Accepted Manuscript Optimization of C/N and current density in a heterotrophic/biofilm-electrode autotrophic denitrification reactor (HAD-BER) Shuang Tong, Nan Chen, Heng Wang, Hengyuan Liu, Chen Tao, Chuanping Feng, Baogang Zhang, Chunbo Hao, Jiaoyang Pu, Jiamin Zhao PII: DOI: Reference:
S0960-8524(14)01233-4 http://dx.doi.org/10.1016/j.biortech.2014.08.117 BITE 13880
To appear in:
Bioresource Technology
Received Date: Revised Date: Accepted Date:
30 June 2014 24 August 2014 26 August 2014
Please cite this article as: Tong, S., Chen, N., Wang, H., Liu, H., Tao, C., Feng, C., Zhang, B., Hao, C., Pu, J., Zhao, J., Optimization of C/N and current density in a heterotrophic/biofilm-electrode autotrophic denitrification reactor (HAD-BER), Bioresource Technology (2014), doi: http://dx.doi.org/10.1016/j.biortech.2014.08.117
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Optimization of C/N and current density in a heterotrophic/biofilm-electrode autotrophic denitrification reactor (HAD-BER) Shuang Tonga,b, Nan Chena,b , Heng Wang b, Hengyuan Liu b, Chen Tao b, Chuanping Fenga,b*, Baogang Zhangb, Chunbo Hao b, Jiaoyang Pu b, Jiamin Zhaob a
Key Laboratory of Groundwater Circulation and Evolution (China University of
Geosciences Beijing), Ministry of Education, Beijing 100083, China b
School of Water Resources and Environment, China University of Geosciences
Beijing, Beijing 100083, China ABSTRACT In this study, central composite design (CCD) and response surface methodology (RSM) were applied to optimize the C/N and current density in a heterotrophic/biofilm-electrode autotrophic denitrification reactor (HAD-BER). Results showed that nitrate could be effectively reduced over a wide range of C/Ns (0.84-1.3535) and current densities (96.8-370.0 mA/m2); however, an optimum C/N of 1.13 and optimum current density of 239.6 mA/m2 were obtained by RSM. Moreover, the HAD-BER performance under the optimum conditions resulted in almost 100% nitrate-N removal efficiency and low nitrite-N and ammonia-N accumulation. Furthermore, under the optimum conditions, H2 generated from water electrolysis matched the CO2 produced by heterotrophic denitrification by stoichiometric calculation. Therefore, CCD and RSM could be used to acquire
*
Corresponding author. Tel.: +86 10 8232 2281; fax: +86 10 8232 1081. E-mail: [email protected] (C. Feng)
optimum operational conditions and improve the nitrate removal efficiency and energy consumption in the HAD-BER. Keywords: Nitrate reduction; HAD-BER; Optimization; Central composite design (CCD); Response surface methodology (RSM) 1.
Introduction Groundwater is one of the main fresh water resources. It has long been used as a
municipal water supply in most countries of the world (Ergas and Reuss, 2001). However, the rapid movement of human civilization towards urbanization, industrialization, and increased agricultural activities has introduced a large amount of nitrate into the environment (Ghafari et al., 2008). In the USA, from 10% to 25% of the groundwater used as drinking water suffers from nitrate concentrations above the maximum allowable contaminant level (10 mg nitrate-N/L) (Liu et al., 2009). In China, the pollution of groundwater with nitrate is even more severe, with nitrate concentrations in groundwater in some rural areas exceeding 130 mg nitrate-N/L (Wan et al., 2009). Nitrate in various sources of groundwater represents a serious risk to the health of humans and animals (Ghafari et al., 2009). Nitrate is reverted to toxic nitrite by microbes in the human body, which reacts with hemoglobin in the blood, converting the hemoglobin to methemoglobin, resulting in methemoglobinemia (Mousavi et al., 2012). Nitrate has also been identified as a cause of gastric cancer (Mirvish, 1977). Accordingly, various agencies have stipulated standards for nitrate-N in drinking water to minimize these health risks: the World Health Organization (WHO) has set a
maximum contaminant level of 10 mg nitrate-N/L (WHO, 2004), and the same value has also been adopted by China (Standards for Drinking Water Quality, GB5749-2006). Hence, various methods have been exploited for nitrate removal, for example, biological processes have high potential for the elimination of nitrate with high stability and reliability (Foglar et al., 2005), which represent a promising solution for controlling nitrate pollution. Denitrification is a microbial process that mitigates nitrate-N pollution by reducing nitrate-N to N2 in hypoxic conditions, utilizing an electron donor such as organic carbon or hydrogen (Schmidt and Clark, 2012). However, traditional hydrogen autotrophic denitrification, using hydrogen as an electron donor, consumes a lot of H2 due to the low H2-delevery rate (Rittmann, 2006), and it has relatively lower efficiency when compared with heterotrophic denitrification (Tang et al., 2012). Whereas, heterotrophic denitrification involves secondary pollution from the added organic carbon source, due to the lack of a carbon source in groundwater (Zhang et al., 2012). Recently, biofilm-electrode reactors (BERs) combining biological and electrochemical methods have become effective nitrate removal devices. BERs are based on hydrogen autotrophic denitrification, which incorporates microorganisms on the cathode, and uses hydrogen produced from water electrolysis as an electron donor for autotrophic denitrification (Zhang et al., 2005). However, traditional BERs consume a lot of electric energy to produce sufficient H2 (Mousavi et al., 2012).
Hence, to improve denitrification efficiency and reduce the consumption of electrical energy, a heterotrophic/biofilm-electrode autotrophic denitrification reactor (HAD-BER) has been developed in previous studies (Zhao et al., 2011; Zhao et al., 2012; Tong et al., 2013). The reactor’s performance was evaluated under varying hydraulic retention times (HRTs) and electric currents (Is). The highest nitrate removal efficiency was achieved at HRT = 8 h and I = 10 mA (Zhao et al., 2011). Likewise, the behaviors of autotrophic, heterotrophic, and autotrophic and heterotrophic denitrification (AD, HD, and HAD) were investigated in the reactor, and the HAD process proved to be superior to single AD and HD (Zhao et al., 2012). It was then demonstrated that the autotrophic bacteria used CO2 generated by heterotrophic processes, representing a synergistic interaction between heterotrophic and autotrophic bacteria in the HAD-BER (Tong et al. 2013). Therefore, it was not necessary to add an inorganic carbon source to the HAD-BER, reducing the cost compared to traditional BERs, which require the addition of an inorganic carbon source such as NaHCO3 (Park et al., 2006). HAD-BERs also greatly improved the efficiency of the reaction, significantly saving electricity (Zhao et al., 2012). Both heterotrophic and biofilm-electrode autotrophic denitrification can be considered two-step reduction processes, in which nitrite is an intermediate byproduct (Park et al., 2008; Ghafari et al., 2009). Nitrite reductase is more sensitive than nitrate reductase to the experimental conditions (Ghafari et al., 2009). For heterotrophic denitrification, a range of factors influence the nitrite accumulation, including organic carbon supply, temperature, oxygen content, and phosphate concentration (Hunter,
2003). At the same time, a deficiency of available carbon might also be the leading factor for nitrite accumulation (Zhang et al., 2012). On the other hand, for biofilm-electrode autotrophic denitrification, a minimum dissolved hydrogen concentration of 0.2 mg/L is a necessity for nitrite reductase, with inhibition at lower concentrations, while nitrate reductase will occur even at lower hydrogen concentrations (0.1mg/L) (Chih et al., 1999). Excessively high hydrogen concentrations have also been reported as another denitrification inhibitor and cause of nitrite accumulation (Islam and Suidan, 1998). Therefore, suitable amounts of carbon and hydrogen must be supplied to cope with the susceptibility of inhibition during the denitrification process. In other words, C/N and current density (as the in situ hydrogen generator) must be maintained at appropriate levels in a HAD-BER or BER. Many researchers have investigated the effects of different parameters and tried to optimize them. For example, Zhao et al. (2011) found the optimal values of single factors (I and HRT) by performing numerous experiments with an intensified BER. Hao et al. (2013) optimized the operating parameters (C/Ns and HRTs) in a three-dimensional BER. The C/N was tested at 0.5, 1.0, 1.5, 2.0, and 3.0, and the HRT at 5, 7, 10, and 12 h. A nitrate removal efficiency of 98.3% was obtained at a C/N ratio of 3.0 and HRT of 7 h. In the studies mentioned above, each individual parameter was evaluated by fixing all other parameters. However, this research approach has some drawbacks, including a large workload and not addressing the interactions between different parameters. The autotrophic denitrification and
heterotrophic denitrification in the HAD-BER influence each other, however, so far, little research has focused on the interaction between C/N and current density. Design of experiments (DOE) and response surface methodology (RSM) are widely used for modeling process parameters. As RSM requires a small number of experiments, it is advantageous over conventional methods. DOEs are suitable for multi-factor experiments to identify correlations between various factors and determine the most favorable conditions for the processes. DOEs include central composite designs (CCDs), Doehlert matrixes (DMs), and Box-Behnken designs (BBDs) (Bingöl et al., 2012). DOEs and RSMs have been employed successfully to optimize the operational parameters in water treatment (Ghafari et al., 2009; Li et al., 2010a). Therefore, in this study, to determine the optimum C/N and current density in the developed HAD-BER, a CCD design method was used to set experimental parameters, and a series of experiments were performed with those experimental parameters. RSM was then employed to determine the optimum conditions (C/N and current density) of the HAD-BER. Then finally, the performance of the HAD-BER was evaluated under those optimum conditions. 2.
Materials and Methods
2.1 Experimental apparatus The HAD-BER used in the current study is similar to one from a previous study (Tong et al., 2013). To keep the HAD-BER at a constant temperature (30 ± 1 ℃), the reactor was kept
in a water bath containing four electric heaters (AT-180, Atman, China). 2.2 Sludge acclimation and synthetic contaminated groundwater preparation A mixed culture containing denitrifying bacteria was originally collected from anaerobic sludge, which was obtained from the Qinghe Sewage Treatment Plant (Beijing, China). Acclimatized denitrifying bacteria, obtained according to a previously developed method (Tong et al., 2013), were used as the inoculums. Hydrogen, provided by water electrolysis, and methanol were used as electron donors for autotrophic and heterotrophic denitrification, respectively. A C/N of 1.25 and current density of 200 mA/m2 were adopted to initiate the HAD-BER (Tong et al., 2013). Biofilm formed gradually and a dark grey color was observed on the cathode module within 30 days. Synthetic contaminated groundwater (per liter of tap water) contained 0.304 g/L NaNO3 and 0.044 g/L KH2PO4, thus the concentration of nitrate-N was 50 mg/L. CH3OH was added as required to maintain the desired C/N. All solid reagents were weighed using an analytical balance (TX223L, Shimadzu, Japan) and liquid reagents were measured via pipettes with specific ranges (20-200μL and 100-1000μL, Nichipet EX-PLUS, Nichiryo, Japan; 0.5-5mL, Research plus, Eppendorf, Germany). All chemical reagents used in the experiments were of analytical grade. 2.3 Experimental design and data analysis Design-Expert (version 8.0) software was used for the statistical design of experiments and data analysis. In this study, CCD and RSM were applied to optimize the operating variables C/N and current density. The ranges of variables were taken
from previous studies (Wang et al., 2009; Zhao et al., 2011; Zhao et al., 2012; Tong et al., 2013). Table 1 showed the CCD including four factorial points (consisting of all possible combinations of the maximum and minimum levels), four axial points (all the factors set to the midpoints, except the center point), and center point with two additional experimental trials as the replicates of this point. In this table, the independent variable levels were presented in terms of the coded levels. The coded values set for C/N (A) and current density (B) at five levels were: –1.414 (–alpha), –1, 0 (central), +1, and +1.414 (+alpha). The C/Ns were 0.6465, 0.75, 1.00, 1.25, and 1.3535, and current densities were 30, 80, 200, 320, and 370 mA/m2. To obtain the sufficient C/N and appropriate current density, three dependent parameters were analyzed as responses: nitrate-N, nitrite-N, and ammonia-N concentrations. The quadratic equation model for predicting the optimal conditions could be expressed according to the following equation: Y = β + ∑ β + ∑
β
+ ∑ ∑
β +
(1)
where i is the linear coefficient, j is the quadratic coefficients, β is the regression coefficient, k is the number of factors studied and optimized in the experiment, and e is the random error. Due to two factors (C/N and current density) being involved in this study (k = 2), the following equation would be derived: Y = β + β + β + β + β + β + Analysis of variance (ANOVA) provided the statistical results and diagnostic checking tests to evaluate the adequacy of the models. The quality for fitting the
(2)
polynomial model was expressed by the coefficient of determination R2. Model terms were evaluated by the P-value (probability) with a 95% confidence level. Respective contour plots (two-dimensional) and surface plot (three-dimensional) were obtained from the responses in the Design-Expert (version 8.0) software, for the HAD-BER performance, based on the effects of C/N and current density with the actual values. 2.4 Experimental procedures and analytical methods 2.4.1 Determination of optimum operating conditions The reactor with domesticated anaerobic bacteria was used in the experimental stage. The HRT and temperature were set at 8 h and 30 ℃, respectively. According to the CCD in section 2.3 (Table 1), 11 experimental runs were performed. Two parallel samples for influent and three parallel samples for effluent were taken every 24 h. The samples were filtered with 0.45 µm cellulose acetate membrane filters before analysis. All the runs lasted for at least seven days, including three days for bacterial adjustment and at least four days for the experiments. In total, the experiments covered a period of 107 days, including 30 days start-up and 77 trial days. Experiments were repeated if there was a sample analysis error greater than 5%. 2.4.2 Validation of optimum operating conditions The validation experiment was performed under the optimal C/N and current density suggested by RSM. The HRT and temperature were maintained at 8 h and 30 ℃, respectively. Nitrate-N concentration in the influent and effluent as well as the nitrite-N and ammonia-N accumulation in the effluent were detected every 24 h. The sample pretreatment was the same as described in 2.4.1. The experiment lasted for 14
days. 2.4.3 Analytical methods Nitrate-N, nitrite-N, and ammonia-N in both the influent and effluent were determined by ultraviolet spectrophotometer (DR 5000, HACH, USA) according to standard methods. 3.
Results and discussion
3.1 Development of models and ANOVA Experiments designed by CCD (Table 1) were performed randomly at the set values of C/N and current density. Data collected for nitrate-N, nitrite-N, and ammonia-N concentrations in effluent from HAD-BER were analyzed, and the relationship between the two operating parameters in the HAD-BER (A: C/N and B: current density) and three important process responses (R1: Nitrate-N concentration, R2: Nitrite-N concentration, and R3: Ammonia-N concentration) were mathematically modeled by RSM according to Eq. (2). Table 2 represents the models in terms of coded and actual values, as well as pertinent ANOVA results. The P-values were used to estimate whether F was large enough to indicate statistical significance and used to check the significance of each coefficient (Bingöl et al., 2012). P-values lower than 0.05 indicated that the model was statistically significant. In table 2, the P-values were
S0960-8524(14)01233-4 http://dx.doi.org/10.1016/j.biortech.2014.08.117 BITE 13880
To appear in:
Bioresource Technology
Received Date: Revised Date: Accepted Date:
30 June 2014 24 August 2014 26 August 2014
Please cite this article as: Tong, S., Chen, N., Wang, H., Liu, H., Tao, C., Feng, C., Zhang, B., Hao, C., Pu, J., Zhao, J., Optimization of C/N and current density in a heterotrophic/biofilm-electrode autotrophic denitrification reactor (HAD-BER), Bioresource Technology (2014), doi: http://dx.doi.org/10.1016/j.biortech.2014.08.117
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Optimization of C/N and current density in a heterotrophic/biofilm-electrode autotrophic denitrification reactor (HAD-BER) Shuang Tonga,b, Nan Chena,b , Heng Wang b, Hengyuan Liu b, Chen Tao b, Chuanping Fenga,b*, Baogang Zhangb, Chunbo Hao b, Jiaoyang Pu b, Jiamin Zhaob a
Key Laboratory of Groundwater Circulation and Evolution (China University of
Geosciences Beijing), Ministry of Education, Beijing 100083, China b
School of Water Resources and Environment, China University of Geosciences
Beijing, Beijing 100083, China ABSTRACT In this study, central composite design (CCD) and response surface methodology (RSM) were applied to optimize the C/N and current density in a heterotrophic/biofilm-electrode autotrophic denitrification reactor (HAD-BER). Results showed that nitrate could be effectively reduced over a wide range of C/Ns (0.84-1.3535) and current densities (96.8-370.0 mA/m2); however, an optimum C/N of 1.13 and optimum current density of 239.6 mA/m2 were obtained by RSM. Moreover, the HAD-BER performance under the optimum conditions resulted in almost 100% nitrate-N removal efficiency and low nitrite-N and ammonia-N accumulation. Furthermore, under the optimum conditions, H2 generated from water electrolysis matched the CO2 produced by heterotrophic denitrification by stoichiometric calculation. Therefore, CCD and RSM could be used to acquire
*
Corresponding author. Tel.: +86 10 8232 2281; fax: +86 10 8232 1081. E-mail: [email protected] (C. Feng)
optimum operational conditions and improve the nitrate removal efficiency and energy consumption in the HAD-BER. Keywords: Nitrate reduction; HAD-BER; Optimization; Central composite design (CCD); Response surface methodology (RSM) 1.
Introduction Groundwater is one of the main fresh water resources. It has long been used as a
municipal water supply in most countries of the world (Ergas and Reuss, 2001). However, the rapid movement of human civilization towards urbanization, industrialization, and increased agricultural activities has introduced a large amount of nitrate into the environment (Ghafari et al., 2008). In the USA, from 10% to 25% of the groundwater used as drinking water suffers from nitrate concentrations above the maximum allowable contaminant level (10 mg nitrate-N/L) (Liu et al., 2009). In China, the pollution of groundwater with nitrate is even more severe, with nitrate concentrations in groundwater in some rural areas exceeding 130 mg nitrate-N/L (Wan et al., 2009). Nitrate in various sources of groundwater represents a serious risk to the health of humans and animals (Ghafari et al., 2009). Nitrate is reverted to toxic nitrite by microbes in the human body, which reacts with hemoglobin in the blood, converting the hemoglobin to methemoglobin, resulting in methemoglobinemia (Mousavi et al., 2012). Nitrate has also been identified as a cause of gastric cancer (Mirvish, 1977). Accordingly, various agencies have stipulated standards for nitrate-N in drinking water to minimize these health risks: the World Health Organization (WHO) has set a
maximum contaminant level of 10 mg nitrate-N/L (WHO, 2004), and the same value has also been adopted by China (Standards for Drinking Water Quality, GB5749-2006). Hence, various methods have been exploited for nitrate removal, for example, biological processes have high potential for the elimination of nitrate with high stability and reliability (Foglar et al., 2005), which represent a promising solution for controlling nitrate pollution. Denitrification is a microbial process that mitigates nitrate-N pollution by reducing nitrate-N to N2 in hypoxic conditions, utilizing an electron donor such as organic carbon or hydrogen (Schmidt and Clark, 2012). However, traditional hydrogen autotrophic denitrification, using hydrogen as an electron donor, consumes a lot of H2 due to the low H2-delevery rate (Rittmann, 2006), and it has relatively lower efficiency when compared with heterotrophic denitrification (Tang et al., 2012). Whereas, heterotrophic denitrification involves secondary pollution from the added organic carbon source, due to the lack of a carbon source in groundwater (Zhang et al., 2012). Recently, biofilm-electrode reactors (BERs) combining biological and electrochemical methods have become effective nitrate removal devices. BERs are based on hydrogen autotrophic denitrification, which incorporates microorganisms on the cathode, and uses hydrogen produced from water electrolysis as an electron donor for autotrophic denitrification (Zhang et al., 2005). However, traditional BERs consume a lot of electric energy to produce sufficient H2 (Mousavi et al., 2012).
Hence, to improve denitrification efficiency and reduce the consumption of electrical energy, a heterotrophic/biofilm-electrode autotrophic denitrification reactor (HAD-BER) has been developed in previous studies (Zhao et al., 2011; Zhao et al., 2012; Tong et al., 2013). The reactor’s performance was evaluated under varying hydraulic retention times (HRTs) and electric currents (Is). The highest nitrate removal efficiency was achieved at HRT = 8 h and I = 10 mA (Zhao et al., 2011). Likewise, the behaviors of autotrophic, heterotrophic, and autotrophic and heterotrophic denitrification (AD, HD, and HAD) were investigated in the reactor, and the HAD process proved to be superior to single AD and HD (Zhao et al., 2012). It was then demonstrated that the autotrophic bacteria used CO2 generated by heterotrophic processes, representing a synergistic interaction between heterotrophic and autotrophic bacteria in the HAD-BER (Tong et al. 2013). Therefore, it was not necessary to add an inorganic carbon source to the HAD-BER, reducing the cost compared to traditional BERs, which require the addition of an inorganic carbon source such as NaHCO3 (Park et al., 2006). HAD-BERs also greatly improved the efficiency of the reaction, significantly saving electricity (Zhao et al., 2012). Both heterotrophic and biofilm-electrode autotrophic denitrification can be considered two-step reduction processes, in which nitrite is an intermediate byproduct (Park et al., 2008; Ghafari et al., 2009). Nitrite reductase is more sensitive than nitrate reductase to the experimental conditions (Ghafari et al., 2009). For heterotrophic denitrification, a range of factors influence the nitrite accumulation, including organic carbon supply, temperature, oxygen content, and phosphate concentration (Hunter,
2003). At the same time, a deficiency of available carbon might also be the leading factor for nitrite accumulation (Zhang et al., 2012). On the other hand, for biofilm-electrode autotrophic denitrification, a minimum dissolved hydrogen concentration of 0.2 mg/L is a necessity for nitrite reductase, with inhibition at lower concentrations, while nitrate reductase will occur even at lower hydrogen concentrations (0.1mg/L) (Chih et al., 1999). Excessively high hydrogen concentrations have also been reported as another denitrification inhibitor and cause of nitrite accumulation (Islam and Suidan, 1998). Therefore, suitable amounts of carbon and hydrogen must be supplied to cope with the susceptibility of inhibition during the denitrification process. In other words, C/N and current density (as the in situ hydrogen generator) must be maintained at appropriate levels in a HAD-BER or BER. Many researchers have investigated the effects of different parameters and tried to optimize them. For example, Zhao et al. (2011) found the optimal values of single factors (I and HRT) by performing numerous experiments with an intensified BER. Hao et al. (2013) optimized the operating parameters (C/Ns and HRTs) in a three-dimensional BER. The C/N was tested at 0.5, 1.0, 1.5, 2.0, and 3.0, and the HRT at 5, 7, 10, and 12 h. A nitrate removal efficiency of 98.3% was obtained at a C/N ratio of 3.0 and HRT of 7 h. In the studies mentioned above, each individual parameter was evaluated by fixing all other parameters. However, this research approach has some drawbacks, including a large workload and not addressing the interactions between different parameters. The autotrophic denitrification and
heterotrophic denitrification in the HAD-BER influence each other, however, so far, little research has focused on the interaction between C/N and current density. Design of experiments (DOE) and response surface methodology (RSM) are widely used for modeling process parameters. As RSM requires a small number of experiments, it is advantageous over conventional methods. DOEs are suitable for multi-factor experiments to identify correlations between various factors and determine the most favorable conditions for the processes. DOEs include central composite designs (CCDs), Doehlert matrixes (DMs), and Box-Behnken designs (BBDs) (Bingöl et al., 2012). DOEs and RSMs have been employed successfully to optimize the operational parameters in water treatment (Ghafari et al., 2009; Li et al., 2010a). Therefore, in this study, to determine the optimum C/N and current density in the developed HAD-BER, a CCD design method was used to set experimental parameters, and a series of experiments were performed with those experimental parameters. RSM was then employed to determine the optimum conditions (C/N and current density) of the HAD-BER. Then finally, the performance of the HAD-BER was evaluated under those optimum conditions. 2.
Materials and Methods
2.1 Experimental apparatus The HAD-BER used in the current study is similar to one from a previous study (Tong et al., 2013). To keep the HAD-BER at a constant temperature (30 ± 1 ℃), the reactor was kept
in a water bath containing four electric heaters (AT-180, Atman, China). 2.2 Sludge acclimation and synthetic contaminated groundwater preparation A mixed culture containing denitrifying bacteria was originally collected from anaerobic sludge, which was obtained from the Qinghe Sewage Treatment Plant (Beijing, China). Acclimatized denitrifying bacteria, obtained according to a previously developed method (Tong et al., 2013), were used as the inoculums. Hydrogen, provided by water electrolysis, and methanol were used as electron donors for autotrophic and heterotrophic denitrification, respectively. A C/N of 1.25 and current density of 200 mA/m2 were adopted to initiate the HAD-BER (Tong et al., 2013). Biofilm formed gradually and a dark grey color was observed on the cathode module within 30 days. Synthetic contaminated groundwater (per liter of tap water) contained 0.304 g/L NaNO3 and 0.044 g/L KH2PO4, thus the concentration of nitrate-N was 50 mg/L. CH3OH was added as required to maintain the desired C/N. All solid reagents were weighed using an analytical balance (TX223L, Shimadzu, Japan) and liquid reagents were measured via pipettes with specific ranges (20-200μL and 100-1000μL, Nichipet EX-PLUS, Nichiryo, Japan; 0.5-5mL, Research plus, Eppendorf, Germany). All chemical reagents used in the experiments were of analytical grade. 2.3 Experimental design and data analysis Design-Expert (version 8.0) software was used for the statistical design of experiments and data analysis. In this study, CCD and RSM were applied to optimize the operating variables C/N and current density. The ranges of variables were taken
from previous studies (Wang et al., 2009; Zhao et al., 2011; Zhao et al., 2012; Tong et al., 2013). Table 1 showed the CCD including four factorial points (consisting of all possible combinations of the maximum and minimum levels), four axial points (all the factors set to the midpoints, except the center point), and center point with two additional experimental trials as the replicates of this point. In this table, the independent variable levels were presented in terms of the coded levels. The coded values set for C/N (A) and current density (B) at five levels were: –1.414 (–alpha), –1, 0 (central), +1, and +1.414 (+alpha). The C/Ns were 0.6465, 0.75, 1.00, 1.25, and 1.3535, and current densities were 30, 80, 200, 320, and 370 mA/m2. To obtain the sufficient C/N and appropriate current density, three dependent parameters were analyzed as responses: nitrate-N, nitrite-N, and ammonia-N concentrations. The quadratic equation model for predicting the optimal conditions could be expressed according to the following equation: Y = β + ∑ β + ∑
β
+ ∑ ∑
β +
(1)
where i is the linear coefficient, j is the quadratic coefficients, β is the regression coefficient, k is the number of factors studied and optimized in the experiment, and e is the random error. Due to two factors (C/N and current density) being involved in this study (k = 2), the following equation would be derived: Y = β + β + β + β + β + β + Analysis of variance (ANOVA) provided the statistical results and diagnostic checking tests to evaluate the adequacy of the models. The quality for fitting the
(2)
polynomial model was expressed by the coefficient of determination R2. Model terms were evaluated by the P-value (probability) with a 95% confidence level. Respective contour plots (two-dimensional) and surface plot (three-dimensional) were obtained from the responses in the Design-Expert (version 8.0) software, for the HAD-BER performance, based on the effects of C/N and current density with the actual values. 2.4 Experimental procedures and analytical methods 2.4.1 Determination of optimum operating conditions The reactor with domesticated anaerobic bacteria was used in the experimental stage. The HRT and temperature were set at 8 h and 30 ℃, respectively. According to the CCD in section 2.3 (Table 1), 11 experimental runs were performed. Two parallel samples for influent and three parallel samples for effluent were taken every 24 h. The samples were filtered with 0.45 µm cellulose acetate membrane filters before analysis. All the runs lasted for at least seven days, including three days for bacterial adjustment and at least four days for the experiments. In total, the experiments covered a period of 107 days, including 30 days start-up and 77 trial days. Experiments were repeated if there was a sample analysis error greater than 5%. 2.4.2 Validation of optimum operating conditions The validation experiment was performed under the optimal C/N and current density suggested by RSM. The HRT and temperature were maintained at 8 h and 30 ℃, respectively. Nitrate-N concentration in the influent and effluent as well as the nitrite-N and ammonia-N accumulation in the effluent were detected every 24 h. The sample pretreatment was the same as described in 2.4.1. The experiment lasted for 14
days. 2.4.3 Analytical methods Nitrate-N, nitrite-N, and ammonia-N in both the influent and effluent were determined by ultraviolet spectrophotometer (DR 5000, HACH, USA) according to standard methods. 3.
Results and discussion
3.1 Development of models and ANOVA Experiments designed by CCD (Table 1) were performed randomly at the set values of C/N and current density. Data collected for nitrate-N, nitrite-N, and ammonia-N concentrations in effluent from HAD-BER were analyzed, and the relationship between the two operating parameters in the HAD-BER (A: C/N and B: current density) and three important process responses (R1: Nitrate-N concentration, R2: Nitrite-N concentration, and R3: Ammonia-N concentration) were mathematically modeled by RSM according to Eq. (2). Table 2 represents the models in terms of coded and actual values, as well as pertinent ANOVA results. The P-values were used to estimate whether F was large enough to indicate statistical significance and used to check the significance of each coefficient (Bingöl et al., 2012). P-values lower than 0.05 indicated that the model was statistically significant. In table 2, the P-values were
