- Email: [email protected]
oxic processes: Improvements to published N2O models
Chemical Engineering Journal 325 (2017) 386–395
Contents lists available at ScienceDirect
Chemical Engineering Journal journal homepage: www.elsevier.com/locate/cej
Mathematical modeling of nitrous oxide (N2O) production in anaerobic/ anoxic/oxic processes: Improvements to published N2O models Xiaoqian Ding a,d, Jianqiang Zhao a,b,⇑, Bo Hu c, Xiaoling Li c, Guanghuan Ge a, Kun Gao a, Ying Chen a,b a
School of Environmental Science and Engineering, Chang’an University, Xi’an 710064, Shaanxi, China Key Laboratory of Subsurface Hydrology and Ecological Effect in Arid Region of Ministry of Education, Xi’an 710064, Shaanxi, China c School of Civil Engineering, Chang’an University, Xi’an 710064, Shaanxi, China d School of Architecture and Civil Engineering, Xi’an University of Science and Technology, Xi’an 710054, Shaanxi, China b
h i g h l i g h t s
g r a p h i c a l a b s t r a c t
An improved model was proposed for
N2O production modeling in the A2O process. Competition for electrons among four denitrification reductases was considered. One affinity constant for XSTO was divided into four constants in the model. The improved model better predicted N2O production and nitrite accumulation. N2O accumulation resulted from the more rapid decline of the N2O reduction rate.
a r t i c l e
i n f o
Article history: Received 21 December 2016 Received in revised form 21 April 2017 Accepted 13 May 2017 Available online 15 May 2017 Keywords: N2O modeling Denitrification on intracellular polymers Electron competition A2O process Model improvement
a b s t r a c t Competition for electrons among different steps of denitrification on intracellular polymers (XSTO) plays a significant role in nitrous oxide (N2O) accumulation in the biological nitrogen removal process. In this work, this electron competition was considered in a mathematical model to predict N2O production in anaerobic/anoxic/oxic sequencing batch reactors (A2O-SBR) for the first time. The affinity constant for intracellular polymers of heterotrophs (K STO ) that was used in previously published models was divided into four affinity constants (K STO;1 , K STO;2 , K STO;3 and K STO;4 ) to represent the ability of each denitrification reductase to compete for intracellular polymers. The improved model was calibrated and validated using experimental data from three independent A2O-SBR systems. The results demonstrated that the modeling predictions strongly agreed with the measured data from all experimental tests under various operational conditions. The modeling results indicated that N2O accumulation resulted from the more rapid decline of the N2O reduction rate than the nitrite reduction rate for the inadequate XSTO in these A2O-SBR systems. The modeling results also suggested that distinguishing affinity constants for intracellular polymers during the four-step denitrification felicitously described a different XSTO distribution in each reduction step, thereby better predicting nitrogen dynamics and N2O production in A2O processes than the published model. The improved model is therefore a preferable tool to gain insight into N2O accumulation in A2O processes. Ó 2017 Elsevier B.V. All rights reserved.
⇑ Corresponding author at: School of Environmental Science and Engineering, Chang’an University, Xi’an 710064, Shaanxi, China. E-mail address: [email protected] (J. Zhao). http://dx.doi.org/10.1016/j.cej.2017.05.082 1385-8947/Ó 2017 Elsevier B.V. All rights reserved.
X. Ding et al. / Chemical Engineering Journal 325 (2017) 386–395
1. Introduction Nitrous oxide (N2O) is one of the most important greenhouse gases. Although its proportion in total greenhouse gas emissions is only 0.03%, it has a global warming potential more than 300-fold greater than carbon dioxide (CO2) [1]; it significantly contributes to the carbon footprint of greenhouse gas emissions [2]. Measurements have demonstrated that substantial amounts of N2O can be emitted from lab-scale experiments and full-scale wastewater treatment plants (WWTPs) [2] during both autotrophic nitrification and heterotrophic denitrification processes [3–6]. Low dissolved oxygen (DO) concentrations, high nitrite accumulations during nitrification and denitrification processes, and a limited availability of biodegradable organic compounds during denitrification are the most significant causes leading to N2O accumulation and emission in biological nitrogen removal (BNR) processes [2,4,7]. Moreover, when carbon sources are limited or carbon/nitrogen (C/N) ratios are low [8,9], the electron supply cannot meet the electron demand for complete denitrification. Under these circumstances, electron competition may occur among four denitrification reductases [10]. Pan et al. [10] also revealed that competition for electrons occurs under carbon-limited as well as carbon-abundant conditions. Furthermore, the slower degradation of polyhydroxybutyrate (PHB) cannot provide adequate electrons for denitrification. Therefore, electron competition likely also occurs when PHB is used as an electron donor [2,7]. Specifically, under electron-competitive conditions, nitrate reductase may have great advantages in capturing electrons when compared to nitrite and N2O reductases [2], which easily leads to nitrite and N2O accumulation [11,12]. As N2O reduction is the last step of denitrification, N2O reductase cannot gain enough electrons, resulting in N2O production during denitrification [2,11,13]. Additionally, some environmental conditions, such as pH, temperature, carbon source and free nitrous acid (FNA) inhibition, can intensify electron competitions and eventually lead to N2O accumulation [10,14]. Therefore, the mechanisms of electron completion on N2O production in the denitrification process are of great importance and deserve more attention in research. Mathematical modeling has been demonstrated to be helpful for testing the mechanisms of pollutant removal in wastewater treatment [15]. Firstly, the Activated Sludge Model for Nitrogen (ASMN) proposed by Hiatt and Grady [16] was successfully developed to predict N2O production by describing heterotrophic denitrification as a four-step process. Subsequently, the Activated Sludge Model for Indirect Coupling of Electrons (ASM-ICE) was developed to represent the electron competition among four denitrification steps [17,18]. However, both the four-step ASMN model and the ASM-ICE model did not consider the role of intracellular polymers in N2O production in denitrification processes. Afterward, the relations between denitrification on intracellular polymers (XSTO, an internal cell storage product of heterotrophic organisms [15]) and N2O accumulation were investigated in several published models [19–21]. These models satisfactorily described XSTO synthesis/consumption, nitrogen removal, and N2O production in denitrification on intracellular polymers, denitrifying phosphorus removal and anaerobic/oxic/anoxic (AOA) processes. However, due to the lack of recognition of the electron competition in denitrification, none of these models considered the competition for electrons in denitrification processes with the intracellular polymers being the sole electron donors. More information should be provided to explore the reaction kinetics for the models aiming to predict N2O production during denitrification on intracellular polymers. The anaerobic/anoxic/oxic (A2O) process, in which phosphorus and nitrogen can be removed simultaneously [22,23], is the most
387
commonly used process in wastewater treatment [24]. In the WWTPs of China, the A2O process accounted for the biggest design treatment capacity of 33.2% [25]. The A2O process undergoes alternating anaerobic/anoxic/oxic conditions to achieve internal cell product storage/consumption and nitrogen removal. Firstly, the readily biodegradable organics (SS) are stored as intracellular polymers by heterotrophic bacteria under anaerobic conditions. Next, nitrate (NO 3 ) is reduced to nitrite (NO2 ), then nitric oxide (NO), N2O, and finally nitrogen gas (N2) in the anoxic stage, using the intracellular polymers as electron donors. Nitrate reductase (Nar), nitrite reductase (Nir), NO reductase (Nor) and N2O reductase (Nos) are the four functional enzymes in this four-step denitrification process [26]. During the aerobic stage, ammonium (NH+4) is successively oxidized to nitrite and nitrate by ammonia oxidizing bacteria (AOB) and nitrite oxidizing bacteria (NOB), respectively. A significant amount of nitrite could accumulate under oxygenlimited conditions. Due to the utilization of the aerobically synthesized PHB for denitrification in the anoxic stage, N2O could be emitted from WWTPs running the A2O process [27]. Moreover, N2O emissions from the full-course BNR processes operated at the oxidation ditch (OD), University of Cape Town (UCT), AO and AOA modes have been previously modeled [19,28–30], but N2O production in an A2O system has never been modeled. Since the application of the A2O process is popular in wastewater treatment, modeling efforts should begin now. In this work, an improved model, in which the competition for storage polymers among four denitrification reductases was firstly introduced, was proposed to investigate the competition for electrons and the mechanisms of N2O production during denitrification in A2O processes. The improved model was verified using experimental data from three independent A2O-SBR systems with different operation characteristics. The new model is expected to be beneficial for improving the prediction of N2O production in A2O processes. 2. Materials and methods 2.1. Model description The mechanisms of XSTO synthesis/consumption and nitrogen conversion dynamics in the A2O process are the same as that in the AOA process. Therefore, the published model for predicting N2O production in AOA systems (the published model for short [19]) was employed in this work. The component definitions, stoichiometry and composition matrix, kinetic rate expression matrix of the published model are presented in Supplementary Material Tables S.1–S.3. The published model by Ding et al. [19], as well as other reported models describing N2O production during denitrification on intracellular polymers [20,21], did not consider the electron competition among the four reduction steps under starving conditions [2,3,11,13]. The affinity constants (K STO ) regarding storage polymers (electron donor) were uniformly used in all four denitrification steps [19–21]. This limitation potentially restrains the model application for N2O prediction in endogenous denitrification processes. In this work, the key kinetic parameter of K STO was divided into four independent affinity constants, i.e., K STO;1 , K STO;2 , K STO;3 and K STO;4 , to describe the distinct electron capturing ability of each denitrification reductase. These four affinity constants, which govern the distributions of intracellular polymers, were introduced into the denitrification process in this study for the first time. Modifications of the kinetic rate expressions (R5.1, R5.2, R5.3 and R5.4) in the published models are presented in Table 1. The parameter values of K STO;1 , K STO;2 , K STO;3 and K STO;4 were calibrated using the experimental data from the anaerobic/anoxic/oxic
388
X. Ding et al. / Chemical Engineering Journal 325 (2017) 386–395
Table 1 Modified kinetic rate expressions in the improved model. Process
Process rate equation
5.1
K S;1 I;H;NO2 ;1 X STO =X H 3 lH;STO;NO;1 K H;OH;OþS2 O K H;NONOþS K STO;1 þX STO =X H K S;1 þSS K I;H;NO ;1 þSNO X H NO
K
S
2
2
K
K
3
3
2
S
2
K
5.2
K S;2 I;H;NO2 ;2 X STO =X H 2 lH;STO;NO;2 K H;OH;OþS2 O K H;NONOþS K STO;2 þX STO =X H K S;2 þSS K I;H;NO ;2 þSNO X H NO
5.3
K S;3 I;H;NO2 ;3 =X H lH;STO;NO;3 K H;OH;OþS2 O K H;NOSNOþSNO K STO;3XSTO þX STO =X H K S;3 þSS K I;H;NO ;3 þSNO X H
2
2
2
2
2
K
2
5.4
2
K
2
K
2
S
2
K
K S;4 I;H;NO2 ;4 =X H lH;STO;NO;4 K H;OH;OþS2 O K H;N NO2þSO N O K STO;4XSTO þX STO =X H K S;4 þSS K I;H;NO ;4 þSNO X H 2
2
2
2
2
2
sequencing batch reactors (A2O-SBR) from the literature. The kinetic rate expressions of autotrophic bacteria were the same as the published model. 2.2. Experimental data Data from three experimental cases studying N2O production in A2O-SBR systems [11,13,31] were employed to test the simulation capability of the improved model. The operational characteristics of these modeling cases are illustrated in Table 2. 2.2.1. Experimental data from Case 1 [13] In this study, a sealed laboratory-scale A2O-SBR (7.5 L) was used to investigate N2O emission during denitrifying phosphorus removal. The reactor was operated with three cycles per day; each cycle included a 90 min anaerobic phase, a 210-min anoxic phase and a 30 min aerobic phase with DO concentrations at 2–4 mg O2 L1. In each cycle, 5.5 L of synthetic wastewater, which contained 448.46 mg of CH3COONa (350 mg of chemical oxygen demand (COD)) per liter and other nutrient elements, was fed into the SBR. At the end of the anaerobic phase, 100 mL of KNO3 solution was added to the reactor, providing an initial NO 3 concentration of 35 mg N L1 for anoxic denitrification. During the steady-state operation, three batch experiments were carried out in the sealed batch reactors (1 L) to explore the effects of carbon source shock on N2O production during the denitrifying phosphorus removal process. Among them, the batch test investigating acetate (HAc) shock on N2O production was operated with the same carbon load and the same initial NO 3 concentration as the parent SBR. More detailed experimental setup and analytical methods can be found in Wang et al. [13]. A typical cycle-running study in the parent SBR and a batch test with acetate as the sole carbon source were employed to calibrate and validate the improved model. 2.2.2. Experimental data from Case 2 [31] This study aimed to evaluate N2O production and denitrification performance of the denitrifying phosphorus removal process under different anaerobic reaction times (AnRTs). The parent laboratoryscale SBR (7.5 L) was operated in the A2O mode, where one standard running cycle was 8 h, containing a 120 min anaerobic period,
a 210 min anoxic period and a 30 min aerobic period. Synthetic wastewater with mixed acetate and propionate (in the molar ratio 3:1) as carbon sources was fed into the SBR at the beginning of the anaerobic period, while the KNO3 solution was added to the reactor at the beginning of the anoxic period. The initial COD concentration was approximately 220 mg COD L1, and the NO 3 concentration was 50 ± 5 mg N L–1 in the reactor. The DO concentration was approximately 2–3 mg O2 L1 in the aerobic phase. Three batch tests investigating the short-term effect of AnRT (i.e., 90 min in Run 1, 120 min in Run 2, 150 min in Run 3) on N2O production during denitrifying phosphorus removal were conducted using three sealed batch reactors (1 L). The synthetic wastewater, the KNO3 solution and the adding methods were the same as in the parent SBR. More detailed experimental setup and analytical methods can be found in Guo et al. [31]. Three batch tests in this modeling case were adopted to validate the improved model structure and parameter values. 2.2.3. Experimental data from Case 3 [11] A laboratory-scale A2O-SBR (7.5 L) was used to investigate PHB formation, N2O production and denitrifying phosphorus removal performance. The DO level in the reactor was controlled within 2–4 mg O2 L1 during the aerobic stage. The experiment consisted of four running phases, of which phase I was the longest. During phase I, the sealed reactor was operated for three cycles per day, with each cycle consisting of a 90 min anaerobic reaction, a 210 min anoxic reaction, and a 30 min aerobic reaction. In each cycle, 5.5 L of synthetic wastewater with HAc as the sole carbon source (approximately 400 mg COD L1) was fed into the reactor and 100 mL of KNO3 solution was added to the reactor, resulting 1 in an initial NO at the start of the anoxic 3 concentration of 40 NL period. More detailed experimental setup and analytical methods can be found in Wang et al. [11]. Experimental data in typical cycles (Cycle 77 and Cycle 119) of phase I were used to validate the model. 2.3. Model calibration and validation Model calibration for optimizing parameter values was performed by fitting the predicted curve to the measured data from the experimental studies. The newly introduced and the most important parameters were estimated in each modeling case. In Case 1, experimental data from the cycle running test were first used to calibrate the model parameters; data from the batch test were then used to validate the calibrated parameter values. In Case 2, experimental data from Run 1 were used to estimate the parameter values, and experimental data from Run 2 and Run 3 were applied to evaluate the improved model. Likewise, the operational results from Cycle 77 and Cycle 119 were used to calibrate and validate the model parameters, respectively. The MATLAB (R2012a) program was selected as the computational platform.
Table 2 Three experimental cases used for model validation. Culture
Carbon source
Nitrogen added
Experimental tests
Case 1
DPAOs(a)
Acetate: 350 mg COD L1
1 NO 3 : 35 mg N L
Case 2
DPAOs(a)
Acetate and propionate (in the molar ratio 3:1): 300 mg COD L1
1 NO 3 : 50 mg N L
Case 3
DPAOs
Acetate: 400 mg COD L1
1 NO 3 : 40 mg N L
(A): cycle running test (B): batch test (A): run 1 with an AnRT(b) of 90 min; (B): run 2 with an AnRT(b) of 120 min; (C): run 3 with an AnRT(b) of 150 min (A): cycle running test on Day 77 (B): cycle running test on Day 119
(a)
DPAOs(a): Denitrifying phosphate accumulating organisms. AnRT(b): Anaerobic reaction time.
389
X. Ding et al. / Chemical Engineering Journal 325 (2017) 386–395
3. Results and discussion 3.1. Model calibration The biological nature of wastewater treatment processes implies that their activated sludge model parameters must be determined according to the local situation [32]. Thus, the most crucial parameters were individually calibrated for each A2O-SBR
study in this work. Competition for electron donors and FNA inhibition have been regarded as the leading causes of N2O accumulation during denitrification [2,14,33]. Consequently, four new affinity constants for the XSTO of four denitrification reductases (K STO;1 , K STO;2 , K STO;3 and K STO;4 ) and four nitrite inhibition constants for NO 3 , NO2 , NO and N2O reductions were calibrated in each modeling case. Simultaneously, the maximum growth rates of heterotrophs in both the aerobic and anoxic stages were also calibrated.
Table 3 Calibrated stoichiometric and kinetic parameters and their values. Symbol
Definition
Unit
Case 1
Case 2
Case 3
lH;STO;O2 lH;STO;NO;1 lH;STO;NO;2 lH;STO;NO;3 lH;STO;NO;4
Aerobic max. growth rate on X STO
d1
3.6(1)
3.6(1)
3.6(1)
Anoxic max. growth rate on X STO for NO 3 reduction
d1
4.5(3)
1.68(2)
1.68(2)
Anoxic max. growth rate on X STO for NO 2 reduction
d1
1.18(3)
0.456(2)
0.576(4)
Anoxic max. growth rate on X STO for NO reduction
d1
3.408(2)
3.408(2)
3.408(2)
Anoxic max. growth rate on X STO for N2O reduction
d
1.36
K STO K STO;1 K STO;2 K STO;3 K STO;4 K I;H;NO2 ;2 K I;H;NO2 ;3 iN;BM
Saturation constant of X STO for aerobic growth Saturation constant of X STO for NO 3 reduction Saturation constant of X STO for NO 2 reduction Saturation constant of X STO for NO reduction Saturation constant of X STO for N2O reduction Nitrite inhibition constant for NO 2 reduction Nitrite inhibition constant for NO reduction N content of microbial organisms
g COD g1 COD g COD g1 COD g COD g1 COD g COD g1 COD g COD g1 COD g N m3 g N m3 g N g1 COD
0.1(3) 0.1(2) 0.1(2) 0.1(2) 0.125(3) 10(2) 10(3) 0.03(3)
1
(3)
(3)
0.82
1.12(3)
0.1(3) 0.1(2) 0.05(3) 0.1(2) 0.112(3) 10(2) 9(3) 0.03(3)
0.1(3) 0.06(3) 0.07(3) 0.1(2) 0.127(3) 8(3) 17(2) 0.05(3)
Source: (1) Zhou et al. [34]; (2) Liu et al. [21]; (3) Estimated in this study; (4) Liu et al. [20].
+ + Fig. 1. Modeling results and measured data in Case 1: (A1) HAc; (A2) NO 3 , NO2 and NH4; (A3) N2O profiles in the cycle running test; (B1) HAc; (B2) NO3 , NO2 and NH4; (B3) N2O profiles in the batch test.
390
X. Ding et al. / Chemical Engineering Journal 325 (2017) 386–395
In total, thirteen parameters were calibrated in three modeling cases, as shown in Table 3. In all three case studies, the calibrated lH;STO;NO;2 (1.18, 0.456 and 0.576 d1) and lH;STO;NO;4 (1.36, 0.82 and 1.12 d1) were lower than lH;STO;NO;1 (4.5, 1.68 and 1.68 d1), suggesting a lower reduction rate of nitrite and N2O than that of nitrate in the three A2O-SBR systems. In addition, since the degradation of XSTO was the direct source of electrons, the newly introduced K STO;1 , K STO;2 , K STO;3 and K STO;4 parameters with respect to XSTO utilizing described the electron capturing abilities of the four denitrification reductases. The calibrated values of the four affinity constants determined the distinct competitivenesses for electrons and electron donors among the four denitrification reductases. The estimated values of K STO;4 (0.125, 0.112 and 0.127 g COD g1 COD) were higher than K STO;1 (0.1, 0.1 and 0.06 g COD g1 COD) and K STO;2 (0.1, 0.05 and 0.07 g COD g1 COD) in all three modeling cases, indicating a weaker affinity for XSTO in N2O reduction com pared with that in NO 3 and NO2 reductions. These results indicated that Nos had a weaker competition for electrons compared to Nar and Nir [26]. Moreover, it has been reported that since the two upstream reduction enzymes are more powerful in competing for electrons compared to the downstream denitrification reductases, the electron priority was favored by Nar and Nir rather than by Nos when the electron supply was limited [10]. Other parameter values not listed in Table 3 were directly obtained from the literature in Ding et al. [19].
3.2. Model validation In Case 1, experimental data from the cycle running test were first used to calibrate the parameter values, in terms of HAc, + NO 3 , NO2 , NH4 and N2O dynamics. The model predictions and measured data in the experiment are illustrated in Fig. 1 A. During the anaerobic period, HAc was removed through XSTO synthesis in the microbial cell. When the nitrate solution was added to the reactor, NO 3 was reduced to NO2 , and a small amount of N2O was produced during the first 90 min of the anoxic stage. After NO 3 was completely reduced, NO2 was further reduced to N2 and N2O. As a result, N2O consistently increased along with NO 2 1 reduction, until NO . At the end of 2 was below to 7.5 mg N L the anoxic phase, when the low nitrite concentration weakened the FNA inhibition on N2O reduction, N2O was reduced to N2, resulting in a decrease in N2O emissions. Due to the N2O production from NH+4 oxidation, N2O emissions slightly increased in the aeration stage (Fig. 1A3). The developed model captured these trends well. After, experimental data from the batch test of Case 1 were used to validate the calibrated parameter values. As Fig. 1B indicated, the profiles of the variables in the batch test were similar to the cycle running result, except for the different N2O production caused by the different operation regimes of the two tests. Since N2O mainly produced during the anoxic period in these A2OSBR systems [11,13,31], the model structure and model parameters
+ + Fig. 2. Modeling results and measured data in Case 2: (A1) Ss; (A2) NO 3 , NO2 and NH4; (A3) N2O profiles in Run 1; (B1) Ss; (B2) NO3 , NO2 and NH4; (B3) N2O profiles in Run 2; + (C1) Ss; (C2) NO 3 , NO2 and NH4; (C3) N2O profiles in Run 3.
X. Ding et al. / Chemical Engineering Journal 325 (2017) 386–395
relating to AOB and NOB were directly adopted from the published model [19]. The modeling results were well matched with the experimental data in this validation case. In Case 2, experimental data from three batch tests were used to further calibrate and validate the improved model (Fig. 2). A set of + SS, NO 3 , NO2 , NH4 and N2O data from Run 1 were first used to calibrate the model parameters. Then, two sets of experimental data from Run 2 and Run 3 were used to evaluate the parameter values. Slightly unlike the reaction processes in Case 1, NO 3 was reduced slowly, with a slow and continual increase in NO 2 accumulation and a progressive growth in N2O production in all three batch tests. The difference in nitrogen dynamics between Case 2 and Case 1 was possibly due to the different carbon sources used in the experiment (Table 2). The results indicated that when the AnRT was extended from 90 to 150 min, the nitrite accumulations increased, strengthening the FNA inhibition on denitrification. Accordingly, the ratio of N2O production to nitrate removal exhibited a slight increase. The model predictions were in good agreement with the measured trends in all three runs. In Case 3, two cycling operation results were used to test the ability of the model to predict N2O production in A2O processes (Fig. 3). The experimental setups in Case 3 were similar to those in Case 1, except for the different carbon loads and nitrate loads in the reactors (Table 2). As a result, the profiles of HAc, NO 3, + NO 2 , NH4 and N2O in Case 3 were similar to those in Case 1. The modeling results satisfactorily agreed with the experimental data
391
from both Cycle 77 and Cycle 119 tests during the model calibration and validation stages. The parameter values that provided the optimum model fittings with the measured data in the three modeling cases are summarized in Table 3 (calibrated in this work) and Supplementary Material Table S.4 (obtained from the literature). The R2 values of the three modeling studies are provided in Table 4. The average value of R2 was 0.9245, indicating that the established model structure and parameter values are robust in their ability to predict HAc/ 2 + SS, NO 3 NO2 , NH4 and N2O profiles in A O-SBR systems under different operational conditions. 3.3. Comparison of modeling results with the published model To test the advantage of the improved model for N2O modeling in A2O processes, the modeling results provided by the improved model and the published model [19] for Case 3 were compared in Fig. 4. During the prediction performed by the published model, experimental data from Cycle 77 were used to calibrate the model parameters, while experimental data from Cycle 119 were used to validate the estimated parameter values. When the established K STO value was 0.1 g COD g1 COD for the published model, the estimated lH;STO;NO;1 , lH;STO;NO;2 , lH;STO;NO;3 and lH;STO;NO;4 values were 1.68 d1, 0.576 d1, 3.408 d1 and 0.93 d1 for the four reduction steps, respectively. Other parameter values calibrated for the published model were the same as the improved model (Table 3). The
+ + Fig. 3. Modeling results and measured data in Case 3: (A1) HAc; (A2) NO 3 , NO2 and NH4; (A3) N2O profiles in Cycle 77; (B1) HAc; (B2) NO3 , NO2 and NH4; (B3) N2O profiles in Cycle 119.
392
X. Ding et al. / Chemical Engineering Journal 325 (2017) 386–395
Table 4 R2 values of the prediction results for three modeling cases. Items
NO 3 NO 2 NH+4 HAc/SS N2O
Case 1
Case 2
Case 3
Cycle running test
Batch test
Run 1
Run 2
Run 3
Cycle 77
Cycle 119
0.9564 0.9556 0.9816 0.8460 0.9469
0.9548 0.9471 0.9879 0.6698 0.9628
0.9720 0.9706 0.9075 0.7692 0.9720
0.9854 0.9861 0.9191 0.8751 0.9889
0.9977 0.9915 0.8831 0.8660 0.9838
0.9708 0.9507 0.8516 0.9640 0.9920
0.8587 0.8435 0.8452 0.8218 0.9824
Fig. 4. Comparison of predicted results by the improved model (solid lines) and the published model (dashed lines) for Case 3: (A1) NO 3 ; (A2) NO2 and (A3) N2O profiles in Cycle 77; (B1) NO 3 ; (B2) NO2 and (B3) N2O profiles in Cycle 119.
results indicated that both models captured the trends of NO 3 , NO2 and N2O variations that occurred in Case 3. However, the fitting errors between the measured data and the predicted data by the published model were relatively larger than that predicted by the improved model. The improved model, by contrast, more success fully reproduced the NO 3 , NO2 and N2O profiles observed in Case 3. Likely, this difference was due to the higher K STO;2 value of 0.1 g COD g1 COD in the published model compared to that in the improved model (0.07 g COD g1 COD), as the XSTO consumption proportions for NO 2 reduction predicted by the published model (11.23% in Cycle 77 and 11.21% in Cycle 119) were lower than those predicted by the improved model (11.96% in Cycle 77 and 11.95% in Cycle 119), as shown in Fig. 5. This was the origin that resulted in a slower rate of nitrite reduction and a larger accumulation of N2O in the modeling results given by the published model, compared to the experimentally observed data. It was suggested that considering the electron competition among Nar, Nir, Nor and Nos by distinguishing the affinity constants for XSTO of
the four denitrification reductases was beneficial to predict the N2O production during denitrification in A2O systems. Thus, competition for XSTO among the four denitrification reductases should be incorporated into the four-step reduction process to correctly model N2O accumulation during denitrification on intracellular polymers. 3.4. Distribution of XSTO consumption during denitrification As N2O mainly accumulated during the anoxic period in these A2O-SBR systems [11,13,31], the consumption of the anaerobically synthesized XSTO during denitrification is of great importance. Based on differing affinities of the four denitrification reductases for XSTO, the distributions of XSTO consumption among the different reduction steps were quantitatively investigated, while attempting to represent the electron distribution during denitrification (Fig. 6). The results indicated that most of the XSTO was consumed during NO 3 reduction; thus, most electron fluxes were captured by Nar
X. Ding et al. / Chemical Engineering Journal 325 (2017) 386–395
Fig. 5. Comparison of distributions of XSTO consumption predicted by the improved model and the published model for Case 3: ‘‘improved” and ‘‘published” in the abscissa refer to the improved model and the published model, respectively.
to denitrify nitrate to nitrite in all three A2O-SBR systems. On the contrary, Nir and Nos gained little XSTO during nitrite reduction and N2O reduction in all three cases. Since Nar receives electrons directly from the pool in the first echelon, while Nir and Nos receive their electrons from the pool in the second echelon
393
[18,35], the electrons provided by XSTO degradation were first captured by Nar for nitrate reduction [2,13]. As Nar and Nor captured more electrons from more XSTO consumption, the denitrification rates for NO 3 and NO reduction were much faster than those for NO 2 and N2O reduction in the three modeling cases (Fig. 7), resulting in a trace amount of NO production (data not shown) and an 2 accumulation of NO 2 and N2O in these A O-SBR systems. Fig. 7 showed the NO and N O reduction rates in the three 2 2 modeling cases. It was found that both the NO 2 and N2O reduction rates declined with time, which can be described by the exponential function. By comparing the values of the function parameters, it was revealed that N2O reduction rates decreased more rapidly than NO 2 reduction rates in all modeling tests. This suggested that the N2O reduction rates (rN2 O ) declined much more rapidly than the NO ) with the consumption of XSTO, resulting 2 reduction rates (r NO 2 in the apparent N2O accumulation in these validation cases. It further confirmed that Nir is more competent at capturing electrons than Nos when the electrons available for denitrification was inadequate [2,10]. Moreover, in all three modeling cases, XSTO consumption distributions among Nar, Nir, Nor and Nos did not vary significantly with operation regimes (cycle running or batch test in Case 1), AnRTs (90 min, 120 min or 150 min in Case 2), or carbon source (HAc in Case 1 and Case 3 or HAc + Pro in Case 2), and remained at approximately 44.31%, 15.13%, 22.05% and 18.52%, respectively. Meanwhile, the N2O production factors (N2O production/NO 3 removed) were different at the same XSTO distributions in each modeling case (Fig. 6). This difference was possibly due to the diverse differences between the rN2 O and rNO2 decline speeds, which might be attributed to the above-mentioned different operating conditions in one modeling case resulting in the slightly different amounts of XSTO production and consumption. This suggested that N2O accumulation in A2O-SBR systems was caused by not only XSTO distributions but also XSTO synthesis as determined by operational conditions.
Fig. 6. Predicted XSTO consumption distributions (A) and measured N2O production factors (B) during denitrification in three modeling cases.
394
X. Ding et al. / Chemical Engineering Journal 325 (2017) 386–395
Fig. 7. Predicted denitrification rates for NO 3 reduction, NO2 reduction, NO reduction and N2O reduction in three modeling cases: (A1) the cycle running test in Case 1; (A2) the batch test in Case 1; (B1) Run 1 in Case 2; (B2) Run 2 in Case 2; (B3) Run 3 in Case 2; (C1) Cycle 77 in Case 3; (C2) Cycle 119 in Case 3.
4. Conclusions In this work, an improved model for A2O processes was proposed to predict nitrogen dynamics and N2O accumulation in both cycle running tests and batch tests. The results demonstrated that the improved model was capable of reproducing the measured data in three A2O-SBR cases. N2O accumulation resulted from the more rapid decline of the N2O reduction rate than the nitrite reduction rate when the XSTO was inadequate during denitrification. The results suggested that the improved model is favorable over the published models in describing N2O production under electron-competitive conditions. The improved model is expected to provide more information for understanding N2O production and electron distribution in A2O processes.
Author disclosure statement The authors declare that they have no conflict of interest. Acknowledgments This work was supported by the Shaanxi Province Science & Technology Development Program (Grant No. 2014K15-03-02) and the National Natural Science Foundation of China (Grant No. 51308050). Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.cej.2017.05.082.
X. Ding et al. / Chemical Engineering Journal 325 (2017) 386–395
References [1] [2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15] [16] [17]
[18]
IPCC, Climate Change 2013: The Physical Science Basis, 2013, Stockholm, Sweden. M.J. Kampschreur, H. Temmink, R. Kleerebezem, M.S.M. Jetten, M.C.M. van Loosdrecht, Nitrous oxide emission during wastewater treatment, Water Res. 43 (2009) 4093–4103. C. Li, J. Zhang, S. Liang, H.H. Ngo, W. Guo, Y. Zhang, Y. Zou, Nitrous oxide generation in denitrifying phosphorus removal process: main causes and control measures, Environ. Sci. Pollut. R. 20 (2013) 5353–5360. P. Wunderlin, J. Mohn, A. Joss, L. Emmenegger, H. Siegrist, Mechanisms of N2O production in biological wastewater treatment under nitrifying and denitrifying conditions, Water Res. 46 (2012) 1027–1037. R.L. Meyer, R.J. Zeng, V. Giugliano, L.L. Blackall, Challenges for simultaneous nitrification, denitrification, and phosphorus removal in microbial aggregates: mass transfer limitation and nitrous oxide production, FEMS Microbiol. Ecol. 52 (2005) 329–338. W. Jia, J. Zhang, H. Xie, Y. Yan, J. Wang, Y. Zhao, X. Xu, Effect of PHB and oxygen uptake rate on nitrous oxide emission during simultaneous nitrification denitrification process, Bioresource Technol. 113 (2012) 232–238. S. Schalk-Otte, R.J. Seviour, J.G. Kuenen, M.S.M. Jetten, Nitrous oxide (N2O) production by Alcaligenes faecalis during feast and famine regimes, Water Res. 34 (2000) 2080–2088. H. Itokawa, K. Hanaki, T. Matsuo, Nitrous oxide production in high-loading biological nitrogen removal process under low COD/N ratio condition, Water Res. 35 (2001) 657–664. H.J. Lu, K. Chandran, Factors promoting emissions of nitrous oxide and nitric oxide from denitrifying sequencing batch reactors operated with methanol and ethanol as electron do. . ., Biotechnol. Bioeng. (2010). Y. Pan, B. Ni, P.L. Bond, L. Ye, Z. Yuan, Electron competition among nitrogen oxides reduction during methanol-utilizing denitrification in wastewater treatment, Water Res. 47 (2013) 3273–3281. Y. Wang, J. Geng, Z. Ren, W. He, M. Xing, M. Wu, S. Chen, Effect of anaerobic reaction time on denitrifying phosphorus removal and N2O production, Bioresour. Technol. 102 (2011) 5674–5684. Y. Zhou, M. Lim, S. Harjono, W.J. Ng, Nitrous oxide emission by denitrifying phosphorus removal culture using polyhydroxyalkanoates as carbon source, J. Environ. Sci. (China) 24 (2012) 1616–1623. Y. Wang, J. Geng, G. Guo, C. Wang, S. Liu, N2O production in anaerobic/anoxic denitrifying phosphorus removal process: the effects of carbon sources shock, Chem. Eng. J. 172 (2011) 999–1007. Y. Zhou, M. Pijuan, R.J. Zeng, Z. Yuan, Free nitrous acid inhibition on nitrous oxide reduction by a denitrifying-enhanced biological phosphorus removal sludge, Environ. Sci. Technol. 42 (2008) 8260–8265. M. Henze, W. Gujer, T. Mino, M.C.M. van Loosdrecht, Activated Sludge Models ASM1, ASM2, ASM2d and ASM3, IWA Publishing, London, UK, 2000. W.C. Hiatt, C.P.L. Grady, An updated process model for carbon oxidation, nitrification, and denitrification, Water Environ. Res. 80 (2008) 2145–2156. Y. Pan, B. Ni, H. Lu, K. Chandran, D. Richardson, Z. Yuan, Evaluating two concepts for the modelling of intermediates accumulation during biological denitrification in wastewater treatment, Water Res. 71 (2015) 21–31. Y. Pan, B.J. Ni, Z. Yuan, Modeling electron competition among nitrogen oxides reduction and N2O accumulation in denitrification, Environ. Sci. Technol. 47 (2013) 11083–11091.
395
[19] X. Ding, J. Zhao, B. Hu, Y. Chen, G. Ge, X. Li, S. Wang, K. Gao, X. Tian, Mathematical modeling of nitrous oxide production in an anaerobic/oxic/ anoxic process, Bioresour. Technol. 222 (2016) 39–48. [20] Y Liu, L. Peng, J. Guo, X. Chen, Z. Yuan, B. Ni, Evaluating the role of microbial internal storage turnover on nitrous oxide accumulation during denitrification, Sci. Rep. (2015). [21] Y. Liu, L. Peng, X. Chen, B. Ni, Mathematical modeling of nitrous oxide production during denitrifying phosphorus removal process, Environ. Sci. Technol. 49 (2015) 8595–8601. [22] F. Fang, B. Ni, W. Li, G. Sheng, H. Yu, A simulation-based integrated approach to optimize the biological nutrient removal process in a full-scale wastewater treatment plant, Chem. Eng. J. 174 (2011) 635–643. [23] T. Pai, Modeling nitrite and nitrate variations in A2O process under different return oxic mixed liquid using an extended model, Process Biochem. 42 (2007) 978–987. [24] Y. Peng, X. Wang, B. Li, Anoxic biological phosphorus uptake and the effect of excessive aeration on biological phosphorus removal in the A2O process, Desalination 189 (2006) 155–164. [25] L. Jin, G. Zhang, H. Tian, Current state of sewage treatment in China, Water Res. 66 (2014) 85–98. [26] C. Li, T. Wang, N. Zheng, J. Zhang, H.H. Ngo, W. Guo, S. Liang, Influence of organic shock loads on the production of N2O in denitrifying phosphorus removal process, Bioresource Technol. 141 (2013) 160–166. [27] Y. Wang, D. Zhou, W. Zhao, L. Ye, L. Li, X. Tan, C. Song, Nitrous oxide emissions from biological wastewater treatment plants: current status and challenges, Acta Sci. Circum. 34 (2014) 1079–1088. [28] B. Ni, L. Ye, Y. Law, C. Byers, Z. Yuan, Mathematical modeling of nitrous oxide (n2o) emissions from full-scale wastewater treatment plants, Environ. Sci. Technol. 47 (2013) 7795–7803. [29] B. Ni, Y. Pan, B. van den Akker, L. Ye, Z. Yuan, Full-scale modeling explaining large spatial variations of nitrous oxide fluxes in a step-feed plugflow wastewater treatment reactor, Environ. Sci. Technol. 49 (2015) 9176–9184. [30] M. Spérandio, M. Pocquet, L. Guo, B. Ni, P.A. Vanrolleghem, Z. Yuan, Evaluation of different nitrous oxide production models with four continuous long-term wastewater treatment process data series, Bioprocess. Biosyst. Eng. 39 (2016) 493–510. [31] G. Guo, Y. Wang, C. Wang, H. Wang, M. Pan, S. Chen, Short-term effects of excessive anaerobic reaction time on anaerobic metabolism of denitrifying polyphosphate-accumulating organisms linked to phosphorus removal and N2O production, Front. Environ. Sci. Eng. 7 (2013) 616–624. [32] P.A. Vanrolleghem, H. Spanjers, B. Petersen, P. Ginestet, I. Takacs, Estimating (combinations of) activated sludge and components by respirometry, Water Sci. Technol. 39 (1999) 195–214. [33] J.H. Ahn, S. Kim, H. Park, B. Rahm, K. Pagilla, K. Chandran, N2O emissions from activated sludge processes, 2008–2009: Results of a national monitoring survey in the United States, Environ. Sci. Technol. 44 (2010) 4505–4511. [34] M. Zhou, J. Gong, C. Yang, W. Pu, Simulation of the performance of aerobic granular sludge SBR using modified ASM3 model, Bioresour. Technol. 127 (2013) 473–481. [35] D. Richardson, H. Felgate, N. Watmough, A. Thomson, E. Baggs, Mitigating release of the potent greenhouse gas N2O from the nitrogen cycle – could enzymic regulation hold the key?, Trends Biotechnol 27 (2009) 388–397.
Contents lists available at ScienceDirect
Chemical Engineering Journal journal homepage: www.elsevier.com/locate/cej
Mathematical modeling of nitrous oxide (N2O) production in anaerobic/ anoxic/oxic processes: Improvements to published N2O models Xiaoqian Ding a,d, Jianqiang Zhao a,b,⇑, Bo Hu c, Xiaoling Li c, Guanghuan Ge a, Kun Gao a, Ying Chen a,b a
School of Environmental Science and Engineering, Chang’an University, Xi’an 710064, Shaanxi, China Key Laboratory of Subsurface Hydrology and Ecological Effect in Arid Region of Ministry of Education, Xi’an 710064, Shaanxi, China c School of Civil Engineering, Chang’an University, Xi’an 710064, Shaanxi, China d School of Architecture and Civil Engineering, Xi’an University of Science and Technology, Xi’an 710054, Shaanxi, China b
h i g h l i g h t s
g r a p h i c a l a b s t r a c t
An improved model was proposed for
N2O production modeling in the A2O process. Competition for electrons among four denitrification reductases was considered. One affinity constant for XSTO was divided into four constants in the model. The improved model better predicted N2O production and nitrite accumulation. N2O accumulation resulted from the more rapid decline of the N2O reduction rate.
a r t i c l e
i n f o
Article history: Received 21 December 2016 Received in revised form 21 April 2017 Accepted 13 May 2017 Available online 15 May 2017 Keywords: N2O modeling Denitrification on intracellular polymers Electron competition A2O process Model improvement
a b s t r a c t Competition for electrons among different steps of denitrification on intracellular polymers (XSTO) plays a significant role in nitrous oxide (N2O) accumulation in the biological nitrogen removal process. In this work, this electron competition was considered in a mathematical model to predict N2O production in anaerobic/anoxic/oxic sequencing batch reactors (A2O-SBR) for the first time. The affinity constant for intracellular polymers of heterotrophs (K STO ) that was used in previously published models was divided into four affinity constants (K STO;1 , K STO;2 , K STO;3 and K STO;4 ) to represent the ability of each denitrification reductase to compete for intracellular polymers. The improved model was calibrated and validated using experimental data from three independent A2O-SBR systems. The results demonstrated that the modeling predictions strongly agreed with the measured data from all experimental tests under various operational conditions. The modeling results indicated that N2O accumulation resulted from the more rapid decline of the N2O reduction rate than the nitrite reduction rate for the inadequate XSTO in these A2O-SBR systems. The modeling results also suggested that distinguishing affinity constants for intracellular polymers during the four-step denitrification felicitously described a different XSTO distribution in each reduction step, thereby better predicting nitrogen dynamics and N2O production in A2O processes than the published model. The improved model is therefore a preferable tool to gain insight into N2O accumulation in A2O processes. Ó 2017 Elsevier B.V. All rights reserved.
⇑ Corresponding author at: School of Environmental Science and Engineering, Chang’an University, Xi’an 710064, Shaanxi, China. E-mail address: [email protected] (J. Zhao). http://dx.doi.org/10.1016/j.cej.2017.05.082 1385-8947/Ó 2017 Elsevier B.V. All rights reserved.
X. Ding et al. / Chemical Engineering Journal 325 (2017) 386–395
1. Introduction Nitrous oxide (N2O) is one of the most important greenhouse gases. Although its proportion in total greenhouse gas emissions is only 0.03%, it has a global warming potential more than 300-fold greater than carbon dioxide (CO2) [1]; it significantly contributes to the carbon footprint of greenhouse gas emissions [2]. Measurements have demonstrated that substantial amounts of N2O can be emitted from lab-scale experiments and full-scale wastewater treatment plants (WWTPs) [2] during both autotrophic nitrification and heterotrophic denitrification processes [3–6]. Low dissolved oxygen (DO) concentrations, high nitrite accumulations during nitrification and denitrification processes, and a limited availability of biodegradable organic compounds during denitrification are the most significant causes leading to N2O accumulation and emission in biological nitrogen removal (BNR) processes [2,4,7]. Moreover, when carbon sources are limited or carbon/nitrogen (C/N) ratios are low [8,9], the electron supply cannot meet the electron demand for complete denitrification. Under these circumstances, electron competition may occur among four denitrification reductases [10]. Pan et al. [10] also revealed that competition for electrons occurs under carbon-limited as well as carbon-abundant conditions. Furthermore, the slower degradation of polyhydroxybutyrate (PHB) cannot provide adequate electrons for denitrification. Therefore, electron competition likely also occurs when PHB is used as an electron donor [2,7]. Specifically, under electron-competitive conditions, nitrate reductase may have great advantages in capturing electrons when compared to nitrite and N2O reductases [2], which easily leads to nitrite and N2O accumulation [11,12]. As N2O reduction is the last step of denitrification, N2O reductase cannot gain enough electrons, resulting in N2O production during denitrification [2,11,13]. Additionally, some environmental conditions, such as pH, temperature, carbon source and free nitrous acid (FNA) inhibition, can intensify electron competitions and eventually lead to N2O accumulation [10,14]. Therefore, the mechanisms of electron completion on N2O production in the denitrification process are of great importance and deserve more attention in research. Mathematical modeling has been demonstrated to be helpful for testing the mechanisms of pollutant removal in wastewater treatment [15]. Firstly, the Activated Sludge Model for Nitrogen (ASMN) proposed by Hiatt and Grady [16] was successfully developed to predict N2O production by describing heterotrophic denitrification as a four-step process. Subsequently, the Activated Sludge Model for Indirect Coupling of Electrons (ASM-ICE) was developed to represent the electron competition among four denitrification steps [17,18]. However, both the four-step ASMN model and the ASM-ICE model did not consider the role of intracellular polymers in N2O production in denitrification processes. Afterward, the relations between denitrification on intracellular polymers (XSTO, an internal cell storage product of heterotrophic organisms [15]) and N2O accumulation were investigated in several published models [19–21]. These models satisfactorily described XSTO synthesis/consumption, nitrogen removal, and N2O production in denitrification on intracellular polymers, denitrifying phosphorus removal and anaerobic/oxic/anoxic (AOA) processes. However, due to the lack of recognition of the electron competition in denitrification, none of these models considered the competition for electrons in denitrification processes with the intracellular polymers being the sole electron donors. More information should be provided to explore the reaction kinetics for the models aiming to predict N2O production during denitrification on intracellular polymers. The anaerobic/anoxic/oxic (A2O) process, in which phosphorus and nitrogen can be removed simultaneously [22,23], is the most
387
commonly used process in wastewater treatment [24]. In the WWTPs of China, the A2O process accounted for the biggest design treatment capacity of 33.2% [25]. The A2O process undergoes alternating anaerobic/anoxic/oxic conditions to achieve internal cell product storage/consumption and nitrogen removal. Firstly, the readily biodegradable organics (SS) are stored as intracellular polymers by heterotrophic bacteria under anaerobic conditions. Next, nitrate (NO 3 ) is reduced to nitrite (NO2 ), then nitric oxide (NO), N2O, and finally nitrogen gas (N2) in the anoxic stage, using the intracellular polymers as electron donors. Nitrate reductase (Nar), nitrite reductase (Nir), NO reductase (Nor) and N2O reductase (Nos) are the four functional enzymes in this four-step denitrification process [26]. During the aerobic stage, ammonium (NH+4) is successively oxidized to nitrite and nitrate by ammonia oxidizing bacteria (AOB) and nitrite oxidizing bacteria (NOB), respectively. A significant amount of nitrite could accumulate under oxygenlimited conditions. Due to the utilization of the aerobically synthesized PHB for denitrification in the anoxic stage, N2O could be emitted from WWTPs running the A2O process [27]. Moreover, N2O emissions from the full-course BNR processes operated at the oxidation ditch (OD), University of Cape Town (UCT), AO and AOA modes have been previously modeled [19,28–30], but N2O production in an A2O system has never been modeled. Since the application of the A2O process is popular in wastewater treatment, modeling efforts should begin now. In this work, an improved model, in which the competition for storage polymers among four denitrification reductases was firstly introduced, was proposed to investigate the competition for electrons and the mechanisms of N2O production during denitrification in A2O processes. The improved model was verified using experimental data from three independent A2O-SBR systems with different operation characteristics. The new model is expected to be beneficial for improving the prediction of N2O production in A2O processes. 2. Materials and methods 2.1. Model description The mechanisms of XSTO synthesis/consumption and nitrogen conversion dynamics in the A2O process are the same as that in the AOA process. Therefore, the published model for predicting N2O production in AOA systems (the published model for short [19]) was employed in this work. The component definitions, stoichiometry and composition matrix, kinetic rate expression matrix of the published model are presented in Supplementary Material Tables S.1–S.3. The published model by Ding et al. [19], as well as other reported models describing N2O production during denitrification on intracellular polymers [20,21], did not consider the electron competition among the four reduction steps under starving conditions [2,3,11,13]. The affinity constants (K STO ) regarding storage polymers (electron donor) were uniformly used in all four denitrification steps [19–21]. This limitation potentially restrains the model application for N2O prediction in endogenous denitrification processes. In this work, the key kinetic parameter of K STO was divided into four independent affinity constants, i.e., K STO;1 , K STO;2 , K STO;3 and K STO;4 , to describe the distinct electron capturing ability of each denitrification reductase. These four affinity constants, which govern the distributions of intracellular polymers, were introduced into the denitrification process in this study for the first time. Modifications of the kinetic rate expressions (R5.1, R5.2, R5.3 and R5.4) in the published models are presented in Table 1. The parameter values of K STO;1 , K STO;2 , K STO;3 and K STO;4 were calibrated using the experimental data from the anaerobic/anoxic/oxic
388
X. Ding et al. / Chemical Engineering Journal 325 (2017) 386–395
Table 1 Modified kinetic rate expressions in the improved model. Process
Process rate equation
5.1
K S;1 I;H;NO2 ;1 X STO =X H 3 lH;STO;NO;1 K H;OH;OþS2 O K H;NONOþS K STO;1 þX STO =X H K S;1 þSS K I;H;NO ;1 þSNO X H NO
K
S
2
2
K
K
3
3
2
S
2
K
5.2
K S;2 I;H;NO2 ;2 X STO =X H 2 lH;STO;NO;2 K H;OH;OþS2 O K H;NONOþS K STO;2 þX STO =X H K S;2 þSS K I;H;NO ;2 þSNO X H NO
5.3
K S;3 I;H;NO2 ;3 =X H lH;STO;NO;3 K H;OH;OþS2 O K H;NOSNOþSNO K STO;3XSTO þX STO =X H K S;3 þSS K I;H;NO ;3 þSNO X H
2
2
2
2
2
K
2
5.4
2
K
2
K
2
S
2
K
K S;4 I;H;NO2 ;4 =X H lH;STO;NO;4 K H;OH;OþS2 O K H;N NO2þSO N O K STO;4XSTO þX STO =X H K S;4 þSS K I;H;NO ;4 þSNO X H 2
2
2
2
2
2
sequencing batch reactors (A2O-SBR) from the literature. The kinetic rate expressions of autotrophic bacteria were the same as the published model. 2.2. Experimental data Data from three experimental cases studying N2O production in A2O-SBR systems [11,13,31] were employed to test the simulation capability of the improved model. The operational characteristics of these modeling cases are illustrated in Table 2. 2.2.1. Experimental data from Case 1 [13] In this study, a sealed laboratory-scale A2O-SBR (7.5 L) was used to investigate N2O emission during denitrifying phosphorus removal. The reactor was operated with three cycles per day; each cycle included a 90 min anaerobic phase, a 210-min anoxic phase and a 30 min aerobic phase with DO concentrations at 2–4 mg O2 L1. In each cycle, 5.5 L of synthetic wastewater, which contained 448.46 mg of CH3COONa (350 mg of chemical oxygen demand (COD)) per liter and other nutrient elements, was fed into the SBR. At the end of the anaerobic phase, 100 mL of KNO3 solution was added to the reactor, providing an initial NO 3 concentration of 35 mg N L1 for anoxic denitrification. During the steady-state operation, three batch experiments were carried out in the sealed batch reactors (1 L) to explore the effects of carbon source shock on N2O production during the denitrifying phosphorus removal process. Among them, the batch test investigating acetate (HAc) shock on N2O production was operated with the same carbon load and the same initial NO 3 concentration as the parent SBR. More detailed experimental setup and analytical methods can be found in Wang et al. [13]. A typical cycle-running study in the parent SBR and a batch test with acetate as the sole carbon source were employed to calibrate and validate the improved model. 2.2.2. Experimental data from Case 2 [31] This study aimed to evaluate N2O production and denitrification performance of the denitrifying phosphorus removal process under different anaerobic reaction times (AnRTs). The parent laboratoryscale SBR (7.5 L) was operated in the A2O mode, where one standard running cycle was 8 h, containing a 120 min anaerobic period,
a 210 min anoxic period and a 30 min aerobic period. Synthetic wastewater with mixed acetate and propionate (in the molar ratio 3:1) as carbon sources was fed into the SBR at the beginning of the anaerobic period, while the KNO3 solution was added to the reactor at the beginning of the anoxic period. The initial COD concentration was approximately 220 mg COD L1, and the NO 3 concentration was 50 ± 5 mg N L–1 in the reactor. The DO concentration was approximately 2–3 mg O2 L1 in the aerobic phase. Three batch tests investigating the short-term effect of AnRT (i.e., 90 min in Run 1, 120 min in Run 2, 150 min in Run 3) on N2O production during denitrifying phosphorus removal were conducted using three sealed batch reactors (1 L). The synthetic wastewater, the KNO3 solution and the adding methods were the same as in the parent SBR. More detailed experimental setup and analytical methods can be found in Guo et al. [31]. Three batch tests in this modeling case were adopted to validate the improved model structure and parameter values. 2.2.3. Experimental data from Case 3 [11] A laboratory-scale A2O-SBR (7.5 L) was used to investigate PHB formation, N2O production and denitrifying phosphorus removal performance. The DO level in the reactor was controlled within 2–4 mg O2 L1 during the aerobic stage. The experiment consisted of four running phases, of which phase I was the longest. During phase I, the sealed reactor was operated for three cycles per day, with each cycle consisting of a 90 min anaerobic reaction, a 210 min anoxic reaction, and a 30 min aerobic reaction. In each cycle, 5.5 L of synthetic wastewater with HAc as the sole carbon source (approximately 400 mg COD L1) was fed into the reactor and 100 mL of KNO3 solution was added to the reactor, resulting 1 in an initial NO at the start of the anoxic 3 concentration of 40 NL period. More detailed experimental setup and analytical methods can be found in Wang et al. [11]. Experimental data in typical cycles (Cycle 77 and Cycle 119) of phase I were used to validate the model. 2.3. Model calibration and validation Model calibration for optimizing parameter values was performed by fitting the predicted curve to the measured data from the experimental studies. The newly introduced and the most important parameters were estimated in each modeling case. In Case 1, experimental data from the cycle running test were first used to calibrate the model parameters; data from the batch test were then used to validate the calibrated parameter values. In Case 2, experimental data from Run 1 were used to estimate the parameter values, and experimental data from Run 2 and Run 3 were applied to evaluate the improved model. Likewise, the operational results from Cycle 77 and Cycle 119 were used to calibrate and validate the model parameters, respectively. The MATLAB (R2012a) program was selected as the computational platform.
Table 2 Three experimental cases used for model validation. Culture
Carbon source
Nitrogen added
Experimental tests
Case 1
DPAOs(a)
Acetate: 350 mg COD L1
1 NO 3 : 35 mg N L
Case 2
DPAOs(a)
Acetate and propionate (in the molar ratio 3:1): 300 mg COD L1
1 NO 3 : 50 mg N L
Case 3
DPAOs
Acetate: 400 mg COD L1
1 NO 3 : 40 mg N L
(A): cycle running test (B): batch test (A): run 1 with an AnRT(b) of 90 min; (B): run 2 with an AnRT(b) of 120 min; (C): run 3 with an AnRT(b) of 150 min (A): cycle running test on Day 77 (B): cycle running test on Day 119
(a)
DPAOs(a): Denitrifying phosphate accumulating organisms. AnRT(b): Anaerobic reaction time.
389
X. Ding et al. / Chemical Engineering Journal 325 (2017) 386–395
3. Results and discussion 3.1. Model calibration The biological nature of wastewater treatment processes implies that their activated sludge model parameters must be determined according to the local situation [32]. Thus, the most crucial parameters were individually calibrated for each A2O-SBR
study in this work. Competition for electron donors and FNA inhibition have been regarded as the leading causes of N2O accumulation during denitrification [2,14,33]. Consequently, four new affinity constants for the XSTO of four denitrification reductases (K STO;1 , K STO;2 , K STO;3 and K STO;4 ) and four nitrite inhibition constants for NO 3 , NO2 , NO and N2O reductions were calibrated in each modeling case. Simultaneously, the maximum growth rates of heterotrophs in both the aerobic and anoxic stages were also calibrated.
Table 3 Calibrated stoichiometric and kinetic parameters and their values. Symbol
Definition
Unit
Case 1
Case 2
Case 3
lH;STO;O2 lH;STO;NO;1 lH;STO;NO;2 lH;STO;NO;3 lH;STO;NO;4
Aerobic max. growth rate on X STO
d1
3.6(1)
3.6(1)
3.6(1)
Anoxic max. growth rate on X STO for NO 3 reduction
d1
4.5(3)
1.68(2)
1.68(2)
Anoxic max. growth rate on X STO for NO 2 reduction
d1
1.18(3)
0.456(2)
0.576(4)
Anoxic max. growth rate on X STO for NO reduction
d1
3.408(2)
3.408(2)
3.408(2)
Anoxic max. growth rate on X STO for N2O reduction
d
1.36
K STO K STO;1 K STO;2 K STO;3 K STO;4 K I;H;NO2 ;2 K I;H;NO2 ;3 iN;BM
Saturation constant of X STO for aerobic growth Saturation constant of X STO for NO 3 reduction Saturation constant of X STO for NO 2 reduction Saturation constant of X STO for NO reduction Saturation constant of X STO for N2O reduction Nitrite inhibition constant for NO 2 reduction Nitrite inhibition constant for NO reduction N content of microbial organisms
g COD g1 COD g COD g1 COD g COD g1 COD g COD g1 COD g COD g1 COD g N m3 g N m3 g N g1 COD
0.1(3) 0.1(2) 0.1(2) 0.1(2) 0.125(3) 10(2) 10(3) 0.03(3)
1
(3)
(3)
0.82
1.12(3)
0.1(3) 0.1(2) 0.05(3) 0.1(2) 0.112(3) 10(2) 9(3) 0.03(3)
0.1(3) 0.06(3) 0.07(3) 0.1(2) 0.127(3) 8(3) 17(2) 0.05(3)
Source: (1) Zhou et al. [34]; (2) Liu et al. [21]; (3) Estimated in this study; (4) Liu et al. [20].
+ + Fig. 1. Modeling results and measured data in Case 1: (A1) HAc; (A2) NO 3 , NO2 and NH4; (A3) N2O profiles in the cycle running test; (B1) HAc; (B2) NO3 , NO2 and NH4; (B3) N2O profiles in the batch test.
390
X. Ding et al. / Chemical Engineering Journal 325 (2017) 386–395
In total, thirteen parameters were calibrated in three modeling cases, as shown in Table 3. In all three case studies, the calibrated lH;STO;NO;2 (1.18, 0.456 and 0.576 d1) and lH;STO;NO;4 (1.36, 0.82 and 1.12 d1) were lower than lH;STO;NO;1 (4.5, 1.68 and 1.68 d1), suggesting a lower reduction rate of nitrite and N2O than that of nitrate in the three A2O-SBR systems. In addition, since the degradation of XSTO was the direct source of electrons, the newly introduced K STO;1 , K STO;2 , K STO;3 and K STO;4 parameters with respect to XSTO utilizing described the electron capturing abilities of the four denitrification reductases. The calibrated values of the four affinity constants determined the distinct competitivenesses for electrons and electron donors among the four denitrification reductases. The estimated values of K STO;4 (0.125, 0.112 and 0.127 g COD g1 COD) were higher than K STO;1 (0.1, 0.1 and 0.06 g COD g1 COD) and K STO;2 (0.1, 0.05 and 0.07 g COD g1 COD) in all three modeling cases, indicating a weaker affinity for XSTO in N2O reduction com pared with that in NO 3 and NO2 reductions. These results indicated that Nos had a weaker competition for electrons compared to Nar and Nir [26]. Moreover, it has been reported that since the two upstream reduction enzymes are more powerful in competing for electrons compared to the downstream denitrification reductases, the electron priority was favored by Nar and Nir rather than by Nos when the electron supply was limited [10]. Other parameter values not listed in Table 3 were directly obtained from the literature in Ding et al. [19].
3.2. Model validation In Case 1, experimental data from the cycle running test were first used to calibrate the parameter values, in terms of HAc, + NO 3 , NO2 , NH4 and N2O dynamics. The model predictions and measured data in the experiment are illustrated in Fig. 1 A. During the anaerobic period, HAc was removed through XSTO synthesis in the microbial cell. When the nitrate solution was added to the reactor, NO 3 was reduced to NO2 , and a small amount of N2O was produced during the first 90 min of the anoxic stage. After NO 3 was completely reduced, NO2 was further reduced to N2 and N2O. As a result, N2O consistently increased along with NO 2 1 reduction, until NO . At the end of 2 was below to 7.5 mg N L the anoxic phase, when the low nitrite concentration weakened the FNA inhibition on N2O reduction, N2O was reduced to N2, resulting in a decrease in N2O emissions. Due to the N2O production from NH+4 oxidation, N2O emissions slightly increased in the aeration stage (Fig. 1A3). The developed model captured these trends well. After, experimental data from the batch test of Case 1 were used to validate the calibrated parameter values. As Fig. 1B indicated, the profiles of the variables in the batch test were similar to the cycle running result, except for the different N2O production caused by the different operation regimes of the two tests. Since N2O mainly produced during the anoxic period in these A2OSBR systems [11,13,31], the model structure and model parameters
+ + Fig. 2. Modeling results and measured data in Case 2: (A1) Ss; (A2) NO 3 , NO2 and NH4; (A3) N2O profiles in Run 1; (B1) Ss; (B2) NO3 , NO2 and NH4; (B3) N2O profiles in Run 2; + (C1) Ss; (C2) NO 3 , NO2 and NH4; (C3) N2O profiles in Run 3.
X. Ding et al. / Chemical Engineering Journal 325 (2017) 386–395
relating to AOB and NOB were directly adopted from the published model [19]. The modeling results were well matched with the experimental data in this validation case. In Case 2, experimental data from three batch tests were used to further calibrate and validate the improved model (Fig. 2). A set of + SS, NO 3 , NO2 , NH4 and N2O data from Run 1 were first used to calibrate the model parameters. Then, two sets of experimental data from Run 2 and Run 3 were used to evaluate the parameter values. Slightly unlike the reaction processes in Case 1, NO 3 was reduced slowly, with a slow and continual increase in NO 2 accumulation and a progressive growth in N2O production in all three batch tests. The difference in nitrogen dynamics between Case 2 and Case 1 was possibly due to the different carbon sources used in the experiment (Table 2). The results indicated that when the AnRT was extended from 90 to 150 min, the nitrite accumulations increased, strengthening the FNA inhibition on denitrification. Accordingly, the ratio of N2O production to nitrate removal exhibited a slight increase. The model predictions were in good agreement with the measured trends in all three runs. In Case 3, two cycling operation results were used to test the ability of the model to predict N2O production in A2O processes (Fig. 3). The experimental setups in Case 3 were similar to those in Case 1, except for the different carbon loads and nitrate loads in the reactors (Table 2). As a result, the profiles of HAc, NO 3, + NO 2 , NH4 and N2O in Case 3 were similar to those in Case 1. The modeling results satisfactorily agreed with the experimental data
391
from both Cycle 77 and Cycle 119 tests during the model calibration and validation stages. The parameter values that provided the optimum model fittings with the measured data in the three modeling cases are summarized in Table 3 (calibrated in this work) and Supplementary Material Table S.4 (obtained from the literature). The R2 values of the three modeling studies are provided in Table 4. The average value of R2 was 0.9245, indicating that the established model structure and parameter values are robust in their ability to predict HAc/ 2 + SS, NO 3 NO2 , NH4 and N2O profiles in A O-SBR systems under different operational conditions. 3.3. Comparison of modeling results with the published model To test the advantage of the improved model for N2O modeling in A2O processes, the modeling results provided by the improved model and the published model [19] for Case 3 were compared in Fig. 4. During the prediction performed by the published model, experimental data from Cycle 77 were used to calibrate the model parameters, while experimental data from Cycle 119 were used to validate the estimated parameter values. When the established K STO value was 0.1 g COD g1 COD for the published model, the estimated lH;STO;NO;1 , lH;STO;NO;2 , lH;STO;NO;3 and lH;STO;NO;4 values were 1.68 d1, 0.576 d1, 3.408 d1 and 0.93 d1 for the four reduction steps, respectively. Other parameter values calibrated for the published model were the same as the improved model (Table 3). The
+ + Fig. 3. Modeling results and measured data in Case 3: (A1) HAc; (A2) NO 3 , NO2 and NH4; (A3) N2O profiles in Cycle 77; (B1) HAc; (B2) NO3 , NO2 and NH4; (B3) N2O profiles in Cycle 119.
392
X. Ding et al. / Chemical Engineering Journal 325 (2017) 386–395
Table 4 R2 values of the prediction results for three modeling cases. Items
NO 3 NO 2 NH+4 HAc/SS N2O
Case 1
Case 2
Case 3
Cycle running test
Batch test
Run 1
Run 2
Run 3
Cycle 77
Cycle 119
0.9564 0.9556 0.9816 0.8460 0.9469
0.9548 0.9471 0.9879 0.6698 0.9628
0.9720 0.9706 0.9075 0.7692 0.9720
0.9854 0.9861 0.9191 0.8751 0.9889
0.9977 0.9915 0.8831 0.8660 0.9838
0.9708 0.9507 0.8516 0.9640 0.9920
0.8587 0.8435 0.8452 0.8218 0.9824
Fig. 4. Comparison of predicted results by the improved model (solid lines) and the published model (dashed lines) for Case 3: (A1) NO 3 ; (A2) NO2 and (A3) N2O profiles in Cycle 77; (B1) NO 3 ; (B2) NO2 and (B3) N2O profiles in Cycle 119.
results indicated that both models captured the trends of NO 3 , NO2 and N2O variations that occurred in Case 3. However, the fitting errors between the measured data and the predicted data by the published model were relatively larger than that predicted by the improved model. The improved model, by contrast, more success fully reproduced the NO 3 , NO2 and N2O profiles observed in Case 3. Likely, this difference was due to the higher K STO;2 value of 0.1 g COD g1 COD in the published model compared to that in the improved model (0.07 g COD g1 COD), as the XSTO consumption proportions for NO 2 reduction predicted by the published model (11.23% in Cycle 77 and 11.21% in Cycle 119) were lower than those predicted by the improved model (11.96% in Cycle 77 and 11.95% in Cycle 119), as shown in Fig. 5. This was the origin that resulted in a slower rate of nitrite reduction and a larger accumulation of N2O in the modeling results given by the published model, compared to the experimentally observed data. It was suggested that considering the electron competition among Nar, Nir, Nor and Nos by distinguishing the affinity constants for XSTO of
the four denitrification reductases was beneficial to predict the N2O production during denitrification in A2O systems. Thus, competition for XSTO among the four denitrification reductases should be incorporated into the four-step reduction process to correctly model N2O accumulation during denitrification on intracellular polymers. 3.4. Distribution of XSTO consumption during denitrification As N2O mainly accumulated during the anoxic period in these A2O-SBR systems [11,13,31], the consumption of the anaerobically synthesized XSTO during denitrification is of great importance. Based on differing affinities of the four denitrification reductases for XSTO, the distributions of XSTO consumption among the different reduction steps were quantitatively investigated, while attempting to represent the electron distribution during denitrification (Fig. 6). The results indicated that most of the XSTO was consumed during NO 3 reduction; thus, most electron fluxes were captured by Nar
X. Ding et al. / Chemical Engineering Journal 325 (2017) 386–395
Fig. 5. Comparison of distributions of XSTO consumption predicted by the improved model and the published model for Case 3: ‘‘improved” and ‘‘published” in the abscissa refer to the improved model and the published model, respectively.
to denitrify nitrate to nitrite in all three A2O-SBR systems. On the contrary, Nir and Nos gained little XSTO during nitrite reduction and N2O reduction in all three cases. Since Nar receives electrons directly from the pool in the first echelon, while Nir and Nos receive their electrons from the pool in the second echelon
393
[18,35], the electrons provided by XSTO degradation were first captured by Nar for nitrate reduction [2,13]. As Nar and Nor captured more electrons from more XSTO consumption, the denitrification rates for NO 3 and NO reduction were much faster than those for NO 2 and N2O reduction in the three modeling cases (Fig. 7), resulting in a trace amount of NO production (data not shown) and an 2 accumulation of NO 2 and N2O in these A O-SBR systems. Fig. 7 showed the NO and N O reduction rates in the three 2 2 modeling cases. It was found that both the NO 2 and N2O reduction rates declined with time, which can be described by the exponential function. By comparing the values of the function parameters, it was revealed that N2O reduction rates decreased more rapidly than NO 2 reduction rates in all modeling tests. This suggested that the N2O reduction rates (rN2 O ) declined much more rapidly than the NO ) with the consumption of XSTO, resulting 2 reduction rates (r NO 2 in the apparent N2O accumulation in these validation cases. It further confirmed that Nir is more competent at capturing electrons than Nos when the electrons available for denitrification was inadequate [2,10]. Moreover, in all three modeling cases, XSTO consumption distributions among Nar, Nir, Nor and Nos did not vary significantly with operation regimes (cycle running or batch test in Case 1), AnRTs (90 min, 120 min or 150 min in Case 2), or carbon source (HAc in Case 1 and Case 3 or HAc + Pro in Case 2), and remained at approximately 44.31%, 15.13%, 22.05% and 18.52%, respectively. Meanwhile, the N2O production factors (N2O production/NO 3 removed) were different at the same XSTO distributions in each modeling case (Fig. 6). This difference was possibly due to the diverse differences between the rN2 O and rNO2 decline speeds, which might be attributed to the above-mentioned different operating conditions in one modeling case resulting in the slightly different amounts of XSTO production and consumption. This suggested that N2O accumulation in A2O-SBR systems was caused by not only XSTO distributions but also XSTO synthesis as determined by operational conditions.
Fig. 6. Predicted XSTO consumption distributions (A) and measured N2O production factors (B) during denitrification in three modeling cases.
394
X. Ding et al. / Chemical Engineering Journal 325 (2017) 386–395
Fig. 7. Predicted denitrification rates for NO 3 reduction, NO2 reduction, NO reduction and N2O reduction in three modeling cases: (A1) the cycle running test in Case 1; (A2) the batch test in Case 1; (B1) Run 1 in Case 2; (B2) Run 2 in Case 2; (B3) Run 3 in Case 2; (C1) Cycle 77 in Case 3; (C2) Cycle 119 in Case 3.
4. Conclusions In this work, an improved model for A2O processes was proposed to predict nitrogen dynamics and N2O accumulation in both cycle running tests and batch tests. The results demonstrated that the improved model was capable of reproducing the measured data in three A2O-SBR cases. N2O accumulation resulted from the more rapid decline of the N2O reduction rate than the nitrite reduction rate when the XSTO was inadequate during denitrification. The results suggested that the improved model is favorable over the published models in describing N2O production under electron-competitive conditions. The improved model is expected to provide more information for understanding N2O production and electron distribution in A2O processes.
Author disclosure statement The authors declare that they have no conflict of interest. Acknowledgments This work was supported by the Shaanxi Province Science & Technology Development Program (Grant No. 2014K15-03-02) and the National Natural Science Foundation of China (Grant No. 51308050). Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.cej.2017.05.082.
X. Ding et al. / Chemical Engineering Journal 325 (2017) 386–395
References [1] [2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15] [16] [17]
[18]
IPCC, Climate Change 2013: The Physical Science Basis, 2013, Stockholm, Sweden. M.J. Kampschreur, H. Temmink, R. Kleerebezem, M.S.M. Jetten, M.C.M. van Loosdrecht, Nitrous oxide emission during wastewater treatment, Water Res. 43 (2009) 4093–4103. C. Li, J. Zhang, S. Liang, H.H. Ngo, W. Guo, Y. Zhang, Y. Zou, Nitrous oxide generation in denitrifying phosphorus removal process: main causes and control measures, Environ. Sci. Pollut. R. 20 (2013) 5353–5360. P. Wunderlin, J. Mohn, A. Joss, L. Emmenegger, H. Siegrist, Mechanisms of N2O production in biological wastewater treatment under nitrifying and denitrifying conditions, Water Res. 46 (2012) 1027–1037. R.L. Meyer, R.J. Zeng, V. Giugliano, L.L. Blackall, Challenges for simultaneous nitrification, denitrification, and phosphorus removal in microbial aggregates: mass transfer limitation and nitrous oxide production, FEMS Microbiol. Ecol. 52 (2005) 329–338. W. Jia, J. Zhang, H. Xie, Y. Yan, J. Wang, Y. Zhao, X. Xu, Effect of PHB and oxygen uptake rate on nitrous oxide emission during simultaneous nitrification denitrification process, Bioresource Technol. 113 (2012) 232–238. S. Schalk-Otte, R.J. Seviour, J.G. Kuenen, M.S.M. Jetten, Nitrous oxide (N2O) production by Alcaligenes faecalis during feast and famine regimes, Water Res. 34 (2000) 2080–2088. H. Itokawa, K. Hanaki, T. Matsuo, Nitrous oxide production in high-loading biological nitrogen removal process under low COD/N ratio condition, Water Res. 35 (2001) 657–664. H.J. Lu, K. Chandran, Factors promoting emissions of nitrous oxide and nitric oxide from denitrifying sequencing batch reactors operated with methanol and ethanol as electron do. . ., Biotechnol. Bioeng. (2010). Y. Pan, B. Ni, P.L. Bond, L. Ye, Z. Yuan, Electron competition among nitrogen oxides reduction during methanol-utilizing denitrification in wastewater treatment, Water Res. 47 (2013) 3273–3281. Y. Wang, J. Geng, Z. Ren, W. He, M. Xing, M. Wu, S. Chen, Effect of anaerobic reaction time on denitrifying phosphorus removal and N2O production, Bioresour. Technol. 102 (2011) 5674–5684. Y. Zhou, M. Lim, S. Harjono, W.J. Ng, Nitrous oxide emission by denitrifying phosphorus removal culture using polyhydroxyalkanoates as carbon source, J. Environ. Sci. (China) 24 (2012) 1616–1623. Y. Wang, J. Geng, G. Guo, C. Wang, S. Liu, N2O production in anaerobic/anoxic denitrifying phosphorus removal process: the effects of carbon sources shock, Chem. Eng. J. 172 (2011) 999–1007. Y. Zhou, M. Pijuan, R.J. Zeng, Z. Yuan, Free nitrous acid inhibition on nitrous oxide reduction by a denitrifying-enhanced biological phosphorus removal sludge, Environ. Sci. Technol. 42 (2008) 8260–8265. M. Henze, W. Gujer, T. Mino, M.C.M. van Loosdrecht, Activated Sludge Models ASM1, ASM2, ASM2d and ASM3, IWA Publishing, London, UK, 2000. W.C. Hiatt, C.P.L. Grady, An updated process model for carbon oxidation, nitrification, and denitrification, Water Environ. Res. 80 (2008) 2145–2156. Y. Pan, B. Ni, H. Lu, K. Chandran, D. Richardson, Z. Yuan, Evaluating two concepts for the modelling of intermediates accumulation during biological denitrification in wastewater treatment, Water Res. 71 (2015) 21–31. Y. Pan, B.J. Ni, Z. Yuan, Modeling electron competition among nitrogen oxides reduction and N2O accumulation in denitrification, Environ. Sci. Technol. 47 (2013) 11083–11091.
395
[19] X. Ding, J. Zhao, B. Hu, Y. Chen, G. Ge, X. Li, S. Wang, K. Gao, X. Tian, Mathematical modeling of nitrous oxide production in an anaerobic/oxic/ anoxic process, Bioresour. Technol. 222 (2016) 39–48. [20] Y Liu, L. Peng, J. Guo, X. Chen, Z. Yuan, B. Ni, Evaluating the role of microbial internal storage turnover on nitrous oxide accumulation during denitrification, Sci. Rep. (2015). [21] Y. Liu, L. Peng, X. Chen, B. Ni, Mathematical modeling of nitrous oxide production during denitrifying phosphorus removal process, Environ. Sci. Technol. 49 (2015) 8595–8601. [22] F. Fang, B. Ni, W. Li, G. Sheng, H. Yu, A simulation-based integrated approach to optimize the biological nutrient removal process in a full-scale wastewater treatment plant, Chem. Eng. J. 174 (2011) 635–643. [23] T. Pai, Modeling nitrite and nitrate variations in A2O process under different return oxic mixed liquid using an extended model, Process Biochem. 42 (2007) 978–987. [24] Y. Peng, X. Wang, B. Li, Anoxic biological phosphorus uptake and the effect of excessive aeration on biological phosphorus removal in the A2O process, Desalination 189 (2006) 155–164. [25] L. Jin, G. Zhang, H. Tian, Current state of sewage treatment in China, Water Res. 66 (2014) 85–98. [26] C. Li, T. Wang, N. Zheng, J. Zhang, H.H. Ngo, W. Guo, S. Liang, Influence of organic shock loads on the production of N2O in denitrifying phosphorus removal process, Bioresource Technol. 141 (2013) 160–166. [27] Y. Wang, D. Zhou, W. Zhao, L. Ye, L. Li, X. Tan, C. Song, Nitrous oxide emissions from biological wastewater treatment plants: current status and challenges, Acta Sci. Circum. 34 (2014) 1079–1088. [28] B. Ni, L. Ye, Y. Law, C. Byers, Z. Yuan, Mathematical modeling of nitrous oxide (n2o) emissions from full-scale wastewater treatment plants, Environ. Sci. Technol. 47 (2013) 7795–7803. [29] B. Ni, Y. Pan, B. van den Akker, L. Ye, Z. Yuan, Full-scale modeling explaining large spatial variations of nitrous oxide fluxes in a step-feed plugflow wastewater treatment reactor, Environ. Sci. Technol. 49 (2015) 9176–9184. [30] M. Spérandio, M. Pocquet, L. Guo, B. Ni, P.A. Vanrolleghem, Z. Yuan, Evaluation of different nitrous oxide production models with four continuous long-term wastewater treatment process data series, Bioprocess. Biosyst. Eng. 39 (2016) 493–510. [31] G. Guo, Y. Wang, C. Wang, H. Wang, M. Pan, S. Chen, Short-term effects of excessive anaerobic reaction time on anaerobic metabolism of denitrifying polyphosphate-accumulating organisms linked to phosphorus removal and N2O production, Front. Environ. Sci. Eng. 7 (2013) 616–624. [32] P.A. Vanrolleghem, H. Spanjers, B. Petersen, P. Ginestet, I. Takacs, Estimating (combinations of) activated sludge and components by respirometry, Water Sci. Technol. 39 (1999) 195–214. [33] J.H. Ahn, S. Kim, H. Park, B. Rahm, K. Pagilla, K. Chandran, N2O emissions from activated sludge processes, 2008–2009: Results of a national monitoring survey in the United States, Environ. Sci. Technol. 44 (2010) 4505–4511. [34] M. Zhou, J. Gong, C. Yang, W. Pu, Simulation of the performance of aerobic granular sludge SBR using modified ASM3 model, Bioresour. Technol. 127 (2013) 473–481. [35] D. Richardson, H. Felgate, N. Watmough, A. Thomson, E. Baggs, Mitigating release of the potent greenhouse gas N2O from the nitrogen cycle – could enzymic regulation hold the key?, Trends Biotechnol 27 (2009) 388–397.