w a t e r r e s e a r c h 6 7 ( 2 0 1 4 ) 3 2 1 e3 2 9

Available online at www.sciencedirect.com

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

Modeling nitrogen removal with partial nitritation and anammox in one floc-based sequencing batch reactor Bing-Jie Ni a,*, Adriano Joss a,b, Zhiguo Yuan a a

Advanced Water Management Centre, The University of Queensland, St. Lucia, Queensland 4072, Australia Eawag, Swiss Federal Institute of Aquatic Science and Technology, Ueberlandstr. 133, 8600 Duebendorf, Switzerland

b

article info

abstract

Article history:

Full-scale application of partial nitritation and anammox in a single floc-based sequencing

Received 18 March 2014

batch reactor (SBR) has been achieved for high-rate nitrogen (N) removal, but mechanisms

Received in revised form

resulting in reliable operation are not well understood. In this work, a mathematical model

15 August 2014

was calibrated and validated to evaluate operating conditions that lead to out-competition

Accepted 23 September 2014

of nitrite oxidizers (NOB) from the SBRs and allow to maintain high anammox activity

Available online 2 October 2014

during long-term operation. The validity of the model was tested using experimental data from two independent previously reported floc-based full-scale SBRs for N-removal via

Keywords:

partial nitritation and anammox, with different aeration strategies at aeration phase

Nitrogen removal

(continuous vs. intermittent aeration). The model described the SBR cycle profiles and

Anammox

long-term dynamic data from the two SBR plants sufficiently and provided insights into the

Dissolved oxygen

dynamics of microbial population fractions and N-removal performance. Ammonium

Nitritation

oxidation and anammox reaction could occur simultaneously at DO range of 0.15

Nitrite oxidizers

e0.3 mg O2 L
1 at aeration phase under continuous aeration condition, allowing simplified

Sequencing batch reactor

process control compared to intermittent aeration. The oxygen supply beyond prompt depletion by ammonium oxidizers (AOB) would lead to the growth of NOB competing with anammox for nitrite. NOB could also be washed out of the system and high anammox fractions could be maintained by controlling sludge age higher than 40 days and DO at around 0.2 mg O2 L
1. Furthermore, the results suggest that N-removal in SBR occurs via both alternating nitritation/anammox and simultaneous nitritation/anammox, supporting an alternative strategy to improve N-removal in this promising treatment process, i.e., different anaerobic phases can be implemented in the SBR-cycle configuration. © 2014 Elsevier Ltd. All rights reserved.

1.

Introduction

Biological nitrogen (N) removal via nitrite is attractive as it reduces the energy demand for aeration and the COD demand for subsequent denitrification. Anaerobic ammonium oxidation * Corresponding author. Tel.: þ61 7 3346 3230; fax: þ61 7 3365 4726. E-mail address: [email protected] (B.-J. Ni). http://dx.doi.org/10.1016/j.watres.2014.09.028 0043-1354/© 2014 Elsevier Ltd. All rights reserved.

(anammox) bacteria are capable of autotrophic oxidation of ammonium to N2 with nitrite as the electron acceptor in the absence of molecular oxygen (Strous et al., 1999). Thus, the Nremoval could be achieved with a close cooperation between aerobic ammonia-oxidizing bacteria (AOB) and anammox, the

322

w a t e r r e s e a r c h 6 7 ( 2 0 1 4 ) 3 2 1 e3 2 9

first producing one of the substrates (nitrite) for the second. Since the discovery of anammox about 20 years ago, combined nitritationeanammox has been confirmed as attractive option for N-removal in high strength wastewater due to the low specific costs for N-removal (Wett, 2006; Siegrist et al., 2008; Joss et al., 2009; Daverey et al., 2013; Jenni et al., 2014). N-removal via the nitritation/anammox route requires less oxygen and no organic carbon substrate, consumes less alkalinity, and produces less biomass than via the traditional nitrification/denitrification route (Jetten et al., 2001; Joss et al., 2009, 2011; Lackner et al., 2014). Due to the slow growth rate of the anammox bacteria (Strous et al., 1999), combining nitritation and anammox have mostly been tested in biofilm or granular reactors (Sliekers et al., 2002, 2003; Wyffels et al., 2004; Feng et al., 2007; Gong et al., 2007; Liu et al., 2008). Several other reactor types have also been proposed for this purpose, many of them designed in two steps (Fux et al., 2002; Feng et al., 2007; Gong et al., 2007; Kuenen, 2008). At first step, part of the ammonium is oxidized with oxygen to nitrite (nitritation) in one reactor, while the anammox reaction takes place at second step separately in another reactor. However, two-step processes segregating partial nitritation and anammox required a pH control in the first step and a good coordination of first and second step (van der Star et al., 2007; Gustavsson et al., 2008). Biofilm or granular anammox systems would need extremely long startup time, e.g., the first full-scale anammox reactor (in Rotterdam, NL) has achieved a rate of 10 kg N m
3 d
1 after a startup period of three years (van der Star et al., 2007). Full-scale application of partial nitritation and anammox in a single floc-based sequencing batch reactor (SBR) has been achieved at different municipal plants (Joss et al., 2009, 2011; Lackner et al., 2014), which confirmed the one-step floc-based process suitable for removing N from ammonium-rich wastewater. Advantages of this one-step reactor combined nitritationeanammox solution included: considerable simplification of reactor control and operation (e.g., no pH control), almost no nitrite accumulation, and high activity comparable to the twostage units together (van der Star et al., 2007; Joss et al., 2009). Compared to biofilm systems, floc-based SBR allows significantly shorter startup times or no startup time. In addition, in floc-based systems, partial nitrification can be controlled by choosing an appropriate sludge age due to the differences in growth rates of AOB and nitrite-oxidizing bacteria (NOB). Although nitritation and anammox has been successfully established in a single SBR system, it could be difficult to prevent the growth of NOB efficiently in the long-term operation (Joss et al., 2009, 2011; Brockmann and Morgenroth, 2010; Lackner et al., 2014). The NOB growth can resume rapidly once with favorable conditions (Antileo et al., 2007; Brockmann and Morgenroth, 2010; Joss et al., 2011), which could compete with anammox for nitrite and thus decrease the overall N-removal efficiencies. As evident by the diversity of approaches to achieve partial nitritation and anammox in one SBR (Joss et al., 2009; 2011), mechanisms responsible for the outcompetition of NOB from the system and maintaining high anammox activity are still not well understood. Therefore, the aim of this study was to evaluate operating conditions that lead to out-competition of NOB, and to allow high N-removal efficiencies during long-term operation using

mathematical modeling. The model was calibrated and validated to describe experimental data from two different previously reported floc-based full-scale SBRs for N-removal via partial nitritation and anammox and provided insights into the dynamics of microbial population fractions and Nremoval performance in the floc-based SBR system.

2.

Materials and methods

2.1.

Mathematical model

A multispecies activated sludge model was applied to describe the microbial interactions under a wide range of operational conditions for N-removal via partial nitritation and anammox, employing the software AQUASIM V2.1 (Reichert, 1998). The model describing two-step nitrificationedenitrification was extended by implementing an anammox process. Four bacterial metabolisms were considered in this model as described by Koch et al. (2000) and Hao et al. (2002a): anammox bacteria, AOB, NOB, and heterotrophs. The model was summarized in Tables S1eS3 of the Supplementary material. The kinetic and stoichiometric parameters are listed in Table S4 of the Supplementary material. The operational temperature of the two full-scale SBRs in this work varied from 18  C to 37  C during the long-term operation. Thus, the effect of temperature on a rate constant relative to a standard temperature (293 K) was included in the model, which can be expressed by a modified Arrhenius equation (Hao et al., 2002b): rT ¼ r293 exp½qT ðT
293Þ; with qT ¼

lnðrT1 =rT2 Þ T1
T2

(1)

where rT is the reaction rate at temperature T; r293 is the reaction rate at the standard temperature (293 K), T is the temperature (K), and qT is the temperature coefficient (K
1), which can be calculated according to Hao et al. (2002b). In this model, the pH variation was not specifically modeled since the pH in both experimental SBR systems was constantly at 7.5 ± 0.5 during the long-term operations, which generally favor the growth of the microorganisms in the systems (Joss et al., 2009). The alkalinity was included as one of the state variables in the model according to Gujer et al. (1999) in order for the detection of possible pH failure. The inhibition of nitrite on anammox organisms was also not included due to the fact that nitrite accumulation level in both full-scale SBRs (~1.5 mg N L
1) was much lower than the inhibitory level previously reported in literature (Jin et al., 2012). However, the model can be modified to include possible pH or/and nitrite inhibitory effects if necessary in future applications, which could be included in the kinetic rate expression of the biological processes in Table S2 (Hellinga et al., 1999). In addition, the possible effects of mass transfer in flocs on substrate conversion were lumped in the model through employing the apparent substrate affinity constants that have been widely applied into floc sludge system reported in literature.

2.2.

Experimental data from two full-scale SBRs

Experimental data from two previously reported floc-based full-scale SBRs for N-removal via partial nitritation and

w a t e r r e s e a r c h 6 7 ( 2 0 1 4 ) 3 2 1 e3 2 9

anammox (Joss et al., 2009) are used for the model evaluations. Two 1400-m3 SBR reactors (18  12  6.5 m) consisting of a completely stirred reactor, a feed pump, a decanter and an aeration unit are operated with the single-SBR process, treating 1800 m3 d
1 with a total nitrogen load of 1250 kg N d
1. Each reactor is equipped with two complete aeration units (800e2000 N m3 h
1 and 1600e4100 N m3 h
1) and two 7 kW stirrers. The process is controlled by a programmable logical controller equipped with online sensors for the water level, ammonium, volumetric air flow control to the aeration unit, soluble oxygen, temperature, pH and conductivity. The SBR cycle always comprises a feeding phase, one aeration phase, one mixing phases, a sedimentation phase and a discharge phase. A complete cycle typically lasts between 6 and 10 h under normal conditions. The maximum feed rate is 220 m3 h
1. A floating decantation unit with a maximum flow of 600 m3 h
1 is responsible for the discharge. Flowproportional 24 h samples are taken from the reactors' inflow and outflow. The detail average composition of the feeding ammonium-rich digester supernatant can be found in previous work (Joss et al., 2009). The SBR cycle profiles and the effluent dynamic data used for model evaluations were taken from the long-term continuous operation (over 300-day) of both full-scale nitritation/anammox SBRs after they both achieved successful startup. The SBR A was operated with continuous aeration and the SBR B with intermittent aeration for comparison. More operational details for both SBRs are given in Joss et al. (2009). The variations of ammonium and DO concentrations in typical SBR cycles, as well as the influent and effluent ammonium, nitrite, and nitrate concentrations in both full-scale SBRs were monitored by online measurements and regular sampling for offline analysis.

2.3.

Model calibration and validation

The model was calibrated using experimental data from the SBR cycle profiles from the full-scale SBR A. The model includes 15 species-specific biochemical processes and over 30 stoichiometric and kinetic parameters, as summarized in Tables S2eS4 of the Supplementary material. Most of these parameters are well established in previous studies. Thus, literature values reported with similar model structure were directly adopted for these parameters (Table S4). Sensitivity analysis reveals that the maximum growth rates of AOB, NOB and anammox bacteria (e.g., mAOB, mNOB and mAN) are the key parameters to predict the data sets available. Parameter estimation based on experimental measurements was then carried out for three parameters, namely mAOB, mNOB and mAN (see Table S4). These three parameter values were estimated through minimizing the sum of squares of the deviations between the experimental measurements and the model predictions for the SBR cycle data of SBR A. The validation step was then carried out with the calibrated model parameters (Table S4) using the long-term monitoring data of effluent ammonium, nitrite, and nitrate concentrations from SBR A with dynamic influent conditions which has not been used to estimate the parameters. To further verify the validity and applicability of the model, we also applied the model to evaluate the experimental data from SBR B using the same parameter values listed in Table S4

323

(without further calibration). The SBR B was operated with different conditions from SBR A (intermittent vs. continuous aeration). At the first place, the experimental SBR cycle profiles from SBR B were compared with the model predicted results. As a further validation, the model and same parameter values were also used to generate additional sets of simulation data for comparison with the measured data of long-term effluent ammonium, nitrite, and nitrate concentrations from SBR B. These extensive evaluations could strongly support the validity and applicability of the model of this work which was very important for further model simulations. The difference of the microbial population fractions and N-removal efficiencies under different aeration conditions (continuous vs. intermittent aeration at aerobic phase) was investigated based on model simulations. With respect to the estimation of active bacteria (AOB, NOB and anammox bacteria) proportions, the simulations were conducted with the same initial and operating conditions to the experiments. In addition, model simulations were also performed to elucidate the effects of DO (0.05e0.5 mg O2 L
1) and solids retention time (SRT, 10e60 days) on the SBR performance.

3.

Results

3.1. Model calibration and validation with data from SBR A Model calibration of this work involved adjusting three key parameter values for the system so that the SBR cycle profiles produced by the model closely agreed with the measured data from SBR A. The three key parameters (i.e., mAOB, mNOB and mAN) were estimated by fitting simulation results to the monitoring data. Fig. 1 shows the model simulations matched the measured ammonium and DO data well (R2 ¼ 0.95). Nitrite accumulation in the SBR cycle was measured to be very low (~1.5 mg N L
1) during the whole operation period, which was also well predicted by the model (Fig. 1). These results supported that the developed model properly captures the relationships among ammonium utilization, nitrite conversion and DO consumption. Model and parameter validation was then based on the comparison between the model predictions and the long-term monitoring data collected from dynamic operation of SBR A which was operated with continuous aeration. Comparisons of the experimental and the model simulated effluent ammonium, nitrite, and nitrate concentrations in SBR A are shown in Fig. 2. There is a close correlation between the measured and the modeled results (R2 ¼ 0.89). Overall, a good prediction is observed with respect to the ammonium, nitrite, and nitrate concentrations and their disturbance. The mean value of the differences between the measured data and model predictions is relatively low, indicating that the kinetic model and parameters for AOB, NOB and anammox bacteria in this work are applicable to simulate this full-scale partial nitritation and anammox SBR system. The low nitrite concentrations predicted at the end of aeration phase (see Fig. 1) confirmed that nitrite consumption by anammox could occur simultaneously to the aeration in SBR A.

324

w a t e r r e s e a r c h 6 7 ( 2 0 1 4 ) 3 2 1 e3 2 9

Feeding Continuous aeration Other cycle phases

200

1.0

+

+

Modelled NH4 0.8

160

-1

+

NH4

120

-

Modelled NO2

0.6

1

-

Measured DO1 Modelled DO

80

0.4

+

-1

DO

DO (mg L )

NH4 and NO2 (mg L )

Measured NH4

0.2

40 -

NO2 0

0

2

4 Time (h)

0.0 6

8

Fig. 1 e Model calibration results using experimental data from SBR cycle profiles of the full-scale SBR A.

The simulated ammonium profiles of SBR A roughly match the trend in the experimental profiles. Moreover, the model is capable of accurately simulating the short-term effects resulting from the variation of DO concentration, in which high simulated peaks of ammonium are observed at the low DO levels (Fig. 2A). However, the predicted nitrate concentrations are lower than the measured ones in some case, as shown in Fig. 2B. Such a discrepancy may be attributed to the fact that heterotrophic denitrification may partially take place at the aerobic phase because of the possible local varying concentrations of oxygen and locally different biomass behavior in experiments.

(A)

120 80

+

-1

NH4 (mg L )

160

40

(B)

30 20

3.2.

Model validation with data from SBR B

To further verify the validity and applicability of the model describing full-scale N-removal via partial nitritation and anammox in floc-based SBR, model predicted results are compared with experimental data from SBR B which examined the intermittent aeration. The model and its parameters (the same parameters shown in Table S4) were first evaluated with the SBR cycle profiles of SBR B. The validation results in Fig. 3 show that the model predictions well match the measured data in terms of ammonium and DO profiles for the validation experiments (R2 ¼ 0.96). Furthermore, the model shows no systematic deviations. The good correspondence for independent data set supports the validity of the developed model for the floc-based SBR with intermittent aeration. The validation of the model and its parameters were further performed with the long-term experimental data from dynamic operation of SBR B. Fig. 4 compares the simulated and experimental results for effluent ammonium, nitrite, and nitrate concentrations during dynamic operation of SBR B. Again, the model captured all the experimental trends. The both simulated and measured ammonium is shown in Fig. 4A and a close correlation is observed (R2 ¼ 0.87). These independent model validation results with both SBRs A and B using one set of parameter values (Figs. 2e4 and Table S4) support the accurate predictions of both aerobic and anaerobic ammonium oxidation activity in the model, and also demonstrates the effectiveness of the parameter values for

-

-1

NO3 (mg L )

0 40

There is a slight deviation between the predicted and observed nitrite concentration, especially after day 100 (Fig. 2C). These differences may be due to an underestimation of either the heterotrophic denitrification or anaerobic ammonium oxidation activity in the simulation, but it is also possible that differences in the oxygen flux caused this deviation. In addition, in SBR A, the nitrate production was underestimated in some cases after day 100. NOB may be more persistent than model predicted: the persistence of Nitrobacter NOBs in bioreactor even under oxygen limited conditions has also been shown by literature (Ahn et al., 2008). Even though the predictions appear difference, the model is able to reasonably predict the nitrite changing trends (Fig. 2C).

10

6

Feeding

200

Other phases +

+

NH4

-

Modelled NO2

120

100

200 300 Time (days)

400

500

Fig. 2 e Model validation results using experimental data from the long-term operation of the full-scale SBR A (measured data, dot; and predicted profiles, line): (A) effluent ammonium concentrations; (B) effluent nitrate concentrations; and (C) effluent nitrite concentrations.

+

0

0.9

1

Measured DO1 Modelled DO

DO

80

0.6

-1

0

1.5

+ Modelled NH4 1.2

160

-

2

Intermittent aeration

Measured NH4

-1

4

NH4 and NO2 (mg L )

-1 -

(C)

DO (mg L )

NO2 (mg L )

0

0.3

40 -

0

NO2 0

2

4 Time (h)

0.0 6

8

Fig. 3 e Model validation results using experimental data from SBR cycle profiles of the full-scale SBR B.

w a t e r r e s e a r c h 6 7 ( 2 0 1 4 ) 3 2 1 e3 2 9

the processes. In addition, the model gives reasonable simulation results for the nitrite and nitrate concentrations (Fig. 4B and C), indicating that the model is appropriate to qualitatively describe the out-competition of NOB in one floc-based SBR system. Nitrite oxidation activity by NOB should has been avoided or limited in nitritation/anammox and its depletion is rate-limiting for the overall N-removal because this process leads to nitrate accumulation and the heterotrophic nitrate reduction is limited by the availability of organic substrate. The agreement between the model simulated and experimental data of SBR B shown in Fig. 4 further confirms the validity of the model. The comparable effluent nitrate concentrations predicted in both SBRs (varying in both in the range of 0e20 mg NO
3N L
1, as shown in Figs. 2 and 4) indicated that the two aeration strategies also perform comparably in terms of nitrite oxidizing activity depletion. However, higher ammonium consumption rate is achieved under continuous aeration (SBR A: 890 mg N L
1 d
1) compare to intermittent aeration (SBR B: 650 mg N L
1 d
1) since under regular operation conditions the simultaneous anammox activity can deplete more ammonium with the continuously produced nitrite from AOB activity during the aeration phase.

3.3.

Aeration strategies at aerobic phase

Experimental and modeling results of SBRs A and B suggested aeration at aerobic phase is important (continuous and intermittent aeration) since even very low oxygen concentrations could inhibit anammox bacteria (Strous et al., 1997). A

+

-1

NH4 (mg L )

100

-

-1

NO3 (mg L )

batch of seven simulations (Fig. 5) was run by keeping the values of DO concentration, but modifying the length of the cycles and the time fraction in which the reactor is aerated to evaluate the effect of aeration strategies at aerobic phase to the active biomass fractions and N-removal efficiencies. The feeding ammonium concentration was set constant to 600 mg L
1 according to the experimental condition. Fig. 5 shows the active bacteria fractions and N-removal efficiencies as a function of the frequency with which the aeration takes place. For the 4-h aeration phase, anammox bacteria dominate at the continuous aeration and 2-h aerobic/ 2-h anoxic aeration; while AOB dominate at the 1-h aerobic/1h anoxic aeration. NOB are outcompeted by AOB at all three cases because the oxygen to nitrogen supply ratio is significantly smaller than what is required for complete nitrification (Fig. 5A). It can be seen that the growth of anammox is the key that sustains the high N-removal efficiencies in the system (Fig. 5B). At the 1-h aerobic/1-h anoxic aeration, anammox growth is not sustained as well as other two cases while AOB can not convert enough ammonium to nitrite for anammox bacteria growth. An opposite distribution profile is observed for the 6-h aeration phase: anammox bacteria are dominant at the intermittent aeration while AOB and NOB are dominant at the continuous aeration (Fig. 5A). Although AOB attain higher specific growth rates and high ammonium removal efficiencies at the continuous aeration than at the intermittent

(A)

80 60 40 20 0

(B)

30 20 10 0

(C)

2.0

-

-1

NO2 (mg L )

325

1.5 1.0 0.5 0.0

0

60

120

180 240 Time (days)

300

360

Fig. 4 e Model validation results using experimental data from the long-term operation of the full-scale SBR B (measured data, dot; and predicted profiles, line): (A) effluent ammonium concentrations; (B) effluent nitrate concentrations; and (C) effluent nitrite concentrations.

Fig. 5 e The reactor performance of SBR with nitritation and anammox as a function of the frequency with which the aeration takes place: (A) the active bacteria fractions; and (B) N-removal efficiencies.

326

w a t e r r e s e a r c h 6 7 ( 2 0 1 4 ) 3 2 1 e3 2 9

The steady-state active bacteria fractions and N-removal performance at different DO concentrations are evaluated with continuous aeration at 4-h aeration phase. The microbial population distributions and conversions to N components together with N-removal performance at different DO are shown in Fig. 6. There is clear increase of total autotrophs (AOB, NOB and anammox bacteria) with a maximum at DO range of 0.15e0.3 mg O2 L
1 (Fig. 6A). At this operational DO range, the anammox bacteria dominate with a fraction up to 70% of the total autotrophs and NOB is almost completely outcompeted, which sequentially results in maximal total Nremoval in the SBR system (Fig. 6B). The total effluent nitrogen concentration reaches a minimum of about 50 mg N L
1 at DO of 0.2 mg O2 L
1, corresponding to an ammonium removal of 95% and 90% for the TN removal. At this point, the amounts of nitrite and nitrate productions are about 20 mg N L
1 and zero. Nitrate production increases with the increasing DO concentrations from 0.3 to 0.5 mg O2 L
1 with a maximum at 0.45 mg O2 L
1 (in Fig. 6B). At this point, the total N removal decrease from maximal 90% to about 20%. At further elevated DO concentrations (larger than 0.45 mg O2 L
1), the nitrification processes shift to nitrate formation. The NOB can effectively compete with AOB for oxygen and with anammox bacteria for nitrite. Nitrate production increases with the outcompeting of anammox bacteria and reach a high value of 450 mg N L
1 (Fig. 6B), in which anammox bacteria are completely outcompeted (Fig. 6A). As shown in Fig. 6A, NOB becomes the important fractions of total autotrophs in the SBR (up to 35%) and almost no anammox bacteria grow. Further increasing DO concentrations reduce the nitrite production significantly because the oxygen supply is significantly larger than what is required for complete nitrification. It thus can be concluded that the DO concentrations has a profound effect on active biomass fractions and N-removal efficiencies in the full scale floc-based SBR with nitritation and

Active bacteria fraction

Total

1500 1200

0.6

900

0.4

600

0.2

300

0.1

0.2 +

(B)

NH4

600

-1

NOB

0.8

0.0 0.0

N concentration (mg L )

Anammox

0.3 -

NO2

-

NO3

0

0.4

0.5

TN

NH4

+

100 80

480

60

360

DO cotrol for high rate N-removal

240 120 0 0.0

0.1

0.2 0.3 0.4 -1 DO concentration (mg L )

40 20

0.5

Removal efficiency (%)

Controlling with DO concentration

AOB

1.0

-1

3.4.

(A)

Total autotrophs (mg L )

aeration (Fig. 5B), NOB growth rates exceed those of anammox bacteria, resulting in NOB domination and low Total N (TN) removal efficiencies. Furthermore, for the 6-h aeration phase, the aeration frequency has a small effect in the active biomass fractions and performance of the reactor. These simulation results suggested that the DO supply rather than aeration strategies might be key element that governing the overall N-removal efficiencies of the floc-based SBR with nitritation and anammox. In addition, experimental results of nitritationeanammox (Wett, 2006; Joss et al., 2009, 2011) showed that at low DO concentrations anammox could occur simultaneously. Thus, the SBR can be operated with a single continuous aeration phase per SBR cycle (e.g., the 4-h aeration phase of Fig. 5) at sufficient anammox activity. In this operating mode (controlling oxygen around 0.2 mg O2 L
1), simultaneous nitrite formation and depletion results in almost no nitrite accumulation. The principal advantages of continuous aeration are its simplicity, better monitoring of reactor performance and ultimately also high performance (Fig. 5). Increased simplicity results from the requirement of significantly less on/off switching of the aerators compare to intermittent aeration.

0

Fig. 6 e The steady-state reactor performance (continuous aeration at 4-h aeration phase) at different DO concentrations: (A) the active bacteria fractions; and (B) Nremoval efficiencies.

anammox. A low DO concentration resulted in a very low total autotrophic biomass dominated by AOB with a low fraction of anammox bacteria growth in the system, but an over high DO concentration would lead to purely AOB and NOB dominated sludge with high nitrate contained effluent due to outcompetition of anammox bacteria by NOB although the ammonium removal efficiency increased. Therefore, controlling with DO concentration (e.g. at DO range of 0.15e0.3 mg O2 L
1) at aeration phase of SBR should be the main strategy for displacing NOB with anammox bacteria, and thus governing the overall N-removal efficiencies.

3.5.

SRT governing anammox activity

The SRT could be another process parameter determining the active bacteria fractions and N-removal efficiencies of the full scale floc-based SBR with nitritation and anammox. The effects of the SRT on the fractions of active biomass are illustrated in Fig. 7A. As expected, the total active autotrophs concentration increases with increasing SRT (Fig. 7A). However, the active bacteria fractions vary greatly with increasing SRT. For instance, at SRT ¼ 60 days, AOB is 26.6% of total active autotrophs, while anammox bacteria is up to 73% and NOB is only 0.4% of total active autotrophs. However, at SRT ¼ 25 days, AOB is about 97% of total active autotrophs, while anammox bacteria is only 3% and NOB is completely washed out. Fig. 7A illustrates vividly the dynamics of the three forms of active autotrophs fractions. The AOB fractions decrease gradually after a SRT of 20 days under the simulation conditions. However, the anammox bacteria do not behave the same as AOB. The highest accumulation of anammox bacteria occurs when AOB is declining: anammox bacteria fractions

327

w a t e r r e s e a r c h 6 7 ( 2 0 1 4 ) 3 2 1 e3 2 9

2000 1600

0.6

1200

0.4

800

0.2

400

10

20 +

NH4

600

30 NO2

40 NO3

50 TN

60 + NH4

480

100 80

360

SRT control to retain Anammox

240 120 0

0

60 40 20

10

20

30 40 SRT (days)

50

60

0

Fig. 7 e The steady-state reactor performance (DO ¼ 0.2 mg O2 L¡1) at different operational SRT of the SBR: (A) the active bacteria fractions; and (B) N-removal efficiencies.

increase when SRT is higher than 20 days, and it stabilized for larger SRT; at SRT ¼ 50 days, anammox bacteria are about 70% of total active autotrophs. The NOB is almost disappeared under all SRT conditions (10e60 days), although it becomes a very small fraction at an SRT of 60 days. The simulation results showed that NOB faction of total autotrophs increased from 0.1% to 1% with the increase of SRT from 30 to 60 days, suggesting that NOB growth would occur in the system with longer SRT. The low fraction of NOB was likely due to the very low DO level of 0.2 mg-O2/L that applied in the simulation considering a relative high oxygen affinity constant of 2.2 mgO2/L for NOB was used in the model. As a whole, NOB can be largely washed out and high anammox fractions can be maintained if the system is operated at a SRT >40 days and by controlling the DO at around 0.2 mg O2 L
1 at aeration phase of SBR. Fig. 7B shows the effluent N components and N-removal performance under different SRT conditions. The lowest effluent TN concentration can be obtained for an SRT >40 days, which is corresponding to the high anammox fractions. In this case, ammonium and nitrite dominates the effluent N, and it can be driven down to about 5% of the influent N. The Nremoval efficiency associated with the high anammox fraction (70% of total active autotrophs) is about 95%. For lower SRT (20e30 days), the effluent N is dominated with nitrite and closer to 60e80% of influent N.

4.

Discussion

Attempts to improve N-removal in one-step partial nitritation and anammox systems have so far consisted in decreasing the

operational DO of aerated phases in the SBR cycle (Joss et al., 2009, 2011) or bulk DO of biofilm/granulation systems for careful microenvironment (concentration gradients within biofilm or granule) control (Hao et al., 2002a). These strategies are based on the assumption that nitritation and anammox occur simultaneously (Joss et al., 2009), in which N-removal is enhanced by decreasing the oxygen penetration in the floc or biofilm/granule. The modeling results describe that simultaneous nitritation/anammox (SNA) is present indeed. However, the results also suggest that contribution of SNA to high N-removal is limited. Fig. 8 shows the relative contribution of SNA to the total N-removal in seven different case simulations. In the continuous aeration (no anoxic mixing phase in the SBR cycle) cases (nos. 1 and 2), total N-removal is lower than 40%, attributing to the SNA. Controlling of DO concentrations (0.1 vs. 0.2 mg O2 L
1) can only slightly increase the total Nremoval efficiency (Fig. 8). When anaerobic periods are introduced, as from cases 3e7, the overall N-removal becomes significantly higher. In addition, controlling of DO concentrations and anaerobic periods length can increase the total N-removal efficiency. However, significant anammox activity occurs during the anaerobic period, when nitritation is absent. In these cases, the nitrite being consumed at the beginning of the anaerobic periods is reminiscent from the aerobic phase. This means that anammox and nitritation occur in alternating nitritation/anammox (ANA). The inhibition of oxygen on anammox may temporarily lead to AOB activity in the outer layer and anammox activity in the inner anaerobic layer under the alternating aerobic/anaerobic conditions. However, this may not necessarily result in a stratification of the biomass itself in the flocs due to the reversibility of oxygen inhibition on anammox in the system (Joss et al., 2011). A homogeneous distribution of the biomass in flocs could not be excluded because of flocs growing and falling apart randomly in the system (Jeanningros et al., 2010). These results show that attempts to increase N-removal by operating the reactor at low DO, supposedly to increase SNA,

100 TN removal efficiency (%)

Total

Removal efficiency (%)

-1

NOB

-1

N concentration (mg L )

Anammox

0.8

0.0

(B)

AOB

1.0

Total autotrophs (mg L )

Active bacteria fraction

(A)

80 60 40

Alternated nitritation/Anammox Simultaneous nitritation/Anammox

Cycling aeration (2, 3, 4, 5, and 6 h, respectively) With anaerobic phase DO 0.3 mg L

DO 0.3 mg L

DO 0.2 mg L

Constant aeration No anaerobic phase DO 0.1 mg L

DO 0.2 mg L

DO 0.2 mg L

DO 0.2 mg L

20 0

1

2

3

4 Case number

5

6

7

Fig. 8 e The N-removal in the floc-based SBR is the result of simultaneous nitritation/anammox (SNA, dark areas in bars) and that resulting from alternated nitritation/ anammox (ANA, white areas in bars).

328

w a t e r r e s e a r c h 6 7 ( 2 0 1 4 ) 3 2 1 e3 2 9

can be only partially successful. In fact, N-removal can be significantly enhanced when operational changes create improved conditions for ANA, such as in case nos. 3, 4, and 5, which have anaerobic periods in the operation cycle. Realizing that N-removal is due to both ANA and SNA (Fig. 8) allows to design alternative strategies to further increase N-removal in full scale floc-based SBR with nitritation and anammox, i.e., different anaerobic phases can be implemented in the SBRcycle configuration (e.g. cases no 4 and 5).

5.

Conclusions

The following conclusions could be drawn from this study on the modeling of full-scale N-removal via partial nitritation and anammox in one floc-based SBR: 1. Continuous aeration at aeration phase of SBR could simplify process control compared to intermittent aeration, and achieve good overall performance with nitritation and anammox occurring simultaneously. 2. DO limitations were identified as key factor for outcompetition of NOB and maintaining high anammox activity in the SBR. A DO range of 0.15e0.3 mg O2 L
1 at aeration phase was crucial to achieve high N removal efficiency while keeping NOB out of the reactor during longterm operation. 3. NOB can be largely washed out of the system and high anammox fractions can be maintained if the system is operated at a SRT >40 days and by controlling the DO at around 0.2 mg O2 L
1 at aeration phase of SBR. 4. N-removal in the full scale floc-based SBR occurs via both alternating nitritation/anammox and simultaneous nitritation/anammox, supporting an alternative strategy to improve N-removal, i.e., different anaerobic phases can be implemented in the SBR-cycle configuration.

Acknowledgments This study was supported by the Australian Research Council through Project DP130103147. Bing-Jie Ni acknowledges the support of Australian Research Council Discovery Early Career Researcher Award (DE130100451).

Appendix A. Supplementary data Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.watres.2014.09.028.

references

Ahn, J.H., Yu, R., Chandran, K., 2008. Distinctive microbial ecology and biokinetics of autotrophic ammonia and nitrite oxidation in a partial nitrification bioreactor. Biotechnol. Bioeng. 100, 1078e1087.

Antileo, C., Roeckel, M., Lindemann, J., Wiesmann, U., 2007. Operating parameters for high nitrite accumulation during nitrification in a rotating biological nitrifying contactor. Water Environ. Res. 79, 1006e1014. Brockmann, D., Morgenroth, E., 2010. Evaluating operating conditions for outcompeting nitrite oxidizers and maintaining partial nitrification in biofilm systems using biofilm modeling and Monte Carlo filtering. Water Res. 44, 1995e2009. Daverey, A., Su, S.H., Huang, Y.T., Chen, S.S., Sung, S., Lin, J.G., 2013. Partial nitrification and anammox process: a method for high strength optoelectronic industrial wastewater treatment. Water Res. 47, 2929e2937. Feng, Y.J., Tseng, S.K., Hsia, T.-H., Ho, C.-M., Chou, W.-P., 2007. Partial nitrification of ammonium-rich wastewater as pretreatment for anaerobic ammonium oxidation (Anammox) using membrane aeration Bioreactor. J. Biosci. Bioeng. 104, 182e187. Fux, C., Boehler, M., Huber, P., Brunner, I., Siegrist, H., 2002. Biological treatment of ammonium-rich wastewater by partial nitritation and subsequent anaerobic ammonium oxidation (anammox) in a pilot plant. J. Biotechnol. 99, 295e306. Gong, Z., Yang, F.L., Liu, S., Bao, H., Hu, S., Furukawa, K., 2007. Feasibility of a membrane-aerated biofilm reactor to achieve single-stage autotrophic nitrogen removal based on Anammox. Chemosphere 69, 776e784. Gujer, W., Henze, M., Mino, T., van Loosdrecht, M.C.M., 1999. Activated sludge model No. 3. Water Sci. Technol. 39, 183e193. Gustavsson, D.J.I., Nyberg, U., la Cour Jansen, J., 2008. Operation for nitritation of sludge liquor in a full-scale SBR. Water Sci. Technol. 58, 429e444. Hao, X.D., Heijnen, J.J., van Loosdrecht, M.C.M., 2002a. Sensitivity analysis of a biofilm model describing a one-stage completely autotrophic nitrogen removal (CANON) process. Biotechnol. Bioeng. 77, 266e277. Hao, X.D., Heijnen, J.J., van Loosdrecht, M.C.M., 2002b. Modelbased evaluation of temperature and inflow variations on a partial nitrificationeAnammox biofilm process. Water Res. 36, 4839e4849. Hellinga, C., van Loosdrecht, M.C.M., Heijnen, J.J., 1999. Model based design of a novel process for nitrogen removal from concentrated flow. Math. Comput. Model. Dyn. Syst. 5, 351e371. Jeanningros, Y., Vlaeminck, S.E., Kaldate, A., Verstraete, W., Graveleau, L., 2010. Fast start-up of a pilot-scale deammonification sequencing batch reactor from an activated sludge inoculum. Water Sci. Technol. 61, 1393e1400. Jenni, S., Vlaeminck, S.E., Morgenroth, E., Udert, K.M., 2014. Successful application of nitritation/anammox to wastewater with elevated organic carbon to ammonia ratios. Water Res. 49, 316e326. Jetten, M.S.M., Wagner, M., Fuerst, J., van Loosdrecht, M.C.M., Kuenen, G., Strous, M., 2001. Microbiology and application of the anaerobic ammonium oxidation (‘anammox’) process. Curr. Opin. Biotechnol. 12, 283e288. Jin, R.C., Yang, G.F., Yu, J.J., Zheng, P., 2012. The inhibition of the Anammox process: a review. Chem. Eng. J. 197, 67e79. Joss, A., Derlon, N., Cyprien, C., Burger, S., Szivak, I., Traber, J., Siegrist, H., Morgenroth, E., 2011. Combined nitritationeanammox: advances in understanding process stability. Environ. Sci. Technol. 45, 9735e9742. € nig, R., Rottermann, K., Joss, A., Salzgeber, D., Eugster, J., Ko Burger, S., Fabijan, P., Leumann, S., Mohn, J., Siegrist, H., 2009. Full-scale nitrogen removal from digester liquid with partial nitritation and anammox in one SBR. Environ. Sci. Technol. 43, 5301e5306. Koch, G., Egli, K., van der Meer, J.R., Siegrist, H., 2000. Mathematical modeling of autotrophic denitrification in a nitrifying biofilm of a rotating biological contactor. Water Sci. Technol. 41, 191e198.

w a t e r r e s e a r c h 6 7 ( 2 0 1 4 ) 3 2 1 e3 2 9

Kuenen, J.G., 2008. Anammox bacteria: from discovery to application. Nat. Rev. Microbiol. 6, 320e326. Lackner, S., Gilbert, E.M., Vlaeminck, S.E., Joss, A., Horn, H., van Loosdrecht, M.C.M., 2014. Full-scale partial nitritation/ anammox experiences e an application survey. Water Res. 55, 292e303. Liu, S., Yang, F., Xue, Y., Gong, Z., Chen, H., Wang, T., Su, Z., 2008. Evaluation of oxygen adaptation and identification of functional bacteria composition for anammox consortium in non-woven biological rotating contactor. Bioresour. Technol. 99, 8273e8279. Reichert, P., 1998. AQUASIM 2.0dComputer Program for the Identification and Simulation of Aquatic Systems. EAWAG, Du¨bendorf, Switzerland. Siegrist, H., Salzgeber, D., Eugster, J., Joss, A., 2008. Anammox brings WWTP closer to energy autarky due to increased biogas production and reduced aeration energy for N-removal. Water Sci. Technol. 57, 383e388. Sliekers, A.O., Derwort, N., Gomez, J.L.C., Strous, M., Kuenen, J.G., Jetten, M.S.M., 2002. Completely autotrophic nitrogen removal over nitrite in one single reactor. Water Res. 36, 2475e2482.

329

Sliekers, A.O., Third, K.A., Abma, W., Kuenen, J.G., Jetten, M.S.M., 2003. CANON and Anammox in a gas-lift reactor. FEMS Microbiol. Lett. 218, 339e344. Strous, M., Kuenen, J.G., Jetten, M.S.M., 1999. Key physiology of anaerobic ammonium oxidation. Appl. Environ. Microbiol. 65, 3248e3250. Strous, M., Van Gerven, E., Kuenen, J.G., Jetten, M.S.M., 1997. Effects of aerobic and microaerobic conditions on anaerobic ammonium-oxidizing (Anammox) sludge. Appl. Environ. Microbiol. 63, 2446e2448. van der Star, W.R.L., Abma, W.R., Blommers, D., Mulder, J.W., Tokutomi, T., Strous, M., Picioreanu, C., Van Loosdrecht, M.C.M., 2007. Startup of reactors for anoxic ammonium oxidation: experiences from the first full-scale anammox reactor in Rotterdam. Water Res. 41, 4149e4163. Wett, B., 2006. Solved upscaling problems for implementing deammonification of rejection water. Water Sci. Technol. 53, 121e128. Wyffels, S., Boeckx, P., Zhang, D., Van Cleemput, O., Chen, G., Verstraete, W., 2004. Nitrogen removal from sludge reject water by a two-stage oxygen-limited autotrophic nitrification denitrification process. Water Sci. Technol. 49, 57e64.