Influence of wastewater composition on nutrient removal behaviors in the new anaerobic–anoxic/nitrifying/induced crystallization process

Abstract

In this study, the new anaerobic–anoxic/nitrifying/induced crystallization (A₂N–IC) system was compared with anaerobic-anoxic/nitrifying (A₂N) process to investigate nutrient removal performance under different influent COD and ammonia concentrations. Ammonia and COD removal rates were very stable in both processes, which were maintained at 84.9% and 86.6% when the influent ammonia varied from 30 mg L⁻¹ to 45 mg L⁻¹ and COD ranged from 250 mg L⁻¹ to 300 mg L⁻¹. The effluent phosphorus always maintained below 0.2 mg L⁻¹ in A₂N–IC, whereas in A₂N the effluent phosphorus concentration was 0.4–1.7 mg L⁻¹, demonstrating that A₂N–IC is suitable to apply in a broader influent COD and ammonia concentration range. Under higher influent COD (300 mg L⁻¹) or lower ammonia conditions (30 mg L⁻¹), the main function of chemical induced crystallization was to coordinate better nutrient ratio for anoxic phosphorus uptake, whereas under high phosphorus concentration, it was to reduce phosphorus loading for biological system. Under the similar influent wastewater compositions, phosphorus release amounts were always lower in A₂N–IC. To clarify the decrease procedure of phosphorus release in the A₂N–IC, the equilibrium between chemical phosphorus removal and biological phosphorus removal in A₂N–IC was analyzed by mass balance equations. During the long-term experiment, some undesirable phenomena were observed: the declining nitrification in post-aerobic tank and calcium phosphorus precipitation in the anaerobic tank. The reasons were analyzed; furthermore, the corresponding improvements were proposed. Nitrification effect could be enhanced in the post-aerobic tank, therefore ammonia removal rate could be increased; and biologically induced phosphorus precipitation could be inhibited by controlling pH at the anaerobic stage, so the phosphorus release and recovery could be improved.

1. Introduction

Eutrophication has become a worldwide and serious problem due to the high levels of phosphorus plus nitrogen in rivers and lakes. The biological nutrient removal system is considered as the economical and sustainable method for phosphorus and nitrogen removal from wastewater. Conventional biological nutrient removal processes, such as anaerobic–anoxic–aerobic (A²O) process, were widely adopted in many municipal wastewater treatment plants. However, chemical oxygen demand (COD) was an essential limiting factor for denitrification and phosphorus release (Ekama and Wentzel, 1999; Wu et al., 2011). Furthermore, there was sludge retention time (SRT) conflict between phosphorus removal organisms (PAOs) and nitrifiers. The anaerobic–anoxic/nitrifying (A₂N) process by utilizing denitrifying phosphorus removal organisms (DNPAOs) could solve the problem (Kuba et al., 1996b), but at different influent wastewater compositions, it could not maintain stable and high phosphorus removal efficiency (Bortone et al., 1996; Kapagiannidis et al., 2012; Wang et al., 2009).

Although excess phosphorus into water bodies caused serious eutrophication problems, phosphorus is an indispensable resource (Van Vuuren et al., 2010), and the phosphorus reserves base would be depleted in the next 100 years (Herring and Fantel, 1993). Therefore, phosphorus recovery from wastewater has been received increasing interest. Calcium phosphorus crystallization process was suitable for phosphorus recovery from domestic wastewater since a broad range of phosphorus concentration is acceptable in the process (Chen et al., 2009; Kim et al., 2006; Song et al., 2007; Song et al., 2006). HAP formation is based on the following reaction.

$$5{\text{Ca}}^{2+}+3{\text{PO}}_4^{3-}+{\text{OH}}^{-}\to {\text{Ca}}_5{({\text{PO}}_4)}_3\text{OH}$$

In order to attain the phosphorus effluent standards and recovering phosphorus simultaneously, in our previous study the incorporated HAP crystallization plus A₂N process, which was named the anaerobic–anoxic/nitrifying/induced crystallization (A₂N–IC) process, was proposed (Shi et al., 2012). The previous study demonstrated satisfactory nutrient removal and phosphorus recovery performance under different feeding phosphorus concentrations. Due to the addition of chemical phosphorus removal pathway, it was easy to understand the better behavior in A₂N–IC under high influent phosphorus loading.

In our previous study, feeding COD and ammonia concentrations were fixed (Shi et al., 2012). However, the influent COD and ammonia were closely related to the electron donors and acceptors for anoxic phosphorus uptake by DNPAOs (Wang et al., 2009). When the influent COD and ammonia concentration were varied, it is still unknown whether A₂N–IC could maintain the stable nitrogen and phosphorus removal, which is very important to its development and application in practical wastewater treatment.

In this study, the characteristics of A₂N–IC on nitrogen and phosphorus removal under the varied COD and ammonia concentrations were investigated. Since the influent phosphorus concentrations were all low, we focused much more on the comparison of removal behaviors in the two processes rather than the recovered product. Moreover, during the long experiment period, new phenomena were found out, and A₂N–IC could be further improved in some aspects. Therefore, the aims of the study were as follows: (1) to evaluate the impacts of influent COD and ammonia concentrations on A₂N–IC nutrient removal efficiencies; (2) to analyze the different functions of chemical induced crystallization under different wastewater compositions; (1) to explain the reasons for the undesirable phenomena and propose improvement approaches for A₂N–IC process; (4) to clarify the equilibrium between chemical phosphorus removal and biological phosphorus removal in A₂N–IC.

2. Materials and methods

2.1. A₂N and A₂N–IC processes set up

Fig. 1 was the configuration of the A₂N–IC process. The section in the dashed box was the IC system, and the other part of the process was A₂N system. Phosphorus was released by DNPAOs in the anaerobic tank, and then was taken up in the anoxic and post-aerobic tank. Ammonia was nitrified in the aerobic tank to nitrate, which served as electron acceptors in the anoxic tank. Phosphorus-rich supernatant from settler I was partly led into the chemical crystallization process for phosphorus recovery. Calcite was used as the seed for crystallization.

Figure 1.

Configuration of A₂N–IC process.

The activated sludge taken from oxidation ditch process in Jiangning Sewage Treatment Plant (Nanjing, China) was acclimated about 90 days in the sequencing batch reactors before transferred into continuous-flow A₂N process. The experiment in this study lasted about 270 days. The flow rates of influent, DNPAOs sludge stream, return sludge stream, nitrifiers sludge stream and reagent tank effluent were 25 L/h, 5 L/h, 5 L/h, 5 L/h and 0.47 L/h, respectively. SRTs for DNPAOs and nitrifiers were 15 days and 25 days. Overall hydraulic retention time (HRT) was about 23 h, and mixed liquid suspended solid (MLSS) was maintained at 3600–4300 mg L⁻¹. The aeration flows for aerobic tank, post-aerobic tank, stripping tank and crystallization reactor were 320 L/h, 250 L/h, 100 L/h and 350 L/h.

2.2. Experimental design

The experiment was divided into six runs. For comparison, in both processes there were three groups of influent COD and ammonia compositions (Table 1). In each run, COD concentration was about either 250 mg L⁻¹ or 300 mg L⁻¹, and ammonia concentration was about either 30 mg L⁻¹ or 45 mg L⁻¹. Total phosphorus (TP) was maintained at about 5.5 mg L⁻¹. Each run was operated in a length of more than two folds of SRT (Ma et al., 2005) before sampling. Between run 3 and run 4, run 1 was repeated for 30 days to supply the same initial condition for the two processes. The calcium dosage in the reagent tank was maintained at 158–187 mg L⁻¹. Calcite seeds were 50 g in each run. Heaters were used to keep the wastewater temperature at about 20 °C.

Table 1.

Influent wastewater compositions in each run.

Run	Process	Average COD/P	Average COD/N	COD/(mg L−1)	${\text{NH}}_4^{+}''–''\text{N}/(\text{mg}{\text{L}}^{-1})$	TP/(mg L−1)
1	A2N	44.7	5.6	250.3 ± 11.9	44.4 ± 2.1	5.6 ± 0.3
2	A2N	52.6	6.8	304.9 ± 13.8	44.7 ± 2.1	5.8 ± 0.4
3	A2N	47.0	8.0	244.6 ± 15.5	30.3 ± 1.3	5.2 ± 0.4
4	A2N–IC	42.1	6.0	240.2 ± 25.1	39.8 ± 3.5	5.7 ± 0.5
5	A2N–IC	55.0	7.2	313.7 ± 14.8	43.5 ± 2.1	5.7 ± 0.4
6	A2N–IC	47.7	8.8	252.6 ± 21.2	28.6 ± 1.6	5.3 ± 0.5

2.3. Analytical methods

According to standard methods (APHA, 1998), $\text{COD,TP,}{\text{NO}}_3^{-}-\text{N,}{\text{NH}}_4{|}^{+}-\text{N}$ and MLSS were analyzed. Calcium was measured by an atomic absorption spectrometer (AAS400, PE, USA). X-ray diffraction (XRD) (X’TRA, Switzerland) was used to identify the substance formed on the anaerobic tank wall.

The saturation index (SI) value and the speciation of different ions in the anaerobic tank were calculated by PHREEQC program. Mass balance equations were utilized to describe the equilibrium procedure between chemical phosphorus removal and biological phosphorus removal from unstable situation to stable condition in A₂N–IC.

DNPAOs were analyzed. Total flora concentration determination, isolation and screening methods were in accordance with the methods reported by Merzouki et al. (1999). Biolog MicroPlates and PCR analysis were used for identification, following the established Biolog procedure (Biolog, 1993) and the published method (Marchesi et al., 1998) respectively.

3. Results

3.1. Identification of the isolates

The total number of bacteria in the DNPAOs sludge was 9 × 10⁷ mL⁻¹. By screening, seven isolates obtained, which were able to denitrify and to accumulate phosphate simultaneously. Five of them were identified as Sphingomonas paucimobilis B (99% identity), Pseudomonas citronellolis (92% identity), P. citronellolis (98% identity), Pseudomonas nitroreducens/azelaica (100% identity) and P. citronellolis (98% identity). The other two isolates could not be identified by Biolog. They were determined by PCR analysis as species of Acinetobacter. Species belonging to the genera Pseudomonas were obtained in the denitrifying phosphorus removal system, which was in agreement with the findings in some other papers (Bao et al., 2007; Merzouki et al., 1999).

3.2. COD and ammonia removal performances under varied influent COD and ammonia concentrations

In the two processes, when the influent ammonia varied from 30 mg L⁻¹ to 45 mg L⁻¹ and COD ranged from 250 mg L⁻¹ to 300 mg L⁻¹, the average COD removal rate was 86.6% (Fig. 2), regardless of influent wastewater composition. The COD removal performance was similar with the results in our previous study (Shi et al., 2012). As shown in Fig. 3, ammonia removal efficiency was maintained at 84.9 ± 3.9%, indicating very stable nitrogen removal rate in both processes. At lower influent ammonia concentration (30 mg L⁻¹), the effluent ammonia always reached the Chinese Discharging Standard for Urban Wastewater Treatment Plants (GB18918-2002) Class I level (below 8.0 mg/L, red straight line in Fig. 3). However, when the influent ammonia was high (45 mg L⁻¹), it was a little above 8.0 mg/L in both processes in several days. Therefore, further improvement on ammonia removal was expected.

Figure 2.

Variations of COD concentration in each run.

Figure 3.

Variations of ammonia concentration in A₂N and A₂N–IC: (a) run 1; (b) run 2; (c) run 3; (d) run 4; (e) run 5 and (f) run 6.

3.3. Phosphorus removal efficiency under varied influent COD and ammonia concentrations

As shown in Fig. 4, A₂N–IC showed better phosphorus removal performance in comparison with A₂N. The effluent phosphorus concentration was always below 0.2 mg L⁻¹ in A₂N–IC, whereas in A₂N the effluent phosphorus concentration was 0.4–1.7 mg L⁻¹. The results demonstrated that A₂N–IC could be applied in a broader influent COD and ammonia concentration range.

Figure 4.

Variations of phosphorus concentration in A₂N and A₂N–IC: (a) run 1; (b) run 2; (c) run 3; (d) run 4; (e) run 5 and (f) run 6.

The percentages of chemical phosphorus removal to total phosphorus removal in run 4, 5 and 6 were 12.6 ± 4.0%, 13.1 ± 2.6%, and 9.2 ± 2.3%, decreasing with the reduction of phosphorus concentration in the anaerobic supernatant. The previous study reported the opposite behavior (Shi et al., 2012) due to the increasing influent phosphorus concentration, but in this experiment, the feeding phosphorus concentration was maintained at about 5.5 mg L⁻¹. The average percentage of chemical phosphorus removal to total phosphorus removal was only about 11.5%, so phosphorus was removed mainly by DNPAOs.

4. Discussion

4.1. The approaches to further improvement on ammonia removal

In both processes, a portion of ammonia with the DNPAOs sludge stream was directly flowed into the anoxic tank, without nitrification and denitrification, so ammonia increased in the anoxic tank. This resulted in the unsatisfactory effluent ammonia under high feeding ammonia concentration. Some reports have found that a portion of ammonia could be nitrified in the post-aerobic tank (Kapagiannidis et al., 2011; Wang et al., 2009). Consequently, the effluent ammonia could achieve the standard. However, in some other studies, no obvious decrease of ammonia was observed at the post-aerobic stage (Li et al., 2006; Torrico et al., 2006). Both phenomena occurred during our experiment. At the start-up of the continuous-flow A₂N process, ammonia was reduced from 4–8 mg L⁻¹ to 2–4 mg L⁻¹ in the post-aerobic tank. However, it lasted only about 5 months and the nitrification function in the post-aerobic tank declined gradually. Considering the longer SRT for nitrifiers, the biofilm mediums were filled in the post-aerobic tank to enrich the nitrifiers, but ammonia removal effect was not improved after 90 days’ accumulation (Fig. 5). One reason was that aerobic mass fraction (Hu et al., 2003) was only about 19%, therefore it was difficult for nitrifiers to grow in the DNPAOs sludge system. Furthermore, in the A₂N process which was used the fixed-biofilm for nitrification (Bortone et al., 1996; Wang et al., 2007; Wang et al., 2004), the detached biofilm from aeration zone could flow into the post-aerobic tank and seed the nitrifiers. Therefore, nitrifiers could find a niche in the DNPAOs sludge system. While in our experiment, the nitrifiers sludge was almost completely separated from the DNPAOs sludge by settler II.

Figure 5.

Biofilm medium in the post-aerobic tank after 90 days’ accumulation.

Consequently, according to the above reasons the nitrification effect in the post-aerobic tank could be improved by the following methods: (a) the post-aerobic tank could be enlarged until the aerobic mass fraction was more than 30%, so nitrifiers would be sustained in the suspended medium liquor (Hu et al., 2003); (b) the suspended biomass nitrification stage could be changed into the fixed-biofilm aerobic stage. It would enhance the nitrification effect in the post-aerobic tank, as well as saving the land occupation and costs. Meanwhile, there were other methods to improve the effluent ammonia removal rates; (c) the effluent could be partly returned to the aerobic tank. It would not only improve the ammonia removal efficiency but also supply more electron acceptors for anoxic phosphorus uptake and (d) the bypass DNPAOs sludge flowed into the anoxic tank could be decreased.

4.2. Effect of varied influent COD and ammonia on denitrifying phosphorus removal

The phosphorus release to COD uptake (P_rel/COD_upt) ratio at the anaerobic stage is a typical parameter for the DNPAOs activity determination (Acevedo et al., 2012; Gu et al., 2008). As listed in Table 2, in A₂N–IC, P_rel/COD_upt ratios of run 5 and run 6 were lower than the ratio of run 4, which indicated that lower ammonia or higher COD influent was not beneficial to DNPAOs. Similar phenomenon was observed in A₂N.

Table 2.

The P_rel/COD_upt ratios in the A₂N and A₂N–IC processes.

Run	Process	Prel/CODupt ratios/(mg P mg−1COD)
1	A2N	0.25 ± 0.02
2	A2N	0.18 ± 0.01
3	A2N	0.20 ± 0.02
4	A2N–IC	0.16 ± 0.03
5	A2N–IC	0.12 ± 0.02
6	A2N–IC	0.11 ± 0.01

For run 3 and 6, the lower influent ammonia resulted in the lower nitrate after nitrification, which was insufficient for anoxic phosphorus uptake. Due to the inadequate electron acceptors, the polyphosphate (poly-P) storage was reduced (Kuba et al., 1996b; Murnleitner et al., 1997), which subsequently caused a negative effect on phosphorus release in next cycle. Consequently, P_rel/COD_upt ratio under anaerobic stage decreased. The anoxic phosphorus uptake amount and the percentage of anoxic phosphorus uptake to total phosphorus uptake were 33.6 ± 1.2 mg L⁻¹ and 85.9 ± 2.8% in run 1, whereas in run 3 they dropped to 23.0 ± 0.9 mg L⁻¹ and 73.4 ± 1.5%, which proved the analysis above.

Although the influent ammonia concentrations were nearly the same in run 4 and run 5, with the increased COD, the P_rel/COD_upt ratio removals exhibited a downward trend. A₂N had the similar phenomenon by comparing run 1 with run 2. Under higher feeding COD concentration, DNPAOs sludge steam flowed from settler I to anoxic tank with more external carbon sources. External carbon source would benefit denitrification by ordinary heterotrophic organisms (OHOs) but not for anoxic phosphorus uptake, therefore OHOs were more competitive than DNPAOs to consume nitrate, which weakened phosphorus uptake by DNPAOs (Hu et al., 2002; Zhang et al., 2010).

When influent wastewater compositions were nearly the same for the two processes, P_rel was obviously lower in A₂N–IC (the reason was explained in Section 4.4). Consequently, in the anoxic tank, the initial phosphorus to nitrate (${\text{PO}}_4^{3-}/{\text{NO}}_3^{-}$) ratios for anoxic phosphorus uptake decreased in A₂N–IC in comparison with the A₂N process (Table 3). It indicated the reduced requirements of electron acceptors for anoxic phosphorus uptake. In A₂N–IC, ${\text{PO}}_4^{3-}/{\text{NO}}_3^{-}$ ratios were 0.80–0.91, which were very similar with optimal result from Zhou et al. (2010) As compared with A₂N, in A₂N–IC the anoxic phosphorus uptake to the consumed nitrate ($\mathrm{\Delta }{\text{PO}}_4^{3-}/\mathrm{\Delta }{\text{NO}}_3^{-}$) ratios were much closer to ${\text{PO}}_4^{3-}/{\text{NO}}_3^{-}$ ratios, indicating better simultaneous phosphorus and nitrogen removal efficiencies. Meanwhile, the percentages of anoxic phosphorus uptake to total phosphorus uptake were increased in the A₂N–IC process (Table 3), due to the better nutrient ratio for anoxic phosphorus uptake by the incorporation of chemical system.

Table 3.

${\text{PO}}_4^{3-}/{\text{NO}}_3^{-}\text{,}\mathrm{\Delta }{\text{PO}}_4^{3-}/\mathrm{\Delta }{\text{NO}}_3^{-}$ and anoxic phosphorus uptake to total phosphorus uptake percentage in two processes.

Run	Process	${\text{PO}}_4^{3-}/{\text{NO}}_3^{-a)}/(\text{mg}\text{P}{\text{mg}}^{-1}\text{N})$	$\mathrm{\Delta }{\text{PO}}_4^{3-}/\mathrm{\Delta }{\text{NO}}_3^{-b)}/(\text{mg}\text{P}{\text{mg}}^{-1}\text{N})$	Anoxic phosphorus uptake to total phosphorus uptake percentage/%
1	A2N	1.19 ± 0.07	1.02 ± 0.05	85.9 ± 2.8
2	A2N	1.11 ± 0.05	0.81 ± 0.06	75.8 ± 2.4
3	A2N	1.41 ± 0.08	0.99 ± 0.04	73.4 ± 1.5
4	A2N–IC	0.91 ± 0.11	0.85 ± 0.08	93.8 ± 3.1
5	A2N–IC	0.80 ± 0.05	0.74 ± 0.05	92.5 ± 2.6
6	A2N–IC	0.91 ± 0.14	0.79 ± 0.14	86.5 ± 3.5

It is noteworthy that under different influent wastewater compositions, A₂N–IC process always showed satisfactory TP removal, but the main function of chemical induced crystallization was distinct. Under high feeding phosphorus concentration, chemical system reduced the phosphorus loading for the biological section, so further increase of feeding phosphorus could be removed by increasing chemical phosphorus removal amount, without affecting biological section. However, under low phosphorus, the main function of chemical section was to adjust the proper nutrient ratio for anoxic phosphorus uptake. The adjustment ability was limited because further increase of influent COD or decrease of influent ammonia could have negative effect on biological section, which could not be solved by increasing chemical phosphorus removal.

4.3. The approach to further improvement on phosphorus recovery

The P_rel/COD_upt ratios at anaerobic stage in six runs were generally lower than some published results (Oehmen et al., 2007; Wang et al., 2011). Meanwhile, a decrease of calcium was observed in the anaerobic tank. Therefore, calcium phosphorus precipitation was suspected to occur. As listed in Table 4, SI values verified the possibility. After more than 270 days operation, a new substance was formed (Fig. 6) on the anaerobic tank wall below the wastewater surface. XRD analysis confirmed that the substance was hydroxyapatite (Fig. 7) or its precursor. This phenomenon was similar with the result obtained by Angela et al. (2011) in an anaerobic/aerobic granular sludge process.

Table 4.

SI values with respect to the possible precipitations in the anaerobic tank of stage 5.

Phase	SI value
Ca3(PO4)2 (ACP)	0.26
CaCO3 (Calcite)	0.22
CaHPO4 (DCPC)	−0.44
CaMg(CO3)2 (Dolomite)	0.64
Ca5 (PO4)3OH (HAP)	11.39
MgNH4PO4 (MAP)	−0.60

Figure 6.

Substance formed on the anaerobic tank wall.

Figure 7.

XRD analysis of the substance formed on the anaerobic tank wall.

The calcium phosphorus precipitation also achieved the aim of phosphorus recovery. However, if A₂N–IC could apply in real wastewater treatment plants, it would be difficult to collect the recovery product in the anaerobic tank. Furthermore, it was not beneficial to chemical induced crystallization, because lower phosphorus and calcium concentration could decrease the recovery efficiency in the induced crystallization reactor.

In our experiment, pH increased to 8.0–8.2 in the anaerobic tank, which was corresponded with several published results (Jeon et al., 2001; Pereira et al., 1996), although some previous studies reported the decrease of pH under anaerobic stage (Spagni et al., 2001; Zeng et al., 2008). Sodium acetate was used as the carbon source in the artificial wastewater, so volatile fatty acid uptake by DNPAOs or GAOs was probably the main reason for pH increase.

pH was a key factor for the calcium phosphorus precipitation (Carlsson et al., 1997; Wang et al., 2007). It was hypothesized that pH value was 7.0 in the anaerobic tank of stage 5. The speciation and SI values under assumed and practical pH were shown in Fig. 8 and Table 5, respectively. The basic species (${\text{CaPO}}_4^{-}\text{,}{\text{CaOH}}^{+}\text{,}{\text{PO}}_4^{3-}$) (Song et al., 2002) increased under higher pH, and SI values with respect to calcium phosphorus precipitations all increased. Consequently, pH values played an important role in the speciation and precipitation. Calcium phosphorus precipitation could be inhibited by controlling pH in the anaerobic tank.

Figure 8.

The speciation of calcium and phosphate under different pH values.

Table 5.

SI values under different pHs.

Phase	pH = 7.0	pH = 8.1
ACP	−2.31	0.44
DCPC	−0.65	−0.45
HAP	6.46	11.76

4.4. The equilibrium between chemical phosphorus removal and biological phosphorus removal in A₂N–IC

As seen from Fig. 4, P_rel was decreased obviously in A₂N–IC under the similar influent wastewater composition conditions. In order to clarify the decrease procedure of phosphorus release in A₂N–IC, the equilibrium between chemical phosphorus removal and biological phosphorus removal in A₂N–IC was analyzed according to mass balance and the metabolic mechanism of DNPAOs.

In the A₂N process, phosphorus uptake could be calculated according to mass balance.

$$M_{\text{upt}}^P=M_{\text{inf}}^P+M_{\text{rel}}^P-M_{\text{eff}}^P$$ (3)

Where $M_{\text{inf}}^P$ is the influent phosphorus amount (mg h⁻¹); $M_{\text{upt}}^P$ is the phosphorus uptake amount in anoxic and post-aerobic tank (mg h⁻¹); $M_{\text{rel}}^P$ is the phosphorus release amount in anaerobic tank (mg h⁻¹); $M_m^P$ is the effluent phosphorus amount (mg h⁻¹). And it is supposed that $M_{\mathit{upt}}^P={\mathit{kM}}_{\mathit{rel}}^P$, where $k$ is the biological phosphorus removal coefficient.

When the crystallization reactor was added into the system at t₀, phosphorus uptake is calculated as shown in Eq. (4).

$$M_{\text{upt}}^{P\prime }=M_{\text{inf}}^P+M_{\text{rel}}^P-M_{\text{chem}}^P-M_{\text{eff}}^{P\prime }(t=t_0)$$ (4)

Where $M_{\text{upt}}^{P\prime }$ is the phosphorus uptake amount at the initial stage of the A₂N–IC (t = t₀) (mg h⁻¹); $M_{\text{eff}}^{P\prime }$ is the effluent phosphorus amount at the initial stage of the A₂N–IC (t = t₀) (mg h⁻¹). If $M_{\text{chem}}^P$ was higher than $M_{\text{eff}}^P$, $M_{\text{eff}}^{P\prime }$ was zero because the limiting phosphorus would be used for biological phosphorus uptake before leaving the system. Consequently, $M_{\text{upt}}^{P\prime }<M_{\text{upt}}^P$. According to the metabolic mechanism of DNPAOs (Kuba et al., 1996a; Murnleitner et al., 1997), the decrease of biological phosphorus uptake resulted in the reduction of poly-P storage and biological phosphorus release. Meanwhile, it led to the decrease of chemical phosphorus removal. Therefore after a tiny time (Δt), $M_{\text{rel}}^P$ and $M_{\text{chem}}^P$ decreased to $M_{\text{rel}}^{P\prime }$ and $M_{\text{chem}}^{P\prime }$, respectively. Biological phosphorus uptake could be calculated as shown in Eq. (5).

$$M_{\text{upt}(1)}^{P\prime }=M_{\text{inf}}^P+M_{\text{rel}}^{P\prime }-M_{\text{chem}}^{P\prime }-M_{\text{eff}(1)}^{P\prime }(t=t_1\text{,}{t}_1=t_0+\mathrm{\Delta }t)\text{.}$$ (5)

The similar phenomenon would occur as shown in Eq. (6) until Eq. (7) was established.

$$M_{\text{upt}(n)}^{P\prime }=M_{\text{inf}}^P+M_{\text{rel}(n-1)}^{P\prime }-M_{\text{chem}(n-1)}^{P\prime }-M_{\text{eff}(n)}^{P\prime }(t=t_n)$$ (6)

$$\left(M_{\text{eff}}^{P\prime }=M_{\text{eff}(1)}^{P\prime }=...=M_{\text{eff}(n-1)}^{P\prime }=0\right)$$ (7)

$$M_{\text{upt}(n)}^{P\prime }\ge k_{(n-1)}^{P\prime }{M}_{\text{rel}(n-1)}^{P\prime }\text{.}$$ (8)

Finally, $M_{\text{upt}}^P$ and $M_{\text{rel}}^P$ decreased to $M_{\text{upt}(n)}^{P\prime }$ and $M_{\text{rel}(n-1)}^{P\prime }$ respectively. The new balance between biological uptake and chemical removal was achieved. In comparison with A₂N, the biological phosphorus removal decrement in A₂N–IC was ${\sum }_{i=1}^{i=n}(M_{\mathit{upt}}^P-M_{\mathit{upt}(i)}^{P\prime }-M_{\mathit{rel}}^P+M_{\mathit{rel}(i)}^P)(t_i-t_{i-1})$.

It is worth noting that the above analysis described the procedure from the steady state of A₂N to the unstable condition of A₂N–IC, and finally to the stable stage of A₂N–IC. It was totally different from the equations in our previous study (Shi et al., 2012), which was used for the distribution of phosphorus for biological and chemical phosphorus removal in the steady A₂N–IC system.

5. Conclusions

Under different influent COD and ammonia composition, COD and ammonia removal rates were very stable, which were maintained at 84.9% and 86.6% in both processes. A₂N–IC showed better phosphorus removal performance compared with A₂N. Phosphorus release decreased due to the incorporation of chemical induced crystallization, which brought better nutrient ratios for anoxic phosphorus uptake. Mass balance equations were used to explain the equilibrium between chemical phosphorus removal and biological phosphorus removal in the A₂N–IC process.

To promote phosphorus recovery and ammonia removal in the case of high influent ammonia, the corresponding approaches were appropriate, pointed out as follows: (1) nitrification effect could be enhanced in the post-aerobic tank. Therefore ammonia removal rate could be increased; (2) calcium phosphorus precipitation could be inhibited by controlling pH at the anaerobic stage, so the phosphorus recovery could be improved.