Removing shadows from consequential LCA ... - ACS Publications

3 Jan 2019 - The results show that simplifying pavement CLCA framework through neglecting real-time changes results in notable diversions in the damag...
1 downloads 0 Views 673KB Size
Subscriber access provided by University of Kansas Libraries

Policy Analysis

Removing shadows from consequential LCA through a time-dependent modeling approach: policy-making in the road pavement sector Hessam Azarijafari, Ammar Yahia, and Ben Amor Environ. Sci. Technol., Just Accepted Manuscript • DOI: 10.1021/acs.est.8b02865 • Publication Date (Web): 03 Jan 2019 Downloaded from http://pubs.acs.org on January 4, 2019

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 27

Environmental Science & Technology

Removing shadows from consequential LCA through a time-dependent modeling approach: policy-making in the road pavement sector Hessam AzariJafaria,b,* , Ammar Yahiab and Ben Amora,* a

Interdisciplinary Research Laboratory on Sustainable Engineering and Eco-design

(LIRIDE), Department of Civil and Building Engineering, Université de Sherbrooke, Sherbrooke, Quebec J1K 2R1, Canada b

NSERC Research Chair on Development and Use of Fluid Concrete with Adapted

Rheology, Department of Civil and Building Engineering, Université de Sherbrooke, 2500 Blvd. de l’Université, Sherbrooke, Quebec J1K 2R1, Canada

* Corresponding author: Department of Civil and Building Engineering, Université de Sherbrooke, 2500 Blvd. de l’Université, Sherbrooke, Quebec J1K 2R1, Canada, Tel: +1 819-821-8000x65504, Email: [email protected] Tel: 819-821-8000x65974 Email address: [email protected] ACS Paragon Plus Environment

1

Environmental Science & Technology

Page 2 of 27

1

Abstract

2

Lack of dynamic accounting in consequential life cycle assessment (CLCA) can keep policy-makers from

3

having an accurate analysis of emission flows over time. In this study, we propose a dynamic CLCA

4

framework to assess the environmental consequences of pavements. Dynamic changes in the demand vector

5

and technosphere matrices were computed using relevant time horizons of affected supply technologies and

6

incorporating time-dependent parameters. A Monte Carlo simulation was then conducted to propagate the

7

variability, model uncertainty, and parameter uncertainty sources of LCI to the damage results. The results

8

show that simplifying pavement CLCA framework through neglecting real-time changes results in notable

9

diversions in the damage results. The environmental benefits of substituting asphalt with concrete are

10

underestimated by 7, 17, and 77% for climate change, human health and resources categories, respectively.

11

A divergence of 114% was also observed in ecosystem quality when using the static framework. Moreover,

12

the lack of accounting for a temporal profile for GHG emissions using static characterization factors leads to

13

a divergence of the GWP benefits of substituting asphalt with concrete of 473 metric tons CO2eq (105%). The

14

uncertainty results show 41-71% contribution of the variance in the damage categories is caused by the

15

variability sources and is primarily attributed to monthly temperature accounting and service life.

16

Keywords: Consequential life cycle assessment; Uncertainty and variability; Dynamic inventory; Indirect

17

effects; Pavement selection; Environmental policy-making.

18

2 ACS Paragon Plus Environment

Page 3 of 27

Environmental Science & Technology

19

Introduction

20

Pavement infrastructures, as pre-requisites for sustained economic development, play important roles in

21

growth and connecting people in different regions. Governments are working to reduce the environmental

22

impacts of transportation enabled by these pavements. Road transportation is the largest contributor to

23

greenhouse gas (GHG) emissions in various geographical contexts 1. To curb these GHG emissions,

24

governments have adopted fuel efficiency standards and regulations but the subject of pavement effect on the

25

road transportation impacts has not yet been fully considered. The environmental impacts of pavement are

26

not limited to the materials and machinery of construction (see e.g., 2 and 3). Pavement characteristics can

27

induce major changes in fuel consumption of the vehicles 4. Thus, pavement selection can be considered an

28

important part of the strategy to reduce the environmental impacts of road transportation. Beyond road

29

transportation, pavement selection can also affect electricity consumption in buildings, particularly in urban

30

zones, by radiative forcing (RF) on ambient temperatures. This influence on temperature leads to changes in

31

electricity consumption, as climate control systems compensate for heat gain or loss in buildings 5.

32

The temporal distribution of impacts over the pavement life cycle is not fully captured in previous studies.

33

Neglecting changes which occur over time can manifest in the results in two ways:

34

1. The improvement in efficiency of fuel consumption and emission factors are disregarded.

35

2. Considering the long use-phase of pavements, a number of parameters change as a function of time

36

and use. Oversimplification in trying to capture these real-time changes in parameters may result in

37

inaccurate estimation of the environmental impacts during the use phase.

38

Additional fuel consumption by vehicles is linked to increasing pavement surface roughness over the life

39

cycle. Surface roughness after initial construction is considered the base value for roughness. Emissions

40

based on fuel consumption are calculated over time based on a progressive increase from that initial

41

roughness. It is useful to analyze the real-time fuel consumption of vehicles between various pavement

42

alternatives in comparative studies. The RF effect of albedo is generally calculated as the RF difference

3 ACS Paragon Plus Environment

Environmental Science & Technology

Page 4 of 27

43

between the pavement reflectivity and reflectivity of a fully reflective surface 1. Given that all the life cycle

44

phases are considered in a relative format, it is coherent to quantify all the burdens based on the same base

45

case to attain a sensible and valid contribution analysis of the phases in the impact categories.

46

Generally, in consequential life cycle assessment (CLCA) framework, an alternative (ALT) case will replace

47

a base case, namely, a business-as-usual (BAU) scenario. Therefore, the BAU case seems to be a consistent

48

baseline for all time-dependent parameters in the life cycle phases of pavements. Pavement-related changes

49

in energy consumption can have indirect effects on the fuel market and consumer behavior. These effects are

50

can be captured, to some degree, by consequential life cycle assessment (CLCA) 6. Moreover, conducting an

51

LCA using a consequential framework prevents isolation of a product system from the rest of the

52

technosphere within an economic framework 7. CLCA frameworks can consider consequences of decisions

53

through system expansion, which is beyond the defined system boundaries, i.e., in the parts of

54

multifunctional processes that have been allocated or simplified through averaging.

55

Various methods have been proposed to conduct a CLCA. One common assumption in these methods is the

56

consideration of a single time-horizon in affected technologies. In fact, the affected technologies are intended

57

to be valid for long time periods and typically involve long-term suppliers. This assumption is usually

58

justified. When a long-term change is planned and announced well in advance of implementation, it produces

59

only long-term effects, i.e., the effects from installation and production on newly installed capacity 8. This

60

may be the case in the maintenance of infrastructure, for example, where decision-makers can predict the

61

work schedules well before new construction. The other justification for isolating the long-term effects of a

62

change is that accumulation of individual demands result in capacity adjustments as a continuous process.

63

This continuous process is a basis for decisions on the provision of capital investment. This justification for

64

simplifying the change effect may not be applicable to the CLCA of products with long service life, such as

65

in pavements. Short-term changes in demand of intermediate flows in initial phases of the life cycle, e.g.,

66

materials for construction of infrastructures, are of limited duration and will be terminated as soon as

67

interventions in this phase are finished. Therefore, decision-makers may not need to procure a continuous 4 ACS Paragon Plus Environment

Page 5 of 27

Environmental Science & Technology

68

production volume for a regulated shape of demand. Second, in constructing infrastructure, the market

69

conditions for intermediate flows used in initial phases of the life cycle have little influence on capacity

70

adjustments since the production capacity is significant and the capital cycle of infrastructures is long 9. In a

71

review paper, Zamagni et al. addressed this time-constant consideration of the affected technologies in the

72

storyline of a product system as a “shadow” on the CLCA concept 10. In fact, by isolating long-term impacts

73

in CLCA of pavements, the results tend to obscure short-term impacts. In addition, using a single time

74

horizon demonstrates only one perspective of the life cycle assessment (LCA) results possibly introducing

75

inadvertent bias and inconsistency into the outcomes. For example, exclusion of short-term effects may

76

result in other overlooked impacts from the short run product system, although the frontier between short and

77

long-term is still in debate 11. The result of evaluating affected suppliers with different time-horizons is

78

referred to as “complex marginal supplier” and takes into account the evolution to changed state from the

79

baseline 12.

80

Another argument against static analysis, not specific to CLCA but also relating to other LCA frameworks, is

81

that the impacts of emissions are merely the arithmetic sum of all the current and future activities. For

82

example, in calculating global warming potential (GWP), the severity of the emissions depends on the GHG

83

type, quantity and timing of the induced GHGs 13, 14. To assess environmental impacts through dynamic

84

characterization factors (CFs), a dynamic life cycle inventory (DLCI) is required as the dynamic accounting

85

of the environmental impacts can produce significant relative differences across product systems. In addition,

86

the dynamic assessment of impacts, such as GWP, can prompt large absolute differences compared to static

87

GWPs calculated against arbitrary time horizons. Examples of attributional LCA of pavement construction

88

can be found in

89

static 100-year GWP results. Unlike the previous steady-state inventory, the use of a consequential DLCI can

90

facilitate the goal of reaching an incremental accuracy of LCA results by integrating dynamic CFs. In this

91

study, we aim to develop a dynamic consequential life cycle inventory for assessing the potential impacts of

92

pavements substitution. We established a consistent approach to investigate and compare the impacts of a

15, 16

showed that a 66% difference in the GWP impacts was observed between dynamic and

5 ACS Paragon Plus Environment

Environmental Science & Technology

Page 6 of 27

93

decision by investigating market reactions to the changes in product demand occurring in through the

94

pavement life cycle. Changes in demand were linked to the time-dependent technosphere and the ecosphere

95

matrices to generate the inventory. Finally, the DLCI results were assigned to dynamic and static CFs of

96

endpoint categories to assess damage to the environment.

97

Methodology

98

Description of the dynamic consequential model for pavements

99

We used the basic equation for calculation of the inventory matrix of a process-based LCA 17 to simplify the

100

dynamic aspect of the CLCA framework, where the BAU scenario substitutes with the ALT scenario

101

comprising the same level of functionality as:

102

ΔLCI(t) = ∫ΔLCI(t)dt = ∫B(t).(A(t)) ―1.Δf(t)dt

103

where ΔLCI(t) represents the life cycle inventory matrix at time t; Δf(t) is a vector that represents the

104

changes in demand due to substituting BAU with ALT at time t; A is the technosphere matrix representing

105

the input intermediate flows required for delivering a unit of process outputs; and B represents the biosphere

106

matrix incorporating the environmental interventions required for each intermediate flow. The dots in this

107

equation represent matrix products. The elements of the biosphere matrix can change in time due to the

108

implementation of emission control systems and regulations, which are well explained in previous published

109

works 18. However, the focus of this study is on the dynamic aspect of consequential inventory that is well

110

reflected in the technosphere matrix. Changes in the technosphere matrix over time have to do with

111

intermediate flows modifications and technology efficiency improvements. In this study, the value of A(t) is

112

computed according to Eq. (2):

113

A(t) = (Abase ∘ [T(t) + R(t)])𝑗

114

The ∘ in Eq. (2) represents a Hadamard product, i.e., an entrywise multiplication of matrices. In addition,

115

Abase represents the quantity of intermediate flows required to fulfill the processes in the first year of service

116

life; on the other hand, T(t) modifies the quantity Abase for the technical efficiency according to the time

Eq. (1)

Eq. (2)

6 ACS Paragon Plus Environment

Page 7 of 27

Environmental Science & Technology

117

horizon. j stands for the short-term or long-term demand for a specific intermediate flow. We defined

118

parameter Tshift as the number of years after beginning the life cycle when change in demand will affect new

119

capacity installation. If t ≥ Tshift, the long-term affected supplier is affected by the demand change through the

120

new capacity installation, while prior to reaching Tshift, the current flexible suppliers will be affected, i.e. the

121

short-term affected supplier. The short and long-term affected suppliers will be identified based on the

122

operating costs, and the market will adjust its capacity to cover the demand. In the T(t) matrices, each

123

element ai,p(t) represents the efficiency improvement for the product i to the processes p at time t and is

124

calculated according to Eq. (3). For example, fuel efficiency of passenger cars and trucks tends to improve

125

over time as reported by the government 19.

126

ai,p(t) = 1 ― [ 𝐸 (𝑡)]

127

Where

128

values more than 0. For example, it can be between 0 and 1 (when

129

consumption of vehicles or can be greater than 1 (when

130

processes. To consider the consumer reaction to the improvement in fuel efficiency, a time-dependent

131

rebound effect R(t) is included in Eq. (2) that modifies the values of T(t) matrix. For example, as fuel

132

efficiency improves, drivers usually counterbalance a percentage of this improvement by purchasing more

133

fuel 20. The elasticity of technology efficiency (𝜂𝐸) is implemented to offset a proportion of the improvement

134

in the technology efficiency. The procedure of calculating the elasticity efficiency is presented in section

135

4.1.1 of the SI.

136

Each element of bi,p, in the R(t) matrix is calculated as:

137

bi,p = [ 𝐸 (𝑡) × 𝜂𝐸(𝑡)]

∂𝐸

∂𝐸 𝐸 (𝑡)

Eq. (3)

i,p

represents the relative change in the technology efficiency. The elements of T(t) can take any

∂𝐸 𝐸

∂𝐸 𝐸

≥ 0) in the case of improvement in fuel

< 0) for output flows of resource extraction

∂𝐸

Eq. (4)

i,p

7 ACS Paragon Plus Environment

Environmental Science & Technology

Page 8 of 27

138

Where 𝜂𝐸(𝑡) represents the efficiency elasticity corresponding to product i. The R(t) matrices have the same

139

dimension of T(t) and Abase.

140

To compute the value of elements in Δf (t), we consider the avoided and additional demands from the BAU

141

and ALT scenarios. Therefore, for a given time t, we have:

142

∆f(t) = ∫t ― 1([∆f(t)]ALT ― [∆f(t)]BAU)dt

143

After calculating the DLCI, dynamic LCA results can be generated by assigning the proper dynamic

144

characterization factors (CFs) to the time-dependent inventory.

145

Description of the Case Study

146

The purpose of this study is to evaluate the environmental consequences of reconstruction of existing

147

pavement infrastructure. We chose a case study where asphalt pavement (BAU scenario) was substituted

148

with jointed plain concrete pavement (ALT scenario) to apply the proposed method. In this study, the

149

functional unit was defined as “providing a path for traffic service over the entire network of the two-lane

150

roads with 20,000±1,000 annual average daily traffic, including 5% of trucks, in the province of Quebec, for

151

50 years”. We determined relevant time horizons for technologies affected by demand changes during the

152

pavement life cycle according to life cycle phases. Figure 1 summarizes the connections between short-term

153

and long-term affected technologies with each life cycle component. Connections include materials,

154

construction, IRI-induced and rigidity-induced fuel consumption, albedo effects [in terms of radiative forcing

155

(RF) and urban heat island (UHI)], lighting, and carbonation, i.e., the ability of concrete to reabsorb CO2,

156

which transform concrete hydrates (i.e. portlandite) to a new product through a chemical process21 (readers

157

are referred to Section 4.3. of the SI to get more information about the carbonation calculation and effective

158

parameters). It should be noted that some life cycle components incorporate both the short-term and long-

159

term effects during the use phase of the pavement since their demands are changing during both periods. In

160

addition, carbonation and RF effects do not intervene with any market product and directly affect the

161

ecosphere through a change in the quantity of CO2 in the system (See section 4.3 and 4.2 in the SI). Thus, the

t

Eq. (5)

8 ACS Paragon Plus Environment

Page 9 of 27

Environmental Science & Technology

162

carbon flows were added directly to inventory results. More information justifying time horizon assignments

163

for the affected flows to each life cycle phase is provided in the attached SI document, page S4-S7. We also

164

used the ecoinvent consequential database v.3.2 for modeling background processes 22.

165

In much of the transportation literature a period of 5–10 year is estimated empirically for short-term effects

166

23.

167

short-term horizons. Certain parameters, such as monthly ambient temperature and downward solar radiation

168

at the surface of the Earth, induce significant variations in the inventory results during a year 24. Hence, a

169

monthly time step was considered to produce the dynamic inventory. In the next step, we identified affected

170

technologies and the determining products in case of a multi-functional process for different time horizons

171

using the step-wise procedure proposed by Weidema 25. Readers can find details of the procedure for finding

172

the affected technologies for each flow in the foreground system in SI, page S10-S33. To calculate demand

173

changes we used a set of dynamic parameters for each life cycle component. We also considered a one-year

174

lag time for the procurement of materials through the supply chain. A comprehensive overview of the

175

demand in BAU and ALT cases for materials production, construction, maintenance and repair (M&R) and

176

end-of-life phases can be found in previous work 26. In addition, a detailed description of the methodology

177

for computing change in demands of use-phase components is presented in SI, page S34-S52. Table 1 is

178

provided to reflect the significance of the difference between the impacts of CLCA and ALCA flows.

179

Readers are referred to Table S6 for a complete list of the affected suppliers.

However, we utilized a sensitivity analysis to assess the effects of the frontier between long-term and

9 ACS Paragon Plus Environment

Environmental Science & Technology

T=0

T=50

Page 10 of 27

Link to appropriate flow Affected technology for construction and M&R materials

Materials production

Affected technology for equipment and fuels

Construction Carbonation

CO2 balance

Radiative forcing (Albedo) Use

Urban heat island (Albedo) Affected technology for fuels and electricity

Excessive fuel consumption (IRI) Excessive fuel consumption (Rigidity) Lighting

Affected technology for equipment and fuels

Maintenance and repair End of Life

180

Affected technology for construction materials, equipment, and fuels

181

Figure 1. Connections between pavement life cycle components and affected technologies and flows with

182

different time horizons. Light green represents short-term horizons, dark green long-term horizons, and the

183

gradient boxes represent the incorporation of both short and long-term horizons in the “link to appropriate

184

flow” box.

10 ACS Paragon Plus Environment

Page 11 of 27

185

Environmental Science & Technology

Table 1. Summary of affected technologies in different time horizons (WCS = West Canadian Select, WTI = Western Texas Intermediate)

186

1

187

2

188

3

189

4

190

*

Intermediate flow (unit) Cement (t)

Short-term affected supplier 1 (kg CO2/Unit) * Dry kiln technology with natural gas (834)

Long-term affected supplier 1 (kg CO2/Unit) * Pre-calciner and pre-heater technology with forest biomass (515)

Current average market 2 (kg CO2/Unit) * Average production of dry and pre-calciner technologies with various fuels (821)

Reinforcing rebars (t)

German manufactured with blast oxygen furnace technology (2998)

Canadian manufactured with blast furnace with carbon capture and storage and top gas recycling technology (1996)

Average of recycled and virgin steel input produced through convertor and electric arc technologies (2251)

Car transportation (km)

WCS crude oil (0.376)3

WTI crude oil (0.283)3

Average of imports from WCS, WTI, and OPEC (0.363)4

Bitumen (t)

WTI crude oil (624)3

Bioasphalt (92)3

Fuel for asphalt production (t)

Light fuel oil (27.1)3

Natural gas (24.2)3

Average of imports from WCS, WTI, and OPEC (322)4 Average of Light and heavy fuel, natural gas, biogas, and propane (28.5)4

Low voltage electricity (kWh)

Gas power plant import (0.447)

Wind farm (0.035)

Average of hydropower, wind, imports (0.025)

consequential long-term dataset was used for background processes modeling. attributional recycled-content data set was used for background processes modeling. system expansion was applied to the multifunctional process. economic allocation was applied to the multifunctional process. gwp100 characterization factors were used.

11 ACS Paragon Plus Environment

Environmental Science & Technology

Page 12 of 27

192

Impact assessment methods

193

We considered the damage categories ecosystem quality (EQ), human health (HH), and resources (R) using

194

the CFs proposed in IMPACT 2002+ 27 to assess the significance of the potential damages of the substitution.

195

In addition, we used the CFs for global warming potential (GWP) with a 100-year time horizon proposed in

196

an IPCC report

197

considered single time-independent CFs for life cycle impact assessments. Nevertheless, the proposed dynamic

198

framework enables ones to link the inventory with a time‐dependent impact assessment method. Therefore,

199

we used dynamic CFs for GWP to evaluate the CC impact through time. Inspired by the approach proposed

200

by Levasseur et al. 13, we assigned the dynamic GWP CFs to time-dependent LCIs obtained by linking the

201

demand changes to the corresponding affected technologies as:

28

to analyze the climate change (CC) impact of the substitution. Previous studies majorly

202 203

Eq. (5)

204 205

where GWP (t) is cumulative GHG emissions at year t, computed by summing instantaneous radiative forcing

206

at any time t caused by all emissions from year 0 to year t (the relative starting time of the emissions is

207

considered as 0 and all the GHG emissions during the year are assumed to be accumulated at the end of year

208

and is assigned to the corresponding CF); k is number of years between 0 and t; and DCF stands for dynamic

209

characterization factor for each GHG elementary flow. The dot in the equation represents the matrix product.

210

It should be noted that, a marginal impact modeling in LCIA stage can be applied as the additional impact per

211

additional unit emission or extraction induced by the product system on top of the existing background system

212

29.

213

considering nonlinearity of impacts based on local conditions like extremely high or low background

214

concentrations to which the demand changes contribute an additional emission. Future work should explore

215

the effects of non-linear LCIA on the consequential modeling.

This nonlinearity integration in LCIA stage allows decision-makers to resolve accuracy issues, such as

12 ACS Paragon Plus Environment

Page 13 of 27

Environmental Science & Technology

216

Treatment of uncertainty and variability sources

217

Analyzing uncertainty and variability sources in LCA results allows one to evaluate and improve the robustness

218

of the conclusions. We classified these sources into a) model uncertainty, the errors exist in the structure of the

219

model via used equations and conditions in the use–phase, which is introduced by simplifying potentially

220

relevant aspects of the real world within the modeling structure; b) data quality uncertainty, the parameter

221

uncertainty related to quality of the inventory; and c) variability, the inherent variations in the parameters 26,

222

30.

223

the Crystal Ball spreadsheet-based application. Analyzing the combined effects of sources of uncertainty and

224

variability provides a comprehensive perspective on the precision of the study. We evaluated the contribution

225

of each source of variability and uncertainty, and its corresponding parameters, to the total uncertainty. This

226

classification helps one to refine and prioritize uncertainty sources or parameters that have the most effect on

227

variance. We computed the normalized squared Spearman rank correlation coefficients to characterize the

228

relative contribution of each input in the variance of the damage categories. The information about the

229

parameters and procedure of uncertainty analysis is explicitly presented in spreadsheet SI and word SI, pages

230

S52 to S54. Mutel et al. proposed to set the number of Monte Carlo iterations 100 times the number of assessed

231

parameters in order to minimize the uncorrelated errors in the results of contribution to variance, i.e. the

232

Spearman rank correlation coefficient 31. The variation of results associated with uncertainty and variability

233

sources in input data was assessed with 200,000 Monte Carlo iterations. Justifications for quantifying

234

characterizations of uncertainty and variability sources are presented in section 5, Page S54-S55 of SI.

235

Results and discussions

236

Dynamic human health, ecosystem quality and resources results

237

Figure 2 shows cumulative, dynamic and static results of the three damage categories representing the

238

difference between the BAU and ALT scenarios. Therefore, it is possible for the results to be either positive

239

(when ALT has a larger impact than BAU) and vice versa. For ease of comparison with other studies, we

240

normalized the results to 1-km of pavement. A dynamic CLCA approach helps to provide a more

We then conducted Monte Carlo simulations to propagate these sources to the four assessed categories using

13 ACS Paragon Plus Environment

Environmental Science & Technology

Page 14 of 27

241

comprehensive estimate but is always limited by the understanding of model practitioner of the pavement

242

system, as well as the model simplifications and assumptions. Therefore, it is important to understand the

243

influence of key dynamic factors on the variables in the model rather than to compute the absolute values for

244

the variables. We observed an increase in environmental damages at the beginning of the life cycle, which

245

coincides with the material production and construction phases in all the damage categories except for R

246

(Figure 2 and 3). The EQ, HH, and CC impacts of producing concrete are greater than those for asphalt,

247

while the feedstock energy embodied in crude oil (51.2 MJ/kg) leads to a higher resource consumption

248

impact than the ALT scenario in early ages.

249

The difference between the environmental damages of two pavement systems for the first 15 years is

250

relatively small compared to that observed at the later ages. Indeed, as we go further in the life cycle, we find

251

larger negative impacts implying an increase in the environmental benefits obtained by substituting the BAU

252

with the ALT scenario. Sharp decreases at the ages of 15 and 28 years are related to the avoided demand for

253

M&R of materials and machinery for the BAU case when a new overlay system will replace an old one. As

254

these M&R schedules occur in the long-term, the long-term asphalt suppliers of the M&R of materials are

255

affected by the decrease in demand. For ecosystem quality, the sharp decreases are more significant than

256

those in other categories since 208 mg aluminum and 16.7 mg zinc, by 1-kg wood ash landfarming, are

257

emitted to the soil, resulting in 2.073 PDF.m2.yr per kg bioasphalt (i.e., the long-term affected technology for

258

bitumen production). Comparing with Steele et al. 32, these toxicity emissions were not observed by since

259

they used the TRACI v.2 method that does not capture terrestrial emissions for aluminum 33. To study the

260

significance of the M&R schedule, a sensitivity analysis of the repair intervals in incorporated (Table S1 in

261

SI). The obtained results showed less than 5% change in the R and EQ damage differences and

262

approximately 10% difference in HH and CC when each interval is postponed or hurried for two years

263

(Figure S31 in SI).

264

While the materials production predominantly affects EQ results, the use phase and, particularly, excessive

265

fuel consumption due to pavement rigidity induce the major differences between the damages in BAU and 14 ACS Paragon Plus Environment

Page 15 of 27

Environmental Science & Technology

266

ALT scenarios (Figure S26 in SI). The structural resistance of the surface layer to distresses, i.e. the modulus

267

of elasticity, is ten times greater in the ALT case compared to the BAU. Therefore, the induced crude oil

268

extraction and petroleum consumption by the change in vehicle fuel demand are altered by the distinction

269

between the two scenarios.

270

It should be noted that the IRI induced fuel consumption is constantly changing as a function of both the

271

pavement use and in M&R repair and at certain ages (e.g., between 28-32-year ages). Hence, we observed an

272

increase in the IRI-induced fuel consumption when substituting the BAU with ALT which offset a part of the

273

environmental credits obtained by the substitution. Fuel consumption due to rigidity follows a steady

274

increase, as shown in Figure S22-S25 in SI, only depending on the monthly temperature. Despite resurfacing

275

the ALT scenario with asphalt materials at age 40, the rigidity of the ALT scenario, as the surface layer from

276

0-40 years and as a base from 40-50 years, reduces the fuel consumption throughout the pavement life cycle.

277

However, after resurfacing the ALT with asphalt materials at the age of 40, other use phase components such

278

as albedo effect, IRI effect, and carbonation that depend on the surface characteristics of pavements,

279

produced similar damage values. Therefore, there are no changes in the demands of the flows for these use-

280

phase components.

281

Another parameter that contributes significantly to the difference between BAU and ALT results is the

282

electricity consumption related to the urban heat island (UHI) effect. Peng et al. reported that the effect of

283

UHI is more intense in summer than winter due to lower solar intensity during winter 5. In another study,

284

Kolokotroni et al. reported the decrease in heating demand in winter can offset the cooling demand induced

285

by UHI effect in London, leading to 90% and 50% reduction in electricity consumption in short and long-

286

term, respectively 34. In this case study, 61% of the whole length of the road with the specified traffic service

287

is considered as UHI effective fraction of the pavement in urban neighborhood 35. Incorporating heating and

288

cooling demand changes due to the UHI effect in long and severe winters and mild summers, such as those

289

prevailing in Quebec, Canada, results in a net increase in electricity demand (42±9 MWh/yr/km pavement).

290

At least 75% of the UHI-induced emissions in the four damage categories took place during the short-term 15 ACS Paragon Plus Environment

Environmental Science & Technology

Page 16 of 27

291

period and were caused by increases in electricity production in a gas power plant in the U.S and in New

292

Brunswick. (i.e., the short-term affected supplier for electricity). The monthly air temperature plays a

293

significant role in UHI-induced electricity consumption. Therefore, we examined the effect of temperature

294

variability in the geographical context of the case study and included this effect in the statistical results. In

295

addition, as a simplification, we considered heterogeneous urban neighborhoods to be uniformly distributed

296

buildings of equal story and size with a constant household number. In future work, to approach the reality of

297

an urban neighborhood, actual geometry conditions will be used to increase the accuracy of the model

298

through geographic information systems.

299

The endpoint results computed by the DLCI presented in Figure 2 and 3 give us a clear perspective of life

300

cycle emissions and consumptions. These results also help us to understand how changes in certain elements

301

of the system boundaries influence both other elements and the pavement supply chain as a whole.

302

Calculating the emissions with a static LCI provided us divergent results. As shown in Figure S29, using a

303

single long-term affected technology for each unit process leads to an underestimation in the environmental

304

benefits of substituting BAU with ALT in the CC, HH, and R categories. While in EQ, we found that

305

singling out the long-term affected technology and ignoring the dynamic aspect of the time-dependent

306

parameters can overestimate the environmental benefits of the BAU-ALT substitution by 114% compared

307

with the dynamic results. However, the existing variation due to the uncertainty and variability sources in

308

static results may make the differences among the dynamic and static results not statistically significant, and

309

further investigation will be required.

16 ACS Paragon Plus Environment

Page 17 of 27

Environmental Science & Technology

Human health (DALY)

1.00 0.50 0.00 -0.50 -1.00 -1.50 -2.00 0

5

10

15

20

25

30

35

40

45

50

Time (Year)

Ecosystem Quality (105 PDF.m2.yr)

310 4.00 2.00 0.00 -2.00 -4.00 -6.00 -8.00 -10.00 -12.00 -14.00 -16.00 0

5

10

15

20

25

30

35

40

45

50

Time (Year)

Resource (106 MJ primary)

311 5 0 -5 -10 -15 -20 -25 -30 0

312

5

10

15

20

25

Time (year)

30

35

40

45

50

5th Percentile Mean Median

17 ACS Paragon Plus Environment

Environmental Science & Technology

313

Page 18 of 27

Figure 2. Accumulated mean, median and 5th and 95th percentile results of ecosystem quality (EQ), human

314 health (HH), and resources (R) based on IMPACT 2002+ LCIA method. Static characterization factors were 315 used for all the scenarios. In the static affected technology scenario, we used a similar modeling framework to 316 that of the dynamic except that we only linked demand to long-term affected technologies. In static use phase 317 modeling with static affected technology, in addition to isolating the affected supplier of long-term 318 technology, we modeled the use-phase with static inputs. A list of input parameters is available in Spreadsheet 319 SI, Table S1). 320 321

Dynamic vs. static GHG results in consideration of CFs

322 Dynamic accounting of GHG emissions results in a large absolute difference when compared to static GWPs 323 calculated against a 100-year arbitrary time horizon. As shown in Figure S27 in SI, we have a substantial 324 methane emission at the beginning of the life cycle, primarily due to the increase in the demand of the surface 325 materials and transportation to the construction site. Looking into the details of the process contribution, we 326 observed that the production of steel rebars by the German manufacturers with blast oxygen furnace 327 technology (the short-term supplier) and their transportation to Canada emits 0.011 kg CH4/kg steel to the new 328 construction of the ALT 36. However, the change in the demand for steel does not continue in the later ages for 329 M&R. We also observe a significant GWP increase up to ten years after the starting time of the emissions 330 related to the UHI-induced electricity consumption. The largest emission of CO2 in ALT and BAU systems 331 comes from the binders. In fact, production of portland cement as a binder for concrete pavement by natural 332 gas using dry kiln technology, i.e. the short-term affected supplier, results in 0.83 kg CO2/kg cement emission 333 (versus 0.51 kg CO2/kg cement by long-term affected technology, i.e. preheater and pre-calciner technology 334 with biomass fuel). The asphalt binder induces 0.43 kg CO2/kg bitumen. Concerning demanded binders for 335 each pavement system, i.e. 613 t cement/km pavement versus 219 t bitumen/km pavement, a significant 336 difference of CO2 flow in the new construction is emerged, which is linked to the difference in binder quantity 337 and CO2 intensiveness of the binder production process. In addition, production of high-voltage electricity 338 through the short-term affected supplier (gas power plant in New Brunswick) emits 0.432 kg CO2/kWh, while 339 the quantity of CO2 emitted by the long-term technologies is much lower than that emitted in short-term (0.033 18 ACS Paragon Plus Environment

Page 19 of 27

Environmental Science & Technology

340 kg CO2/kWh) 36. Shifting from the short-term to the long-term supplier of electricity production after 10 years 341 can explain the sharp reduction in the UHI contribution in the GWP results, which is presented in Figure S22. 342 Since the lifetime of CH4 is much shorter than CO2 and N2O 37, its corresponding RF will diminish sooner than 343 other evaluated GHGs after emission. Hence, the major contribution of GWP after methane emissions, comes 344 from CO2 emissions in the years after the materials production and construction phases. 345

One should note that the dynamic CFs results may be more or less than the static CFs results depending on

346 the timing and quantity of emitted GHGs. In fact, limiting the dynamic CF results of GHG emissions to an 347 arbitrary time-horizon brings about the consideration of GHG emissions over a smaller period (e.g., emissions 348 occurring at year 10, 20, or 50 are effective over a period of 90, 80, or 50 years, respectively for a fixed 100349 year time horizon). The difference between static CF and dynamic CF results lays in this shorter effective time 350 of emissions. In fact, the later the emissions occur in dynamic CFs results, the shorter the time horizon used 351 for assessing their impact on global warming when we limit the period of assessment. As illustrated in Figure 352 S27 and manifested in the static CFs results, the positive quantity of GHGs induced in the early ages of the life 353 cycle, was larger in the cumulative dynamic 100-year results compared to static CFs. However, the cumulative 354 GHG emissions at later ages are negative, implying environmental benefits of the substitution during the use 355 phase. The GHG emissions of the use phase resulted in a lower GWP score than that of static CFs. This is 356 because the use phase emissions have an effect over a shorter period of time than those in the materials 357 production and construction phases. In other studies, Levasseur et al.13 compared the substitution of gasoline 358 with corn ethanol and observed a similar reduction in dynamic CF results compared to those with static CFs. 359 This reduction was due to the extensive GHG emissions in the ethanol scenario during the production phase, 360 which is amplified since it occurs at the first year of the analysis period. In a wastewater treatment plant, 361 Shimako et al. reported that when the major GWP impacts come from methane and there was no difference 362 between the dynamic and static CFs results after 100 years since the RF effect of methane is greatly reduced in 363 100 years 38. On the other hand, a comparative study of alternatives for buildings considered the dynamic and 364 static CFs for calculating the buildings GWP 39. The results showed that using the dynamic CFs tends to

19 ACS Paragon Plus Environment

Environmental Science & Technology

Page 20 of 27

365 decrease the impact of wood houses compared to other materials, such as concrete. This decrease is because 366 the EOL emissions occur 100 years after construction, therefore there are no CFs for EOL emissions with a 367 fixed time horizon of 100 years, and the GWP related to the EOL incineration is not captured.

600.00

473.5 metric ton CO2eq

400.00 200.00 0.00

Metric ton CO2eq

-200.00 -400.00 -600.00 -800.00

-1,000.00 -1,200.00 -1,400.00 0

368 369

50

100

150

200

250

Time (Year)

5th Percentile Median Dynamic GWP Static use phase modeling + Static affected technology

300

350

400

450

500

Mean 95th Percentile Static affected technology

Figure 3. Comparison of static GWP fixed CFs for time horizons of 100 years (black continuous line) based

370 on IPCC 2013 and dynamic CFs results (blue continuous line) for the substitution of ALT with BAU. The total 371 cumulative results at the end of life of the pavement were considered constant for the years after 50 for the 372 purpose of comparison with dynamic results. In the static affected technology scenario, we used a similar 373 modeling framework to the dynamic, except that we only linked the demand to long-term affected 374 technologies; static CFs were included. In static use-phase modeling with static affected technology, in 375 addition to limiting the affected supplier to long-term technology and using static CFs, we computed the use376 phase impacts with static inputs. A list of input parameters is available in Spreadsheet SI, Table S1. 377 20 ACS Paragon Plus Environment

Page 21 of 27

Environmental Science & Technology

378

Individual and combined analysis of uncertainty and variability sources

379

Figure 4 illustrates a hierarchical contribution analysis of variance at two levels of uncertainty and variability

380 sources and parameters for the four damage categories. Variability of data plays the leading role in the statistical 381 variation of the results in the CC, HH, and EQ categories. The thickness of surface overlays and the ALT surface 382 milling can change due to the equipment conditions and the technology used for M&R. This variation in the 383 overlay thicknesses in the M&R and construction phases might be considered negligible (e.g., ±2 cm variation 384 in asphalt overlay for M&R) as it can induce a variation of 234 m3 of asphalt materials. Gregory et al. 4 examined 385 the variation of pavement thicknesses for initial construction and found this input to be one of ten parameters 386 that contribute most to GWP results. Adding M&R thickness variations, as we did in this study, we can consider 387 the thickness variation impact through the life cycle and we obtain results which are more significant than those 388 obtained by Gregory et al. In addition, the pavement service life, which is one of the key parameters in reference 389 flows calculation, resulted in an 11-15% contribution in the variance of the damage categories. The air 390 temperature variability contributed 8-16% to the variance of the damage results; less for CC and more for R 391 category. Temperature affects the rigidity of pavements and determines whether the UHI effect induces cooling 392 and heating demands, which are the two of the heaviest weighted components in all the damage category results 393 (See Figure S26 in SI). 394

The quality of the data obtained from proxy processes is also significant according to Figure 4, particularly for

395 the R category. Car and truck fuel consumption dependent on IRI and rigidity of pavements plays a major role 396 in parameter uncertainty contribution. The basic uncertainty factor with a significant value of the 0.12, i.e., the 397 intrinsic uncertainty related to the epistemic errors in sampling size

40

associated with the foreground of the

398 transport-related unit processes is the major reason for the 23-43% contribution to the variance of the results. 399 Dividing the contributed parameters among the related life cycle phases, we observed that more than 90% of the 400 total variance comes from the model, variability and data quality of parameters related to the IRI component as 401 illustrated in Figure S28. Therefore, we can consider this component a priority when resources are available for 402 model refinement. 21 ACS Paragon Plus Environment

Environmental Science & Technology

Page 22 of 27

403 One should note that the uncertainty results of this stage do not incorporate the variation of parameters, 404 assumptions, and modeling in the goal and scope definition and the impact assessment method. As reported by 405 Linkov and Burmistrov, the model uncertainty of exposure point concentrations, temporal changes in exposure 406 point concentrations, and particularly, application to future scenarios, formulation, model implementation, and 407 parameter selection originating from subjective interpretation, can extensively affect the selection made through 408 an impact assessment 41. Moreover, the uncertainty related to consequential LCA modeling was not incorporated 409 to the variation sources. Indeed, in prospective LCA studies, uncertainties are to some extent quantifiable due 410 to incomplete scientific knowledge. One way to improve the uncertainty analysis of this study consists in 411 developing a method to incorporate modeling uncertainty associated with the consequential method, where 412 affected suppliers are determined. In fact, framing the model in different ways to identify the affected 413 technologies may result in a wide range of calculation outputs. For example, looking at electricity generation in 414 Denmark, Mathiesen et al. showed the uncertainty of the CLCA methodology by providing specific examples 415 in different time frames 42. The use of these alternative models is one of few available techniques to treat this 416 model uncertainty

41.

The inclusion of these uncertainty sources to the previously discussed sources of

417 uncertainty and variability might result in a limitation in the capabilities of LCA to serve as a reliable tool for 418 this assessment. Time-dependent variation of the uncertainty distributions are also another interesting area for 419 further studies to complement the dynamic-LCA methodology. Several factors, such as indeterminacies related 420 to the future prediction of the environment and technosphere behaviors, can influence the properties of the 421 probability distribution and therefore, they can be dynamically considered in the analysis. In this study, although 422 the mean values of certain parameters, abbreviated as CAFT in SI 1, are constantly changing in time, the other 423 dependent parameters, such as standard deviations remained constant as a limitation of the uncertainty analysis.

22 ACS Paragon Plus Environment

Page 23 of 27

Environmental Science & Technology

424 Pij

Human health

Climate change

425 Ecosystem quality

IRI

Resources

426 427

Figure 4. Sources of variability and uncertainty and inputs to the variance in each category based on

428 correlation coefficients of the normalized squared Spearman rank. LT= Long-term, ST= Short-term, IRI= 429 International roughness index, M&R= Maintenance and repair, BAU= Business-as-usual scenario, Pij= 430 Dimension functions in rigidity model, Ta= Atmospheric transmittance factor. 23 ACS Paragon Plus Environment

Environmental Science & Technology

431

Page 24 of 27

The use of the dynamic consequential model implicates pavement alternatives as a means to transition to lower

432 environmental footprints for road transportation. To achieve the energy saving potential of pavements, necessary 433 policy measures include a revision of subsidies for energy household consumption and innovative road 434 transportation technologies. The use-phase of pavements and, particularly, the responses of suppliers to demands 435 can be effectively influenced by interventions in the energy sector. In regard to the pavement use phase, the 436 interaction of the model parameters in this phase with the inherent variations in the world, i.e., the true 437 differences in time-dependent and spatial-related parameters or user behavior and is manifested in uncertainty 438 analysis, implies the limitation of improving the accuracy of the results. Although not established as a policy in 439 Quebec province, there is a probability of developing a mature market for electric vehicles in the long term. 440 Depending on the cost of infrastructure development or operation, the electric vehicles might be the affected 441 supplier in different time horizons. Particularly in regions with investments in low-cost clean technologies of 442 electricity generation, the use phase-induced demands for vehicles fuels may revert the conclusion when an 443 alternative scenario manifests its environmental benefits through the fuel-saving properties. The evidence for 444 the importance of the future energy strategy is in electricity provision in Quebec, as we show for the UHI effect, 445 where the wind electricity projection in the long term diminished the impact of heating demand increase resulted 446 from shifting from darker to a lighter color pavement. Nonetheless, to avoid underscoring the impacts of the 447 demand changes a more precise analysis is required, since the fluctuation and peak demand in intermediate 448 flows, such as electricity, need to disaggregate intra-annual data 43. This discussion can be more meaningful 449 underneath the scenario considering urban densification in the studied region, reflecting the importance of urban 450 planning. 451

Acknowledgment

452

The authors thank the anonymous dedicated reviewers for their helpful and constructive criticism and their

453 support of the approach taken. The authors are also grateful to the Natural Sciences and Engineering Research 454 Council of Canada (NSERC) for its financial support through its ICP program and Fonds de Recherche du 455 Québec-Nature et Technologie (FRQNT) through its Merit Scholarship Program for International Students (V1). 24 ACS Paragon Plus Environment

Page 25 of 27

Environmental Science & Technology

456

Associated content

457

Parameters, probability distributions, and statistical values are presented in spreadsheet SI1 (XLSX file). SI2

458 (PDF file) includes the details of calculations and case study specifications.

459

References

460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496

1. AzariJafari, H.; Yahia, A.; Ben Amor, M., Life cycle assessment of pavements: reviewing research challenges and opportunities. J. Clean. Prod. 2016, 112, Part 4, 2187-2197. 2. FHWA Pavement Notebook Chapter 1: Pavement Policy. https://www.fhwa.dot.gov/pavement/notebook/chapter01.cfm (February 15th, 2018), 3. Gschösser, F.; Wallbaum, H., Life Cycle Assessment of Representative Swiss Road Pavements for National Roads with an Accompanying Life Cycle Cost Analysis. Environ. Sci. Technol. 2013, 47, (15), 84538461. 4. Gregory, J. R.; Noshadravan, A.; Olivetti, E. A.; Kirchain, R. E., A Methodology for Robust Comparative Life Cycle Assessments Incorporating Uncertainty. Environ. Sci. Technol. 2016, 50, (12), 63976405. 5. Peng, S.; Piao, S.; Ciais, P.; Friedlingstein, P.; Ottle, C.; Bréon, F.-M.; Nan, H.; Zhou, L.; Myneni, R. B., Surface Urban Heat Island Across 419 Global Big Cities. Environ. Sci. Technol. 2012, 46, (2), 696-703. 6. Gong, J.; You, F., Consequential Life Cycle Optimization: General Conceptual Framework and Application to Algal Renewable Diesel Production. ACS Sustainable Chemistry & Engineering 2017, 5, (7), 5887-5911. 7. Yang, Y.; Heijungs, R., On the use of different models for consequential life cycle assessment. Int J Life Cycle Assess 2018, 23, (4), 751-758. 8. Schmidt, J. H., Life cycle assessment of five vegetable oils. J. Clean. Prod. 2015, 87, 130-138. 9. Habert, G.; Billard, C.; Rossi, P.; Chen, C.; Roussel, N., Cement production technology improvement compared to factor 4 objectives. Cem. Concr. Res. 2010, 40, (5), 820-826. 10. Zamagni, A.; Guinée, J.; Heijungs, R.; Masoni, P.; Raggi, A., Lights and shadows in consequential LCA. Int. J. Life Cycle Assess. 2012, 17, (7), 904-918. 11. Earles, J. M.; Halog, A., Consequential life cycle assessment: a review. Int J Life Cycle Assess 2011, 16, (5), 445-453. 12. A. Zamagni, P. B., P.L. Porta, R. Buonamici, P. Masoni, J. Guinée, R. Heijungs, T. Ekvall, R. Bersani, A. Bieńkowska and U. Pretato Critical review of the current research needs and limitations related to ISO-LCA practice ENEA, The Italian National Agency on new Technologies, Energy and the Environment: Deliverable D7 of work package 5 of the CALCAS project 2008. 13. Levasseur, A.; Lesage, P.; Margni, M.; Deschênes, L.; Samson, R., Considering Time in LCA: Dynamic LCA and Its Application to Global Warming Impact Assessments. Environ. Sci. Technol. 2010, 44, (8), 31693174. 14. Kendall, A.; Price, L., Incorporating Time-Corrected Life Cycle Greenhouse Gas Emissions in Vehicle Regulations. Environ. Sci. Technol. 2012, 46, (5), 2557-2563. 15. Chen, X.; Wang, H., Life cycle assessment of asphalt pavement recycling for greenhouse gas emission with temporal aspect. J. Clean. Prod. 2018, 187, 148-157. 16. Yu, B.; Sun, Y.; Tian, X., Capturing time effect of pavement carbon footprint estimation in the life cycle. J. Clean. Prod. 2018, 171, 877-883.

25 ACS Paragon Plus Environment

Environmental Science & Technology

497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544

Page 26 of 27

17. Mutel, C. L.; Hellweg, S., Regionalized Life Cycle Assessment: Computational Methodology and Application to Inventory Databases. Environ. Sci. Technol. 2009, 43, (15), 5797-5803. 18. Collinge, W. O.; Landis, A. E.; Jones, A. K.; Schaefer, L. A.; Bilec, M. M., Dynamic life cycle assessment: framework and application to an institutional building. Int J Life Cycle Assess 2013, 18, (3), 538552. 19. NRC 2018 Fuel Consumption Guide. http://www.nrcan.gc.ca/energy/efficiency/transportation/carslight-trucks/buying/7487 (February 1st 2018), 20. Moshiri, S.; Aliyev, K., Rebound effect of efficiency improvement in passenger cars on gasoline consumption in Canada. Ecol. Econ. 2017, 131, 330-341. 21. Galan, I.; Andrade, C.; Mora, P.; Sanjuan, M. A., Sequestration of CO2 by Concrete Carbonation. Environ. Sci. Technol. 2010, 44, (8), 3181-3186. 22. ecoinvent, Ecoinvent v.3.2 database. In Swiss Centre for Life Cycle Inventories, Ed. Zurich and Dubendorf, Switzerland, 2015. 23. Goodwin, P.; Dargay, J.; Hanly, M., Elasticities of Road Traffic and Fuel Consumption with Respect to Price and Income: A Review. Transport Reviews 2004, 24, (3), 275-292. 24. Xu, X.; Gregory, J.; Kirchain, R., The Impact of Pavement Albedo on Radiative Forcing and Building Energy Demand: Comparative Analysis of Urban Neighborhoods. In Transportation Research Board 97th Annual Meeting, Transportation Research Board: Washington D.C., 2018. 25. Weidema, B. P. Market information in life cycle assessment; Danish Environment Protection Agency: Copenhagen, Denmark, 2003. 26. AzariJafari, H.; Yahia, A.; Amor, B., Assessing the individual and combined effects of uncertainty and variability sources in comparative LCA of pavements. Int J Life Cycle Assess 2018, 23, (9), 1888-1902. 27. Jolliet, O.; Margni, M.; Charles, R.; Humbert, S.; Payet, J.; Rebitzer, G.; Robenbaum, R. K., IMPACT 2002 + : A New Life Cycle Impact Assessment Methodology. Int J Life Cycle Assess 2003, 8, (6), 324-330. 28. Myhre, G.; Shindell, D.; Bréon, F.-M.; Collins, W.; Fuglestvedt, J.; Huang, J.; Koch, D.; Lamarque, J.-F.; Lee, D.; Mendoza, B.; Nakajima, T.; Robock, A.; Stephens, G.; Takemura, T.; Zhang, H., Anthropogenic and Natural Radiative Forcing. In Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change, Stocker, T. F.; Qin, D.; Plattner, G.-K.; Tignor, M.; Allen, S. K.; Boschung, J.; Nauels, A.; Xia, Y.; Bex, V.; Midgley, P. M., Eds. Cambridge University Press: Cambridge, United Kingdom and New York, NY, USA, 2013; pp 659–740. 29. Rosenbaum, R. K.; Hauschild, M. Z.; Boulay, A.-M.; Fantke, P.; Laurent, A.; Núñez, M.; Vieira, M., Life Cycle Impact Assessment. In Life Cycle Assessment Theory and Practice, Springer Nature: Cham, Switzerland, 2018; pp 167-270. 30. Lloyd, S. M.; Ries, R., Characterizing, Propagating, and Analyzing Uncertainty in Life-Cycle Assessment: A Survey of Quantitative Approaches. J. Ind. Ecol. 2007, 11, (1), 161-179. 31. Mutel, C. L.; de Baan, L.; Hellweg, S., Two-Step Sensitivity Testing of Parametrized and Regionalized Life Cycle Assessments: Methodology and Case Study. Environ. Sci. Technol. 2013, 47, (11), 5660-5667. 32. Steele, P.; Puettmann, M. E.; Penmetsa, V. K.; Cooper, J. E., Life-Cycle Assessment of Pyrolysis BioOil Production. Forest Products Journal 2012, 62, (4), 326-334. 33. Bare, J.; Young, D.; Qam, S.; Hopton, M.; Chief, S., Tool for the Reduction and Assessment of Chemical and other Environmental Impacts (TRACI). Washington, DC: US Environmental Protection Agency 2012. 34. Kolokotroni, M.; Ren, X.; Davies, M.; Mavrogianni, A., London's urban heat island: Impact on current and future energy consumption in office buildings. Energy and Buildings 2012, 47, 302-311. 35. Canada Statistics Road Transportation, Table RO1: National Highway System. https://www.tc.gc.ca/eng/policy/anre-menu-3042.htm (February 12th, 2018), 36. ecoinvent, Ecoinvent v.3.4 database. In Swiss Centre for Life Cycle Inventories, Ed. Zurich and Dubendorf, Switzerland, 2017. 26 ACS Paragon Plus Environment

Page 27 of 27

545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561

Environmental Science & Technology

37. IPCC IPCC Guidelines for National Greenhouse Gas Inventories. http://www.ipccnggip.iges.or.jp/public/2006gl/pdf/0_Overview/V0_0_Cover.pdf 38. Shimako, A. H.; Tiruta-Barna, L.; Bisinella de Faria, A. B.; Ahmadi, A.; Spérandio, M., Sensitivity analysis of temporal parameters in a dynamic LCA framework. Sci. Total Environ. 2018, 624, 1250-1262. 39. Fouquet, M.; Levasseur, A.; Margni, M.; Lebert, A.; Lasvaux, S.; Souyri, B.; Buhé, C.; Woloszyn, M., Methodological challenges and developments in LCA of low energy buildings: Application to biogenic carbon and global warming assessment. Build. Environ. 2015, 90, 51-59. 40. Muller, S.; Lesage, P.; Ciroth, A.; Mutel, C.; Weidema, B.; Samson, R., The application of the pedigree approach to the distributions foreseen in ecoinvent v3. Int J Life Cycle Assess 2014, 1-11. 41. Linkov, I.; Burmistrov, D., Model Uncertainty and Choices Made by Modelers: Lessons Learned from the International Atomic Energy Agency Model Intercomparisons†. Risk Anal. 2003, 23, (6), 1297-1308. 42. Mathiesen, B. V.; Münster, M.; Fruergaard, T., Uncertainties related to the identification of the marginal energy technology in consequential life cycle assessments. J. Clean. Prod. 2009, 17, (15), 13311338. 43. Amor, M. B.; Lesage, P.; Pineau, P.-O.; Samson, R., Can distributed generation offer substantial benefits in a Northeastern American context? A case study of small-scale renewable technologies using a life cycle methodology. Renewable and Sustainable Energy Reviews 2010, 14, (9), 2885-2895.

562 563

27 ACS Paragon Plus Environment