AEROBIC GRANULATION IN A SEQUENCING BATCH

REACTOR

J. J. BEUN

1

*, A. HENDRIKS

1

, M. C. M. VAN LOOSDRECHT

1

*

M

,

E. MORGENROTH

2

*

M

, P. A. WILDERER

2

*

M

and J. J. HEIJNEN

1

*

M

1

Department of Biochemical Engineering, Kluyver Laboratory for Biotechnology, Delft University of

Technology, Julianalaan 67, 2628 BC Delft, The Netherlands and

2

Institute of Water Quality Control

and Waste Management, Technical University Munich, 85748 Garching, Germany

(First received February 1998; accepted in revised form October 1998)

AbstractÐIn a sequencing batch reactor (SBR) granules of aerobic heterotrophic microorganisms were

cultured. The e€ect of di€erent operational conditions on the formation of these aerobic granules were

studied. The time allowed for settling was the main parameter to select for growth of bacteria in well

settling granules. Both a short HRT and a relative high shear were found favorable for granulation. A

substrate loading rate of 7.5 kg COD/(m

3

day) was applied. This led to formation of granules with an

average diameter of 3.3 mm and a biomass density of 11.9 gVSS/l

granule

. Based on microscopic obser-

vations a hypothesis for the granulation process was formulated. The reactor was started up without

any carrier material present. At the beginning ®lamentous fungal pellets dominated the reactor. These

pellets functioned as an immobilization matrix in which bacteria could grow out to colonies. After a

certain time the fungal pellets fell apart due to lysis in the inner part of the pellets, the bacterial colo-

nies could now remain in the reactor because they were large enough to settle suciently fast. These

colonies further grew out to granules. This paper shows that granule formation in aerobic reactors is

feasible and can be exploited to increase the volumetric conversion capacity of such reactors. # 1999

Elsevier Science Ltd. All rights reserved

Key wordsÐgranulation, SBR, aerobic granules, shear, settling time

INTRODUCTION

Most wastewater treatment systems have some dis-

advantages, e.g. high surplus biomass production,

low ¯exibility with respect to ¯uctuating loading

rates, a large area requirement for reactors and es-

pecially settlers and a relatively low volumetric con-

version capacity (0.5±2 kg COD/(m

3

day) for

activated sludge or biorotor systems). For anaerobic

processes much more compact reactors have been

developed [e.g. 40 kg COD/(m

3

day) for the UASB

(up¯ow anaerobic sludge blanket) reactor (Lettinga

et al., 1993)]. In these reactors the biomass is grow-

ing as well settling granules, which allow the ac-

cumulation of high amounts of active biomass in

the reactor. Moreover no settlers are needed

because sludge separation is integrated in the

UASB reactor itself. The mechanism of granulation,

however, is still subject of discussion. Since granula-

tion is almost entirely studied in the context of

methanogenic systems, it is regularly hypothesized

that the speci®c syntrophic bacterial interactions in

this process are the main cause of the granulation.

On the other hand, it can be stated that in the

UASB reactor the microorganisms have to grow in

a granule because otherwise they would be washed

out due to the continuous upward liquid velocity in

the UASB. Granulation is not only restricted to

methanogens. Granulation by acidifying bacteria

(Beeftink, 1987) nitrifying bacteria (De Beer et al.,

1993), denitrifying bacteria (Van der Hoek, 1988)

and aerobic heterotrophs (Tijhuis et al., 1994; Van

Benthum et al., 1996) has been observed. All these

observations have been done in a continuously

operated system. For many applications a discon-

tinuous operation is advantageous. In these sequen-

cing batch reactors (SBR) aerobic granules can also

be formed (Morgenroth et al., 1997). An extra ad-

vantage for the application of granules is here that

the settling (or stand still) time required is very

short due to the high settling velocity of granules.

The SBR in this study is a bubble column in

which wastewater is treated aerobically in a cycle of

a few hours. The biomass in the reactor consists of

aerobic granules. At the beginning of every cycle a

certain amount of wastewater is added to the reac-

tor. Then aeration and conversion take place. At

the end of the cycle the aeration is switched o€ and

the granules are allowed to settle for a few minutes

only. After settling, the upper part of the reactor

content is removed as clari®ed e‚uent. In this way

COD-removal and sludge settling take place in the

Wat. Res. Vol. 33, No. 10, pp. 2283±2290, 1999

# 1999 Elsevier Science Ltd. All rights reserved

Printed in Great Britain

0043-1354/99/$ - see front matter

PII: S0043-1354(98)00463-1

*Author to whom all correspondence should be addressed.

[Tel.: [phone]; fax: [phone]; e-mail:

[e-mail]].

2283

same reactor. This results in a single reactor with a

high concentration of granular sludge and therefore

high volumetric conversion rates. Besides, the

dynamic process conditions enable the SBR sludge

to withstand ¯uctuating wastewater streams. It

should be remarked that the particle size should

remain small in order to prevent serious di€usion

limitation in the granules.

The primary design criterium for the SBR is

based on the assumption that sludge granules will

be formed if ¯ocs are washed out. Sludge granules

have a high settling velocity compared to sludge

¯ocs, because granules are more dense. So granules

require less time to settle than ¯ocs. Therefore the

time allowed for settling in the SBR cycle is the

main design parameter. A short settling period will

eventually select for biomass particles with a high

settling velocity.

Because the settling velocity is an important

selection criterium, a high H/D ratio (column

height/column diameter) is advantageous. A high

H/D ratio and the absence of an external settler

results in a reactor with a small footprint. Besides,

the good settling characteristics allow a short stand-

still time for settling, allowing more time for bio-

logical puri®cation.

The aim of this research was to test the possi-

bility to obtain granules in the SBR and to ®nd op-

erational conditions leading to good granulation.

This means that smooth and dense granules have to

be formed. Experiments with the BAS (bio®lm air-

lift suspension)-reactor (Tijhuis et al., 1994; Kwok

et al., 1996; Van Benthum et al., 1996) already

pointed out that shear and surface substrate loading

play an important role in formation of smooth and

well settling bio®lm particles (Van Loosdrecht et

al., 1995).

In this study it was decided to pay attention to

the following aspects: (1) selection of granules based

on wash out of suspended and ¯oc-forming bac-

teria, (2) e€ects of substrate loading and shear on

the formation of stable and well settling particles.

The minimal settling velocity, v

set

min

, was varied

because this was the selection tool as explained

above. The minimal settling velocity in the reactor

was enforced by ®xing the settling time according

to: v

set

min

=settling height/settling time.

The hydraulic retention time, HRT, was varied

because this in¯uences the washout of suspended

biomass. The HRT has to be smaller than the 1/

m

max

to suppress suspended biomass growth. The

super®cial gas velocity, v

sg

, was varied because this

contributes to the shear and the COD loading was

varied because this in¯uences the accumulation of

biomass.

MATERIALS AND METHODS

Reactor set-up

A schematic representation of the reactor set-up is given

in Fig. 1.

The glass reactor had a working volume of 2.25 or 2.5 l

(Table 2). The internal diameter of the column was 5.6 cm,

the total height was 150 cm. E‚uent was drawn at 50 cm

from the bottom, so 1.25 l was left in the reactor after

e‚uent withdrawal. The reactor was open so that a natu-

ral mixed population could develop. As inoculum sludge

from a standard SBR for COD removal was used. The ex-

periments were performed at room temperature (20228C).

Fig. 1. Schematic representation of the SBR.

J. J. Beun et al.

2284

Air was introduced by a ®ne bubble aerator in the bottom

of the column. A mass ¯ow controller system was used to

keep the air¯ow constant.

Medium

A synthetic wastewater with the following composition

was used: Ethanol 0.40 g/l, NH

4

Cl 0.16 g/l, K

2

HPO

4

0.58 g/l,

KH

2

PO

4

0.23 g/l,

MgSO

4

7H

2

O

0.09 g/l,

CaCl

2

2H

2

O 0.07 g/l, EDTA 0.02 g/l, trace solution 1 ml/l.

This gives a total COD of 0.83 gCOD/l, a TKN of

0.04 gN/l and a total P of 0.16 gP/l.

The composition of the trace solution was: FeCl

3

6H

2

O

1.5 g/l, H

3

BO

3

0.15 g/l, CuSO

4

5H

2

O 0.03 g/l, KI 0.03 g/l,

MnCl

2

4H

2

O

0.12 g/l,

Na

2

MoO

4

2H

2

O

0.06 g/l,

ZnSO

4

7H

2

O 0.12 g/l, CoCl

2

6H

2

O 0.15 g/l. (Smolders et

al., 1995).

Experimental set-up

The reactor was operated with a cycle length of 3 or

4 h. The timing of a cycle is given in Table 1. The dur-

ation of each phase in a cycle is given. Aeration is on

during in¯uent adding and is o€ during e‚uent withdra-

wal and settling.

During the experiments, CO

2

-concentration in the o€-

gas, pH and incidentally DO were measured online.

An overview of the operational conditions applied in

di€erent periods is given in Table 2.

Analytical procedures

Ethanol concentrations were determined using gas chro-

matography (Chrompack CP9001 containing an Hewlet

Packard column). Carbon dioxide content of the incoming

air and the o€gas of the reactor was measured with an in-

frared carbon dioxide analyzer (Beckman Instruments

870). Ammonium concentrations were measured spectro-

photometrically (630 nm) with an auto-analyzer (Skalar

5010).

The biomass density (the biomass concentration in the

granules) was determined as follows: from a sample of

granules (100 ml), the water was removed by ®ltrating

(45 mm ®lter) and the granules were added to 20 ml of

demineralized water. The total volume of the granules

could now be determined by measuring the total volume

of water plus granules (volumetric displacement method).

Hereafter the dry weight of these granules was determined

by drying the sample for at least 24 h at 1058C. The bio-

mass density could then be calculated by dividing the dry

weight of the granules by the total volume of the granules.

The biomass concentration in the reactor was calculated

from the biomass density, the biomass bedvolume and the

bedporosity (E). The biomass bedvolume of the reactor

was determined daily by settling the granules for 5 min

and reading the volume of the settled granules directly

from the volume indication on the column. The bedporos-

ity was assumed to be 0.4 (Tijhuis et al., 1994).

The biomass concentration in the reactor could be

calculated as follows: biomass concentration in reactor =

(1 ÿ E)
bedvolume
biomass density/volume reactor.

The sludge retention time (SRT) could be calculated by

dividing the amount of biomass removed with the e‚uent

per day by the amount of biomass in the reactor. The bio-

mass concentration in the e‚uent was determined by ®l-

trating the e‚uent using a 45 mm ®lter and drying the

®lter for at least 24 h at 1058C.

The ash content of the biomass was measured by burn-

ing the biomass for 1 h at 6008C.

Changes in morphology of the granules were followed

by Image Analysis. From a representative sample of gran-

ules the following parameters were measured:
. Dav = average ferret diameter

. Area = particle surface of the projection of a granules

on a ¯at surface

. Shape = capriciousness of the particle surface

(=4p area/circumference

2

; 0 = line, 1 = circle)

. Aspect = roundness of the particle

(=min. ferret diameter/max. ferret diameter;

0 = line, 1 = circle)

Between 50 and 100 granules were analyzed. Of every

parameter, a minimum and a maximum value was deter-

mined, the average value, the standard deviation and the

standard error. Photographs of the granules were taken to

follow and to show the progress of granulation under

di€erent operational conditions.

RESULTS

General observations

The reactor was started up by adding 10 ml of

suspended, non-settling cells from a COD removing

SBR. Also during start-up the settling time was

kept short. After inoculation of the reactor, highly

®lamentous granules were formed in several days.

From observations both in the reactor and under

the microscope it could be concluded that the gran-

ules in this ®rst stage were formed by fungi. These

granules were not stable at all and broke up into

pieces after a few days. Subsequently a large part of

Table 1. Timing of a 3-h cycle and a 4-h cycle

Successive phases

3-h Cycle

4-h Cycle

E‚uent withdrawal

1 min (1 l)

1 min (1.25 l)

Adding in¯uent

2 min (1 l)

2 min (1.25 l)

Aeration

177 min

237 min

Settling

2 min

2 min

Total cycle length

180 min

240 min

Table 2. Overview of the operational conditions

Period

Cycle length (h)

Volume (l)

HRT (h)

COD load

(kg COD/(m

3

d))

Super®cial gas

velocity (m/s)

Minimal settling

velocity (m/h)

1

4

2.5

8

2.5

0.014

15

2

4

2.5

8

2.5

0.020

15

3

3

2.25

6.75

2.5

0.020

24

4

3

2.25

6.75

2.5

0.041

24

5

3

2.25

6.75

7.5

0.020

24

6

3

2.25

6.75

5

0.041

24

7

3

2.25

6.75

5

0.041

12

a

8

3

2.25

6.75

7.5

0.041

12

a

a

In these two periods the settling period was 4 min instead of 2 min.

Aerobic granulation in a SBR

2285

the biomass was washed out and a new granulation

occurred. The granules formed in this second stage

hardly contained any ®laments and consisted domi-

nantly of bacteria.

In Table 3 an overview of the results is given. A

period is called stable if no granule degradation

occurred and the biomass was not washed out. As

can be seen in Table 3, during all periods the shape

factor of the granules was constant around 0.45

and the aspect ratio was constant around 0.79.

These parameters appeared to be independent of

the settling velocity, the super®cial air velocity, the

COD loading and the hydraulic retention time.

The biomass density was 11.9 g/l. This was lower

than reported for granules or bio®lms formed in

airlift reactors, for which a biomass density of 20±

30 g/l (Kwok et al., 1996) and 15±20 g/l (Tijhuis et

al., 1994) has been reported.

A typical pattern of CO

2

in the o€gas, NH

4

+

and

ethanol concentrations and pH during one SBR-

cycle are given in Fig. 2. The pH was ¯uctuating a

little bit around 6.5. In all periods ethanol was con-

verted completely. Ethanol was consumed with a

constant, maximum rate until it was depleted. The

CO

2

produced is given in percentage in the o€gas

(=l CO

2

/l o€-gas/100). From the graph it is clear

that when ethanol was consumed the CO

2

o€-gas

concentration was maximal. NH

4

+

was consumed

during the whole cycle which indicates that biomass

growth occurred also when ethanol was absent. No

nitri®cation was taking place. After ethanol de-

pletion CO

2

was still produced due to conversion of

storage compounds (Van Loosdrecht et al., 1997).

Best period

The conditions applied in period 7 led to the best

granulation. The granules were stable and there was

a relative high amount of biomass accumulation in

the reactor up to a biomass concentration of 3.2 g/l

and a settled bed-volume of 1125 ml (this is 90% of

the volume under the e‚uent pipe). The SRT

gradually increased during this period (21 days)

from 1.8 day to 3.4 day. This was mainly due to the

fact that the bed-volume increased. The granules

were relative large with a D

av

of 3.3 mm. A photo-

graph of these granules is shown below (Fig. 3).

Comparing the results of the di€erent periods the

e€ects of several operational conditions can be

made clear.

COD loading

At a high COD loading there was, as expected,

more biomass growth. When the other parameters

(minimal settling velocity, super®cial gas velocity

and hydraulic retention time) were optimal, this

biomass could accumulate and the biomass concen-

Table 3. Overview of the results

Period Weeks Stable

Bedvolume

(ml)

Biomass concentration

(g/l)

Sludge retention time

(d)

Average diameter

(mm)

Shape factor

(±)

Aspect ratio

(±)

1

7

no

397

1.2

1.9

3.220.9

0.4620.02

0.86

2

10

no

464

1.3

2.8

3.020.5

0.5420.12

0.8520.04

3

8

yes

572 (525 4 850)

a

2.0 (1.3 4 2.7)

a

5.8 (2.1 4 11.7)

a

1.920.4

0.4120.06

0.6920.05

4

6

yes

581

2.0

3.1

2.120.3

0.4220.05

0.7320.02

5

0.3

no

733

4.0

0.6

2.0

0.37

0.78

6

0.3

no

663

2.2

1.3

2.5

0.43

0.78

7

3

yes

967 (700 4 1125)

a

2.8 (2.5 4 3.2)

a

2.7 (1.8 4 3.4)

a

3.3

0.50

0.81

8

1

yes

1158

n.m.

n.m.

4.6

0.44

0.81

a

Bedvolume increased signi®cantly during this period.

Fig. 2. Typical changes in concentrations during one cycle of the SBR reactor: (Q) NH

4

conc.; (W)

ethanol conc.; (ÐÐÐ) CO2 produced; (- - -) pH.

J. J. Beun et al.

2286

tration in the reactor increased. This was the case

in period 7 and 8. The COD loading did not appear

to have a direct e€ect on granulation within the

range tested. However it in¯uenced the ®nal form

of the granules (see below).

Hydraulic retention time

A low HRT should suppress suspended biomass

growth, due to wash-out of this suspended biomass

(Tijhuis et al., 1994). The HRT of 8 h in period 1

and 2 did not seem to be short enough to suppress

suspended biomass growth in the bubble column.

In combination with a low super®cial gas velocity

these operational conditions did not result in a

stable operation. Decreasing the HRT to 6.75 h

solved the problem and the stability of the granular

biomass in the bubble column improved. In a simi-

lar reactor set-up using sewage, Morgenroth et al.

(1997) also came to the conclusion that a shorter

HRT is bene®cial for granulation.

Minimal settling velocity

A high minimal settling velocity of 24 m/h could

be applied only when the COD load was low,

2.5 kg COD/(m

3

d)(period 4). At this COD-loading

rate less biomass accumulated in the reactor, so a

higher minimal settling velocity could be applied

without the granules hindering each others settling

too much.

When the minimal settling velocity was 24 m/h at

a COD load of 5 kg COD/(m

3

day) or higher, the

amount of granules was increased. Due to this the

settling was hindered too much to allow for a good

separation. The biomass did not settle below the

e‚uent pipe, so that part of it was washed out with

the e‚uent. Because the e‚uent pipe was ®xed, the

bedvolume could not become more than 1.25 l.

Decreasing the minimal settling velocity, so increas-

ing the settling time, led to better settling and more

biomass accumulation in the reactor.

Super®cial gas velocity

The gas ¯ow is the main cause of shear in the

reactor. At a relative high super®cial gas velocity of

0.041 m/s detachment of ®lamentous outgrowth

from the surface of the granules took place so that

more smooth granules were formed. The detached

biomass did not settle fast enough to be retained in

the reactor and was washed out with the e‚uent.

Only the granules were retained in the reactor for

weeks. This can be illustrated with the image analy-

zer data obtained from period 4, presented in

Table 4. The e‚uent contained small particles only

whereas the granules were retained in the reactor

(Fig. 4).

At a super®cial gas velocity of 0.041 m/s it was

even possible to increase the COD load to 7.5 kg

COD/(m

3

day) when the settling time was long

enough. Granulation occurred in this case and

Fig. 3. Photograph of the granules during period 7 (Dav = 3.3 mm).

Aerobic granulation in a SBR

2287

because of the high shear smooth, dense and stable

granules were formed with an average diameter of

4.6 mm (period 8).

A low super®cial gas velocity of 0.014 or

0.020 m/s did not lead to stable granules. This was

observed in periods 1 and 2. The settling time was

long enough and the COD load was low in both

cases but nevertheless the granules were not stable.

DISCUSSION

This and previous research (Morgenroth et al.,

1997) showed that it is possible to form granular

sludge in a SBR process. This could be partly

caused by the fact that the substrate is added pulse-

wise. This in general leads to a better sludge

settling. However by applying a short settling time,

i.e. selecting for biomass particles with a high

settling velocity, it was possible to obtain a granular

sludge in the reactor. This and other research

(Beeftink, 1987; Van der Hoek, 1988; De Beer et

al., 1993; Tijhuis et al., 1994) clearly show that

sludge granules can be formed by a wide variety of

organisms. Clearly granulation is not restricted to

certain microbiological groups, but related to the

way reactors are operated. Probably the same hy-

pothesis as proposed in the formation of bio®lms in

air-lift reactors can be used (Tijhuis et al., 1994).

Di€usion limitation increases going from suspended

biomass, to ¯occulated biomass to bio®lm or granu-

lar biomass. The latter biomass type will therefore

always grow slower then suspended cells. Only by

preventing the accumulation of suspended cells (by

the HRT) or ¯ocs (by the settling velocity) proper

granules (and bio®lms) will be formed.

Based on the microscopic observations done

during the research we can propose a mechanism

for the formation of granules in an aerobic reactor

without the presence of a carrier material. This pro-

posed mechanism is schematically depicted in Fig. 5.

After inoculation with bacterial sludge from a

COD-removing SBR fungi become dominating.

Fungi easily form mycelial pellets which settle very

well and can be retained in the reactor. Bacteria do

not possess that property and will be washed out

almost completely. Therefore, during start-up, the

biomass in the reactor will consist mainly of ®la-

mentous mycelial pellets. Due to the shear in the

reactor, detachment of the ®laments on the surface

of the pellets takes place and the pellets become

more compact. The pellets grow out to a diameter

of 5±6 mm and then they lyse probably due to oxy-

gen limitation in the inner part of the pellet. The

Table 4. Particle characteristics during period 4

Parameter

Reactor

E‚uent

Shape factor [±]

0.42

0.47

Aspect ratio [±]

0.73

0.74

Average diameter [mm]

2.1

1.2

Fig. 4. Photograph of the granules during period 4 (Dav = 2.1 mm).

J. J. Beun et al.

2288

mycelial pellets seem to function as an immobiliz-

ation matrix in which the bacteria can grow out to

colonies. When the mycelial pellets fall apart due to

lysis of the inner part of the pellets, the bacterial

colonies can maintain themselves because now they

are large enough to settle. These microcolonies

further grow out to granules, leading eventually to

a bacterial dominated population in the reactor.

It should be noticed that this proposed mechan-

ism is based on experiments in a reactor which was

started up with a small amount of suspended, non-

settling cells as an inoculum. If an inoculum should

be used which consists of ¯ocs and/or small gran-

ules, the mechanism will be di€erent. This can

already be con®rmed by observations from recent

experiments.

The granulation in the SBR described here shows

resemblance to the formation of granular bio®lms

in the continuously operated bio®lm airlift suspen-

sion (BAS) reactor (Tijhuis et al., 1994; Kwok et

al., 1996). The applied COD load in the SBR was

relatively low [7.5 kg COD/(m

3

day)] compared to

up to 20 kg COD/(m

3

day) in the BAS reactors.

This does not imply that a higher COD loading is

not possible in the SBR; it simply has not been

tried. The super®cial gas velocity in the BAS reactor

was 0.044 m/s, which is almost the same as in the

SBR which was operated as bubble column. The

HRT however was much shorter in the BAS reac-

tor, 40 min. Under these conditions stable and

dense bio®lms were formed with a biomass density

of 20 g/l and the biomass concentration in the BAS

reactor was 4 g/l. In other experiments with the

BAS reactor (Tijhuis et al., 1994) a biomass concen-

tration of 2 g/l was obtained and a biomass density

of 20 g/l when the COD load was 5 kg COD/

(m

3

day). In these experiments the HRT was also

less than 1 hour and the super®cial gas velocity in

the riser was again around 0.044 m/s.

The operational conditions applied in these SBR

experiments were comparable to those applied in

the BAS reactor. The most important di€erence

was the HRT, which was much longer in the SBR

experiments (6.75 h). The most important di€erence

in the results of these two reactors is that the bio-

mass density of the granules obtained in the SBR

was lower (11.9 g/l). The biomass concentration in

the SBR was also somewhat lower (3 g/l) than in

the BAS reactor.

In the BAS reactors the substrate load and the

shear appeared to be the dominant governing fac-

tors for stable bio®lm formation (Van Loosdrecht

et al., 1995; Kwok et al., 1996). In the SBR exper-

iments these factors were of importance as well. A

high COD loading leads easily to outgrowth of ®la-

ments, that hinder the settling and lead to unstable

reactor operation (period 5). If however the higher

loading rate is balanced by a high shear, more com-

pact granules are formed and a stable situation can

be reached (period 8). In the granule SBR, however,

also the time allowed for settling is as crucial,

mainly because it ®xes the amount of sludge ac-

cumulation. A too long settling time will result in

the formation of ¯occulated biomass. A too short

settling time does not lead to the accumulation of

sucient granules due to the hindered settling in

the lower compartment. If not enough granular

sludge is accumulated the loading stays too high

and no stable situation is reached.

CONCLUSIONS

In this study it has been shown that formation of

granular sludge in a SBR is possible. The dominant

requirement hereto is a short settling time, allowing

Fig. 5. Proposed mechanism of granulation after the start up of a SBR reactor with a short settling

time.

Aerobic granulation in a SBR

2289

for the wash-out of ¯occulated sludge. However the

settling time should be enough to allow for the hin-

dered settling of the granules to below the e‚uent

discharge point. The e€ect of sludge loading and

shear appear to be in the same trend as for bio®lm

formation in bio®lm airlift suspension reactors,

however more detailed research is needed. Also the

e€ect of the in¯uent intervals should be part of

future research.

The results show that it may be possible to

achieve more compact SBR processes. Due to the

short settling time a larger part of the cycle can be

used for process activity. Due to the granular

sludge high sludge concentrations can be reached,

e€ectively due to the low sludge volume index.

REFERENCES

Beeftink H. H. (1987) Anaerobic Bacterial Aggregates.

Ph.D.

thesis,

University

of

Amsterdam,

The

Netherlands.

De Beer D., van der Heuvel J. C. and Ottengraf S. P. P.

(1993) Microelectrode measurements in nitrifying aggre-

gates. Appl. Env. Microbiol. 59, 573±579.

Kwok W. K., van Loosdrecht, M. C. M. and Heijnen, J.

J. (1996) Application of a bio®lm airlift suspension reac-

tor for acetic acid removal. Internal report, Delft

University of Technology, Delft, The Netherlands.

Lettinga G., Hulsho€ Pol L. W. and Zeeman, G. (1993)

Biologische Waterzuivering, AnaeÈrobe Zuivering (biologi-

cal Wastewater Treatment, Anaerobic Treatment).

Department of Environmental Technology, Wageningen

Agricultural University, The Netherlands.

Morgenroth E., Sherden T., van Loosdrecht M. C. M.,

Heijnen J. J. and Wilderer P. A. (1997) Aerobic granu-

lar sludge in a sequencing batch reactor. Water Res.

31(12), 3191±3194.

Smolders G. J. F., Klop J., van Loosdrecht M. C. M. and

Heijnen J. J. (1995) A metabolic model of the biological

phosphorus removal process. I. E€ect of the sludge

retention time. Biotechnol. Bioeng. 48, 222±233.

Tijhuis L., van Loosdrecht M. C. M. and Heijnen J. J.

(1994) Formation and growth of heterotrophic aerobic

bio®lms on small suspended particles in airlift reactors.

Biotechnol. Bioeng. 44, 595±608.

Van Benthum W. A. J., Garrido-FernaÂndez J. M., Tijhuis

L., van Loosdrecht M. C. M. and Heijnen J. J. (1996)

Formation and detachment of bio®lms and granules in

a nitrifying bio®lm airlift suspension reactor. Biotechnol.

Prog. 12(6), 764±772.

Van der Hoek J. P. (1988) Granulation of denitrifying

sludge. In: Granular aerobic sludge, ed. G. Lettinga, A.

J. B. Zehnder, J. T. C. Grotenhuis and L. W. Hulsho€

Pol. Pudoc, Wageningen, The Netherlands, pp. 203±210.

Van Loosdrecht M. C. M., Eikelboom D., Gjaltema A.,

Mulder A., Tijhuis L. and Heijnen J. J. (1995) Bio®lm

structures. Water Sci. Tech. 32(8), 35±43.

Van Loosdrecht M. C. M., Pot M. A. and Heijnen J. J.

(1997) Importance of bacterial storage polymers in bio-

processes. Water Sci. Tech. 35(1), 41±47.

J. J. Beun et al.

2290