Available online at www.sciencedirect.com

Bioresource Technology 99 (2008) 7444–7449

Aerobic granulation under the combined hydraulicand loading selection pressures

Yao Chen a,b, Wenju Jiang a,*, David Tee Liang b, Joo Hwa Tay b

a College of Architecture and Environment, Sichuan University, Chengdu 610065, PR Chinab Institute of Environmental Science and Engineering, Nanyang Technological University, Singapore 637723, Singapore

Received 4 December 2007; received in revised form 20 February 2008; accepted 20 February 2008Available online 2 April 2008

Abstract

Two SBR reactors were set up to investigate the feasibility of aerobic granulation under the combined selection pressures of hydraulicshear force and substrate loading. Aerobic granulation was studied at superficial upflow air velocity of 3.2 and 2.4 cm/s under an organicloading rate (OLR) range of 6.0–15.0 kg COD/m3 d. Good reactor performance and well granule characteristics were achieved in a wideOLR range from 6.0 high up to 15.0 kg COD/m3 d at 3.2 cm/s. While under the velocity of 2.4 cm/s, stable operation was limited in theOLR range of 6.0–9.0 kg COD/m3 d and failed to operate with granule deterioration under further higher OLRs. The optimal combi-nation of hydrodynamic shear force and loading selection pressure was demonstrated to be an important factor that influence aerobicgranulation and govern the granule characteristics and reactor performance.� 2008 Elsevier Ltd. All rights reserved.

Keywords: Aerobic granulation; Sequencing batch reactor (SBR); Combined selection pressure; Hydraulic shear force; Organic loading rate (OLR)

1. Introduction

Aerobic granulation technology has been developed andstudied widely for the last two decades (Beun et al., 1999;Morgenroth et al., 1997; Peng et al., 1999; Tay et al.,2001b). Aerobic granules are aggregates resulted from nat-urally self-immobilization of the bacteria under aerobiccondition. Compared with conventional activated sludge,aerobic granules have the advantages of the clear-outshape, compact and strong microbial structure, fast start-up, good settling ability, high biomass retention and abilityto withstand a high organic loading rate (OLR).

Due to the economical and technical advantages, aerobicgranules-based bioreactors have been given more interestsin the removal of biodegradable organic matter, nitrogenand phosphor (Beun et al., 2001; Lin et al., 2003). To be ableto apply aerobic granulation technology widely and reli-ably, the formation and stability of granules are essential

0960-8524/$ - see front matter � 2008 Elsevier Ltd. All rights reserved.

doi:10.1016/j.biortech.2008.02.028

* Corresponding author. Tel.: +86 28 8540 3016; fax: +86 28 8540 5613.E-mail address: wenjujiang@scu.edu.cn (W. Jiang).

for successful operation. Previous researches have shownthat hydraulic selection pressure and substrate loading rateare two key factors that influence the formation, structureand stability of aerobic granules (Kim et al., 2008; Moyet al., 2002; Tay et al., 2001a; Zheng et al., 2006). In col-umn-type bioreactors, aeration is the main source of oxygensupply and a major cause of hydrodynamic turbulence andhydraulic shear force. Hydraulic shear force caused by aer-ation can be a direct indicator of hydraulic selection pres-sure and quantified by superficial upflow air velocity. It isan easy, effective and comparable way to control the gran-ulation in laboratory-scale experiments (Beun et al., 1999;Liu and Tay, 2002; Tay et al., 2004a).

The OLR is another important operational parameterthat can significantly influence the microbial ecology andprocess performance of aerobic granules systems. It canaffect the granulation process by selecting and enrichingdifferent bacterial species and influencing the size, settlingability and bioactivity of the granules. A moderate OLRwas found to favor the development of stable aerobicgranules (Tay et al., 2004b,c). Poor stability of aerobic

Y. Chen et al. / Bioresource Technology 99 (2008) 7444–7449 7445

granules under high OLRs was also reported in other stud-ies (Liu and Liu, 2006; Zheng et al., 2006). However, a highOLR is desirable in biological wastewater treatment sys-tems as it is favorable of treating high-strength wastewaterswith small footprints.

Therefore, considering the effects of hydraulic selectionpressure and the stressed pressure of substrate loading onaerobic granulation, it is feasible to apply the combinationof hydraulic and substrate loading selection pressures tooptimize aerobic granulation. To evaluate the impact ofthe combined hydraulic shear force and substrate loadingrate on aerobic granulation and the feasibility of stableoperation, two sequencing batch reactors (SBR) were setup to study the granulation processes, reactor performanceand granule characteristics. This work could contribute toa better understanding of stable operation with aerobicgranules at optimal operational parameters.

2. Methods

2.1. Experimental set-up

Two column-type acrylic reactors (R1 and R2) with aworking volume of 1.1 l were used as SBR reactors to cul-tivate aerobic granules. Each reactor had the same geomet-rical configuration with an internal diameter of 52 mm.Fine air bubbles for aeration were introduced through anair diffuser in the bottom of the column and the air flowrate of 4.0 l/min and 3.0 l/min were applied for R1 andR2, respectively. Thus it resulted in a hydraulic shear force,in terms of superficial upflow air velocity of 3.2 cm/s in R1and 2.4 cm/s in R2. Each reactor was operated for 4 h percycle sequentially: 5 min of influent filling, 225 min of aer-ation, 5 min of settling and 5 min of effluent withdrawal.Effluent was discharged at the middle sampling port ofthe working column and the discharge volume ratio was50% per cycle. The hydraulic retention time (HRT) waskept for 8 h. OLR was increased stepwise from 6.0 to 9.0,12.0 and 15.0 kg COD/m3 d or until the failure of opera-tion in both reactors. The detailed experimental conditionsof the reactors were shown in Table 1.

Table 1Detailed experimental conditions of the sequencing batch reactors

Reactor Operationaldays

Superficial upflowair velocity (cm/s)

OLR(kg COD/m3 d)

R1 Day 1–118 3.2 6.0Day 119–184 3.2 9.0Day 185–335 3.2 12.0Day 336–390 3.2 15.0

R2 Day 1–118 2.4 6.0Day 119–184 2.4 9.0Day 185–263 2.4 12.0

2.2. Medium

The synthetic wastewater consisted of sodium acetate assole carbon source and the compositions of the wastewaterwere as follows: NaAC � 3H2O, 4270 mg/l; NH4Cl, 776 mg/l; KH2PO4, 89 mg/l, and other necessary elements weresimilar to that microelement solution detailed elsewhere(Tay et al., 2002). This gave an initial substrate concentra-tion COD of 2000 mg/l and an OLR of 6.0 kg COD/m3 dfor the reactors. COD concentration was increased step-wise to 3000, 4000 and 5000 mg/l by proportionally adjust-ing the concentration of each chemical ingredient to attainhigher OLRs of 9.0, 12.0 and 15.0 kg COD/m3 d. Increasesin OLR were implemented when the reactor had reachedsteady state and stabilized for at least 4 weeks.

2.3. Seeding

Both R1 and R2 were started up with 50% of activatedsludge from a local municipal wastewater treatment plant.The seed sludge was grayish brown in color and had a sus-pended solid (SS) concentration of 5.02 g/l, sludge volumeindex (SVI) of 187.3 ml/g SS and a mean particle size of120 lm.

2.4. Analytical methods

Wastewater samples taken from the reactors were ana-lyzed for COD, SS, volatile suspended solids (VSS) andSVI in accordance to the standard methods (APHA,1998). The particle size was measured by a laser particlesize analysis system (MasterSizer 2000, Malvern Instru-ments, UK) with a range of 0.02–2000 lm and granulemorphology was determined by an image analysis (IA) sys-tem (Olympus SZX9, Japan) with Image Pro Plus software(Media Cybernetics, L.P. version 4.0, USA).

The biomass density was determined according to themethod described by Beun et al. (1999). The test proceduredescribed by Ghangrekar et al. (1996) was used for thedetermination of the granular strength. The results areexpressed in terms of an integrity coefficient (IC). Thehigher the IC, the greater the strength of granules. Fractaldimension (Df) is used to determine the internal structureof aerobic granules according to the method reported byChen et al. (2007). A high Df indicates compact aggregateswhereas low values correspond to more ‘loose’ aggregates.

3. Results

3.1. Aerobic granulation

Fig. 1 shows the evolution of granule size and profile ofbiomass concentration in both reactors (R1 and R2) underdifferent combinations of hydraulic and loading selectionpressures. As shown in Fig. 1, after 10 days of inoculation,

Fig. 1. Evolution of aerobic granule size and profile of biomass concentration in SBR reactors under various combinations of selection pressures, (a) R1(3.2 cm/s, OLR = 6.0–15.0 kg COD/m3 d) and (b) R2 (2.4 cm/s, OLR = 6.0–9.0 kg COD/m3 d).

7446 Y. Chen et al. / Bioresource Technology 99 (2008) 7444–7449

tiny granules appeared in both reactors and stabilized fromDay 40 under the initial OLR of 6.0 kg COD/m3 d.Granule size and biomass concentration were varied insig-nificantly and maintained within a certain range in the fol-lowing period. Good performance was also presented withhigh COD removal rate, low biomass concentration in theeffluent and good settling ability in both reactors.

After 80-day’s stable operation, OLR were increased to9.0 kg COD/m3 d in both reactors. Granule size and bio-mass concentration were gradually increased and reacheda new stable stage approximate 20–27 days after the newOLR was applied. During the following 6 weeks, matureaerobic granules were well performed and shown similarcharacteristics in both R1 and R2. Compared to the sub-strate loading condition of 6.0 kg COD/m3 d, larger granulesize and higher biomass concentration were maintained inboth reactors under OLR of 9.0 kg COD/m3 d.

From Day 185, OLR were stepwise increased to12.0 kg COD/m3 d. As shown in Fig. 1, R1 and R2 wereundergone a serious granule disintegration soon afternew loading condition was applied. Granules in both reac-tors were disintegrated from big particles to fluffy flocs andwashed out of the reactors. Simultaneously biomass con-centration in two reactors dropped quickly from over10 g/l to a fairly low level of 1–2 g/l. Due to different

hydraulic shear forces, the behavior of R1 and R2 were dif-ferent in sequent operation.

With relatively higher hydraulic shear force, R1 (3.2 cm/s) was recovered from poor circumstance after 3 weeks’acclimation under the new OLR of 12.0 kg COD/m3 d.New granules were formed with particle size graduallyincreasing and stabilized at about 950 lm. Meanwhile,biomass concentration in R1 was escalated to a higherlevel (around 28.0 g/l) than previous operation stage(Fig. 1a). It took 90 days for R1 to be stabilized againand this was much longer than the acclimation stage atlower OLRs.

On the contrary, under relatively lower hydraulic selec-tion pressure (2.4 cm/s), R2 was failed to accommodate tothe new operational condition with higher OLR of12.0 kg COD/m3 d and could not run smoothly thereafter.As illustrated in Fig. 1b, with OLR of 12.0 kg COD/m3 d,after 2 weeks’ disintegration, granules in R2 were reformedand particle size was increased rapidly in the following 2weeks. Large-sized filamentous granules were dominatedin the reactor and finally turned into porous bioparticles.Biomass concentration in R2 was also fluctuated duringthe sequent operation. And R2 showed poor performancewith low COD removal (70–80%), low biomass concentra-tion (around 3.0 g/l), high SS in the effluent (3.5–4.0 g/l)

Y. Chen et al. / Bioresource Technology 99 (2008) 7444–7449 7447

and high SVI (130–162 ml/g SS). Granular sludge deterio-rated completely after 70 days running in R2.

From Day 336, OLR was increased to 15.0 kg COD/m3 d in R1. Granular sludge soon adapted to the highOLR and became stable within 27 days. Slightly smallergranule size and much higher biomass concentration waspresented in this stage compared with previous stages(Fig. 1a).

3.2. Reactor performance and aerobic granule characteristics

Table 2 summarized the reactor performance and char-acteristics of aerobic granules under various combinationsof hydraulic and organic loading selection pressures atsteady state.

As shown in Table 2, the biomass concentration in thereactors increased as OLR increased under same hydraulicoperational condition, e.g. from 4.69 g/l at 6.0 kg COD/m3 d to 43.06 g/l at 15.0 kg COD/m3 d under the identicalshear force of 3.2 cm/s in R1, and from 7.16 g/l at6.0 kg COD/m3 d to 12.69 g/l at 9.0 kg COD/m3 d in R2(2.4 cm/s). The same phenomena were observed in thechange of biomass concentration in the effluent. Thechange of biomass concentration indicated the stimulationof substrate loading increments on granule sludge growth.In spite of the biomass concentration change, the CODremoval rate of steady states in two reactors was similarlyhigh and fluctuated insignificantly. The high value of CODremoval presented good performance and high bioactivityin both reactors.

In general, as OLR increased, aerobic granules becamebigger and denser in R1, i.e. with larger particle size andlower SVI value, although there was a slight decrease inparticle size from 950 to 854 lm when OLR increased from12.0 to 15.0 kg COD/m3 d. Granules with clear-outer mor-phology and similar appearance were dominated in bothreactors at various steady states. Aspect ratio, which repre-sents the particle’s regularity (0 = line and 1 = circle), wasalmost unchangeable when the standard deviations wastaken into account. It was shown that once mature aerobic

Table 2Reactor performance and characteristics of aerobic granules under different O

Reactor R1

Superficial upflow air velocity (cm/s)

3.2

Reactor performance

OLR (kg COD/m3 d) 6.0 9.0 12.0SS (g/l) 4.69 ± 0.42 10.56 ± 0.75 28.3VSS (g/l) 4.38 ± 0.43 9.50 ± 0.71 26.7Effluent SS (g/l) 0.51 ± 0.10 0.55 ± 0.11 0.65COD removal (%) 96.0 ± 1.1 98.5 ± 0.7 97.7

Granule characteristics

Particle size (lm) 342 ± 48 516 ± 44 950Df 2.07 ± 0.04 1.69 ± 0.05 1.75SVI (ml/g) 27 ± 6 25 ± 3 14 ±Aspect ratio 0.58 ± 0.03 0.60 ± 0.03 0.61

granules were formed, changes in morphology of the gran-ules were negligible in spite of which combinations of selec-tion pressure they sustained. Lots of cavities were presentin the mature granules surface and the irregular morphol-ogy enabled aerobic granules with less mass transfer resis-tance and promotion of diffusion effectiveness.

To differentiate granules generated in different opera-tions, Df is a more sophisticated way than the visualdescription (Bellouti et al., 1997). Df reflected the internalstructure of fractal aggregates. Higher Df means compactand dense aggregates while lower Df indicates loose or por-ous structure (Guan et al., 1998; Mu and Yu, 2006). Thechange of Df indicated that the granular structure becamerelatively looser and more porous than previous stage,which decreased from 2.07 to around 1.70–1.76 at shearforce of 3.2 cm/s when OLR increased from 6.0 to15.0 kg COD/m3 d and from1.96 to 1.76 at shear force of2.4 cm/s (OLR from 6.0 to 9.0 kg COD/m3 d). The rela-tively looser structure allowed well penetration of nutrientsinto the granule interior and enhanced mass diffusion. Ithelped to maintain growth rate and metabolism of the bac-teria in the granules, and probably enabled the granules tosustain the high OLRs.

Moreover, it was demonstrated from Table 2 that underthe same OLR, biomass concentration in the reactor wasdeclined with the increase of shear force. This indicatedthat more sludge could retain in the reactor under lowhydraulic shear force. Particle size and SVI value were alsodecreased along with the increase of shear force that indi-cated the high hydraulic shear force was beneficial todevelop dense granules and improve sludge settling ability.

Fig. 2 shows the comparison of biomass density andgranule strength of stable aerobic granules under variouscombinations of selection pressure. As shown in Fig. 2,granule density was escalated as OLR increased underthe identical shear force and declined with the decrease ofshear force at the same substrate loading condition. ForR1 (3.2 cm/s), granule density and strength were increasedfrom 41.9 to 103.7 g/l and 97–99% with increase of OLR,respectively. Otherwise, granule density was slightly differ-

LRs at steady state in the reactors

R2

2.4

15.0 6.0 9.04 ± 1.47 43.06 ± 4.66 7.16 ± 0.93 12.69 ± 0.786 ± 1.67 40.36 ± 4.38 6.45 ± 0.81 11.44 ± 0.73± 0.15 0.71 ± 0.07 0.50 ± 0.08 0.56 ± 0.09± 0.6 98.0 ± 0.5 97.0 ± 0.9 98.3 ± 0.7

± 33 854 ± 25 557 ± 60 560 ± 50± 0.06 1.76 ± 0.05 1.96 ± 0.02 1.76 ± 0.073 8 ± 1 41 ± 6 38 ± 5± 0.05 0.62 ± 0.02 0.56 ± 0.02 0.58 ± 0.03

Fig. 2. Comparison of biomass density and granule strength in SBRreactors under various combinations of selection pressures.

7448 Y. Chen et al. / Bioresource Technology 99 (2008) 7444–7449

ent (25.6–28.7 g/l) between OLR of 6.0 and 9.0 kg COD/m3 d in R2 and was smaller than those in R1.

4. Discussion

In nature, selection pressures for aerobic granulation aretrigger forces that play a crucial role in granulation processand further influence the granular characteristics and reac-tor performance (Qin et al., 2004; Wang et al., 2007). Theeffect of either hydraulic shear force, in terms of superficialupflow air velocity, or stressed substrate loading, in termsof OLR, was investigated to show the importance on suc-cessful operation of aerobic granule systems.

High hydraulic selection pressure exerts in systems willretain denser, heavier, more compact and smooth granules,as well as improved metabolic activity (Chen et al., 2007).No granular sludge was observed with low superficialupflow air velocity less than 1.2 cm/s (Tay et al., 2001a),while a much stronger hydraulic pressure triggered seriousbiomass washout and led to reactor failure (Pan et al.,2004). In this study, dense and good settling ability aerobicgranules were developed with increasing hydraulic shearforce under same OLR conditions. As shown in Table 2and Fig. 2, better settling ability presented by smallerSVI value, higher granule density and strength were foundunder relatively higher shear force than those under lowshear force.

Substrate concentration is found to exert a microbialselection pressure and affects granulation process and per-formance of granules by selection and enriching differentbacterial species (Moy et al., 2002). Moreover, only moder-ate OLR can sustain stable aerobic granules systems inSBR reactors. The size of aerobic granules generallyincreases with OLR, and leads to a decrease in densityand an increase in porosity, which may further cause insta-bility of reactor operation (Moy et al., 2002; Tay et al.,2004b,c). Similar phenomena were found in this study thatgranule size was increased with OLR increase at the identi-

cal superficial upflow air velocity, while more biomass wasdeveloped in reactors with stimulation growth under highersubstrate loading.

Otherwise, high selection pressures may lead to microor-ganisms’ maladjustment and system failure. Thus the opti-mal combination of those two forces would be crucial toenhance the application of aerobic granules for treatmentof high-strength wastewaters in an economical operation.

As illustrated in Fig. 2 and Table 2, stable aerobic gran-ules could be formed and maintained in various combina-tions of hydraulic and loading selection pressures in bothreactors, e.g. from 6.0 to 9.0 kg COD/m3 d in superficialupflow air velocity of 3.2 or 2.4 cm/s, respectively. Bothreactors acclimated to the combined selection pressuresrapidly and stabilized again with good performance andwell granular characteristics. It was indicated that the com-bined effects of hydraulic and loading forces could main-tain steady and reliable system operation in moderateOLR of 6.0–9.0 kg COD/m3 d at either shear force of 3.2or 2.4 cm/s.

When OLR was further increased from 9.0 to12.0 kg COD/m3 d, the stronger OLR applied than previ-ous stage upset the balance and caused dramatically differ-ent outcome under the operations of two hydraulic shearforces. Both reactors underwent a sharp re-selection pro-cess with granule disintegration and sludge washout. Withrelatively higher hydraulic shear force (3.2 cm/s), the reac-tor overcame the drawback of higher OLR on granulationand re-formed stable aerobic granules for steady operation.However, the re-selection and acclimation process wasmore violent and time-consuming than previous stages oflower OLRs. Based on the re-selection process, the reactorunder relatively higher shear force (3.2 cm/s) could accom-modate soon after OLR continuously increased to15.0 kg COD/m3 d. New balance was established betweenhigh hydraulic and loading selection pressures to maintainstabilization of system running.

Compared to it, although new granules were appearedin the reactor, the relatively lower hydraulic shear force(2.4 cm/s) could not balance the overgrowth of granulescaused by higher OLR and granules were deterioratedcompletely. It was in accordance with the previousresearches (Moy et al., 2002; Zheng et al., 2006). Zhenget al. (2006) found that under shear force of 2.0 cm/s, thecompact bacteria-dominated aerobic granules were not sta-ble and transited gradually to large-sized filamentous gran-ules and disintegrated. Moy et al. (2002) indicated that theacetate-fed granules could not sustain high OLR(9.0 kg COD/m3 d) and disintegrated under shear force of2.0 cm/s. As shown in these studies, with relatively lowershear force (2.0 cm/s), only moderate OLR could sustainstable aerobic granules. Further increase of OLR only ledto system failure.

The large size of granule at high OLR is demonstratedto be a result of its high growth rate (Yang et al., 2004),while high shear force would lead to more collision amongparticles and friction between particle and liquid, finally

Y. Chen et al. / Bioresource Technology 99 (2008) 7444–7449 7449

developed small size aerobic granules. It implied that highhydraulic shear force could mediate the size increasing pro-cess caused by the increasing OLR through the violent par-ticle–particle and particle–liquid collision. Withoutsufficient external hydraulic shear force, balance betweenhydraulic and loading selection pressures was broken, thusthe overgrowth of granular sludge caused by high OLRwas out of control and led to failure of system operation.In addition, the interaction of hydrodynamic shear forceand loading selection pressure also influenced aerobic gran-ulation by selection of adaptable bacterial community andregulated the metabolic pathway so as to maintain stabil-ization of reactor systems. With higher OLR of 12.0–15.0 kg COD/m3 d, only higher hydraulic shear force(3.2 cm/s) could maintain smooth reactor operation.

5. Conclusions

Good reactor performance and well granule characteris-tics could maintain and operate under the hydraulic shearforce of 3.2 cm/s in a wide OLR range from 6.0 kg COD/m3 d high up to 15.0 kg COD/m3 d. While under thehydraulic shear force of 2.4 cm/s, the OLR was limited to6.0–9.0 kg COD/m3 d for stable operation and failed tooperate when OLRs increased further. The optimal combi-nation of hydrodynamic shear force and loading selectionpressure was demonstrated to be an important factor thatinfluence aerobic granulation and govern the granule char-acteristics and reactor performance.

References

APHA, 1998. Standard methods for the examination of water andwastewater, 20th ed. American Public Health Association, Washing-ton, DC, USA.

Bellouti, M., Alves, M.M., Novais, J.M., Mota, M., 1997. Flocs vs.granules: differentiation by fractal dimension. Water Res. 31 (5), 1227–1231.

Beun, J.J., Heijnen, J.J., Van Loosdrecht, M.C.M., 2001. N-removal in agranular sludge sequencing batch airlift reactor. Biotechnol. Bioeng. 75(1), 82–92.

Beun, J.J., Hendriks, A., Van Loosdrecht, M.C.M., Morgenroth, E.,Wilderer, P.A., Heijnen, J.J., 1999. Aerobic granulation in a sequenc-ing batch reactor. Water Res. 33 (10), 2283–2290.

Chen, Y., Jiang, W., Liang, D.T., Tay, J.H., 2007. Structure and stabilityof aerobic granules cultivated under different shear force in sequencingbatch reactors. Appl. Microbiol. Biotechnol. 76 (5), 1199–1208.

Ghangrekar, M.M., Asolekar, S.R., Ranganathan, K.R., Joshi, S.G.,1996. Experience with UASB reactor start-up under different operatingconditions. Water Sci. Technol. 34 (5–6), 421–428.

Guan, J., Waite, T.D., Amal, R., 1998. Rapid structure characterization ofbacterial aggregates. Environ. Sci. Technol. 32 (23), 3735–3742.

Kim, I.S., Kim, S.-M., Jang, A., 2008. Characterization of aerobicgranules by microbial density at different COD loading rates.Bioresour. Technol. 99 (1), 18–25.

Lin, Y.M., Liu, Y., Tay, J.H., 2003. Development and characteristics ofphosphorus-accumulating microbial granules in sequencing batchreactors. Appl. Microbiol. Biotechnol. 62 (4), 430–435.

Liu, Y., Liu, Q.S., 2006. Causes and control of filamentous growth inaerobic granular sludge sequencing batch reactors. Biotechnol. Adv. 24(1), 115–127.

Liu, Y., Tay, J.H., 2002. The essential role of hydrodynamic shear force inthe formation of biofilm and granular sludge. Water Res. 36 (7), 1653–1665.

Morgenroth, E., Sherden, T., Van Loosdrecht, M.C.M., Heijnen, J.J.,Wilderer, P.A., 1997. Aerobic granular sludge in a sequencing batchreactor. Water Res. 31 (12), 3191–3194.

Moy, B.Y.P., Tay, J.H., Toh, S.K., Liu, Y., Tay, S.T.L., 2002. Highorganic loading influences the physical characteristics of aerobic sludgegranules. Lett. Appl. Microbiol. 34 (6), 407–412.

Mu, Y., Yu, H.Q., 2006. Rheological and fractal characteristics ofgranular sludge in an upflow anaerobic reactor. Water Res. 40 (19),3596–3602.

Pan, S., Tay, J.H., He, Y.X., Tay, S.T.L., 2004. The effect of hydraulicretention time on the stability of aerobically grown microbial granules.Lett. Appl. Microbiol. 38 (2), 158–163.

Peng, D.C., Bernet, N., Delgenes, J.P., Moletta, R., 1999. Aerobicgranular sludge – a case report. Water Res. 33 (3), 890–893.

Qin, L., Tay, J.-H., Liu, Y., 2004. Selection pressure is a driving force ofaerobic granulation in sequencing batch reactors. Process Biochem. 39(5), 579–584.

Tay, J.H., Liu, Q.S., Liu, Y., 2001a. The effects of shear force on theformation, structure and metabolism of aerobic granules. Appl.Microbiol. Biotechnol. 57 (1–2), 227–233.

Tay, J.H., Liu, Q.S., Liu, Y., 2001b. Microscopic observation of aerobicgranulation in sequential aerobic sludge blanket reactor. J. Appl.Microbiol. 91 (1), 168–175.

Tay, J.H., Liu, Q.S., Liu, Y., 2002. Aerobic granulation in sequentialsludge blanket reactor. Water Sci. Technol. 46 (4–5), 13–18.

Tay, J.H., Liu, Q.S., Liu, Y., 2004a. The effect of upflow air velocity on thestructure of aerobic granules cultivated in a sequencing batch reactor.Water Sci. Technol. 49 (11–12), 35–40.

Tay, J.H., Pan, S., He, Y.X., Tay, S.T.L., 2004b. Effect of organic loadingrate on aerobic granulation. I: Reactor performance. J. Environ. Eng.ASCE 130 (10), 1094–1101.

Tay, J.H., Pan, S., He, Y.X., Tay, S.T.L., 2004c. Effect of organic loadingrate on aerobic granulation. II: Characteristics of aerobic granules. J.Environ. Eng. ASCE 130 (10), 1102–1109.

Wang, X.H., Zhang, H.M., Yang, F.L., Xia, L.P., Gao, M.M., 2007.Improved stability and performance of aerobic granules under stepwiseincreased selection pressure. Enzyme Microbiol. Technol. 41 (3), 205–211.

Yang, S.F., Liu, Q.S., Tay, J.H., Liu, Y., 2004. Growth kinetics of aerobicgranules developed in sequencing batch reactors. Lett. Appl. Micro-biol. 38 (2), 106–112.

Zheng, Y.M., Yu, H.Q., Liu, S.H., Liu, X.Z., 2006. Formation andinstability of aerobic granules under high organic loading conditions.Chemosphere 63 (10), 1791–1800.