Life cycle climate impacts of the US concrete pavement network

Resources, Conservation and Recycling 72 (2013) 76–83

Contents lists available at SciVerse ScienceDirect

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

Life cycle climate impacts of the US concrete pavement network Alexander Loijos ∗ , Nicholas Santero 1 , John Ochsendorf Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, MA 02139, USA

a r t i c l e

i n f o

Article history: Received 17 August 2012 Received in revised form 22 December 2012 Accepted 27 December 2012 Keywords: Life cycle assessment (LCA) Pavements Concrete Greenhouse gases (GHGs) Global warming potential (GWP) Urban infrastructure Road network

a b s t r a c t Life cycle assessment (LCA) offers a comprehensive approach to evaluate and improve the environmental impacts of pavements. First, a general pavement LCA methodology is created that describes the concepts necessary to conduct a comprehensive pavement LCA. Second, the methodology is applied to the life cycle of concrete pavements to quantify current emissions across the road network. System boundaries are drawn to include all phases of the pavement life cycle – materials production, construction, use, maintenance, and end of life. Greenhouse gas emissions are quantified for twelve functional units, which evaluate average conditions for each major roadway classification in the United States. The results present the relative contribution of each component in the life cycle, the annual emissions occurring during the 40-year analysis period, and the sensitivity of these results to model parameters. It is found for all roads that the majority of emissions occur in year one – from cradle-to-gate materials production, and pavement construction – primarily due to cement production. The results are most sensitive to traffic volume, and then to parameters affecting the cement production. Based on emissions and their sensitivity, the LCA results suggest three broad reduction approaches: reducing embodied emissions, reducing use phase emissions, and reducing end-of-life emissions. © 2013 Elsevier B.V. All rights reserved.

1. Introduction The construction, operation, and maintenance of the United States roadway system are responsible for substantial energy and resource consumption. The three primary types of pavements are asphalt, concrete, and composite pavements, which comprise approximately 4.2 million kilometers of paved public roads in the United States (Federal Highway Administration, 2008). In addition to the need for continually maintaining public roads, this network has been growing each decade, requiring substantial investment for maintenance and new construction. This vast network has significant environmental and economic impacts for the nation and the planet. The cumulative environmental impact of the road network is unknown, though significant greenhouse gases are released during the construction and operation of pavements. Annually, 320 million metric tons of raw materials go into the construction, rehabilitation, and maintenance of this system on average (Holtz and Eighmy, 2000). The current system of paved roads in the United States handles a volume of traffic on the order of 13 billion vehicle kilometers per day (Bureau of Transportation Statistics, 2008). Road transport contributed the most greenhouse gases (GHGs) of any transport

∗ Corresponding author. Current address: 4001 EastSide Calpella Road, Ukiah, CA 95482, USA. Tel.: +1 415 971 1666. E-mail addresses: [email protected], [email protected] (A. Loijos). 1 Currently address: PE International, 344 Boylston Street, Boston, MA 02216, USA. 0921-3449/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.resconrec.2012.12.014

mode in 2007, accounting for 83% of emissions from the transportation sector and 27% of all emissions in the U.S. (Environmental Protection Agency, 2009). Due to the high environmental and economic impact of pavements, there is growing interest in rigorously quantifying the life-cycle performance of pavements. From 1996 to 2010, an average of at least one study has been published per year that applies the life cycle assessment methodology to pavements (examples: Athena Institute, 2006; Stripple, 2001; Häkkinen and Mäkelä, 1996). Several “green” rating systems are also currently under development for pavements, such as those described in Greenroads and FHWA, following the success of the building green rating system Leadership in Energy and Environmental Design, known as LEED (The Greenroads Foundation, 2011; Federal Highway Administration, 2011a; U.S. Green Building Council, 2011). Improving the sustainability of pavements requires a better understanding of how this infrastructure impacts the natural environment. This research uses life cycle assessment (LCA) to investigate the pavement life cycle, emphasizing the methods and impacts associated with new and reconstructed concrete pavements, which comprise approximately 12% of US paved roads (Federal Highway Administration, 2008). GHG emissions are quantified using global warming potential (GWP) characterization factors. Each relevant phase and component of the pavement life cycle is evaluated in order to comprehensively quantify current emissions and identify opportunities for emission reductions. To effectively characterize a wide breadth of pavements, twelve related, but independent, road designs are analyzed. These designs

A. Loijos et al. / Resources, Conservation and Recycling 72 (2013) 76–83

represent each roadway classification by the Federal Highway Administration (FHWA), ranging from rural local roads to urban interstates. This allows for an evaluation of multiple roadway functions and enables the capacity to estimate impacts across the entire U.S. pavement network. The specific objectives of this article are as follows: 1. Demonstrate a comprehensive methodology that puts forth good-practice concepts for conducting a pavement LCA; 2. Quantify life-cycle GHG emissions of concrete pavements in order to capture the contribution for (a) a range of different roadway classes, (b) each relevant life-cycle component, (c) each year during the 40-year analysis period, and (d) the national GHG emissions of all new concrete pavements that are constructed each year; 3. Use the contribution of the different life-cycle components, and the sensitivity of the overall GHG emissions to various parameters, to demonstrate opportunities for life-cycle emissions reductions. 2. Methodology 2.1. Background and approach This research follows the LCA approach formalized and supported by the International Organization for Standardization (ISO) 14040 series (International Organization for Standardization, 2006a,b). The ISO guidelines describe a generalized approach for LCA, but leave the development of product-specific frameworks in the hands of stakeholders in their respective fields. Mapping the life cycle, developing functional units, drawing systems boundaries, and mining data are left to the discretion and challenge of individual practitioners. A recent review of the pavement LCA literature found that there are still notable framework gaps and inconsistencies amongst 12 existing pavement LCA studies, including issues with the functional units, system boundaries, goals, scopes, and data (Masanet et al., 2010). Of these 12 studies, not a single study included all five of the primary life cycle phases evaluated, and half of them excluded more than one of the following primary phases: materials production, construction, use, maintenance and rehabilitation, and end-of-life. The present study aims to broadly include all five life cycle phases, as well as comprehensively include as many of the components of these phases as possible. The goal and scope of the pavement LCA play an important role in determining proper functional unit selection, system boundaries, and data sources. The European Commission’s International Reference Life Cycle Data System handbooks recognize that needs differ between pavement LCAs, distinguishing between microlevel decisions (such as a comparison between two products) and macro-level decisions (such as large scale policy decision support) (European Commission, 2010). For instance, a project-level comparative assessment may select a single functional unit for ease of comparison, and draw system boundaries that exclude lighting, carbonation, or other components that are assumed to be equal amongst competing alternatives. Conversely, a policy-level assessment such as the present study may subdivide the pavement network into representative classes for functional units, and rely on national average data. Because the supporting science is continually uncovering new knowledge relating pavements to environmental impact, it is expected that boundaries and data sources will be adjusted as necessary to reflect the current state of the science. The specific functional units, system boundary, and data sources used for this research are discussed in the following Sections 2.2 and 2.3.

77

The methodology used to conduct a pavement LCA is itself a valuable contribution, regardless of the numerical results or conclusions. Documenting assumptions, disclosing data sources, and clearly defining goals and scopes serve to establish the framework, or individual methodology, that is used for a particular study. Häkkinen and Mäkelä (1996), Stripple (2001), and Athena Institute (2006) are examples of pavement LCAs that provide reasonably transparent methodologies, documenting elements that enable proper review, validation and reproducibility of modeling procedures and results. While the boundary decisions, functional units, and other study-specific decisions are subject to critique, transparent methodologies allow for the audience to understand the rationale behind the authors’ decisions, leading to more robust conclusions and reproducible results. Table S1 describes the basic transparency requirements that should be part of a pavement life cycle analysis. This research prioritizes transparency by providing descriptions, values, and rationales for each of the elements listed in Table S1 as they relate to the project goals and scope. 2.2. Scope This article analyzes the GHG emissions of new and reconstructed concrete pavements. The following subsections describe the functional unit, the system boundary, and the impact assessment method used to meet the objectives presented in this section. The functional unit is a reference unit that allows for consistent comparisons between different products and comparison of results across different studies. This research adopts multiple functional units in order to characterize the various classifications of concrete pavement roadways in the United States. Representative structures for each FHWA roadway classification are developed and analyzed over one centerline-kilometer for the respective traffic loadings presented in Table 1. Centerline lengths are used so that each of the twelve classifications (six rural and six urban) can be evaluated based on their cross-sectional geometric and material design. The number of lanes, average passenger and truck traffic, and lane widths for rigid pavements is taken from the Highway Statistics 2008 publication to define the units of service or functional units (Federal Highway Administration, 2008). Based on this data, structures are derived using American Association of State Highway Officials (AASHTO) pavement design methods (AASHTO, 1993, 2004). Example rural and urban structures are depicted in Fig. S1. Summaries of all analyzed roadways are found in Table 1, assumed parameters based on FHWA and AASHTO design methods are found in Table S2, and corresponding material masses are found in Table S5. A step-by-step breakdown of the FHWA data and derivation of the AASHTO ‘93 design procedure is found in Loijos (2011). The concrete mix uses 335 kg/m3 blended cement (90% portland cement, 10% coal fly ash), a water-to-cement ratio of 0.45, and crushed aggregate for the remaining material, with a value of 2350 kg/m3 for the material density (American Concrete Pavement Association, 2011). The fly ash substitution value is based on an estimated national average utilization of fly ash in concrete in 2008 (American Coal Ash Association, 2008; United States Geological Survey, 2008). It is important to note that the 10% fly ash is not necessarily a typical replacement rate for concrete mixes due to potentially poor resistance to alkali silica reaction (ASR). While the average value is used here to meet the stated objectives (i.e., represent gross national averages), a project-specific concrete LCA should acknowledge that a minimum 15% replacement rate is probably more realistic and corresponds to many state DOTs thresholds for the material density (American Concrete Pavement Association, 2011). Moreover, supplementary cementitious materials (SCMs) other than fly ash are also commonly used in concrete mixes, but are not considered here.

78

A. Loijos et al. / Resources, Conservation and Recycling 72 (2013) 76–83

Table 1 Analyzed pavement designs. Rural roadways

Function definition AADTa AADTTb Total lanes Lane width (m)

Principal arterial

Minor arterial

Major collector

Minor collector

Local

Interstate

Freeway

Principal arterial

Minor arterial

Collector

Local

22,074 4415 4c 3.7

6414 706 2 3.7

3084 308 2 3.7

1219 85 2 3.4

574 40 2 3.4

177 12 2 2.7

78,789 6303 6c 3.7

53,809 2152 4c 3.7

19,631 785 4 3.7

9729 389 2 3.7

4221 169 2 3.4

980 39 2 2.7

2.4 203 152 32

2.4 191 152 –

1.8 152 152 –

1.5 127g 0 –

0.6 102g 0 –

3.0 305 152 38

1.2/3.0d 279 152 38

2.4e 216 152 32

2.4e 178 152 32

2.4e,f 165 0 –

2.1e,f 127g 0 –

Corresponding AASHTO design Shoulder width (m) 1.2/3.0d Slab thickness (mm) 292 Base thickness (mm)h 152 i Dowel gauge (mm) 38 a b c d e f g h i

Urban roadways

Interstate

AADT: annual average daily traffic (two way). AADTT: annual average daily truck traffic (two way). Two carriageways with separating median. Inner/outer shoulder width. Includes aggregate in concrete, base, and foreslope elements. Minimum foreslope of 4 H:1 V is used. Urban curb and gutter design with no foreslope. Shoulders are parking lanes. These pavements are thinner than some states allow. However, the AASHTO ‘93 design procedure was still followed to remain consistent. A conventional granular base of 152 mm (6 in.) is assumed for roads with more than 1 million ESALs. Dowel length is 0.46 m, lateral spacing is 0.23 m, steel density is 7850 kg/m3 , and concrete slab length is 4.57 m.

The analysis period begins at initial pavement construction and continues through 40 years of operation, which includes two rehabilitation activities (at years 20 and 30), and ends at recycling and disposal at the end of life. Concrete rehabilitation includes 4% slab replacement and complete surface grinding. While concrete pavements often last more than 40 years, the end of life is included in order to evaluate preferred waste management practices. The analysis period and rehabilitation schedules and activities are based on the most common responses in surveys conducted by FHWA and the Mississippi Department of Transportation (DOT) of other U.S. state agencies’ life cycle cost analysis procedures, which reflect their experience on how long pavements can predictably last (Rangaraju et al., 2008; Mississippi Department of Transportation, 2007). While a 40-year analysis period is appropriate for this study, the results and conclusions for comparative LCAs (such as those comparing asphalt and concrete alternatives) are more sensitive to the analysis period. These functional units are meant to represent average concrete structures for each of the FHWA roadway classifications. In reality, concrete pavement designs will vary significantly from one pavement to the next, even if the basic structural inputs are the same. Regional climate, local design practices, budget, service life, material availability, and other factors play a role in the design process. There is also significant variation within each roadway classification, making it difficult to adopt a single representative structure. For instance, urban interstates routinely support between 30,000 and 130,000 vehicles per day, but the weighted average (79,000) is used in this analysis (Federal Highway Administration, 2008). Fig. 1 presents the phases and components included within the system boundaries for this study. Each phase of the life cycle is represented: materials production, construction, use, maintenance, and end of life. The phases are broken down into multiple components for each life-cycle phase. Distinguishing a roadway LCA from a pavement LCA necessitates allocating certain components based on their differential impact, relative to some baseline. For example, vehicle fuel consumption is only allocated to a pavement based on roughness increases over the life cycle, as the majority of fuel consumption is attributable to the vehicle life cycle, and only a marginal amount is caused by the pavement. Thus, the pavement roughness at initial construction is taken to be the baseline roughness, and GHG emissions from

fuel consumption are calculated based on the progressive increase from that initial roughness. Fuel consumption due to the structural deflection of the pavement is excluded from this section due to the assumption that deflections did not change over the life cycle, as well as limitations with the accuracy of existing methods (Santero et al., 2011). This differential approach ensures that impacts are only allocated to the pavement that are caused by the pavement itself. This is applied to pavement albedo as well, which is light in color (˛ = 0.40) at initial construction and each time it is newly grinded, but is only attributed GWP as it darkens to a minimum albedo of ˛ = 0.25 by year 20 at an assumed constant rate (Masanet et al., 2010). Lighting requirements remain constant during the analysis period, making the baseline lighting demand and final lighting demand equal, and their associated emissions are assumed to be zero for the baseline scenario, but is included so as to evaluate the effect of pavement albedo on reducing lighting needs (in Section 3.4). Similarly, emissions due to normal traffic are not included, but the traffic delays due to construction and rehabilitation activities are attributed to the pavement. It must be noted that comparative assessments may need to establish different baselines in order to capture fuel consumption, albedo, and lighting differences between pavement types. As such, the approach and results are not sufficient for comparative purposes, such as comparing to alternative materials like asphalt pavements. Also not included in the scope of this analysis are the following elements deemed insignificant (i.e. less than 1% for most roads) to the life cycle: capital goods production (excavation and paving machinery, production plant equipment, oil refinery infrastructure, etc.), production of roadway lighting hardware, road paint production and application, and joint sealant. GHG emissions are normalized using their GWP, as measured in carbon dioxide equivalents (CO2 e). GWP is a common characterization method to evaluate the impacts of GHG emissions on climate change. This research uses the Center for Environmental Studies of the University of Leiden’s “CML” factors for GWP characterization, updated for the November 2009 report by the Intergovernmental Panel on Climate Change (CML, 2011). While climate change is a preeminent environmental issue, it is important to acknowledge that other impact categories (e.g., human health impact, water consumption, energy consumption) need to be considered for a comprehensive sustainability framework for pavements.

A. Loijos et al. / Resources, Conservation and Recycling 72 (2013) 76–83

79

Fig. 1. Concrete pavement life-cycle phases and their components included in system boundary.

2.3. LCA model The life cycle inventory (LCI) data for the analysis is taken from published literature and LCI databases. Data was chosen from a variety of sources in order to meet ISO data quality requirements for the goal and scope, giving utmost priority to: temporal, geographical, and technological representativeness, precision, completeness, consistency, and having undergone critical review (International Organization for Standardization, 2006b). The supporting information contains a summary of key values used in this research, including GWP emission factors (in Table S8), transportation modes and distances (in Table S9), and use phase assumptions and calculations (in Table S10). A more detailed description and derivation of the data used in this research is found in Loijos (2011). It is important to note that there is inherent uncertainty and variability in the values utilized. For example, GHG emissions from cement production tend to vary significantly based on the type of kiln and energy source; the cement CO2 e factor used here represents average U.S. emissions based on a 2006 study (Marceau et al., 2006). The model parameters are evaluated in a sensitivity analysis, which tests the influence of the various input parameters within a range of possible alternative values. Parameterization within the utilized GaBi life cycle assessment software allows for estimation of data variability and includes a sensitivity analysis tool that enables understanding of the sensitivity of the results to variation of the input parameters (GaBi, 2011). Since this variation is expressed as a percentage in GaBi, a limitation is that this assumes all parameters vary equally in the positive direction as in the negative direction, as well as being uniformly distributed. In actuality, the distribution of values for a given parameter may not be symmetric. Despite this limitation, a coarse sensitivity analysis still offers value and is provided in Section 3. 3. Results The results include an estimation of the life-cycle GWP for each of the twelve FHWA roadway classifications, broken down by the

impact associated with each life-cycle component. A time series of emissions is shown for one structure in order to show the contribution for each year during the 40-year analysis period. Each representative structure is extrapolated across all lane-kilometers in its respective roadway classification in order to estimate a nationwide footprint for new concrete pavements on an annual basis. Lastly, a sensitivity analysis is conducted to evaluate the variation in these results due to various elements in the concrete pavement life cycle. 3.1. Impacts for the roadway classifications The quantities of life-cycle GHG emissions, expressed in GWP, for twelve different classifications of new concrete pavements are presented in Fig. 2. For clarity purposes, components with smaller contributions are grouped together, which are disaggregated in Table S6. These results are specific to new concrete construction, although the data and results can be extended to reconstruction with only minor adjustments. These results are representative for typical scenarios within each roadway classification, although the results for a given project may differ from those of the broader class. This variability has largely been captured and is presented in the sensitivity analysis below. Total life-cycle GWP ranges from 440 Mg CO2 e/km on the rural local road to 6670 Mg CO2 e/km on the urban interstate. This is largely due to the fact that interstates are more massive structures, both in terms of the thickness of the concrete slab, as well as the width across the road. For example, the representative rural local road is 102 mm thick, with two 3.4 m wide lanes, and two 0.6 m wide shoulders, whereas the urban interstate has 305 mm of concrete, with three 3.6 m wide lanes in each direction, and two inner and two outer 3.0 m wide shoulders. Traffic is the other primary driver of the variation across structures, which affects the fuel consumed due to roughness. When looking at the life-cycle GWP contributions for each road type, the first notable feature is that cement production emissions are the largest contribution for every one of the twelve structures.

80

A. Loijos et al. / Resources, Conservation and Recycling 72 (2013) 76–83 ---------- Rural Roadways ----------

----------- Urban Roadways ----------

7000

GWP/km (Mg CO2e)

6000 5000 Components contribung less than 3% each

4000 3000

Components contribung between 3-8% each

2000

Albedo 1000 Fuel Consumed from Roughness

0

Cement Producon

Fig. 2. Life-cycle GWP per km of new concrete pavements for twelve roadway classifications. Components contributing less than 3% each include (in decreasing order): traffic delay, cement transport, aggregate mining, pavement demolition, concrete mix transport, onsite construction equipment, diamond grinding, water production, concrete mixing, steel transport, fly ash, pavement lighting, and carbonation. Those contributing 3–8% include aggregate transport, end-of-life disposal, end-oflife transport, and steel production (for those functional units that have steel).

The contribution of cement production ranges from 43% (for the urban interstate) to 56% (for the rural local road) of the total lifecycle emissions. The second largest contribution is “fuel consumed from roughness” in every case except for the rural minor collector, rural local road, and the urban local road, where end of life disposal is the second largest contribution. For numerical results for all roadway classes, see Table S6. 3.2. Time series emissions While the majority of life-cycle GHG emissions are due to cradleto-gate materials production and pavement construction at the beginning of the analysis period, there are several effects occurring continuously throughout the use phase (albedo, carbonation, and fuel consumption due to roughness) and several one-time events after production (rehabilitation, traffic delay, and end of life demolition, transport, recycling and disposal). This is shown for the case of urban interstates in Fig. 3, which plots emissions over the analysis period. The initial emissions in year one – from cradle-to-gate materials production and pavement construction – dominate the time series of emissions, at 55% of the total for urban interstates, which grows as a proportion of the total as the roadway class gets smaller. The

Cradle-to-gate materials production and pavement construction

3700

End of life demolition, recycling, and disposal

GWP/km (Mg CO2e)

3600 Percent of total

700 600

Carbonation (decreasing rate), fuel consumption from roughness, and albedo (increasing rates)

45%

500

3.3. Extrapolation to the entire network The results representing each of twelve roadway classifications are extrapolated based on the number of average lane-kilometers of annual construction for that classification and into the future for a period of 50 years in order to obtain a national estimate for the annual GHG emissions due to new concrete pavement construction. Table S7 shows the number of kilometers for each of the five primary classifications, as well as the growth rate over time sampled from the most recent ten year period (adapted from FHWA (2008)). During this period, rural roads constitute approximately 43% of the entire network by length, and urban roads 57%. It is assumed the concrete network grew at the same rate as the overall network, as no disaggregated growth data is available. Adjusting for the extent of each roadway classification, as well as adjustments for lane counts and lane widths by using mean rather than mode values (see Table S2), the results in Fig. S2 represent the GHG emissions of all new concrete roads built on average per year in the U.S. These emissions total 3.1 Tg CO2 e per year, or approximately 0.05% of total U.S. annual GHG emissions (Environmental Protection Agency, 2009). The rural network contributes 1.5 Tg CO2 e per year (48% of total), and the urban network contributes 1.6 Tg CO2 e per year (52% of total). Urban interstates contribute 0.5 Tg CO2 e per year (17% of total). Collector roads on both rural and urban networks have the smallest contribution to the national emissions due to the fact that growth is negligible (0.03% per year) on this roadway classification. While this is a small proportion of overall national GHG emissions, this approximation does not account for a large proportion of the pavement network (namely asphalt pavements, composite pavements, and asphalt rehabilitation on concrete pavements), nor the emissions from vehicle and goods damage, nor the potential impact of consequential effects of construction, such as induced road transport traffic. It is expected that the inclusion of these networks and activities would significantly increase the overall footprint.

55% Diamond grinding and traffic delay

400 300 200 100 0

second largest one-time contribution is from end-of-life demolition, transport, recycling and disposal, contributing 9%. This largely comes from transport and landfill emissions, which are highly variable depending on waste management practices. While much of the concrete and aggregate is transported to an aggregate stock yard, as assumed here, this assumption is high in cases where the pavement demolition coincides with reconstruction and the materials are reused on site. Additionally, landfilled concrete waste is normally buried in a carbon dioxide free environment, but improved waste management practices can take advantage of carbonation at the end of life to further reduce life-cycle GWP. Carbonation is more active initially after new pavement construction and each rehabilitation activity, and then the effect slows, per Fick’s law of diffusion; and roughness related emissions increase up until pavement grinding occurs. GHG emissions from grinding and traffic delay at each rehabilitation event contributes 1% of life-cycle GWP for urban interstates.

2008

2018

2028 Year

2038

Fig. 3. Life-cycle GWP per km per year over the 40-year analysis period for an urban interstate built in 2008.

3.4. Sensitivity analysis In order to provide a more robust assessment, the variability of input data was identified during the life-cycle inventory procedure, and incorporated into the model such that a sensitivity analysis can capture the uncertainty of the results. The nominal value and sensitivity variation for each parameter are presented in Table S11. The sensitivity variation approximates one standard deviation from the nominal value ( ± ), such that approximately 68% of available data points are within this interval. The sensitivity variation for traffic is asymmetric, while all other parameters are modeled to

A. Loijos et al. / Resources, Conservation and Recycling 72 (2013) 76–83

81

Fig. 4. Sensitivity of overall life-cycle GWP to the ten most influential parameters on six rural (left) and six urban (right) roadway classifications (parameters and their variation in Table S11).

vary uniformly in the positive and negative direction. In actuality, the distribution of values for a given parameter may not be symmetric. Only those parameters which exceed an effect of 2% change on a majority of the roadway classifications are included in the sensitivity results. The results of the sensitivity analysis are shown in Fig. 4. One of the variables that the results are most sensitive to is the traffic volume that the pavement structure supports, reported as AADT. Instead of assuming an average traffic volume and associated concrete pavement structure for each roadway classification, the pavement thickness, number of lanes, width of shoulders, and fuel consumption due to roughness, are varied according to traffic volume distributions within each classification. The variability of GWP for each roadway classification in Fig. S3 spans between the first and fifth sextiles of AADT, which approximates one standard deviation from the mean, as described by Loijos (2011). Some of the roadway classifications show large variability, including the rural interstate, urban interstate, urban other freeway/expressway, urban other principal arterial, and urban minor arterial. This is partially because of large traffic variability on these networks, but also because the number of lanes in the representative structures changes, accounting for the asymmetry in many of the error bars. The results become more sensitive to certain parameters moving from smaller to larger roads (such as regional climate variability of the pavement’s international roughness index (IRI) over time), while other parameters are more important on the smaller roads (e.g., outer shoulder width, carbonation rate, pavement albedo). This is primarily due to the fact that the parameters to which the results are more sensitive correspond to those life-cycle components that contribute a larger proportion of the overall emissions. In general, smaller roads are sensitive to parameters which relate to the materials production, since this has a larger overall contribution to the total. They are also sensitive to use phase components that are driven by surface area, like carbonation and pavement albedo. Larger roads are sensitive to traffic-related parameters, since the roughness and traffic delay components comprise a larger proportion of overall emissions.

4. Discussion This section includes two primary parts: (1) a qualitative discussion of the shortcomings of the above approach, as well as ways to address them in the future, and (2) a summary of the most relevant and interesting key findings that are drawn as conclusions. 4.1. Limitations Defining the goal of the LCA is an important framing step that enables the research to focus on a clear question, but at the same time constrains the research to a narrow domain. The first limiting step is the focus on greenhouse gases. The research field will benefit from a similar quantification of other impact categories, such as land use, water consumption, energy consumption, ecotoxicity, human health impacts, natural resource consumption and social impacts. Limiting the present LCA to pavements alone rather than roadway transportation removes the need to account for potentially large systemic impacts such as induced traffic due to road expansion, increased urban sprawl, and increased reliance on goods transport by vehicle rather than other more efficient means. Also, pavements have an important effect on vehicle fuel consumption, which is a complex interaction that is not fully understood by the science at present (University of California Pavement Research Center, 2010). For this reason, structural deflection, and the pavement’s effect on air temperature and vehicle drag, are not included, but may be significant. This article models a multi-tiered supply chain for a complex pavement network that has within it many different sources of regional, temporal, and technological variability. The representative structures that are here analyzed incorporate many assumptions in an attempt to derive the “average” structure for each roadway classification. The need to represent an entire class of roadways using a single average structure is a necessary, but significant, simplification. For instance, single values were selected for all structures for the design life, concrete mix design, flexural strength, drainage coefficient, and elastic modulus of concrete, as well as for

82

A. Loijos et al. / Resources, Conservation and Recycling 72 (2013) 76–83

other life cycle concerns such as traffic delay management, pavement albedo, and AADT. For pavements that have parameters which deviate from this average, project-specific analyses can lead to significantly different results. 4.2. Key findings The key contribution of this LCA is the quantification of greenhouse gases in the life cycle of concrete pavements in the United States. This is done for six urban and six rural roadway classifications, all on a per-kilometer basis over a 40-year analysis period, and on a U.S. network-wide basis for each year, extrapolating 50 years into the future. Emissions are quantified so as to accomplish the following: evaluate different roadway classifications; show the contribution of different components of the life cycle; show the relative emissions from each year of the analysis period; and quantify national emissions for all new concrete pavements constructed each year. The sensitivity of these results to the various model parameters is also evaluated, which suggest opportunities for emissions reduction. The majority of emissions occur during materials production, transportation, and end of life (excluding entire use phase), constituting between 64% and 80% on all roads. Cement production has the largest single life-cycle GHG contribution on all roads: from 43% on urban interstates to 56% on rural local roads. Between 51% (urban interstates) and 63% (rural other principal arterials) of the total emissions occur in year one – from cradle-togate materials production, and pavement construction. The second largest one-time contribution occurs at end of life demolition, transport, recycling and disposal, which contributes between 6% and 13% of total GHG emissions. GHG emissions from grinding and traffic delay at each rehabilitation event (at years 20 and 30) contribute between 1% and 2% of life-cycle GWP. The life-cycle GHG emissions for all new concrete pavements constructed in the U.S. are approximately 3.1 Tg CO2 e per year, or about 0.05% of total national emissions in 2009. Of the model parameters analyzed for sensitivity, the results are most sensitive to traffic volume, varying the results by up to 60%. The results are also particularly sensitive to parameters affecting the cement emissions, such as shoulder width, lane width, and cement emission factor, as well as aggregate transportation distance, and use phase parameters such as IRI at year 20 and pavement albedo. Based on the life-cycle contributions of GHG emissions, the LCA results suggest three broad reduction approaches: reducing embodied emissions, reducing use phase emissions, and reducing emissions at end of life. The MEPDG design procedure has been identified as an opportunity for reducing embodied GHG emissions by preventing overdesign. Another effective strategy involves mix design optimization, by replacing cement with supplementary cementitious materials such as coal fly ash, blast-furnace slag, and silica fume (Tikalsky et al., 2011). Use phase strategies that can reduce emissions include reducing pavement roughness, which is currently the second largest life-cycle contribution on most roadway classifications. By increasing the pavement albedo, more of the sun’s incoming light is reflected back out of the atmosphere, presenting a geoengineering strategy that reduces GWP (Masanet et al., 2010; University of California Pavement Research Center, 2010). Lastly, concrete’s ability to directly absorb carbon dioxide through carbonation is an opportunity once concrete is crushed at the end of life, and either stockpiled or recycled as base material (Dodoo et al., 2009). Reducing the traffic volume on US roads would benefit each of these broad approaches, and additionally reduce the expansion of the network. By quantifying the global warming potential of concrete pavements in the United States using life cycle assessment, this

article demonstrates the variation in emissions due to different roadway types, quantifies the relevant emissions for each phase of the pavement life, and identifies opportunities for reductions. Acknowledgments This research was carried out for the Concrete Sustainability Hub at the Massachusetts Institute of Technology and was sponsored by the Portland Cement Association (PCA) and the Ready Mixed Concrete Foundation (RMC) Research and Education Foundation. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.resconrec.2012.12.014. References AASHTO. AASHTO guide for the design of pavement structures. Washington, DC, 1993. AASHTO. A policy on geometric design of highways and streets. 5th ed. Washington, DC: Association of State Highway and Transportation Officials; 2004. American Coal Ash Association. 2008 Coal combustion product (CCP) production & use survey report. Aurora, CO, 2009. American Concrete Pavement Association. ACPA application library: agency practices explorer [online], http://1734298.sites.myregisteredsite.com/apps/; 2011 [accessed April 2011]. Athena Institute. A life cycle perspective on concrete and asphalt roadways: embodied primary energy and global warming potential. Ottawa, Ontario, 2006. Bureau of Transportation Statistics.Moore W, editor. Transportation statistics annual report. Washington, DC: U.S. Department of Transportation; 2008. CML. CML-IA characterisation factors (updated November 2010) [online], University of Leiden, Institute of Environmental Sciences, http://cml.leiden. edu/software/data-cmlia.html; 2011 [accessed July 2011]. Dodoo A, Gustavsson L, Sathre R. Carbon implications of end-of-life management of building materials. Resource Conservation and Recycling 2009;53(5): 286–376. Environmental Protection Agency. 2009 U.S. greenhouse gas inventory report [online], Washington, DC, http://www.epa.gov/climatechange/emissions/ downloads09/GHG2007entire report-508.pdf; 2009 [accessed February 2010]. European Commission. International reference life cycle data system (ILCD) handbook – general guide for life cycle assessment – detailed guidance, EUR 24708 EN. Luxembourg: Publications Office of the European Union; 2010. Federal Highway Administration. Highway statistics [online], Washington, DC, http://www.fhwa.dot.gov/policyinformation/statistics/2008/; 2008 [accessed July 2011]. Federal Highway Administration. INVEST: infrastructure voluntary evaluation sustainability tool [online], https://www.sustainablehighways.org/; 2011a [accessed July 2011]. GaBi, version 4. Life cycle assessment software. Stuttgart, Germany: PE International; 2011. Häkkinen T, Mäkelä K. Environmental adaption of concrete: environmental impact of concrete and asphalt pavements. Espoo, Finland: VTT Technical Research Centre of Finland; 1996. Holtz K, Eighmy T. Scanning European advances in the use of recycled materials in highway construction. Public Roads 2000;64(1):34–40. International Organization for Standardization. ISO 14040: Environmental management – Life cycle assessment – Principles and framework. Geneva, Switzerland, 2006a. International Organization for Standardization. ISO 14044: Environmental management – Life cycle assessment – Requirements and guidelines. Geneva, Switzerland, 2006b. Loijos A. Life cycle assessment of concrete pavements: Impacts and opportunities, S.M. Thesis, Cambridge, MA: Massachusetts Institute of Technology, 2011. Marceau M, Nisbet M, VanGeem M. Life cycle inventory of portland cement manufacture. Skokie, IL: Portland Cement Association; 2006. Masanet E, Santero N, Horvath A. Life cycle assessment of pavements: a critical review of existing research. Berkeley, CA: Lawrence Berkeley National Laboratory; 2010. Mississippi Department of Transportation. Life cycle cost analysis for pavement type selection [online], Jackson, MS, http://research.transportation.org/ Pages/LifeCycleCostAnalysisforPavementTypeSelection.aspx; 2007 [accessed April 2011]. Rangaraju PR, Amirkhanian S, Guven Z. Life cycle cost analysis for pavement type selection, FHWA-SC-08-01. Washington, DC: Federal Highway Administration; 2008.

A. Loijos et al. / Resources, Conservation and Recycling 72 (2013) 76–83 Santero N, Loijos A, Akbarian M, Ochsendorf J. Methods, impacts, and opportunities in the concrete pavement life cycle. Cambridge, MA: Massachusetts Institute of Technology; 2011. Stripple H. Life cycle assessment of road: a pilot study for inventory analysis. Gothenburg, Sweden: IVL Swedish Environmental Research Institute; 2001. The Greenroads Foundation. The greenroads rating system [online], http://www.greenroads.us/; 2011 [accessed July 2011]. Tikalsky P, Taylor P, Hanson S, Ghosh P. Development of performance properties of ternary mixtures: laboratory study on concrete, DTFH61-06-H-00011. Ames, IA: National Concrete Pavement Technology Center; 2011.

83

U.S. Green Building Council. Policy brief: energy performance in the LEED rating system [online], http://www.usgbc.org/ShowFile.aspx?DocumentID=8946; 2011 [accessed July 2011]. United States Geological Survey. 2008 minerals yearbook: sand and gravel, construction [online], Reston, VA, http://minerals.usgs.gov/minerals/pubs/ commodity/sand & gravel construction/; 2008 [accessed July 2011]. University of California Pavement Research Center. Pavement life cycle assessment workshop [online], Institute of Transportation Studies Berkeley, http://www.ucprc.ucdavis.edu/P-LCA/; 2010 [accessed May 2011].