use of life-cycle analysis to msw - delware

7
Use of Life-Cycle Analysis To Support Solid Waste Management Planning for Delaware P. OZGE KAPLAN,* S. RANJI RANJITHAN, AND MORTON A. BARLAZ Department of Civil, Construction, and Environmental Engineering, North Carolina State University, Raleigh, North Carolina 27695 Received July 3, 2008. Revised manuscript received November 29, 2008. Accepted December 2, 2008. Mathematical models of integrated solid waste management (SWM) are useful planning tools given the complexity of the solid waste system and the interactions among the numerous components that constitute the system. An optimization model was used in this study to identify and evaluate alternative plans for integrated SWM for the State of Delaware in consideration of cost and environmental performance, including greenhouse gas (GHG) emissions. The three counties in Delaware were modeled individually to identify efficient SWM plans in consideration of constraints on cost, landfill diversion requirements, GHG emissions, and the availability of alternate treatment processes (e.g., recycling, composting, and combustion). The results show that implementing a landfill diversion strategy (e.g., curbside recycling) for only a portion of the population is most cost-effective for meeting a county- specific landfill diversion target. Implementation of waste-to- energy offers the most cost-effective opportunity for GHG emissions reductions. Introduction The cost and environmental implications (e.g., energy consumption, greenhouse gas (GHG) emissions) of solid waste management (SWM) are important societal issues. SWM costs are borne by the public, either through use fees or taxes. SWM also has environmental impacts resulting from waste collection, separation, treatment processes such as composting and combustion, and landfill disposal (1). The beneficial use of waste, for either energy recovery or material recovery, can result in both revenue and avoided emissions (2, 3). An integrated analysis must be conducted to assess the net cost and net environmental effects of (1) an SWM program constituted of a set of municipal solid waste (MSW) process choices that interactively affect system-wide waste flow and (2) SWM policies that constrain the system (e.g., banning items such as yard waste from landfills and banning waste processing options such as waste combustion). Thus, policymakers face the challenge of developing and imple- menting integrated SWM programs that represent an ap- propriate use of public funds while considering emissions and energy consumption. Mathematical models of integrated SWM can serve as planning tools given the complexity of the system, the interactions among the numerous components that con- stitute the system, and the number of potential SWM alternatives. While numerous models have been described (4-9), and several case studies have been conducted in Europe (10-14), the number of case studies applying SWM planning models in the United States is limited (15). The objective of this study was to evaluate alternative plans for integrated SWM in the State of Delaware considering cost and environmental performance, particularly GHG emissions. This study was conducted to assist the Delaware Solid Waste Authority (DSWA) conduct its periodic (every 10 years) evaluation of the statewide SWM program and development of a long-term plan. The next section sum- marizes the integrated solid waste management decision support tool (ISWM-DST), a life-cycle model that was utilized to analyze potential SWM programs considering combina- tions of curbside recycling, yard waste composting, and combustion with energy recovery, i.e., waste-to-energy (WTE), to divert waste from landfills (16-18). The subsequent section describes the modeling approach tailored for urban and rural counties in Delaware and input data development. Analyses are then presented in which system cost and environmental performance are explored at increasing diversion constraints. Initially cost and then GHG emissions are used as the model objective. Finally, we describe how the model results can be applied to advance SWM planning for Delaware. Model Description The ISWM-DST is a steady-state deterministic optimization model that represents the flow of individual MSW compo- nents from generation through collection, separation for recycling at materials recovery facilities (MRFs), treatment (e.g., yard waste composting and WTE), and landfill disposal as described previously (16-18). A summary is provided here, and Table S1 of the Supporting Information (SI) gives additional resources. The ISWM-DST includes (1) process models for estimating cost (including revenue from recy- clables and energy recovery), energy consumption, and life- cycle emissions associated with each SWM unit operation, (2) a mathematical programming-based integrated system model that embeds the waste mass flow equations, and (3) a linear programming (LP) model solver (CPLEX) (Figure S1, SI). The process models compute a set of cost and life-cycle emission coefficients per mass of waste item handled in a process using a combination of default and site-specific data. There are process models for waste collection, separation, treatment, and disposal. In addition, there are process models for electrical energy production and the conversion of recyclables into new products (i.e., remanufacturing). An offset analysis is used to calculate the environmental benefits or added burdens from the conversion of recycled materials to new products and from the generation of electricity from landfill gas and WTE (19). All unit processes are integrated, and the mass balance is represented by a series of waste flow equations that may be solved for the minimum value of cost, net energy consumption, or emissions of selected pollutants. The ISWM- DST tracks 30 air- and water-borne pollutants and optimizes on seven air pollutants (CO, CO 2 , CH 4 , NO x , SO x , PM, and greenhouse gas equivalents [GHEs]), cost, and energy consumption. Recently, the capability to consider the effect of uncertain input parameters on model outputs was * Corresponding author present address: Research Fellow, National Risk Management Research Laboratory, U.S. Environmental Protec- tion Agency, Mail Drop E305-02, Research Triangle Park, NC 27711; phone: (919) 541-5069; fax: (919) 541-7885; e-mail: Kaplan.Ozge@ epa.gov. Environ. Sci. Technol. 2009, 43, 1264–1270 1264 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 5, 2009 10.1021/es8018447 CCC: $40.75 2009 American Chemical Society Published on Web 01/29/2009

Upload: chan-kian

Post on 12-Jan-2016

2 views

Category:

Documents


0 download

DESCRIPTION

LCA

TRANSCRIPT

Use of Life-Cycle Analysis ToSupport Solid Waste ManagementPlanning for DelawareP . O Z G E K A P L A N , * S . R A N J I R A N J I T H A N ,A N D M O R T O N A . B A R L A Z

Department of Civil, Construction, and EnvironmentalEngineering, North Carolina State University,Raleigh, North Carolina 27695

Received July 3, 2008. Revised manuscript receivedNovember 29, 2008. Accepted December 2, 2008.

Mathematical models of integrated solid waste management(SWM) are useful planning tools given the complexity of the solidwaste system and the interactions among the numerouscomponents that constitute the system. An optimization modelwas used in this study to identify and evaluate alternativeplans for integrated SWM for the State of Delaware inconsideration of cost and environmental performance, includinggreenhouse gas (GHG) emissions. The three counties inDelaware were modeled individually to identify efficient SWMplans in consideration of constraints on cost, landfill diversionrequirements, GHG emissions, and the availability of alternatetreatment processes (e.g., recycling, composting, andcombustion). The results show that implementing a landfilldiversion strategy (e.g., curbside recycling) for only a portionof the population is most cost-effective for meeting a county-specific landfill diversion target. Implementation of waste-to-energyoffersthemostcost-effectiveopportunityforGHGemissionsreductions.

Introduction

The cost and environmental implications (e.g., energyconsumption, greenhouse gas (GHG) emissions) of solidwaste management (SWM) are important societal issues.SWM costs are borne by the public, either through use feesor taxes. SWM also has environmental impacts resulting fromwaste collection, separation, treatment processes such ascomposting and combustion, and landfill disposal (1). Thebeneficial use of waste, for either energy recovery or materialrecovery, can result in both revenue and avoided emissions(2, 3). An integrated analysis must be conducted to assessthe net cost and net environmental effects of (1) an SWMprogram constituted of a set of municipal solid waste (MSW)process choices that interactively affect system-wide wasteflow and (2) SWM policies that constrain the system (e.g.,banning items such as yard waste from landfills and banningwaste processing options such as waste combustion). Thus,policymakers face the challenge of developing and imple-menting integrated SWM programs that represent an ap-propriate use of public funds while considering emissionsand energy consumption.

Mathematical models of integrated SWM can serve asplanning tools given the complexity of the system, theinteractions among the numerous components that con-stitute the system, and the number of potential SWMalternatives. While numerous models have been described(4-9), and several case studies have been conducted inEurope (10-14), the number of case studies applying SWMplanning models in the United States is limited (15).

The objective of this study was to evaluate alternativeplans for integrated SWM in the State of Delaware consideringcost and environmental performance, particularly GHGemissions. This study was conducted to assist the DelawareSolid Waste Authority (DSWA) conduct its periodic (every 10years) evaluation of the statewide SWM program anddevelopment of a long-term plan. The next section sum-marizes the integrated solid waste management decisionsupport tool (ISWM-DST), a life-cycle model that was utilizedto analyze potential SWM programs considering combina-tions of curbside recycling, yard waste composting, andcombustion with energy recovery, i.e., waste-to-energy(WTE), to divert waste from landfills (16-18). The subsequentsection describes the modeling approach tailored for urbanand rural counties in Delaware and input data development.Analyses are then presented in which system cost andenvironmental performance are explored at increasingdiversion constraints. Initially cost and then GHG emissionsare used as the model objective. Finally, we describe how themodel results can be applied to advance SWM planning forDelaware.

Model DescriptionThe ISWM-DST is a steady-state deterministic optimizationmodel that represents the flow of individual MSW compo-nents from generation through collection, separation forrecycling at materials recovery facilities (MRFs), treatment(e.g., yard waste composting and WTE), and landfill disposalas described previously (16-18). A summary is provided here,and Table S1 of the Supporting Information (SI) givesadditional resources. The ISWM-DST includes (1) processmodels for estimating cost (including revenue from recy-clables and energy recovery), energy consumption, and life-cycle emissions associated with each SWM unit operation,(2) a mathematical programming-based integrated systemmodel that embeds the waste mass flow equations, and (3)a linear programming (LP) model solver (CPLEX) (Figure S1,SI). The process models compute a set of cost and life-cycleemission coefficients per mass of waste item handled in aprocess using a combination of default and site-specific data.There are process models for waste collection, separation,treatment, and disposal. In addition, there are process modelsfor electrical energy production and the conversion ofrecyclables into new products (i.e., remanufacturing). Anoffset analysis is used to calculate the environmental benefitsor added burdens from the conversion of recycled materialsto new products and from the generation of electricity fromlandfill gas and WTE (19).

All unit processes are integrated, and the mass balanceis represented by a series of waste flow equations that maybe solved for the minimum value of cost, net energyconsumption, or emissions of selected pollutants. The ISWM-DST tracks 30 air- and water-borne pollutants and optimizeson seven air pollutants (CO, CO2, CH4, NOx, SOx, PM, andgreenhouse gas equivalents [GHEs]), cost, and energyconsumption. Recently, the capability to consider the effectof uncertain input parameters on model outputs was

* Corresponding author present address: Research Fellow, NationalRisk Management Research Laboratory, U.S. Environmental Protec-tion Agency, Mail Drop E305-02, Research Triangle Park, NC 27711;phone: (919) 541-5069; fax: (919) 541-7885; e-mail: [email protected].

Environ. Sci. Technol. 2009, 43, 1264–1270

1264 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 5, 2009 10.1021/es8018447 CCC: $40.75 2009 American Chemical SocietyPublished on Web 01/29/2009

incorporated (20), which enables a post-optimization un-certainty analysis to be conducted.

The functional unit for the system is the management of1 Mg of MSW set out for collection. MSW includes wastegenerated in single-family residential, multifamily residential,and commercial sectors as defined by the U.S. EPA (21).Unique waste generation data may be provided for each oftwo distinct areas in the residential and multifamily sectorsand ten distinct commercial generation points.

Modeling Approach and Data Development for the Stateof DelawareScenario Definition. Model scenarios were constructed torepresent current practice and to explore the implicationsof increased landfill diversion by recycling, composting, andWTE on total system cost and emissions (Table 1). In the firstset of scenarios, the goal was to identify cost-effective SWMplans for different levels of waste diversion, which weremodeled as incrementally increasing diversion requirements.As described in Table 1, cases 1-5 were analyzed withdifferent combinations of unit processes enabled. In thesecond set of scenarios, the goal was to determine SWMplans to minimize GHEs in megagram carbon equivalents(eq 1) for different levels of cost.

GHE (Mg)) [mass of CO2-fossil (Mg)] × 12/44+21 × [mass of CH4(Mg)] × 12/16 (1)

where 21 is used to convert the mass of CH4 to CO2 equivalentsand 12/44 and 12/16 convert CO2 and CH4 to an equivalentmass of C, respectively.

SWM plans were identified at incrementally increasingcost targets starting with the cost of current practice (case2). All scenarios were evaluated separately for each ofDelaware’s three counties. A follow-up paper will describehow these county-specific strategies were combined toconstruct and analyze statewide integrated strategies.

Data Development. Delaware is comprised of threecounties (Figure S2, SI). New Castle County (NCC) is themost densely populated with 64% of the state’s 783600 people.Kent and Sussex Counties are largely rural. The wastegeneration rate and composition data were adopted fromstate waste characterization reports (22, 23). Per capita wastegeneration was estimated to be 1.04 kg person-1 day-1,excluding durable items. This rate was assumed to be constantstatewide and independent of whether a resident lived in theresidential or multifamily sector. Totals of 21%, 10%, and

10% of the residential population reside in multifamilydwellings in NC, Kent, and Sussex Counties, respectively.

The number of collection locations in the multifamilysector was calculated by estimating that one dumpster willserve 40 multifamily housing units, resulting in 1028 collectionlocations in NCC. The per-location commercial MSW gen-eration rate was computed from the ratio of commercialMSW generation to the number of commercial locations.Commercial waste generation data and the number ofcommercial locations were obtained from public records(23, 24). Waste generation and composition data are sum-marized in Tables S2 and S3 (SI).

Waste generation in NCC was modeled using two resi-dential sectors, one multifamily sector and one commercialsector. Two residential sectors were required to representdifferences in average distances from collection routes tothe facilities (i.e., transfer station, landfill) as ∼10% of thecounty’s waste flows through a transfer station. Residentialsector 2 in NCC represents the southern region that is servedby a transfer station. Kent and Sussex Counties wererepresented by one residential, one multifamily, and onecommercial sector.

Approximately 20% of MSW generated in Delaware iscurrently recovered via the state’s drop-off program plus therecovery of source-separated recyclables from the com-mercial sector (22, 23). There was essentially no curbsidecollection of recyclables or WTE at the time of this study. Forevaluation of future SWM scenarios in which curbsidecollection of recyclables and composting were enabled, itwas necessary to estimate capture rates for these programs.It was assumed that if a residential curbside recycling programwere to be implemented, then recovery rates would be higherthan the national average rates, which represent the averageof all states, including some that do not have a recyclingprogram. Material-specific recovery rates in Delaware wereset 20% greater than the national average rates (21). Inaddition, the rate of participation in potential future resi-dential curbside collection programs was assumed to be 80%.Additional input data are presented in Tables 2 and S4 andS5 (SI).

Uncertainty Analysis. For a given countywide strategy,uncertainty in the cost and life-cycle emissions was estimatedusing uncertainty propagation procedures (20). Probabilitydensity functions (PDFs) for selected model inputs were onthe basis of experience and expert judgment. Cumulativedensity functions (CDFs) for cost and GHE were used to assessthe robustness of the countywide SWM strategies. Finally, acorrelation analysis was conducted to understand the relativesignificance of uncertainty in each input parameter. Theuncertain inputs and their assumed PDFs are presented inTable S6 (SI).

Surrogate Environmental Indicator Parameters. A rep-resentative indicator parameter for environmental perfor-mance was identified to (1) simplify the presentation andanalysis of the results and (2) be consistent with the abilityof ISWM-DST to minimize only one pollutant at a time.Analyses were conducted in which GHEs were minimizedfor different cost constraints, and the results show that GHEis a reasonable surrogate for emissions of multiple pollutants(Figure 1). Correlation coefficients (r2) were above 0.9 whenconsidering the trend of GHE with that of energy consump-tion and all air pollutants except CO, which did not correlatewell with any other pollutant (Table S7, SI).

ResultsCost-Effective SWM Strategies. The results of model analysesin which different combinations of unit operations wereenabled are presented in this section. (As the results for Kentand Sussex Counties were similar, detailed results for KentCounty are presented in the SI.) When all waste is buried in

TABLE 1. Description of Model Scenarios

case description

Model Objective: Least Cost(1) landfill only all waste buried in a landfill(2) current practice recyclables recovered through

voluntary drop-off only(3) recycling waste diversion by both curbside

recyclables collection and wastesorting at a mixed waste MRFare enabled

(4) recycling +composting

as in case 3 plus the separatecurbside collectionof yard waste isenabled

(5) recycling +composting + WTE

as in case 4 plus WTE is enabled

Model Objective: Least Greenhouse Gas Equivalents(6) recycling +

composting + WTEas in case 5

(7) recycling +composting

as in case 4

VOL. 43, NO. 5, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 1265

a landfill (case 1), the resulting annual cost is calculated tobe $37.2 million, $19.8 million, and $34.8 million for NC,Kent, and Sussex Counties, respectively. The correspondingemissions are presented in Figure 2 and in Tables S8-S10(SI). Interestingly, the avoided emissions associated with theconversion of landfill gas (LFG) to energy resulted in netnegative emissions for PM, SOx, and CO2-fossil, with thelargest benefit occurring in NCC where travel distances arethe shortest. A negative emission means that the avoidedemissions exceed the emissions attributable to waste col-lection and processing.

For current practice (case 2), which results in 18-20%diversion using recyclables drop-off and collection of com-mercial recyclables, net system costs including revenues fromthe sale of recyclables for NC, Kent, and Sussex Countieswere $39.7 million, $21.4 million, and $36.9 million, respec-tively. The corresponding emissions are presented in Figure2 and in Tables S11-S13 (SI).

The ISWM-DST was next used to identify cost-effectivewaste management strategies in which landfill diversion wasconstrained to match current practice as well as higher levels.Solid waste operations that were enabled in addition to thosecurrently used include (1) a mixed waste MRF in whichrecyclables are recovered from MSW using a combination ofhand sorting and mechanical separation and (2) curbside

collection of recyclables that are processed in an MRF (case3). Within curbside collection, alternatives to sort at eitherthe curb during collection or an MRF were enabled.

In NCC, all recyclables were recovered through the drop-off program with increasing use of a mixed waste MRF toachieve up to 28% diversion (Figure 2a) because utilizationof a mixed waste MRF was estimated to be cheaper thanimplementation of curbside recycling. As the diversion rate

TABLE 2. Summary of Key Model Inputs

parameter default value

Collectionrefuse collection

frequency1 time per week

curbside recyclablescollection frequency

1 time per week

time from collectionto transfer station,min

10 for urban, 30 for rural

time from collectionto MRF, min

15 for urban,30 for rural

time from collectionto compost, min

15 for urban,30 for rural

time from collectionto WTE, min

10 for urban,80 for rural

time from collectionto LF, min

15 for urban,30 for rural

time from transfer stationto WTE, km

45

MRFmaterials market prices Table S4 (SI)separation efficiency

for mixed waste MRF (%)55 for each

recyclableseparation efficiency

for commingled MRF (%)94 for glass, 100 for

all other items

WTEbasic design mass burnheat rate, BTU/(kW h) 18 000

(∼19% efficiency)ferrous recovery

rate (%)90

utility sectoroffset

baseload coal andnatural gas

Landfillbasic design per EPA regulationstime frame for emissions

estimates, years100

gas collectionefficiency (%)

0 in years 1-2,50 in year 3, 70 in year 4,80 in years 5-100

gas managementscheme

conversion toelectrical energy

utility sectoroffset

baseload coal andnatural gas

FIGURE 1. Correlation between GHE and other pollutants. Thedata plotted represent least-GHE SWM strategies for NewCastle County in which all unit operations were enabled. Anegative value means that the avoided emissions exceeded theemissions from waste management.

FIGURE 2. Variation of mass flow and GHE based on use oflandfill only (0% diversion, case 1), current practice (case 2),and an alternative in which a mixed waste MRF and curbsidecollection of commingled recyclables are enabled (case 3): (a)New Castle County, (b) Sussex County.

1266 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 5, 2009

increased to 28.5%, curbside collection of recyclables wasutilized in residential sector 1 and the multifamily sector torecover more material. When the model objective was set tomaximize diversion, commingled recycling was utilized inresidential sector 2 to increase diversion to 28.91%. The MRFbeing farther away from the collection routes resulted inlonger transportation distances, which escalated the totalcost (Figure 2 and Table S11, SI). At maximum diversion(28.91%), the use of a mixed waste MRF decreased slightly,and the recyclables collected at curbside were sorted by thecrew rather than at a MRF. The shift in waste processingchoices between 28.5% and 28.91% diversion is controlledby slightly different assumptions about material losses inMRFs that receive commingled versus presorted recyclables.While mathematically correct, this diversion increase is likelyinsignificant in practice. Of course, the maximum attainablediversion rate depends on the model inputs specifyingparticipation and capture rates (Table S4, SI). Between 20%and 28% diversion, the cost increases uniformly, after whichit increases sharply with the implementation of curbsiderecycling (Figure 2). GHEs decrease consistently as diversionincreases due to benefits from remanufacturing offsets andreduced landfill emissions (Figure 2a).

The results for Sussex and Kent Counties are similar tothose for NCC except that GHEs and several other pollutantsreach minima at less than maximum diversion (Figure 2b,Tables S12 and S13, SI). GHEs increase with the implemen-tation of curbside recycling due to the rural character ofthese counties, causing increased emissions associated withadditional collection vehicles.

Yard waste composting was enabled for the next set ofcost-effective analyses (case 4). As in case 3, drop-off recyclingand a mixed waste MRF were utilized until 28% diversion,after which composting and then finally curbside recyclablescollection were implemented at the maximum diversion(Figure 3a). Composting, which was relatively cheaper, wasutilized before curbside recycling. Again, the cost increasessharply when curbside recycling is included. Between 28%and 32% diversion, GHE does not change because noadditional recyclables are recovered, resulting in no changesin the corresponding remanufacturing offsets. Compostingresults in increased CO2-fossil associated with collection andfacility operation (Tables S14 and S15, SI). These CO2-fossilemissions are approximately balanced by the reduced massburied in a landfill though this result is sensitive to the mannerin which LFG is managed (i.e., flare vs energy recovery) andits collection efficiency. While there are benefits associatedwith compost as a product in certain applications, this studydid not attribute offsets to the use of compost. The resultsfor Sussex and Kent Counties (Figure 3b, Table S15, SI) showtrends similar to those for NCC, but GHEs increase with theimplementation of composting and then curbside recyclablescollection due to the greater transport distances.

To complete the scenarios with cost as the objectivefunction, WTE was enabled with curbside recycling andcomposting (case 5). A new WTE facility is assumed to belocated in NCC, and transfer stations are assumed to beavailable in Kent and Sussex Counties (Figure S2, SI). Landfilldiversion is now defined to include waste processed by WTE,excluding the resultant ash. For NCC, once the maximumdiversion achievable via only the drop-off program is realized,WTE is increasingly utilized to achieve 85% diversion (Figure4a). Increased diversion above 85% was obtained by firstutilizing a mixed waste MRF, followed by composting andthen curbside recyclables collection. Composting was se-lected over WTE to maximize diversion because of additionaldisposal needs for the ash generation in WTE. In practicethis is inconsequential. Curbside recycling increases diversionas noncombustibles (e.g., glass and aluminum), which wouldotherwise be counted as ash for landfill disposal, are diverted

from WTE. The sharp cost increase at 88% diversion is dueto the inclusion of more costly programs to capture morematerial (Figure 4). Interestingly, these programs result inonly slight increases in diversion, but with a sharply highercost and an increase in GHEs.

The major difference in the results for the rural countiesin case 5 is that a mixed waste MRF was utilized at smallerdiversion targets. Recyclables were recovered by sortingmixed waste, after which the residual was transported to aWTE facility. This is cost-effective because the waste mustbe transported, at greater cost, to northern Delaware forcombustion while a mixed waste MRF is located closer to thepoint of waste generation (Figure S2, SI). The capital costsof WTE are such that, realistically, only one facility would belocated in Delaware, and a location near the industrializedarea (i.e., NCC) was assumed. The increase in presortedrecycling at 87% diversion in Sussex County is from presortedcommercial material. Composting and curbside recyclingwere only selected at the maximum diversion rate (Figure4b). As for NCC, GHE achieved a minimum at 88% diversionbefore composting and curbside recycling were utilized(Figure 4a).

Minimum GHE SWM Strategies. The objective of thisanalysis was to minimize GHE at increasing cost targets,starting with the cost of current practice. First, all processeswere enabled as in case 5. In NCC, WTE was the most cost-effective way to minimize GHEs until $60 million year-1, afterwhich more expensive processes were utilized with a slightdecrease in GHEs (Figure S3, SI). The use of curbside recyclingresults in the recovery of slightly more recyclable materialsthan a mixed waste MRF, yielding increased remanufacturingGHE offsets. Interestingly, although composting was enabled,it was not utilized. WTE is the most effective GHE-reducing

FIGURE 3. Variation of mass flow and GHE for alternate SWMstrategies in which curbside recyclables collection and yardwaste composting are enabled: (a) New Castle County, (b)Sussex County (case 4).

VOL. 43, NO. 5, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 1267

option because the recovered energy offsets the generationof electricity from fossil fuels. In contrast to waste manage-ment choices for NCC, increasing quantities of waste wereprocessed in a mixed waste MRF prior to entering WTE inKent and Sussex Counties (Figure S3, Table S18, SI). Theutilization of both the mixed waste MRF and WTE dependedon the cost constraint. When GHE is minimized without acost constraint, only a minimal improvement in GHEs isachieved by using commercial and multifamily recycling(Figure S3, Table S19, SI).

A scenario was explored in which GHEs were minimizedwithout WTE to represent current regulations that prohibitWTE use in Delaware. In NCC, recyclables drop-off is utilizedinitially followed by the use of a mixed waste MRF, curbsiderecyclables collection, and finally composting as the costtarget is increased (Figure S4, SI). Composting is selectedover a landfill at costs higher than $55 million year-1 althoughno emissions offset is assigned to the compost product. Thisis explained by the assumed decay rate of grass in landfillsand the LFG collection efficiency that dictates how muchgas is captured over time. The assumed decay rate for grassis relatively high (k ) 0.09 year-1), and no gas is assumed tobe collected during the first 2 years. As such, some gasproduction attributable to grass is released to the atmospherein the early years, making composting more favorable froma GHE standpoint.

In the rural counties, only a base case and minimum GHEscenario were considered because the difference in costamong these scenarios was less than 5%. For Sussex County,the primary difference between least GHE and the base case

scenario is the utilization of commingled curbside recyclingin the multifamily sector along with some additional recoveryat a mixed waste MRF (Figure S4, SI). Both of these unitoperations serve to complement the existing residential drop-off recycling program with decreasing GHEs. Diversion inthe least GHE case is 34.5%, 23.1%, and 24.5% for NC, Kent,and Sussex Counties, respectively.

Sensitivity to Varying Recyclable Market Prices. Selectedstrategies in case 5 for NCC were analyzed to evaluate whetherincreased revenue from recyclable material sales wouldincrease the use of curbside recycling. The original andupdated recyclables market prices are presented in Table S5(SI). Case 5 scenarios with diversion rates of 40% and 85%were rerun (Figure S5, SI). Despite the increased prices,curbside recycling was not selected, and WTE was stillpreferred for meeting the diversion targets. The changes insystem cost were 3-4%, which is insignificant relative to theaccuracy of the model.

Uncertainty Analysis. For NCC, the expected cost ofcurrent practice when considering uncertainty is $39 million,with a range of $32.6 million to $47.4 million and a 38%likelihood of exceeding the deterministic cost of $39.7 million(Figure S6, SI). The expected GHE of current practice is 18830MTCE year-1, with a range of 15674-21415 and a 3%likelihood of exceeding the deterministic estimate of 20726MTCE year-1 (Figure S7, SI).

Table 3 shows a subset of the most strongly correlateduncertain input parameters to the selected model outputs(cost, energy consumption, and GHE). These results can beused to prioritize the input parameters for which better dataare most needed. Comparison of CDFs for multiple SWMalternatives can be used to consider robustness as part ofSWM alternative selection.

DiscussionDifferences among SWM Strategies for Urban and RuralCounties. The higher population density in NCC resultedoverall in less costly SWM strategies. When identifying cost-effective diversion strategies with curbside recyclables col-lection enabled, unit costs at maximum diversion were $174,$507, and $631 Mg-1 in NC, Kent, and Sussex Counties,respectively. There are two caveats to this analysis. First,urban areas in Kent County (e.g., Dover) may behave morelike NCC in some respects. Second, DSWA does not controlthe manner in which cities and counties collect refuse andrecyclables, but rather manages the waste after collection.Cities and counties may implement a variety of collectionalternatives that are not optimal.

The strategy with the lowest GHE for NCC (case 6) resultsin a 74665 MTCE year-1 reduction at an incremental cost of

FIGURE 4. Variation of mass flow and GHE for alternate SWMstrategies in which curbside recyclables collection, yard wastecomposting, and WTE are enabled: (a) New Castle County, (b)Sussex County (case 5). The cost and GHE for the 88%diversion case for NCC were disaggregated into individualcomponents of the waste management system in Table S17 (SI).

TABLE 3. Correlation Factors for Uncertain Input ParametersThat Are Strongly Correlated to Cost, Energy Consumption,and GHG Emissionsa

New Castle Sussex

cost commercial residualcollection-loading timeat one service stop

compacted wastedensity in thelandfill

0.745 -0.956energy

consumptionheat rate in

combustion facilitycompacted waste

density in thelandfill

0.855 -0.932greenhouse gas

equivalentsCO2-fossil emissions

savings from aluminumremanufacturing

compacted wastedensity in thelandfill

-0.603 -0.747a Negative correlation indicates an inverse relation

between the input parameter and the output.

1268 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 5, 2009

$33.4 million year-1 relative to current practice. In contrast,the net decrease in GHEs and cost increase for Sussex Countyare 21695 MTCE year-1 and $6.4 million, respectively. Whilethe GHE reduction in Sussex County is slightly more cost-effective, there is less waste and therefore a smaller overallreduction potential. In NCC, emission reductions wererealized via both curbside recycling and WTE. In contrast,the implementation of curbside recycling increased emissionsin Sussex County (case 3) relative to scenarios with diversionlower than the maximum diversion.

Effects of Cost and Diversion Targets on Waste ProcessChoices. The ISWM DST is not constrained to apply a singlewaste process to an entire sector equally. In most solutions(e.g., Figure 2), a process that is more expensive than drop-off is only utilized to the extent required to meet a diversionconstraint. Thus, only a fraction of the total population maybe served by, for example, curbside collection to achieve thetarget diversion. Similarly, WTE was used for only a fractionof the total waste when minimizing GHE subject to a costtarget (Figure S3, SI).

In practice, it may be difficult to convince a communityof the rationale for providing only some residents with, forexample, curbside collection while expecting others to utilizedrop-off bins, or having some waste disposed in a landfillwhile other waste is treated by WTE. This is an example ofa situation where the optimal strategy could be judgedpolitically or socially infeasible. Alternative strategies thatare only incrementally more expensive than the optimalsolution, but utilize maximally different sets of facilities, canbe developed (16). This is expected to yield more efficientstrategies that may include politically or socially more viableoptions.

Counterintuitive Insights Gained through Modeling.One advantage of a mathematical analysis of a complexsystem is that it may result in outcomes that are not intuitive.When the objective was to minimize cost at a desireddiversion level, the model was able to identify a creativeapproach in which some waste was first processed througha mixed waste MRF prior to flowing to WTE (Figure 4b). Thistandem processing accomplishes the following: (1) allowsfor recovery of noncombustible recyclables (e.g., glass andaluminum) that were not captured via the drop-off programand (2) reduces the quantity of waste to be transported tothe WTE facility from the rural counties.

Ultimately, a decision-maker must determine the mostsuitable SWM plan in consideration of competing cost,environmental, and social/political considerations. Whilethere are many cases that could be examined, Table 4summarizes three key parameters for an SWM strategy at adiversion level just prior to where the cost increases sharply.Clearly, the most significant reductions in GHEs can be

realized when WTE is utilized, albeit at a higher cost. Whilethe objective of case 5 was to minimize cost at varyingdiversion constraints, the objective of case 6 was to minimizeGHE emissions. For an expenditure of $50 million year-1,GHE emissions of -17200 and -31300 MTCE year-1 arerealized for NCC in cases 5 and 6, respectively, at diversionlevels of 60% and 57% (Figures 4 and S3, SI).

This study quantifies the tradeoffs among cost, diversion,and environmental performance by using a life-cycle plan-ning tool to evaluate multiple alternatives for SWM inDelaware. The resultant trends are similar to results reportedfor several European case studies (10-14). While this studyprovides a quantitative and systematic basis for evaluatingcost, diversion, and GHE objectives for SWM choices andtheir tradeoffs, specific decisions must be made as to thedirection of future SWM. Such a decision may also involvepolitical and other subjective considerations. The quantitativeresults here are envisioned to provide the necessary infor-mation to screen for technically superior strategies that couldform the basis for such a decision-making process. Whenmaking final decisions that constrain the array of alternativesto be considered, a more narrow set of SWM alternativesshould be selected for detailed engineering analysis beforea strategy is implemented. In subsequent work, methods aredescribed to develop optimal statewide strategies based oncombinations of the county-specific alternatives describedhere.

Supporting Information AvailableWaste composition and recyclables capture rate, uncertainparameters and their distributions, and tables of mass andemissions data for each scenario. This material is availablefree of charge via the Internet at http://pubs.acs.org.

Literature Cited(1) U.S. EPA. Solid Waste Management and Greenhouse Gases: A

Life-Cycle Assessment of Emissions and Sinks, September 2006.http://epa.gov/climatechange/wycd/waste/SWMGHGreport.html (accessed July 3, 2008).

(2) Solano, E.; Dumas, R. D.; Harrison, K. W.; Ranjithan, S.; Barlaz,M. A.; Brill, E. D. Life cycle-based solid waste managements2.Illustrative applications. J. Environ. Eng. 2002, 128, 993–1005.

(3) Thorneloe, S. A.; Weitz, K.; Jambeck, J. Application of the USdecision support tool for materials and waste management.Waste Manage. 2007, 27, 1006–1020.

(4) Eriksson, O.; Frostell, B.; Bjorklund, A.; Assefa, G.; Sundqvist,J.-O.; Granath, J.; Carlsson, M.; Baky, A.; Thyselius, L. ORWAREsAsimulation tool for waste management. Resour., Conserv. Recycl.2002, 36, 287–307.

(5) Finnveden, G.; Bjorklund, A.; Moberg, A.; Ekvall, T. Environ-mental and economic assessment methods for waste manage-ment decision-support: Possibilities and limitations. WasteManage. Res. 2007, 25, 263–269.

(6) Hellweg, S.; Doka, G.; Finnveden, G.; Hungerbuhler, K. Assessingthe eco-efficiency of end-of-pipe technologies with the envi-ronmental cost efficiency indicatorsA case study of solid wastemanagement. J. Ind. Ecol. 2005, 9, 189–203.

(7) Huang, G. H.; Chi, G. F.; Li, Y. P. Long-term planning of anintegrated solid waste management system under uncertaintysI.Model development. Environ. Eng. Sci. 2005, 22, 823–834.

(8) Huang, G. H.; Chi, G. F.; Li, Y. P. Long-term planning of anintegratedsolidwastemanagementsystemunderuncertaintysII.A North American case study. Environ. Eng. Sci. 2005, 22, 835–853.

(9) Rechberger, H.; Brunner, P. H. A new, entropy based methodto support waste and resource management decisions. Environ.Sci. Technol. 2002, 36, 809–816.

(10) Emery, A.; Davies, A.; Griffiths, A.; Williams, K. Environmentaland economic modeling: A case study of municipal solid wastemanagement scenarios in Wales. Resour., Conserv. Recycl. 2007,49, 244–263.

(11) Eriksson, O.; Reich, M. C.; Frostell, B.; Bjorklund, A.; Assefa, G.;Sundqvist, J. O.; Granath, J.; Baky, A.; Thyselius, L. Municipalsolid waste management from a systems perspective. J. Clean.Prod. 2005, 13, 241–252.

TABLE 4. Cost, Emissions, and Diversion for a Waste Manage-ment Strategy Displaying Near-Optimal Characteristicsa

cost, millionsof dollars year-1

GHE,MTCE year-1 diversion, %

New Castle Countycurrent practice 39.7 20700 20case 3 45.6 13900 28case 4 51 11200 35case 5 57.8 -33000 85

Sussex Countycurrent practice 36.9 11400 19case 3 36.8 8500 26case 4 38 4400 27case 5 41.6 -5600 87

a Data are for a diversion level just prior to the level atwhich costs escalate sharply.

VOL. 43, NO. 5, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 1269

(12) Finnveden, G.; Bjorklund, A.; Reich, M. C.; Eriksson, O.; Sorbom,A. Flexible and robust strategies for waste management inSweden. Waste Manage. 2007, 27, S1-S8.

(13) Kirkeby, J. T.; Birgisdottir, H.; Bhander, G. S.; Hauschild, M.;Christensen, T. H. Modelling of environmental impacts of solidwaste landfilling within the life-cycle analysis program EASE-WASTE. Waste Manage. 2007, 27, 961–970.

(14) Reich, M. C. Economic assessment of municipal waste man-agement systemssCase studies using a combination of life cycleassessment (LCA) and life cycle costing (LCC). J. Clean. Prod.2005, 13, 253–263.

(15) Chester, M.; Martin, E.; Sathaye, N. Energy, greenhouse gas,and cost reductions for municipal recycling systems. Environ.Sci. Technol. 2008, 42, 2142–2149.

(16) Harrison, K. W.; Dumas, R. D.; Solano, E.; Barlaz, M. A.; Brill,E. D.; Ranjithan, S. R. A decision support system for developmentof alternative solid waste management strategies with life-cycleconsiderations. ASCE J. Comput. Civ. Eng. 2001, 15, 44–58.

(17) Solano, E.; Ranjithan, S.; Barlaz, M. A.; Brill, E. D. Life cycle-based solid waste managements1. Model development. J.Environ. Eng. 2002, 128, 981–992.

(18) Harrison, K. W.; Dumas, R. D.; Barlaz, M. A.; Nishtala, S. R. Alife-cycle inventory model of municipal solid waste combustion.J. Air Waste Manage. Assoc. 2000, 50, 993–1003.

(19) McDougall, F. R.; White, P. R.; Frankie, M.; Hindle, P. IntegratedSolid Waste Management: A Life Cycle Inventory; BlackwellScience: Oxford, U.K., 2001.

(20) Kaplan, P. O.; Barlaz, M. A.; Ranjithan, S. R. Life-cycle-basedsolid waste management under uncertainty. J. Ind. Ecol. 2004,8, 155–172.

(21) U.S. EPA. Waste Characterization of Municipal Solid Waste inthe United States: 2000 Update; EPA-530-R-02-001; Washington,DC, 2002.

(22) Delaware Solid Waste Authority. Project Final Report for theDelaware Solid Waste Characterization Study; File No. 0296001;SCS Engineers: Long Beach, CA, 1997.

(23) Delaware Solid Waste Authority. Assessment of Delaware SolidWaste Discards in 2000 and the Potential Recycling of Materials;Franklin Associates, Ltd.: Prairie Village, KS, 2002.

(24) Delaware Department of Labor. 1998 Annual Averages by MajorIndustry SectorsDelaware. http://www.delawareworks.com/oolmi/information/data/EmpWages/IND1998YR.shtml (ac-cessed July 3, 2008).

ES8018447

1270 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 5, 2009