Article pubs.acs.org/est
Mathematical Modeling of Nitrous Oxide Production during Denitrifying Phosphorus Removal Process Yiwen Liu, Lai Peng, Xueming Chen, and Bing-Jie Ni* Advanced Water Management Centre, The University of Queensland, St. Lucia, Brisbane, QLD 4072, Australia S Supporting Information *
ABSTRACT: A denitrifying phosphorus removal process undergoes frequent alternating anaerobic/anoxic conditions to achieve phosphate release and uptake, during which microbial internal storage polymers (e.g., Polyhydroxyalkanoate (PHA)) could be produced and consumed dynamically. The PHA turnovers play important roles in nitrous oxide (N2O) accumulation during the denitrifying phosphorus removal process. In this work, a mathematical model is developed to describe N2O dynamics and the key role of PHA consumption on N2O accumulation during the denitrifying phosphorus removal process for the first time. In this model, the four-step anoxic storage of polyphosphate and four-step anoxic growth on PHA using nitrate, nitrite, nitric oxide (NO), and N2O consecutively by denitrifying polyphosphate accumulating organisms (DPAOs) are taken into account for describing all potential N2O accumulation steps in the denitrifying phosphorus removal process. The developed model is successfully applied to reproduce experimental data on N2O production obtained from four independent denitrifying phosphorus removal study reports with different experimental conditions. The model satisfactorily describes the N2O accumulation, nitrogen reduction, phosphate release and uptake, and PHA dynamics for all systems, suggesting the validity and applicability of the model. The results indicated a substantial role of PHA consumption in N2O accumulation due to the relatively low N2O reduction rate by using PHA during denitrifying phosphorus removal.
■
INTRODUCTION Denitrifying phosphorus removal is a cost-effective and sustainable process to achieve simultaneous phosphorus and nitrogen removal from wastewater.1 This process is based on the enrichment of denitrifying polyphosphate accumulating organisms (DPAOs) through alternating anaerobic/anoxic conditions.2 DPAOs are capable of oxidizing intracellular polymers stored at the anaerobic stage and utilizing nitrate or nitrite (instead of oxygen) as electron acceptors, which provide energy for phosphorus uptake at the anoxic stage.3,4 A denitrifying phosphorus removal process would substantially decrease the aeration demanded for phosphorus uptake and significantly reduce the external carbon source required for denitrification, as well as the excess sludge production.5 Nitrous oxide (N2O) can be produced and accumulated as a significant intermediate of denitrification during the denitrifying phosphorus removal process,6−10 which has raised increasing concerns owing to its potent greenhouse gas effect and its ability to deplete stratospheric ozone.11,12 It has been reported that the amount of N2O production in a denitrifying phosphorus removal system ranged from 2.3% to 21.6% of the influent nitrogen load.13−16 The accumulated N2O during the anoxic phase would be stripped out during subsequent aeration, contributing to significant N2O emission from wastewater treatment.17 Therefore, understanding N2O pro© XXXX American Chemical Society
duction in denitrifying phosphorus removal is of great importance to the overall carbon footprint of the wastewater treatment system. Denitrification processes by DPAOs in denitrifying phosphorus removal systems compulsorily utilize intracellular storage poly-β-hydroxyalkanoates (PHAs) as a carbon source, which is a rate-limiting substrate in terms of regulating microbial growth. The slower degradation nature of PHA may cause substrate competition between the four key denitrifying enzymes, i.e., nitrate (NO3−) reductase (Nar), nitrite (NO2−) reductase (Nir), nitric oxide (NO) reductase (Nor), and N2O reductase (Nos), in the four steps of denitrification (from NO3− to N2 via NO2−, NO and N2O), which might potentially lead to intermediates accumulation (e.g., N2O). Schalk-Otte et al.18 reported that the increase of N2O production from denitrification coincided with the onset of PHA consumption upon COD depletion. Li et al.19 observed dissolved N2O concentrations in the presence of PHAs as the sole carbon source decreased by four times when acetate was added. Zhou et al.13 also suggested that the degradation of PHAs potentially lead to N2O accumulation. These results Received: April 1, 2015 Accepted: June 26, 2015
A
DOI: 10.1021/acs.est.5b01650 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology
Experimental Data for Model Evaluation. Experimental data from four cases with different operational conditions14−16,27 studying N2O dynamics during denitrifying phosphorus removal were used to test the predictive abilities of the developed model. Case I:15 DPAOs were enriched at 22 ± 2 °C in a sealed labscale sequencing batch reactor (SBR) fed with synthetic wastewater mainly containing acetate and phosphate. The SBR was operated under alternating anaerobic (90 min)/anoxic (210 min, initial N-NO3− concentration of 50−60 mg N/L) conditions. Two batch experiments were conducted with this culture in two 2.4-L sealed batch reactors at different anaerobic reaction times, i.e., 90 min (Batch I) and 120 min (Batch II), respectively. The anoxic reaction time was the same for both batch reactors, i.e., 210 min, initiated by adding KNO3 solution to give an initial N-NO3− concentration of about 55 mg N/L. Samples were taken periodically for NO3−, NO2−, PO43−, PHA, and N2O analysis. Case II:16 DPAOs were acclimatized at 20 ± 1 °C in a sealed lab-scale SBR fed with synthetic wastewater mainly containing acetate, propionate, and phosphate and operated under alternating anaerobic (120 min)/anoxic (210 min) conditions. After completion of the anaerobic phase, KNO3 solution was added to initiate the anoxic phase with an initial N-NO3− concentration of 50 ± 5 mg N/L. Two batch tests were performed in two closed reactors with an anaerobic reaction time of 90 min (Batch I) and 150 min (Batch II), respectively, at a constant anoxic (initial N-NO3− concentration of 50 mg N/L) reaction time of 210 min. NO3−, NO2−, PO43−, PHA, and N2O were measured. Case III:14 DPAOs were cultivated at 25−30 °C in a sealed lab-scale SBR fed with synthetic wastewater mainly containing acetate and phosphate and operated under alternating anaerobic (90 min)/anoxic (210 min, initial N-NO 3 − concentration of 35 mg N/L) conditions. The steady-state cycle profile of this culture was used for model evaluation in terms of N2O accumulation during denitrifying phosphorus removal. Furthermore, one batch test was carried out in a sealed reactor with the same conditions as those for the cycle operation. Samples were taken periodically for NO3−, NO2−, PO43−, PHA, and N2O analysis. Case IV:27 DPAOs were enriched at 24 ± 1 °C in a sealed SBR fed with synthetic wastewater mainly containing acetate, propionate, and phosphate and operated under alternating anaerobic (90 min)/anoxic (180 min, initial N-NO 3 − concentration of 40 mg N/L) conditions. Two batch tests were conducted at influent COD concentrations of 240 mg/L (Batch I) and 400 mg/L (Batch II) in 1.4-L batch reactors, respectively, using the same conditions as those for the SBR operation. Samples were taken periodically for NO3−, NO2−, and PO43− analysis. The dissolved N2O concentration was monitored online using a N2O microsensor (Unisense, Denmark). Testing the Predictive Power of the Model. The developed model includes 27 stoichiometric and kinetic parameters as summarized in Table S3 in the SI. About 21 of these model parameter values are well established in previous studies. Thus, literature values were directly adopted for these parameters (SI Table S3). The NO reduction related parameters are beyond the ability of measurement since NO was not added or measured in any tests given its toxicity to bacteria. Indeed, NO reduction is usually prioritized by bacteria to avoid its toxicity; thus, a relatively high value of μDPAO3 and a
indicated PHA utilization might play an essential role in N2O production during the denitrifying phosphorus removal process. Mathematical models are widely applied to predict phosphorus removal during wastewater treatment.20−26 However, comparatively little effort has been dedicated to model the N2O dynamics during the denitrifying phosphorus removal process although considerable amounts of N2O production in such a process have been demonstrated.8,14,15,27 To date, the majority of existing models that have been proposed describe the denitrifying phosphorus removal process as a one-step denitrification21,25 or two-step denitrification,28 without consideration of N 2 O production. Although a three-step denitrifying phosphorus removal model was proposed with the consideration of N2O as an intermediate,24 this model was neither calibrated nor validated with N2O production data from denitrifying phosphorus removal systems. This study aims to develop a new denitrifying phosphorus removal model for describing N2O production in this process, particularly during the anoxic PHA utilization phase. The validity and applicability of the developed model are tested by comparing simulations with experimental data on N2O production from four independent denitrifying phosphorus removal study reports with different experimental conditions.
■
MATERIALS AND METHODS Model Development. The denitrifying phosphorus removal process undergoes frequent alternating anaerobic/ anoxic conditions to achieve phosphate release and uptake.10 In this process, the readily biodegradable substrate is stored in the form of PHAs by DPAOs using the energy generated during the release of polyphosphate under anaerobic conditions. Under the subsequent anoxic conditions, DPAOs uptake phosphate in the form of polyphosphate using the energy obtained from anoxic growth of DPAOs on the stored PHAs. The model developed in this work considered the four-step (from NO3− to N2 via NO2−, NO, and N2O) anoxic phosphorus uptake and four-step anoxic growth on PHAs to describe all potential N 2O accumulation steps in the denitrifying phosphorus removal process. The developed model synthesizes all these relevant reactions and describes the relationships among four biomass groups, DPAOs (XDPAO), PHAs (XPHA), polyphosphate (XPP), and residual inert biomass (XI), and seven soluble compounds, phosphate (SPO4), NO3− (SNO3), NO2− (SNO2), NO (SNO), N2O (SN2O), N2 (SN2), and readily degradable substrate (Ss). The units are g-N m−3 for all nitrogenous species, g-P m−3 for phosphorus species, and g-COD m−3 for other compounds. Three groups of biological processes (see Tables S1 and S2 in the Supporting Information (SI) for the kinetic and stoichiometric matrices) were considered, namely, anaerobic storage of XPHA (Process 1), anoxic storage of XPP (Processes 2−5), and anoxic growth of XDPAO on XPHA (Processes 6−9), the latter two modeled as four sequential denitrifying processes from NO3− to N2 via NO2−, NO, and N2O with individual reaction-specific kinetics. In addition, anoxic endogenous respiration of DPAOs (Processes 10−13) and lysis of the internal storage compounds (polyphosphate and PHAs, Processes 14−21) were also included. Table S3 in the SI lists the definitions, values, units, and sources of all parameters used in the developed model. B
DOI: 10.1021/acs.est.5b01650 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology Table 1. Best-Fit Parameters with 95% Confidence Intervals Describing N2O Accumulation in Four Case Studies Parameters
Case I
Case II
Case III
Case IV
qPHA μDPAO1 μDPAO2 μDPAO4
0.53 ± 0.04 0.070 ± 0.001 0.019 ± 0.001 0.018 ± 0.001
− 0.110 ± 0.012 0.032 ± 0.004 0.016 ± 0.003
− 0.109 ± 0.003 0.016 ± 0.002 0.008 ± 0.001
0.48 ± 0.04 0.110 ± 0.006 0.023 ± 0.001 −
Figure 1. Model evaluation with experimental data from Case I:15 (A) NO3−, NO2−, and N2O profiles from Batch I; (B) PO43− and PHA dynamics from Batch I; (C) NO3−, NO2−, and N2O profiles from Batch II; and (D) PO43− and PHA dynamics from Batch II.
Figure 2. Model evaluation with experimental data from Case II:16 (A) NO3−, NO2−, and N2O profiles from Batch I; (B) PO43− and PHA dynamics from Batch I; (C) NO3−, NO2−, and N2O profiles from Batch II; and (D) PO43− and PHA dynamics from Batch II.
low value of KNO from literature are used in this model (Table S3) to ensure there is no accumulation of NO for all cases. The
remaining four parameters, i.e., rate constant for anaerobic storage of XPHA (qPHA), anoxic growth rate of DPAOs on nitrate C
DOI: 10.1021/acs.est.5b01650 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology
Figure 3. Model evaluation with experimental data from Case III:14 (A) NO3−, NO2−, and N2O profiles from batch test; (B) PO43− and PHA dynamics from batch test; (C) NO3−, NO2−, and N2O profiles from cycle profile; and (D) PO43− and PHA dynamics from cycle profile.
The developed model and calibrated parameter set (Table 1) were then further tested for their ability to predict N2O dynamics in another data set of Case I, i.e., Batch II under a longer anaerobic reaction time (120 min). The model predictions and the experimental results are shown in Figure 1C and 1D. The validation results showed that the model predictions well matched the measured data of NO3−, NO2−, N2O, PO43−, and PHA concentrations in the validation experiment, which supports the validity of the developed model. Model Evaluation with Experimental Data of Case II. The experimental results obtained from Case II were also used to evaluate the developed model in terms of NO3−, NO2−, N2O, PO43−, and PHA dynamics. In this case, experimental data from Batch I (Figure 2A and 2B) were used to obtain key kinetic parameter values related to N2O accumulation during denitrifying phosphorus removal (μDPAO1, μDPAO2, and μDPAO4), and the resulting parameter values were used to simulate profiles for comparison with another data set (Batch II, Figure 2C and 2D) as a validation. To simplify the calibration process, the same parameter value of qPHA as employed in Case I were used without any further calibration. The values of the estimated parameters in this case are listed in Table 1. As shown in Figure 2, NO3−, NO2−, PO43−, and PHA dynamics were similar to Case I, except with higher N2O accumulation (ca. 7.6 mg-N/L). The model captured these trends reasonably well. In addition, a slightly higher N2O accumulation (ca. 8.5 mg-N/L) was predicted in Figure 2C, in contrast to Figure 2A (ca. 7.8 mg-N/L) due to the higher PHA formation, further suggesting that N2O accumulation during denitrifying phosphate removal could be regulated by the anaerobic PHA synthesis process. Model Evaluation with Experimental Data of Case III. The developed model was then evaluated with the NO3−, NO2−, N2O, PO43−, and PHA data from the batch test (as calibration) and cycle study (as validation) in Case III, with
(μDPAO1), anoxic growth rate of DPAOs on nitrite (μDPAO2), and anoxic growth rate of DPAOs on N2O (μDPAO4), which are the key parameters governing the N2O accumulation, are then calibrated using experimental data (SI Table S3). Parameter values were estimated by minimizing the sum of squares of the deviations between the measured data and the model predictions using the secant method embedded in AQUASIM 2.1d.29 For each case, the model was first calibrated with one set of experimental data and then validated through simulation of the N 2 O production for other sets of experimental data (not used for model calibration) with the obtained best-fit parameter values. Parameter estimates for all four case studies are listed in Table 1.
■
RESULTS Model Evaluation with Experimental Data of Case I. The model is first calibrated to describe the experimental data from Batch I in Case I. The predicted nitrate, nitrite, N2O, PO43−, and PHA profiles with the established model are illustrated in Figure 1A and 1B, along with the experimental results. After 90 min of anaerobic phosphate release and PHA storage, nitrate was added to initiate denitrifying phosphorus uptake using stored PHA by DPAOs during the anoxic stage. N2O increased gradually at the beginning, along with the accumulation of nitrite from nitrate reduction (Figure 1A). Afterward, N2O did not show significant change (ca. 2 mg-N/ L), consistent with the relatively stable nitrite concentrations due to the unavailability of PHA in the remaining period (Figure 1A and B). Our model captured all these trends reasonably well. The good agreement between these simulated and measured data supported that the developed model properly captures the relationships among N2O dynamics, nitrogen reduction, phosphate release and uptake, and PHA turnovers for Batch I of Case I. The calibrated parameter values giving the optimum model fittings with the experimental data are listed in Table 1. D
DOI: 10.1021/acs.est.5b01650 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology
Figure 4. Model evaluation with experimental data from Case IV:27 (A) NO3−, NO2−, and N2O profiles from Batch I; (B) PO43− and PHA dynamics from Batch I; (C) NO3−, NO2−, and N2O profiles from Batch II; and (D) PO43− and PHA dynamics from Batch II.
three parameters (μDPAO1, μDPAO2, and μDPAO4) being calibrated (Table 1). There was a higher nitrite accumulation (ca. 20 mgN/L) in this case, leading to a N2O accumulation of 4−6 mgN/L. Our model predictions mostly matched the experimental results in Figure 3, again supporting the validity of the developed N2O model. However, there was a difference between the model predictions and experimental data of PHA from 2.5 h in the cycle study (model validation, Figure 3C), possibly due to the unexpected measurement errors resulting in the abnormal experimental data. Model Evaluation with Experimental Data of Case IV. In the last case, the experimental results obtained with different organic loadings (Batch I for calibration; Batch II for validation) were also used to evaluate the developed model in Figure 4. Three key parameter (qPHA, μDPAO1, and μDPAO2) values were calibrated for this culture (Table 1). The good agreement between simulations and measured results further indicated the developed model is also able to describe the continuous N2O accumulation and subsequent utilization process (Figure 4A and C). Different from other cases, there was no N2O accumulation at the end of the study, accompanied by negligible nitrite accumulation from 3.5 h.
not applicable to describe N 2O dynamics during this process.21,24 In this work, a new mathematical model based on the wellknown activated sludge model (ASM) is developed to describe the N2O production during the denitrifying phosphorus removal process for the first time. The validity of this developed model was confirmed by four independent case studies.14−16,27 The set of best-fit parameter values are shown in Table 1. The parameter values obtained were robust in their ability to predict nitrate, nitrite, N2O, phosphate, and PHA dynamics under different operational conditions, indicating that the developed N2O model is applicable for different wastewater systems. The developed model and the results reported in this work are useful to design and optimize a biological simultaneous phosphate and nitrogen removal process in terms of N2O emission. It should be noted that a 1% increase in N2O emission would induce a 30% increase in the carbon footprint during the wastewater treatment.41 Therefore, the model of this work would be very useful for accurate estimation and possible mitigation of N2O emission from denitrifying phosphorus removal systems. In addition, the developed simple model can be easily integrated with other ASM-based N2O models for nitrification and ordinary denitrification processes to describe overall N2O dynamics in wastewater treatment systems.34,35,42,43 Table 1 shows model parameter values vary across the examined systems, although they are, consistently, within a relatively narrow range. Differences might be due to differences in DPAO communities under different reactor conditions. The estimated value of qPHA is 0.53 h−1, which is higher than the literature reported value (0.25−0.37 h−1),21,26 likely due to a higher anaerobic PHA storage rate for these DPAO cultures. Furthermore, modeling results indicated that a substantial amount of N2O could be accumulated during the denitrifying phosphorus removal process due to the relatively lower N2O reduction rate by DPAOs when utilizing PHA. As shown in
■
DISCUSSION Denitrifying phosphorus removal is a promising process for simultaneous phosphate and nitrogen removal from wastewater.1,6,30 However, recent studies have demonstrated substantial N 2 O accumulation during this process.7,14,15,19,27,31−33 Modeling of N2O production is of significance for understanding and predicting N2O emissions,34−40 which can be a powerful tool for supporting operation optimization and development of mitigation strategies during the denitrifying phosphorus removal process. However, the previously proposed denitrifying phosphorus removal models completely ignored N2O production and are E
DOI: 10.1021/acs.est.5b01650 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology Table 1, in all four case studies μDPAO4 (0.008−0.018 h−1) was much lower than μDPAO1 (0.070−0.110 h−1) and μDPAO2 (0.016−0.032 h−1) (Table 1), suggesting the lower competition capacity for electrons from internal storage polymers of nitrous oxide reductase.8 Therefore, an appropriate anoxic reaction time is required to achieve complete denitrification (N2) for N2O mitigation. In addition, μDPAO4 (0.008−0.018 h−1) is also about one magnitude smaller than the widely used anoxic growth rate on N2O and readily biodegradable carbon sources by ordinary heterotrophic denitrifiers, e.g., 0.134 h−1,34 again supporting the fact that microbial internal storage (e.g., PHAs) is a rate-limiting substrate in terms of regulating DPAO growth and could be a major contributor to N2O production by DPAOs.18 In our model, the potential growth of denitrifying glycogen accumulating organisms (DGAOs) is not considered. This simplification is reasonable, as DPAOs are predominant cultures in all four case studies, while DGAOs are usually undesired communities10 and have been washed out from the studied denitrifying phosphorus removal systems. The possible existence of different groups of DPAOs performing different steps of denitrification24 was not specifically considered for model simplification, which were lumped into the four-step denitrification processes by DPAOs in this model. These assumptions may be revised in the future, if more information on N2O production by DPAOs and DGAOs becomes available. The denitrification processes of ordinary heterotrophic bacteria were also not considered in the current model. This is acceptable due to the fact that the cultures in all cases were operated under the alternating anaerobic/anoxic conditions to enrich DPAOs, which largely limited the growth of ordinary heterotrophic denitrifiers. However, the heterotrophic denitrification could be easily incorporated into the model, if heterotrophic activity would be included in the system. In addition, the inhibition of free nitrous oxide (FNA) on nitrous oxide reductase of DPAOs was not included in the developed model due to the relatively low FNA level (
