A laboratory fed-batch reactor has been used to study under controlled conditions the performance of partial nitritation/anammox for the 200,000 m3/day step-feed activated sludge process at the Changi Water Reclamation Plant, Singapore. The similarity of the concentrations of NH4, NO2, NO3, PO4, suspended chemical oxygen demand (sCOD), pH, and alkalinity (ALK) between the on-site process and laboratory reactor illustrates that the laboratory fed-batch reactor can be used to simulate the site performance. The performance of the reactor fed by primary effluent illustrated the existence of anammox and heterotrophic denitrification and apparent excessive biological phosphorus removal as observed from the site. The performance of the reactor fed by final effluent proved the presence of anammox process on site. Both the laboratory reactor and on-site process showed that higher influent 5-day biochemical oxygen demand/total nitrogen (BOD5/TN) (COD/TN) ratio increases the nitrogen removal efficiency of the process.

INTRODUCTION

Changi Water Reclamation Plant (WRP), which was commissioned in 2007 (Daigger et al. 2008), is currently the largest WRP in Singapore with a treatment capacity of 800,000 m3/day of municipal wastewater. There are four parallel trains (reactors) of step-feed activated sludge (SFAS) process with identical configuration, working volume, separate settling tanks, and return activated sludge (RAS) facilities. Each train treats 200,000 m3/d of wastewater. Given that the Changi WRP has the lowest final effluent nitrogen (total nitrogen, TN < 5 mgN/L) and significant nitrite concentration (1.1 mg NO2-N/L) compared with the other activated sludge processes in Singapore (Ou et al. 2012; Cao et al. 2014a), four comprehensive sampling programs in Train 2 were carried out from 2011 to 2012. The results demonstrated significant presence of autotrophic nitrogen removal in the activated sludge process (Cao et al. 2013, 2014a).

Lotti et al. (2015) recently evaluated the kinetic parameters for anammox bacteria growing at the temperature of Changi WRP (30 °C) and in a system with an anoxic sludge retention time (SRT) of 3 d. The results showed that flocculent/free suspended cell anammox bacteria, dominated by Candidatus Brocadia sp. 40, were able to be retained at such a relatively low SRT (Lotti et al. 2015), thus supporting the hypothesis that suspended or free cells of anammox bacteria, which have a short doubling time and fast growth rates compared with granular anammox bacteria, can play a significant role in the nitrogen removal processes in Changi WRP (Cao et al. 2013). Quantitative polymerase chain reaction (q-PCR) tests (using modified primers from primer pair of Amx694F2/Amx960R) used to study the sludge taken from the Changi WRP found the concentration (16S rRNA) of potential anammox bacteria to be in the order of between 1.0 × 106 to 7.1 × 106 copies/mL (He 2014). The main species identified was a close relative of Candidatus Brocadia sp. 40, with 98% sequence similarity (He 2014; Lee et al. 2014) to the organism in the high rate anammox reactor at Delft University of Technology (TU Delft) (Lotti et al. 2015).

To further improve and understand the mainstream partial nitritation and anammox performance of the SFAS process, the first step is the development of a cost-effective laboratory system enabling the study of the performance of full-scale processes. The overall objectives of scale-down were to study the effects of short-term changes of hydraulic flow, an increase in carbon loading during the early phase of heavy rain period, uneven feed distribution and enriching anammox bacteria, etc. on the process performance. Given the plug flow feature of the SFAS process, a fed-batch reactor was adopted to scale down the full-scale process. This paper presents the first set of fed-batch reactor studies, which illustrates the feasibility of using a fed-batch reactor to simulate the performance of the full-scale process and the effect of short-term changes of influent BOD5/TN (COD/TN) ratio on the process performance (BOD5: 5-day biochemical oxygen demand; COD: chemical oxygen demand).

MATERIALS AND METHODS

Reactor and operation

The reactor used had a total volume of 10 L, height 29.1 cm, and a maximum liquid volume of 8.5 L. A heater was used to control the water temperature inside the reactor at 30 ± 2 °C, which was similar to that of the full-scale plant. The reactor operation was designed according to the site conditions. Corresponding to five step-feeds in the full-scale process, there were five pairs of anoxic and aerobic phases, each lasting 1.6, 1.3, 1.0, 0.8, and 0.6 h during a complete reaction cycle. The reactor was inoculated with 3.75 L of activated sludge from the RAS [0.1 mgN/L, PO4-P: 12 mgP/L, ALK: 80 mg/L as CaCO3, pH: 6.5, mixed liquor suspended solids (MLSS): 6,000–7,000 mg/L] and 1.3 L of the primary effluent. The initial MLSS concentration in the laboratory batch reactor was approximately 4,500 mg/L, which is similar to that in the first basin of the full-scale process. According to the time sequence of the site operation, 1.3 L of primary effluent was fed four times at the above-mentioned time internals. A ceramic diffuser was used for aeration, and the dissolved oxygen (DO) was controlled at 1.5–2.0 mgO2/L during the aerobic phase. When aeration was stopped the DO dropped within 1 minute to below 0.1 mgO2/L, which was lower than the DO measured at the top of the anoxic compartment at Changi WTP (0.1–0.16 mgO2/L). Samplings were at the end of each anoxic phase, which was also considered to be the beginning of the aerobic phase, as well as at the end of each aerobic phase. The volume and measured composition of both the sludge and primary effluent were used to calculate the composition at the start of the anoxic phase. In the study, the reactor fed by primary effluent (reactor 1), which had a high BOD5/TN (COD/TN) ratio, was used to simulate the site performance; while the reactor fed by final effluent (reactor 2) dosed with NH4 and alkalinity and which had a much lower BOD5/TN (COD/TN) ratio, was used to study the effect of influent BOD5/TN (COD/TN) ratio on the process performance.

Analytical methods

Standard methods according to APHA (1999) were adopted in the analyses including: NH4: ISE (ions selective electrode) (4500–NH3D); NO3: ISE (4500–NO3D); total Kjeldahl nitrogen (TKN): (4500_Norg B); COD (and sCOD): closed reflux by Hach Vial (5220D); total phosphorus (TP) and PO4: conventional colorimeter (4500–PD); alkalinity: titration (2320B) and MLSS (2540D). NO2 was analyzed according to USEPA (1987) by colorimeter (NitriVer3 Nitrite Reagent Powder Pillows from HACH) (4500–NO2B). DO and temperature were measured by portable meter (YSI 85D). pH was measured by portable meter (HQ40d with Rugged). For all the liquid composition analysis the samples were filtered (0.45 μm, Whatman) on site and sent to the laboratory for analysis within an hour.

RESULTS AND DISCUSSION

Performance of fed-batch reactor fed with primary effluent

Characteristics of primary effluent

As illustrated in Table 1, most of the parameters measured on 17 April 2014 are in the ranges of the average site values in 2014 except NH4-N and TN, which are lower than the averages. The respective influent BOD5/TN and COD/TN ratios are 3.9 and 11, in the same range (BOD5/TN: 3.9; and COD/TN: 9.5). The alkalinity is in the low range, a feature of municipal wastewater in Singapore (Cao et al. 2008).

Table 1

Parameters of the primary effluent used as influent in the fed-batch reactor 1 on 17 April 2014 and average of site primary effluent data in 2014 (all in mg/L except pH)

Parameter COD sCOD BOD5 TSS NH4-N NO2-N NO3-N TN PO4-P pH ALK (as CaCO3
Value in this test 330 182 118 106 23.4 0.1 0.1 30.0 7.9 7.0 145 
Average of the site data 2014a 365 (±36) 201 (±16) 154 (±21) 116 (±14) 33 (±3) NA NA 39 (±3) 4.1 (±0.6) 7.2 (±0.1) 171(±8) 
Parameter COD sCOD BOD5 TSS NH4-N NO2-N NO3-N TN PO4-P pH ALK (as CaCO3
Value in this test 330 182 118 106 23.4 0.1 0.1 30.0 7.9 7.0 145 
Average of the site data 2014a 365 (±36) 201 (±16) 154 (±21) 116 (±14) 33 (±3) NA NA 39 (±3) 4.1 (±0.6) 7.2 (±0.1) 171(±8) 

COD: chemical oxygen demand; sCOD: suspended chemical oxygen demand; BOD5: 5-day biochemical oxygen demand; TSS: total suspended solids.

aThe values are means ± standard deviations.

Nitrogen profile

Figure 1 demonstrates that the NH4-N concentration was increased at the start of the anoxic period, as compared with the concentration at the end of the preceding aerobic periods, due to primary effluent feeding; while at the same time, NO2-N and NO3-N were reduced due to dilution from the primary effluent feeding. The common features were: in the aerobic periods, NH4-N decreased, NO2-N produced concomitantly with NO3-N as a result of ammonium oxidation (and little assimilation); in the anoxic zones, NH4-N decreased concomitantly with reduction of NO2-N and NO3-N except in the first anoxic phase where NH4-N was increased due to dominance of ammonification, cell decay and lysis as anammox activity was due to low NO2 concentration (<1 mg NO2-N/L) in the primary effluent and RAS.
Figure 1

Nitrogen profile through batch reaction cycle. AN, anoxic phase; AE, aerobic phase. Number sequence: in, starting; out, end.

Figure 1

Nitrogen profile through batch reaction cycle. AN, anoxic phase; AE, aerobic phase. Number sequence: in, starting; out, end.

Partial nitrification and nitritation in aerobic phase

Figure 2 shows the following. (i) A significant fraction of ammonium was removed in the aerobic phases. The average percentage was 95.1%, which was higher than 72.2% of the average for the full-scale process (Cao et al. 2014a). The higher NH4-N removal efficiency is likely to be due to the lower NH4-N concentration in the feed of the test compared with the site average in 2014, and the ideal plug-flow conditions simulated in the batch reactor versus the non-ideal plug flow on site even with a large length/width ratio. (ii) The NO2-N concentration was always much higher than NO3-N at the ends of the aerobic phases. The average nitrite accumulation ratio (NAR) is 81.3%, which is slightly higher than 76.0% of the average for the full-scale plant data (Cao et al. 2014a). The results demonstrate a well-established nitrite shunt in the aerobic phase (Blackburne et al. 2008).
Figure 2

NH4 removal and NAR through batch reaction cycle (%).

Figure 2

NH4 removal and NAR through batch reaction cycle (%).

Ammonium and nitrite/nitrate removal in anoxic phase

Figure 3 presents the NH4-N mass removal in the four final anoxic phases (small contribution from the first anoxic phase). There was clear ammonium removal in the anoxic phase, pointing to the potential presence of anammox conversion. The total NH4-N removal in the anoxic phase was 45.8 mg NH4-N, [Σ(ΔNH4-N × q), q: liquor volume in the reactor] per reaction cycle through the anammox pathway that would result in a total nitrogen removal of 105 mgN according to the stoichiometric coefficient of anammox reaction (Lotti et al. 2014). Corresponding to the total nitrogen input of 169 mgN (based on the feed volume and concentration) the anammox based autotrophic nitrogen removal contribution (62% of total nitrogen input) was higher than 38%, which is the average derived from mass balances according to regular monitoring data in 2014 and the averages of nitrification and NAR from four full-scale plant sampling programs (Cao et al. 2014b). This indicates a significant contribution of nitrogen removal through anammox-based conversion. Favorable conditions in the laboratory fed-batch experiment (such as minimal DO in the anoxic zone and strict plug-flow representation) and potentially higher anammox bacteria population in the sludge sampling day, may be the main causes for the greater contribution of autotrophic denitrification measured.
Figure 3

NH4-N removal in four anoxic phases.

Figure 3

NH4-N removal in four anoxic phases.

Excessive biological phosphorus removal

Figure 4 shows the PO4-P and sCOD concentration profiles in all the individual phases. Regular patterns were observed: PO4-P increased in the anoxic phases due to PO4 release concomitantly with volatile fatty acids (VFA)-sCOD uptake. Similar to the full-scale plant results (Cao et al. 2013), PO4 was released and VFA-sCOD was taken up in the anoxic zone. The up-take of VFA-sCOD reduced the COD usage for denitritation, which favored the anammox reaction.
Figure 4

PO4-P and sCOD profiles through batch reaction cycle. AN, anoxic phase; AE, aerobic phase. Number sequence: in, starting; out, end.

Figure 4

PO4-P and sCOD profiles through batch reaction cycle. AN, anoxic phase; AE, aerobic phase. Number sequence: in, starting; out, end.

pH and alkalinity

As shown in Figure 5, apparent pH reduction was observed in all the aerobic phases due to nitrification. Corresponding alkalinity reduction was observed. pH increase was observed in all anoxic phases, likely due to denitrification and PO4-P release, but alkalinity increase was only observed during the first and second anoxic phases. The reasons for the irregular pattern of alkalinity data were not clear although the release of K+, Mg2+, Ca2+ concomitantly with PO4 in the anoxic phase (Neethling et al. 2006) causing the interaction of the different buffer systems present (bicarbonate and phosphate) may be one of the factors.
Figure 5

pH and alkalinity profile through batch reaction cycle. AN, anoxic phase; AE, aerobic phase. Number sequence: in, starting; out, end.

Figure 5

pH and alkalinity profile through batch reaction cycle. AN, anoxic phase; AE, aerobic phase. Number sequence: in, starting; out, end.

Final effluent quality

Comparing the effluent data with the average site data of 2014 (Table 2), the fed-batch test 1 achieved nearly the same excellent effluent quality in terms of nitrogen removal: the total soluble inorganic nitrogen was 4.8 mgN/L (summation of NH4-N, NO2-N, and NO3-N), which was similar to 3.8 mgN/L, the average value of 2014 site data. Similar to site data, the low NO3 and NO2 were due to conventional denitrification in addition to nitrogen short-cut and likely the anammox process during the reaction cycle. By assuming soluble non-biodegradable organic ammonium nitrogen of 0.5 mg/L in the final effluent, according to Cao et al. (2008), and ignoring the nitrogen in the solids of final effluent the total soluble nitrogen in the final effluent was 5.3 mg/L. This corresponds to a total nitrogen removal efficiency of 82% which is similar to the site data of 89% in 2014.

Table 2

Parameters of the final effluent of the fed-batch reactor 1 on 17 April 2014 and averages of the final effluent site data in 2014 (all in mg/L except pH)

Parameter sCOD TSS NH4-N NO2-N NO3-N PO4-P pH ALK (as mg CaCO3./L) 
Value in this test 45 17 0.1 3.7 1.0 2.3 6.9 34 
Average of site data in 2014a 29 (±6) 7.3 (±1.5) 2.4 (±0.6) 0.5 (±0.2) 0.9 (±0.7) 1.7 (±0.7) 7.1 (±0.1) 67 (±10) 
Parameter sCOD TSS NH4-N NO2-N NO3-N PO4-P pH ALK (as mg CaCO3./L) 
Value in this test 45 17 0.1 3.7 1.0 2.3 6.9 34 
Average of site data in 2014a 29 (±6) 7.3 (±1.5) 2.4 (±0.6) 0.5 (±0.2) 0.9 (±0.7) 1.7 (±0.7) 7.1 (±0.1) 67 (±10) 

aThe values are means ± standard deviations.

Fate of nitrogen

Given that the total soluble nitrogen in the final effluent is 5.3 mg/L and the total influent nitrogen is 30 mgN/L, the nitrogen of the final effluent equals 18% of total nitrogen input. Accounting for 18% in final effluent and 62% removal by anammox-based nitrogen removal, the remaining 20.0% of total nitrogen input was from heterotrophic denitrification, ordinary heterotrophic organism metabolism and growth. These data demonstrate that (some) anammox bacteria can coexist with denitrifiers. It has been shown before that anammox bacteria can effectively compete for VFA (formate, acetate, and propionate) as electron donor for nitrate/nitrite reduction with ordinary heterotrophic denitrifiers (Kartal et al. 2008, 2012; Winkler et al. 2012; Guillén et al. 2014).

Performance of fed-batch reactor fed with final effluent

Characteristics of final effluent

To test for nitrogen removal without the presence of the readily biodegradable COD, the reactor was fed with final effluent from Changi WRP amended with ammonium bicarbonate (NH4HCO3) to obtain 37.1 mgNH4-N/L and NaOH to increase the pH and alkalinity. The influent BOD5/TN ratio was 0.12 (COD/TN ratio: 2.1). Table 3 lists the influent composition used in this test.

Table 3

Parameters of the final effluent after NH4-N, pH, and ALK adjustment used as influent in the fed-batch reactor 2 test on 23 July 2014 (all in mg/L except pH)

Parameter COD sCOD BOD5 TSS NH4-N NO2-N NO3-N TN PO4-P pH ALK (as mg CaCO3−1
Value 87 42 26 37 0.1 0.8 41 1.9 8.7 180 
Parameter COD sCOD BOD5 TSS NH4-N NO2-N NO3-N TN PO4-P pH ALK (as mg CaCO3−1
Value 87 42 26 37 0.1 0.8 41 1.9 8.7 180 

Nitrogen profile

NH4-N was oxidized concomitantly with NO2-N and NO3-N production in all the aerobic phases (Figure 6). However, the NH4 removal percentages dropped from the third aerobic phase, where little nitrification occurred, towards the end of the reaction cycle. This low nitrification efficiency during the last two aerobic phases was observed in another fed-batch experiment with influent of a similar composition. This phenomenon is contradictory to the assumption that ammonium oxidizing bacteria (AOB) activity and rate would be enhanced under low BOD5/TN (COD/TN) ratio (de Clippeleir et al. 2013). Alkalinity at the end of several aerobic phases was approximately 20 mg/L (as CaCO3) only which is 7–10 mg/L (as CaCO3) less than those measured at the end of most aerobic phases of fed-batch reactor 1 (Figure 5) and about 30 mg/L (as CaCO3) less than the Changi WRP site data (Cao et al. 2013). Low alkalinity appears to be one of the reasons but more studies are needed. In the anoxic phase, NH4-N reduced concomitantly with reduction of NO2-N and NO3-N except in the fifth anoxic phase where no apparent NO2-N reduction occurred.
Figure 6

Nitrogen profile through batch reaction cycle. AN, anoxic phase; AE, aerobic phase. Number sequence: in, starting; out, end.

Figure 6

Nitrogen profile through batch reaction cycle. AN, anoxic phase; AE, aerobic phase. Number sequence: in, starting; out, end.

Partial nitrification and nitritation in aerobic phase

Results from calculations presented in Figure 7 indicate the following: (i) 80% of ammonium was removed in the aerobic phases, which was higher than the average of 72% for the full-scale process (Cao et al. 2014a), but lower than the average of 95% of fed-batch reactor 1 test. (ii) The NO2-N concentration was always much higher than NO3-N at the end of the aerobic phases. The average NAR was 65%, lower than the 76% average for the full-scale plant data (Cao et al. 2014a) and 81% for the fed-batch reactor 1. This low NAR obtained under low influent BOD5/TN (COD/TN) ratio appears to be contradictory to de Clippeleir et al. (2013), probably owing to low alkalinity and inorganic carbon limitation on nitrification (Tokutomi et al. 2006), and short-term change of the BOD5/TN (COD/TN) ratio. However, the results demonstrate that nitrite shunt in the aerobic phase was maintained.
Figure 7

NH4 removal and nitrite accumulation ratios (NAR) through batch reaction cycle (%).

Figure 7

NH4 removal and nitrite accumulation ratios (NAR) through batch reaction cycle (%).

Ammonium and nitrite/nitrate removal in anoxic phase

The total NH4-N removal in the anoxic phase was 31 mgNH4-N [Σ(ΔNH4-N × q)] per reaction cycle (Figure 8). Nitrogen removal through the autotrophic path was 73 mgN according to the stoichiometric coefficient of anammox reaction (Lotti et al. 2014). Corresponding to the total nitrogen input of 233 mgN (based on the feed volume and measured concentration of the final effluent) the contribution of anammox based nitrogen removal was 31% of total nitrogen input neglecting the small amount of NH4 used for cell synthesis. This was less than 56% removal for the fed-batch reactor 1. The activity of anammox bacteria in the fed-batch reactor 2 was not as high as in the fed-batch reactor 1. This was perhaps due to less abundance or activity of anammox population in sludge of reactor 2 as compared to that of reactor 1 as the experiments were not conducted on the same days or because anoxic conditions were less favorable due to the absence of readily biodegradable COD. However, contribution of anammox based nitrogen removal (31%) is close to the average of 38% derived from mass balance according to the regular monitoring data in 2014 and the averages of nitrification and NAR from four full-scale plant sampling programs in Changi WRP (Cao et al. 2014a). Again, this strongly points to the presence of anammox based N-removal.
Figure 8

NH4-N removal in anoxic phase.

Figure 8

NH4-N removal in anoxic phase.

PO4-P

Little biodegradable COD, VFA and low concentration PO4-P (1.9 mgPO4-P/L. Table 3) were available in the final effluent and RAS. This led to no apparent P release in the anoxic phases and P uptake in the aerobic phases indicating limited PAO activity in this test.

pH and alkalinity

Figure 9 shows pH reduction in the aerobic phases due to nitrification except for the fourth aerobic phase. Corresponding alkalinity reduction was observed. However, the alkalinity at the end of aerobic phases for fed-batch reactor 2 was about 7–10 mg/L (as CaCO3) less than that of the fed-batch reactor 1, although the alkalinity in the feed (180 mg CaCO3/L, Table 3) was higher than that of the fed-batch reactor 1 (145 mg CaCO3/L, Table 1). This pH increase in all anoxic phases proves the fact, i.e., NH4-N removal through anammox pathway and the presence of anammox process on site considering that: (i) this pH increase was not due to heterotrophic denitrification unlike reactor 1, little readily biodegradable COD was available in reactor 2; and (ii) the NH4-N removal under low DO due to simultaneous nitrification and denitrification (SND) could be excluded as SND would cause pH reduction. However, alkalinity presented irregulars pattern as in reactor 1.
Figure 9

pH and alkalinity through reaction cycle. AN, anoxic phase; AE, aerobic phase. Number sequence: in, starting; out, end.

Figure 9

pH and alkalinity through reaction cycle. AN, anoxic phase; AE, aerobic phase. Number sequence: in, starting; out, end.

Final effluent quality

Given the total soluble N in the effluent (Table 4) and by assuming 0.5 mg/L soluble non-biodegradable organic ammonium nitrogen in the final effluent according to Cao et al. (2008), the total soluble nitrogen in reactor 2 test is 27 mgN/L. It is much higher than the 5.3 mgN/L in reactor 1 and the site average value of 4.3 mgN/L at Changi WRP in 2014. This high nitrogen was attributed to: (i) high residual NH4-N of 11 mgN/L versus 0.1 mgN/L of reactor 1 and 2.4 mgN/L of the site data; and (ii) high NO2-N of 11 mgN/L versus 3.7 mgN/L of reactor 1 and 0.5 mgN/L of the site data and high NO3-N of 5.1 mgN/L versus 1.0 mgN/L of reactor 1 and 0.9 mgN/L of the site data. Low heterotrophic denitrification efficiency due to lack of carbon could be the main cause. The total nitrogen removal efficiency was 34%. The ratio of ΔNO3-N produced/ΔNH4-N removed was 16%. Given little biodegradable COD in the feed and released nitrogen from cell decay, this fed-batch reactor 2 test clearly demonstrates the functions of heterotrophic denitrification and anammox process in nitrogen conversion.

Table 4

Effluent parameters of the fed-batch reactor test 2 on 23 July 2014 (all in mg/L except pH)

Parameter sCOD TSS NH4-N NO2-N NO3-N PO4-P pH ALK (as mgCaCO3/L) 
Value 39 NM* 11 11 5.1 6.7 33 
Parameter sCOD TSS NH4-N NO2-N NO3-N PO4-P pH ALK (as mgCaCO3/L) 
Value 39 NM* 11 11 5.1 6.7 33 

*Not measured.

Fate of nitrogen

Given the nitrogen in the final effluent of 27 mgN/L and the total influent nitrogen of 41 mgN/L, the nitrogen of the final effluent contributed to 66% of total nitrogen input. Accounting for 65.8% in final effluent and 31% by anammox-based nitrogen removal, 3.0% of total nitrogen input was contributed for autotrophic growth ignoring little heterotrophic denitrification.

Comparisons of fed-batch reactors 1 and 2 and site performance

Data in Table 5 indicate that high nitrogen removal was achieved in the fed-batch test 1 with high BOD5/TN (COD/TN) influent ratio, which was similar to the site data. Nitrogen removal contribution through the autotrophic path in the fed-batch test 2 was close to the average of the site data. The contribution of heterotrophic denitrification and growth of fed-batch reactor 1 was less than the site average due to its higher contribution of anammox-based autotrophic removal; while fed-batch reactor 2 had the least (3%) contribution since there was little contribution from heterotrophic denitrification. Three sets of data demonstrate the significant contribution of the anammox autotrophic nitrogen path under different influent BOD5/TN (COD/TN) ratio conditions. The data also illustrate that high BOD5/TN (3.9) (COD/TN ratio: 9.5–11) enhances total nitrogen removal efficiency through conventional denitrification, which was similar to that reported by Han et al. (2014), although that presented here was observed under short-term conditions.

Table 5

Comparisons of autotrophic N removal contribution, nitrogen distributions, and total N removal efficiency of two fed-batch tests and site data in 2014

Parameter BOD5/TN (COD/TN) Anammox-based N removal contribution, % Heterotrophic denitrification and growth contributions, % Nitrogen in the final effluent % N removal efficiency, % 
Reactor 1 3.9 (11) 62 20 18 82 
Reactor 2 0.12 (2.1) 31 3.0 66 34 
Site data* 3.9 (9.5) 38* 50 12 89 
Parameter BOD5/TN (COD/TN) Anammox-based N removal contribution, % Heterotrophic denitrification and growth contributions, % Nitrogen in the final effluent % N removal efficiency, % 
Reactor 1 3.9 (11) 62 20 18 82 
Reactor 2 0.12 (2.1) 31 3.0 66 34 
Site data* 3.9 (9.5) 38* 50 12 89 

CONCLUSIONS

The similarity of nitrogen, phosphorus, and pH/alkalinity profiling between the full-scale process in the Changi WRP and the laboratory fed-batch tests fed by primary effluent illustrates that a laboratory fed-batch reactor is able to simulate full-scale performance. It provides a cost-effective tool to investigate the effects of some short-term perturbations in operating conditions, such as variations of feed flow and characteristics, DO in aerobic/anoxic zones, primary effluent feed distribution, etc. on the performance of the process. The performance of the reactor fed by primary effluent illustrated the existence of anammox and heterotrophic denitrification and apparent excessive biological phosphorus removal as observed from the site. The performance of the reactor fed by final effluent proved the presence of the anammox process on site. The results of two laboratory fed-batch tests illustrate the effect of influent BOD5/TN (COD/TN) ratio on the effluent nitrogen quality: nitrogen removal efficiency was enhanced by a high ratio of BOD5/TN (COD/TN) due to conventional denitrification.

REFERENCES

REFERENCES
APHA
1999
Standard Methods for the Examination of Water and Wastewater
.
20th edition, American Public Health Association
,
Washington, DC, USA
.
Cao
Y. S.
Ang
C. M.
Raajeevan
K. S.
Wah
Y. L.
(eds)
2008
Performance of the full-scale activated sludge process
. In:
Biological Nitrogen Removal Activated Sludge Process in Warm Climates: Full-scale Process Investigation, Scaled-down Laboratory Experimentation and Mathematical Modelling
.
IWA Publishing
,
London
,
UK
.
Cao
Y. S.
Kwok
B. H.
Yong
W. H.
Chua
S. C.
Wah
Y. L.
Yahya
A. G.
2013
The main stream autotrophic nitrogen removal in the largest full scale activated sludge process in Singapore: Process analysis
. In:
Proceedings of WEF/IWA Nutrient Removal and Recovery 2013: Trends in Resource Recovery and Use
,
July 28–31
,
Vancouver, BC
,
Canada
.
Cao
Y. S.
Kwok
B. H.
Noraini
A. Z.
Lau
C. L.
Zulkifli
I.
Chua
S.,C.
Wah
Y. L.
Yahya
A. G.
2014a
The mainstream partial nitritation-anammox nitrogen removal in the largest activated sludge process and comparisons with other BNR activated sludge process in Singapore
. In:
Proceedings of IWA World Water Congress
,
September 21–26
,
Lisbon
,
Portugal
.
Cao
Y.
Kwok
B. H.
Zhou
Y.
Lee
Z.
Liu
Y.
He
J.
van Loosdrecht
M. C. M.
Daigger
G. T.
Lay
W.
Chua
S. C.
Wah
Y. L.
Ghani
Y. A.
2014b
Activated sludge nitrogen removal in warm climates: from conventional to innovative processes
. In:
Proceedings of IWA Global Challenges: Sustainable Wastewater Treatment and Resource Recovery
,
October 26–30
,
Kathmandu
,
Nepal
.
Daigger
G. T.
Nicholson
G. A.
Koh
C. L. Y.
Moh
W. H.
Young
J. C.
Ghani
Y. A.
Yong
W. H.
2008
Start-up and initial operation of Singapore's 800,000 m3/day Changi Water Reclamation Plant
. In:
Proceedings of IWA–PUB Water Convention Conference
,
June 23–27
,
SIWW
,
Singapore
.
de Clippeleir
H.
Vlaeminck
S. E.
de Wilde
F.
Daeninck
K.
Mosquera
M.
Boeckx
P.
Verstraete
W.
Boon
N.
2013
One-stage partial nitritation/anammox at 15 °C on pretreated sewage: feasibility demonstration at lab-scale
.
Appl. Microbiol. Biotechnol.
97
(
23
),
10199
10210
.
Guillén
J. A. S.
Yimman
Y.
Vazquez
C. M. L.
Brdjanovic
D.
van Lier
J. B.
2014
Effects of organic carbon source, chemical oxygen demand/N ratio and temperature on autotrophic nitrogen removal
.
Water Sci. Technol.
69
(
10
),
2079
2086
.
Han
M. H.
de Clippeleir
H.
Regmi
P.
Al-Omari
A.
Wett
B.
Vlaeminck
S. E.
Bott
C.
Murthy
S.
2014
Carbon-nitrogen-energy nexus governs technology decision for resource efficient sewage treatment
. In:
Proceedings of IWA Global Challenges: Sustainable Wastewater Treatment and Resource Recovery
,
October 26–30
,
Kathmandu
,
Nepal
.
He
J.
2014
q-PCR Quantification of Anammox Abundance of activated sludge from Changi Water Reclamation Plant. Internal report
.
Kartal
B.
van Niftrik
L.
Rattray
J.
van de Vossenberg
J. L.
Schmid
M. C.
Sinninghe Damst
J.
Jetten
M. S.
Strous
M.
2008
Candidatus ‘Brocadia fulgida’: an autofluorescent anaerobic ammonium oxidizing bacterium
.
FEMS Microbiol. Ecol.
63
(
1
),
46
55
.
Kartal
B.
van Niftrik
L.
Keltjens
J. T.
Op den Camp
H. J.
Jetten
M. S.
2012
Anammox--growth physiology, cell biology, and metabolism
.
Adv. Microb. Physiol.
60
,
211
262
.
Lee
Z.
Qin
Y.
de Araujo
C.
Zhou
Y.
Liu
Y.
2014
Preliminary Analysis of Microbial Community in Changi Water Reclamation Plant. Internal report
.
Lotti
T.
Kleerebezem
R.
Lubello
C.
van Loosdrecht
M. C. M.
2014
Physiological and kinetic characterization of a suspended cell anammox culture
.
Water Res.
60
,
1
14
. DOI:10.1016/j.watres.2014.04.017.
Lotti
T.
Kleerebezem
R.
Abelleira-Pereira
J.
Abbas
B.
van Loosdrecht
M. C. M.
2015
Faster through training: the anammox case
.
Water Res.
81
,
261
268
.
Neethling
J. B.
Bakke
B.
Benisch
M.
2006
Factors Influencing the Reliability of Enhanced Biological Phosphorus Removal (WERF Research Report Series)
.
WERF/ IWA Publishing
,
London
,
UK
.
Ou
G. J.
Kwok
B. H.
Zhang
Y. N.
Cao
Y. S.
2012
Excellent nitrogen removal performance of a full-scale step-feed activated sludge process of municipal wastewater treatment in Singapore
. In:
Proceedings of Water Convention
,
July 1–4, 2012
,
SIWW
,
Singapore
.
Tokutomi
T.
Kiyokawa
T.
Shibayarna
C.
Harada
H.
Ohashi
A.
2006
Effect of inorganic carbon on nitrite accumulation in an aerobic granule reactor
.
Water Sci. Technol.
53
(
12
),
285
294
.
USEPA
1987
Standard Methods for the Examination of Water. EPA 520/5-84-006
,
United States Environmental Protection Agency
,
NC, USA
.