reti

, PBuil

2 July 2012Accepted 1 August 2012Available online 15 September 2012

impacts arising from the extraction, material, use and end-of-lifephases of all consumables, equipment and energy used for theremediation, and can be quantied by a life cycle assessment (LCA)(Lemming et al., 2010a; Morais and Delerue-Matos, 2010). An LCAalso provides a framework where the tradeoff between primary

scenarios. Lemming et al. (2010b) compared the primary andsecondary impacts of three remediation technologies (bioremedi-ation, thermal remediation, and ex-situ remediation) in a lowpermeability clayey till setting.

Here, we examine the in-situ mass destruction methods ofenhanced reductive dechlorination (ERD) and chemical oxidation(ISCO) as these technologies are not well represented in the LCAliterature. Only one study, Cadotte et al. (2007) has presented anassessment of the environmental impacts of the ISCO remediation

* Corresponding author. Tel.: 45 4525 1595; fax: 45 4593 2850.

Contents lists available at

Journal of Environm

ls

Journal of Environmental Management 112 (2012) 392e403E-mail address: gile@env.dtu.dk (G. Lemming).1. Introduction

Contaminated sites are usually remediated because they areassessed to pose a risk to the local environment (groundwater,surface water, terrestrial ecosystems) and to users of the site(indoor air risk, risk of direct contact with soil, etc.). The reductionof the local risks, also termed the primary impacts associated witha contaminated site, however, is done at the expense of increasingthe secondary impacts on health and the environment. Thesesecondary environmental impacts are the local, regional and global

impacts (the local contamination risk) and increased local, regionaland global impacts from remediation activities (secondary impacts)can be assessed. Environmental assessment methods such as lifecycle assessments are increasingly being required by environ-mental authorities when determining management strategies forcontaminated sites. Recent studies applying LCA for the assessmentof secondary and primary impacts of remediation include Sparreviket al. (2011a) who employed a LCA for the case of contaminatedsediment management, which showed that a do-nothing scenariocaused a much lower environmental impact than active cappingKeywords:RemediationContaminated sitesGroundwaterLife cycle assessmentChlorinated solventsNatural attenuationChemical oxidationEnhanced reductive dechlorinationDecision support0301-4797/$ e see front matter 2012 Elsevier Ltd.http://dx.doi.org/10.1016/j.jenvman.2012.08.002compared options are (i) long-term monitoring (ii) in-situ enhanced reductive dechlorination (ERD), (iii)in-situ chemical oxidation (ISCO) with permanganate and (iv) long-term monitoring combined withtreatment by activated carbon at the nearby waterworks. The life cycle assessment included evaluation ofboth primary and secondary environmental impacts. The primary impacts are the local human toxicimpacts due to contaminant leaching into groundwater that is used for drinking water, whereas thesecondary environmental impacts are related to remediation activities such as monitoring, drilling andconstruction of wells and use of remedial amendments. The primary impacts for the compared scenarioswere determined by a numerical risk assessment and remedial performance model, which predicted thecontaminant mass discharge over time at a point of compliance in the aquifer and at the waterworks. Thecombined assessment of risk reduction and life cycle impacts showed that all management options resultin higher environmental impacts than they remediate, in terms of person equivalents and assumingequal weighting of all impacts. The ERD and long-term monitoring were the scenarios with the lowestsecondary life cycle impacts and are therefore the preferred alternatives. However, if activated carbontreatment at the waterworks is required in the long-term monitoring scenario, then it becomes unfa-vorable because of large secondary impacts. ERD is favorable due to its low secondary impacts, but only ifleaching of vinyl chloride to the groundwater aquifer can be avoided. Remediation with ISCO caused thehighest secondary impacts and cannot be recommended for the site.

2012 Elsevier Ltd. All rights reserved.Article history:Received 17 April 2012Received in revised formA comparative life cycle assessment is presented for four different management options for a tri-chloroethene-contaminated site with a contaminant source zone located in a fractured clay till. TheIs there an environmental benet fromCombined assessments of the risk reduc

Gitte Lemming*, Julie C. Chambon, Philip J. BinningDepartment of Environmental Engineering, Technical University of Denmark, Miljoevej,

a r t i c l e i n f o a b s t r a c t

journal homepage: www.eAll rights reserved.mediation of a contaminated site?on and life cycle impact of remediation

oul L. Bjergding 113, DK-2800 Kgs. Lyngby, Denmark

SciVerse ScienceDirect

ental Management

evier .com/locate/ jenvman

technology and it focuses on the use of Fentons reagent as theoxidizing agent for groundwater contaminants. In this study weconsider the application of two other oxidants (permanganate andpersulfate) for source zone remediation of trichloroethene (TCE) ata Danish clay till site. On-site activated carbon treatment ofcontaminated groundwater in combination with remediation orcontainment by pump-and-treat has been considered in earlierstudies (Higgins and Olson, 2009; Bayer and Finkel, 2006). A newlong-term monitoring scenario combined with activated carbontreatment at the downstream waterworks is included asa management alternative to remediation of the site in this study.The long-term monitoring scenarios included in this study aremotivated by discussions in the scientic literature (Lemming et al.,2010b; Sparrevik et al., 2011a) and among eld practitionersregarding the benet of active remediation strategies at contami-nated sites, requiring long time frames (e.g. clay till sites) and/orsignicant energy and resources for site cleanup.

Thus, the purpose of this paper is to compare four alternativemanagement options for the TCE-contaminated site using anapproach combining groundwater risk assessment, remediationperformance modeling, and life cycle assessment (Fig. 1). Themanagement options are: (i) long-term monitoring with no activesource remediation, (ii) in-situ bioremediation by ERD, (iii) ISCOusing permanganate or persulfate as oxidizing agent, and (iv) long-term monitoring combined with activated carbon treatment at thewaterworks. For each management alternative a groundwater riskassessment and a remedial performance model is employed toanswer the following questions:

The model results are then employed with a life cycle assess-ment to determine the primary and secondary environmentalimpacts of each management alternative and thereby evaluate theoverall environmental benet of each alternative. Such informationcan be used as the basis for comparison and selection of manage-ment options for decision makers. Note that in this assessment weexclude the evaluation of tertiary impacts, i.e. the environmentalimpacts associated with the future use of the site (Lesage et al.,2007).

2. Materials and methods

2.1. Site description

The concept of combined risk and life cycle assessment isillustrated for a contaminated site located at Sortebrovej in Tom-merup, Denmark (Fig. 2). The site is contaminated by tri-chloroethene (TCE), which is mainly located in a fractured clay tillbetween 13 and 23 m below ground surface (mbgs) and in a thinupper sand aquifer located beneath the clay till. A conceptualgeological model and a transect of the source zone is shown inFig. 2. The contamination at Sortebrovej poses a risk to the drinkingwater abstraction wells at the Tommerup waterworks where wateris extracted from the regional groundwater aquifer. The watersupply wells are located 200 m northeast of the contaminantsource zone. The remediation targets an 11 m deep treatment zonewhich covers a horizontal area of 750 m2 in the clay till and1500 m2 in the sand. The contaminant mass of TCE was estimatedto 23.4 kg based on source zone calculations (Chambon et al., 2011).

wa

ent

valu

Hu

Se

G. Lemming et al. / Journal of Environmental Management 112 (2012) 392e403 393 What is the timeframe required to reach the remedial target? How does the groundwater concentration vary in time at thedownstream compliance point?

How does the contaminant concentration vary in time at thedownstream waterworks?

Conc

entra

tion

Time

Conc

entra

tion

Concentrations at the POC and

Remedial performance modeling and risk assessm

POCWW

Time

Mas

s/flu

x

Mass depletion in the source

Combined e

Normalised impact

Human toxicity (non cancer)Human toxicity (cancer)

Primary impacts (local toxic impacts) Fig. 1. Concept for combined evaluation of remedial performance, risk assessmeFull scale remediation by in-situ enhanced reductive dechlori-nation (ERD) was started at the site in 2006 with the addition ofa fermentable electron donor (emulsied soybean oil and lactate)and specic degrader organisms (Dehalococcoides) to the targettreatment zone in order to stimulate the complete microbial

Time

terworks

Inventory of secondary emissions

and resource use

Inventory of primary emissions

Impact assessment

Tim

efra

me

Accu

mu

late

d in

take

ation

Normalised impact

Global warming potentialAcidifaction

Aquatic eutrophicationEcotoxicity

man toxicity (non cancer)Human toxicity (cancer)

etc

Life cycle assessment

condary impacts (Local, regional and global impacts)nt and life cycle assessment. POC: Point of compliance. WW: Waterworks.

menG. Lemming et al. / Journal of Environ394degradation of TCE to ethene via the degradation productsdichloroethene (DCE) and vinyl chloride (VC). This study is partlyretrospective since an ERD remediation has already been imple-mented at the site. However, site remediation is on-going and thetimeframe for cleanup is not known (Manoli et al., 2012).

2.2. Management options

Four management options are considered for this site: long-term monitoring; remediation by enhanced reductive dechlorina-tion (ERD); remediation by in-situ chemical oxidation (ISCO), andlong-term monitoring combined with activated carbon treatmentat the waterworks.

In the long-term monitoring option, no active source zoneremediation takes place and TCE is only removed from the sourcezone by dissolution in inltrating water. No evidence of naturallyoccurring TCE transformation is observed at the site, therefore nodegradation was assumed in this scenario.

ERD is a bioremediation technology, involving the injection ofan electron donor (a fermentable substrate such as lactate, soybeanoil or molasses), and of specic degrader organisms (Dehalo-coccoides) to the subsurface. ERD stimulates the sequentialdechlorination from TCE to dichloroethene (DCE), vinyl chloride

Fig. 2. Location of the Sortebrovej site and water supply wells in Tommerup. The transconcentrations [mg/L] and the conceptual local geology and fracture setup used in the molocated 100 m downstream of the site.tal Management 112 (2012) 392e403(VC) and ethene, which is non-toxic and easily degraded (Scheutzet al., 2008). The generated degradation products (DCE and VC)are more mobile than TCE and will therefore be transported fasterout of the source zone than TCE (Chambon et al., 2010). Thesubstrate employed at the site was an emulsied oil substratecontaining soybean oil (60%), lactate (12%) and emulsiers.

ISCO enables a direct destruction of TCE to chloride, water andcarbon dioxide by the addition of a strong oxidant (in this casepotassium permanganate or sodium persulfate) and no formationof lower chlorinated ethenes is expected (Yan and Schwartz, 1999).A fraction of the amended oxidant is consumed by organicconstituents in the subsurface sediments. This consumption istermed natural oxidant demand (NOD), and can be very large inclayey tills (Hnning et al., 2007a).

The fourth management option employs a combination oflong-term monitoring of the site and activated carbon treatmentat the waterworks. Typical drinking water treatment in Denmarkonly involves aeration and sand ltration. The introduction ofactivated carbon treatment at the waterworks results in the use ofadditional power at the waterworks because of the need forincreased pumping and the UV-treatment that is required afterthe drinking water has passed through the activated carbon lter(Cowi, 2009).

ect runs along the groundwater ow direction and shows the initial aqueous TCEdel. POC: Point of compliance for assessing groundwater quality criteria. The point is

Due to the large depth to the source zone (>13 m), excavationfollowed by ex-situ treatment is not a viable option for this site.

2.3. LCA approach

The functional unit, which denes the service compared in thelife cycle assessment, is dened here to be the management of thetarget treatment zone which leads to a 99% removal of thecontaminant mass. The mass removal target ensures that theDanish groundwater quality criterion for TCE of 1 mg/L is met at thepoint of compliance 100 m downstream from the site.

The life cycle assessment considers the extraction of rawmaterials, the manufacturing, use and end-of-life phases forconsumables, equipment, energy etc. used in the differentmanagement options (see Fig. 3). In order to simplify the assess-ment, the materials used for pumps, mixing tanks and containersfor amendments were disregarded as they were assumed to resultin only negligible contributions to impacts due to relatively lowamounts of materials used and the high direct reuse rate. Theuptake of injection wells in the closure phase of the project wasdisregarded due to uncertainty of the fate of these wells. If uptakewas included, the impact would be equal for the ERD and ISCOsystem, whereas the long-term monitoring scenario would not beaffected. Furthermore, the CO2 emission caused by the degradationof TCE was excluded from the inventory due to its minor contri-bution. The Life Cycle Impact assessment method applied wasEDIP2003 (Hauschild and Potting, 2005) for non-toxic impacts andUSEtox (Rosenbaum et al., 2008) for the toxic impact categories.All results are normalized to person equivalents (PE) by dividingwith the average impact from a European citizen in 2004 (Laurentet al., 2011a, 2011b).

2.4. Life cycle inventory

The amount of remedial amendments required for ERD and ISCOwas estimated using data on site properties (see Table 1) and themodel-predicted time frames for cleanup (see Section 3.1). Thesubstrate demand for ERD was estimated to be the amount ofelectron donor required to react with the contaminant mass, thenative electron acceptors in the source zone (mainly Fe(III) andsulfate), and the dissolved electron donors transported to thesource zone with inltrating water during the treatment period(mainly sulfate). A safety factor of 2 was applied in the calculation.During the fermentation of the organic donor, all surplus donor wasexpected to be fermented to methane and discharged to theatmosphere.

The required amount of oxidant for ISCOwas estimated to be thesum of the natural oxidant demand (NOD) and the oxidant requiredtomineralize themass of TCE in the source zonewith a safety factorof 2. Previous laboratory batch experiments determined the

Use phase End-of-life

Decomposition of amendments

bacterial culture (ERD) and oxidant (ISCO)

EmissionsLocal toxic emissions Emissions

Monitoring

Table 1Site properties used for the inventory in Table 2.

Parameter (unit) Clay till Sand

Source zone area (m2) 750 1500Source zone depth (m) 10 1Porosity 0.3 0.3Bulk density (kg/L) 1.8 1.8Inltration rate (mm/year) 75 75Aqueous sulfate concentration in

source zone (mg/L)50 50

Solid phase iron (III) concentrationin source zone (mg/kg)

210 210

G. Lemming et al. / Journal of Environmental Management 112 (2012) 392e403 395High density poly-ethene (HDPE)

Production phase

Bentonite

Gravel

Steel

Remedial amendments

Emissions

Transport

Remedial amendments cover substrate and

Raw materials and energy

Pumps

Containers

Injection wells

System boundary

CO from TCE degradation

Closure and uptake of wellsFig. 3. System boundaries of tInstallation of injection wells

Groundwater extraction

Injection of oxidant/ substrate

Personnel transport

Treatment of soil from borings

Transport

Raw materials and energy

Raw materials and energy

Steel recycling (80%)

HDPE recycling (80%)he life cycle assessment.

The Ecoinvent database v.2 (Frischknecht et al., 2007) providedinventory data for the background processes (such as electricityproduction, steel production, transportation, potassium perman-ganate production and soybean oil production), and a full list of theapplied Ecoinvent processes are found in Table A4 in SI. TheEcoinvent data was supplemented with additional data from theliterature for the remediation-specic processes (lab analysis, ex-situ soil treatment, bioculture production and activated carbonproduction), see details in SI.

2.5. Modeling of timeframes and primary impacts for managementoptions

A numerical reactive transport model was used to assess thedevelopment over time of source zone concentrations and massdischarge to the groundwater aquifer for each management option

Table 2Inventory data for amendments added to the subsurface.

Parameter (unit) ERD lowrate

ERD highrate

ISCO

Substrate demand (kg) a 3700 5500 eGroundwater use for dilution

of substrate (m3) b65 110 e

Bioculture demand (kg) c 219 219 eIn-situ methane generation

potential (kg CH4/kg substrate) d0.18 0.18 e

KMnO4 demand (tonnes) e e 525In-situ CO2 generation

potential (kg CO2/kg KMnO4)e e 0.21

Groundwater use for dissolution (m3) e e e 52,500

a Assuming a H2 yield of 0.63 kg/kg substrate.b A 10% solution of the substrate is added to the clay till and a 1% solution to the

sand (Fyns Amt, 2006).c The bacterial culture, KB1, is only added once to the subsurface, i.e. during the

G. Lemming et al. / Journal of Environmental Management 112 (2012) 392e403396average NOD for the site to be 18 kg KMnO4 per tonne of soil in claytill samples (Hnning and Bjerg, 2003). For the sand aquifer a NODvalue of 1 kg KMnO4 per tonne soil was applied using typical valuesfor Danish aquifer materials (Hnning et al., 2007a). The carbondioxide generated from the reaction between the oxidant and thesoil was included in the inventory and assumed to be released tothe atmosphere. Table 2 lists the inventory data for amendmentsused in ERD and ISCO.

The long-term management scenario combined with treat-ment of the abstracted groundwater at the waterworks usesgranular activated carbon (GAC). An adsorption capacity of 2% (w/w) was assumed for the adsorption of TCE on to the GAC.Compared to traditional Danish drinking water treatment withaeration and sand ltration, adding activated carbon ltrationleads to an increased energy use of 10% or 0.025 kWh per m3

(Cowi, 2009).Table 3 provides an overview of the activities included in the

LCA of each scenario. More details regarding the energy use for thedrilling of wells, injection and pumping is found in Table A1 in

rst injection round, and is expected to sustain itself during the remediation period.Therefore the amount is the same for the two ERD scenarios. The actual amount ofKB1used at the Sortebrovej site is applied in the LCA.d With a soybean oil and lactate content of 60 and 12% respectively, the methane

generation potential becomes 0.18 kg/kg of substrate added.e The permanganate is diluted to a 10 g/L solution.Supporting Information (SI). The table also shows the material usedper meter well and the amount of activated carbon consumed.Monitoring activities for all scenarios include transportation(40 km return), pumping of groundwater and subsequent analysesin a laboratory. The transportation distances and site visitfrequencies for the scenarios are found in Table A2 in SI.

Table 3Overview of main activities included in each management scenario.

Long-term monitoring Enhanced reductive dechlorination (ERD) In-sit

Monitoring at anddownstream of the sitea

38 Injection wells (materials, installationand transport)Pumping and injectionb of substrateand biocultureProduction, transport and decompositionof substrate and biocultureMonitoring at and downstreamof the sitea,c

Personnel transportation duringinjection campaigns

38 Injand trPumpProduof oxiMoniPersocamp

a Annual monitoring frequency.b Injection campaigns every 5th year based on ndings in Manoli et al. (2012).c Biannual monitoring frequency in the two years following each injection campaign.(ERD, ISCO and long-termmonitoring). The source zone model wascoupled to a groundwater model to predict the downstreamcontaminant concentrations at a point of compliance located 100mfrom the site (Figs. 1 and 2). The source zone model calculates thediffusion dominated contaminant transport from the clay matrix tothe fracture and the advectiveedispersive transport of contami-nants in the vertical fractures. The reactive model includes thesequential reductive dechlorination of TCE (trichloroethene) to itsdaughter products DCE (dichloroethene), VC (vinyl chloride) andethane, and is described by Monod kinetics. The model approach issimilar to the model described in Lemming et al. (2010b) withaddition of a thin upper sand aquifer at this site (see cross section inFig. 2). Model input parameters are available in the SI, Table A5.

The remedial amendments (substrate and oxidants) are injectedby gravity feed into the injection wells in the treatment zone. Theinjected volume is expected to mainly ow into the high perme-ability sand stringers and fractures in the clay till, and from there todiffuse into the clay till and form a reaction zone around the sandstringers (see Fig. 2). The geological proles from the site suggestthat there is approximately 1 sand stringer per meter of depth. Theextent of the reactive zones in ERD were set to 5 cm based onndings from clay till cores at the site (Manoli et al., 2012). For ISCO,a reaction zone of 10 cm was assumed based on ndings fromanother Danish clay till site (Hnning et al., 2007b). In the sandaquifer, the remedial amendments are assumed to be homoge-neously distributed and reactions are assumed to occur in theentire volume.

For ERD two different sets of degradation rates were applied.These were derived based on analysis of two different laboratorybatch tests data sets. The rates are referred to as low and highrate respectively and can be seen in SI, Table A5.

u chemical oxidation (ISCO) Long-term monitoring and activatedcarbon treatment

ection wells (materials, installationansport)ing and injectionb of oxidantction, transport and decompositiondanttoring at and downstream of the sitea,c

nnel transportation during injectionaigns

Electricity use for pumping and UVtreatment at waterworksActivated carbon useMonitoring at and downstreamof the sitea

rate at the waterworks Q (m3): C(t) J(t)/Q. The mass dischargeis the sum of the mass discharge from the source zone to theupper and the regional aquifer, assuming that all contaminantmass ends up in the water abstracted at the water supply. Theaccumulated intake of TCE and degradation products for the1800 people supplied from the waterworks is estimated byintegrating over time the drinking water concentrations multi-plied by the ingested groundwater volume (1.4 L/day). USEtoxhealth effect factors are used to convert the intake to non-cancerand cancer effects.5

10

15

20

25Co

nta

min

ant m

ass

(kg)

a

Mass depletion in the

treatment zone

G. Lemming et al. / Journal of Environmental Management 112 (2012) 392e403 3970

3

4

5

6

ntra

tion

(g/

L)

b

Total concentrations at

100 m2.5.1. Primary impactsThe modeled time variation of mass discharge of TCE, DCE

and VC is used to estimate the concentration in the groundwaterabstracted at the Tommerup waterworks located 200 m down-stream of the source. The exposure concentration C (g/m3) asa function of time t in the drinking water is estimated bydividing the mass discharge J (g/year) by the annual pumping

Table 4Model-estimated timeframes for mass removal and compliance with groundwater qualitaquifer.

Timeframe (years)

99% mass removalin source

Compliance withcriteria at POC

Long-term monitoring 670 830ERD (low rate) 200 400ERD (high rate) 90 10a

ISCO 80 10a

a The monitoring period was set to be 10 years even though the results indicate that

0

1

2

Tota

l con

ce

0.0

0.1

0.2

0.3

0.4

0.5

0 200 400 600 800 1000

VC co

nce

ntra

tion

(g/

L)

Time (years)Monitoring ERD (low rate)ERD (high rate) ISCO

c

VC concentrations at

100 m

TCE criterion

VC criterion

Fig. 4. Model results showing the (a) contaminant mass in the treatment zone, (b)contaminant concentrations at the POC in the groundwater aquifer 100 m downstreamof the source (sum of TCE, DCE and VC), and (c) VC concentrations at 100 m. Note thedifferent scales on the y-axes.y criteria at the POC, and model-estimated leached amounts of contaminants to the

Leached amounts of contaminants (kg)

TCE DCE VC

23.2 0 00.05 0.3 1.10.2 0.02 0.0070.007 0 03. Results and discussion

3.1. Timeframe for 99% mass removal in source and downstreammonitoring

The model results (Fig. 4) show that if remediation is notconducted at the site, then it takes almost 700 years before thecontamination has leached out of the source zone. Remediationwith ISCO reduces this timeframe to 80 years, whereas ERDtakes between 90 years (ERD high rate) and 200 years (ERD lowrate).

The groundwater concentration at the point of compliance100 m downstream from the source was modeled for all scenariosand compared with the groundwater quality criteria (see Fig. 4). Inthe long-term monitoring scenario, TCE exceeds the criterion of1 mg/L for a period of approximately 800 years. Remediation withboth ERD and ISCO result in sufciently low concentrations ofchlorinated ethenes to comply with the quality criterion of 1 mg/Lfor the sum of chlorinated ethenes. However, for the low rate ERDscenario the vinyl chloride concentration exceeds the groundwaterquality criterion of 0.2 mg/L for a period of 300 years, and reacheslevels up to 0.5 mg/L.

In addition to the timeframes for removal of the source zonemass, the model results were used to estimate the necessarytimeframe for monitoring in the source zone (set equal to massremoval time) as well as downstream in the groundwater (set equalto the timeframe for complying with the groundwater qualitycriteria at the point of compliance 100 m downstreams) in eachscenario. Results are summarized in Table 4.

The modeled mass discharge (g/year) of contaminants to theaquifer over time is shown in Fig. A1 in the SI. Based on these resultsthe accumulated amounts of TCE, DCE and VC leaching togroundwater were calculated by integrating the mass dischargeresults over time and are presented in Table 4. The model resultsshow, that in the long-term monitoring scenario, 23 kg of TCE willleach to the groundwater aquifer, whereas remediation with ERDwill reduce this to 1.3 kg of chlorinated ethenes (ERD low rate) and0.2 kg of chlorinated ethenes (ERD high rate). In the low rate ERDscenario the emission of 1.4 kg of chlorinated ethenes is mainlycomposed of VC. Remediation with ISCO reduces the emission ofTCE to a negligible level.there is no exceedance of the quality criteria at the POC.

00.05

0.1

0.15

0.2

0.25

0.3

0.35

0 200 400 600 800 1000

Tota

l con

cent

ratio

n (g

/L)

Time (years)Monitoring ERD (low rate)ERD (high rate) ISCO

0.00

0.01

0.02

0.03

0.04

0.05

0 200 400 600 800 1000

Conc

entra

tion

(g/L)

Time (years)

Sum TCE DCE VC

b

ERD (low rate):

concentrations at waterworks

a

Total concentrations at

waterworks

), an

G. Lemming et al. / Journal of Environmental Management 112 (2012) 392e4033983.2. Exposure concentration at the waterworks

The resulting total concentrations (sum of TCE, DCE and VC) inthe waterworks for the long-term monitoring scenario are shownin Fig. 5, which also shows the individual component concentra-tions (TCE, DCE and VC) for the ERD low rate scenario. The resultsshow that the concentrations at the waterworks do not exceed theDanish drinking water quality criteria of 1 mg/L (sum of chlorinatedethenes) and 0.2 mg/L (vinyl chloride) for any of the scenarios. Thisis due to signicant dilution by uncontaminatedwater abstracted inthe well eld. If such dilution at the drinking water well can beaccepted despite the exceedance of the groundwater qualitycriterion, then the long-term monitoring scenario is a possiblemanagement option. However, if there are other point sources inthe catchment of the well eld then the total concentration mayexceed the criterion and an additional treatment at the waterworksmay be required.

3.3. Secondary life cycle impacts

The secondary life cycle impacts in terms of normalizedimpacts in PE (person equivalents) are compared for the vemanagement options in Fig. 6. Remediation by ISCO result in thegreatest environmental impacts in all impact categories and theyare on average a factor of 16 larger than those of the long-termmonitoring scenario. For ERD with a high rate, the magnitudes

Fig. 5. (a) Contaminant concentrations at the waterworks (sum of TCE, DCE and VCof the environmental impacts are of comparable size to those ofthe long-term monitoring scenario except for the ozone forma-tion impact which is signicantly higher for ERD. If the low

0

20

40

60

80

100

120

140

160

GlobalWarming

Ozoneformation

Acidification Terrestrialeutrophication

Aeutro

Nor

mal

ized

resu

lt (P

E)

Long-term monitoring ERD (high rate) ERD (low

Fig. 6. Life cycle assessment results for the secondary impacts in PE (reaction rate occurs in the ERD process, the remediation time-frame is doubled and the environmental impacts are increased byan average of 70%. Furthermore, the results reveal thatcombining the long-term monitoring scenario with activatedcarbon treatment at the waterworks will increase impacts of thisscenario by a factor of 8.

3.4. Primary life cycle impacts

Table 5 lists the primary toxic impacts due to human ingestionresulting from the leaching of contaminants to drinking water. Inthe long-term monitoring scenario no source remediation is con-ducted and so 23.2 kg of TCE will leach to the groundwater leadingto a human toxic (cancer) impact of 15 PE. If source zone remedi-ation is applied with ERD (high rate) or ISCO then this primaryimpact is reduced to very low levels (

mentas monitoring (transportation, sampling and lab analyses) is theonly activity included and is therefore responsible for all impactsexcept in the human toxicity (cancer) category, where theleaching of TCE to the aquifer accounts for 17% of the totalimpact.

For the high rate ERD (timeframe of 90 years), the injectionwells (installation and materials) are responsible for between 30and 65% of the different impacts. The remaining impacts stemmostly from the substrate, monitoring and the personneltransport during injection campaigns, whereas the biocultureand the pumping and injection only result in minor contribu-tions. In Fig. 7 the impacts from the substrate are dividedbetween the production and transport, and in-situ methaneformation. The in-situ methane formation from the fermentationof the substrate results in a very large global warming impact(31% of total) and ozone formation (43% of total), whereas theproduction of the soybean based substrate causes signicantaquatic eutrophication due to the application of fertilizer duringthe cultivation.

For the low rate ERD (200 years), the injection wells also makea signicant contribution to the total impact, but the contributionin percent is smaller than for the high rate scenario due to thelonger timeframe requiring additional substrate, monitoring, andpersonnel transport for substrate injection. More importantly, thevinyl chloride (VC) generated in the ERD process contributessignicantly to the human toxic impacts (42% of the non-carcinogenic effects and 15% of the carcinogenic effects) in thisERD scenario.

The potassium permanganate used in the ISCO scenario isresponsible for 88e92% of the impacts in all impact categories,except the aquatic eutrophication category where it contributesonly 62%. Injection wells contribute 2e11% of impacts, andgroundwater abstraction and injection 3e5% of all impacts,except for aquatic eutrophication, where it contributes to 24% ofthe impact. The large environmental impact of potassiumpermanganate is due to the large amount required (525 tonnes).The environmental impacts associated with the application ofpermanganate are mainly due to the energy required for itsproduction and a minor contribution is due to the road trans-portation of permanganate to Denmark. The in-situ generation ofcarbon dioxide due to oxidation of the solid organic carbon in theclay till sediment constitutes 13% of the global warming

Table 5Primary impacts in person equivalents (PE) of the 4 scenarios. The scenario long-term monitoring with activated carbon treatment is omitted, since it has noprimary impacts.

Long-termmonitoring

ERD(low rate)

ERD(high rate)

ISCO

Human toxicity (non-cancer) e a 7.5 0.1 e a

Human toxicity (cancer) 15 35 0.4 0.15Sum (PE) 15 42.5 0.5 0.15

a TCE has no non-cancer health effects according to the USEtox database.

G. Lemming et al. / Journal of Environpotential.In the long-term monitoring scenario combined with activated

carbon treatment at the waterworks, monitoring activities arerequired for 600 years in the source zone and 830 years in thedownstream groundwater (Table 4) and are responsible for only12e15% of the total impacts as seen in Fig. 7. The increased elec-tricity use at the waterworks due to additional pumping and UV-treatment is the main reason for the very high impacts seen inthis scenario. The activated carbon contributes to 8% of the ozoneformation impact, but contributes little to the remaining impactsdue to the relatively small amount (1150 kg) needed to remove theca. 23 kg of TCE.4. Discussion

4.1. Secondary impacts

The life cycle assessment of the secondary impacts of theassessed scenarios showed that long-term monitoring and ERD arethe management options that have the lowest life cycle impacts,even though these options have very extended timeframes.However, long-term monitoring without actual source removalresults in elevated drinking water concentrations at the water-works, where activated carbon treatment might be necessary toensure safe drinking water. In this case, the environmental impactsof the long-term monitoring scenario increase by a factor of 8 anddisfavors this option. In this case study, the environmental burdenfrom the added electricity use at the waterworks was allocatedbased on the mass discharge from the contaminated site and thewater volume this would contaminate up to the MCL. This resultedin a contaminated water volume of 23 million m3, which isequivalent to 170 years of drinking water production at thiswaterworks. If the drinking water was already contaminated abovethe MCL due to other contaminated sites, the burden allocated tothis site might be lower. It should also be noted that the assessmentassumes that the current technologies remain valid for the longduration of the scenarios. This is a conservative assumption astreatment technologies and electricity productionwill develop overtime. In the assessment of ERD it was assumed that all methanegenerated by the fermentation of the substrate was released to theatmosphere. This is a worst case assumption. The treatment zone atthe studied site is located 13 m below ground with an extendedunsaturated zone above it. It is therefore likely that a part of themethane will be reduced by methane oxidizing organisms in thetop soil and this can reduce the impact scores for global warmingand ozone formation signicantly (refer to Fig. 7).

The secondary impacts of ISCO aremuch larger (by up to a factorof 16) than those for ERD and long-termmonitoring due to the largeamount of oxidizing agent needed. Even though the timeframe forISCO is long due to the mass transfer limitations in the clay till, theimpacts due to site visits for monitoring and injection are minorwhen comparedwith that due to the oxidizing agent. This nding isconsistent with the LCA study by Cadotte et al. (2007) who studiedalternatives for groundwater remediation of diesel oil includingISCO with Fentons reagent. Here large amounts of Fentons reagent(14,250 tonnes) generated considerable secondary impacts due toits production. Sodium persulfate (Na2S2O8) is an alternativeoxidant for chemical oxidation. For this site a persulfate demand of300 tonnes was estimated based on an average clay till NOD of11 kg Na2S2O8/tonne determined in laboratory batch tests(Hnning and Bjerg, 2003) and a NOD for sand of 0.27 kg Na2S2O8/tone (Tsitonaki et al., 2010). Fig. 8 compares the environmentalimpacts of two ISCO scenarios applying permanganate (525 tonnes)and persulfate (300 tonnes) respectively. The persulfate scenarioshows a reduction in global warming, ozone formation, terrestrialeutrophication and respiratory impacts by 20e40% compared tothe permanganate scenario, mainly due to a lower energy use forthe production of the smaller amount of persulfate. However, at thesame time acidication increases by 80%, the aquatic eutrophica-tion by a factor of 5, and the ecotoxicity and human toxicity by10e30% (see Fig. 8) due to the high contribution to these impactcategories from persulfate production.

4.2. Primary impacts

The long-term monitoring scenario causes the largest release ofcontaminant mass to groundwater over its timeframe (23.2 kg of

al Management 112 (2012) 392e403 399TCE), whereas remediation by ERD reduces this to 1.1 kg VC and

Global Warming

Ozone formation (Human)Acidification

Terrestrial eutrophication

Aquatic eutrophication

Respiratory inorganics

Ecotoxicity freshwater

Human toxicity (non-cancer)Human toxicity (cancer)

Global Warming

Ozone formation (Human)Acidification

Terrestrial eutrophication

Aquatic eutrophication

Respiratory inorganics

Ecotoxicity freshwater

Human toxicity (non-cancer)Human toxicity (cancer)

Global Warming

Ozone formation (Human)Acidification

Terrestrial eutrophication

Aquatic eutrophication

Respiratory inorganics

Ecotoxicity freshwater

Human toxicity (non-cancer)Human toxicity (cancer)

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

Global Warming

Ozone formation (Human)Acidification

Terrestrial eutrophication

Aquatic eutrophication

Respiratory inorganics

Ecotoxicity freshwater

Human toxicity (non-cancer)Human toxicity (cancer)

Contribution to impacts

Injection wells Monitoring Pumping and injectionSubstrate: Production and transport Substrate: In situ CH4 generation BioculturePermanganate: Production and transport Permanganate: In situ CO2 generation Person transport (injection)Primary impact DCE Primary impact VC Activated CarbonExtra electricity use at waterworks

ER

D (

hig

h rate

90

y)

IS

CO

(80

y)

ER

D (L

ow

ra

te

2

00

y)

Lo

ng

-te

rm

mo

nito

rin

g a

nd

ac

tiv

ated

c

arb

on

Fig. 7. Contribution to total impacts (in percent of total PE) of the sub-parts of the remediation systems.

G. Lemming et al. / Journal of Environmental Management 112 (2012) 392e403400

050

100

150

200

250

300

350

Global warming

Ozone formation

Acidif ication Terrestrial eutrophication

Aeutro

Nor

mal

ized

resu

lt (P

E)

ISCO (permanganate)

Fig. 8. Life cycle comparison of ISCO using per

G. Lemming et al. / Journal of Environment0.3 kg of DCE (low rate scenario). The release of contaminants to theaquifer is negligible in the high rate ERD and ISCO scenarios (seeTable 4). However, when looking at the human toxicity of theseemissions (Table 5), the low rate ERD scenario actually causes anincreased release of toxicity compared with the long-term moni-toring because of the high toxicity of VC. Thus, the results indicatethat there is actually a risk of increasing the toxic release to theaquifer when remediating with ERD.

ISCO reduces the primary impacts from TCE efciently,however the addition of the large amount of oxidant may bedetrimental as it can contain heavy metal impurities and lead tomobilization of naturally occurring subsurface metals as well asprecipitation of manganese dioxide (Crimi and Siegrist, 2003). Thetoxic impact of this potential emission of heavy metal impuritiesand potential mobilization of naturally occurring metals wasdisregarded in the assessment of primary impacts as it is likelythat the metals will be demobilized before they reach the water-works. However, in comparison to the original aim of treatinga source zone of 23.4 kg of TCE, the addition of heavy metals dueto heavy metal impurities could be signicant as described in theSI, Table A6.

4.3. Is there an environmental benet from remediating the site?

Fig. 9 compares the total primary and secondary environmentalimpacts of each management scenario in terms of PE, calculated bythe summation of the normalized impacts, i.e. assuming equalweighting of each impact type. The long-term monitoring scenarioresults in a generation of 15 PE of primary impacts due to the on-0 200 400 600 800 1000 1200

Long-term monitoring and activated carbon

ISCO

ERD (low rate)

ERD (high rate)

Long-term monitoring

Total impacts (PE)Primary impacts (sum) Secondary impacts (sum)

2250 PE

Fig. 9. Comparison of total primary impacts (in PE) and secondary impacts (in PE) ofthe ve management options. Note that the normalized impacts are a summation ofthe individual impacts, i.e. assuming equal weighting of each impact type.site leaching of TCE to the groundwater and 140 PE of secondaryimpacts due to the monitoring related impacts of transportation,sampling and laboratory analyses.

ERD with a high rate is the only option that can remove theprimary impact of TCE leaching without increasing the secondaryimpacts. ISCO and long-termmonitoring with activated carbon alsoremove the primary impact, but does it at the expense of anincrease in the total secondary impacts by a factor of 8 (long-termmonitoring and activated carbon) and 16 (ISCO). Thus, only ERD canbe said to provide an overall environmental benet compared tothe long-term monitoring scenario. This however, depends on thebiodegradation rate; if the rate of the sequential dechlorination islow, there is a risk that vinyl chloride will leach to the aquifer (ERDlow rate) and that the primary impact will increase due to ERD. Inthis case, the overall benet of ERD is negative and long-termmonitoring is the preferred action.

The life cycle impact of the scenario based on long-termmonitoring combined with activated carbon treatment at thewaterworks depends on the allocation of the added environmentalburden of water treatment between the contaminated sites in thedrinking water catchment. In a catchment with many contami-nated sites, this option is likely to be more favorable than in thecurrent study, because the impact of water treatment is shared bymany contaminated sites. Compared to other solutions at thewaterworks (such as establishing a new well eld, use of surfacewater or desalination of marine water) activated carbon treatmentwas found to be the option with the lowest environmental impactin a Danish study (Cowi, 2009).

None of the compared management scenarios result in a netreduction in environmental impacts in terms of PE. Therefore, it

quatic phication

Respiratory inorganics

Ecotoxicity f reshwater

Human toxicity (non-cancer)

Human toxicity (cancer)

ISCO (persulfate)

1650

PE

1500

PE

manganate and persulfate as an oxidant.

al Management 112 (2012) 392e403 401could be argued that there is no environmental benet fromremediating the site. However, the evaluation of whether or notthere is an environmental benet is not straightforward. Intro-ducing a stakeholder panel in order to elicit weights between thedifferent impacts quantied in the LCA might result in a muchlarger weight on the local primary impacts than on the secondaryimpacts which are aggregated impacts over a larger geographicalscale (Sparrevik et al., 2011b). The high secondary impacts obtainedfor the ISCO technology and the long-term monitoring with acti-vated carbon treatment, however, indicate that these managementscenarios are not well balanced by the actual local impact removed.

Note, that in this study we only considered primary impacts dueto groundwater contamination and subsequent human exposurevia drinking water. At this site there is no human exposure viaindoor air or via direct soil contact. The benet of remediating thesite would increase if multiple exposure routes were present.Finally, the assessments in this paper are based on remediation ofa clay till site. Such remediation requires very long time frames and

mena large amount of reactants, which obviously disfavors someremedial actions. In high permeability settings with faster reme-diation, other conclusions may be reached.

4.4. Optimization of remediation timeframes

The long timeframes for remediation with ERD and ISCO at thissite are mainly a consequence of the clay till geology and the factthat the horizontal sand stringers where the reaction zones occurare spaced 1 m apart. This makes the diffusion from the clay matrixto the sand stringer a limiting step for the cleanup. A modelsensitivity study (Chambon et al., 2011) showed that increasing theinjection interval over depth by a factor of 4 (injection every 25 cm)can decrease the remediation time by a factor of at least 2, i.e. theremediation time of ERD and ISCO can be decreased to approxi-mately 40 years. Field tests with tracers suggests that a closelyspace injection in clay till may be achieved using direct pushtechnology (Christiansen et al., 2012). Electrokinetic-enhanceddelivery of microorganisms and substrate for ERD is alsocurrently under development (Mao et al., 2012) and may help toovercome the mass transfer limitations in clay. A shorter remedi-ation time reduces the impact of monitoring, personnel transport,and substrate use. Thus ERD can become more competitive withlong-term monitoring if the eld execution is optimized. ISCO willalso be more attractive if it is assessed over a shorter time period;however the secondary impacts will not be reduced signicantlysince they are driven by the large oxidant amount required.

5. Conclusions

A comparison of four different management scenarios (long-term monitoring, ERD, ISCO and long-term monitoring with acti-vated carbon treatment at the waterworks) for a contaminated sitewas performed using LCA combined with remedial performancemodeling. Two ERD scenarios were included with differing degra-dation rates for reductive dechlorination (ERD low rate and ERDhigh rate). If the source zone is not remediated and only long-termmonitoring employed, the model calculations show that thegroundwater quality criteria will be exceeded at a compliance point100 m downstream of the site for a very long time (approximately800 years). The drinking water concentration at the waterworkswill not be exceeded due to dilution with uncontaminatedgroundwater in the well eld at the waterworks.

With ERD, the TCE contamination risk is eliminated, but in thelow rate scenario, there is a risk that the production of vinyl chlo-ride causes the groundwater quality criteria to be exceeded for 300years. Due to the higher toxicity of vinyl chloride compared to TCE,there is therefore a risk of increasing the primary toxic impactscompared to the situation where all TCE leaches to the aquifer overtime. With the high rate ERD, ISCO and long-term monitoring withactivated carbon treatment scenarios, all primary toxic areremoved. This, however, is achieved at the expense of the genera-tion of much greater secondary impacts in the ISCO and the long-term monitoring scenario. ISCO generates especially high levels ofsecondary impacts due to the applied permanganate. Changing tosodium persulfate, an alternative oxidizing agent, does not improvethe environmental prole of ISCO.

A combined assessment of primary and secondary environ-mental impacts was made in this study. It was found that in termsof person equivalents (PE) the aggregated environmental impactgenerated on the global, regional and local scales was greater thanthe local environmental impact removed in all the assessedscenarios. Overall, long-termmonitoring and ERD were found to bethe preferable management options as they resulted in the lowest

G. Lemming et al. / Journal of Environ402secondary environmental impacts. However, in order to be a viablemanagement option, regulatory bodies must accept that the long-term monitoring option causes the groundwater quality criterionfor TCE to be exceeded for 800 years. If long-term monitoring iscombined with activated carbon treatment at the waterworks, thisoption will lead to larger secondary environmental impacts thanERD, but will still be preferable to the ISCO scenario. Finally ERD isonly favorable if it can be ensured that it does not cause anysignicant leaching of VC to the groundwater. If such leachingoccurs, the benets of ERD are lost because even small amounts ofVC result in very high primary toxic impacts.

It should be noted, that the conclusions in this paper are drawnon the basis of a case study with a clay till site geology resulting ina long remediation timeframe. In more permeable settings withfaster remediation, other conclusions may be reached. The ndingsfrom this study show that risk assessment and life cycle assessmentof management options for a contaminated site can be combined toprovide a more holistic assessment and an improved basis fordecisions on the selection of a management strategy for a contam-inated site.

Acknowledgments

The work was funded by the Region of Southern Denmark, theTechnical University of Denmark and REMTEC, Innovative REMe-diation and assessment TEChnologies, Danish Council for StrategicResearch. Mette Christophersen and Jan Petersen (Region ofSouthern Denmark), Claus Westergaard (Orbicon), Arkil A/S andGlibstrup A/S are acknowledged for providing valuable data on thesite and for the LCA.

Appendix A. Suppoting information

Supporting information related to this article can be found athttp://dx.doi.org/10.1016/j.jenvman.2012.08.002.

References

Bayer, P., Finkel, M., 2006. Life cycle assessment of active and passive groundwaterremediation technologies. Journal of Contaminant Hydrology 83, 171e199.

Cadotte, M., Deschnes, L., Samson, R., 2007. Selection of a remediation scenario fora diesel-contaminated site using LCA. The International Journal of Life CycleAssessment 12, 239e251.

Chambon, J.C., Broholm, M.M., Binning, P.J., Bjerg, P.L., 2010. Modeling multi-component transport and enhanced anaerobic dechlorination processes ina single fracture-clay matrix system. Journal of Contaminant Hydrology 112,77e90.

Chambon, J., Lemming, G., Manoli, G., Broholm, M.M., Christophersen, M.,Binning, P.J., Bjerg, P.L., 2011. Design requirements for ERD and ISCO: how closeand how fast to achieve an effective remediation? Levering af reaktive stoffer ilavpermeable aejringer e its a contact sport! ATV Jord og Grundvand 74,21e32 2011. Kgs. Lyngby.

Christiansen, C.M., Damgaard, I., Broholm, M.M., Kessler, T., Bjerg, P.L., 2012. Direct-push delivery of dye tracers for direct documentation of solute distribution inclay till. Journal of Environmental Engineering 138, 27e37.

Cowi, August 2009. LCA af vandforsyningsalternativer (LCA of alternatives for watersupply). In: Kbenhavns Energi og Hvidovre Forsyning (In Danish).

Crimi, M.L., Siegrist, R.L., 2003. Geochemical effects on metals following perman-ganate oxidation of DNAPLs. Ground Water 41, 458e469.

Frischknecht, R., Jungbluth, N., Althaus, H.-J., Doka, G., Dones, R., Heck, T.,Hellweg, S., Hischier, R., Nemecek, T., Rebitzer, G., Spielmann, M., Wernet, G.,2007. Overview and Methodology. In: Ecoinvent Report No. 1. Swiss Centre forLife Cycle Inventories, , Dbendorf.

Fyns Amt, 2006. Afvrgeforanstaltninger. Dokumentationsrapport. Lok. nr. 485-20.Tidligere maskinfabrik Sortebrovej 26, 5690 Tommerup. Remediation docu-mentation report by Orbicon A/S for Fyns Amt. December 2006 (In Danish).

Hauschild, M.Z., Potting, J., 2005. Spatial Differentiation in Life Cycle ImpactAssessment e the EDIP-2003 Methodology. In: Environmental News No. 802005. Environmental Protection Agency, Danish Ministry of the Environment.

Higgins, M.R., Olson, T.M., 2009. Life-cycle case study comparison of permeablereactive barrier versus pump-and-treat remediation. Environmental Science &Technology 43, 9432e9438.

tal Management 112 (2012) 392e403Hnning, J., Bjerg, P.L., 2003. Kemisk Oxidation. Laboratorieforsg e Sortebrovej.Milj og Ressourcer DTU. Technical University of Denmark (In Danish).

Hnning, J., Broholm, M.M., Bjerg, P.L., 2007a. Quantication of potassiumpermanganate consumption and PCE oxidation in subsurface materials. Journalof Contaminant Hydrology 90 (3e4), 221e239.

Hnning, J., Broholm, M.M., Bjerg, P.L., 2007b. Role of diffusion in chemical oxida-tion of PCE in a dual permeability system. Environmental Science & Technology41, 8426e8432.

Laurent, A., Lautier, A., Rosenbaum, R.K., Olsen, S.I., Hauschild, M.Z., 2011a.Normalization references for Europe and North America for application withUSEtox (TM) characterization factors. International Journal of Life Cycle Assess-ment 16, 728e738.

Laurent, A., Olsen, S.I., Hauschild, M.Z., 2011b. Normalization in EDIP97 andEDIP2003: updated European inventory for 2004 and guidance towardsa consistent use in practice. The International Journal of Life Cycle Assessment16, 401e409.

Lemming, G., Hauschild, M.Z., Bjerg, P.L., 2010a. Life cycle assessment of soil andgroundwater remediation technologies: literature review. International Journalof Life Cycle Assessment 15, 115e127.

Lemming, G., Hauschild, M.Z., Chambon, J., Binning, P.J., Bulle, C., Margni, M.,Bjerg, P.L., 2010b. Environmental impacts of remediation of a trichloroethene-contaminated site: life cycle assessment of remediation alternatives. Environ-mental Science & Technology 44, 9163e9169.

Lesage, P., Ekvall, T., Deschenes, L., Samson, R., 2007. Environmental assessment ofBrowneld rehabilitation using two different life cycle inventory models e part1: methodological approach. International Journal of Life Cycle Assessment 12,391e398.

Manoli, G., Chambon, J., Broholm, M.M., Scheutz, C., Binning, P.J., Bjerg, P.L., 2012. Aremediation performance model for enhanced metabolic reductive dechlori-nation of Chloroethenes in fractured clay till. Journal of Contaminant Hydrology131 (1e4), 64e78.

Mao, X., Wang, J., Ciblak, A., Cox, E.E., Riis, C., Terkelsen, M., Gent, D.B.,Alshawabkeh, A.N., 2012. Electrokinetic-enhanced bioaugmentation for

remediation of chlorinated solvents contaminated clay. Journal of HazardousMaterials 213-214, 311e317.

Morais, S.A., Delerue-Matos, C., 2010. A perspective on LCA application in siteremediation services: critical review of challenges. Journal of HazardousMaterials 175, 12e22.

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-the UNEP-SETAC toxicity model: recommended characterisation factors for humantoxicity and freshwater ecotoxicity in life cycle impact assessment. Interna-tional Journal of Life Cycle Assessment 13, 532e546.

Scheutz, C., Durant, N., Dennis, P., Hansen, M.H., Jrgensen, T., Jakobsen, R., Cox, E.,Bjerg, P.L., 2008. Concurrent ethene generation and growth of dehalococcoidescontaining vinyl chloride reductive dehalogenase genes during an enhancedreductive dechlorination eld demonstration. Environmental Science & Tech-nology 42, 9302e9309.

Sparrevik, M., Saloranta, T., Cornelissen, G., Eek, E., Fet, A.M., Breedveld, G.D.,Linkov, I., 2011a. Use of life cycle assessments to evaluate the environmentalfootprint of contaminated sediment remediation. Environmental Science &Technology 45, 4235e4241.

Sparrevik, M., Barton, D.N., Oen, A.M.P., Sehkar, N.U., Linkov, I., 2011b. Use of mul-ticriteria involvement processes to enhance transparency and stakeholderparticipation at Bergen Harbor, Norway. Integrated Environmental Assessmentand Management 7, 414e425.

Tsitonaki, A., Mosbk, H., Bjerg, P.L., Petri, B., Siegrist, R.L., Crimi, M., 2010. In situchemical oxidation of contaminated soil and groundwater using persulfate:a review. Critical Reviews in Environmental Science and Technology 40,55e91.

Yan, Y.E., Schwartz, F.W., 1999. Oxidative degradation and kinetics of chlorinatedethylenes by potassium permanganate. Journal of Contaminant Hydrology 37,343e365.

G. Lemming et al. / Journal of Environmental Management 112 (2012) 392e403 403