Efficiency assessment of urban wastewater treatment plants in China: Considering greenhouse gas emissions

G Model

ARTICLE IN PRESS

RECYCL-3434; No. of Pages 9

Resources, Conservation and Recycling xxx (2016) xxx–xxx

Contents lists available at ScienceDirect

Resources, Conservation and Recycling journal homepage: www.elsevier.com/locate/resconrec

Full length article

Efficiency assessment of urban wastewater treatment plants in China: Considering greenhouse gas emissions Siyu Zeng a,b , Xing Chen a , Xin Dong a,∗ , Yi Liu a a b

School of Environment, Tsinghua University, Beijing 100084, China Environmental Simulation and Pollution Control State Key Joint Laboratory, School of Environment, Tsinghua University, Beijing 100084, China

a r t i c l e

i n f o

Article history: Received 8 September 2016 Received in revised form 15 November 2016 Accepted 15 December 2016 Available online xxx Keywords: Wastewater treatment plant Efficiency assessment Distance function Greenhouse gas emission Benchmarking

a b s t r a c t Wastewater treatment plants (WWTPs) are high-cost facilities for improving the urban water environment and facilitating resource recycle but with inevitable negative externality. To comprehensively assess urban WWTP performance, a distance function approach was configured to quantify the efficiency with capital cost and energy consumption as inputs, removals of four types of pollutants as desirable outputs, and emission of greenhouse gases (GHGs) as undesirable output. Adding both direct and indirect GHG emissions into the efficiency metrics would help decision makers obtain a more profound understanding of urban WWTPs’ contribution to both aquatic and atmospheric environments. The method was applied to 1079 urban WWTPs across China adopting eight major biological technologies. The average efficiency score was 0.322, implying that GHG emissions could decrease by 32.2% if all plants worked efficiently. Eight plants were deemed the most efficient and formed a frontier of the best practices, while 12 plants were the most inefficient with distances from the frontier larger than 0.650. The parameterized distance function could be used to set a benchmark system for the performance surveillance of urban WWTPs. The integrated efficiency assessment considering multiple dimensions and statistical analysis on a large sample allowed us to reveal reasons for efficiency gaps. Statistic test results showed that plants scale, technology, and capacity of tertiary treatment were significant for explaining efficiency disparities. Large-scale plants, plants with the bioreactors or the anaerobic-anoxic-oxic processes, and plants without tertiary treatment processes tended to be more efficient, showing the advantage in co-benefiting water pollutants and GHG control. © 2016 Elsevier B.V. All rights reserved.

1. Introduction Wastewater treatment plants (WWTPs) are recognized as fundamental tools for improving the urban water environment (Pasqualino et al., 2009). More than that, WWTPs also play an essential role in resource conservation and recycling in the ecosystem, in terms of wastewater reclamation and reuse (Pasqualino et al., 2009), nutrients recycling (Tidåker et al., 2006), and sewage sludge treatment and disposal (Suh and Rousseaux, 2002). However, the operation of WWTPs inevitably causes some negative externalities, such as acidification and eutrophication of recipient water bodies and emission of greenhouse gases (GHGs) (Flores-Alsina et al., 2011; Lassaux et al., 2007). When removing pollutants from raw wastewater to comply with effluent standards, the WWTP is also well known as an energy-intensive facility

∗ Corresponding author. E-mail address: [email protected] (X. Dong).

(Hernandez-Sancho et al., 2011a). Given that the constructions and technical upgrades of WWTPs always involve high costs (Dasgupta et al., 2001), plant managers and local governments have keen interests in simultaneously improving the performance and restricting costs of WWTPs (Molinos-Senante et al., 2010). Nevertheless, various aspects influence the behavior of WWTPs. To enhance the overall performance of WWTPs and reduce their negative impacts, an integrated assessment of WWTPs in which all technical, economic, and environmental aspects are considered is the first crucial step (Hoibye et al., 2008). In a productive economy, efficiency is applied to describe the optimal use of available resources under existing technology (Hernández-Sancho and Sala-Garrido, 2009) and assess performances in different areas. Popular methods for efficiency analysis concerning WWTPs include life cycle assessment (LCA), multipleobjective evaluation, and indicator-based method. LCA takes the whole treatment process including sludge treatment into account and evaluates performance of case WWTPs from perspectives of cost, energy and chemical consumption, nutrient loading, and GHG

http://dx.doi.org/10.1016/j.resconrec.2016.12.005 0921-3449/© 2016 Elsevier B.V. All rights reserved.

Please cite this article in press as: Zeng, S., et al., Efficiency assessment of urban wastewater treatment plants in China: Considering greenhouse gas emissions. Resour Conserv Recy (2016), http://dx.doi.org/10.1016/j.resconrec.2016.12.005

G Model RECYCL-3434; No. of Pages 9 2

ARTICLE IN PRESS S. Zeng et al. / Resources, Conservation and Recycling xxx (2016) xxx–xxx

emissions (Pasqualino et al., 2009; Dong et al., 2014; Lorenzo-Toja et al., 2015; Rahman et al., 2016). The approach of multipleobjective evaluation usually considers aspects such as operating cost, water quality, effluent standard, and microbiological risks as objective functions for ranking WWTPs (Flores-Alsina et al., 2010; Kalbar et al., 2012). As to indicator-based method, different hierarchical indicators, e.g. environmental, personnel, physical, operational, quality of service and economic and financial indicators are often selected to evaluate WWTP performance (Balkema et al., 2002; Matos and Association, 2003; Quadros et al., 2010). It can be difficult to apply these methods because they all require detailed inventory data and are technically complicated (SalaGarrido et al., 2011). Additionally, when only partial cases are evaluated, using the results from these methods to make inferences about general conclusions requires extra caution (Zhong et al., 2011). An alternative for WWTP assessment is the distance function approach, first proposed by (Pittman, 1983) and (Färe et al., 1989), through which pollution emissions are incorporated into the traditional Shephard’s production function (Shephard et al., 1970) to measure productivity and efficiency. Since its proposal, the distance function approach has been widely used in the estimation of productivity indexes, technological efficiency, shadow prices of pollutants, and marginal abatement costs of pollutants, owing to its advantages, including joint modeling of both desirable and undesirable outputs (Diaz-Balteiro and Romero, 2008), easily obtained quantitative input/output data, and compact results of efficiency capturing interactions among technologies and side effects (Lee et al., 2014). Several studies used the distance function approach to assess WWTPs performances (Tupper and Resende, 2004; MolinosSenante et al., 2010; Molinos-Senante et al., 2011b; Lorenzo-Toja et al., 2015) in which different technical, economic and environmental aspects were taken into consideration in an integrated model. Hence, the distance function approach was selected and employed in the present study because it can provide a more complete picture of production processes (Murty et al., 2006). Regarding the undesirable output of the WWTP, GHG emissions have drawn greater attention recently since it is identified as one of the largest minor sources (Doorn et al., 1997). Since the International Panel on Climate Change (IPCC) stated that wastewater treatments are biogenic sources, few studies investigating WWTP performance involving GHG emissions (Tupper and Resende, 2004; Molinos-Senante et al., 2010; Lorenzo-Toja et al., 2015) have been conducted. With the increasing criticism that the GHG emission of WWTPs was underestimated, paying more attention to this specific source is necessary (Bani Shahabadi et al., 2009; Foley et al., 2010; Yoshida et al., 2014). Adding the GHG emission into the assessment of WWTPs leads to a rethinking of the results of traditional methods. By adding this new angle, we could obtain an integrated analysis of WWTPs and perform comparisons of different WWTP performances to facilitate the reduction of GHG emissions. In China, the number of WWTPs has increased dramatically during the last three decades in urban area. By 2013, 3513 urban WWTPs had been built and the total treatment capacity had reached 1.25 × 108 m3 /d. The demand for future constructions and technical upgrades of urban WWTPs is still significant. Despite the extensive literature on WWTP assessments, integrated analyses of urban WWTPs in China from economic and environmental perspectives remain scarce. Therefore, any attempt to address additional evidence to improve urban WWTP performance is vital and timely. With the distance function approach, this paper aims to assess the current efficiency of China’s urban WWTPs by considering the reduction of pollution load, as well as related costs, energy consumption, and GHG emissions. Factors that affect efficiency scores are discussed as well, because a large sample of 1079 urban WWTPs

and a parameterized distance function allow us to reveal reasons for efficiency gaps.

2. Methods 2.1. Definition of input and output The assessment unit for computation with the distance function approach was each urban WWTP. In this study, the wastewater treatment process was considered as a joint-product process by certain inputs, while the output set with both desirable and undesirable outputs led to positive and negative environmental impacts accordingly. According to specific research aims and scopes, different inputs (e.g. cost, energy, chemicals, staff), positive impacts (e.g. treated water, pollutants removal), and negative impacts (e.g. effluent loads, pharmaceutical and personal care products) of urban WWTPs were chosen flexibly, when applying the approach of distance function (Tupper and Resende, 2004; Molinos-Senante et al., 2010; Lorenzo-Toja et al., 2015). This study sought to capture the efficiency of urban WWTPs with the aim of obtaining vital information for improving environmental performance and valuing the reduction potential of GHG emissions, resulting in the determination of two inputs, four desirable output and one undesirable output. Two inputs were considered: fixed-asset investment per volume of treated wastewater (x1 , RMB Yuan/m3 ) and energy consumption per volume of treated wastewater (x2 , kWh/m3 ). Electricity (energy consumption) is actually the largest operation cost of WWTPs in China (Jin et al., 2014). These two inputs could represent both capital cost and operational expenditure for WWTPs. Desirable outputs were defined with the concentration differences of pollutants between the influent and effluent, because the pollutant removal from WWTPs is a primary function of the treatment process and contributes a positive effect to the environment. Four types of pollutants were involved, including chemical oxygen demand (COD), suspended solids (SS), total nitrogen (TN), and total phosphorus (TP), and the corresponding outputs were denoted as y1 (mg/L), y2 (mg/L), y3 (mg/L), and y4 (mg/L), respectively. To measure the negative environmental impact of urban WWTPs, GHG emissions per volume of treated wastewater (z1 , mg/L) was treated as an undesirable output in this study. WWTPs produce GHGs during the treatment process and through energy consumption, called direct (or on-site) and indirect (or off-site) emissions respectively. Both direct and indirect GHG emissions were calculated when quantifying the undesirable outputs. Previous studies have shown a wide range of intensities of GHG emissions in WWTPs, affected by influent load, temperature, treatment process, and operating conditions (Kampschreur et al., 2009; Corominas et al., 2012). As a result, high uncertainty tends to exist when determining the GHG emissions, although various quantification methods are reported in the literature including the modeling approach (Bani Shahabadi et al., 2010; Rodriguez-Garcia et al., 2012), onsite observation (Wang et al., 2011), and estimation with emission factors statistically (Pan et al., 2011). In this study, for the large sample size and lack of data, the method using emission factors was more suitable, though the plant-specific emissions factors were not practically available. Consequently, GHG emissions including both direct emissions (GHGdirect , mg CO2 -eq/L) and indirect emissions (GHGindirect , mg CO2 -eq/L) were estimated following the method derived from IPCC guidelines (CHANGE, 2006), so that the undesirable output can be determined as z1 = GHGdirect +GHGindirect . The equation for estimating direct GHG emissions is as follows: GHGdirect = (TOW × B0 × MCF-R) × 25 + TN × EF N 2 O × 298

(1)

Please cite this article in press as: Zeng, S., et al., Efficiency assessment of urban wastewater treatment plants in China: Considering greenhouse gas emissions. Resour Conserv Recy (2016), http://dx.doi.org/10.1016/j.resconrec.2016.12.005

G Model

ARTICLE IN PRESS

RECYCL-3434; No. of Pages 9

S. Zeng et al. / Resources, Conservation and Recycling xxx (2016) xxx–xxx Table 1 Indirect GHG emission factor in different grid (105 mg CO2 -eq/kWh).

EFCO2-eq ,i

Northeast

Northwest

Central

North

East

South

11.15

8.23

6.49

10.62

8.01

6.68

where TOW (mg BOD/L), annual total organics in wastewater on the basis of influent BOD concentration; B0 (mg CH4 /mg BOD), maximum CH4 producing capacity, was set as the default value of 0.6; MCF, methane correction factor, was set as 0.3 for those centralized aerobic treatment plants which were not well managed; R (mg CH4 /L), methane recovery, was set as 0 since most WWTPs in China did not build methane recovery system; TN (mg/L), influent TN concentration; EFN 2 O , emission factor for N2 O, was set as the default value of 0.00786; and GHGdirect is converted as the CO2 -equivalent emission (mg CO2 -eq/L) with weights (25 for CH4 , 298 for N2 O) based on the 100-year global warming potentials. The GHG emissions from electricity consumption in urban WWTP reflect the indirect GHG emissions. But the sources for generating electricity in China are diverse, including coal, oil, liquefied natural gas, hydropower, atomic, solar, and etc. More than that, the indirect GHG emissions depending on the electricity generation mix for different grids in China vary drastically. It is essential to identify the electricity source for each urban WWTP and calculate indirect emissions with different emission factors accordingly. In this study, the electricity provided in the same regional grid is considered homogeneous. Based on the geographical location of each WWTP and the distribution area serviced by each grid, different emission factors are applied. The estimation equation of indirect GHG emissions is as follows: GHGdirect = x2 × EF CO 2 -eq ,i × 1000i = 1, . . ., 6

(2)

3

put consecutively. The directional output distance function (Dk ) for this WWTP was achieved in a quadratic parametric form with 43 parameters, including ˛0 , ˛i , ˇj ,  1 , ip , jq , 11 , ıij , i1 , ωj1 (i = 1,2; j = 1,2,3,4; p = 1,2; q = 1,2,3,4), as shown in Eq. (3).

→

Dk = DO (x1k , x2k , y1k , y2k , y3k , y4k , z1k ; 1, 1) 2 

= ˛0 +

(˛i xik ) +

i=1 2 2 1  

2

4   j=1



 1 1   2 jq yjk yqk + 11 z1k + 2 2 4

4

ip xik xpk +

i=1 p=1 4 2   



ˇj yjk + 1 z1k + (3)

j=1 q=1 2  

ıij xik yjk +

i=1 j=1

(i1 xik z1k ) +

i=1

4  

ωj1 yjk z1k



j=1

The distance function constructs a frontier of the best practices comprising the most efficient units among observation samples. To determine parameter values, the sum of the distances between the frontier and the individual observations were minimized; the parameters should meet the restrictions of feasibility, monotonicity, translation property, and symmetry condition, as shown in Eq. (4). Matlab was used to estimate parameters by solving the optimization programming problems. To ensure convergence when solving Eq. (4), the observation data were normalized by the average values of corresponding variables.

(kWh/m3 ), electricity consumption per volume of treated

where x2 wastewater; EFCO 2 -eq ,i (mg CO2 -eq/kWh), indirect GHG emission factor for WWTPs serviced by the grid i; i = 1, . . ., 6, represented Northeast, Northwest, Central, North, East, and South Grid, respectively. EFCO 2 -eq ,i for different grids were based on values reported by Song et al. (2013) and showed in Table 1. The determination of EFCO 2 -eq in each grid took account of the type and composition of power (e.g. thermal power, hydropower, wind power, and nuclear power), power change among grids, and transmission loss. 2.2. Distance function and its solution The distance function method was a multi-objective and integrated assessment method. It was used to examine the efficiency level of each urban WWTP compared with the most efficient treatment process using existing technology. In accordance with the definition of inputs and outputs before-mentioned, the most efficient treatment processes were those that minimized all input resources, removed the largest amount of pollutants, and released the least amount of GHGs. It means that this assessment would not only take into account the water quality improvement capability of urban WWTP, but also consider the resource consumption and GHG emission. In essence, the thought of Pareto Optimal was used to evaluate and rank the efficiency of urban WWTP. Among different types of distance functions, including input distance function, output distance function, and directional distance function (Färe and Grosskopf, 2000), directional output distance function (Färe et al., 2005; Färe et al., 2006) was selected and applied to the evaluation because it allowed for the reduction of undesirable outputs and the estimation of output possibilities with a fixed input. Assume that N WWTPs are under assessment. For the kth WWTP (1 < k < N), observations were denoted as (x1k , x2k , y1k , y2k , y3k , y4k , z1k ) for two inputs, four desirable outputs, and one undesirable out-

min

N 

(Dk − 0)

k=1

s.t. (i) Dk ≥ 0, k = 1, . . ., N (ii) ∂Dk /∂xik ≥ 0, i = 1, 2; k = 1, . . ., N (iii) ∂Dk /∂yjk ≤ 0, j = 1, . . ., 4; k = 1, . . ., N (iv) ∂Dk /∂z1k ≥ 0, k = 1, . . ., N (v)

4 

ˇj − 1 = −1; 4  j=1

(4)

jq − ωj1 = 0, j = 1, . . ., 4;

q=1 4

j=1

11 −

4 

ωj1 = 0;



ıij − i1 = 0, i = 1, 2

j=1

/ p; jq = qj , j = / q (vi) ip = pi , i =

The function value Dk itself is a measurement of the efficiency of WWTP k because it simultaneously accounts for the reduction of GHG emission and improvements in pollutant removal with fixedasset investment and energy consumption as inputs. The lower the value of the plant’s distance function was, the higher the efficiency of the plant. In particular, if Dk = 0, then this particular plant was located on the frontier and reached the highest efficiency level. Feasible but inefficient plants would take on values greater than 0. To analyze the efficiency contribution of specific inputs and outputs of WWTPs, the ratios of outputs to inputs were used to

Please cite this article in press as: Zeng, S., et al., Efficiency assessment of urban wastewater treatment plants in China: Considering greenhouse gas emissions. Resour Conserv Recy (2016), http://dx.doi.org/10.1016/j.resconrec.2016.12.005

G Model

ARTICLE IN PRESS

RECYCL-3434; No. of Pages 9

S. Zeng et al. / Resources, Conservation and Recycling xxx (2016) xxx–xxx

4

Fig. 1. Distribution of inputs and GHG emissions.

investigate the details of plant efficiency, namely, efficiency subindicators, which were defined as follows: yik ,i = 1, ..., 4,k = 1, ..., N x1k z EP5k = 1k ,k = 1, ..., N x1k y EEik = ik ,i = 1, ..., 4,k = 1, ..., N x2k z EE5k = 1k ,k = 1, ..., N x2k

EPik =

(5)

the ratios of the tertiary treatment capacity to total wastewater treatment capacity were between 5% and 100%. The number of sample WWTPs located in Northeast Grid, Northwest Grid, Central Grid, North Grid, East Grid, and South Grid were 37, 42, 262, 240, 296, and 202 respectively. The sample covered the majority of treatment technologies, a large range of plant sizes, and sufficient geographical distribution in China; thus, can be used to represent current status of WWTPs in China. 2.4. Explanatory factors

2.3. Samples and data A total of 1924 urban WWTPs were inventoried in this case study to develop the efficiency assessment. The data of fixed-asset investment, energy consumption, pollutants removal amount of COD, SS, TN, and TP were self-reported by each WWTP for the year of 2011 and compiled in the relevant yearbooks (China Urban Water Association, 2012). The direct and indirect GHG emission of each plant was estimated following Eqs. (1) and (2). Samples with missing values in either inputs or outputs were eliminated, and 1186 urban WWTPs had complete data format and were retained for this evaluation. In particular, samples with values located outside the range of mean plus/minus three times the standard deviation, which were treated as outliers, i.e. 107 urban WWTPs were considered to be outliers, were also removed. As a result, the sample for assessment included 1079 urban WWTPs that are located in 30 provinces in China. The treatment capacity of sample WWTPs ranged between 500 m3 /d and 1700,000 m3 /d. Eight of the most widely used technologies in China were adopted by the sample plants, including conventional activated sludge (CAS), anoxic-oxic (A/O), anaerobicanoxic-oxic (A2 /O), oxidation ditch (OD), sequencing batch reactor (SBR), biological filter (BF), biological contact oxidation (BCO), and membrane bioreactor (MBR). Among those, 153 WWTPs contained the tertiary treatment processes after the secondary biological processes, such as the biofilter and ultrafilter. The total scale of wastewater reclamation of these plants was 4.34×106 m3 /d, while

Once the efficiency scores were estimated for the plants, the next step was identifying explanatory factors that affected the WWTP efficiencies. In related literature, many factors could affect the performance of WWTPs, including loading, technology, capacity, with or without the tertiary treatment for wastewater reuse purpose, and seasonality (Muga and Mihelcic, 2008; Sala-Garrido et al., 2011; Hernandez-Sancho et al., 2011a,b; Molinos-Senante et al., 2011a). In the present study, plant scale, the technology, and containing wastewater reclamation process (tertiary treatment process) were assumed relevant to efficiency and were thus examined. Kruskal-Wallis test was used to validate the statistical significance of differences among multiple groups (Sueyoshi and Aoki, 2001; Botti et al., 2009), while Mann-Whitney U test was used to validate the statistical significance of the difference between two groups. 3. Results and discussion 3.1. Characteristics of sample WWTPs The inputs and outputs of the sample plants were summarized. Distributions of fixed-asset investment, energy consumption, and GHG emissions are shown in Fig. 1. The average and median of fixed-asset investment were 26 and 12 RMB Yuan/m3 respectively. The average and median of energy consumption were 0.30 and 0.26 kWh/m3 respectively. The average concentration of total GHG emission was 1098 mg/L. Average pollutant removal rates of COD, SS, TN, and TP of sample plants were 81%, 90%, 54%, and 67%, while

Table 2 Parameters estimated in the distance function. Parameter

Value

Parameter

Value

Parameter

Value

Parameter

Value

␣0 ␣1 ␣2 ␤1 ␤2 ␤3 ␤4 ␥1 ␩11 ␩12 ␩21

0.102 0.000 0.035 −0.006 −0.008 −0.256 −0.009 0.722 0.000 −0.001 −0.001

␩21 ␭11 ␭12 ␭13 ␭14 ␭21 ␭22 ␭23 ␭24 ␭31 ␭32

0.002 −0.080 0.002 −0.033 0.013 0.002 0.019 −0.005 −0.041 −0.033 −0.005

␭33 ␭34 ␭41 ␭42 ␭43 ␭44 ␮11 ␦11 ␦12 ␦13 ␦14

0.062 0.022 0.013 −0.041 0.022 0.006 −0.133 0.000 0.000 −0.001 0.003

␦21 ␦22 ␦23 ␦24 11 21 ␻11 ␻21 ␻31 ␻41 –

0.004 −0.003 −0.007 −0.002 0.002 −0.009 −0.098 −0.024 0.046 −0.057 –

Please cite this article in press as: Zeng, S., et al., Efficiency assessment of urban wastewater treatment plants in China: Considering greenhouse gas emissions. Resour Conserv Recy (2016), http://dx.doi.org/10.1016/j.resconrec.2016.12.005

G Model

ARTICLE IN PRESS

RECYCL-3434; No. of Pages 9

S. Zeng et al. / Resources, Conservation and Recycling xxx (2016) xxx–xxx

Fig. 2. Distribution of distance function values for all of WWTPs.

average levels of COD, SS, TN, and TP removal during the treatment process were 294, 201, 22, and 3 mg/L, respectively. As for the efficiency sub-indicators, the medians of removals of four pollutants and GHG emission per fixed-asset investment (EP1 to EP5 ) were 24, 16, 2, 0.2, and 88 g/RMB Yuan, respectively, while the medians of removals of four pollutants and GHG emission per electricity consumption (EE1 to EE5 ) were 1078, 747, 78, 10, and 3959 g/kWh, respectively. 3.2. Efficiency analysis Following the methods above, efficiency scores were obtained for sample WWTPs. All 43 parameters estimated are shown in Table 2. The distribution of WWTP efficiency is shown in Fig. 2. The average efficiency level of all 1079 WWTPs was 0.322, which indicated that the plants under assessment could decrease GHG emissions and increase pollutant removal by 32.2% if they all worked efficiently. The standard deviation of efficiency was 0.120. Results from this study support the idea that WWTPs can operate with the same fixed-asset investment and energy inputs, but have better performance in terms of pollutant removal and the reduction of GHG emissions. Distances between most WWTPs and efficient ones ranged between 0.250 and 0.450, indicating a remarkable potential for improvement. The efficiency scores of eight WWTPs were equal to 0, indicating that these plants obtained the best efficiency level and were located on the best practices frontier. These plants support the idea that with less fixed-asset investments and energy consumptions, WWTPs can remove more pollutants and release less GHGs. The characteristics of these plants are shown in Table 3. Generally speaking, these plants had lower fixed-asset investment, energy consumption, and GHG emissions, whereas their pollutant removal rates were higher than those of other plants. Taking plant No.5 as an example, although its fixed-asset investment, electricity consump-

5

tion and GHG emissions were around the median level, the high pollutants removal contributed benefits to the efficiency value. Its influent concentration of COD, SS and TN were higher than 80% plants, but the effluent concentration of these pollutants were lower than 85% plants. The pollutants removal rates of COD, SS, TN and TP of No.5 WWTP were 92%, 98%, 70% and 67%. On the contrary, 12 plants were identified as the most inefficient plants, with scores larger than 0.650. Compared with the efficient plants, the average fixed-asset investment, electricity consumption, and GHG emissions of these 12 plants were 2.1, 2.5, and 1.5 times higher, respectively. The average removal rates of COD, SS, TN, and TP were 77%, 84%, 40%, and 60%, which were clearly lower than those of other WWTPs. Taking plant No.1072 as an example which used OD technology with the scale of 10,000 m3 /d, 38 RMB Yuan/m3 and 0.54 kWh/m3 were used to remove 280 mg/L COD, 170 mg/L SS, 5 mg/L TN and 2.0 mg/L TP and the GHG emissions were 1444 mg CO2 eq/L. The pollutants removal rates of COD, SS, TN and TP were, 70%, 85%, 14% and 67%. The extremely higher fixedasset investment and energy consumption, the poor performance of pollutants removal process and large GHG emissions together made the worst efficiency. The parameterized distance function allowed us to set a benchmark system for performance supervision and guide the development of the whole industry. For example, the 15th percentiles of two efficiency sub-indicators related to undesirable outputs (EP5 and EE5 ) and the 85th percentiles of eight efficiency sub-indicators related to desirable outputs (EP1 -EP4 and EE1 -EE4 ) could be used to form a theoretically efficient plant for benchmarking. For 1 m3 of wastewater, the benchmark plant would invest 5.29 RMB Yuan and consume 0.17 kWh of electricity so that it could remove 390 g COD, 260 g SS, 33 g N, and 4 g P, and release 796 g CO2 as well. The corresponding distance function value of such a plant was 0.016, equivalent to the 1 st percentile of efficiency. 3.3. Explanatory factors 3.3.1. Treatment capacity The average treatment capacity of sample WWTPs was 5 × 104 m3 /d. To identify the relationship between treatment capacity and efficiency, all samples were categorized by treatment capacity into five groups within the range of 0–1, 1–2, 2–5, 5–10, and >10 × 104 m3 /d. The numbers of WWTPs in the five groups were 176, 230, 401, 189, and 83 respectively. The average value of the distance function of each group was 0.349, 0.335, 0.317, 0.312, and 0.275, respectively. With the increase of plant size, the distance function had a decreasing trend, indicating the increase of efficiency. The standard deviations of the efficiency values of each group was 0.142, 0.110, 0.114, and 0.117 respectively, which meant that the fluctuation of efficiency for the WWTPs with the size larger than 1 × 104 m3 /d is smaller. The boxplots for distance function values of different plant groups are illustrated in Fig. 3. Kruskal-Wallis test was applied to determine whether the differences among plant

Table 3 Characteristics of the most efficient plants. No

1 2 3 4 5 6 7 8

Fixed-asset (RMB Yuan/m3 )

6.7 17.9 54.6 15.5 11.7 18.7 2.3 6.3

Electricity Consumption (kWh/m3 )

0.35 0.37 0.73 0.19 0.31 0.19 0.18 0.12

Pollutants Removal (mg/L) COD

SS

TN

TP

540 440 400 390 550 700 400 210

280 280 470 160 490 320 390 240

60 20 25 50 35 35 45 25

5.5 3.5 12.0 6.5 2.0 2.0 4.5 2.0

GHG Emissions (mg CO2 eq/L)

Technology

Plant Size (m3 /d)

1615 534 1995 931 1131 1638 1095 362

BCO CAS MBR A2 /O A2 /O A2 /O Other OD

40,000 37,500 10,000 70,000 75,000 40,000 50,000 30,000

Please cite this article in press as: Zeng, S., et al., Efficiency assessment of urban wastewater treatment plants in China: Considering greenhouse gas emissions. Resour Conserv Recy (2016), http://dx.doi.org/10.1016/j.resconrec.2016.12.005

G Model

ARTICLE IN PRESS

RECYCL-3434; No. of Pages 9

S. Zeng et al. / Resources, Conservation and Recycling xxx (2016) xxx–xxx

6

Fig. 3. Distance function values of different treatment capacity groups.

Table 4 Statistic tests of different groups of WWTPs. Indicator

D EP1 EP2 EP3 EP4 EP5 EE1 EE2 EE3 EE4 EE5

P-value of Statistic Test Difference among plant scale

Difference among treatment technology

Difference among different capacity of tertiary treatment

0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000

0.043 0.183 0.081 0.062 0.118 0.090 0.012 0.073 0.071 0.003 0.003

0.055 0.000 0.000 0.003 0.030 0.000 0.089 0.157 0.100 0.044 0.152

scales were significant and possible reasons for the said difference. The p values listed in Table 4 reveal that the efficiencies of the five treatment capacity groups were significantly different (p < 0.05). Moreover, the differences of efficiency sub-indicators among the five groups were all significant, indicating that the magnitude of scale had comprehensive effects on plant behavior. Comparing WWTPs located in the large size group (>100,000 m3 /d) with those in the small size group (≤10,000 m3 /d), both the average values and standard deviations of fixed-asset investment, electricity consumption and GHG emission of large size plant were lower, while there was no significant difference on the removal of pollutants, as shown in Table 5. It can be found that

Fig. 5. Boxplots of distance function values of different technologies.

plants with treatment capacities larger than 100,000 m3 /d had better and more reliable performance. On the other hand, WWTPs with treatment capacities less than 10,000 m3 /d were worse at neither average efficiency nor performance stability. In fact, smallscale WWTPs tended to have higher costs due to scale economies (Hernandez-Sancho et al., 2011b), show higher consumption of electricity because of fluctuations in influent loading (Lee et al., 2008), and lack continuous monitoring (Lorenzo-Toja et al., 2015) to obtain better techno-economic performance. In addition, due to the management difficulty, the regional difference in the operation and maintenance of small-scale WWTPs was much more obvious, leading to higher efficiency fluctuation. 3.3.2. Treatment technology Fig. 4 demonstrates the percentages of the number and total treatment capacity of WWTPs adopting different technologies. Accordingly, A2 /O, OD, and SBR are the most popular technologies in China. The values of distance functions of different technologies are summarized in Fig. 5, which shows that the averages of MBR, A2 /O, SBR, BF, CAS, BCO, OD and A/O were 0.304, 0.305, 0.318, 0.318, 0.322, 0.331, 0.336, and 0.340, respectively. A Kruskal-Wallis test was applied among eight different technologies. Results indicated that efficiency differences among eight technologies were significant, as shown in Table 4. The efficiency sub-indicators were tested as well to explore the possible reasons for the different performance levels. Except for EP1 and EP4 , other economic indicators and all energy consumption indicators showed statistical differences under the significant level of 10%. The p values for GHG emission related indicators EP5 and EE5 were also less than 0.1, indicating

Fig. 4. Percentage of different WWTP technologies.

Please cite this article in press as: Zeng, S., et al., Efficiency assessment of urban wastewater treatment plants in China: Considering greenhouse gas emissions. Resour Conserv Recy (2016), http://dx.doi.org/10.1016/j.resconrec.2016.12.005

G Model

ARTICLE IN PRESS

RECYCL-3434; No. of Pages 9

S. Zeng et al. / Resources, Conservation and Recycling xxx (2016) xxx–xxx

7

Table 5 Comparison of the large size group and the small size group. Plant Size (m3 /d)

Mean Std.

<10,000 > 100,000 <10,000 > 100,000

Distance function value

0.35 0.27 0.14 0.12

Fixed-asset (RMB Yuan/m3 )

46 9 47 13

Electricity Consumption (kWh/m3 )

0.37 0.22 0.21 0.08

Pollutants Removal (mg/L)

GHG Emissions (mg CO2 eq/L)

COD

SS

TN

TP

296 275 101 98

206 194 75 70

22 23 11 12

3.1 3.2 2.2 2.4

1173 990 380 288

Table 6 Details of two example plants. Plant No.

Distance function value

Tertiary (Yes or No)

Fixed-asset (RMB Yuan/m3 )

Electricity Consumption (kWh/m3 )

602 1056

0.35 0.52

No Yes

100 234

0.22 1.14

the importance of GHG emissions control from the perspective of technology selection. Moreover, the p values of economic oriented indicators were basically larger than those of energy oriented indicators. The energy consumption efficiency turned out to be dominant for the significant difference in the distance function value among different technologies. Therefore, the reduction of energy consumption is reasonably meaningful in the operation of WWTPs in China. Among the eight types of technologies, WWTPs adopting MBR and A2 /O had the best average efficiency (i.e., the lowest average distance as 0.304 and 0.305 respectively). Since the distance function value quantified the possibilities of reduction of GHG emissions and expansion of pollutants removal amounts with fixed inputs, the efficiency analysis showed the great potential of MBR and A2 /O in terms of achieving an optimal balance among investment, energy consumption, pollutant removal, and GHG emissions. Both technologies display a co-benefit advantage in realizing the goals of aquatic and air quality improvements. There were 28 plants with MBR technology in this study. These plants located in 16 provinces and the average capacity was 3 × 104 m3 /d. Compared with plants adopting other technologies, although MBR plants had the highest average fixed-asset investment of 33 RMB Yuan/m3 , the distinguished advantage in high pollutant removals, especially SS (219 mg/L in average) and TP (3.4 mg/L in average), made MBR show the highest average efficiency. In our study, direct GHG emissions of plants with MBR were relatively low, leading to a moderate level of total GHG emissions, though MBR was claimed as a technology with high energy consumption (Tolkou and Zouboulis, 2016). Regarding the detailed performance of MBR, the average values of removals of four pollutants and GHG emissions per fixed-asset investment (EP1 to EP5 ) were 37, 25, 2, 0.4, and 123 g/RMB Yuan, while the average values of sub-indicators EE6 to EE10 were 1236, 890, 78, 13, and 4214 g/kWh, respectively. On the other hand, MBR was a new technology with less practical observations in the sample; thus, the range of efficiency of the MBR plants was quite large. The performance of MBR might be more stable and improved with more applications spread in China. In fact, MBR has gained considerable attention due to its remarkable performance in the removal of nutrients and been regarded as one of the most advanced technologies which would possibly be used widely in the future in China (Zheng et al., 2010). Nevertheless, the cost of MBR might be the primary concern for local governments and utility companies. According to the results of efficiency assessment, it is suggested that MBR be promoted for the sake of protecting sensitive water bodies, as it is more suit-

Pollutants Removal (mg/L)

GHG Emissions (mg CO2 eq/L)

COD

SS

TN

TP

275 390

194 240

23 27

3.2 3.5

990 1894

able for those with high environmental requirements and economic affordability as well. As one of the most widely applied biological treatment technologies in China, the A2 /O process was adopted in 216 sample plants located in 25 provinces. The average treatment capacity of A2 /O plants was 7.4 × 104 m3 /d, indicating that most A2 /O plants have larger capacities. Different from plants adopting MBR, plants with A2 /O process outperformed other plants because of the moderate level of investment and electricity input (25 RMB Yuan/m3 and 0.30 kWh/m3 in average) and high COD removal (310 mg/L in average). Moreover, A2 /O achieved best pollutants removal rate of COD and TP and tended to produce less CH4 and N2 O and remove more nutrients during the treatment process (Soda et al., 2013; Yan et al., 2014). Regarding the detailed performance of A2 /O, the average values of removals of four pollutants and GHG emission per fixed-asset investment (EP1 to EP5 ) were 39, 26, 3, 0.4, and 141 g/RMB Yuan, respectively. Furthermore, the average values of sub-indicators EE6 to EE10 were 1193, 808, 93, 12, and 4232 g/kWh, respectively. Compared with MBR, it is more practical to promote A2 /O in areas where the fixed-asset investment cost is a considerable constraint while the reduction of pollution load is not extremely required. 3.3.3. Tertiary treatment for water reuse purposes There are 153 plants with tertiary treatment processes among the samples with 4.34 × 106 m3 /d of the total scale of wastewater reclamation. Tertiary treatment processes in WWTPs are built to remove the excess nutrients and other contaminants, generally for the purpose of water reuse (Asano and Levine, 1996). Wastewater reclamation and reuse is an alternative source for water supply (Chu et al., 2004), especially in areas where water resources are scarce and population and economic growth is rapid like China (Yi et al., 2011). Generally, the discharge standard of pollutants for the tertiary treatment was stricter than the secondary treatment for water reuse purposes, requiring the tertiary treatment not only reduce common pollutants, such as COD, SS, TN and TP, but also reduce other pollutants, such as pathogens, pharmaceuticals and personal ˜ care products (Munoz and Caus, 2005). To achieve more comprehensive pollution control targets, plants with the tertiary treatment process would usually increase fixed-asset investment, energy and chemical consumption, GHG emissions, and other management costs (Molinos-Senante et al., 2010; Fine and Hadas, 2012). As for sample plants in this study, although the average fixed-asset investment of plants with tertiary treatment processes (22 RMB Yuan/m3 ) was lower than it of plants without tertiary treatment processes (26 RMB Yuan/m3 ), the corresponding average distance function

Please cite this article in press as: Zeng, S., et al., Efficiency assessment of urban wastewater treatment plants in China: Considering greenhouse gas emissions. Resour Conserv Recy (2016), http://dx.doi.org/10.1016/j.resconrec.2016.12.005

G Model RECYCL-3434; No. of Pages 9 8

ARTICLE IN PRESS S. Zeng et al. / Resources, Conservation and Recycling xxx (2016) xxx–xxx

due to the fact that the advantage of wastewater reclamation in the resources conservation and pollution reduction were not fully considered in the distance function, the results still suggest that the conflict between the benefit of advanced wastewater treatment and its externality should be emphasized from the holistic perspective. 4. Conclusion

Fig. 6. Boxplots of distance function values of plants with tertiary treatment.

value, electricity consumption, GHG emissions, and COD, SS, TN, TP removals for plants with tertiary treatment processes (0.342, 0.33 kWh/m3 , 1243 mg/L, 324 mg/L, 215 mg/L, 24 mg/L, and 3.0 mg/L) were all higher than those of plants without tertiary treatment processes (0.319, 0.29 kWh/m3 , 1074 mg/L, 289 mg/L, 199 mg/L, 22 mg/L, and 2.9 mg/L). Taking No. 1 plant as an example of tertiary treatment plants, the total treatment capacity was 40,000 m3 /d and 12.5% (5000 m3 /d) of treated wastewater was reclaimed and recovered for water reuse purpose. The distance value of No. 1 plant equaled to 0, indicating the current best practice towards sustainability among plants with tertiary treatment process. The details of No.1 plant were listed in Table 3. The fixed-asset investment was around the first quartile of all plants. Although the electricity consumption and GHG emissions were higher than 76% and 91% plants, the high pollutants removal of COD, SS, TP, and TN contributed benefits to the efficiency value. The pollutants removal rates of COD, SS, TN and TP of No.1 WWTP were 90%, 93%, 75% and 79%, respectively. The details of 2 specific WWTPs were shown in Table 6. Plant No. 1056 with tertiary treatment process removed much more pollutants (i.e. COD, SS, TN and TP), but it also had higher fixed-asset investment, electricity consumption and GHG emissions than plant No. 602. According to the essence of distance function method i.e. integrated assessment, the distance function value of Plant No. 1056 with tertiary treatment process was bigger than Plant No. 602 without tertiary treatment. It meant that Plant No. 602 without tertiary treatment was more efficient. The results indicated that the benefit of effluent quality improvement might not be enough to compensate for the disadvantages of high energy inputs and GHG emissions, under our assessment framework which concentrated only on four types of bulk pollutants. To further analyze the effect of containing the tertiary treatment process on the efficiency, we divided all the sample plants into three groups: 926 WWTPs without capacity of tertiary treatment, 83 WWTPs with a small capacity of tertiary treatment (less than 50% of total amount of treated wastewater) and 70 WWTPs with a large capacity of tertiary treatment (greater than 50% of total amount of treated wastewater), whose average distance function values were 0.319, 0.324, and 0.363, respectively, as shown in Fig. 6. With the increase of the capacity of tertiary treatment, the performance went bad remarkably. The results of Kruskal-Wallis test showed the difference of efficiency (distance function value) among three groups was significant (p < 0.1), while both economic cost and energy consumption are meaningful for the explanation of the efficiency gap (Table 4). Although the plants with the tertiary treatment processes were “punished” in the assessment, partially

An extensive assessment regarding the efficiency of 1079 urban WWTPs throughout China was performed. With the approach of distance function, positive and negative environmental impacts, economic cost, and energy utilization of WWTPs were considered together. With GHG emission taken as an undesirable output of WWTPs, the co-benefit of controlling water pollution and mitigating climate change was verified in the assessment. The results are useful in explaining the differences among WWTPs and their potential in performance improvement from an integrated perspective. The parameterized distance function could also be used for benchmarking purposes. By identifying factors affecting efficiency on the basis of the statistical analysis of a large sample, including plant scale, treatment technology adopted, the capacity of tertiary treatment, this study provides further information for future improvement in the design and operation of urban WWTPs. Despite the relevance of the results in terms of understanding gaps among urban WWTPs or their potential for increasing their efficiency and, therefore, their environmental performance from an integrated perspective, it is suggested future research should focus on the extension of the input and output set, e.g., the consideration of chemical consumption and its potential ecological hazard. Due to the variability in the performance, e.g. the daily oscillation of effluent quality, uncertainty in the assessment is also worth discussing. Acknowledgment This work was supported by the National Natural Science Foundation of China (51308320). References Asano, T., Levine, A.D., 1996. Wastewater reclamation, recycling and reuse: past, present, and future. Water Sci. Technol. 33, 1–14. Balkema, A.J., Preisig, H.A., Otterpohl, R., Lambert, F.J.D., 2002. Indicators for the sustainability assessment of wastewater treatment systems. Urban Water 4, 153–161. Bani Shahabadi, M., Yerushalmi, L., Haghighat, F., 2009. Impact of process design on greenhouse gas (GHG) generation by wastewater treatment plants. Water Res. 43, 2679–2687. Bani Shahabadi, M., Yerushalmi, L., Haghighat, F., 2010. Estimation of greenhouse gas generation in wastewater treatment plants − model development and application. Chemosphere 78, 1085–1092. Botti, L., Briec, W., Cliquet, G., 2009. Plural forms versus franchise and company-owned systems: a DEA approach of hotel chain performance. Omega 37, 566–578. CHANGE, I.P.O.C., 2006. 2006 IPCC Guidelines for National Greenhouse Gas Inventories. China Urban Water Association, 2012. Urban Drainage Statistics Yearbook 2012. China Urban Water Association, Beijing, China (in Chineses). Chu, J., Chen, J., Wang, C., Fu, P., 2004. Wastewater reuse potential analysis: implications for China’s water resources management. Water Res. 38, 2746–2756. Corominas, L., Flores-Alsina, X., Snip, L., Vanrolleghem, P.A., 2012. Comparison of different modeling approaches to better evaluate greenhouse gas emissions from whole wastewater treatment plants. Biotechnol. Bioeng. 109, 2854–2863. Dasgupta, S., Huq, M., Wheeler, D., Zhang, C.H., 2001. Water pollution abatement by Chinese industry: cost estimates and policy implications. Appl. Econ. 33, 547–557. Diaz-Balteiro, L., Romero, C., 2008. Valuation of environmental goods: a shadow value perspective. Ecol. Econ. 64, 517–520. Dong, J., Chi, Y., Tang, Y., Wang, F., Huang, Q., 2014. Combined life cycle environmental and exergetic assessment of four typical sewage sludge treatment techniques in China. Energy Fuels 28, 2114–2122. Doorn, M., Strait, R., Barnard, W., Eklund, B., 1997. Estimates of Global Greenhouse Gas Emissions from Industrial and Domestic Wastewater Treatment. Final

Please cite this article in press as: Zeng, S., et al., Efficiency assessment of urban wastewater treatment plants in China: Considering greenhouse gas emissions. Resour Conserv Recy (2016), http://dx.doi.org/10.1016/j.resconrec.2016.12.005

G Model RECYCL-3434; No. of Pages 9

ARTICLE IN PRESS S. Zeng et al. / Resources, Conservation and Recycling xxx (2016) xxx–xxx

Report, September 1994–March 1997. Pechan (EH) and Associates, Inc., Durham, NC (United States). Färe, R., Grosskopf, S., 2000. Theory and application of directional distance functions. J. Product. Anal. 13, 93–103. Färe, R., Grosskopf, S., Lovell, C.K., Pasurka, C., 1989. Multilateral productivity comparisons when some outputs are undesirable: a nonparametric approach. Rev. Econ. Stat. 71, 90–98. Färe, R., Grosskopf, S., Noh, D.-W., Weber, W., 2005. Characteristics of a polluting technology: theory and practice. J. Econ. 126, 469–492. Färe, R., Grosskopf, S., Weber, W.L., 2006. Shadow prices and pollution costs in U.S. agriculture. Ecol. Econ. 56, 89–103. Fine, P., Hadas, E., 2012. Options to reduce greenhouse gas emissions during wastewater treatment for agricultural use. Sci. Total Environ. 416, 289–299. Flores-Alsina, X., Gallego, A., Feijoo, G., Rodriguez-Roda, I., 2010. Multiple-objective evaluation of wastewater treatment plant control alternatives. J. Environ. Manage. 91, 1193–1201. Flores-Alsina, X., Corominas, L., Snip, L., Vanrolleghem, P.A., 2011. Including greenhouse gas emissions during benchmarking of wastewater treatment plant control strategies. Water Res. 45, 4700–4710. Foley, J., De Haas, D., Yuan, Z., Lant, P., 2010. Nitrous oxide generation in full-scale biological nutrient removal wastewater treatment plants. Water Res. 44, 831–844. Hernández-Sancho, F., Sala-Garrido, R., 2009. Technical efficiency and cost analysis in wastewater treatment processes: a DEA approach. Desalination 249, 230–234. Hernandez-Sancho, F., Molinos-Senante, M., Sala-Garrido, R., 2011a. Energy efficiency in Spanish wastewater treatment plants: a non-radial DEA approach. Sci. Total Environ. 409, 2693–2699. Hernandez-Sancho, F., Molinos-Senante, M., Sala-Garrido, R., 2011b. Cost modelling for wastewater treatment processes. Desalination 268, 1–5. Hoibye, L., Clauson-Kaas, J., Wenzel, H., Larsen, H.F., Jacobsen, B.N., Dalgaard, O., 2008. Sustainability assessment of advanced wastewater treatment technologies. Water Sci. Technol. 58, 963. Jin, L., Zhang, G., Tian, H., 2014. Current state of sewage treatment in China. Water Res. 66, 85–98. Kalbar, P.P., Karmakar, S., Asolekar, S.R., 2012. Selection of an appropriate wastewater treatment technology: a scenario-based multiple-attribute decision-making approach. J. Environ. Manage. 113, 158–169. Kampschreur, M.J., Temmink, H., Kleerebezem, R., Jetten, M.S., van Loosdrecht, M.C., 2009. Nitrous oxide emission during wastewater treatment. Water Res. 43, 4093–4103. Lassaux, S., Renzoni, R., Germain, A., 2007. Life cycle assessment of water from the pumping station to the wastewater treatment plant. Int. J. Life Cycle Assess. 12, 118–126. Lee, M.W., Hong, S.H., Choi, H., Kim, J.-H., Lee, D.S., Park, J.M., 2008. Real-time remote monitoring of small-scaled biological wastewater treatment plants by a multivariate statistical process control and neural network-based software sensors. Process Biochem. 43, 1107–1113. Lee, S.-c., Oh, D.-h., Lee, J.-d., 2014. A new approach to measuring shadow price: reconciling engineering and economic perspectives. Energy Econ. 46, 66–77. Lorenzo-Toja, Y., Vázquez-Rowe, I., Chenel, S., Marín-Navarro, D., Moreira, M.T., Feijoo, G., 2015. Eco-efficiency analysis of Spanish WWTPs using the LCA + DEA method. Water Res. 68, 651–666. Matos, R., Association, I.W., 2003. Performance Indicators for Wastewater Services. IWA Publishing. Molinos-Senante, M., Hernández-Sancho, F., Sala-Garrido, R., 2010. Economic feasibility study for wastewater treatment: a cost–benefit analysis. Sci. Total Environ. 408, 4396–4402. Molinos-Senante, M., Hernández-Sancho, F., Sala-Garrido, R., 2011a. Cost–benefit analysis of water-reuse projects for environmental purposes: a case study for Spanish wastewater treatment plants. J. Environ. Manage. 92, 3091–3097. Molinos-Senante, M., Hernández-Sancho, F., Sala-Garrido, R., Garrido-Baserba, M., 2011b. Economic feasibility study for phosphorus recovery processes. Ambio 40, 408–416.

9

˜ Munoz, J., Caus, J., 2005. Maximizing the environmental benefits in the operation of a WWTP of a seaside tourist area, Technical workshop: the integration of reclaimed water in water resource management. Muga, H.E., Mihelcic, J.R., 2008. Sustainability of wastewater treatment technologies. J. Environ. Manage. 88, 437–447. Murty, M., Kumar, S., Paul, M., 2006. Environmental regulation, productive efficiency and cost of pollution abatement: a case study of the sugar industry in India. J. Environ. Manage. 79, 1–9. Pan, T., Zhu, X.-D., Ye, Y.-P., 2011. Estimate of life-cycle greenhouse gas emissions from a vertical subsurface flow constructed wetland and conventional wastewater treatment plants: a case study in China. Ecol. Eng. 37, 248–254. Pasqualino, J.C., Meneses, M., Abella, M., Castells, F., 2009. LCA as a decision support tool for the environmental improvement of the operation of a municipal wastewater treatment plant. Environ. Sci. Technol. 43, 3300–3307. Pittman, R.W., 1983. Multilateral productivity comparisons with undesirable outputs. Econ. J. 93, 883–891. Quadros, S., Rosa, M.J., Alegre, H., Silva, C., 2010. A performance indicators system for urban wastewater treatment plants. Water Sci. Technol. 62, 2398–2407. Rahman, S.M., Eckelman, M.J., Onnis-Hayden, A., Gu, A.Z., 2016. Life-cycle assessment of advanced nutrient removal technologies for wastewater treatment. Environ. Sci. Technol. 50, 3020–3030. Rodriguez-Garcia, G., Hospido, A., Bagley, D.M., Moreira, M.T., Feijoo, G., 2012. A methodology to estimate greenhouse gases emissions in Life Cycle Inventories of wastewater treatment plants. Environ. Impact Assess. Rev. 37, 37–46. Sala-Garrido, R., Molinos-Senante, M., Hernández-Sancho, F., 2011. Comparing the efficiency of wastewater treatment technologies through a DEA metafrontier model. Chem. Eng. J. 173, 766–772. Shephard, R.W., Gale, D., Kuhn, H.W., 1970. Theory of Cost and Production Functions. Princeton University Press, Princeton, NJ. Soda, S., Arai, T., Inoue, D., Ishigaki, T., Ike, M., Yamada, M., 2013. Statistical analysis of global warming potential, eutrophication potential, and sludge production of wastewater treatment plants in Japan. J. Sustain. Energy Environ. 4, 33–40. Song, R., Zhu, J., Hou, P., Wang, H., 2013. Getting Every Ton of Emissions Right: An Analysis of Emission Factors for Purchased Electricity in China. World Resources Institute Working Paper. http://www.wri.org/publication/gettingevery-ton-emissions-right. Sueyoshi, T., Aoki, S., 2001. A use of a nonparametric statistic for DEA frontier shift: the Kruskal and Wallis rank test. Omega-Int. J. Manag. Sci. 29, 1–18. Suh, Y.-J., Rousseaux, P., 2002. An LCA of alternative wastewater sludge treatment scenarios. Resour. Conserv. Recycl. 35, 191–200. Tidåker, P., Kärrman, E., Baky, A., Jönsson, H., 2006. Wastewater management integrated with farming − an environmental systems analysis of a Swedish country town. Resour. Conserv. Recycl. 47, 295–315. Tolkou, A., Zouboulis, A., 2016. Effect of climate change in WWTPs with a focus on MBR infrastructure. Desalin. Water Treat. 57, 2344–2354. Tupper, H.C., Resende, M., 2004. Efficiency and regulatory issues in the Brazilian water and sewage sector: an empirical study. Util. Policy 12, 29–40. Wang, J., Zhang, J., Xie, H., Qi, P., Ren, Y., Hu, Z., 2011. Methane emissions from a full-scale A/A/O wastewater treatment plant. Bioresour. Technol. 102, 5479–5485. Yan, X., Li, L., Liu, J., 2014. Characteristics of greenhouse gas emission in three full-scale wastewater treatment processes. J. Environ. Sci. 26, 256–263. Yi, L., Jiao, W., Chen, X., Chen, W., 2011. An overview of reclaimed water reuse in China. J. Environ. Sci. 23, 1585–1593. Yoshida, H., Mønster, J., Scheutz, C., 2014. Plant-integrated measurement of greenhouse gas emissions from a municipal wastewater treatment plant. Water Res. 61, 108–118. Zheng, X., Zhou, Y., Chen, S., Zheng, H., Zhou, C., 2010. Survey of MBR market: trends and perspectives in China. Desalination 250, 609–612. Zhong, W., Yuan, W., Li, S.X., Huang, Z., 2011. The performance evaluation of regional R&D investments in China: an application of DEA based on the first official China economic census data. Omega 39, 447–455.

Please cite this article in press as: Zeng, S., et al., Efficiency assessment of urban wastewater treatment plants in China: Considering greenhouse gas emissions. Resour Conserv Recy (2016), http://dx.doi.org/10.1016/j.resconrec.2016.12.005