge can lead to high biomass retention in the reactor because of
the compact and dense structure of the granules. Biomass concentrations as high as 6.0 to
12.0 g/l have been obtained in SBRs operated with a volumetric exchange ratio of 50%
(Tay et al., 2002a,c). The feasibility of applying aerobic granulation technology for the
treatment of high-strength organic wastewaters was demonstrated by Moy et al. (2002),
who examined the ability of aerobic granules to sustain high organic loading rates by
introducing step increases in organic loading only after the COD removal efficiencies had
stabilized at values of >89% for at least 2 weeks. Aerobic granules cultivated this way on
glucose were exposed to organic loading rates that were gradually raised from 6.0 to 9.0,
12.0, and 15.0 kg COD/m3 day. Aerobic granules were able to sustain the maximum
organic loading rate of 15.0 kg COD/m3 day while removing more than 92% of the COD.
The granules initially exhibited a fluffy loose morphology dominated by filamentous
bacteria at low loadings and evolved into smooth irregular shapes characterized by folds,
crevices, and depressions at higher loadings. These irregularities were thought to allow for
Y. Liu, J.-H. Tay / Biotechnology Advances 22 (2004) 533–563 547

better diffusion and penetration of nutrients into the interior of the granule. Mass transfer
of nutrients was also enhanced by the higher substrate concentration that existed in the
bulk wastewater at higher loadings. These factors enabled the aerobic granules to sustain
high organic loading rates without compromising the granule integrity.

2.6.2. Simultaneous organic and nitrogen removal


Complete nitrogen removal involves nitrification and denitrification. Nitrite and nitrate
produced from nitrification are reduced to gaseous nitrogen by denitrifiers. Yang et al. (in
press, 2004a) investigated the simultaneous removal of organics and nitrogen by aerobic
granules. Heterotrophic, nitrifying, and denitrifying populations were shown to success-
fully coexist in microbial granules. Increased substrate N/COD ratio led to significant shifts
among the three populations within the granules. Coexistence of heterotrophic and
nitrifying populations in aerobic granules was also observed by Jang et al. (2003).
Enhanced activities of nitrifying and denitrifying populations were achieved in granules
developed at high substrate N/COD ratios; however, the heterotrophic populations in
granules tended to decrease with the increase of substrate N/COD ratio. Dissolved oxygen
(DO) concentration had a pronounced effect on the efficiency of denitrification by
microbial granules and a certain level of mixing was necessary for ensuring sufficiency
of mass transfer between the liquid and granules during denitrification (Yang et al., 2003b).
Similar phenomena were reported by Beun et al. (2001). It appears that complete organics
and nitrogen removal can be efficiently and stably achieved in a single granules-based SBR.

2.6.3. Phosphorous removal


Environmental regulations in many jurisdictions require a reduction of phosphorus
concentration in wastewater to levels of 0.5 to 2.0 mg/l before discharge. The well-known
enhanced biological phosphorus removal (EBPR) process removes P without the use of
chemical precipitation and is an economical and reliable option for P removal from
wastewater. The EBPR process operates on the basis of alternating anaerobic and aerobic
conditions with substrates feeding limited to the anaerobic stage. Most EBPR processes
are based on suspended biomass cultures and require large reactor volumes. Although full-
scale experience shows a strong potential of the EBPR, difficulties in assuring stable and
reliable operation have also been recognized. Often, the reasons for failure of biological
phosphorus removal are not clear (Barnard et al., 1985; Bitton, 1999).
In attempts to overcome the problems associated with the conventional bioremoval of P,
Lin et al. (2003) successfully developed phosphorus-accumulating microbial granules in
SBRs operated at substrate P/COD ratios ranging from 1/100 to 10/100 by weight. The
soluble COD and PO4 – P profiles showed that the granules had typical P-accumulating
characteristics, with concomitant uptake of soluble organic carbon and the release of
phosphate in the anaerobic stage, followed by rapid phosphate uptake in the aerobic stage.
The size of P-accumulating granules exhibited a decreasing trend with the increase of
substrate P/COD ratio. The structure of the granules became more compact and dense as
the substrate P/COD ratio increased. The P uptake by granules was in the range of 1.9% to
9.3% by weight, or comparable to that of the conventional enhanced biological phospho-
rus removal (EBPR) processes. These results will certainly spur the further development of
novel granule-based EBPR technologies.
548 Y. Liu, J.-H. Tay / Biotechnology Advances 22 (2004) 533–563

2.6.4. Phenolic wastewater treatment


Phenol is a toxic and inhibitory substrate, but also a carbon source for the bacteria. The
consequence of the presence of phenol in biological wastewater treatment is process
instability, which can lead to the washout of the microorganisms (Allsop et al., 1993). In
low concentrations, phenol is biodegradable, but high concentrations can kill phenol-
degrading bacteria. Industrial wastewaters from fossil fuel refining, pharmaceutical and
pesticide processing are the major sources of phenolic pollution. Jiang et al. (2002, 2004)
investigated the feasibility of treating phenol-containing wastewater with aerobic granular
sludge. Granular sludge is less susceptible to toxicity of phenol because much of the
biomass in the granules is not exposed to the same high concentration as present in the
wastewater. The phenol-degrading aerobic granules displayed an excellent ability to
degrade phenol (Jiang et al., 2002, 2004). For an influent phenol concentration of 500
mg/l, a stable effluent phenol concentration of less than 0.2 mg/l was achieved in the
aerobic granular sludge reactor (Jiang et al., 2002, 2004). The high tolerance of aerobic
granules to phenol can be exploited in developing compact high-rate treatment systems for
wastewaters loaded with a high concentration of phenol. Aerobic granules may prove
powerful bioagents for removing other inhibitory and toxic organic compounds from high-
strength industrial wastewaters. Aerobic granules appear to be highly tolerant of toxic
heavy metals (Xie, 2003).

2.6.5. Biosorption of heavy metals by aerobic granules


Heavy metals are often found in a wide variety of industrial wastewaters. More
stringent metal concentration limits are being established in view of their relatively high
toxicity. Many biomaterials have been tested as biosorbents for heavy metal removal.
These include marine algae, fungi, waste activated sludge, and digested sludge (Lodi et al.,
1998; Taniguchi et al., 2000; Valdman and Leite, 2000). In view of the physical
characteristics of aerobic granules as discussed earlier, these granules are ideal biosorbent
for heavy metals. The granules are physically strong and have large surface area and high
porosity for adsorption. In addition, the granules can be easily separated from the liquid
phase after biosorption capacity is exhausted. The biosorption of Zn2 + and Cd2 + by
aerobic granules has been reported (Liu et al., 2002, 2003c,d). The biosorption of Zn2 +
was shown to relate to both the initial Zn2 + and granule concentrations (Liu et al., 2002).
The concentration gradient of Zn2 + was the main driving force for Zn2 + biosorption by the
granules. The maximum biosorption capacity for Zn2 + was 270 mg/g of granules. For
Cd2 +, this capacity was 566 mg/g (Liu et al., 2003d).

3. Anaerobic granulation technology

3.1. Mechanisms and model for anaerobic granulation

A number of models for anaerobic granulation have been developed over the past 20
years to enhance the understanding of the mechanisms of anaerobic granulation. These
models mainly include the inert nuclei model, divalent cation-bridge model, proton
translocation – dehydration model, extracellular polymer model, ‘‘Spaghetti’’ model,
Y. Liu, J.-H. Tay / Biotechnology Advances 22 (2004) 533–563 549

syntrophic microcolony model, thermodynamic models, quorum sensing model, etc. These
models are discussed by Liu et al. (2003i). The factors influencing anaerobic granulation
are discussed next.

3.2. Factors influencing anaerobic granulation and its performance

A major problem associated with the upflow anaerobic sludge blanket (UASB) reactors
is the long start-up period (2 –4 months or longer) required for the development of
anaerobic granules. Enhanced granulation processes are desirable to reduce the space time
requirements of bioreactors. This section discusses the main factors that influence
anaerobic granulation and its performance.

3.2.1. Upflow liquid velocity and hydraulic retention time


It has been observed that anaerobic granulation can proceed well at relatively high
liquid upflow velocity, but does not occur under conditions of low hydrodynamic shear
(Alphenaar et al., 1993; Arcand et al., 1994; O’Flaherty et al., 1997; Alves et al., 2000).
According to Alphenaar et al. (1993), granulation in UASB reactors is favored by a
combination of high upflow liquid velocity and short hydraulic retention time (HRT).
Usually, the effects of upflow liquid velocity on anaerobic granulation are explained by the
selection pressure theory (Hulshoff Pol et al., 1988). A long HRT accompanied with a low
upflow liquid velocity may allow dispersed bacterial growth and be less favorable for
microbe granulation. In contrast, a short HRT combined with a high upflow liquid velocity
can lead to washout of nongranulation competent bacteria and thus promote sludge
granulation. These findings are consistent with those reported for aerobic granulation.
Evidence shows that flocculent anaerobic sludge can be converted to a relatively active
anaerobic granular sludge by manipulating hydraulic stress and the settleability of the
anaerobic granules. Granules with improved SVI and sludge settling velocity develop
when the upflow liquid velocity is increased (Noyola and Mereno, 1994). While the
upflow liquid velocity has a significant positive effect on the mean granule size, its effect
on the specific washout rate of the smaller particles is insignificant (Arcand et al., 1994).

3.2.2. Organic loading rate


The organic substrate loading rate (OLR) describes the degree of starvation of the
microorganisms in a biological system. At a low OLR, microorganisms are subject to
nutrient starvation, while a high OLR sustains fast microbial growth (Bitton, 1999).
Evidence shows that anaerobic granulation can be accomplished by gradually increasing
the OLR during the start-up (Hulshoff Pol, 1989; Kosaric et al., 1990; Campos and
Anderson, 1992; Tay and Yan, 1996). It is critical to select a reasonably high OLR during
start-up, to ensure rapid granulation and a stable treatment process. A simple and practical
strategy for rapid start-up of anaerobic granular sludge reactors is to increase the OLR to
attain only 80% reduction of biodegradable chemical oxygen demand (COD) with
supplementary monitoring of effluent for washout of suspended solids (de Zeeuw, 1988;
Fang and Chui, 1993). However, if the applied OLR is too high during start-up of UASB
reactors, the biogas production rate increases and this causes hydrodynamic turbulence and
can lead to washout of the seed sludge from the reactor. This often leads to unsuccessful
550 Y. Liu, J.-H. Tay / Biotechnology Advances 22 (2004) 533–563

start-up of a UASB reactor. As discussed earlier, compared to aerobic granulation, the


anaerobic granulation is highly sensitive to the organic loading applied. This is probably
because of different growth characteristics of aerobic and anaerobic bacteria.

3.2.3. Characteristics of feed


Characteristics of the feed are considered a key factor influencing the formation,
composition, and structure of anaerobic granules. As discussed earlier, the effect of feed on
aerobic granulation seems to be insignificant because of fast growth of aerobic bacteria.
Based on their free energy of oxidation, organic substrates can be roughly classified into
high-energy and low-energy feeds. During the UASB start-up period, high-energy
carbohydrate feeding can sustain the acidogens and facilitate the formation of extracellular
polymers. The more readily the acidogens take up and metabolize the substrate, the more
rapidly the proton pumps will be activated, and sooner the methanogens will obtain the
substrate (Tay et al., 2000). Thus, the rapid growth of acidogens due to the presence of
high-energy substrate in the influent would facilitate the overall process of sludge
granulation in the UASB reactors.
The granules grown on volatile fatty acid mixture (acetate, propionate, and butyrate)
under mesophilic conditions can be classified into three distinct types according to the
predominant acetate utilizing methanogens present: (1) rod-type granules, which are
mainly composed of rod-shaped bacteria in fragments of about four to five cells
resembling Methanothrix; (2) filament-type granules, which consist predominantly of
long multicellular rod-shaped bacteria; and (3) sarcina-type granules, which develop when
a high concentration of acetic acid is maintained in the reactor (Hulshoff Pol et al., 1983;
de Zeeuw, 1984). A trend has been observed towards increasing diversity of methanogenic
subpopulations with an increasing complexity of the waste composition. At least four
distinct microcolonies have been observed in granules treating brewery wastewater (Wu,
1991). One of these microcolonies was composed of Methanothrix-like rods only, while
the other microcolonies consisted of hydrogen – carbon dioxide utilizing Methanobacte-
rium-like rods juxtapositioned with three different rods-shaped synthrophs (Hickey, 1991).
Full-scale UASB experience confirms that anaerobic sludge granulation occurs in many
different types of wastewaters. Because of the extremely low growth rate of anaerobic
bacteria, the energy content of the substrate are important for anaerobic granulation;
however, the complexity of substrate also exerts a selection pressure on the microbial
diversity in anaerobic granules. This selection pressure may in turn influence the formation
and microstructure of granules through its effect on the food chain and community
signaling communications.

3.2.4. Seed sludge


Theoretically any medium containing the proper bacterial flora can be used as seed
sludge for a UASB reactor. Potential seed sludges include manure, fresh water sediments,
septic tank sludge, digested sewage sludge, and surplus sludge from anaerobic treatment
plants. The start-up of UASB reactors using digested sewage sludge as seed may take
several months (de Zeeuw, 1984). A long start-up period is the major deterrent to the use
of UASB systems. With respect to the use of digested sewage sludge as seed, de Zeeuw
(1984) found that heavy, relatively inactive sludge was preferred over lighter, more active
Y. Liu, J.-H. Tay / Biotechnology Advances 22 (2004) 533–563 551

sludge because of differences in washout. Two types of sludge washouts were distin-
guished. These were erosion washout and sludge bed washout. Sludge bed erosion
washout is a selective washout based on differences in settleability. Although digested
sewage sludge is usually used for the start-up of UASB reactors, other types of seed
sludges have been used successfully when granular sludge for seeding is unavailable. Wu
et al. (1987) used aerobic activated sludge from a sewage treatment plant and primary
sludge from an aerobic plant treating textile dyeing wastewater for start-up.
Addition of a small amount of granules to nongranular inoculum is proven to stimulate
the granulation process (Hulshoff Pol et al., 1983; Xu and Tay, 2002). This is probably a
consequence of supplying a specific inoculum that is responsible for granulation. UASB
systems can also be started up using existing granules. This lends an advantage to the
UASB start-up process, although a successful start-up is not dependent on the use of
precultivated granules. When feasible, inoculation with a large amount of seed granular
sludge from a healthy UASB reactor is desirable. However, the availability of granular
seed sludge is limited and the purchase and transport of the inoculum can be expensive.
Hulshoff Pol et al. (1983) reported that the addition of crushed granular methanogenic
sludge to digested sewage in a UASB reactor fed with acetate and propionate can give rise
to the development of methanogenic sludge granules with a diameter of 1 to 2 mm.
Two different types of sludges may develop on the same medium depending on the
source of the inoculum. Xu and Tay (2002) used methanol-precultured anaerobic sludge to
inoculate a UASB reactor. This approach accelerated the formation of embryonic granules
in a laboratory-scale UASB reactor. The granulation process reached its postmaturation
stage about 15 to 20 days ahead of the control reactor. Use of methanogen-enriched seed
sludge for UASB inoculation can reduce the time required for start-up. It seems certain
that anaerobic granulation can be expedited simply by manipulating the composition of
seed sludge. This approach can greatly facilitate the start-up of UASB reactors. It is still
not entirely clear as to which species in seed sludge contribute the most to anaerobic
granulation.

3.2.5. Addition of polymer and cations


Synthetic and natural polymers have been widely used in coagulation/flocculation
processes. These polymers are known to promote particle agglomeration and have been
used to enhance the formation of anaerobic granules. El-Mamouni et al. (1998) found that
the supplementation with the polymer chitosan (a polymer that is similar in structure to
polysaccharides) significantly enhanced the formation of anaerobic granules in the UASB-
like reactors. For example, the granulation rate in the chitosan-containing reactor was 2.5-
fold higher than that in the control reactor without the polymer supplementation. The
polymer enhanced granules had about the same specific activity of methane production as
the granules formed without the polymer. Polymeric chains enhance flocculation by
bridging microbial cells. Such initial microbial nuclei are the first step in microbial
granulation. In cationic polymer-assisted anaerobic granulation processes, it has been
observed that the start-up period required for the development of granular sludge blanket
can be shortened significantly compared to when no polymers are used (Uyanik et al.,
2002). Two mechanisms appear to be involved in polymer enhancement of anaerobic
granulation. The addition of polymers to anaerobic systems likely changes the surface
552 Y. Liu, J.-H. Tay / Biotechnology Advances 22 (2004) 533–563

properties of bacteria to promote association of individual cells. Polymer may also form a
relatively solid and stable three-dimensional matrix within which bacteria multiply and
daughter cells are then confined. The polymer additives appear to play a similar role as do
the naturally secreted extracellular polymeric substances (EPS) in aggregating anaerobic
sludge.
Evidence shows that the presence of divalent and trivalent cations ions, such as Ca2 +,
Mg2 +, Fe2 +, and Fe3 +, helps bind negatively charged cells together to form microbial
nuclei that promote further granulation (Mahoney et al., 1987; Schmidt and Ahring, 1993;
Teo et al., 2000; Yu et al., 2001). De Zeeuw (1984) reported that the rate of sludge
granulation was significantly enhanced at a calcium concentration of 100 mg/l in the
wastewater. Similarly, Mahoney et al. (1987) observed that granule formation was
stimulated by the presence of calcium in a concentration range of 100 to 200 mg/l.
Research by Teo et al. (2000) showed that an increase in Ca2 + concentration from 0 to 80
mg/l substantially improved the strength of anaerobic granules, as indicated by a 60%
decrease in turbidity. The role of Ca2 + in anaerobic granulation processes is still uncertain.
Calcium concentrations exceeding about 500 mg/l (Guiot et al., 1988; Thiele et al., 1990;
Yu et al., 2001) are detrimental to granulation. At high calcium concentrations, problems
such as the precipitation of calcium on the surface of granules and accumulation of
calcium inside anaerobic granules with consequent reduced microbial activity have been
reported (Yu et al., 2001).

3.2.6. Temperature
As a core microbial component of anaerobic granules, methanogenic bacteria grow
slowly in wastewater and their generation times range from 3 days at 35 jC to as high as
50 days at 10 jC (Bitton, 1999). When the reactor temperature is below 30 jC, the activity
of methanogens is seriously reduced. This is the main reason why mesophilic UASB
reactors must be operated at a temperature of 30 to 35 jC for successful functioning. In
addition, sludge washout and deterioration of COD removal efficiency have been reported
in UASB reactors when temperature is increased in steps from 37 to 55 jC (Fang and Lau,
1996). Lepisto and Rintala (1999) further reported that effluent quality from a UASB
reactor operated at 70 jC was lower than that from reactors operated at 35 jC and 55 jC.
High temperatures are known to encourage the growth of suspended biosolids; however,
extremely high temperatures inhibit bacterial growth. Extreme thermophilic UASB
reactors (i.e., temperature above 55 jC) seem not to be practicable because of the
additional energy that is required to maintain the high temperature and the relatively poor
effluent quality. A high-temperature operation is also difficult to control.
Recently, attention has been given to the impact of low temperature on the performance
of anaerobic granular sludge reactors (Angenent et al., 2001; Lettinga et al., 2001; Lew et
al., 2003; Singh and Viraraghavan, 2003). Singh and Viraraghavan (2003) showed that
COD removal efficiency can be as high as 70 to 90% in a UASB reactor operated at 11 jC
with a hydraulic retention time of 6 h. Similarly, the expanded granular sludge bed (EGSB)
reactors have been shown to be practicable systems for anaerobic treatment of mainly
soluble and preacidified wastewaters at temperatures of 5 to 10 jC (Lettinga et al., 2001).
In addition, anaerobic migrating blanket reactors (AMBRs) have also been successfully
applied to treat low-strength wastewaters at low temperatures (Angenent et al., 2001).
Y. Liu, J.-H. Tay / Biotechnology Advances 22 (2004) 533–563 553

Clearly, therefore, anaerobic granular sludge systems are suitable for the treatment of
municipal wastewater at low and moderate temperatures.

3.2.7. pH
Based on the sequence of anaerobic reaction, microbial species involved can be roughly
divided into the following three categories: (a) bacteria responsible for hydrolysis; (b)
acid-producing bacteria; and (c) methane-producing bacteria. In general, the acid-produc-
ing bacteria tolerate a low pH and have an optimal pH of 5.0 to 6.0; however, most
methane-producing bacteria can function optimally in a very narrow pH range of 6.7 to 7.4
(Bitton, 1999). This explains why pH is more inhibitory to methane-producing bacteria
than to acidogenic bacteria in UASB reactors. Once the reactor pH falls outside 6.0 to 8.0,
the activity of methane-producing bacteria is reduced to a low level and this decline in
activity in turn poses serious operational problem that can lead to the failure of the reactor.
Under normal operation conditions, the pH reduction caused by acid-producing bacteria
can be buffered by bicarbonate produced by the methane-producing bacteria.

3.3. Characteristics of anaerobic granules

3.3.1. Microstructure
Based on the microscopic observations, a multilayer model for anaerobic granulation
was initially proposed by MacLeod et al. (1990) and Guiot et al. (1992). According to this
model, the microbiological composition of granules is different in each layer. The inner
layer mainly consists of methanogens that may act as nucleation centers that are necessary
for the initiation of granule development. H2-producing and H2-utilizing bacteria are
dominant species in the middle layer and a mix of species, including rods, cocci, and
filamentous bacteria, predominates in the outermost layer. The conversion of a target
organic compound to methane depends on the spatial organization of the methanogens and
other species in UASB granules. The layered structure of UASB granules is further
evidenced by immunological and histologic methods (Ahring et al., 1993; Lens et al.,
1995); dynamic models (Arcand et al., 1994); studies with microelectrodes (Santegoeds et
al., 1999); and fluorescence in situ hybridization using 16S rRNA-targeted oligonucleo-
tides (Sekiguchi et al., 1999; Tagawa et al., 2000).
A distinct layered structure has been found also in the methanogenic –sulfidogenic
aggregates, with sulfate-reducing bacteria in the outer 50 to 100 Am and methanogens in the
inner part (Santegoeds et al., 1999). Unlike the initial multilayer model proposed by
MacLeod et al. (1990), recent research shows that UASB granules have large dark
nonstaining centers in which neither archaeal nor bacterial signals are observed (Rocheleau
et al., 1999). These nonstaining centers may be formed as a result of the accumulation of
metabolically inactive, decaying biomass and inorganic material (Sekiguchi et al., 1999).
Granules with a homogeneous nonlayered structure have also been reported (Groten-
huis et al., 1991; Fang et al., 1995). Filamentous microorganisms were predominant
throughout such granules. Some researchers have argued that layered and nonlayered
microstructures can be developed with carbohydrates and proteins as substrates, respec-
tively (Fang et al., 1995). This phenomena is said to originate in the different initial steps
involved in the degradation of carbohydrates and proteins. The carbohydrate degradation
554 Y. Liu, J.-H. Tay / Biotechnology Advances 22 (2004) 533–563

to small molecules is faster than the subsequent degradation of the intermediates. In


contrast, the initial step in the protein degradation is a rate-limiting step. Furthermore,
different types of granules can also form on a given substrate (Daffonchio et al., 1995). It
has been suggested that the granular microstructure is dependent on the degradation
kinetics of the substrate. In other words, different dominating catabolic pathways may give
rise to different granules (Daffonchio et al., 1995; Schmidt and Ahring, 1996).

3.3.2. Methanogenic activity


Methane-producing bacteria are one of the major species in anaerobic granules. In most
research on anaerobic granules, the activity of methane-producing bacteria has been used
to quantify the metabolic activity of the granules. In general, the methanogenic activity can
be defined as the methane production per unit biomass per unit time or the methane
production per unit reactor volume per unit time. The methanogenic activity can be used to
evaluate the performance of a reactor and as an indicator of toxic or inhibitory effects on
anaerobic granules.

3.3.3. Physicochemical properties


Research suggests that cell surface hydrophobicity plays a crucial role in both aerobic
and anaerobic granulation (Tay et al., 2000; Teo et al., 2000; Liu et al., 2003b). Increasing
cell surface hydrophobicity causes a decrease in the excess Gibbs energy of the surface
and further facilitates cell-to-cell interactions leading to a stable structure. Some environ-
mental factors are known to influence cell surface hydrophobicity. These factors include
starvation, oxygen level, selection pressure, and the ionic strength of the medium (Rouxhet
and Mozes, 1990; Castellanos et al., 2000; Liu et al., 2003b; Qin et al., 2004).
Biosolids washed out from the UASB reactors have been found to be more hydrophilic
than the sludge retained in the reactor (Mahoney et al., 1987). This seems to indicate that
in the presence of a selection pressure such as a high liquid upflow velocity, micro-
organisms having a high surface hydrophobicity can be self-immobilized to form denser
aggregates that remain in the reactor. Anaerobic granules tend to become weaker as the
surface negative charge increases (Quarmby and Forster, 1995). In general, the surface of
microorganisms is negatively charged at physiological pH values. The maximum cell
immobilization strength was observed at the isoelectric point of the cells (Rouxhet and
Mozes, 1990; Liu, 1995). Therefore, it can be concluded that the anaerobic granulation is
closely correlated with cell surface properties.
The size of anaerobic granules has a dual effect on the performance of a UASB system.
If the granule is too small, it is likely to be easily washed out of the reactor and this leads to
operational instability. On other hand, a large size of granules reduces mass transfer within
it. The large anaerobic granules in UASB reactors are usually associated with fluffy
bacteria. These large granules can be easily lost with the effluent because of their low
density. The size and density of anaerobic granules depend on many factors including
hydrodynamic conditions, COD loading rate, and the microbial species involved. In
industrial practice, a narrow size distribution of granules is preferred and medium-sized
granules with a diameter of 1.0 to 2.0 mm seem the most attractive. The density of
anaerobic granules indicates their compactness. In UASB reactors, anaerobic granules can
grow up to 2 to 5 mm or larger and can range in specific gravity from 1.033 to 1.065
Y. Liu, J.-H. Tay / Biotechnology Advances 22 (2004) 533–563 555

(Pereboom and Vereijken, 1994; Tay and Yan, 1996). The relatively high specific gravity
of individual anaerobic granules allows them to settle rapidly. This permits good
separation of solids and liquid in a compact separator.
The strength of anaerobic granules strongly influences the stability of granules. A high
strength means a more compact and stable structure of the granules. Quarmby and Forster
(1995) found that high COD loading rates result in low-strength anaerobic granules. This
is expected. The strength of anaerobic granules depends on many factors including the
microbial diversity, organics loading rate, the feed, exopolysaccharide production, and
hydrodynamic shear forces. High-strength anaerobic granules are desirable in industrial
applications.

3.4. Anaerobic granulation in other types of reactors

Almost all research on anaerobic granulation has been carried out in UASB reactors.
The feasibility of other types of bioreactors in the development of anaerobic granular
sludge has not been sufficiently demonstrated. The reasons for this are not clear. This
section discusses some of the work done on anaerobic granules-based bioreactors other
than UASBs.

3.4.1. Anaerobic continuous stirred tank reactor


Anaerobic granules have been successfully produced in continuously stirred tank
reactors (CSTR), but they disappeared within 3 weeks when the reactor was incubated
statically instead of being agitated (Vanderhaegen et al., 1992). This seems to indicate that
hydrodynamic shear forces play a crucial role in maintaining the integrity of granular
sludge, as has been confirmed for aerobic granulation. Thus, anaerobic granulation may
not depend on the type of reactor, but on how it is operated. Of course, the hydrodynamic
flow patterns in UASB and CSTR are different and the question of how the nature of flow
affects the development of anaerobic granules remains to be answered definitively.

3.4.2. Internal circulation reactor


A major problem of UASB reactors is the washout of seed sludge during the start-up
phase. A new generation of more advanced anaerobic reactor systems has been developed
recently in attempts to overcome the washout-related operational problems of the
conventional UASB reactors. One advanced design is the ‘‘internal circulation’’ (IC)
reactor. This is characterized by biogas separation in two stages within a reactor that has a
high height-to-diameter ratio and a gas-driven internal circulation of the effluent
(Pereboom and Vereijken, 1994). The IC reactor consists of two interconnected UASB
compartments on top of each other. Most of the biogas is produced in the first stage that is
located in the bottom part of the IC reactor. The gas is trapped in gas hoods and then rises
through the riser zone to a gas –liquid separator placed on top of the reactor. The biogas
production thus drives an internal circulation flow through the airlift action (Chisti, 1998).
This results in excellent mixing in the bottom zone of the reactor. In the second stage on
top of the reactor, the biomass retention and effluent polishing take place.
Its design makes the IC reactor more capable of handling high upflow liquid and gas
velocities to allow the treatment of low-strength effluents at short hydraulic retention
556 Y. Liu, J.-H. Tay / Biotechnology Advances 22 (2004) 533–563

times. In addition, the treatment of high-strength effluents at very high volumetric loading
rates is feasible. Furthermore, the high turbulence and sufficient mixing characteristics of
the IC reactors reduce susceptibility to clogging. In general, the performance of IC reactors
is comparable with or better than that of UASBs for high-strength industrial wastewater
treatment. The IC reactor has been successfully used to commercially treat a wide variety
of wastewaters (Habets et al., 1997; Driessen and Yspeert, 1999).

3.4.3. Expanded granular sludge bed (EGSB) reactor


Anaerobic expanded and fluidized bed reactors have been developed for treating
wastewaters. These reactors rely on dense carrier particles (e.g., sand, activated carbon)
that support a microbial biofilm for degrading the waste. Expanded and fluidized bed
reactors are anaerobic biofilm systems and not granular sludge systems. Zoutberg and de
Been (1997) devised a new type of anaerobic reactor, the expanded granular sludge bed
(EGSB) reactor. The most important feature of the EGSB reactor is the sludge granulation
without carrier materials. Granular sludge could grow and be maintained under high liquid
and gas velocities in the EGSB reactor. The EGSB is suitable for treating both industrial
and municipal wastewaters (Zoutberg and de Been, 1997; Lettinga et al., 2001). However,
the maintenance of expanded sludge bed in EGSB is relatively difficult because of an
absence of solid carriers.

3.4.4. Anaerobic sequencing batch reactor (ASBR)


The major differences between ASBR and UASB are the following: (1) a feed
distribution system is not required in ASBR; (2) there is no three-phase separator in the
ASBR; (3) an upflow hydraulic pattern is absent in ASBR; and (4) the ASBR is operated
in discontinuous mode. Wirtz and Dague (1996) successfully cultivated a granular sludge
blanket in an ASBR. Angenent and Sung (2001) found that the mixed liquor volatile
suspended solids retention in the ASBR was 2.5-fold higher than that in the UASB.
Furthermore, the performance in terms of COD removal in the ASBR was comparable to
that of the UASB. Evidence suggests that at low organic loading rates, the performances of
continuous UASB and anaerobic SBR are quite similar; however, continuous UASB
reactors perform better than the anaerobic SBRs at high organic loading rates (Kennedy
and Lentz, 2000).

3.4.5. Anaerobic migrating blanket reactor


Angenent and Sung (2001) reported a novel anaerobic wastewater treatment system, the
anaerobic migrating blanket reactor (AMBR). This is a continuously fed, compartmen-
talized reactor that does not require an elaborate gas –solid separator and systems for feed
distribution. Anaerobic granules developed in the AMBR tended to be darker in color,
smaller, and denser than granules formed in a UASB reactor operated under similar
conditions. The AMBR has some advantages over the UASB reactor. For example, the
AMBR has a low biomass migration rate, less chance of short-circuiting, efficient removal
of poorly biodegradable compounds, and can be operated in step feed mode for high-
strength wastewaters during shock loads. However, the internal structure of AMBR is
more complex than for the UASB. For example, the AMBR requires multipoint
mechanical mixing to improve feed distribution and prevent clogging by sludge.
Y. Liu, J.-H. Tay / Biotechnology Advances 22 (2004) 533–563 557

3.5. Industrial application of anaerobic granulation technology

Kassam et al. (2003) analyzed global trends in the industrial use of anaerobic
wastewater treatment systems. Their data showed exponential growth in the use of
industrial anaerobic wastewater systems worldwide up to the mid-1990s. After 1994,
the number of annual installations declined, but it has remained relatively constant over the
last 3 years. As compared to conventional biological processes, the anaerobic granules-
based biosystems have the benefits of: (1) being simple in construction and operation; (2)
requiring no power from external grid; (3) being compact; (4) generating a low amount of
biological sludge; (5) having a high treatment efficiency; (6) being low in capital and
operating costs; (7) requiring no oxygen; and (8) generating methane fuel. The UASB
technology has been successfully applied to treat industrial wastewater from pulp/paper
industry, the food industry, breweries, distilleries, and the chemical industry. However, an
analysis of the market trends shows that the traditional UASB systems are being phased
out in favor of the high-capacity and high-rate systems such as the EGSB and IC (Lettinga
et al., 2001; Kassam et al., 2003).

4. Future work

This work reviewed the state of the art of biogranulation technology developed for
wastewater treatment. Although extensive work has been done in this area, future research
needs to look into some of the following aspects:

(1) In view of extremely long start-up period of anaerobic granular sludge reactors,
strategies need to be examined for accelerating anaerobic granulation.
(2) Apparently, the selection pressure is a main driving force of aerobic granulation (Qin et
al., 2004), but this needs to be elucidated conclusively.
(3) The question of whether the formation of anaerobic, aerobic granules and biofilms are
subject to the similar mechanisms needs to be addressed.
(4) Aerobic granulation has been observed only the SBRs. The feasibility of attaining
aerobic granulation in continuous culture systems needs to be investigated.
(5) Compared to anaerobic granules, aerobic granules have relatively low stability because
of their fast growth rate. Liu et al. (2004) showed that selecting slow-growing bacteria
in aerobic granules improved the stability of the granules. It is desirable to develop a
practical strategy for improving the stability of aerobic granules by manipulating
operational conditions or through selecting for slow-growth bacteria.
(6) Biogranule-associated bacteria live in a confined space. One advantage of the struc-
tured microbial granules is the ability to acquire transmissible, genetic elements at
accelerated rates. Rapid evolution by horizontal transfer of genetic material has been
observed in biofilms (Watnick and Kolter, 2000). Such transfer likely occurs also in
granules, but this needs to be examined.
(7) It would be interesting to look at the feasibility of transplanting engineered species into
microbial granules to tailor microbial granules for treating specific types of
wastewaters.
558 Y. Liu, J.-H. Tay / Biotechnology Advances 22 (2004) 533–563

Acknowledgements

M.A. Nay, S.F. Yang, and Q.S. Liu are thanked for assisting with the collection of
materials for this work.

References

Allsop PJ, Chisti Y, Moo-Young M, Sullivan GR. Dynamics of phenol degradation by Pseudomonas putida.
Biotechnol Bioeng 1993;41:572 – 80.
Alphenaar PA, Visser A, Lettinga G. The effect of liquid upflow velocity and hydraulic retention time on
granulation in UASB reactors treating wastewater with a high-sulphate content. Bioresour Technol 1993;
43:249 – 58.
Alves M, Cavaleiro AJ, Ferreira EC, Amaral AL, Mota M, da Motta M, Vivier H, Pons MN. Characterization by
image analysis of anaerobic sludge under shock conditions. Water Sci Technol 2000;41:207 – 14.
Angenent LT, Sung S. Development of anaerobic migrating blanket reactor (AMBR), a novel anaerobic treatment
system. Water Res 2001;35:1739 – 47.
Angenent LT, Banik GC, Sung SW. Anaerobic migrating blanket reactor treatment of low-strength wastewater at
low temperatures. Water Environ Res 2001;73:567 – 74.
Arcand Y, Guitot SR, Desrochers M, Chavarie C. Impact of the reactor hydrodynamics and organic loading on the
size and activity of anaerobic granules. Chem Eng J Biochem Eng J 1994;56:23 – 35.
Arching BK, Schmidt JE, Winther-Nielsen M, Macario AJL, de Macario EC. Effect of medium compo-
sition and sludge removal on the production, composition and architecture of thermophilic (55 jC)
acetate-utilizing granules from an uplow anaerobic sludge blanket reactor. Appl Environ Microbiol
1993;59:2538 – 45.
Barnard JL, Stevens GM, Leslie PL. Design strategies for nutrient removal plants. Water Sci Technol 1985;
17:233 – 42.
Ben-Jacob E, Cohen I, Levine H. Cooperative self-organization of microorganisms. Adv Phys 2000;49:395 – 554.
Beun JJ, Hendriks A, van Loosdrecht MCM, Morgenroth E, Wilderer PA, Heijnen JJ. Aerobic granulation in a
sequencing batch reactor. Water Res 1999;33:2283 – 90.
Beun JJ, van Loosdrecht MCM, Heijnen JJ. Aerobic granulation. Water Sci Technol 2000;41:41 – 8.
Beun JJ, Heijnen JJ, van Loosdrecht MCM. N-removal in a granular sludge sequencing batch airlift reactor.
Biotechnol Bioeng 2001;75:82 – 92.
Bitton G. Wastewater microbiology. New York: Wiley-Liss, 1999.
Bos R, van de Mei HC, Busscher HJ. Physico-chemistry of initial microbial adhesive interactions—its mecha-
nisms and methods for study. FEMS Microbiol Rev 1999;23:179 – 230.
Bossier P, Verstraete W. Triggers for microbial aggregation in activated sludge? Appl Microbiol Biotechnol 1996;
45:1 – 6.
Calleja GB. Microbial aggregation. Florida: CRC Press; 1984.
Campos CMM, Anderson GK. The effect of the liquid upflow velocity and the substrate concentration on the
startup and steady state period of lab-scale UASB reactors. Water Sci Technol 1992;25:41 – 50.
Castellanos T, Ascencio F, Bashan Y. Starvation-induced changes in the cell surface of Azospirillum. FEMS
Microbiol Ecol 2000;33:1 – 9.
Chisti Y. Pneumatically agitated bioreactors in industrial and environmental bioprocessing: hydrodynamics,
hydraulics and transport phenomena. Appl Mech Rev 1998;51:33 – 112.
Chisti Y. Shear sensitivity. In: Flickinger MC, Drew SW, editors. Encyclopedia of bioprocess technology:
fermentation, biocatalysis, and bioseparation, vol. 5. New York: Wiley; 1999a. p. 2379 – 406.
Chisti Y. Mass transfer. In: Flickinger MC, Drew SW, editors. Encyclopedia of bioprocess technology: fermen-
tation, biocatalysis, and bioseparation, vol. 3. New York: Wiley; 1999b. p. 1607 – 40.
Chudoba J. Control of activated sludge filamentous bulking. Water Res 1985;19:1017.
Costerton JW, Irvin RT, Cheng KJ. The bacterial glycocalyx in nature and disease. Annu Rev Microbiol 1981;
35:299 – 324.
Y. Liu, J.-H. Tay / Biotechnology Advances 22 (2004) 533–563 559

Davies DG, Parsek MR, Pearson JP, Iglewski BH, Costerton JW, Greenberg EP. The involvement of cell-to-cell
signals in the development of a bacterial biofilm. Science 1998;280:295 – 8.
Daffonchio D, Thavessri J, Verstraete W. Contact angle measurement and cell hydrophobicity of granular sludge
from upflow anaerobic sludge bed reactors. Appl Environ Microbiol 1995;61:3676 – 80.
De Zeeuw WJ. Acclimatization of anaerobic sludge for UASB reactor start-up, PhD thesis. Wageningen, The
Netherlands: Agricultural University of Wageningen; 1984.
De Zeeuw WJ. Granular sludge in UASB-reactors. In: Lettinga G, Zehnder AJB, Grotenhuis JTC, Hulshoff Pol
LW, editors. Granular ananerobic sludge: microbiology and technology. The Netherlands: Wageningen; 1988.
p. 132 – 45.
Driessen W, Yspeert P. Anaerobic treatment of low, medium and high strength effluent in the agro-industry. Water
Sci Technol 1999;40:221 – 8.
El-Mamouni R, Leduc R, Guiot SR. Influence of synthetic and natural polymers on the anaerobic granulation
process. Water Sci Technol 1998;38:341 – 7.
Etterer T, Wilderer PA. Generation and properties of aerobic granular sludge. Water Sci Technol 2001;43:19 – 26.
Fang HHP, Chui HK. Maximum COD loading capacity in UASB reactors at 37 jC. J Environ Eng 1993;119:
103 – 19.
Fang HHP, Chui HK, Li YY. Effect of degradation kinetics on the microstructure of anaerobic biogranules. Water
Sci Technol 1995;32:165 – 72.
Fang HPP, Lau IWC. Start-up of the thermophilic (55 jC) UASB reactors using different mesophilic seed sludges.
Water Sci Technol 1996;34:445 – 52.
Grotenhuis JTC, van Lier JB, Plugge CM, Stams AJM, Zehnder AJB. Effect of ethylene glycol-bis
(h-aminoethylether)-N,N-tetraacetic acid (EGTA) on stability and activity of methanogenic granular sludge.
Appl Microbiol Biotechnol 1991;36:109 – 14.
Guiot SR, Gorur SS, Bourque D, Samson R. Metal effect on microbial aggregation during upflow anaerobic
sludge bed-filter (UBF) reactor start-up. In: Lettinga G, Zehnder AJB, Grotenhuis JTC, Hulshoff Pol LW,
editors. Granular ananerobic sludge: microbiology and technology. Wageningen: GASMAT, Lunteren; 1988.
p. 187 – 94.
Guiot SR, Pauss A, Costerton JW. A structured model of the anaerobic granules consortium. Water Sci Technol
1992;25:1 – 10.
Gurdon JB, Bourillot PY. Morphogen gradient interpretation. Nature 2001;413:797 – 803.
Habets LHA, Engelaar AJHH, Groeneveld N. Anaerobic treatment of inuline effluent in an internal circulation
reactor. Water Sci Technol 1997;35:189 – 97.
Hickey RF. Startup, operation, monitoring and control of high-rate anaerobic treatment systems. Water Sci
Technol 1991;24:207 – 55.
Hulshoff Pol LW. The phenomenon of granulation of anaerobic sludge, PhD thesis. Wageningen, The Nether-
lands:Agricultural University of Wageningen; 1989.
Hulshoff Pol LW, de Zeeuw WJ, Velzebber CTM, Lettinga G. Granulation in UASB-reactors. Water Sci Technol
1983;8/9:291 – 304.
Hulshoff Pol LW, Heijnekamp K, Lettinga G. The selection pressure as a driving force behind the granulation of
anaerobic sludge. In: Lettinga G, Zehnder AJB, Grotenhuis JTC, Hulshoff Pol LW, editors. Granular ana-
nerobic sludge: microbiology and technology. Wageningen: GASMAT, Lunteren; 1988. p. 153 – 61.
Jang A, Yoon YH, Kim IS, Kim KS, Bishop PL. Characterization and evaluation of aerobic granules in sequenc-
ing batch reactor. J Biotechnol 2003;105:71 – 82.
Jiang HL, Tay JH, Tay STL. Aggregation of immobilized activated sludge cells into aerobically grown microbial
granules for the aerobic biodegradation of phenol. Lett Appl Microbiol 2002;35:439 – 45.
Jiang HL, Tay JH, Liu Y, Tay STL. Ca2 + augmentation for enhancement of aerobically grown microbial granules
in sludge blanket reactors. Biotechnol Lett 2003;25:95 – 9.
Jiang HL, Tay JH, Tay STL. Changes in structure, activity and metabolism of aerobic granules as a microbial
response to high phenol loading. Appl Microbiol Biotechnol 2004;63:602 – 8.
Kassam ZA, Yerushalmi L, Guiot SR. A market study on the anaerobic wastewater treatment systems. Water
Sci Technol 2003;143:179 – 92.
Kennedy KJ, Lentz EM. Treatment of landfill leachate using sequencing batch and continuous upflow anaerobic
sludge blanket (UASB) reactors. Water Res 2000;34:3640 – 56.
560 Y. Liu, J.-H. Tay / Biotechnology Advances 22 (2004) 533–563

Kos B, Suskovic J, Vukovic S, Simpraga M, Frece J, Matosic S. Adhesion and aggregation ability of probiotic
strain Lactobacillus acidophilus M92. J Appl Microbiol 2003;94:981 – 7.
Kosaric N, Blaszczyk R, Orphan L, Valladares J. The characteristics of granules from upflow anaerobic sludge
blanket reactors. Water Res 1990;24:1473 – 7.
Lens P, de Beer D, Cronenberg C, Ottengraf S, Verstraete W. The use of microsensors to determine distributions
in UASB aggregates. Water Sci Technol 1995;31:273 – 80.
Lepisto R, Rintala J. Extreme thermophilic (70 jC), VFA-fed UASB reactor: performance, temperature response,
load potential and comparison with 35 and 55 jC UASB reactors. Water Res 1999;33:3162 – 70.
Lettinga G, van Velsen AFM, Hobma SW, de Zeeuw W, Klapwijk A. Use of the upflow sludge blanket (USB)
reactor concept for biological waste water treatment especially for anaerobic treatment. Biotechnol Bioeng
1980;22:699 – 734.
Lettinga G, Rebac S, Zeeman G. Challenge of psychrophilic anaerobic wastewater treatment. Trends Biotechnol
2001;19:363 – 70.
Lew B, Belavski M, Admon S, Tarre S, Green M. Temperature effect on UASB reactor operation for domestic
wastewater treatment in temperate climate regions. Water Sci Technol 2003;48:25 – 30.
Lin YM, Liu Y, Tay JH. Development and characteristics of phosphorous-accumulating granules in sequencing
batch reactor. Appl Microbiol Biotechnol 2003;62:430 – 5.
Liu Y. Adhesion kinetics of nitrifying bacteria on various thermoplastic supports. Colloids Surf, B Biointerfaces
1995;5:213 – 9.
Liu Y, Tay JH. The essential role of hydrodynamic shear force in the formation of biofilm and granular sludge.
Water Res 2002;36:1653 – 65 (24).
Liu Y, Yang SF, Tan SF, Lin YM, Tay JH. Aerobic granules: a novel zinc biosorbent. Lett Appl Microbiol 2002;
35:548 – 51.
Liu QS, Tay JH, Liu Y. Substrate concentration-independent aerobic granulation in sequential aerobic sludge
blanket reactor. Environ Technol 2003a;24:1235 – 43.
Liu Y, Yang SF, Liu QS, Tay JH. The role of cell hydrophobicity in the formation of aerobic granules. Curr
Microbiol 2003b;46:270 – 4.
Liu Y, Lin YM, Yang SF, Tay JH. A balanced model for biofilms developed at different growth and detachment
forces. Process Biochem 2003c;38:1761 – 5.
Liu Y, Yang SF, Xu H, Woon KH, Lin YM, Tay JH. Biosorption kinetics of cadmium (II) on aerobic granular
sludge. Process Biochem 2003d;38:995 – 9.
Liu Y, Xu HL, Yang SF, Tay JH. A general model for biosorption of Cd2 +, Cu2 + and Zn2 + by aerobic granules.
J Biotechnol 2003e;102:233 – 9.
Liu Y, Yang SF, Tay JH. Elemental compositions and characteristics of aerobic granules cultivated at different
substrate N/C ratios. Appl Microbiol Biotechnol 2003f;61:556 – 61.
Liu Y, Lin YM, Yang SF, Tay JH. A balanced model for biofilms developed at different growth and detachment
forces. Process Biochem 2003g;38:1761 – 5.
Liu Y, Yang SF, Qin L, Tay JH. A thermodynamic interpretation of cell hydrophobicity in aerobic granulation.
Appl Microbiol Biotechnol 2003h;64:410 – 5.
Liu Y, Xu HL, Yang SF, Tay JH. The mechanisms and models for anaerobic granulation in upflow anaerobic
sludge blanket reactor. Water Res 2003i;37:661 – 73.
Liu Y, Yang SF, Tay JH. Improved stability of aerobic granules through selecting slow-growing nitrifying
bacteria. J Biotechnol 2004;108:161 – 9.
Lodi A, Solisoio C, Converti A, Del Borghi M. Cadmium, zinc, copper, silver and chromium(III) removal from
wastewaters by Sphaerotilus natans. Bioprocess Eng 1998;19:197 – 203.
MacLeod FA, Guiot SR, Costerton JW. Layered structure of bacterial aggregates produced in an upflow anaerobic
sludge bed and filter reactor. Appl Environ Microbiol 1990;56:1598 – 607.
Mahoney EM, Varangu LK, Cairns WL, Kosaric N, Murray RGE. Effect of Ca2 + on microbial aggregation during
UASB reactor start-up. Water Sci Technol 1987;19:249 – 60.
McSwain, BS, Irvine, RL, Wilderal, PA. The effect of intermittent feeding on aerobic granule structure. 5th Interna-
tional Conference on Biofilm Systems by International Water Association. South Africa: Cape Town; 2003.
Meyer RL, Saunders AM, Zeng RJ, Keller J, Blackall LL. Microscale structure and function of anaerobic –
aerobic granules containing glycogen accumulating organisms. FEMS Microbiol Ecol 2003;45:253 – 61.
Y. Liu, J.-H. Tay / Biotechnology Advances 22 (2004) 533–563 561

Meyer RL, Saunders AM, Zeng RJ, Keller J, Blackall LL. Microscale structure and function of anaerobic –
aerobic granules containing glycogen accumulating organisms. FEMS Microbiol Ecol 2003;45:253 – 61.
Mishima K, Nakamura M. Self-immobilization of aerobic activated sludge—a pilot study of the aerobic upflow
sludge blanket process in municipal sewage treatment. Water Sci Technol 1991;23:981 – 90.
Morgenroth E, Sherden T, van Loosdrecht MCM, Heijnen JJ, Wilderer PA. Aerobic granular sludge in a
sequencing batch reactor. Water Res 1997;31:3191 – 4.
Morvai L, Mihaltz P, Czako L. The kinetic basis of a new start-up method to ensure the rapid granulation of
anaerobic sludge. Water Sci Technol 1992;25:113 – 22.
Moy BYP, Tay JH, Toh SK, Liu Y, Tay STL. High organic loading influences the physical characteristics of
aerobic sludge granules. Lett Appl Microbiol 2002;34:407 – 12.
Ng PH. Storage stability of aerobic granules cultivated in aerobic granular sludge blanket reactor. Final year
report of Bachelor of Engineering 2002. Singapore: Nanyang Technological University; 2002.
Noyola A, Mereno G. Granulation production from raw waste activated sludge. Water Sci Technol 1994;30:
339 – 46.
O’Flaherty V, Lens PN, de Beer D, Colleran E. Effect of feed composition and upflow velocity on aggregate
characteristics in anaerobic upflow reactors. Appl Microbiol Biotechnol 1997;47:102 – 7.
Ohashi A, Harada H. Adhesion strength of biofilm developed in an attached-growth reactor. Water Sci
Technol 1994;29:10.
Peng D, Bernet N, Delgenes JP, Moletta R. Aerobic granular sludge—a case report. Water Res 1999;33:890 – 3.
Pereboom JHF, Vereijken TLFM. Methanogenic granule development in full scale internal circulation reactors.
Water Sci Technol 1994;30:9 – 21.
Pratt LA, Kolter R. Genetic analysis of bacterial biofilm formation. Curr Opinion Microbiol 1999;2:598 – 603.
Prescott LM, Harley JP, Klein DA. Microbiology. 4th edition. Boston: McGraw-Hill; 1999.
Pringle JH, Fletcher M. Influence of substratum wettability on attachment of fresh bacteria to solid surface. Appl
Environ Microbiol 1983;45:811 – 7.
Qin L, Tay JH, Liu Y. Selection pressure is a driving force of aerobic granulation in sequencing batch reactors.
Process Biochem 2004;39:579 – 84.
Quarmby J, Forster CF. An examination of the structure of UASB granules. Water Res 1995;29:2449 – 54.
Rocheleau S, Greer CW, Lawrence JR, Cantin C, Laramee L, Guiot SR. Differentiation of Methanosaeta concilii
and Methanosarcina barkeri in anaerobic mesophilic granular sludge by in situ hybridization and confocal
scanning laser microscopy. Appl Environ Microbiol 1999;65:2222 – 9.
Rouxhet PG, Mozes N. Physical chemistry of the interaction between attached microorganisms and their support.
Water Sci Technol 1990;22:1 – 16.
Santegoeds CM, Damagaad LR, Hesselink C, Zopfi J, Lens P, Muyzer G, et al. Distribution of sulfate-reducing
and methanogenic bacteria in anaerobic aggregates determined by microsensor and molecular analysis. Appl
Environ Microbiol 1999;65:4618 – 29.
Schmidt JE, Ahring BK. Effects of magnesium on thermophilic acetate-degrading granules in upflow anaerobic
sludge blanket (UASB) reactor. Enzyme Microb Technol 1993;15:304 – 10.
Schmidt JE, Ahring BK. Granular sludge formation in upflow anaerobic sludge blanket (UASB) reactors. Bio-
technol Bioeng 1996;49:229 – 46.
Schwarzenbeck N, Erley R, Wilderer PA. Growth of aerobic granular sludge in a SBR-system treating wastewater
rich in particulate matter. 5th International conference on biofilm systems, 14 – 19 September, Cape Town,
South Africa.
Sekiguchi Y, Kamagata Y, Nakamura K, Ohashi A, Harada H. Fluorescence in situ hybridization using 16S
rRNA-targeted oligonucleotides reveals localization of methanogenes and selected uncultured bacteria in
mesophilic and thermophilic sludge granules. Appl Environ Microbiol 1999;65:1280 – 8.
Shin HS, Lim KH, Park HS. Effect of shear stress on granulation in oxygen aerobic upflow sludge reactors. Water
Sci Technol 1992;26:601 – 5.
Singh KS, Viraraghavan T. Impact of temperature on performance, microbiological, and hydrodynamic aspects of
UASB reactors treating municipal wastewater. Water Sci Technol 2003;48:211 – 7.
Tagawa T, Syutsubo K, Sekiguchil Y, Ohashi A, Harada H. Quantification of methanogen cell density in
anaerobic granular sludge consortia by fluorescence in-situ hybridization. Water Sci Technol 2000;42:
77 – 82.
562 Y. Liu, J.-H. Tay / Biotechnology Advances 22 (2004) 533–563

Taniguchi J, Hemmi H, Tanahashi K, Amano N, Nakayama T, Nishim T. Zinc biosorption by a zinc-resistant


bacterium, Brevibacterium sp. Strain, HZM-1. Appl Microbiol Biotechnol 2000;54:581 – 8.
Tay JH, Yan YG. Influence of substrate concentration on micronial selection and granulation during start-up of
upflow anaerobic sludge blanket reactors. Water Environ Res 1996;68:1140 – 50.
Tay JH, Xu HL, Teo KC. Molecular mechanism of granulation: I. H+ translocation – dehydration theory.
J Environ Eng 2000;126:403 – 10.
Tay JH, Liu QS, Liu Y. Microscopic observation of aerobic granulation in sequential aerobic sludge blanket
reactor. J Appl Microbiol 2001a;91:168 – 75.
Tay JH, Liu QS, Liu Y. The effects of shear force on the formation, structure and metabolism of aerobic granules.
Appl Microbiol Biotechnol 2001b;57:227 – 33.
Tay JH, Liu QS, Liu Y. The role of cellular polysaccharides in the formation and stability of aerobic granules. Lett
Appl Microbiol 2001c;33:222 – 6.
Tay JH, Liu QS, Liu Y. Aerobic granulation in sequential sludge blanket reactor. Water Sci Technol 2002a;
46:13 – 8.
Tay JH, Yang SF, Liu Y. Hydraulic selection pressure-induced nitrifying granulation in sequencing batch reactors.
Appl Microbiol Biotechnol 2002b;59:332 – 7.
Tay JH, Liu QS, Liu Y. Characteristics of aerobic granules grown on glucose and acetate in sequential aerobic
sludge blanket reactors. Environ Technol 2002c;23:931 – 6.
Tay JH, Ivanov V, Pan S, Tay STL. Specific layers in aerobically grown microbial granules. Lett Appl
Microbiol 2002d;34:254 – 7.
Tay STL, Ivanov V, Yi S, Zhuang WQ, Tay JH. Presence of anaerobic Bacteroides in aerobically grown microbial
granules. Microb Ecol 2002e;44:278 – 85.
Tay JH, Tay STL, Ivanov V, Pan S, Liu QS. Biomass and porosity profile in microbial granules sued for aerobic
wastewater treatment. Lett Appl Microbiol 2003a;36:297 – 301.
Tay JH, Pan S, He YX, Tay STL. Effect of organic loading rate on aerobic granulation: Part II. Characteristics of
aerobic granules. J Environ Eng. 2003b [in press].
Tay JH, Liu QS, Liu Y. Shear force influences the structure of aerobic granules cultivated in sequencing batch
reactor. 5th International conference on biofilm systems, 14 – 19 September, Cape Town, South Africa.
Teo KC, Xu HL, Tay JH. Molecular mechanism of granulation: II. Proton translocating activity. J Environ Eng
2000;126:411 – 8.
Thiele JH, Wu WM, Jain MK, Zeikus JG. Ecoengineering high rate biomathanation system: design of improved
syntrophic biomathanation catalysis. Biotechnol Bioeng 1990;35:990 – 9.
Toh SK, Tay JH, Moy BYP, Ivanov V, Tay STL. Size-effect on the physical characteristics of the aerobic granule
in a SBR. Appl Microbiol Biotechnol 2003;60:687 – 95.
Tsuneda S, Nagano T, Hoshino T, Ejiri Y, Noda N, Hirata A. Characterization of nitrifying granules produced in
an aerobic upflow fluidized bed reactor. Water Res 2003;37:4965 – 73.
Uyanik S, Sallis PJ, Anderson GK. The effect of polymer addition on granulation in an anaerobic baffled reactor
(ABR): Part I. Process performance. Water Res 2002;36:933 – 42.
Valdman E, Leite SGF. Biosorption of Cd, Zn and Cu by Saragssum sp. waste biomass. Bioprocess Eng
2000;22:171 – 3.
Vanderhaegen B, Ysebaert E, Favere K, van Wambeke M, Peeters T, Panic V, et al. Acidogenesis in relation to in-
reactor granule yield. Water Sci Technol 1992;25:21 – 30.
Watnick P, Kolter R. Biofilm, city of microbes. J Bacteriol 2000;182:2675 – 9.
Wimpenny JWT, Colasanti R. A unifying hypothesis for the structure of microbial biofilms based on cellular
automaton models. FEMS Microbiol Ecol 1997;22:1 – 16.
Wirtz RA, Dague RR. Enhancement of granulation and start-up in anaerobic sequencing batch reactor. Water
Environ Res 1996;68:883 – 92.
Wu WM. Technological and microbiological aspects of anaerobic granules, PhD dissertation. Michigan,
U.S.A.:Michigan State University; 1991.
Wu WM, Hu JC, Gu XS, Gu GW. Cultivation of anaerobic granular sludge in UASB reactors with aerobic
activated sludge as seed. Water Res 1987;21:789 – 99.
Xie S. Metabolic response of aerobic granules to chromium(III). Final year report of Bachelor of Engineering.
Singapore:Nanyang Technological University; 2003.
Y. Liu, J.-H. Tay / Biotechnology Advances 22 (2004) 533–563 563

Xu HL, Tay JH. Anaerobic granulation with methanol-cultured seed sludge. J Environ Sci Health Part A
2002;37:85 – 94.
Yang SF, Tay JH, Liu Y. Effect of substrate N/COD ratio on the formation of aerobic granules. J Environ Eng
2003a [in press].
Yang SF, Liu Y, Tay JH. A novel granular sludge sequencing batch reactor for removal of organic and nitrogen
from wastewater. J Biotechnol 2003b;106:77 – 86.
Yang SF, Liu QS, Tay JH, Liu Y. Growth kinetics of aerobic granules developed in sequencing batch reactors. Lett
Appl Microbiol 2004a;38:106 – 12.
Yang SF, Tay JH, Liu Y. Inhibition of free ammonia to the formation of aerobic granules. Biochem Eng J
2004b;17:41 – 8.
Yi S, Tay JH, Maszenan AM, Tay STL. A culture-independent approach for studying microbial diversity in
aerobic granules. Water Sci Technol 2003;47:283 – 90.
Yu HQ, Tay JH, Fang HHP. The role of calcium in sludge granulation during UASB reactor start-up. Water Res
2001;35:1052 – 60.
Zhu J, Wilderer PA. Effect of extended idle conditions on structure and activity of granular activated sludge.
Water Res 2003;37:2013 – 8.
Zoutberg GR, de Been P. The biobed EGSB (expanded granular sludge bed) system covers shortcoming of the
upflow anaerobic sludge blanket reactor in the chemical industry. Water Sci Technol 1997;35:183 – 8.

You might also like