ARTICLE IN PRESS

WAT E R R E S E A R C H

42 (2008) 2166 2176

Available at www.sciencedirect.com

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

Demonstration of nitrogen removal via nitrite in a sequencing batch reactor treating domestic wastewater
Richard Blackburne, Zhiguo Yuan, Jurg Keller
Advanced Water Management Centre, The University of Queensland, Brisbane 4072, Australia

ar t ic l e i n f o
Article history: Received 8 July 2007 Received in revised form 22 November 2007 Accepted 25 November 2007 Available online 3 December 2007 Keywords: Nitritation Wastewater treatment Nitrication Two-step nitrication model Aerobic duration control Nitrite pathway

abs tra ct
Nitrogen removal via nitrite, as opposed to the traditional nitrate, may be benecial for carbon-limited biological wastewater treatment plants. However, reliable termination of nitrication at nitrite (nitritation) has proved difcult in the treatment of domestic wastewater. In this study, nitritation was attained in a sequencing batch reactor (SBR) with pre-denitrication treating domestic wastewater (total Kjeldahl nitrogen (TKN) concentration of about 43 mgN L1) by aerobic duration control. The aerobic duration control strategy terminates aeration upon completion of ammonium oxidation with accumulated nitrite still remaining. The SBR was purposefully operated such that the inuence of other known selection factors for nitritation was absent. The process proved effective in achieving a steady state whereby over 80% nitritation was sustained. Investigation of the cause of nitritation by a calibrated ammonium and nitrite oxidation model showed aerobic duration control as the key factor leading to nitritation. & 2007 Elsevier Ltd. All rights reserved.

1.

Introduction

The traditional biological nitrogen removal processes utilised in wastewater treatment plants (WWTPs) involve the oxidation of ammonium (NH+) to nitrate (NO) (nitrication) and 4 3 then reduction with an organic carbon source (chemical oxygen demand (COD)) to nitrogen gas (N2) (denitrication). Both nitrication and denitrication possess nitrite (NO) as 2 an intermediate. Hence, if nitrication is stopped at nitrite (nitritation), then complete denitritation from nitrite to nitrogen gas can be achieved (although there may be situations where denitritation is not always complete and hence some N2O may be produced). Nitrogen removal via nitrite may yield up to a 25% reduction in aeration and 40% reduction in COD requirements. The reduction in COD requirements would be particularly advantageous for WWTPs treating wastewater decient in COD for the required nitrogen removal via nitrate. However, the difculty to utilise nitrogen removal via nitrite lies in achieving specic inhibition or
Corresponding author. Fax: +61 7 3365 4726.

removal of the nitrite oxidising bacteria (NOB; those that oxidise nitrite to nitrate) while retaining ammonia oxidising bacteria (AOB; those that oxidise ammonia to nitrite), thereby attaining nitritation. In recent years, reports of stable nitritation processes from activated sludge have appeared frequently in the literature and some suggested factors to achieve nitritation in these processes include: 1. Wash out of NOBs Wash out of NOBs have been reported in a continuous conguration (such as a chemostat) operated at a sludge retention time (SRT) which is less than the critical SRT of NOBs but greater than the critical SRT of AOBs. This situation seems to be best achieved at elevated temperatures (3040 1C) (SHARON process; Hellinga et al. (1998)) or low dissolved oxygen (DO) concentrations (Blackburne et al., 2007a; Sliekers et al., 2005) where the maximum specic growth rate of AOBs is higher than that of NOBs.

E-mail address: j.keller@awmc.uq.edu.au (J. Keller). 0043-1354/$ - see front matter & 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2007.11.029

ARTICLE IN PRESS
WAT E R R E S E A R C H

42 (2008) 2166 2176

2167

Nomenclature

COD chemical oxygen demand (mgO2 L1) TKN total Kjeldahl nitrogen (mgN L1) XAOB, XNOB biomass concentration of AOB and NOB,
respectively (gVSS L1) mAOB, mNOB growth rate of AOB and NOB, respectively (d1) bAOB, bNOB decay rate of AOB and NOB, respectively (d1)

SNO3 nitrate-nitrogen concentration (mgN L1) LNH4 ; LNO2 load of ammonium and nitrite nitrogen,

rAOB, rNOB nitrogen oxidation rate of AOB and NOB, SRT V SNH4 SNO2
respectively (gN gVSS1 d1) sludge retention time (d) volume (L) ammonium-nitrogen concentration (mgN L ) nitrite-nitrogen concentration (mgN L1)
1

respectively (gN d1) YAOB, YNOB biomass yield of AOB and NOB, respectively (gVSS gN1) Ks;NH4 ; Ks;NO2 substrate half saturation constants with respect to ammonium and nitrite, respectively (mgN L1) Ko,AOB, Ko, NOB oxygen half saturation constants of AOB and NOB, respectively (mgO2 L1) So oxygen concentration (mgO2 L1)

MLSS mixed liquor suspended solids concentration VSS

(mg L1) volatile suspended solids concentration (mg L1)

2. Batch processes with nitrite accumulation Sustained nitritation has also been reported in (sequencing) batch-type reactor operation (Yoo et al., 1999; Fux et al., 2003; Lai et al., 2004; Peng et al., 2004). The termination of aeration prior to or at the completion of ammonium oxidation in a process with nitrite accumulation has been suggested as a key factor leading to sustained nitritation (Yoo et al., 1999; Fux et al., 2003; Peng et al., 2004). 3. Low dissolved oxygen (DO) concentration AOBs have been suggested to outcompete NOBs under low DO concentrations based on a reportedly higher oxygen afnity of AOBs compared with NOBs, thereby resulting in nitritation (Hanaki et al., 1990; Laanbroek and Gerards, 1993). Also at low DO concentration, some processes allow for simultaneous nitrogen oxidation and reduction likely via nitrite (e.g. simultaneous nitrication and denitrication; Pochana and Keller, (1999) or OLAND; Kuai and Verstraete, (1998) or completely autotrophic nitrogen removal over nitrite; Third et al. (2001) or nitrier denitrication; Kampschreur et al.,( 2006)). In these situations, nitrite reducing bacteria (e.g. denitriers, anaerobic AOB) may out-compete the NOBs for nitrite thereby disadvantaging the NOBs by limiting the availability of their substrate (Hanaki et al., 1990; Kuai and Verstraete, 1998; Pochana and Keller, 1999; Third et al., 2001; Wyffels et al., 2004). Sequencing batch reactors (SBRs) have become quite common for nitrogen removal wastewater treatment. Aerobic duration control, whereby aeration is terminated as soon as ammonium oxidation is complete, is particularly simple to implement in SBRs (e.g. the reports listed in number 2 above). For high nitrogen concentration wastewaters, nitritation has been achieved with the use of aerobic duration control in SBRs (Fux et al., 2003; Lai et al., 2004). However, for low strength, domestic-type wastewater nitritation has only been claimed by Peng et al. (2004) with the use of aerobic duration control. However, Peng et al. (2004) used a relatively high nitrogen concentration (93 mgNH4-N L1) in their domestic wastewater and a high volumetric exchange ratio (VER 0.65) so that reactor nitrogen concentrations were high compared

with typical domestic SBR processes and more akin to high nitrogen concentration SBR processes. Aerobic duration control as the major selection factor for nitritation in the above quoted studies cannot be directly concluded. All studies have additional NOB inhibitory compounds such as free ammonia or free nitrous acid (see Anthonisen et al. (1976) for description of NOB inhibitory compounds). Some processes also make use of high temperature which increases the maximum growth rate of AOBs relative to NOBs (Hellinga et al., 1998; Fux et al., 2003; Lai et al., 2004; Peng et al., 2004), and effectively amplify nitrite accumulation during the batch as the AOBs oxidise substrate faster than the NOBs. The contribution of the above factors towards sustaining nitritation in the quoted studies is difcult to quantify but are certainly favourable factors. Also, all processes included simultaneous or post-denitrication with externally added COD (Yoo et al., 1999; Fux et al., 2003; Lai et al., 2004; Peng et al., 2004), which may further disadvantage the NOBs, as noted previously. The aim of this paper is to demonstrate the effectiveness of aerobic duration control in attaining sustained nitritation using a laboratory-scale SBR process mimicking a domestic wastewater full-scale SBR. As such, conditions summarised above that favour AOBs were eliminated as far as possible to allow assessing the contribution of aerobic duration control alone to achieve nitritation.

2.
2.1.

Materials and methods

Sequencing batch reactor conguration

The SBR was operated with aerated and non-aerated periods to allow for nitrication and denitrication. Each cycle consisted of the following: 30 min settling, 10 min decant, 80 min anoxic feed, 105 min aerobic react and 15 min non-aerated waste. When aerobic duration control was implemented, the aerobic react period decreased due to the effect of terminating aeration towards the end of this period. With a working volume of 10 L, 2 L of wastewater was fed each cycle, which results in a VER of 0.2. Domestic wastewater (characteristics given below) was

ARTICLE IN PRESS
2168
WAT E R R E S E A R C H

42 (2008) 2166 2176

supplied by a peristaltic pump (Masterex LS, model 7553, USA) during the entire anoxic period, resulting in a ll time ratio (FTR) of 0.33. Mixing was provided by an overhead stirrer (IKA RW20.n, IKA Works Asia, Malaysia). DO concentration was measured on-line (YSI model 5739, Yellow Springs, Ohio, USA) and controlled via a Programmable Logic Controller (PLC) to remain between 1.8 and 2.2 mg/L during the aeration period using ON/OFF control. pH was measured on-line (Ionode IJ44, TPS, Australia) and recorded but not controlled and remained around neutral values (pH 7.170.3). The SRT was typically maintained at 8 days; however, poor sludge settling events (noted in Section 3) caused additional biomass loss and the wastage ow was adjusted temporarily during and after these periods to recover nitrication, resulting in higher SRTs. With an SRT of 8 days, the corresponding mixed liquor concentration was within the range of 12001700 mg L 1. The oxygen uptake rate (OUR) was measured during the aerobic period of the SBR by determining the slope of the decrease in DO concentration (during the aeration OFF period of the DO control). The OUR-based aerobic duration control algorithm was programmed into the PLC. When the OUR value dropped below a threshold OUR value, the aeration was terminated and the DO concentration decreased to zero. To optimise the aerobic duration control, the threshold OUR value was calibrated manually every 23 days to correspond with the completion of ammonium oxidation. The ammonium concentration at the end of the cycle was determined by measuring the efuent using a Merck spectrophotometer (Model SQ118, Darmstadt, Germany) and the associated Merck Ammonium Kit. Calibration of the threshold OUR value was achieved by aiming to maintain the efuent ammonium concentration at about 1 mgN L1, which can be considered as a good efuent quality. The threshold OUR was altered by 12 mgO2 L1 h1 every 23 days based on the comparison of the actual measured efuent ammonium concentration and the above set-point. The threshold OUR varied between 15 and 25 mgO2 L1 h1. The reactor operation consisted of several phases that are briey summarised below: Start-up (days 175) Phase A (days 75173) The SBR was operated to steady state and the nal cycle conguration was determined. The SBR was operated to achieve maximal nitrogen removal via nitrate. The aerobic duration control was implemented in the middle of this period (day 117). Formic acid was added as a pulse addition in the last 10 minutes of the anoxic period of the SBR cycle. The aerobic duration control operated during this period. The threshold OUR value at which aeration was terminated was optimised manually every 23 days. The optimised threshold OUR value was implemented and the reactor was operated to achieve steady state nitrogen removal via nitrite.

Further details of each operation phase are discussed in Section 3.

2.2.

Wastewater characteristics

The wastewater used in this study was collected weekly from a WWTP in Brisbane, Australia, which receives predominantly domestic wastewater from Brisbane City. This WWTP has on-site primary sedimentation and the wastewater was collected after this initial step to minimise the inuent solids addition to the laboratory SBR and provide a low COD to nitrogen ratio wastewater. The wastewater was stored in a cold room (temperature controlled to about 4 1C) over the duration of the week, which was effective in preserving the COD and nitrogen characteristics. The total Kjeldahl nitrogen (TKN) and COD concentrations and COD/TKN ratio measured over the course of the reactor operation are shown in Fig. 1. The wastewater temperatures were typical of a warmtemperate region (1825 1C) and typical of domestic wastewater with an average nitrogen concentration of approximately 43 mgTKN L1. The average COD/TKN ratio was 8.6 mgCOD mgN1, although some periods had considerably lower COD/TKN ratios over several weeks, as shown in Fig. 1. Given the empirical ratio for good denitrication of about 8.6 mgCOD mgN1 (extracted from the activated sludge model number 1; Henze et al. (2000)), operation at COD/TKN ratios below the average may be considered as COD limited if removing nitrogen via nitrate.

2.3.

Analytical methods

Mixed liquor suspended solids concentration/volatile suspended solids concentration (VSS) were determined using Whatman GF/A lters and in accordance with standard methods (APHA, 1992). COD was determined using a Merck spectrophotometer (Model SQ118, Darmstadt, Germany) and Merck COD Kit. TKN was determined according to standard methods (APHA, 1992) and ammonium, nitrite and nitrate were measured with a ow injection analyser (FIA) (Lachat QuikChem 8000, Milwaukee, Wisconsin, USA).

2.4.

Two-step nitrication model

Phase B (days 173194)

Phase C (days 194246) Phase D (days 246341)

A two-step nitrication model describing the aeration period of the SBR operation, with and without aerobic duration control, was coded into MATLAB V 6.5 (Mathworks, Natick, Massachusetts, USA) to evaluate the nitritation experiments. Only the aerobic period was assessed since denitritation proceeded effectively from nitrite (see Section 3). Nitrication was split into the two sub-processes, ammonium oxidation and nitrite oxidation, which is different from other models (e.g. ASM, (Henze et al., 2000)) where nitrication is considered as a singlestep process. The model equations used are shown in Table 1.

The key assumptions in the model included:

1. Denitrication was not COD-limited and was complete at the end of the anoxic period (i.e. SNO2 SNO3 0). This was

ARTICLE IN PRESS
WAT E R R E S E A R C H

42 (2008) 2166 2176

2169

600
COD TKN COD/TKN

30 27.5 25 22.5 20 17.5 15 Average COD/TKN 12.5 10 7.5 5 2.5 0 400 COD/ TKN

500 TKN; COD (mg.L-1) 400 300 200 100 0 0 100 200 Duration (days) 300

Fig. 1 COD, TKN and COD/TKN wastewater characteristics. Table 1 Model equations and descriptions for the aerobic period only of the SBR operation Equation description
Change in AOB biomass with time Change in NOB biomass with time Change in NH4 concentration with time Change in NO2 concentration with time Change in NO3 concentration with time Monod-based growth kinetics for AOBs Monod-based growth kinetics for NOBs

Table 2 Default model parameters Model parameter

Ks,AOB Ks,NOB Ko,AOBa Ko,NOBa bAOB; bNOB SRT V YAOB YNOB mMAX b AOB mMAX b NOB
a

Value

Units

Reference

Equation
dXAOB XAOB mAOB XAOB bAOB XAOB dt SRT dXNOB XNOB mNOB XNOB bNOB XNOB dt SRT dSNH4 m XAOB AOB dt Y AOB dSNO2 mAOB XAOB mNOB XNOB dt Y AOB Y NOB dSNO3 mNOB XNOB dt Y NOB mAOB mMAX AOB SNH4 SO KS;NH4 SNH4 KO;AOB SO SNO2 SO KS;NO2 SNO2 KO;NOB SO

0.75 0.15 0.03 0.4 0.05 8 10 0.17 0.1 0.65 0.65

mgN L1 mgN L1 mgO2 L1 mgO2 L1 days1 Days L gVSS gN1 gVSS gN1 days1 days1

Alleman (1984) Manser (2005) Blackburne et al. (2007a) Blackburne et al. (2007a) Henze et al. (2000) Blackburne et al. (2007b) Blackburne et al. (2007b) Calibrated Calibrated

mNOB mMAX NOB

These values were determined for enriched AOB and NOB cultures with small ocs which had negligible oxygen mass transfer resistance (i.e. assumed to be close to the inherent biological values). The values were not adjusted in this model to account for likely mass transfer resistance than due to larger ocs. The inuence of this will be identical for both AOB and NOB real Ko values as mass transfer resistance can be assumed identical for both. b The model was calibrated against the data from the cyclic studies performed during phases B (full nitrication) and D (80% nitritation) (data not shown), hence these values were determined accordingly.

consistent with experimental data from the SBR operation when nitrogen removal was via nitrite (data not shown). 2. The assimilation of ammonium for cellular growth was not included in the model. This assumption was necessary since the growth of heterotrophic bacteria was not considered in the model. And the inuence of cellular assimilation of ammonium would be the same for both nitrifying bacterial groups, hence not favouring either group. 3. Decay or lysis of bacteria only occurs during the aerobic period. This is supported by the literature (eg. Nowak et al., 1994; Siegrist et al., 1999; Lee and Oleszkiewicz, 2003)

reporting much higher decay rates under aerobic conditions compared with anoxic conditions. Also, since both AOB and NOB bA values are assumed identical, there will be no favouring of either group predicted by this model. 4. Aeration is switched off when the ammonium concentration reaches 0.5 mgN L1, which functions as an approximation for the aerobic duration control. Although the same accuracy for the ammonium threshold could not be achieved during the experimental work, this should be considered as an optimal detection point for the completion of nitrication, and is expected to be achievable with

ARTICLE IN PRESS
2170
WAT E R R E S E A R C H

42 (2008) 2166 2176

reliable and accurate on-line ammonium sensors available for full-scale applications. 5. Ammonication is not included in the model rather only inorganic nitrogen is considered. The wastewater used had a high ammonium to TKN percentage of $83% and ammonication will occur during the unaerated (and ll) period of the cycle. The default model parameters are shown in Table 2. Floccular mass transfer dynamics have been shown to affect the half saturation kinetic constants, Ko and Ks (Mueller et al., 1966; Bakti and Dick, 1992; Beccari et al., 1992; Gapes, 2003; Blackburne et al., 2007a). Ammonium needs to diffuse from bulk liquid into ocs before being used by AOBs. In contrast, nitrite is produced within ocs and therefore has a much less diffusion distance before reaching NOBs. Consequently, in the model a higher, mass transfer affected Ks value for AOBs was used compared with the Ks value for NOBs, which was chosen as a value corresponding to a situation where mass transfer effects are largely removed (i.e. the likely Ks value in the oc).

(around 56 mgCOD mgN1), conrming the difcult nature of this wastewater for full nitrogen removal via nitrate. The OUR measurement and an initial OUR threshold value was implemented on day 117 for the aerobic duration control but was ineffective in achieving nitrite accumulation. This is attributed to a potential requirement of pre-existing nitrite accumulation during the cycle, which was not observed in this case. Initial nitrite accumulation during a batch is common to all other publications that use aerobic duration control to achieve nitritation (Yoo et al., 1999; Fux et al., 2003; Lai et al., 2004; Peng et al., 2004). Conceptually, nitrite accumulation ensures that some nitrite is remaining after the termination of the aerobic period, which results in a lower amount of substrate being used by the NOBs compared with the amount of substrate (ammonium) that is oxidised by the AOBs. This suggests that nitrite accumulation is important to start an effective aerobic duration control.

3.2.

Nitrogen removal via nitrite

3.
3.1.

Results and discussion

Nitrogen removal via nitrate

The SBR achieved an inconsistent degree of nitrogen removal via nitrate prior to the aerobic duration control implementation as shown by the percent nitrogen removal in Fig. 2. The start up period for the SBR was up to day 75, after which the SBR operating conguration was xed. The SBR achieved over 95% ammonium oxidation for the most part of phase A. However, denitrication was much less reliable than nitrication, which caused variation in the percent nitrogen removal (Fig. 2). The maximal percent nitrogen removed was 90%. Periods with less percent removal generally corresponded to wastewater batches with a low COD/TKN ratio

To initiate some nitrite accumulation, a temporary addition of a selective inhibitor for NOBs was considered. Selective NOB inhibitory compounds assessed included free ammonia, free hydroxylamine and formic acid. The threshold free ammonia concentration for NOB inhibition has been reported in the literature to be between 0.02 and 0.7 mgN L1 (as summarised in Philips et al. (2002)). Given that the initial ammonium concentration at the start of the aerobic period was measured to be less than 10 mgN L1, the pH would need to be raised to a value of about 8.3 to achieve 0.7 mgN L1 of free ammonia. Such an elevated pH may result in negative impacts on nitrication performance and, as the ammonium is oxidised, the corresponding free ammonia concentration quickly decreases below the threshold inhibitory value. Adding more ammonium, which would also increase the free ammonia concentration was considered to be not a feasible option as, in reality, remaining unoxidised ammonium would leave in

100
% nitrogen removed COD/TKN

15 14 13 12 11 10 9 8 7 6 60 80 100 120 140 Duration (days) 160 5 180 COD/TKN

90 80 % nitrogen removed 70 60 50 40 30 20 10 0

Fig. 2 Percent nitrogen removal from the SBR operated with nitrication and denitrication via nitrate (phase A) and the corresponding COD/TKN ratio.

ARTICLE IN PRESS
WAT E R R E S E A R C H

42 (2008) 2166 2176

2171

the efuent and violate a WWTP ammonium discharge licence. Therefore, free ammonia inhibition was concluded as not suitable for this application. Free hydroxylamine is an intra-cellular intermediate of AOBs and is thought to accumulate extra-cellularly from AOB cells, due to low DO concentrations (Yang and Alleman, 1992). Extra-cellular free hydroxylamine is very effective in bringing about nitrite accumulation with about 15% nitrite accumulation observed at a free hydroxylamine concentration of 0.2 mgNH2OH-N L1 (Blackburne et al. (2004)). However, hydroxylamine seems to react with nitrite, which results in a lack of control over the actual degree of inhibition (Stuven et al., 1992; Blackburne et al., 2004). Therefore, free hydroxylamine inhibition is also considered not suitable in this case. Eilersen et al. (1994) showed that formic acid at a concentration of around 100 mgHCOOH L1 could completely inhibit NOBs while showing no adverse effect on the AOB activity. Formic acid is already used at some WWTPs for air diffuser cleaning and therefore dosing of it should not necessarily require further infrastructure. However, formic acid will be oxidised in the aerobic period by heterotrophs, reducing the concentration with time. However, it seems to be effective in inhibiting NOBs even when added in the anoxic period/zone, thereby reducing the immediate oxidation impact. Formic acid is of comparable cost to some organic carbon sources currently used to enhance denitrication. For the benets of control, cost and availability, dosing of formic acid was considered the most favourable. To achieve some nitrite accumulation without completely inhibiting NOBs, formic acid was added to a reactor concentration of 40 mgHCOOH L1, which was expected to attain about 40% inhibition of nitrite oxidation, considering Eilersen et al.s (1994) data. As illustrated in Fig. 3, phase B, nitrite accumulation was attained during the formic acid addition period (20 days). Although the effect seems somewhat variable, formic acid is thought to be the primary factor in

establishing the observed nitrite accumulation. The peaks in ammonium concentrations during phase B were due to tuning of the threshold OUR value, which proved difcult because of the interference of the oxidation of formic acid on the OUR measurement. Furthermore, the efuent ammonium concentration was found to be very sensitive to threshold OUR value, which made the optimisation of the latter very difcult. During phase C, nitrite accumulation through aerobic duration control alone could be achieved at about 40% nitrite with a NOx concentration of around 5 mg L1. However, the tuning of the threshold OUR remained experimentally difcult partly due to the changing wastewater concentrations. The initial variations in efuent ammonium and the restricted nitrite fraction of about 40% indicate that the OUR threshold value used for the aerobic duration control was still not optimal during this phase. Eventually, during phase D (Fig. 4), an optimum threshold OUR value was achieved and the nitrite fraction of total oxidised nitrogen increased to a level of 80%, which was maintained for about 60 days. During this period a relatively steady process performance was obtained, with near-complete ammonium oxidation (except in the last 20 days due to settling problems and solids washout). This demonstrates that nitritation and denitritation was achievable as the main nitrogen removal process. It seemed difcult to attain more than 80% nitritation probably due to the slow removal of the remaining NOBs. The percent nitrogen removal was signicantly more stable than nitrogen removal via nitrate (as in Fig. 2) with the minimum percent nitrogen removal being 77% and the maximum 94%. During this phase (excluding the poor settling event at the end of phase D), an average of 42 mgN L1 TKN was measured in the inuent while an average of 6.4 mgN L1 TN remained in the efuent, which results in an average nitrogen removal of 84%. The observed stability in nitrogen removal performance was achieved despite the weekly batches of wastewater frequently having very low

25 Nitrogen concentration (mgN.L-1)

100

20

80 Percent (NO2/NOX)

15

60

10

40

20

0 170

190

210 Duration (days)

Effluent NOx-N

230

0 250

Effluent NH4-N

Effluent NO2-N

%NO2

Fig. 3 Addition of formic acid to inhibit nitrite oxidation (B) and tuning of the threshold OUR value (C).

ARTICLE IN PRESS
2172
WAT E R R E S E A R C H

42 (2008) 2166 2176

25 Nitrogen concentration (mgN.L-1)

100

20

80 Percent (NO2/NOx) OUR (mg.L.h-1)

15

Poor settling event

60

10

40

20

0 240

0 260 280 300 Duration (days)

Effluent NOx-N

320

340

Effluent NH4-N

Effluent NO2-N

%NO2

Fig. 4 Attainment of a nitritation culture and subsequent removal of nitrogen via nitrite. Reasons for poor settling were not studied in this work.

Anoxic feed PO4-P; NH4-N; NO3-N; NO2-N (mg.L-1) 10 Settle and Decant

Aerobic Aeration off at OUR = 16.5 mg.L.h-1 30

24

18

12

0 0 30 60 90 120 150 Duration (mins)

NH4-N NO2-N NO3-N

180

210

0 240

PO4-P

OUR

Fig. 5 Nitrogen and phosphorous concentrations sampled from a cycle study on day 277.

inuent COD/TKN ratio (around 6 mgCOD mgN1 in the period between days 290 and 320). Under these signicantly CODlimited conditions, it would be expected that the percent nitrogen removal would uctuate as seen in phase A (Fig. 2) due to incomplete denitrication in the anoxic period. Hence, nitrogen removal via nitrite could achieve an improved stable operating performance with this wastewater compared with the traditional nitrogen removal via nitrate. Due to the reduced COD requirements from denitritation some COD was remaining occasionally at the end of the anoxic period. This created a short anaerobic period, which led to phosphorus release and subsequent biological phosphorus uptake during the following aerobic period as shown in Fig. 5, a cycle study on day 277. This shows that nitrogen

removal via nitrite can reduce the COD requirement sufciently so that biological phosphorus removal may even be possible despite the relatively low COD/TKN ratio of the wastewater.

3.3.

Model-based assessment of SBR results

A two-step dynamic nitrication model was used to simulate the SBR operation to investigate possible causes of nitritation. Of particular interest was to gain insight into the following: 1. Verify that aerobic duration control alone is the selection factor for nitritation; 2. the time required to reach a high percent nitritation (greater than 80%);

ARTICLE IN PRESS
WAT E R R E S E A R C H

42 (2008) 2166 2176

2173

60 XAOB 50 Concentration (mg.L-1) Turn on aerobic duration control %NO2/NOx

120

100

40

80 %NO2/NOx

30

60

20 Day = 470 10 XNOB 0 0 200 400 800 600 1000 Duration (days) 1200 1400

40

20

0 1600

Fig. 6 Model results showing percent nitritation with and without aerobic duration control. The %NO2/NOX curve is a direct representation of the increase in nitrite fraction of NOx but also the percent removal of NOBs.

3. the methods to implement initial nitrite accumulation and why this is required to achieve nitritation with aerobic duration control; and 4. the effect of AOB and NOB bacterial kinetic parameters on achieving nitritation. Using only the aerobic duration control to select for nitritation, the same general trend in percent nitritation as shown in the experimental data (phases C and D) was obtained from the model (Fig. 6). Achieving a high percentage of nitritation (above 80%) (Fig. 6) takes far more time than the usually assumed 3 sludge ages to achieve steady-state operation. For example, the time taken to achieve 80% nitritation from full nitrication would be approximately 300 days, if no initial inhibitor of NOBs is used. This is indeed a reection of the slow NOB removal rate observed in the experimental results. Possible explanations for this slow change in performance are

 since both NOBs and AOBs attain growth opportunities

during the aerated period of the cycle, the net loss of NOBs per cycle is small; removal of bacteria through sludge wastage is inherently a slow process due to long turn-over time of sludge, which is 8 days in this case.

acid addition. This difference may be attributed to nitrite mass transfer dynamics. The model determines the nitrite formed within the biological ocs at the actual cellular scale. However, the concentration of nitrite measured experimentally is the bulk liquid concentration. This measured value is affected by the rate of diffusion through the oc and the rate of nitrite oxidation which is directly proportional to the number of NOBs in the ocs. Therefore, when the number of NOBs is large the rate of nitrite oxidation inside the oc can be faster than the rate of nitrite diffusion out of the oc, and hence nitrite accumulation is not observed in the bulk liquid. Hence, with selective inhibition of the NOBs, the rate of nitrite oxidation becomes comparatively less signicant than diffusion and the model is more representative of measured data. This situation is maintained when the NOB population is already diminished and a signicant nitrite concentration is accumulated in the reactor. This data further support the notation that an initial nitrite accumulation will be required for effective aerobic duration control. As noted previously, the AOBs and NOBs both attain growth opportunities during the aerated period. Aerobic duration control is effective only if the AOBs obtain more growth opportunity than the NOBs. This will occur only if the AOB reaction rate (rAOB) is greater than the NOBs (rNOB) for a given equal available mass of nitrogen substrate. The reaction rate of both AOBs and NOBs are directly proportional to the respective growth rates (Eqs. (1) and (2)): mAOB Y AOB rAOB , Y AOB LNH4 =bAOB 1=SRT YNOB rNOB . Y NOB LNO2 =bNOB 1=SRT (1)

To reduce the substantial time required to achieve a high level of nitritation (greater than 80%), the addition of an NOB inhibitor can be effective. A secondary effect of the NOB inhibitor is to attain the initial nitrite accumulation in a cycle which, as noted from the experimental studies, seemed necessary to begin attaining nitritation. The model predicts nitrite accumulation which is unlike the experimental results that showed no initial nitrite accumulation in each cycle without an external inhibitor added. Fig. 7 shows the discrepancy between the model and experimental results for a single cycle before nitritation was initiated by formic

mNOB

(2)

Assuming that bAOB, bNOB and SRT are all identical for both NOBs and AOBs, the growth rate is equal to a constant bAOB 1=SRT=LNH4 multiplied by the rate of reaction (written for AOBs but the same for NOBs) for a given mass of nitrogen substrate. Hence, the overall relative growth rates

ARTICLE IN PRESS
2174
WAT E R R E S E A R C H

42 (2008) 2166 2176

NH4-N NH4-N Model

NO2-N NO2-N Model

NO3-N NO3-N Model

9 Nitrogen concentration (mgN.L-1) 8 7 6 5 4 3 2 1 0 0 20 40 60 80 Duration (minutes) 100 120 Cut aeration

Fig. 7 Experimental and model data depicting the aerobic and waste periods of the SBR with OUR control implemented (day 138).

are considered the most signicant parameters in determining the effectiveness of aerobic duration control. The overall growth rate is affected by the maximum growth rate value, the Ko value and the Ks value, as shown in the Monod kinetics equation (Table 1). If the growth rate of the NOBs is indeed greater than that of the AOBs, as is commonly assumed (Henze et al., 2000), then sustainable nitritation would not be possible using aerobic duration control since the condition that the AOB reaction rate is greater than the NOB reaction rate would be violated as the NOBs would simply outgrow the AOBs during the aerated period. The growth rate values of both NOBs and AOBs can be determined with knowledge of the maximum growth rate, ammonium, nitrite and DO concentrations. To assess the impact of relative maximum growth rate and Ks values of both NOBs and AOBs, steady-state simulations were used. The effect of the DO concentration and different Ko values (to those used in Table 2) was not considered in this investigation as the oxygen concentration was generally kept above the range where it has a major rate-limiting effect. The exception to this approximation would be at times when COD was carried over from anoxic to aerobic periods, potentially resulting in a competition for oxygen between heterotrophic and autotrophic bacteria. This situation was not considered in this model as COD carry over occurred infrequently. However, a constant difference in oxygen afnity, due to the relative Ko values of the AOBs and NOBs, was included in all simulations with the model, given the Ko value is negligible for the AOBs but 0.4 mgO2 L1 for the NOBs (from Blackburne et al. (2007a)). The results are summarised in Fig. 8, which shows the inuence of the relative maximum growth rate on percent nitritation attainable through aerobic duration control to be more sensitive than the relative Ks values. The area on the left of the curve indicates relative maximum growth rates of AOBs and NOBs, which would likely lead to full nitrication even with aerobic duration control implemented. The area to the right of the curve indicates that nitritation would be observed.

For the Ks values selected for AOBs and NOBs in the model (ratio 0.2), the curve indicates that 100% nitritation would be achieved when the maximum growth rates are equal, while for the maximum growth rate of NOBs exceeding the AOBs, some extent up to 100% of nitrication would be expected. This supports the notion that the growth rate of NOBs should be greater than that of the AOBs for examples of full nitrication, as is commonly accepted. However, this could not be the case in the experiments in this paper, as nitritation would not have been attainable. Our hypothesis for the observed experimental nitritation is that the maximum growth rate of AOBs was greater than that of NOBs, leading to nitritation. This hypothesis is supported by anecdotal evidence from other aerobic duration control reports, such as Peng et al., 2004, which showed a requirement for high temperature, which would favour the AOB growth rate (Hellinga et al., 1998), for the achievement of nitritation. Since we did not require another ongoing selection pressure for AOBs, such as high temperature, it may have been that, after the initial nitritation induction, we selected an AOB culture which had an inherently high growth rate. Alternatively, we may have selected for ocs with minimal mass transfer resistances (i.e. small ocs) which would effectively reduce the ammonium Ks values, thereby raising the AOB growth rate.

4.

Conclusions

This study aimed to investigate aerobic duration control to achieve nitrogen removal via nitrite for a laboratory-scale representation of a domestic wastewater SBR treatment process. The key ndings are: 1. After initially inducing 40% nitrite accumulation with formic acid addition, nitritation could be increased steadily to a sustained 80% nitritation using only the aerobic duration control as a selection pressure for AOBs over

ARTICLE IN PRESS
WAT E R R E S E A R C H

42 (2008) 2166 2176

2175

100 90 80 Percent nitritation 70 60 50 40 30 20 10 0 0 0.2

Model Ks values

0.4 0.8 0.6 1 Relative maximum growth rate of AOB to NOB

1.25 0.2 0.125 0.75

1.2

KsNOB/KsAOB:

Fig. 8 Summarised results from the model with variable comparative maximum growth rates and Ks values of NOBs to AOBs. The solid line corresponds to the Ks values used in the model, while the other types of lines correspond to other Ks values considered.

2.

3.

4.

5.

NOBs. Indeed, aerobic duration control was conrmed as the key selection factor for nitritation by a calibrated twostep nitrication model. The model indicated that the relative growth rates of AOBs and NOBs heavily inuence the degree of nitritation attained. It was conrmed in this study again that the growth rate of NOBs must be less than that of the AOBs to successfully achieve nitritation. The requirement for accumulated nitrite to remain at the end of aeration to ensure the effectiveness of aerobic duration control was proposed as the explanation for this. The model indicated that it should take a substantial time (in the order of years) to completely remove the NOBs using aerobic duration control as the sole selection factor. However, addition of an NOB inhibitor has the effect of decreasing the time required to achieve a high level of nitritation. As well, starting nitritation experimentally was found to be difcult unless an NOB inhibitor was used. Aerobic duration control appears quite effective in attaining nitritation for this installation but its application is probably limited to batch or plug ow processes with a clearly apparent change in the ammonium concentration prole. Other ammonium oxidation congurations with low ammonium concentration at all times (such as an oxidation ditch) may not allow a sufciently large difference in the AOB and NOB growth rates due to the inuence of the high ammonium Ks values relative to nitrite Ks values. The process robustness could be enhanced with a better measure than OUR for the aerobic duration control. OUR is clearly affected by processes other than ammonium oxidation (e.g. heterotrophic oxidation of COD). Volcke et al., 2006 proposed (based on simulation studies) a method to control the extent of nitritation in a SHARON reactor using a cascade feedback control loop with an oxygen set point and pH control. This type of system may also be applicable to this application. Alternatively, the direct

measurement of the ammonium concentration to determine the end of ammonium oxidation may be effective but will rely on a robust ammonium probe, which, in the past, has not been available.

Acknowledgements
Funding for this work was from an Australian Research Council grant, project number: DP0210502.
R E F E R E N C E S

Alleman, J.E., 1984. Elevated nitrite occurrence in biological wastewater treatment systems. Water Sci. Technol. 17, 409419. Anthonisen, A.C., Loehr, R.C., Prakasma, T.B.S., Srinath, E.G., 1976. Inhibition of nitrication by ammonia and nitrous acid. J. Water Pollut. Control Fed. 48, 835852. APHA, 1992. Standard Methods for the Examination of Water and Wastewater. American Public Health Association. Bakti, N.A.K., Dick, R.I., 1992. A model for a nitrifying suspendedgrowth reactor incorporating intraparticle diffusion limitation. Water Res. 26, 16811690. Blackburne, R., Carvalho, G., Yuan, Z.G., Keller, J., 2004. Selective production of nitrite using hydroxylamine as inhibitor of nitrite oxidation. Water Environ. Manage. Ser., 189196. Blackburne, R., Yuan, Z., Keller, J., 2007a. Partial nitrication to nitrite using low dissolved oxygen concentration as the main selection factor. Biodegradation, in press, doi:10.1007/ S10532-007-9136-4. Blackburne, R., Vadivelu, V.M., Yuan, Z., Keller, J., 2007b. Determination of growth rate and yield of nitrifying bacteria by measuring carbon dioxide uptake rate. Water Environ. Res. 79, 24372445. Beccari, M., Dipinto, A.C., Ramadori, R., Tomei, M.C., 1992. Effects of dissolved oxygen and diffusion resistances on nitrication kinetics. Water Res. 26, 10991104.

ARTICLE IN PRESS
2176
WAT E R R E S E A R C H

42 (2008) 2166 2176

Eilersen, A.M., Henze, M., Kloft, L., 1994. Effect of volatile fatty acids and trimethylamine on nitrication in activated sludge. Water Res. 28, 13291336. Fux, C., Lange, K., Faessler, A., Huber, P., Grueniger, B., Siegrist, H., 2003. Nitrogen removal from digester supernatant via nitriteSBR or SHARON? Water Sci. Technol. 48, 918. Gapes, D., 2003. External and internal mass transfer in biological treatment systems. Ph.D. Thesis, Department of Chemical Engineering. The University of Queensland. Australia. Hanaki, K., Wantawin, C., Ohgaki, S., 1990. Nitrication at low levels of dissolved oxygen with and without organic loading in a suspended growth reactor. Water Res. 24, 297302. Hellinga, C., Schellen, A.A.J.C., Mulder, J.W., van Loosdrecht, M.C.M., Heijnen, J.J., 1998. The SHARON Process: an innovative method for nitrogen removal from ammonium-rich waste water. Water Sci. Technol. 37, 135142. Henze, M., Gujer, W., Mino, T., van Loosdrecht, M.C., 2000. Activated Sludge Models ASM1, ASM2, ASM2d and ASM3. IWA Publishing. Kampschreur, M.J., Tan, N.C.G., Picioreanu, C., Jetten, M.S.M., Schmidt, I., van Loosdrecht, M.C.M., 2006. Role of nitrogen oxides in the metabolism of ammonia-oxidising bacteria. Biochem. Soc. Trans. 34, 179181. Kuai, L.P., Verstraete, W., 1998. Ammonium removal by the oxygen-limited autotrophic nitrication-denitrication system. Appl. Environ. Microbiol. 64, 45004506. Laanbroek, H.J., Gerards, S., 1993. Competition for limiting amounts of oxygen between Nitrosomonas europaea and Nitrobacter winogradskyi grown in mixed continuous cultures. Arch. Microbiol. 159, 453459. Lai, E., Senkpiel, S., Solley, D., Keller, J., 2004. Nitrogen removal of high strength wastewater via nitritation/denitritation using a sequencing batch reactor. Water Sci. Technol. 50, 2733. Lee, Y., Oleszkiewicz, J.A., 2003. Effects of predation and ORP conditions on the performance of nitriers in activated sludge systems. Water Res. 37, 42024210. Manser, R., 2005. Population dynamics and kinetics of nitrifying bacteria in membrane and conventional activated sludge plants. Ph.D. Thesis, Swiss Federal Institute for Environmental Science and Technology, Swiss Federal Institute of Technology. Mueller, J.A., Voelkel, K.G., Boyle, W.C., 1966. Nominal diameter of oc related to oxygen transfer. J. Sanit. Eng. Div. (ASCE) 10, 4756. Nowak, O., Schweighofer, P., Svardal, K., 1994. Nitrication inhibitiona method for the estimation of actual maximum

autotrophic growth rates in activated sludge systems. Water Sci. Technol. 30, 919. Peng, Y.Z., Chen, Y., Peng, C.Y., Liu, M., Wang, S.Y., Song, X.Q., Cui, Y.W., 2004. Nitrite accumulation by aeration controlled in sequencing batch reactors treating domestic wastewater. Water Sci. Technol. 50, 3543. Philips, S., Laanbroek, H.J., Verstraete, W., 2002. Origin, causes and effects of increased nitrite concentrations in aquatic environments. Rev. Environ. Sci. BioTechnol. 1, 115141. Pochana, K., Keller, J., 1999. Study of factors affecting simultaneous nitrication and denitrication (SND). Water Sci. Technol. 39, 6168. Siegrist, H., Brunner, I., Koch, G., Phan, L.C., Le, V.C., 1999. Reduction of biomass decay rate under anoxic and anaerobic conditions. Water Sci. Technol. 39, 129137. Sliekers, A.O., Haaijer, S.C.M., Stafsnes, M.H., Kuenen, J.G., Jetten, M.S.M., 2005. Competition and coexistence of aerobic ammonium and nitrite oxidising bacteria at low oxygen concentrations. Appl. Microbiol. Biotechnol. 68, 808817. Stuven, R., Vollmer, M., Bock, E., 1992. The impact of organic matter on nitric oxide formation by Nitrosomonas europaea. Arch. Microbiol. 158, 439443. Third, K.A., Sliekers, A.O., Kuenen, J.G., Jetten, M.S.M., 2001. The CANON system (completely autotrophic nitrogen-removal over nitrite) under ammonium limitation: interaction and competition between three groups of bacteria. Syst. Appl. Microbiol. 24, 588596. Volcke, E.I.P., van Loosdrecht, M.C.M., Vanrolleghem, P.A., 2006. Controlling the nitrite: ammonium ratio in a SHARON reactor in view of its coupling with an Anammox process. Water Sci. Technol. 53, 4554. Wyffels, S., Boeckx, P., Pynaert, K., Zhang, D., Van Cleemput, O., Chen, G., Verstraete, W., 2004. Nitrogen removal from sludge reject water by a two-stage oxygen-limited autotrophic nitrication denitrication process. Water Sci. Technol. 49, 5764. Yang, L., Alleman, J.E., 1992. Investigation of batchwise nitrite buildup by an enriched nitrication culture. Water Sci. Technol. 26, 9971005. Yoo, H., Ahn, K., Lee, H., Lee, K., Kwak, Y., Song, K., 1999. Nitrogen removal from synthetic wastewater by simultaneous nitrication and denitrication (SND) via nitrite in an intermittently-aerated reactor. Water Res. 33, 145154.