Integrated Life-Cycle Assessment and Life-Cycle Cost Analysis Model for Concrete Bridge Deck Applications Alissa Kendall1; Gregory A. Keoleian2; and Gloria E. Helfand3 Abstract: An integrated life-cycle assessment and life-cycle cost analysis model was developed and applied to enhance the sustainability of concrete bridge infrastructure. The objective of this model is to compare alternative bridge deck designs from a sustainability perspective that accounts for total life-cycle costs including agency, user, and environmental costs. A conventional concrete bridge deck and an alternative engineered cementitious composite link slab design are examined. Despite higher initial costs and greater materialrelated environmental impacts on a per mass basis, the link slab design results in lower life-cycle costs and reduced environmental impacts when evaluated over the entire life cycle. Traffic delay caused by construction comprises 91% of total costs for both designs. Costs to the funding agency comprise less than 3% of total costs, and environmental costs are less than 0.5%. These results show life-cycle modeling is an important decision-making tool since initial costs and agency costs are not illustrative of total life-cycle costs. Additionally, accounting for construction-related traffic delay is vital to assessing the total economic cost and environmental impact of infrastructure design decisions. DOI: 10.1061/共ASCE兲1076-0342共2008兲14:3共214兲 CE Database subject headings: Life cycles; Environmental issues; Bridge design; Fiber reinforced materials; Concrete pavements; Traffic delay; Construction costs.

Introduction Concrete pavements and structures such as bridges are fundamental components of our transportation network, and thus also fundamental to economic vitality and personal mobility. Yet, the American Society of Civil Engineers estimates that the poor conditions of United States roads cost users $117.2 billion in added operating costs and time lost in traffic delay annually 共ASCE 2005兲. Poor roadway conditions persist despite economic and material investment in highways and roads of approximately $64.6 billion and 260 million metric tons of concrete annually in the United States 共FHWA 2002; Kelly 1998兲. The most recent highway bill signed into law, the Safe, Accountable, Flexible, Efficient Transportation Equity Act: A Legacy for Users 共SAFETEA-LU兲, authorizes $25.2 billion for interstate maintenance alone through the year 2009, and $21.6 billion for preventive maintenance and improvements on highway bridges through 2009 共FHWA 2005兲. 1 Assistant Professor, Dept. of Civil and Environmental Engineering, Univ. of California, 1 Shields Ave., Davis, CA 95616; formerly, Doctoral Student, Center for Sustainable Systems, School of Natural Resources and Environment, Univ. of Michigan, 440 Church St., Dana Building, Ann Arbor, MI 48109. 2 Associate Professor, Center for Sustainable Systems, School of Natural Resources and Environment, Univ. of Michigan, 440 Church St., Dana Building, Ann Arbor, MI 48109 共corresponding author兲. E-mail: [email protected] 3 Associate Professor, School of Natural Resources and Environment, Univ. of Michigan, 440 Church St., Dana Building, Ann Arbor, MI 48109. Note. Discussion open until February 1, 2009. Separate discussions must be submitted for individual papers. The manuscript for this paper was submitted for review and possible publication on June 22, 2006; approved on October 9, 2006. This paper is part of the Journal of Infrastructure Systems, Vol. 14, No. 3, September 1, 2008. ©ASCE, ISSN 1076-0342/2008/3-214–222/$25.00.

The magnitude of investment demonstrated by these allocations underscores the need to approach road building and repair from a new perspective—long term and preventive, rather than short term and corrective. Life-cycle assessment 共LCA兲 and life-cycle cost analysis 共LCCA兲 methodologies provide the means for this kind of evaluation. An integrated LCA and LCCA model 共LCA-LCCA兲 was developed to provide a holistic assessment of the economic costs of concrete infrastructure applications, in this case a highway bridge deck. LCA is a framework designed to evaluate the environmental performance of a product or process throughout its life cycle, including raw material acquisition, production, use, final disposal or recycling, and the transportation needed between these phases 共ISO 1997兲. Often, LCA elucidates unseen environmental and social burdens incurred over a product or system’s lifetime. By quantifying environmental and social burdens, the LCA model provides the data necessary for comprehensive LCCA. The LCA model was developed prior to integration with the LCCA model. Its methods and results are described in Keoleian et al. 共2005兲. Like the LCA model, the LCCA model analyzes the costs associated with all phases of an infrastructure application throughout its life cycle. LCCAs vary in scope and depth, accounting for different kinds of costs and benefits. For example, LCCA may account only for agency costs, which are the costs incurred by the funding agency; it may account for user costs in addition to agency costs, such as costs incurred by motorists who are delayed or detoured by construction related traffic; and LCCA, more rarely, may include environmental costs, such as the pollution damage costs associated with construction processes. According to a study conducted on behalf of the New Jersey Department of Transportation 共DOT兲, only 12.5% of state DOTs apply any sort of LCCA on bridges 共Ozbay et al. 2003兲. However, LCCA could prove extremely useful in bridge applications because bridges require significant capital investment but also con-

214 / JOURNAL OF INFRASTRUCTURE SYSTEMS © ASCE / SEPTEMBER 2008

Downloaded 19 Jan 2009 to 141.211.206.116. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright

siderable investment for maintenance and rehabilitation during the bridge’s service life. Other studies have developed life-cycle costing methods for roads, highways, and highway infrastructure. Ehlen identified life-cycle costs for polymer-reinforced concrete bridges that include agency and user costs 共driver delay, vehicle operating, and vehicle accident costs兲 and third party costs 共Ehlen 1999兲. Ehlen classified third party costs as the upstream environmental costs associated with construction materials 共pollution from mining, processing, and transportation兲 and the downstream environmental costs related to construction activities such as runoff. While Ehlen noted the increasing importance of third party costs, they were not quantified and environmental impacts from construction related traffic delay were not identified either. In addition to LCCA studies like Ehlen’s, there has been a considerable effort to model the expected maintenance costs of civil infrastructure. Frangopol and others have done extensive work to optimize bridge maintenance costs using reliability-based optimization methods to predict and reduce life-cycle costs for bridges. These analyses examine and optimize agency costs, and generally neglect user, third party, and environmental considerations 共Enright and Frangopol 1999; Estes and Frangopol 2001; van Noortwijk and Frangopol 2004兲. Despite neglecting nonagency costs, these studies develop valuable methods for predicting and optimizing maintenance and rehabilitation schedules. Maintenance and rehabilitation schedules are extremely important in LCCAs since they determine the number and timing of construction events that drive life-cycle cost results. There is also existing software for applying LCCA to bridges, namely BLCCA, created by the National Cooperative Highway Research Program and funded by the Transportation Research Board, and BridgeLCC created by the Office of Applied Economics, Building and Fire Research Laboratory at the National Institute of Standards and Technology 共Ehlen 2003; Hawk 2002兲. These software applications were reviewed in a 2004 Federal Highway Administration 共FHWA兲 report 共FHWA 2004a; Palisade Corporation 2000兲. Both BLCCA and BridgeLCC provide user and agency cost calculations. In addition, BridgeLCC allows the software user to enter third party cost parameters, and BLCCA includes vulnerability costs, which are the expected costs associated with extreme events, such as earthquakes or explosions, that are unlikely but possibly disastrous. Both software programs have uncertainty modeling capabilities. Some limitations and weaknesses noted by the FHWA include a complex user interface and limited uncertainty models for BLCCA, and for BridgeLCC, a lack of details in the cost output and the opinion that BridgeLCC does not reflect “realistic bridge-related experience.” Other models have also been developed for LCCA of pavements, rather than bridges, such as RealCost, developed by the FHWA Office of Asset Management. This program accounts for user and agency costs and allows for both deterministic and probabilistic output 共FHWA 2004b兲. None of these software tools were specifically designed for bridge deck decision making, but all are useful points of reference to understand previous measures that computerized and automated LCCA for road and highway infrastructure. The objective of this research was to create a model that was highly tailorable, where new materials, new designs, and multiple bridge deck sites could be modeled. Input variables include component materials, material durability, and the bridge deck repair and rehabilitation schedule. In addition, traffic parameters, fuel economy improvements, and roadway and construction work zone characteristics can be manipulated and are integrated with

the traffic model. The emissions model and fuel economy model can be updated to replicate the climate characteristics and regional fleet characteristics for the location of the bridge deck under evaluation. Moreover, the integrated LCA-LCCA model developed here is unique in its capacity to calculate life-cycle user, environmental, and agency costs, including the upstream burdens of material and fuel production. This ability to quantify upstream burdens and environmental and user costs is an additional capability not available in other models.

Model Description LCA-LCCA model development began with the LCA model. Within the LCA model, a life-cycle inventory 共LCI兲 is performed. A LCI evaluates the inputs and outputs of the system under consideration by examining each phase of the system’s life cycle. The life-cycle phases include material production, consisting of the acquisition and processing of raw materials; distribution, which accounts for transport of materials and equipment to and from the construction site; construction and rehabilitation of the bridge deck, including all construction processes and construction related congestion effects; use of the bridge deck, which models vehicular travel over the bridge during its service life; and finally end of life, which assesses demolition of the bridge deck, transportation of the material to a landfill or recycling facility, and processing of the materials. LCI datasets for each input are required. A LCI dataset provides the life-cycle information for a given material or energy source. For example, the LCI dataset for cement provides the total primary energy needed to produce a unit of cement. Total primary energy includes the energy required for extraction, refining, and production of the energy source. The dataset also quantifies nonfuel material inputs, such as the mass of limestone required, and all the inputs and outputs required to extract and process these raw materials. The sources for these datasets may be found in Keoleian et al. 共2005兲. The LCA modeling required integration of three additional models: a vehicle emissions model, MOBILE6 developed by the U.S. Environmental Protection Agency 共EPA兲; a construction equipment model, NONROAD, also developed by the EPA; and a traffic flow model developed at the University of Kentucky 共Kentucky Transportation Center 2002; U.S. EPA 2000, 2002兲. The LCCA model, whose methods were first developed by Chandler, but enhanced in this model, was integrated with the LCA model 共Chandler 2004兲. The LCCA model uses some of the same modeling parameters and user inputs to the LCA model, such as the bridge deck system specifications, material specifications, traffic flow rate, and the construction timeline. The LCCA model also uses some of the results of the LCA model to calculate environmental and user costs. The cost model requires inputs for pollution damage costs, the value of lost time to personal and commercial vehicles delayed in traffic, costs of agency construction activities, and discount rates for environmental, user, and agency costs. Fig. 1 shows the integrated model framework.

System Definition The LCA-LCCA model was applied to a highway overpass bridge deck, and two alternative designs that could be used to construct it. The first design uses conventional mechanical steel expansion joints, and the second design applies engineered cementitious composite 共ECC兲 link slabs in place of conventional expansion

JOURNAL OF INFRASTRUCTURE SYSTEMS © ASCE / SEPTEMBER 2008 / 215

Downloaded 19 Jan 2009 to 141.211.206.116. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright

Fig. 1. Integrated LCA-LCC model flow diagram

joints. ECC is an advanced cement-based material that has the advantages of concrete such as great compressive strength, but also is capable of ductile behavior like a metal. ECC is considered more durable than conventional concrete and the link slab design is expected to eliminate key failure modes associated with the conventional expansion joint design. Thus, the ECC link slab design is expected to increase the life of the bridge deck and reduce the number of repairs necessary over the bridge deck life 共Li et al. 2003; Li 2003兲. The bridge examined in this application was modeled after an overpass in southeast Michigan at the intersection of two highways, M-14 and US-23. During nonconstruction periods the bridge has two lanes with one-way average annual daily traffic flow 共AADT兲 of 35,000 vehicles with no annual growth rate. The structure is 0.16 km 共0.1 mi兲 long and contains eight expansion joints in the case of the conventional design, and eight link slabs in the ECC link slab design. Fig. 2 depicts the two bridge deck joint designs evaluated in this study. The bridge substructure is assumed to be 30 years old and is undergoing its first deck replacement at the beginning of this analysis. The time horizon for the analysis is 60 years. At the end of 60 years the bridge substructure will be 90 years old and is expected to need replacement. The study evaluates three repair processes performed on the deck; bridge deck replacement, shallow overlay, and pothole filling. The repair and maintenance schedules, shown in Fig. 3, reflect the assumption that only these three repair and reconstruction activities take place for both systems. This timeline is a simplification of all possible repair processes that could be undertaken, and excludes some maintenance activities. The conventional deck design requires another deck replacement after 30 years, but the ECC link slab deck design is expected to last twice as long, meeting the bridge’s remaining 60 years of life. The timeline shown in Fig. 3 illustrates a doubling of bridge deck life, 20 rather than 15 year intervals for deck overlays, and a doubling in time between pothole patching events for the ECC system compared to the conventional system. A doubling of deck life for the ECC system significantly influences modeling results. However, as noted before, the link slab design eliminates key failure modes initiated at the expansion joint in the conventional design. A key failure mode associated with the expansion joint is caused by degradation of the joint’s seal, a strip of rubber sandwiched between the two sides of the steel expansion joint. When the seal deteriorates, water and salts creep into the joint and initiate corrosion on the reinforcing bar, which eventually causes

cracking and potholes on the surface of the deck. By eliminating this failure mode, the link slab design extends the life of the bridge deck and deck surface. Research supports this doubling of bridge deck life when the ECC link slab system is treated as a jointless bridge. Jointless bridges are designed with no expansion joints or similar mechanisms such as link slabs, and like the link slab design, eliminate the expansion joint failure mode. A study in New York concluded that jointless bridges remained in acceptable conditions about twice as long as comparable jointed bridges 共Yanev and Chen 1993兲.

Key Cost Parameters Agency Costs Agency costs were provided by a Michigan construction company that requested to remain anonymous. Table 1 shows the cost breakdown for the three construction activities modeled. These costs reflect material cost assumptions of $130.80/m3 共$100/ yd3兲 for conventional concrete and $329.99/m3 共$250/ yd3兲 for ECC. This ECC cost is the expected cost once the material is more widely used and mass produced; the cost of this material upon introduction would be higher. Environmental Costs In this study pollution damage costs were used to characterize environmental costs. These costs are calculated for key pollutants

Fig. 2. Bridge deck with ECC link slab and conventional mechanical steel expansion joint

216 / JOURNAL OF INFRASTRUCTURE SYSTEMS © ASCE / SEPTEMBER 2008

Downloaded 19 Jan 2009 to 141.211.206.116. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright

Fig. 3. Bridge deck rehabilitation timelines

including six of the seven criteria pollutants specified by the EPA, and three primary greenhouse gases 共GHGs兲. The criteria pollutant costs were derived from Banzhaf and colleagues, except for the cost of volatile organic compounds 共VOCs兲 which was provided by Matthews and Lave 共Banzhaf et al. 1996; Matthews and Lave 2000兲. The GHG costs were adapted from Tol’s value of $60 per metric ton 共t兲 of carbon 共Tol 1999兲. The Banzhaf study estimated damage costs from Midwestern coal-fired electric power plants. While the regional aspects of this study are well suited to this application, it represents cost estimates from point sources, not mobile sources, and a significant portion of the criteria pollutants generated over the bridge deck life cycle comes from mobile vehicle sources. In general vehicle emissions emitted at ground level rather than from an elevated stack result in increased exposure for humans and thus greater damage to health, so the costs estimated by Banzhaf can be considered conservative for this LCCA application 共Lai and Thatcher 2000兲. Pollutant damage costs from Banzhaf et al. 1996 are based on “morbidity health values, mortality risks, and 关willingness to pay兴 to avoid mortality risks” 共Banzhaf et al. 1996兲. The VOC damage cost, taken from Matthews and Lave, is based on a mean value from five studies on social damage costs for VOCs. Matthews and Lave emphasize the most important effects of pollution are human health effects, but do not specify the basis for VOC damage cost estimates of each source used in the average. Tol’s marginal cost of carbon is based on results from the climate framework for uncertainty, negotiation and distribution 共FUND兲 model. FUND integrates social factors such as population distribution, technology, and economics with climate modeling 共The Research Unit for Sustainability and Global Change Universität Hamburg 1999兲. The marginal cost of carbon as a GHG is developed based on a cost benefit and cost effectiveness

analysis of different possible climate and emissions scenarios. Costs resulting from climate change, such as the economic impacts of sea level rise and the migration of affected populations, as well as human health impacts such as increased exposure to tropical diseases like malaria, and heat and cold stress, are evaluated and weighed against the costs of emissions reductions 共Tol et al. 2003兲. The marginal cost of carbon must be converted to apply to the GHGs evaluated in this study, which include carbon dioxide 共CO2兲, methane 共CH4兲, and nitrous oxide 共N2O兲. Each of these gases has a different global warming potential reflecting the degree to which they trap heat in the earth’s atmosphere, with the baseline defined by the effects of CO2. The cost per metric ton of carbon can be converted to cost per metric ton CO2共$ / tCO2兲 by multiplying by a factor of 44/12, the ratio of the molecular weights of CO2 and carbon. CH4, and N2O have global warming potentials of 23 and 296, respectively, so these factors then multiply the cost of CO2 for their cost per metric ton 共Joos and McFarland 2001兲. Table 2 shows the pollution damage cost estimates used in the LCCA model for each pollutant. In all cases, pollutant damage costs were adjusted to 2003 United States dollars. Pollution damage costs are difficult to calculate, and all have a significant amount of uncertainty associated with them. Those pollutants that contribute to global warming pose perhaps even more difficulties for cost estimation. In order to address the uncertainty in pollution damage costs, Monte Carlo simulation was performed on pollution costs. User Costs User costs consist of three types; traffic delay, increased risk of traffic crashes, and increased vehicle operating costs due to con-

Table 1. Agency Cost Breakdown for ECC Bridge Deck System Rehabiliation Construction event

Deck type

Total cost 共$兲

Material cost 共$兲

Labor cost 共$兲

Equipment cost 共$兲

Deck replacement

ECC link slab Conventional ECC link slab Conventional ECC link slab Conventional

431,186 385,963 100,521 185,102 654 654

97,038 89,557 15,615 31,625 92 92

229,439 192,448 58,556 117,413 471 471

51,427 47,409 16,591 22,007 92 92

Resurfacing Patching

JOURNAL OF INFRASTRUCTURE SYSTEMS © ASCE / SEPTEMBER 2008 / 217

Downloaded 19 Jan 2009 to 141.211.206.116. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright

Table 2. Air Pollution Damages Costs by Impacted Region

Table 3. Life-Cycle Costs for Bridge Deck System Designs

Average cost

Conventional system

共2003 US$/t兲

共$兲

Pollutant name

Urban

Urban fringe

Rural

Global

Particulate matter Nitrogen oxides Sulfur dioxides Carbon monoxide Lead VOC Carbon dioxide Nitrous oxide Methane

6,144 156 170 2 3,955 1,960 — — —

2,750 65 88 1 2,059 1,960 — — —

800 19 21 0 480 1,960 — — —

— — — — — — 21 7,112 384

struction. User delay, or time spent in construction related traffic, dominates user costs. Delay costs for passenger automobiles, single unit trucks, and combination trucks are $13.61, $21.78, and $26.21, respectively. These costs are based on estimates developed in a Federal Highway Administration study 共Walls and Smith 1998兲. The costs are in 1996 dollars and are brought forward using the consumer price index to 2003 dollars. User delay is calculated based on the time spent in construction related congestion above what would be spent traveling the equivalent distance in normal traffic flow conditions. Increased traffic crashes related to construction work zones and increased distance traveled when detours are used to avoid construction zones also contribute to user costs. Data for work zone accidents in the state of Michigan were used to calculate the increased cost for users in construction work zones due to a higher risk of fatality and injury compared to roadway use when no construction zone is in place 共Michigan Department of State Police 1994–2001; Michigan Department of Transportation 2002兲. The result is an estimated $0.08 per vehicle kilometer traveled 共VKT兲 关$0.13 per vehicle mile traveled 共VMT兲兴 in the construction zone and a $0.06 increased cost per additional VKT 共$0.09 increased cost per additional VMT兲 traveled when a detour is taken. These calculations assume that the relative risk of crashes is proportional to the distance traveled through a work zone versus distance traveled on typical roadway with no construction. The last element of user cost is based on increased vehicle operating costs when a construction work zone is in place. Vehicles delayed in construction related traffic have higher fuel consumption and thus higher fuel costs compared to normal flow conditions. If vehicles avoid congestion by detouring, they travel further than those that stay on the highway and also have increased fuel consumption. Increased fuel consumption in the work zone is estimated based on the difference in fleet average fuel economy between highway fuel economy and city fuel economy for passenger vehicles and heavy duty trucks 共Bradley 2000; Hellman and Heavenrich 2003兲. Fuel costs are based on a 10 year average 共1993–2003兲 of retail fuel costs, approximately $0.40/L for gasoline and $0.35/L for diesel fuel in constant 2003 dollars 共Davis and Diegel 2004兲. Fuel costs comprise less than 2% of total user costs so fuel price volatility is unlikely to affect results. Discount Rate The discount rate used in this model is based on values recommended by the United States Office of Management and Budget

Agency cost 640,000 User cost 21,000,000 Environ. costs 100,000 Total costs 22,000,000 a Calculated from prerounded costs.

ECC system

ECC advantagea

共$兲

共%兲

450,000 18,000,000 80,000 19,000,000

29 14 21 15

共OMB兲 and is estimated at a real discount rate of 4% for user and agency costs 共Office of Management and Budget 2005兲. Some goods, typically including environmental goods, may be discounted at a different rate than private market transactions due to a concern that society is underinvesting in these goods 共Gramlich 1990兲. Pollution damage costs here are subject to a variation of exponential discounting, a sliding discount rate that accounts for the immediate, near, and medium future. This scale was developed by Weitzman, who conducted a survey of over 2,000 leading economists and created the following method: for the immediate future, Years 1–5, a 4% discount rate is used; for the near future, Years 6–25, a 3% discount rate is used; and for the medium future, Years 26–75, a 2% discount rate is used 共Weitzman 2001兲. The reason for a sliding scale is that, given the significant uncertainty in environmental impacts and their costs, appropriate rates of return on capital many years into the future are unknown 共Weitzman 1998兲. The selection of a discount rate can be a topic of controversy in LCCA, since there are no hard and fast rules regarding discount rate selection. In the system evaluated here, since costs are incurred over a 60-year service life, the selection of a discount rate significantly influences results. While the OMB discount rates are considered to be a reasonable assumption of discount rates for public infrastructure investment, opinions vary on what discount rate should be applied. Because of this uncertainty, sensitivity analysis was performed on discount rate selection.

Results Life-Cycle Cost Results The results for life-cycle costs, shown in Table 3, demonstrate that overall the ECC link slab system has a cost advantage over the conventional system in all categories assessed. These costs are based on the 60-year service schedule for construction events shown in Fig. 3. The repair and rehabilitation timeline drives the results shown above. Despite higher initial costs for the ECC system, shown in Table 1 for bridge deck replacement costs, the more frequent repair and replacement rates for the conventional system mean that over time the conventional system accumulates costs that exceed ECC system costs. User costs overwhelmingly dominate total life-cycle costs, which are made up of agency, user, and environmental costs. Environmental costs are notably small compared with agency and user costs. Of these user costs, time lost to vehicles delayed in construction related traffic account for 94% of all user costs and 91% of total life-cycle costs in both cases. Essentially, the magnitude of the cost results is driven by parameters for traffic and traffic modeling. For example, on a road with lower traffic vol-

218 / JOURNAL OF INFRASTRUCTURE SYSTEMS © ASCE / SEPTEMBER 2008

Downloaded 19 Jan 2009 to 141.211.206.116. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright

Fig. 5. Life-cycle costs at nominal discount rates from 0 to 10%

Fig. 4. Life-cycle and vehicle delay costs with increasing AADT

umes but similar capacity, the agency and environmental costs are expected to have more influence on results. Backup at a work zone is caused by traffic volumes exceeding the construction zone capacity. Thus, if traffic volume is lower, construction events would be less likely to cause congestion and user costs would drop significantly. In all cases, user costs refer only to traffic impacts as they differ from nonconstruction traffic conditions. Because of the dominance of traffic related costs on results, sensitivity analysis was run on AADT, a key parameter in determining traffic congestion. Sensitivity to AADT Variations Fig. 4 shows the total life-cycle cost results, as well as the cost of vehicle delay, as one-way AADT varies from 25,000 to 35,000 共one-way traffic兲 with no annual growth rate. The figure shows a dramatic increase in total life-cycle costs, driven by the cost of vehicle delay, between AADT values of 30,000 and 35,000. This suggests that user costs may vary greatly from one project to the next, and collecting accurate AADT data and accurately assessing road capacity can be particularly important on stretches of road where capacity loss during construction causes excessive delays, as in the 35,000 AADT case. DOTs may want to spend more time and effort on managing traffic by providing the public with information far in advance and providing sufficient opportunities for motorists to detour or plan trips differently if delays are imminent. Reducing the number and duration of repair and rehabilitation events would also be important for reducing traffic delay impacts.

The better cost performance of the conventional system at higher discount rates can be explained by the construction schedule. For the initial construction process, the ECC system performs poorly compared with the conventional system. However, as time goes on, the conventional system requires more frequent repair activities. The cost of these future repair activities are greatly affected by the discount rate. For example the agency cost of a conventional bridge deck replacement, scheduled to take place in Year 30, is $385,963. When discounted at 4% the present value of the cost of the bridge replacement today is approximately $124,000, when discounted at 7.2% the cost is $51,000, and when discounted at 10% the present value falls to $24,000. So, despite the more frequent repairs needed in the conventional system, as the discount rate grows these repairs become less significant and total costs are dominated by the costs incurred in the first year. In this sensitivity analysis environmental costs are discounted in the same manner 共no sliding scale of discount rate兲 and at the same rate as all other costs. Fig. 5 shows the agency, user, and environmental life-cycle costs. The conventional system initially costs 59% more than the ECC system at a 0% discount rate, but with a 10% discount rate the conventional system costs less than the ECC system by 9%. The break even discount rate for total life-cycle costs, where both systems cost approximately $16,500,000, is 7.2%. The ECC sys-

Sensitivity to Discount Rate Selection As mentioned previously, selecting a discount rate in LCCA can be a polemic issue. Moreover, the results of an LCCA, especially for long-lived systems such as public infrastructure, can vary greatly based on the selection of a discount rate. Because bridge management is funded by government agencies, using the OMB’s estimates for the discount rate is a reasonable method of approximation. However, even the OMB’s estimates have varied greatly over the last 2 decades. For this reason, the LCCA was run at discount rates from 0 to 10%. While the ECC bridge deck system has a cost advantage in the base case scenario, and all lower discount rate values, the two systems become equal in total lifecycle costs at about 7.2%, and at higher discount rates the conventional system gains a cost advantage over the ECC system.

Fig. 6. Total life-cycle pollution damage costs

JOURNAL OF INFRASTRUCTURE SYSTEMS © ASCE / SEPTEMBER 2008 / 219

Downloaded 19 Jan 2009 to 141.211.206.116. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright

Table 4. Probability Distribution Functions for Pollution Damage Costs 共2003 United States $/t兲 Air pollutant name

Probability distribution function

Carbon dioxide Methane Nitrous oxide Carbon monoxide Lead Nitrogen oxides Particulate matter ⬍10 ␮ Sulfer oxides Volatile organic compounds

Lognormal: mean= 26, SD= 76 Lognormal: mean= 600, SD= 1,700 Lognormal: mean= 7,900, SD= 22,000 Uniform: range= 0.09− 1.51 Uniform: range= 1,865− 2,253 Uniform: range= 38− 91 Uniform: range= 2,243− 3,258 Uniform: range= 51− 125 Uniform: range= 210− 5,767

tem environmental cost advantage at a constant 4% discount rate is only 10%, notably different from the 21% cost advantage shown in Table 3 where a sliding scale for discounting is applied. Despite the change in environmental cost advantage, the effect of the discount rate change on ECC system total life-cycle cost advantage is less than 1%. Pollution Damage Costs Pollution damage costs are small compared to the agency and user cost results. However, only air pollution damage costs were included in these results, while the damage costs of water pollution and runoff were not. Fig. 6 shows the environmental damage costs for the two bridge deck systems, broken down by the different air pollutants analyzed in this LCCA. CO2 and VOCs dominate the total pollution damage costs. For the conventional system, CO2 accounts for 76% of total emissions cost. Seventy nine percent of the CO2 emissions result from fossil fuel combustion in the traffic phase, about 20% from the materials phase, and only 1% from all other phases combined. The contribution of GHGs other than CO2 amount to less than 1% of total environmental costs and only slightly more than 1% of total GHG costs, despite their higher global warming potentials. The ECC system has similar results. CO2 accounts for 69% of all emissions cost, where 83% of the emissions result from the traffic phase, 16% from the materials phase, and only 1% from all other lifecycle phases. For VOCs, the costs are even more dependent on traffic phase results; nearly 100% in the conventional case and 99% in the ECC. Monte Carlo Simulation Applied to Pollution Damage Costs LCCA, especially of long lived systems, is subject to inherent uncertainty. Much of this uncertainty is simply because of difficulty in predicting exactly what events and conditions will exist

in the future. While most parameters in this model are subject to a degree of uncertainty, the uncertainty for damage costs of pollutants, especially GHGs, is high. This is primarily a result of the complexity of the science and economics behind climate change issues. For example, which types of costs are examined, how far into the future the cost of pollution is assessed, and which climate models are used all affect the magnitude of damage costs. Due to the uncertainty in damage cost estimates, Monte Carlo simulation was used to develop a range of costs for pollution damage over the service life of the bridge. Monte Carlo simulation requires that each variable, in this case pollution damage costs, be defined by a probability distribution. Table 4 shows the probability distributions used in the Monte Carlo simulation for all of the assessed pollutants. For criteria pollutant costs, a uniform distribution defined by the range in costs provided in their source articles was used. In all cases, damage costs for urban fringe areas were used when possible. In Banzhaf et al. 共1996兲 the damage cost ranges represent the 90% confidence interval created in that study’s Monte Carlo simulation. Despite large ranges, the damage costs are quite conservative compared to estimates for mobile source emissions. For example, in a study by Delucchi, the estimated lower bound for PM10 costs is $9,750/t, still higher than the upper bound in the costs reported by Banzhaf et al. 1996 共Delucchi 1998兲. Consequently, the environmental cost estimated in this analysis is likely a conservative estimate. GHG costs for this study were calculated using Tol’s 1999 value of 60 $/tC. Tol did not include a range of uncertainty for this estimate. In a 2005 study, Tol performed a review of previous estimates for the marginal cost of CO2, and provided central estimates from each source, along with other factors such as author weights, whether the study was subjected to peer review, the time horizon of the study, etc. 共Tol 2005兲. For this analysis, the central estimates from all peer reviewed sources were taken, unweighted, and a best fit curve created within Crystal Ball, the software used to perform the Monte Carlo simulation 共Decisioneering 2004兲. Results from the Monte Carlo simulation are based on a trial of 2,000 runs, and costs were discounted using Weitzman’s sliding scale of discount rates. Table 5 shows a summary of results from the simulation for the ECC and conventional system. Not surprisingly, results for total environmental costs were highly correlated with the cost of greenhouse gas emissions. Within the 90% confidence interval for total environmental costs, there is a significant range of potential costs. GHGs, and CO2 in particular, drive this observed variance in total environmental costs. For the conventional system, GHGs contribute 86% of the variance, and 77% for the ECC system. These results show that compared with user costs, environmental costs are small, even when uncertainty is taken into ac-

Table 5. Monte Carlo Simulation Results

GHG cost Environ. cost

Bridge design

Mean 共$兲

Median 共$兲

90% certainty 共$兲

ECC link slab Conventional ECC link slab Conventional

68,000 94,000 100,000 130,000

22,000 31,000 65,000 71,000

1,700–290,000 2,300–400,000 2,300–290,000 3,400–400,000

220 / JOURNAL OF INFRASTRUCTURE SYSTEMS © ASCE / SEPTEMBER 2008

Downloaded 19 Jan 2009 to 141.211.206.116. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright

count. However, at the upper limit of the 90% certainty interval environmental costs are equal to about half that of the total agency costs.

Conclusion This paper describes the development of an integrated LCALCCA model and its application to a concrete bridge deck. Two potential bridge decks were evaluated: a conventional mechanical steel expansion joint design and an ECC link slab design. Model results show reduced costs for the ECC link slab design over the 60-year service life while the real discount rate is less than 7.2%. The superior results for the ECC life cycle costs are driven by the repair and rehabilitation schedule, which is based on an approximate doubling of time between repair events over the conventional system. User costs dominate other costs, and comprise more than 90% of total life cycle costs for both systems. Sensitivity analysis of AADT and traffic related costs show that at lower traffic volumes, the dominance of user cost drops to a magnitude where agency and environmental costs play a significant role in the results. By accounting for user costs in this LCCA, a stronger argument for the ECC system can be made. While agency costs are also lower for the ECC system, the savings in avoided user costs are much greater in magnitude and augment the cost advantage of the ECC system. Because of significant uncertainty in damage costs, an uncertainty analysis using Monte Carlo simulation was performed. Results show that total environmental costs are highly correlated with the cost of carbon, and that the range of uncertainty is great. At the upper end of the 90% certainty range, the environmental costs approach about half those of agency costs, which is significant given that conservative damage cost estimates were used, and only air pollutants, and not water effluents, were modeled. This study highlights the importance of evaluating costs of road infrastructure over time. A short-term perspective would lead to selection of the conventional bridge deck despite the ECC link slab system’s life-cycle cost advantage. These results also demonstrate the importance of including user costs in road infrastructure accounting. While the importance of these factors will vary in other applications, this study demonstrates that a LCCA including both time dimensions and costs external to the funding agency can indicate different net benefits for a project than a conventional cost analysis. Future model development will include enhanced uncertainty modeling, sensitivity analysis to the construction timeline, and incorporation of additional construction event types. In addition, the integrated LCC-LCCA model is being applied to other road infrastructure applications including pavement overlays.

Acknowledgments This research was funded through a NSF MUSES Biocomplexity Program Grant 共Grant Nos. CMS-0223971 and CMS-0329416兲. Materials Use: Science, Engineering, and Society 共MUSES兲 supports projects that study the reduction of adverse human impact on the total interactive system of resource use, the design and synthesis of new materials with environmentally benign impacts on biocomplex systems, as well as the maximization of efficient use of materials throughout their life cycles.

References ASCE. 共2005兲. “Report card for America’s infrastructure.” Reston, Va. Banzhaf, H. S., Desvousges, W. H., and Johnson, F. R. 共1996兲. “Assessing the externalities of electricity generation in the Midwest.” Resource Energ. Econ., 18, 395–421. Bradley, R. 共2000兲. “Technology roadmap for the 21st century truck program.” U.S. Department of Energy, Energy Efficiency and Renewable Energy, Washington, D.C. Chandler, R. F. 共2004兲. “Life-cycle cost model for evaluating the sustainability of bridge decks: A comparison of conventional concrete joints and engineered cementitious composite linkslabs.” Master of Science, Univ. of Michigan, Ann Arbor, Mich. Davis, S. C., and Diegel, S. W. 共2004兲. Transportation energy databook, United States Department of Energy, Center for Transportation Analysis, Oak Ridge National Laboratory, Oak Ridge, Tenn. Decisioneering Inc. 共2004兲. Crystal ball 7, Denver. Delucchi, M. A. 共1998兲. “Summary of theory, data, methods, and results.” UCD-ITS-RR-96-3 (1), Institute of Transportation Studies, Univ. of California, Davis, Davis, Calif. Ehlen, M. A. 共1999兲. “Life-cycle costs of fiber-reinforced-polymer bridge decks.” J. Mater. Civ. Eng., 11共3兲, 224–230. Ehlen, M. A. 共2003兲. “BridgeLCC: Life-cycle costing software for the preliminary design of bridges.” National Institute of Standards and Technology, Gaithersburg, Md. Enright, M. P., and Frangopol, D. M. 共1999兲. “Maintenance planning for deteriorating concrete bridges.” J. Struct. Eng., 125共12兲, 1407–1414. Estes, A. C., and Frangopol, D. M. 共2001兲. “Minimum expected costoriented optimal maintenance planning for deteriorating structures: Application to concrete bridge decks.” Reliab. Eng. Syst. Saf., 73, 281–291. Federal Highway Administration 共FHWA兲. 共2002兲. “2002 status of the nation’s highways, bridges, and transit.” Washington, D.C. Federal Highway Administration 共FHWA兲. 共2004a兲. “Bridge life cycle cost analysis research report. Part I: Review of existing bridge life cycle cost analysis and uncertainty modeling software.” Rep. to Federal Highway Administration Bridge Management Information Systems Laboratory, Washington, D.C. Federal Highway Administration 共FHWA兲. 共2004b兲. RealCost: User’s manual, Office of Asset Management, Washington, D.C. Federal Highway Administration 共FHWA兲. 共2005兲. “A summary of highway provisions in SAFETEA-LU.” Washington, D.C. Gramlich, E. M. 共1990兲. A guide to benefit-cost analysis, Prentice-Hall, Englewood Cliffs, N.J. Hawk, H. 共2002兲. “BLCCA.” Life-cycle cost analysis for bridges. N.C.H.R. Program, Transportation Research Board, Washington, D.C. Hellman, K. H., and Heavenrich, R. M. 共2003兲. “Light-duty automotive technology and fuel economy trends: 1975 through 2003.” EPA420R-03-006, United State Environmental Protection Agency, Washington, D.C. International Organization for Standardization 共ISO兲. 共1997兲. “Environmental management—Life cycle assessment—Principles and framework.” ISO 14040, Geneva. Joos, F., and McFarland, M., eds. 共2001兲. “Atmospheric chemistry and greenhouse gases.” Climate change 2001: The scientific basis, Chap. 4, Intergovernmental Panel on Climate Change, New York, 239–288. Kelly, T. 共1998兲. “Crushed cement concrete substitution for construction aggregates—A materials flow analysis.” Circular 117, U.S. Geological Survey, Reston, Va. Kentucky Transportation Center. 共2002兲. “The cost of construction delays and traffic control for life-cycle cost analysis of pavements.” KTC-02-07/SPR197-99 & SPR218-00-1F, Univ. of Kentucky, Lexington, Ky. Keoleian, G. A., et al. 共2005兲. “Life cycle modeling of concrete bridge design: Comparison of engineered cementitious composite link slabs and conventional steel expansion joints.” J. Infrastruct. Syst., 11共1兲, 51–60.

JOURNAL OF INFRASTRUCTURE SYSTEMS © ASCE / SEPTEMBER 2008 / 221

Downloaded 19 Jan 2009 to 141.211.206.116. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright

Lai, A. C. K., and Thatcher, T. L. 共2000兲. “Inhalation transfer factors for air pollution health risk assessment.” J. Air Waste Manage. Assoc., 50共9兲, 1688–1699. Li, V., Fischer, G., Kim, Y., Lepech, M., Qian, S., Weimann, M., and Wang, S. 共2003兲. “Durable link slabs for jointless bridge decks based on strain-hardening cementitious composites.” Project Rep. to the Michigan Department of Transportation, Michigan DOT, Lansing, Mich. Li, V. C. 共2003兲. “On engineered cementitious composites 共ECC兲: A review of the material and its applications.” J. Adv. Concr. Technol., 1共3兲, 215–230. Matthews, H. S., and Lave, L. B. 共2000兲. “Applications of environmental valuation for determining externality costs.” Environ. Sci. Technol., 24共8兲, 1390–1395. Michigan Department of State Police. 共1994–2001兲. “Michigan traffic crash facts—Statewide.” Office of Highway Safety Planning, Lansing, Mich. Michigan Department of Transportation. 共2002兲. “Michigan transportation facts and figures 2001. Highways.” Lansing, Mich. Office of Management and Budget. 共2005兲. “Discount rates for costeffectiveness, lease, purchase and related analyses. Appendix C to guidelines and discount rates for benefit-cost analysis of federal programs.” Office of Management and Budget, Washington, D.C. Ozbay, K., Jawad, D., Parker, N. A., and Hussain, S. 共2003兲. “Life cycle cost analysis: State-of-the-practice vs state-of-the-art.” Proc., 83rd Annual Meeting of the Transportation Research Board, Transportation Research Board, Washington, D.C. Palisade Corporation. 共2000兲. “@RISK: Risk analysis and simulation add-in for Microsoft Excel, professional edition.” Palisade Corporation, Newfield, N.Y. The Research Unit for Sustainability and Global Change Universität

Hamburg. 共1999兲. “FUND: The climate framework for uncertainty, negotiation and distribution.” Vol. 1.6, Univ. of Hamburg, Hamburg, Germany. Tol, R. S. J. 共1999兲. “The marginal costs of greenhouse gas emissions.” Energy, 20共1兲, 61–81. Tol, R. S. J. 共2005兲. “The marginal damage costs of carbon dioxide emission: An assessment of the uncertainties.” Energy Policy, 33共16兲, 2064–2074. Tol, R. S. J., Heintz, R. J., and Lammers, P. E. A. 共2003兲. “Methane emission reduction: An application of FUND.” Clim. Change, 57共1–2兲, 71–98. U.S. Environmental Protection Agency 共EPA兲. 共2000兲. “NONROAD.” Draft, Ann Arbor, Mich. U.S. Environmental Protection Agency 共EPA兲. 共2002兲. “MOBILE 6.” Ann Arbor, Mich. van Noortwijk, J. M., and Frangopol, D. M. 共2004兲. “Two probabilistic life-cycle maintenance models for deteriorating civil infrastructures.” Probab. Eng. Mech., 19, 345–359. Walls, J., III, and Smith, M. R. 共1998兲. “Life-cycle cost analysis in pavement design—Interim technical bulletin.” FHWA-SA-98-079, U.S. Department of Transportation, Federal Highway Administration, Washington, D.C. Weitzman, M. L. 共1998兲. “Why the far-distant future should be discounted at its lowest possible rate.” J. Envir. Econom. Manage., 36共3兲, 201–208. Weitzman, M. L. 共2001兲. “Gamma discounting.” Am. Econ. Rev., 91共1兲, 260–271. Yanev, B., and Chen, X. 共1993兲. “Life-cycle performance of New York City bridges.” Transportation Research Record. 1389, Transportation Research Board, Washington, D.C., 17–24.

222 / JOURNAL OF INFRASTRUCTURE SYSTEMS © ASCE / SEPTEMBER 2008

Downloaded 19 Jan 2009 to 141.211.206.116. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright