w a t e r r e s e a r c h 5 6 ( 2 0 1 4 ) 2 9 2 e3 0 3

Available online at www.sciencedirect.com

ScienceDirect journal homepage: www.elsevier.com/locate/watres

Influence of data collection schemes on the Life Cycle Assessment of a municipal wastewater treatment plant Hiroko Yoshida*, Julie Clavreul, Charlotte Scheutz, Thomas H. Christensen Department of Environmental Engineering, Technical University of Denmark, 2800 Kgs. Lyngby, Denmark

article info

abstract

Article history:

A Life Cycle Assessment (LCA) of a municipal wastewater treatment plant (WWTP) was

Received 20 November 2013

conducted to illustrate the effect of an emission inventory data collection scheme on the

Received in revised form

outcomes of an environmental impact assessment. Due to their burden in respect to data

27 February 2014

collection, LCAs often rely heavily on existing emission and operational data, which are

Accepted 7 March 2014

gathered under either compulsory monitoring or reporting requirements under law. In this

Available online 16 March 2014

study, an LCA was conducted using three input data sources: Information compiled under compulsory disclosure requirements (the European Pollutant Release and Transfer Regis-

Keywords:

try), compliance with national discharge limits, and a state-of-the-art emission data

Wastewater

collection scheme conducted at the same WWTP. Parameter uncertainty for each collec-

Sewage sludge

tion scheme was assessed through Monte Carlo simulation. The comparison of the results

Life cycle assessment

confirmed that LCA results depend heavily on input data coverage. Due to the threshold on

Environmental information

reporting value, the E-PRTR did not capture the impact for particulate matter emission,

Compulsory disclosure requirement

terrestrial acidification, or terrestrial eutrophication. While the current practice can capture more than 90% of non-carcinogenic human toxicity and marine eutrophication, an LCA based on the data collection scheme underestimates impact potential due to limitations of substance coverage. Besides differences between data collection schemes, the results showed that 3e13,500% of the impacts came from background systems, such as from the provisioning of fuel, electricity, and chemicals, which do not need to be disclosed currently under E-PRTR. The incidental release of pollutants was also assessed by employing a scenario-based approach, the results of which demonstrated that these nonroutine emissions could increase overall WWTP greenhouse gas emissions by between 113 and 210%. Overall, current data collection schemes have the potential to provide standardized data collection and form the basis for a sound environmental impact assessment, but several improvements are recommended, including the additional collection of energy and chemical usage data, the elimination of a reporting threshold, the expansion of substance coverage, and the inclusion of non-point fugitive gas emissions. ª 2014 Elsevier Ltd. All rights reserved.

* Corresponding author. Miljoevej. Building 113, Room 226, 2800 Kgs. Lyngby, Denmark. Tel.: þ45 4525 1508. E-mail address: [email protected] (H. Yoshida). http://dx.doi.org/10.1016/j.watres.2014.03.014 0043-1354/ª 2014 Elsevier Ltd. All rights reserved.

w a t e r r e s e a r c h 5 6 ( 2 0 1 4 ) 2 9 2 e3 0 3

1.

Introduction

Life Cycle Assessment (LCA) has in recent years gained interest from the wastewater and sewage sludge treatment sector (Friedrich et al., 2007; Corominas et al., 2013a,b; Yoshida et al., 2013a,b,c). An LCA aims to quantify, organize, and translate environmental emissions from all involved processes into one or multiple environmental impact indicators. LCAs have been used, for example, to assess alternative wastewater and sewage sludge treatment technologies from an environmental point of view, thereby enabling the quantitative evaluation of trade-offs (e.g. nutrient recovery versus health risk, upgrade of nutrient recovery versus energy, and chemical consumption). One of the most critical issues in performing an LCA is establishing reliable inventory data. In practice, generating site-specific monitoring data for all substances known to cause adverse health and environmental impacts is prohibitively expensive, and in many cases it is not even feasible, despite the fact that data gaps relating to flows between unit processes and emissions into the environment introduce epistemic uncertainty in LCA results and therefore a systemic underestimation of environmental impacts (Finnveden, 2000; Huijbregts et al., 2001). Hence, the integrity of the LCA depends largely on the utilization of currently available operational and emission data and the assumptions made to fill data gaps (Bjorklund, 2002; Huijbregts et al., 2001; Reap et al., 2008). The question remains as to how significantly the outcome of an LCA study depends on available data (e.g. source, quantity, quality) and the assumptions made to close obvious data gaps. Due to the nature of the operation, pollution control facilities such as wastewater treatment plants (WWTPs) already collect a range of data concerning flows and emissions as part of everyday operational schemes and in order to fulfill public reporting obligations. In Denmark, as also is the case in many other countries, specific plants are approved or licensed subject to the fulfillment of certain emission standards. Actual emissions, for example the quality of treated wastewater prior to discharge, must be monitored and reported regularly. Yet, limitations in line with such data lie in substance coverage, since not all substances known to have adverse environmental or health impacts are regulated, while some emission pathways are also exempted from the requirement. In addition, these monitoring data are often not organized or made easily accessible to the public. In Europe, industrial entities - including WWTPs - of a certain size are required to report environmental emissions and transfers of pollutants via air, water, and waste to the European Pollutant Release and Transfer Register (E-PRTR). This web-based register, which is aimed at informing the public about the release of pollutants from industrial facilities, replaces and improves upon the former EPER register implemented under Council Directive 96/61/EC and later codified as 2008/1/EC (EEA, 2013). Currently some 28,000 industrial facilities report their environmental data to the E-PRTR every three years. A threshold for minimal production capacity applies, though the European Commission assumes that the E-PRTR covers 90% of industrial emissions for 91 substances in Europe (Wursthorn et al., 2011). Adopting an institutionally backed

293

data collection scheme such as E-PRTR for LCA would not only reduce the burden of data collection, but also ensure the standardization of data collection and reporting schemes across industrial sectors, as well as provide timely inventory updates. These changes would expand the application of LCAs by evaluating data coverage and quality captured by current data collection schemes that require sector-by-sector evaluations. Hence, the goal of this study was to illustrate in a quantitative way how the basis of inventory data affects the outcome of a WWTP LCA by using a specific WWTP located in Copenhagen, Denmark, as a case study. We used three levels of information for establishing inventory data, from using routinely reported data to using data from an advanced, specific monitoring campaign performed at the plant over more than one year. We used: (L1) the E-PRTR reporting guideline, (L2) emissions monitoring mandated by Danish regulations, and (L3) emissions data from a state-of-the-art measurement campaign. This study evaluated on-site emissions from WWTPs, as the current E-PRTR does not require industry practitioners to report on energy and material consumption. However, in order to place the results into the perspective of a conventional LCA, emissions from up- and downstream processeswere included in one scenario (L3þ). The propagation of uncertainties was also conducted, in order to evaluate the influence of variations in measurements on the outcome of LCA.

2.

Methods

The study follows the four steps defined in the ISO standard 14040 (2006), namely goal and scope definition, a Life Cycle Inventory (LCI), a Life Cycle Impact Assessment (LCIA), and the interpretation of results. This section details the first three steps and a review of the uncertainty analysis methods used herein. The results and their interpretation are presented in Section 3. The paper focuses on the three data collection schemes, while detailed documentation on the assumptions and parameter values is provided in supporting information (SI).

2.1.

Goal and scope

The objective of this study was to assess the possibility of adopting the current compulsory environmental disclosure requirement (E-PRTR) when conducting a wastewater LCA, by quantifying potential environmental impacts associated with the operation of Avedøre WWTP, located southwest of Copenhagen, Denmark, under the three input data collection schemes. The reasons for conducting the study included methodological development, and the implications of the results were limited to the discussion on the viability of emission data collection schemes. Since the E-PRTR requires only on-site emissions from WWTPs, an assessment of the construction and demolition phases was not included in this study. The system under study is an urban WWTP serving 265,000 inhabitants (SI-1) and is equipped with primary and secondary wastewater treatment systems utilizing advanced nitrogen

294

w a t e r r e s e a r c h 5 6 ( 2 0 1 4 ) 2 9 2 e3 0 3

Table 1 e Impact categories and normalization references used in the impact assessment. Impact category

Abbre-viation

Global warming

GW

Acidification

AC

Eutrophication, terrestrial

TE

Eutrophication, aquatic (marine)

ME

Photochemical ozone formation

POF

Ecotoxicity (freshwater)

ET

Human toxicity, cancer effects

HTc

Human toxicity, non-cancer effects Particulate matter

HTnc

a b

PM

LCIA method IPCC (Forster et al. 2007) Accumulated Exceedance (Seppa¨la¨ et al., 2006; Posch et al., 2008) Accumulated Exceedance (Seppa¨la¨ et al., 2006; Posch et al., 2008) EUTREND model as implemented in ReCiPe (Struijs et al., 2009) ReCiPe 2008 (Van Zelm et al., 2008) USEtox (Rosenbaum et al., 2008) USEtox (Rosenbaum et al., 2008) USEtox (Rosenbaum et al., 2008) Humbert et al. (2009, 2011)

Unit

Normalization reference

8.10 E þ 3 for reference year 2010 (Laurent et al., 2013) kg SO2-eq 4.96 E þ 1 for reference year 2000 (Laurent et al., 2013) kg NOx-eq 1.15 E þ 2 reference year 2000 (Laurent et al., 2013) kg N-eq 9.38 reference year 2000 (Laurent et al., 2013) kg NMVOC 5.67 E þ 1 referece year 2000 (Laurent et al., 2013) 6.65 E þ 2 for referece year 2010 CTUea (Laurent et al., 2013) 5.42E-5 for referece year 2010 CTUhb (Laurent et al., 2013) CTUh 1.10E-3 for reference year 2010 (Laurent et al., 2013) kg PM2.5-eq 2.76 for reference year 2000 (Laurent et al., 2013) kg CO2-eq

CTUe stands for comparative toxic units for eco-toxicity and is described as potentially affected fraction of species (PAF)  m3  day. CTUh stands for comparative toxic units for human toxicity and it described as disease cases.

and phosphorus removal. Treated effluent is discharged into the ocean, while generated sludge is anaerobically digested, dewatered, and incinerated. Ashes from bag filters and electrostatic precipitators are disposed into an on-site monolandfill, and biogas from the anaerobic digester is utilized for electricity and process heating. The functional unit of this study involved the treatment of the daily average inflow of wastewater (15% industrial, 85% municipal) entering the WWTP in 2011 (79,700 m3 d1). The composition of the wastewater is available in SI-2. An emission inventory was compiled based on the Avedøre WWTP operation in 2011. Since the time scope of this assessment is limited to the operational phase, the construction, maintenance, and demolition of the plants were excluded from the overall assessment, while a time horizon of 100 years was applied for the impact assessment. All emissions from the treatment of wastewater are deemed to happen instantaneously, so emission timing was not considered in this study. Since environmental monitoring and reporting is conducted primarily for benchmarking purposes and not in order to facilitate decision making directly, following the classification of the International Reference Life Cycle Data System (ILCD, 2011), the attributional approach was elected. The multi-functionality problem was addressed using system expansion, including substitution, for upstream and downstream processes. The system boundary of this study included all processes between the arrival of wastewater into the plant and final release into the environment (effluent emissions into the ocean, and fugitive and stack emissions into the air). Leachate from the ash landfill is collected and sent back to the head of the wastewater plant, which was considered a form of internal recycling. As the landfilling of inert material (i.e. sludge ash) is excluded from the scope of an E-PRTR, ash landfill emissions were not included in the assessment. The provisioning of chemicals, fuel, and electricity in upstream

processes, as well as the electricity substitution as a result of utilizing biogas, was also included. Primary data (WWTP emissions and the consumption of materials and energy) were based on plant operational data and extensive site-specific data collection efforts, which took place between 2011 and 2013 (Yoshida et al., 2013b, 2013c). Secondary data (production of materials and energy) were taken from the Ecoinvent database v2.2, in line with spatial, temporal, and technological relevance. A list of secondary data sources can be found in SI-3. Midpoint impacts were assessed in this study. The choice of an LCIA method for each impact category was made based on the recommendation made by the ILCD, which provided a list of LCIA methods considered to be the best at the time the study commenced (Hauschild et al., 2013). The nine impact categories included in the study are presented in Table 1. Some impact categories recommended by the ILCD guideline were excluded in this study. First, marine and terrestrial ecotoxicity and photochemical ozone formation impacts on vegetation were excluded, since no recommendations were made for those impact categories (Hauschild et al., 2013). Secondly, land use, water use, resource depletion, and impacts on human health and ecosystems from ionizing radiation were excluded, as they are beyond the scope of the E-PRTR and no relevant information was collected under the current scheme. Concerning ozone depleting this impact category was not included, even though ozone-depleting hydro-fluorocarbons are included in the list of pollutants under E-PRTR, urban WWTPs are not required to monitor or report on these substances to the registry. The EPRTR does require urban WWTPs to report any emissions of N2O, which is also a major ozone depletion substance (Ravishankara et al., 2009). However, N2O was not listed in the ILCD recommended LICA method for ozone depletion (the steady-state ODPs by WMO, 1999), and hence the impact ozone-depletion was eliminated from the list of impact categories.

w a t e r r e s e a r c h 5 6 ( 2 0 1 4 ) 2 9 2 e3 0 3

Concerning global warming (GW), only CO2 (fossil), CH4, and N2O were included, as there is no evidence of considerable amounts of any other greenhouse gases (GHGs) being emitted from WWTPs. The results were normalized by the amount of impact potential exerted by an average European over a year, presented as the ‘person equivalent’ (PE). The normalization reference used in this study can be found in Table 1. The LCA modeling was conducted using a recently developed mass flow-based LCA tool, EASETECH (Environmental Assessment System for Environmental TECHnologies). As presented in more detail by Clavreul et al. (2013), this model allows for a detailed modeling of each substance flow throughout the entire system as well as the relationships between flow concentrations and emissions. In addition, the EASETECH model enables extensive uncertainty propagation, as discussed further in section 2.3.

2.2.

Input data and modeling approach

Four levels of inventory data were considered in this study, and a graphical overview of emission data for each level is presented in Fig. 1. Under E-PRTR, a municipal WWTP is only required to report emissions via the discharge of treated wastewater and a few emissions into the air (L1). Alternatively, discharge limits are set on effluent, stack emissions from waste incineration, the combustion of biogas, and the landfilling of incineration ash (L2). Under the state-of-the-art data collection scheme, specific data for each unit process and substance flows were traced throughout the WWTP (L3). Since no energy and chemical consumption data are reported under E-PRTR, background emissions were not included in the L1, L2, or L3 scenarios, thus allowing for the comparison of three different data gathering schemes for on-site emissions. Lastly, the inventory for L3 was expanded to include electricity and chemical consumption, hence the requirement for upstream and downstream emissions (L3þ). The comparison between L3 and L3þ illustrates the contribution of background systems. Table 2 tabulates the elements and pollutants that were included in the three different data collection schemes. Details of each data collection method are discussed in the following subsections.

2.2.1. (L1)

Compulsory disclosure of environmental information

Sixty-five industrial activities in nine categories are currently subject to E-PRTR regulation. E-PRTR tracks the movements and emissions of a total of 91 substances (six GHGs, 11 other gases, eight heavy metals, 23 pesticides, 20 chlorinated organic substances, 16 other organic substances, and seven inorganic substances). These emissions are divided into three environmental media types, each of which has a specific list of substances: 60 substances for air, 71 for water, and 61 for land. The number of reporting targets is further limited based on the type of industrial activity; for example, an urban WWTP with a capacity of 100,000 equivalent people or more ought to report 18 substances for air emissions and 42 substances for water. For incineration and biogas combustion facilities, production capacity thresholds are applied: A minimum capacity of 50 metric tons per day of managed non-hazardous waste and a minimum energy production capacity of 50 MW. Each

295

substance has a specific reporting threshold value based on its potency, and if the measured emissions are below this threshold value, the facilities are not obliged to report to EPRTR. A list of substance emissions included in L1 is provided in SI-4. A number of substances are exempted from measurement obligations, and their emissions are estimated based on operational data and empirical emission factors, rather than relying on site-specific emission data. For instance, E-PRTR recommends several internationally approved calculation methods to quantify GHG emissions, and the IPCC guideline is currently the only effective approach for applying to wastewater treatment. In the IPCC method, carbon contained in wastewater is assumed to be biogenic in origin, and CO2 emissions are therefore excluded from the reporting guideline. Fugitive emissions of CH4 and N2O are estimated based on organic matter and nitrogen removal rates. As an alternative to these calculation methods, E-PRTR also accepts sitespecific estimations of GHGs based on operational data. The estimation of GHG emissions is detailed in SI-5.

2.2.2.

Compliance with regulatory discharge limits (L2)

The WWTP monitors emissions of pollutants into water and the air, in order to comply with environmental regulations and then to report these monitoring data to the authorities every year. Under Danish regulations, the WWTP is divided into four entities: Wastewater treatment and sludge stabilization, biogas combustion, sludge incineration, and inert material landfill, each of which is subject to specific discharge limits. The major statutory requirement against the chemical pollution of surface water in the EU is the Water Framework Directive (WFD 2000/60/EC), which calls for the establishment of a list of pollutants posing a significant risk to or via the aquatic environment in Europe. Five metals (Cd, Hg, Ni, Pb, and Zn) are prioritized as major pollutants, and different standards are set for inland surface waters as well as for transitional, coastal, and territorial waters. Member states have to comply with the standards and verify that concentrations of substances of concern do not accumulate in sediment and/or biota. Besides these high-priority pollutants, discharges of As, Cr, and Cu are regulated in Denmark, and there is no difference between the number of inorganic pollutants covered under the L1 and L2 schemes (SI-5). Emissions of particulates, namely SO2, NOx, HCl, HF, NH4, and CO, are measured continuously at the sludge incinerator stack. Besides these conventional air pollutants, the contents of heavy metals, dioxins, and PAH are analyzed and reported twice a year. Emission data for unburnt CH4, CO, and NOx from biogas utilization are monitored, but there is no regulation on the emission of heavy metals. Furthermore, there is no mandatory measurement of biogas leakage from WWT facilities (SI-6).

2.2.3. State-of-the-art data collection (L3) and inclusion of background systems (L3þ) Beyond what is monitored at the WWTP, to comply with environmental regulations, the following emissions data were developed to cover the possible range of compounds included in the impact assessment methodologies:

296

w a t e r r e s e a r c h 5 6 ( 2 0 1 4 ) 2 9 2 e3 0 3

Fig. 1 e Graphical overview of the three data collection schemes applied in the LCA study.

 Unlike assumptions made by the IPCC reporting guidelines, it is reported that the 4-14%proportion of carbon entering WWTPs is fossil in origin (Law et al., 2013). Based on the radiocarbon method (ASTM-D6866-12), it was determined that 14% of carbon in wastewater influent to the Avedøre WWTP was of fossil origin (SI-7)  A plant-integrated emission measurement of N2O and CH4 was conducted combining a tracer release with downwind plume measurements. This enabled the quantification of physical leakages and other diffusive gas emissions (Yoshida et al., 2013b). The rate of methane leakage was

quantified to be 3.0% of the generated methane, while 0.48% of N in incoming wastewater was emitted as N2O.  All inorganic pollutants, characterized in version 1.00 of the USEtox method (15 for human toxicity and 16 for ecotoxicity potential; Rosenbaum et al., 2008), were included in the inventory based on the substance flow analysis conducted at the Avedøre WWTP (Yoshida et al., 2013c). In the L3þ scenario the full extent of data collection was implemented, including upstream processes (e.g. for electricity consumption and the use of treatment chemicals) and

297

w a t e r r e s e a r c h 5 6 ( 2 0 1 4 ) 2 9 2 e3 0 3

Table 2 e List of pollutants and their inclusion in the three data collection schemes (GW: global warming potential, AC: acidification, TE: terrestrial eutrophication, AE: marine aquatic eutrophication, POF: photochemical ozone formation, HT: human toxicity, PM: particulate matter/respiratory inorganics, ET: eco-toxicity). Pollutant

Emissions to air L1 (E-PRTR) L2 (Danish L3 (state-of-theregulation) art LCA)

CO2 (fossil) CH4 N 2O CO NH4 NOx SOx Particulate Matter NMVOC N P Ag As Ba Be Cd Co Cr Cu Hg Mo Ni Pb Sb Se Sn Tl Zn V

Emissions to water Impact categories

Xa Xa Xb Xb Xb Xb

Xa Xa X X X X X

X X X X X X X X

GW GW/POF GW POF AC/TE/AE/PM AC/TE/AE/POF/PM AC/POF/PM PM

Xb

X

X

POF

Xc

X Xc Xc Xc X

Xc X Xc

Xc Xc

Xf X6 Xf Xf X Xf Xf Xf X for both biogas and incineration Xf Xf X X for biogas Xf for incineration Xf Xf Xf Xf Xf

ET/HT ET/HT ET/HT ET/HT ET/HT ET/HT ET/HT ET/HT ET/HT ET/HT ET/HT ET/HT ET/HT ET/HT ET/HT ET/HT ET/HT ET/HT

L1 (E-PRTR) L2 (Danish L3 (State-of-the- Impact regulation) art LCA) categories

X X

X X

X

Xd

X

X

X X X

Xd Xd X

X X

X X

X

X

X X X X X X X X X X X

ME FE ET/HT ET/HT ET/HT ET/HT ET/HT ET/HT ET/HT ET/HT ET/HT

Xe X X X

ET/HT ET/HT ET/HT ET/HT

Xe X Xe X X

ET/HT ET/HT ETP/HTP ET/HT ET/HT

a

Calculated based on the IPCC 2006 reporting guideline. Urban WWTPs are required to report those values even though the actual emission from the plant is below the threshold reporting value set by E-PRTR and no value are reported to the online registry. c Excluded from the inventory because regulations are only set for the sum of all heavy metals detected. d Included in the inventory because regulated under Danish regulation. However, these metals are not included in the list of priority pollutants set by the European Commission. e Excluded from the inventory since the concentration was under detection limit. f Excluded from the inventory due to the limitation of mass balance approach used in Yoshida et al. (2013a). Emissions to air of these metals were smaller than uncertainty ranges accompanied with the input and output flows. b

downstream processes (electricity production as a result of biogas combustion). The energy recovery efficiency was 37.5%. All heat generated from the combined heat and power process (CHP) was used for heating the anaerobic digester or released into the air, and thus it was excluded from the analysis. These parameter values were based on operational data from 2011 and a measurement campaign conducted from 2011 through 2012. A detailed explanation of parameter values and the technical modeling approach is included in SI-8 and SI-12.

2.3.

Uncertainty analysis methods

The uncertainty of the results for the four scenarios (L1, L2, L3, and L3þ) was evaluated by Monte Carlo analysis. In order to make a conservative estimate of uncertainties associated with

parameter values, uniform distribution was applied to the emissions. For parameters measured and recorded daily, normal or triangular distribution was assigned. In cases where the concentration of certain elements was below measurement detection limits, a uniform distribution between 0 and the detection limit was assigned. One hundred thousand Monte Carlo simulation runs were executed using random samples for each parameter based on their probability distributions (SI-13) Background system uncertainty was included in the assessment in two respects: Parameter uncertainty embedded in the Ecoinvent database and uncertainty as a result of data quality. Parameter uncertainties embedded in the Ecoinvent database were assessed by conducting a Monte Carlo simulation for each external process, in order to obtain the distribution

298

w a t e r r e s e a r c h 5 6 ( 2 0 1 4 ) 2 9 2 e3 0 3

of normalized impacts by encompassing the embedded uncertainty in each dataset. Uncertainty as a result of data quality was assessed based on the pedigree matrix approach adapted by the Ecoinvent data quality guideline (Weidema et al., 2012). The quality of each external process was evaluated based on its reliability, completeness, and temporal, geographical, and technological correlations to the target system (SI-14) Uncertainty analysis was performed on scenario L3þ to assess the influence of data variability and epistemic uncertainty, following the recommendation made by Clavreul et al. (2012). The sensitivity analysis involved varying the value of each parameter one-by-one by applying a 10% variation. The relative change in the result was then divided by the relative change in the parameter value, in order to calculate a sensitivity ratio (SR). Besides the emission data under normal and steady operational conditions, E-PRTR also calls upon the industry to report the incidental or non-routine release of pollutants. A scenario analysis was also conducted to estimate the effects of incidental and non-routine releases of pollutants on the overall environmental performance of the WWTP. Two scenarios were considered in terms of impacts on GW (on L3þ):  A first scenario investigated the case of the sub-optimal operation of the biological nitrogen removal (BNR) process. Direct N2O emission measurements showed that a suboptimal BNR operation could lead to the accumulation of NO3 in the reactor and elevate emissions of N2O, reaching 4.27% of nitrogen in wastewater influent through BNR instead of 0.48% under normal conditions (Yoshida et al., 2013c).  A second scenario investigated the incidental release of biogas. From spring through early summer, the WWTP had been experiencing foaming in its anaerobic digester. This accumulation of microbial foam on the surface of the activated sludge led to the inefficient use of the digester volume and lower gas recovery (Ganidi et al., 2009). In order to prevent overpressure in the anaerobic digester tank, biogas was released from the valve located at the top of the digester. As much as 32.7% of methane generated can be lost through pressure releases over these four months or so, while methane leakage remains at 3.0% of the methane generated for the rest of the year.

3.

Results

3.1. Comparison of results obtained from the different data collection schemes Comparisons of the normalized impacts (in PE and % contribution) for L1, L2, L3, and L3þ are presented graphically in Fig. 2. Table SI-13-1 in the supporting material also provides a numerical summary of the results. In general, impacts increase gradually as the data collection scheme becomes more comprehensive, though for the impact categories POF, ME, and HTnc, discrepancies between L1 and L3 were less than 5% and almost negligible. In particular, the difference in ME between L1 and L3 was less than 1% in median value, thus establishing that a legally bound data collection scheme can sufficiently

capture the impact of this category. No impacts were observed for PM, AC, or TE under L1, since the emission of substances contributing to these impact categories were not reported. Much larger discrepancies were observed for the impact categories GW (133%), HTc (6515%), and ET (5.74  1011%) between L1 and L3 median impact potential. L2 was not sufficient for the impact categories PM, AC, and TE, where less than 30% of impacts were captured, though for HTc, POF, and ET, considerable improvements were observed from L1 to L2. While the inclusion of the background system did not affect the outcome concerning ME, a marked increase in overall impacts was observed between L3 and L3þ for all the other impact categories. For instance, including the background system increased the GW of the Avedøre WWTP by 202%, and in some extreme cases, such as ET and PM, the normalized impact increased by 1350% and 821%, respectively. The uncertainty range for each impact category also widened, as more parameters were covered under the data collection scheme. One exception was GW and POF, for which the L1 and L2 results were accompanied by a larger uncertainty range than L3 and even L3þ, due to the fact that IPCC guidelines cite the uncertainty range for each parameter from 10 to 30% of the most probable value, which was used to calculate N2O and CH4 emissions for L1 and L2. Including external studies widens the uncertainty range further, though the median value of the outcome remains unchanged, since normal distribution is assumed when introducing uncertainty into an external process. The data reported by the Ecoinvent dataset were accompanied by a variation coefficient of 10.2e214.6%. In addition, based on the assessment of data quality for each external process, an additional 1.2e12.1% of uncertainty factors were introduced (SI-14).

3.2.

Results of the full LCA (L3þ)

Normalized impacts for the full LCA (L3þ) are presented in Fig. 3. The results can be put into perspective, as the daily influent flow corresponds to the wastewater generated by 726 inhabitants over a year. For example, the calculated impacts on GW were 5.4 PE/daily influent wastewater, so the impacts for one person’s annual wastewater generation are 7.4 mPE, i.e. 0.74% of the total impacts of an average person in Europe over a year. Similarly for ME, which has the largest normalized impact, the contribution of municipal wastewater treatment is about 7.1% of a person’s annual impact. Fig. 3 also shows the contributions of each unit process in wastewater and sludge treatment plants. The main contributors to environmental impacts vary depending on the impact categories. For instance, fugitive emissions from wastewater treatment processes contribute 19.5% of the total GW, while biogas leakage represents 43.8%, 23.4%, and 3.7% of the impact potential of TE, AC, and GW, respectively. As for ME, 95.7% of its potential impact was associated with the discharge of effluent into the ocean. Nonetheless, emissions embedded in the provision of electricity, and emission savings from electricity production through biogas utilization, occupy the major share of all impact categories. In particular, over 90% of biogas was sent to combined heat and power production, and in this assessment

w a t e r r e s e a r c h 5 6 ( 2 0 1 4 ) 2 9 2 e3 0 3

299

Fig. 2 e Normalized impacts obtained for data collection schemes L1, L2, L3 and L3D. The boxes present the 25% and 75% percentiles as well as the median while the error bars show the minimal and maximal values obtained (GW: global warming; AC: acidification; TE: terrestrial eutrophication; ME: marine eutrophication, marine; POF: photochemical ozone formation; ET: ecotoxicity; HTc: human toxicity, cancer; HTnc: human toxicity, non-cancer; PM: particulate matter).

it was the sole source of emission savings and enabled the lowering of impacts by 1.6e27.5% depending on the specific impact category.

3.3.

Perturbation and scenario analysis (L3þ)

A perturbation analysis was performed to test the sensitivity of the L3þ results against various parameters. Fig. 4 shows a tornado diagram of the sensitivity ratios for the 19 most sensitive global warming parameters (SR higher than 0.02 or a change in a parameter value of 10% leads to an increase in a global warming impact of 0.2%). Besides energy consumption, the most important parameters for on-site emissions were related to the composition of wastewater influent (C, N, and P load into the WWTPs and fossil carbon content) and CH4 and N2O emission rates. Similarly, even though the assessment included 65 parameters, relatively small numbers of parameters turned out to be highly sensitive to each impact category (SI-15).

Fig. 5 illustrates the results of the scenario analysis used to capture the impacts of operating conditions on the plant’s environmental performance. An increase in N2O emissions, due to sub-optimal BNR operation, led to an increase in the net global warming impact by 210%, while increased biogas leakage made a 156% impact on global warming. The impact of elevated N2O emissions is limited to GW in this study, but the loss of biogas from the anaerobic digester also contributed to the emission of ammonia into the atmosphere, which increased the AC and TE of the system four- and six-fold, respectively.

4.

Discussion

E-PRTR has in many ways the potential to provide accessible, standardized, and up-to-date emission data for conducting LCAs, as it already specifies the fate of pollutants (air, water, and soil) and provides mass loading data rather than concentrations, which are reported under current monitoring and

300

w a t e r r e s e a r c h 5 6 ( 2 0 1 4 ) 2 9 2 e3 0 3

Fig. 3 e Detailed normalized impacts (in PE and percent contribution) for data collection scheme L3D (GW: global warming; AC: acidification; TE: terrestrial eutrophication; ME: marine eutrophication, marine; POF: photochemical ozone formation; ET: ecotoxicity; HTc: human toxicity, cancer; HTnc: human toxicity, non-cancer; PM: particulate matter). Percent contribution is given in the absolute numbers.

permit requirements (Kaenzig et al., 2011). However, the present study demonstrated that the current E-PRTR scheme is not sufficient for creating a complete LCI of wastewater treatment plants. The most notable issue is that the current E-PRTR does not collect energy and treatment chemical consumption data. Emissions from external processes such as the provision of fuel, electricity, and chemicals used for treating wastewater

and sludge are equally or more important than on-site emissions. The trade-offs between the burden from treatment chemical production and improved performance of WWTP has been emphasized by the previous LCA studies on the wastewater treatment (i.e. Foley et al., 2010; Corominas et al., 2013a,b). Some national level mandates, such as Danish Green Account, require business entities to disclose major energy, water and

Fig. 4 e Sensitivity ratios (SRs) for the global warming impact of scenario L3D. In total, 65 parameters were assessed, but 44 parameters had a SR value less than 0.02 and are not shown.

w a t e r r e s e a r c h 5 6 ( 2 0 1 4 ) 2 9 2 e3 0 3

Fig. 5 e Scenario analysis for suboptimal operation of BNR process and biogas utilization.

raw material consumption figures (Thy, 2003). Since these operational data are collected routinely for billing purposes, the expansion of E-PRTR to include such data would be a feasible endeavor. Despite its importance, the LCI data on the treatment chemical specific to wastewater treatment process are limited. For instance, in this study the production of raw material was used as a surrogate for polymer coagulants. Even when an LCI data are available, it would not guarantee that the dataset is well correlated to the target systems,. As reported by Brogaard et al. (2014), it is evident that huge variation can be found between the LCIs for the same product. As it is major source of impact as well as uncertainties, it is vital to create the more complete and reliable LCI database for treatment chemicals used at the WWTP to improve overall quality of LCA. Beyond the WWTPs, the inclusion of such data would also provide information about the connections between each industrial sector. Once such information was laid on top of the Economic Input-Output table, the data in the E-PRTR could be used to improve the overall data quality of external processes by means of a hybrid LCA (Murray et al., 2008; Stokes and Horvath, 2010). The primary purpose of a WWTP is to control pollutant discharge into water bodies, and current monitoring and reporting requirements are sufficient in capturing those impact categories. However, current monitoring and reporting requirements still have limitations from two perspectives, namely emissions coverage and the method used to quantify these emissions. In the first case, emission coverage is limited by both substance coverage under the E-PRTR and the threshold for reporting requirements. For instance, NOx and SOx are listed under ‘reportable pollutants’, but the amount of emission in our case study was below the reporting threshold and hence its effect was not captured by L1. In some cases, entire emission pathways were ignored under the current monitoring and reporting scheme (e.g. ammonia and hydrogen sulfide emissions via biogas leakage). Lifting the reporting threshold for these pollutants, which are monitored under discharge requirements, and the re-evaluation of emission

301

pathways on a sector-by-sector basis could improve the completeness of data included in the E-PRTR. In contrast, differences in global warming potential impacts originate mainly from the method used to quantify fugitive emissions. In L1, fugitive emissions were assessed indirectly by providing emission factors per person or per kg of BOD removed by the plant, while L2 incorporated data on noncombusted methane from biogas utilization facilities (CHPs, boilers, and flaring devices), though the differences were infinitesimal. L3 included site-specific data such as N2O emissions from the biological nitrogen removal process, the physical leakage of methane, and fossil carbon content. The importance of collecting these site-specific emissions data was confirmed by the contribution and sensitivity analyses. Especially, it is fairly recently that an attempt to evaluate the presence of fossil carbon in wastewater and sewage sludge has been made (Griffith et al., 2009, Law et al., 2013), and the source and fate of fossil C in WWTP are not well identified. Fossil C load could be tied to the kinds of industrial entities present in the service area, as well as other pollutants present in wastewater influent. Including information on such pollutant ‘transfer’ could be another way of expanding the EPRTR database. Parameter uncertainty analysis showed that the uncertainty range increases in line with the number of input parameters, because the majority of parameters were regarded as independent, as sufficient observations were not available to establish correlations. An understanding of emission pathways and correlations between parameters would therefore help to reduce uncertainties associated with detailed LCA studies. Scenario analysis revealed that the environmental performance of the WWTP depended greatly on the plant’s operation. Wastewater treatment is a dynamic process and often experiences deviations in its operating conditions due to external factors (e.g. heavy rain events, changes in organic carbon and nutrient loads). E-PRTR asks the reporting entity to submit information on not only deliberate and routine pollutant releases, but also accidental and non-routine releases (accidental releases resulting from an uncontrolled development in the course of the operation, while non-routine activities include extraordinary activities that are carried out under a controlled operation, such as shutdown and start-up processes before and after maintenance activities). The inclusion of such emissions would help capture the full environmental impact and therefore ought to be encouraged. During this analysis some substances were found to be regulated as pollutants under E-PRTR or current monitoring requirements, but their impact was not captured by current ILCD-recommended Life Cycle impact assessment methods. These include, for example, HCl and HF air emissions resulting from incineration, and COD/BOD and Cl emissions into the aquatic environment. These could be areas for further methodological development in LCA.

5.

Conclusion

An LCA of a municipal WWTP was conducted to illustrate the effect of inventory data collection on the outcomes of environmental impact assessments. Compulsory disclosure

302

w a t e r r e s e a r c h 5 6 ( 2 0 1 4 ) 2 9 2 e3 0 3

requirements and the monitoring of emission discharges under regulations could provide a basis for measuring the direct impacts of plant operations, including even the incidental release of pollutants. However, with the current data gathering scheme, the use of this information as the basis of an LCA could result in the gross underestimation of environmental impacts associated with the WWTP. Several modifications are recommended in order to provide a full picture of environmental performance. Operational data such as fuel usage, electricity, and treatment chemicals should be included in order to include the background emissions. Better quantification of non-point fugitive emissions is especially important, while emission quantification methods should be standardized to eliminate bias caused by the quantification approach. Finally, in order to improve the completeness of the dataset, the reporting threshold should be eliminated for any parameters, which are already provided for under monitoring requirements.

Appendix A. Supplementary data Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.watres.2014.03.014

references

Bjorklund, A.E., 2002. Survey of approaches to improve reliability in LCA. Int. J. Life Cycle Assess. 7, 64e72. Brogaard, L.K., Damgaard, A., Jensen, M.B., Barlaz, M., Christensen, T.H., 2014. Evaluation of Life Cycle Inventory data for Recycling Systems. Resour. Conserv. Recyl. (in press). Clavreul, J., Guyonnet, D., Christensen, T.H., 2012. Quantifying uncertainty in LCA-modelling of waste management systems. Waste Manag. 32, 2482e2495. Clavreul, J., Baumeister, H., Christensen, T.H., Damgaard, A., 2014. EASETECH, an Environmental Assessment System for Environmental Technologies. Environ. Modell. Softw. (under review). Corominas, L., Foley, J., Guest, J.S., Hospido, A., Larsen, H.F., Morera, S., Shaw, A., 2013a. Life cycle assessment applied to wastewater treatment: state of the art. Water Res. 47 (15), 5480e5492. Corominas, L., Larsen, H.F., Flores-Alsina, X., Vanrolleghem, P.A., 2013b. Including life cycle assessment for decision-making in controlling wastewater nutrient removal systems. J. Environ. Manag. 128, 759e767. EEA, 2013. E-PRTR e The European Pollutant Release and Transfer Register. Welcome to E-PRTR. Available. http://prtr.ec.europa. eu (accessed 12.06.13.). Finnveden, G., 2000. On the limitations of life cycle assessment and environmental system analysis tools in general. Int. J. Life Cycle Assess. 5 (4), 229e238. Foley, J., de Haas, D., Hartley, K., Lant, P., 2010. Comprehensive life cycle inventories of alternative wastewater treatment systems. Water Res. 44 (5), 1654e1666. Forster, P., Ramaswamy, V., Artaxo, P., Berntsen, T., Betts, R., Fahey, D.W., Haywood, J., Lean, J., Lowe, D.C., Myhre, G., Nganga, J., Prinn, R., Raga, G., Schulz, M., Van Dorland, R., 2007. Changes in atmospheric constituents and in radiative forcing. In: Climate Change 2007: the Physical Science Basis. Contribution of Working Group I to the Fourth Assessment

Report of the Intergovernmental Panel on Climate Change. In: Solomon, S., Qin, D., Manning, M., Chen, Z., Marquis, M., Averyt, K.B., Tignor, M., Miller, H.L. (Eds.). Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA. Friedrich, E., Pillay, S., Buckley, C.A., 2007. The use of LCA in the water industry and the case for an environmental performance indicator. Water S. Afr. 33 (4), 443e452. Ganidi, N., Tyrrel, S., Cartmell, E., 2009. Anaerobic digestion foaming causes e a review. Bioresour. Technol. 100, 5546e5554. Griffith, D.R., Barnes, R.T., Raymond, P.A., 2009. Input of fossil carbon from wastewater treatment plants to U.S. rivers and oceans. Environ. Sci. Technol. 43 (15), 5647e5651. Hauschild, M.Z., Goedkoop, G., Guine´e, J., Heijungs, R., Huijbregts, M., Jolliet, O., Margni, M., de Schryver, A., Humbert, S., Laurent, A., Sala, S., Pant, R., 2013. Identifying best existing practice for characterization modelling in life cycle impact assessment. Int. J. Life Cycle Assess. 18, 683e697. Huijbregts, M.A.J., Norris, G., Bretz, R., Ciroth, A., Maurice, B., von Bahar, B., Weidema, B., de Beaufort, A.S.H., 2001. Framework for modelling data uncertainty in life cycle inventories. Int. J. Life Cycle Assess. 6, 127e132. Humbert, S., Marshall, J.D., Shaked, S., Spadaro, J.V., Nishioka, Y., Preiss, P., McKone, T.E., Horvath, A., Jolliet, O., 2011. Intake fraction for particulate matter: recommendations for life cycle impact assessment. Environ. Sci. Technol. 45, 4808e4816. International Reference Life Cycle Data System, 2011. International Reference Life Cycle Data System (ILCD) Handbook: General Guide for Life Cycle Assessment e Detailed Guidance. IPCC, 2006. IPCC Guidelines for National Greenhouse gas Inventories, Prepared by the National Greenhouse gas Inventories Programme. In: Eggleston, H.S., Buendia, L., Miwa, K., Nagara, T., Tanabe, K. (Eds.). IGES, Japan. ISO, 2006. ISO 14040: Environmental Management e Life Cycle Assessment e Principles and FrameworkInternational Standards Organisation. Kaenzig, J., Friot, D., Saade´, M., Margni, M., Jolliet, O., 2011. Using life cycle approaches to enhance the value of corporate environmental disclosures. Bus. Strategy Environ. 20, 38e54. Laurent, A., Hauschild, M.Z., Golsteijn, L., Fontes, M., Simas, J., Wood, R., 2013. Prosuite Deliverable 5.2: Normalisation factors for Environmental, Economic and Socio-economic Indicators. Copenhagen, Denmark, p. 54. Law, Y., Jacobsen, G.E., Smith, A.M., Yuan, Z., Lant, P., 2013. Fossil organic carbon in wastewater and its fate in treatment plants. Water Res. 47 (14), 5270e5281. Murray, A., Horvath, A., Nelson, K.L., 2008. Hybrid life-cycle environmental and cost inventory of sewage sludge treatment and end-use scenarios: a case study from China. Environ. Sci. Technol. 42, 3163e3169. Posch, M., Seppa¨la¨, J., Hettelingh, J.P., Johansson, M., Margni, M., Jolliet, O., 2008. The role of atmospheric dispersion models and ecosystem sensitivity in the determination of characterisation factors for acidifying and eutrophying emissions in LCIA. Int. J. Life Cycle Assess. 13 (6), 477e486. Ravishankara, A.R., Daniel, J.S., Portmann, R.W., 2009. Nitrous oxide (N2O): the dominant ozone-depleting substance emitted in the 21st century. Science 2 (326), 123e125. Reap, J., Roman, F., Duncan, S., Bras, B., 2008. A survey of unresolved problems in life cycle assessment, part 1: goal and scope and inventory analysis. Int. J. Life Cycle Assess. 13, 290e300. Rosenbaum, R.K., Bachmann, T.M., Gold, L.S., Huijbregts, M.A.J., Jolliet, O., Juraske, R., Koehler, A., Larsen, H.F., MacLeod, M., Margni, M., McKone, T.E., Payet, J., Schuhmacher, M., van de Meent, D., Hauschild, M.Z., 2008. USEtox e the UNEP SETAC toxicity model: recommended characterisation factors for

w a t e r r e s e a r c h 5 6 ( 2 0 1 4 ) 2 9 2 e3 0 3

human toxicity and freshwater ecotoxicity in life cycle impact assessment. Int. J. Life Cycle Assess. 13, 532e546. Seppa¨la¨, J., Posch, M., Johansson, M., Hettelingh, J.P., 2006. Country-dependent characterisation factors for acidification and terrestrial eutrophication based on accumulated exceedance as an impact category indicator. Int. J. Life Cycle Assess. 11 (6), 403e416. Stokes, J.R., Horvath, A., 2010. Supply-chain environmental effects of wastewater utilities. Environ. Res. Lett. 5. Struijs, J., Beusen, A., van Jaarsveld, H., Huijbregts, M.A.J., 2009. Chapter 6 aquatic eutrophication. In: Goedkoop, M., Heijungs, R., Huijbregts, M.A.J., De Schryver, A., Struijs, J., Van Zelm, R. (Eds.), ReCiPe 2008 a Life Cycle Impact Assessment Method Which Comprises Harmonisedcategory Indicators at the Midpoint and the Endpoint Level. Report I: Characterisation, vol. 6. January 2009. Thy, C., 2003. The Danish green accounts: experience and internal effects. In: Bennett, M., Rikhardsson, M., Schaltegger, S. (Eds.), Environmental Management Accounting: Purpose and Progress. Kluwer Academic Publishers, pp. 115e127. Van Zelm, R., Huijbregts, M.A.J., Den Hollander, H.A., Van Jaarsveld, H.A., Sauter, F.J., Struijs, J., VanWijnen, H.J., Van de Meent, D., 2008. European characterization factors for human

303

health damage of PM10 and ozone in life cycle impact assessment. Atmos. Environ. 42, 441e453. Weidema, B.P., Bauer, C., Hischier, R., Mutel, C., Nemecek, T., Reinhard, J., Vadenbo, C.O., Wernet, G., 2012. Overview and Methodology e Data Quality Guideline for the Ecoinvent Database Version 3. Availab at: http://www.ecoinvent.org/ fileadmin/documents/en/Data_Quality_Guidelines/01_ DataQualityGuideline_v3_Final.pdf (accessed 26.02.14.). Wursthorn, S., Poganietz, W.R., Schebek, L., 2011. Economicenvironmental monitoring indicators for European countries: a disaggregated sector-based approach for monitoring ecoefficiency. Ecol. Econ. 70, 487e496. Yoshida, H., Christensen, T.H., Scheutz, C., 2013a. Life cycle assessment of sewage sludge management: a review. Waste Manag. Res. 31 (11), 1083e1101. Yoshida, H., Christensen, T.H., Guildal, T., Scheutz, C., 2013b. A comprehensive substance flow analysis of a municipal wastewater and sludge treatment plant. Chemosphere (in press). Yoshida, H., Mønster, J., Scheutz, C., 2013c. A Plant-Integrated Measurement Of Nitrous Oxide And Methane Emission From Municipal Wastewater Treatment Plant (in preparation).