Effect of polymeric substrate on sludge settleability

w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 2 6 3 e2 7 3

Available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/watres

Effect of polymeric substrate on sludge settleability Anto´nio M.P. Martins a,1, O¨zlem Karahan a,b, Mark C.M. van Loosdrecht a,c,* a

Department of Biotechnology, Delft University of Technology, Julianalaan 67, 2628 BC Delft, The Netherlands Istanbul Technical University, Faculty of Civil Engineering, Environmental Engineering Department, 34469 Maslak Istanbul, Turkey c KWR watercycle research institute, Groningenhaven 7, 3422 PE Nieuwegein, The Netherlands b

article info

abstract

Article history:

The study aims to evaluate the role of a polymeric substrate (starch) on sludge settleability.

Received 5 April 2010

Despite being an important COD component of the wastewater, the relationship between

Received in revised form

polymeric substrates and bulking sludge has been hardly studied. The polymers are hydro-

13 July 2010

lysed at a rate smaller than the consumption rate of monomers. This means that the soluble

Accepted 18 July 2010

substrate resulting from hydrolysis is likely to be present at growth rate limiting concen-

Available online 27 July 2010

trations. According to the kinetic selection theory this leads to bulking sludge. However, a recently postulated theory suggests that, strong diffusion limited micro-gradients of

Keywords:

substrate concentration inside flocs lead to bulking sludge, and not a low substrate

Bulking sludge

concentration as such. If the polymeric COD is first incorporated in the sludge floc and

Hydrolysis

afterwards hydrolysed in the sludge floc then there is essentially no substrate gradient inside

Kinetic selection

the biological flocs. The experiments showed that conditions leading to bulking sludge with

Sludge settleability

monomers (glucose) did not lead to bulking when starch was used. A bulking sludge event

Starch

was even cured just by substituting the monomer with starch. These results are clearly in line

Storage

with a diffusion gradient e based theory for bulking sludge. Nevertheless, flocs growing on starch are more open, fluffy and porous than flocs formed on maltose or glucose, most likely because the starch needs to be hydrolysed at the surface of the micro-colonies forming the flocculated sludge. Some additional observations on occurrence of filamentous bacteria in oxygen diffusion limited systems are also discussed in this manuscript. ª 2010 Elsevier Ltd. All rights reserved.

1.

Introduction

Particulate substrate, often denoted as slowly biodegradable COD, is an important fraction of the total COD present in the wastewater. For instance, in The Netherlands particulate COD varies usually between 30 and 50% of the total COD (Kruit et al., 1994) while in other countries even higher values are reported: 70e90% in South Africa (Casey et al., 1999) and about 90% in Switzerland (Kappeler and Gujer, 1994). This type of substrate has a high molecular weight and is supposed to

undergo cell external hydrolysis before becoming available for consumption (growth and storage) by bacteria (Gujer et al., 1999). Polymers such as proteins, lipids and polysugars are major components of this COD fraction. Hydrolysis is widely considered the rate-limiting step of the overall COD removal process, and as such included in general activated sludge models (Gujer et al., 1999). This means that hydrolytic products, i.e. soluble substrate, are consumed at a higher rate than they are produced and, likely, will be available for the microorganisms at low (growth rate limiting)

* Corresponding author. Department of Biotechnology, Delft University of Technology, Julianalaan 67, 2628 BC Delft, The Netherlands. Tel.: þ31 15 2781618. ¨ . Karahan), M.C.M.vanLoosdrecht@ E-mail addresses: [email protected] (A.M.P. Martins), [email protected] (O tnw.tudelft.nl (M.C.M. van Loosdrecht). 1 ´ guas do Algarve, S.A., Rua do Repouso 10, 8000-302 Faro, Portugal. present address: A 0043-1354/$ e see front matter ª 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2010.07.055

264

w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 2 6 3 e2 7 3

concentrations. According to the kinetic selection theory (Chudoba et al., 1973) these low soluble substrate concentrations give competitive advantages to filamentous bacteria, leading to bulking. Recently, an alternative hypothesis states that it is not the substrate concentration as such but the microgradients of substrate concentration inside biological flocs that play a more important role in the competition between filamentous and non-filamentous bacteria (Kappeler and Gujer, 1994; Martins et al., 2004a,b). Hydrolysis is generally found to occur at floc level (Dold et al., 1991; San Pedro et al., 1994; Mino et al., 1995; Goel et al., 1998, 1999; Mosquera-Corral et al., 2003; Karahan et al., 2006). Therefore, it can be expected that there will be no real gradients in substrate concentration inside the floc, although the substrate concentration inside the floc will be very low. The objective of this study was to evaluate the effect of polymeric substrate on the development of filamentous bacteria. Hitherto, well-controlled lab-scale systems have been used, as this allow a proper “scale-down” of the conditions bacteria experience in full-scale wastewater treatment plant containing a selector (Martins et al., 2003a, 2004c). For this study fully aerobic systems, fed with a polymeric substrate (potato starch) and easily biodegradable soluble monomers (glucose or maltose) were used as control systems. The use of different feeding periods and dissolved oxygen concentrations allowed to simulate different bulk liquid substrate and oxygen gradients and a variable relative size of the selector.

2.

Material and methods

2.1.

Experimental setup

The experiments were performed in seven different sequencing batch reactors (SBRs) with 2 L working volume. The reactors were controlled and monitored online by a Bioprocessor (Applikon bioprocessor ADI 1030, Schiedam, The Netherlands) connected to the Biodacs data acquisition program (Applikon, Schiedam, The Netherlands). The systems were controlled at a temperature of 20  C, and pH of 7.0 using 1 N HCl or 1 N NaOH.

Each reactor was operated continuously for 60e90 days, in cycles of 4 h with 10 min of aerobic mixing time (0e10 min in the cycle), 3 min (SBR 1, 2, 3, 6 and 7) or 50 min (SBR 4 and 5) of aerobic filling time (10e13 min or 10e60 min in the cycle), 140 min of aerobic reaction time (10e150 min in the cycle), 2 min of sludge withdrawal (148e150), 80 min of settling time (150e230 min in the cycle) and 10 min of effluent discharge time (230e240 min in the cycle). At the end of the cycle 1 L effluent was pumped out of the reactors, resulting in a volumetric exchange ratio of 0.5 and in a hydraulic residence time of 8 h. The applied organic loading rate was 1.2 g chemical oxygen demand (COD) L1 day1 in all experiments. It was tried to keep the solid retention time (SRT) at 10 days but due to a deficient control of the sludge withdrawal pump the actual SRT in pseudo-steady state was considerably different in SBR 2 (7.0 days) and 3 (15.6 days) (Table 1). The length of the aerobic feed phase (or the aerobic fill time ratio, calculated as the quotient of the time for aerobic fill and the total time of one cycle), the type of organic substrate and the mixed liquor dissolved oxygen concentration in the feast phase, when external substrate is present (comparable to the selector), were the operational parameters that were varied in SBR setups (Table 1). Systems with a short feed period (pulse feed in 3 min in SBRs 1, 2, 3, 6, 7, in Table 1) had initially a high substrate concentration as in a process with a plug flow selector with full substrate removal. The other systems, i.e. SBRs 4 and 5, can be compared to a (over-dimensioned) completely mixed selector, with low substrate concentrations, followed by a completely mixed reactor. The reactors were stirred with two standard geometry sixblade turbines. In each cycle two different stirrer speeds were used: 300 rpm during the first hour (allowing a fast mixing and good aeration in the feast phase) and 150 rpm during the remainder famine phase of the cycle (to decrease the turbulence, minimizing the floc break-up). In normal operation, the bulk liquid dissolved oxygen concentration in the feast phase 1 (Sfeast O2 ) was maintained at different levels (i.e., >2.0 mg O2 L 1 in SBR 1e5; <0.2 mg O2 L in SBR 6e7) by adjusting the airflow rate (i.e., 1.0 NL min1 in SBR 1e5 and 0.25 NL min1 in SBR 6e7) and stirring. In the famine phase, when external

Table 1 e Operational parameters e aerobic fill time ratio, organic substrate, dissolved oxygen concentration in the feast phase (Sfeast O2 ), biomass concentration, solids retention time, feast period, sludge loading rate, and floc loading rate e in pseudo-steady state systems. SBR system

Aerobic fill phase Time (min)

1 2 3

10e13 10e13 10e13

4 5

10e60 10e60

6 7

10e13 10e13

Fill time ratio (%) 1.25 1.25 1.25 20.8 20.8 1.25 1.25

Sfeast Sludge Solids Feast Sludge Floc Organic O2 mg O2 L1 content retention period loading rate loading rate substrate in mg TSS L1 time day min kgCOD kgTSS1 kgCOD kgTSS1 normal cyclesa day1 day1 Starch Maltose Starch:maltose (COD basis 1:1) Starch Glucose; starch after 40 days Starch Glucose

>2 >2 >2

4260 3500 5300

11.3 7.0 15.6

>2 >2

4100 3700

8.8 9.5

<0.2 <0.2

5020 4040

11.3 10.3

4.7 3.0 4.6

0.28 0.34 0.22

50 50

0.29 0.32

21 14.5

0.24 0.30

23 27 18 1.4 1.6 19 24

a When the pseudo-steady state was reached transient responses to dump fill of starch, maltose, glucose and mixture starch-maltose on a COD basis 1:1, were measured in each system.

w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 2 6 3 e2 7 3

substrate is depleted, the dissolved oxygen concentration was in all SBR’s always above 2 mg O2 L1 to prevent any effect of oxygen limitation in this phase (Table 1).

2.2.

Feed solution

The synthetic wastewater used as influent was sterilized at 110  C during 40 min. Depending on the SBR system (Table 1) it contained potato starch (filterable through 0.2 mm pore size filters after sterilization, Fulka 85643), maltose (Sigma 5885), or glucose (J. T. Baker 0115) 12.5 CmM (400 mg COD L1) as the carbon source. Glucose was selected as the representative monomer of saccharides, and maltose was used since it is the main compound (a di-mer) generated through the enzymatic hydrolysis of starch. The rest of the nutrients were added by the feed solution containing NH4Cl 1.5 mM (21 mg NL1), KH2PO4 0.48 mM (15 mg PL1), MgSO4$7H2O 0.37 mM, and 1 mL. 1 L1 influent of the following trace solution: EDTA 50 g L , 1 1 ZnSO4$7H2O 22 g L , CaCl2$2H2O 8.18 g L , MnCl2$4H2O 5.06 g L1, FeSO4$7H2O 4.99 g L1, (NH4)6Mo7O24$4H2O 1.1 g L1, CuSO4$5H2O 1.57 g L1, CoCl2$6H2O 1.61 g L1.

2.3.

Activated sludge inoculums

A mixture of fresh sludge from a domestic wastewater treatment plant (1.5 L) and sludge adapted to potato starch (0.5 L), which has been stored at 4  C during 6 months, was used as inoculum in SBRs 1 and 3. The other systems, i.e., SBRs 2, 4, 5, 6 and 7, were inoculated with a mixture of fresh activated sludge (0.3 L) and sludge (1.7 L) coming from SBRs 3, 1, 4, 5 and 2, respectively.

2.4.

Calculation procedures

Achievement of a pseudo-steady state was evaluated from a constant total organic carbon (TOC) and biomass concentration (given by the volatile suspended solids) in the reactor, and when the dynamic pattern in dissolved oxygen during a cycle (i.e., the length of the feast phase) no longer changed. When the organic substrate was consumed the dissolved oxygen concentration rapidly increased. This transition point was used to determine the feast and famine periods. In pseudo-steady state, full SBR cycles were analysed for kinetic characterization of the sludge. This was done through dump filling the SBR systems (the influent was instantaneously added at cycle time of 10 min) with different sugars: starch, maltose, glucose and mixture starchemaltose on a COD basis 1:1. Therefore, kinetics was measured both in normal operational conditions (with the same type of sugar as in pseudosteady state systems) and in transient operational conditions (with a different type of sugar). The experiments were performed in triplicates. The reported data are the average, the standard deviation was always below 7% for clarity this was however not included in the tables. Before each sampling cycle, 100 mg allylthioureum was added to the reactor to prevent interference of nitrification in the measurements. The behaviour of the reactors was characterized by sludge volume index (SVI), specific organic substrate uptake rate in the ), specific poly-glucose storage compounds feast phase (qfeast s production rate and fraction of poly-glucose produced in feast the feast phase (qfeast polyglucose and fpolyglucose ), fraction of sugars

265

consumed and used for synthesis of poly-glucose (Sugarfeast polyglucose ), and specific oxygen consumption rate in ). Sugarfeast the feast phase (qfeast polyglucose was calculated as the O feast feast ratio between qs =qpolyglucose and the maximum yield of polyglucose storage compounds on glucose (0.91 Cmol Cmol1, Dircks et al., 2001). The first samples were taken just before addition of the feed. The last sample was taken just before the settling phase. The feast phase started with the feed phase, and ended when the organic substrate was fully consumed and, simultaneously, internal poly-glucose degradation started. The elemental biomass composition was assumed to be CH1.56O0.59N0.19 (Dircks et al., 2001). The poly-glucose content of the active biomass was calculated as the amount of sugars measured at a certain time in the cycle minus the sugars measured after 24 h of famine aerobic conditions, which were assumed to be mainly intracellular non-stored sugar compounds. All the remaining calculation procedures are described elsewhere (Martins et al., 2003a; Dircks et al., 2001).

2.5.

Hydrolytic activity

The location of hydrolytic activity during starch degradation was determined in batch experiments with (initial ratio substrate/microorganisms of 45 mg starch g1MLVSS) and without biomass. Tests in the absence of biomass were performed using 1 L of mixed liquor collected after 30 min of settling (cycle time of 180 min) and filtered throughout 0.2 mm pore size Gelmann filters. Nitrogen gas was sparged in the bulk liquid during the batch experiments and pH was controlled at 7.0. A spike of potato starch was applied and samples were taken over the time. Starch and TOC (total and soluble) were measured and the hydrolytic activity in the bulk liquid was calculated. In the presence of biomass the hydrolytic activity was detected by staining with iodine. The hydrolytic activity was also evaluated after inactivation of the metabolic activity of the biomass, either by sparging the bulk liquid with nitrogen or adding formaldehyde (1 mL in 100 mL of activated sludge).

2.6.

Analysis

Starch, glucose, poly-glucose storage compounds, soluble and total organic carbon, poly-b-hydroxybutyrate (PHB), ammonium, mixed liquor total and volatile suspended solids, sludge volume index (SVI), pH, dissolved oxygen, oxygen uptake rate, carbon dioxide and hydrolytic activity were measured during the experiments. Carbon dioxide in the incoming air and in the off-gas was measured online with an infrared carbon dioxide analyser. For glycogen determination 4.5 mL of sample was added to 0.5 mL of 6 M HCl. The samples were then digested at 100  C during 5 h. After cooling, the samples were centrifuged and the glucose concentration was measured with highpressure liquid chromatography (HPLC). Starch was determined by the starcheiodine complex (SIC) method (San Pedro et al., 1994). The remaining analytical methods used are as described in (Martins et al., 2003a). Microscopic analysis of the activated sludge was performed according to the reference manuals (Jenkins et al., 1993; Eikelboom, 2000). The method of subjective scoring of filamentous bacteria abundance (Jenkins et al., 1993) was used to

266

w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 2 6 3 e2 7 3

quantify the abundance of filamentous bacteria present in the samples. Whenever available, fluorescent 16S rRNA probes (Interactiva, Ulm, Germany) were used to identify specific filamentous bacteria. The samples were fixed with paraformaldehyde for Gram-negative bacteria analysis or with ethanol for Gram-positive bacteria analysis. Fluorescent probes were hybridised in situ according to (Manz et al., 1992). Table 2 shows a list of oligonucleotide probes used in this study. Samples for fluorescent in situ hybridisation (FISH) to detect Gram-positive bacteria were treated with 0.1% lysozyme during 15 min at 37  C in 10 mM phosphate buffer, pH 6.5. The effect of the type of substrate, i.e., polymer or monomer, on floc morphology was quantitatively evaluated in the systems SBR 1, 2 and 3 by an image analyser (Galai Cue 2 v. 4.7) coupled with stereo microscopy by a CCD camera. Equivalent diameter, superficial area, aspect ratio and shape factor were the measured parameters. Six hundred particles per sample were analysed in each sample and the reported results are an average of five samples.

Results

3.1.

General observations

The average values of some traditional operating SBR parameters, such as the fill-time ratio, dissolved oxygen concentration in the feast phase, biomass concentration, solids retention

Table 2 e Oligonucleotides probes used in this study.

EUB338

Specificity

Many but not all bacteria EUB338-II Planctomycetales EUB338-III Verrucomicrobiales HGC69a High-GC, Actinobacteria LGC354A, LGC354B, Low-GC, Firmicutes LGC354C ALF968 a-Proteobacteria BET42a b-Proteobacteria GAM42a g-Proteobacteria CF319a Cytophaga-Flavobacterium of Bacteroidetes MPA60, MPA223, ‘Microthrix parvicella’ MPA645 SNA23a ‘Sphaerotillus natans’ PAO462, PAO651, ‘Candidatus PAO846 Accumulibacter phosphatis’ GAOQ431, ‘Candidatus GAOQ989 Competibacter phosphatis’ AMAR839 Genus Amaricoccus ‘G-Bacteria’ TET63 Tetrasphaera spp. ‘G-Bacteria’ actino-1011 A High-GC group closely related to Tetrasphaera sp. NlimII175, Eikeloom morphotype NlimII192 Nostocoida limicola II

Reference Amann et al., 1990 Daims et al., 1999 Daims et al., 1999 Roller et al., 1994 Meier et al., 1999 Manz Manz Manz Manz

et al., et al., et al., et al.,

Sludge characteristics

The activated sludge from the wastewater treatment plant, which was used as partial inoculum in SBRs 1 and 3, contained Microthrix parvicella (confirmed with specific 16S rRNA probe, excessive amounts) and a range of other filamentous bacteria (mainly Types 0041/0675 and Thiothrix sp., very common) in high number. Consequently, the initial SVI in these systems was considerably high (320 mL g1) (Fig. 1). The abundance of these bacteria decreased and after two weeks the SVI was less than 130 mL g1. Except in SBR 5 well settling sludge (SVI < 100 mL g1) was always obtained. Pulse fed systems with monomers (e.g., maltose in SBR 2) produced a slightly better SVI (in the range 30e50 mL g1) than pulse fed systems with starch (SVI in the range 80e100 mL g1, SBR 1 and 6). The only bulking event was registered in the system fed during a long period with glucose (SBR 5, SVI of 340 mL g1 after 40 days). Changing the substrate to starch led to a considerable improvement in the sludge settleability, with SVI decreasing to less than 120 mL g1 after 10 days. Sphaerotilus natans was the filamentous bacteria responsible for the bulking event in SBR5 (Fig. 2c). After starch has been used in the feed solution S. natans was out competed by other bacteria and decreased in number. Higher organisms were also commonly observed in the systems. While rotifers (metazoa) were abundant in systems

1992 1992 1992 1996

Erhart et al., 1997

400 300 -1

Probe name

3.2.

SVI (ml g )

3.

time, feast period, sludge loading rate, and floc loading rate are shown in Table 1. Pseudo-steady state was generally reached one month after inoculation. Except in SBRs 4 and 5 the substrate was always present in excess during the feast phase. The soluble substrate uptake rates were, however, not always the maximum rate due to a slow hydrolysis of polymeric substrate (SBRs 1 and 3) or to oxygen limitation (SBRs 6 and 7). Only in SBR 2 the substrate, maltose, was taken up at the maximum rate. In SBRs 4 and 5 the addition of substrate was relatively slow and the uptake rates were determined by the addition rate. SBR 4 simulates the slow release of soluble substrate due to hydrolysis.

Wagner et al., 1994 Crocetti et al., 2000

Crocetti et al., 2002

Maszenan et al., 2000

200 100 0 0

Kong et al., 2001 Liu et al., 2001

Liu and Seviour, 2001

15

30 45 60 75 Experiment time (days)

90

Fig. 1 e Sludge settleability expressed as SVI during the whole operating period: (B) SBR 1; ( ) SBR 2; (D) SBR 3; (O) SBR 4; (,) SBR 5; (B) SBR 6; (>) SBR 7. The arrow indicates the day (40) when the feed in SBR 5 was changed from glucose to starch.

w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 2 6 3 e2 7 3

267

Fig. 2 e Photomicrographs of fluorescent in situ hybridization (FISH) (a, c, and d) and a phase contrast micrograph (b) of sludge at pseudo-steady state in: a and b pulse fed system with starch (SBR 1); c long fed system with glucose (SBR 5); d pulse fed system with glucose at low dissolved oxygen concentration (SBR 7). The length of the bars corresponds to: a and c 20 mm; b 100 mm; d 10 mm. fed with starch (Fig. 2b), vorticella spp. (protozoa) predominated in the monomers fed systems. The floc morphology was partially quantified in SBRs 1, 2 and 3 (Table 3). Compact, smooth and larger shaped flocs (equivalent diameter usually larger than 250 mm) were observed in the maltose fed system (SBR 2). All the starch fed systems were dominated by smaller and porous flocs (equivalent diameter usually smaller than 200 mm). Although not the dominant microorganisms, short and coiled filaments (trichome length usually smaller than 200 mm) resembling the morphotype Nostocoida limicola II, placed inside the floc, were commonly observed in the starch fed systems. This filamentous bacterium belonged to the Gram-positive bacteria with high DNA G þ C content group (positive signal with the probe HGC69a, Fig. 2a) but it did not hybridize with the specific probes designed for its detection (Liu and Seviour, 2001). In all the systems the dominant bacterial group was the Gram-positive high DNA G þ C content group, followed by the b- and asubclass of Proteobacteria (Bet42a and Alf968 16S rRNA probes),

Table 3 e Morphological characteristics of the flocs in SBR systems 1, 2 and 3. Parameter

Projected area Aspect ratio Shape factor Equivalent diameter

Units

mm2

mm

SBR system 1

2

3

0.02 0.62 0.31 160

0.04 0.64 0.49 350

0.02 0.64 0.33 190

roughly estimated as being respectively 50%, 30% and 5% of the total bacteria population (EUB338 þ II þ III 16S rRNA probes). G-bacteria belonging to the genus Amaricoccus of the a-subclass of Proteobacteria were not detected. In the pulse fed system with glucose at low dissolved oxygen concentration coccoid cells predominantly in tetrad arrangements were clearly dominant. These bacteria also hybridized positively with the HGC69a 16S rRNA probe (Fig. 2d). No hybridization occurred with the oligonucleotide probes TET63 and action-1011, specific respectively for Tetrasphaera spp. ‘G-Bacteria’ and a high-GC group closely related to Tetrasphaera spp.

3.3.

Cycle behaviour

Bulk liquid dissolved oxygen concentration and carbon dioxide profiles had a similar trend throughout the cycles of the different experiments (Fig. 3). In the first 10 min the bulk liquid dissolved oxygen concentration and the stripping rate of carbon dioxide increased due to mixing and aeration of the sludge. When the organic substrate was added the dissolved oxygen concentration decreased due to organic substrate consumption. When the carbon source was consumed the dissolved oxygen concentration rapidly increased, indicating the end of the feast phase. The stripping of carbon dioxide decreased in the feed phase due to low bicarbonate content in the feed solution. Carbon dioxide is produced at high rates in the feast phase. The increase in stripping of carbon dioxide could still be measured in the feast phase. Opposite behaviour was observed when all the substrate was consumed. The decrease in dissolved oxygen concentration and stripping of

268

w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 2 6 3 e2 7 3

Fig. 3 e Typical dissolved oxygen and carbon dioxide profiles in the pseudo-steady state pulse fed system with maltose and starch (SBR 3) when exposed to a dump fill of glucose (e), maltose (d), and starch ( ). The arrows indicate the decrease of (a) oxygen consumption and (b) carbon dioxide production.

carbon dioxide after 1 h was caused by the decrease in stirring speed. During settling and withdrawal phases the mixing and the aeration were stopped and dissolved oxygen concentration decreased. Carbon dioxide was not measured during these phases because no aeration was applied. All the other monitored parameters showed a typical feastefamine cycle behaviour (Fig. 4). When the substrate was present, a linear decrease of substrate and a linear increase of poly-glucose were detected. The observed growth yield for starch was slightly higher (0.6 Cmol Cmol1) than for maltose and glucose (0.5 Cmol Cmol1), which corresponds to the lower net carbon dioxide production with starch (Fig. 3b). The observed yield of poly-glucose storage compounds on the substrate also changed with the type of oligosaccharide. While maltose gave a yield of 0.84 Cmol Cmol1 (SBR 2) starch pulse fed system gave a storage yield of 0.70 Cmol Cmol1 (SBR 1). When the substrate was present in limiting concentrations (SBR 4 and 5) the storage yield decreased about two times (0.3e0.4 Cmol Cmol1), indicating less storage and relative more substrate being used for growth in these conditions. A very small fraction of PHB was also formed, usually less than 8% of the total carbon present in the influent. In the famine phase the storage compounds were slowly consumed. Phosphate was released in the feast phase in the range of 11 (SBR 1)e44 (SBR 7) Pmmol Cmol1, indicating hydrolysis of

Fig. 4 e Typical change in concentrations during a pulse feed cycle with glucose at (a) high dissolved oxygen concentration (SBR 3) and (b) low dissolved oxygen concentration (SBR 7): (,) glucose; (A) poly-glucose storage 3L compounds; (3) PHB; (B) NHD 4 ; (:) PO4 . poly-P to produce energy for the active transport of saccharides through the cytoplasmic membrane. In the glucose pulse fed system operated at low dissolved oxygen concentration (SBR 7) more phosphate was released in the feast phase (about 0.35 Pmmol L1) than in the remaining systems (usually less than 0.15 Pmmol L1). Phosphorus removal was observed in SBR 7 since the phosphate concentration in the effluent was usually lower than 1 mg PL1. The phosphorous content of the biomass in this system was 5% of the cellular dry weight. Specific gene probes for phosphorus accumulating organisms (PAOs) and glycogen non-polyphosphate accumulating organisms (GAOs) (Crocetti et al., 2000, 2002) were applied but no positive signal was obtained, indicating the absence of these organisms. Neisser staining revealed the presence of positive granules, presumably poly-P, inside the dominant coccoid gram-positive bacteria with high DNA G þ C content, predominantly arranged in packages of tetrads.

4.

Discussion

4.1. Hydrolytic activity, kinetics and storage aspects of starch conversion In pseudo-steady-state systems the starch was quickly and completely adsorbed on the biomass, as observed by the iodine

w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 2 6 3 e2 7 3

test (blue spots on the floc surface and not in the bulk liquid, indicating the formation of the starcheiodine complex at floc level). This indicates that the hydrolysis was strongly associated with the activated sludge flocs. Starch was hydrolysed at a rate in the range of 500 (SBR 4)e2100 (SBR 1) mg starch L1 h1. Glucose and maltose (analysed by HPLC), and other soluble sugars (by total organic carbon) were not detected in the bulk liquid. This indicates that the hydrolytic products are consumed close to the site of production, i.e. at floc level. The hydrolysis products did not diffuse outside of the floc as hypothesized by others (Ekama and Marais, 1986). The maximum hydrolytic activity measured in the cell-free bulk liquid of SBRs 1, 3, 4 and 6 varied from 5 to 33 mg starch L1 h1, indicating that a very small fraction of starch (less than 7%) could be hydrolysed in the bulk liquid. Starch hydrolysis in the bulk liquid followed first order reaction kinetics, with a reaction rate constant in the range of 0.02 (SBR 1)e0.2 (SBR 3) h1. Inhibition of cell metabolism in the sludge flocs gave a similar hydrolytic activity, which is in line with studies indicating that most of the enzymes are extracellular and immobilized in the exopolymeric substances (EPS) matrix (Frølund et al., 1995). The maximum hydrolysis rate obtained in this study (2100 mg starch L1 h1; 0.47 Cmol Cmol1 h1; 11.9 gCOD gCOD1 d1) is slightly higher than the rates reported in literature (0.30 Cmol Cmol1 h1 in San Pedro et al., 1994 and 3 gCOD gCOD1 d1 in Gujer et al., 1999). The high hydrolysis rate was, however, four to five time times smaller than the maximum soluble monomer uptake rate (2.51 Cmol Cmol1 h1 for maltose in this study, and 2.02 Cmol Cmol1 h1 for glucose in Dircks et al., 2001). A similar trend was observed for the poly-glucose specific production rates. Storage in starch pulse fed systems was slightly less than that of monomer fed systems, indicating that more growth occurred in the former just as if the monomer had been dosed at a slower rate. Even so, the storage capacity with starch was very high (almost 80% of the starch was stored, similar to the result of 75% storage obtained by Karahan et al., 2006), indicating that storage of polymeric substrates can occur as implicitly assumed in the Activated Sludge Model No. 3 (Gujer et al., 1999). Furthermore, these results confirm that storage polymers are an intrinsic part of microbial physiology and ecology of activated sludge processes (van Loosdrecht et al., 1997; Goel et al., 1998). The results of the kinetic characterization of the activated sludge from the different systems are summarized in Table 4. The specific substrate uptake rate in system SBR 1 reflects the rate-limiting step, i.e. hydrolysis of starch. In the long fed systems (SBRs 4 and 5) or with low dissolved oxygen in the feast phase (SBRs 6 and 7) the specific substrate uptake rate reflects mainly the feeding pattern of the system. When to these systems a pulse feed was applied to and oxygen was not limiting the specific substrate uptake rate increased more than two times. The obtained rates were, however, considerably lower than the rates obtained in the continuously pulse fed systems (e.g. 0.44 Cmol1 Cmol1 h1 in SBR 5, where the maximum glucose uptake rate in pulse fed systems was 2.02 Cmol1 Cmol1 h1, in Dircks et al., 2001).

4.2.

Microbiology aspects

Bulking sludge only occurred when the monomer was added at a strong limiting rate (SBR 5, fill time ratio of 25%;

269

q=qmax ¼ 0.10 Cmol Cmol1), similar to observations of previous s reports (Martins et al., 2003a,b). S. natans was the dominant filamentous bacterium in SBR 5 and it was apparently absent at conditions of low dissolved oxygen concentration (SBR 7). These observations are not in agreement with the widely known relationship between filamentous bacteria, causative conditions and control measures (Jenkins et al., 1993), in which S. natans is classified as being a bacterium commonly found in environments with low dissolved oxygen concentration and abundance of soluble substrates. Although they might be correct for some bacteria such relationships have still a high degree of uncertainty and precaution should be a rule whenever they are applied. Surprisingly in strongly diffusion limited oxygen environments, with high concentration of soluble substrate in the bulk liquid (SBR 7), filamentous bacteria were not dominant and well settling sludge was always obtained. These results are conflicting with previous observations in similar conditions, but in which acetate was used as carbon source instead of glucose (Martins et al., 2003b). It might well be that on glucose-fed systems easily a population developed that could use polyphosphate as energy source for glucose uptake and storage; compensating for the lack of oxygen. This aspect was not further investigated but, as shown in other studies (Martins et al., 2004c), the presence of bio-P activity clearly induces a well settling sludge. With the currently available FISH probes we could not identify the involved bacteria. It shows however how complicated the relation between microbial ecology and bulking sludge is.

4.3.

Relation polymeric substrate and sludge settleability

The model polymeric substrate used in this study (potato starch) did not lead to bad settling sludge in any of the experiments. Conditions leading to bulking with soluble monomers (i.e. slowly fed systems simulating a low substrate concentration in the bulk liquid during substrate uptake as occurring in completely mixed systems) did not produce bulking sludge with polymeric substrate. Replacing the soluble monomer (glucose) with starch even cured the bulking event (Fig. 1). In the experiments filamentous organisms belonging to the Gram-positive bacteria with high DNA G þ C content group were commonly present inside the flocs in the systems fed with starch (Fig. 2a). Although not dominant, their presence together with smaller, porous, fluffy and irregularly shaped and, apparently, less compact flocs led to a slightly higher SVI (80e100 mL g1) than in the soluble substrate fed systems (e.g. SBR 2, SVI ¼ 30e50 mL g1). The occurrence of a more open floc structure in the presence of hydrolysable substrates might be the result of the fact that most of the starch adsorbs to the EPS e bulk liquid interface, where it has to be hydrolysed before the monomers can diffuse into the EPS matrix and reach the microbial cells. If the micro-colonies in a floc get too large the bacteria in the interior of the microfloc would not get any substrate and will decay. It could also be that filamentous bacteria create a very open floc matrix like often observed (Wile´n et al., 2003). Since the filamentous bacteria did not extend outside the flocs we propose that the effect is mainly due to the formation of smaller microflocs that flocculate into the larger floc structure. Interestingly in biofilms (Mosquera-Corral et al., 2003) and aerobic granular sludge (de Kreuk et al., submitted for publication) the

270

w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 2 6 3 e2 7 3

Table 4 e Average specific rates in normal and pulse feed cycles, fraction of poly-glucose storage compounds produced in the feast phase and fraction of sugar consumed and used for synthesis of poly-glucose. All SBR 5 measurements were made during the bulking episode. Values in bold indicate maximum rates or fractions. Values between brackets were calculated. Parameter

Unit

C source

SBR system 1

qfeast s

qfeast polyglucose

feast

qO2

qfeast s qmax s

qfeast polyglucose qfeast s

qfeast polyglucose qmax polyglucose

feast fpolyglucose

f Sugarfeast polyglucose

a b c d e f

Cmol Cmol1 h1

Cmmol Cmol1 h1

O2 mmol Cmol1 h1

Cmol Cmol1

Cmol Cmol1

Cmol Cmol1

Cmol Cmol1

%

Starch Maltose Starch þ maltose Glucose Starch Maltose Starch þ maltose Glucose Starch Maltose Starch þ maltose Glucose Starch Maltose Starch þ maltose Glucosed Starch Maltose Starch þ maltose Glucose Starch Maltose Starch þ maltose Glucosee Starch Maltose Starch þ maltose Glucose Starch Maltose Starch þ maltose Glucose

0.47 0.40 0.44 0.42 0.33 0.32 0.32 0.33 12 27 23 33 1.0 0.16 0.85 0.21 0.70 0.80 0.73 0.79 1.0 0.15 0.84 0.22 33 37 29 36 77 88 80 86

2 2.51 1.01 2.12 0.72 46 50 1.0 0.50 0.84 0.71 1.0 0.48 65 46 93 78

3

4a

0.47 0.59 0.52 0.53 0.35 0.52 0.38 0.37 17 27 24 47 1.0 0.24 1.0 0.26 0.74 0.88 0.73 0.70 1.0 0.25 1.0 0.25 34 41 29 32 82 97 80 77

0.07 (0.18)

5b

6c

7c

0.10 (0.28) 0.39

0.24 0.01 (0.07)

0.10 (0.44)

0.38

0.20 (0.65)

0.17 0.15 27 (25)

0.03 (0.26)

0.13 (0.40) 7 (20)

34 44 0.15 (0.38)

24 (49)

39 0.21 (0.60)

14 (31)

0.16 0.12 0.14 (0.39)

0.05 (0.22)

0.19

0.10 (0.32)

0.44 0.63 0.03 (0.21)

0.30 (0.59)

0.65 (0.62)

0.08 0.10 12 (19)

0.02 (0.17)

0.09 (0.27)

30 21 16 (43)

20 (48)

31 (30)

48 69

33 (65)

71 (68)

after a pulse of potato starch. in cycles with a pulse feed of glucose (bulking sludge event). in cycles with Sfeast > 2.5 mg O2 L1. O2 max qs ¼ 2.02 Cmol Cmol1 h1 (Dircks et al., 2001). 1 1 qmax h (Dircks et al., 2001). polyglucose ¼ 1.49 Cmol Cmol max Ys;polyglucose ¼ 0.91 Cmol Cmol1 (Dircks et al., 2001).

use of starch instead of maltose/sugar lead to a more open and porous biofilm/granule surface. Also here it can be reasoned that the polymeric substrates have to be hydrolysed before the monomers can diffuse in the EPS matrix, a more open structure gives enough sites for the polymer to be adsorbed and hydrolysed (de Beer and Stoodley, 1995). This is further discussed in de Kreuk et al. (submitted for publication). A generalization is not yet possible since biofilms growing on proteins versus amino acids showed a different behaviour (Mosquera-Corral et al., 2003), indicating that more research needs to be done. For instance, the localisation and activity of different enzymes (e.g. glucosidases and proteases) need more investigation. Bulking sludge has been reported in some activated sludge systems fed with complex synthetic solutions or the particulate fraction of municipal wastewater (Lakay et al., 1999). Nothing was mentioned in these studies about the place where hydrolysis occurs. The reported occurrence of bulking events with particulate COD could be the result of an important fraction of

particulate substrate being hydrolysed in the bulk liquid, which certainly gives advantages to filamentous bacteria. For instance, in biofilm systems the hydrolytic activity has been found mainly associated with the bulk liquid in unsteady-state periods (Mosquera-Corral et al., 2003), conditions which are also likely to occur in complex dynamic activated sludge systems. These aspects emphasize the importance of localizing the hydrolytic activity, i.e. whether the enzymes are associated with the biological floc (i.e. cell surface or EPS matrix) or released to the bulk liquid, whenever bulking sludge is reported in systems fed with particulate substrates.

4.4.

Diffusion theory as basis for bulking sludge

The occurrence of concentration gradients over the sludge floc due to diffusion limitation were recently hypothesized as being the most important factor for the development of filamentous bacterial structures, and, eventually, bulking sludge, and not

w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 2 6 3 e2 7 3

5.

271

Conclusion

Particulate substrates are incorporated in the sludge floc and subsequently hydrolysed. Despite that this will result in a low concentration of soluble substrate (hydrolysed monomers), and the risk of selection of filamentous organisms, bulking sludge did not occur. It is hypothesized that this is due to the fact that the hydrolysis products are uniformly distributed inside the floc. When substrate is taken up at low concentrations form the bulk solution this would lead to gradients over the floc giving advantage to filamentous organisms which extend outside the floc. Consequently occurrence of substrate gradients, easily developing at low substrate concentrations, seem to be more important then a kinetic selection for the proliferation of filamentous bacteria.

Acknowledgment Fig. 5 e Schematic representation of soluble substrate concentration (Cs) in the floc and in the bulk liquid after applying a pulse feed of monomer (maltose or glucose) or polymer (starch). Phase-contrast photomicrograph of typical biological floc at pseudo-steady state in the pulse fed system with starch (SBR 1).

the substrate concentration as such (Kappeler and Gujer, 1994; Martins et al., 2004a,b; Lou and De Los Reyes, 2008). Bacterial morphology and bacterial physiology are also important factors but the trigger for the development of different types of bacterial structures is the presence of gradient-governed environments, typical of substrate limited activated sludge systems. According to this theory in the presence of substrate diffusion limitation inside the biological flocs, filamentous bacterial structures and, thus, filamentous bacteria, have a higher outgrowth velocity because they grow preferentially in one direction, and not in three directions as floc forming bacterial structures (Martins et al., 2004b). If substrate diffusion limitation does not exist, more regularly shaped and compact bacterial structures are expected. The experimental observations in this study are in line with this theory. In pseudo-steady state systems starch hydrolysis takes place inside the flocs, not giving rise to strong gradients of substrate concentration over the floc radius (Fig. 5). The concentration of monomers at floc level is however low and rate limiting, a condition which according the kinetic selection theory would lead to bulking sludge (Chudoba et al., 1973). Filamentous bacteria were detected in the starch fed systems but they remained inside compact micro-colonies and did not show the typical floc morphology behaviour of bulking sludge, similar to observation by Liao et al. (2004). Note that for bulking sludge to occur only a small fraction of filamentous bacteria is sufficient to raise the SVI from 100 to 200 mL/g. Therefore, it is likely that the soluble substrate concentration, although potentially selective for filamentous bacteria, as such is not the most important factor for bulking sludge. The micro-gradients of substrate concentration as hypothesized by the diffusion based theory are a more likely cause of sludge bulking.

The authors gratefully acknowledged the assistance rendered by the analytical and technical staff of the Department of Biochemical Engineering. The technical support given by the Erasmus student Stefania Ortu from University of Cagliari in this study is highly appreciated. Anto´nio Martins received financial support from the Portuguese State in the context of PRAXIS XXI by the Doctoral Scholarship BD/19538/99.

Nomenclature COD chemical oxygen demand, mg L1 FISH fluorescent in situ hybridisation feast fraction of poly-glucose storage compounds fpolyglucose produced in the feast phase, Cmol Cmol1 PHB poly-b-hydroxybutyrate specific oxygen consumption rate in the feast phase, qfeast O O2 mmol Cmol1 h1 qfeast polyglucose specific production rate of poly-glucose storage compounds in the feast phase, Cmol Cmol1 h1 qmax polyglucose maximum specific production rate of poly-glucose storage compounds, Cmol Cmol1 h1 feast qs specific substrate uptake rate in the feast phase, Cmol Cmol1 h1 max qs maximum specific substrate uptake rate, Cmol Cmol1 h1 feast SO2 bulk liquid dissolved oxygen concentration in the feast phase, mg O2 L1 SBR sequencing batch reactor SRT solids retention time, day Sugarfeast polyglucose fraction of sugars consumed and used for synthesis of poly-glucose storage compounds, % SVI sludge volume index, mL g1

references

Amann, R.I., Binder, B.J., Olson, R.J., Chisholm, S.W., Devereux, R., Stahl, D.A., 1990. Combination of 16S rRNA-targeted oligonucleotide probes with flow cytometry for analyzing

272

w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 2 6 3 e2 7 3

mixed microbial populations. Applied Environmental Microbiology 56 (6), 1919e1925. de Beer, D., Stoodley, P., 1995. Relation between the structure of an aerobic biofilm and transport phenomena. Water Science and Technology 32 (8), 11e18. Casey, T.G., Wentzel, M.C., Ekama, G.A., 1999. Filamentous organism bulking in nutrient removal activated sludge systems. Paper 11: a biochemical/microbiological model for proliferation of anoxic-aerobic (AA) filamentous organisms. Water SA 25 (4), 443e451. Chudoba, J., Grau, P., Ottova´, V., 1973. Control of activated-sludge filamentous bulking-II. Selection of microorganisms by means of a selector. Water Research 7 (10), 1389e1398. Crocetti, G.R., Hugenholtz, P., Bond, P.L., Schuler, A., Keller, J., Jenkins, D., Blackall, L.L., 2000. Identification of polyphosphate-accumulating organisms and design of 16S rRNA-directed probes for their detection and quantitation. Applied Environmental Microbiology 66 (3), 1175e1182. Crocetti, G.R., Banfield, J.F., Keller, J., Bond, P.L., Blackall, L.L., 2002. Glycogen-accumulating organisms in laboratory-scale and full-scale wastewater treatment processes. Microbiology 148 (11), 3353e3364. Daims, H., Bruhl, A., Amann, R., Schleifer, K.-H., 1999. The domain-specific probe EUB338 is insufficient for the detection of all bacteria: development and evaluation of a more comprehensive probe set. Systematic and Applied Microbiology 22, 434e444. Dircks, K., Beun, J.J., Van Loosdrecht, M.C.M., Heijnen, J.J., Henze, M., 2001. Glycogen metabolism in aerobic mixed cultures. Biotechnology Bioengineering 73, 85e94. Dold, P.L., Fleit, E., Han, J., 1991. Hydrolysis of a (1e4) glucan bonds in activated sludge mixed bacterial communities. Environmental Technology 12 (10), 871e879. Eikelboom, D.H., 2000. Process Control of Activated Sludge Plants by Microscopic Investigation. IWA Publishing, London, UK. Ekama, G.A., Marais, G.V.R., 1986. The implications of the IAWPRC hydrolysis hypothesis on low F/M bulking. Water Science and Technology 18 (6), 11e19. Erhart, R., Bradford, D., Seviour, R.J., Amann, R., Blackall, L.L., 1997. Development and use of fluorescent in situ hybridization probes for the detection and identification of ‘Microthrix parvicella’ in activated sludge. Systematic and Applied Microbiology 20, 310e318. Frølund, B., Griebe, T., Nielsen, P.H., 1995. Enzymatic activity in the activated-sludge floc matrix. Applied Microbiology Biotechnology 43, 755e761. Goel, R., Mino, T., Satoh, H., Matsuo, T., 1998. Intracellular storage compounds, oxygen uptake rates and biomass yield with readily and slowly degradable substrates. Water Science and Technology 38 (8e9), 85e93. Goel, R., Mino, T., Satoh, H., Matsuo, T., 1999. Modeling hydrolysis processes considering intracellular storage. Water Science and Technology 39 (1), 97e105. Gujer, W., Henze, M., Mino, T., Van Loosdrecht, M., 1999. Activated sludge model no. 3. Water Science and Technology 39 (1), 183e193. Jenkins, D., Richard, M.G., Daigger, G.T., 1993. Manual on the Causes and Control of Activated Sludge Bulking and Foaming, second ed. Lewis, Boca Raton, Fla. Kappeler, J., Gujer, W., 1994. Verification and applications of a mathematical model for aerobic bulking. Water Research 28 (2), 303e310. Karahan, O., Martins, A., Orhon, D., van Loosdrecht, M.C.M., 2006. Experimental evaluation of starch utilization mechanism by activated sludge. Biotechnology and Bioengineering 93 (5), 964e970. Kong, Y.H., Beer, M., Seviour, R.J., Lindrea, K.C., Rees, G.N., 2001. Structure and functional analysis of the microbial community

in an aerobic: anaerobic sequencing batch reactor (SBR) with no phosphorus removal. Systematic and Applied Microbiology 24, 597e609. de Kreuk, M.K., Kishida, N., Tsuneda, S., van Loosdrecht, M.C.M. Behavior of polymeric substrates in an aerobic granular sludge system. Water Research, submitted for publication. Kruit, J., Boley, F., Jacobs, L.J.A.M., Wouda, T.W.M., 1994. Prediction of the O2 conditions in the selector. Water Science and Technology 29 (7), 229e237. Lakay, M.T., Hulsman, A., Ketley, D., Warburton, C., De Villiers, M., Casey, T.G., Wentzel, M.C., Ekama, G.A., 1999. Filamentous organism bulking in nutrient removal activated sludge systems. Paper 7: exploratory experimental investigations. Water SA 25 (4), 383e396. 1999. Liao, J., Lou, I., De Los Reyes III, F.L., 2004. Relationship of speciesspecific filament levels to filamentous bulking in activated sludge. Applied and Environmental Microbiology 70 (4), 2420e2428. Liu, W.-T., Nielsen, A.T., Wu, J.-H., Tsai, C.-S., Matsuo, Y., Molin, S., 2001. In situ identification of polyphosphate- and polyhydroxyalkanoate-accumulating traits for microbial populations in a biological phosphorus removal process. Environmental Microbiology 3 (2), 110e122. Liu, J.R., Seviour, R., 2001. Design and application of oligonucleotide probes for fluorescent in situ identification of the filamentous bacterial morphotype Nostocoida limicola in activated sludge. Journal of Environmental Microbiology 3 (9), 551e560. van Loosdrecht, M.C.M., Pot, M.A., Heijnen, J.J., 1997. Importance of bacterial storage polymers in bioprocesses. Water Science and Technology 35 (1), 41e47. Lou, I.C., De Los Reyes III, F.L., 2008. Clarifying the roles of kinetics and diffusion in activated sludge filamentous bulking. Biotechnology and Bioengineering 101 (2), 327e336. Manz, W., Amann, R., Ludwig, W., Wagner, M., Schleifer, K.-H., 1992. Phylogenetic oligodeoxynucleotide probes for the major subclasses of proteobacteria: problems and solutions. Systematic and Applied Microbiology 15, 593e600. Manz, W., Amann, R., Ludwig, W., Vancanneyt, M., Schleifer, K., 1996. Application of a suite of 16S rRNA-specific oligonucleotide probes designed to investigate bacteria of the phylum cytophaga-flavobacter-bacteroides in the natural environment. Microbiology 142 (5), 1097e1106. Martins, A.M.P., Heijnen, J.J., van Loosdrecht, M.C.M., 2003a. Effect of feeding pattern and storage on the sludge settleability under aerobic conditions. Water Research 37 (11), 2555e2570. Martins, A.M.P., Heijnen, J.J., Van Loosdrecht, M.C.M., 2003b. Effect of dissolved oxygen concentration on sludge settleability. Applied Microbiology Biotechnology. Martins, A.M.P., Pagilla, K., Van Loosdrecht, M.C.M., 2004a. Filamentous bulking sludge e a critical review. Water Research 38 (4), 793e817. Martins, A.M.P., Picioreanu, C., Heijnen, J.J., van Loosdrecht, M.C.M., 2004b. Three-dimensional dual-morphotype species modeling of activated sludge flocs. Environmental Science and Technology 38 (21), 5632e5641. Martins, A.M.P., Heijnen, J.J., Van Loosdrecht, M.C.M., 2004c. Bulking sludge in biological nutrient removal systems. Biotechnology and Bioengineering 86 (2), 125e135. Maszenan, A.-M., Seviour, R.J., Patel, B.K.C., Wanner, J., 2000. A fluorescently-labelled r-RNA targeted oligonucleotide probe for the in situ detection of G-bacteria of the genus Amaricoccus in activated sludge. Journal Applied Microbiology 88 (5), 826e835. Meier, H., Amann, R., Ludwig, W., Schleifer, K.-H., 1999. Specific oligonucleotide probes for in situ detection of a major group of gram-positive bacteria with low DNA GþC content. Systematic and Applied Microbiology 22, 186e196.

w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 2 6 3 e2 7 3

Mino, T., San Pedro, D.C., Matsuo, T., 1995. Estimation of the rate of slowly biodegradable COD (SBCOD) hydrolysis under anaerobic, anoxic and aerobic conditions by experiments using starch as model substrate. Water Science and Technology 31 (2), 95e103. Mosquera-Corral, A., Montras, A., Heijnen, J.J., van Loosdrecht, M.C.M., 2003. Degradation of polymers in a biofilm airlift suspension reactor. Water Research 37 (3), 485e492. Roller, C., Wagner, M., Amann, R., Ludwig, W., Schleifer, K.-H., 1994. In situ probing of gram-positive bacteria with high DNA G þ C content using 23S rRNA-targeted oligonucleotides. Microbiology 140, 2849e2858.

273

San Pedro, D.C., Mino, T., Matsuo, T., 1994. Evaluation of the rate of hydrolysis of slowly biodegradable COD (SBCOD) using starch as substrate under anaerobic, anoxic and aerobic conditions. Water Science and Technology 30 (11), 191e199. Wagner, M., Amann, R., Kampfer, P., Assmus, B., Hartmann, A., Hutzler, P., Springer, N., Schleifer, K.-H., 1994. Identification and in-situ detection of gram-negative filamentous bacteria in activated sludge. Systematic and Applied Microbiology 17, 405e417. Wile´n, B.-M., Jin, B., Lant, P., 2003. Impacts of structural characteristics on activated sludge floc stability. Water Research 37 (15), 3632e3645.