Evaluation of the Impact of Pavement Roughness on …docs.trb.org/prp/12-2872.pdfEvaluation of the Impact of Pavement Roughness on Vehicle Gas Emissions in Baltimore County ... 12

Evaluation of the Impact of Pavement Roughness on

Vehicle Gas Emissions in Baltimore County

Celine Kalembo1, Mansoureh Jeihani

2, Anthony Saka

3

1Department of Transportation and Urban Infrastructures Studies

School of Engineering, Morgan State University

1700 E. Coldspring Lane, Baltimore, Maryland 21251

Tel: 410-878-7604

e-mail: [email protected]




Tel: 443-885-1873

Fax: 443-885-8324


Corresponding Author




Tel: 443-885-1871

Fax: 443-885-8324


Submission Date: July 31, 2011

Number of Words in the text: 5,474

Number of Tables: 5

Number of Figures: 0

TRB 2012 Annual Meeting Original paper submittal - not revised by author.

ABSTRACT 1

Several gases originating from the land transportation mode are considered as greenhouse 2 gases (GHG) and therefore are among contributing elements of global warming. Both the 3 amount and type of gas emitted depend on factors such as vehicle type, ambient temperature, 4

vehicle age, fuel type and vehicle speed. Among all GHG, carbon dioxide (CO2) will be the 5 focus of this study, as it is the most widespread GHG produced by automobiles. On the other 6 hand, the pavement condition, in terms of its roughness, could be one factor affecting vehicle 7 speeds. With pavement roughness affecting vehicle speeds and speed affecting GHG 8 emissions, one may suggest that the roadway pavement roughness may indirectly affect CO2 9

emission rates. The objective of this study is to investigate the correlation between CO2 10 emissions and the pavement roughness. The pavement roughness measurement used in this 11 study is the International Roughness Index (IRI). IRI data were obtained from the Maryland 12 State Highway Administration. Speed data were collected on selected roads in Baltimore 13

County (Maryland). CO2 emissions quantities were computed using MOVES2010a, a vehicle 14 emission modeling software program. There are two steps involved in determining the IRI 15 and CO2 emissions relationship: (1) verifying that IRI numbers have an impact on vehicle 16 speeds and (2) verifying that vehicle speeds have an impact on CO2 emissions. The results 17

from the analysis indicate a slight increase in the mean speed value from roads in poor 18 condition to roads in either fair or good condition and therefore a decrease in CO2 emissions. 19 The findings of this research could help agencies to properly allocate roadway maintenance 20

funds in the intention of reducing the environmental impacts associated with pavement 21 roughness. 22

23


Celine Kalembo, Mansoureh Jeihani, Anthony Saka 3

INTRODUCTION 1

Topics on greenhouse gases (GHG) have emerged significantly worldwide over the 2

past few years. GHG come from different sources. They are either from natural sources or 3 from sources made by humans (1). There are four main GHG emitted because of human 4 activities: carbon dioxide (CO2), methane (CH4), nitrous oxide (N2O), and ozone depleting 5 compounds (2). Among these four, carbon dioxide, which is produced by automobiles, is the 6 most widespread greenhouse gas according to the Environmental Protection Agency (EPA) 7

(3). The five major fuel consuming sectors contributing to CO2 emissions from fossil fuel 8 combustion are electricity generation, transportation, industrial, residential, and commercial 9 (4). In 2008, the transportation sector’s contribution to global warming represented about 10 27% of the US GHG emissions (5). It is stated in the same source that transportation is the 11 largest source of CO2 and accounts for 47 percent of the net increase in total emissions in the 12

U.S. However, in the State of Maryland, motor vehicle emissions represents about a third of 13

the total ozone forming emissions and greenhouse gas emissions (6). Factors that may affect 14 the amount of gas emissions include, but are not limited to, vehicle type, engine size, ambient 15

temperature, congestion, traffic flow, vehicle age, fuel type, frequent deceleration and 16 acceleration, travelled distance and vehicle speed. This paper presents some of the studies 17 that show that in general the slower the vehicle, the greater the fuel consumption and thus the 18 greater the gas emission rates. 19

Several studies have shown that GHG contribute to global warming. In an effort to 20 minimize the effects of global warming on the planet, some developed countries including 21

those that are major CO2 contributors took several measures. Some of the measures have had 22 positive impacts in certain regions. In the United States (US), there was a remarkable 6.4 23 percent drop in CO2 Equivalent emissions from 2008 to 2009(4). Although measures are in 24

action and the results are being satisfactory in certain parts of the planet, additional efforts 25 should be provided to accelerate the reduction in GHG emissions at the global level as well as 26

the national level. If GHG emissions issues are not addressed efficiently, uncommon 27 variations in the climate change are expected to continue and will have tremendous impacts 28

on both the environment and on people’s health (7). There has been a global increase in the 29 production of carbon dioxide over the past 30 years in most of the sectors (8). Global 30

temperatures are expected to continue to rise as human activities continue to add carbon 31 dioxide, methane, nitrous oxide, and other greenhouse (or heat-trapping) gases to the 32

atmosphere (9). 33 Studies have indicated that nearly 60 percent of the emissions resulted from gasoline 34

consumption are from personal vehicle use (10). With a high demand in passenger vehicles 35 worldwide, GHG emission is likely to remain a major issue. If roads are in poor conditions, 36 more drivers use their brakes more than usual compared to if they were driving on a much 37

smoother road. Frequent vehicle braking is believed to have major impacts on fuel 38 consumption. Heavy braking and quick acceleration can reduce fuel economy by up to 33 39

percent on the highway and 5 percent around town (11). Roughness has an influence on 40 vehicle fuel consumption, tire/pavement contact, vehicle occupant comfort (due to high 41 frequency vibrations) and noise (12). 42

The main objective of this study is to evaluate GHG emissions of vehicles as a 43 function of IRI. Specifically, the study evaluated the relationship between IRI and vehicular 44

speed, and subsequently estimated the resulting GHG emissions. GHG emissions units will 45 be in terms of CO2 equivalents. The international standard practice is to express greenhouse 46 gases in CO2 equivalents (13). The results of this study can be used as a guide to better 47 allocate roadway maintenance funds in order to reduce GHG emissions associated with the 48

pavement surface condition. 49



LITERATURE REVIEW 1

Although the goal of this study is to evaluate the correlation between pavement roughness 2

and GHG emissions, it is necessary to take into consideration the vehicle speed as it relates to 3 both the pavement roughness and gas emissions. Several research papers discuss topics on 4 pavement roughness, vehicle speed, and gas emissions. However, based on the findings, it 5 appears that very little has been done in finding the correlation between IRI and GHG 6

emissions. 7

International Roughness Index 8 Pavement roughness is a product of the surface texture (14). Moreover, based on wavelengths 9

the surface texture is categorized into four domains: unevenness, megatexture, macrotexture, 10 and microtexture. The International Roughness Index (IRI) is the most common index for 11 megatexture (15). It is a mathematical algorithm that takes the profile measurements from a 12

number of “response-type road roughness measuring systems” (RTRRMSs) and calculates 13 the suspension deflection that would be observed from the standard corner suspension of a 14 car (known as a “quarter car”) (16). The simulated suspension motion is accumulated and 15 divided by the distance traveled to give an index with units of slope, typically 16 meter/kilometer or inch/mile. The World Bank developed this unit of measurement in the 17

1980s. 18

19 The Maryland State Highway Administration (MDSHA) generates the International 20

Roughness Index (IRI) by measuring the longitudinal surface profile in both the left and right 21 wheel paths using an Automatic Road Analyzer (ARAN) (17). Ride quality data is provided 22 for each 1/100th mile pavement segment, and is summarized into 1/10th mile sections (17). 23

IRI values, expressed in units of inch/ mile, are used to determine the pavement condition 24 classifications. MDSHA uses five categories of IRI: very good (0<IRI<60), good 25

(60≤IRI<95), fair (95≤IRI≤170), mediocre (170<IRI≤220), and poor (220<IRI≤640). 26

Pavement Roughness, Operating Speed and Operating Costs 27 Few studies were found on the topics of pavement roughness, operating speed and 28

operating costs. According to their finding, these studies showed a link between pavement 29 roughness and operating speed; however, the results were inconsistent. A link was also 30

shown between Pavement Roughness and Operating Costs. 31 In 1985 Paterson and Watanatada showed that at less than 160 NAASRA Roughness 32

Count (NRC), travel speeds are relatively insensitive to roughness with safety and speed limit 33 considerations tending to dominate (18). A value of 160 NRC corresponds to about 6 m/km 34 IRI when using Equation 1. Equation 2 is used to convert m/km to in/mi. Therefore 6 m/km is 35

equivalent to about 386 in/mi. 160 NRC (386 in/mi) indicates a pavement in poor condition. 36 37

NAASRA (counts/km) = 26.49 × Lane IRI (m/km) −1.27 (Equation 1) (19) 38

miin

mile

incheskm

m 36.636213.0

37.391 (Equation 2) 39

40 In 1996, Symonds et al. indicated that travel speeds are not affected by pavement 41

roughness if the pavement roughness is less than 120 NRC (290 in/mi) (20). 42

A study on the relationship between roughness measurements and operating speed 43 showed that drivers slow down as road roughness increases (18). The correlation between 44 pavement roughness and vehicle operating costs by consideration of the fuel, oil, and tire 45 consumption and the maintenance repair and depreciation cost was also investigated (21). 46



Rougher pavements have been associated with higher operating costs (22). Operating costs 1

tend to be higher when driving on roads that are rough because of frequent stops and starts 2 (22). 3

Based on the above studies, it appears that there is a relationship between pavement 4 roughness and operation speed. However, due to the variations in the results from one study 5

to another, it is unclear how strong the relationship is. 6

Pavement Roughness and Gas Emissions 7 A study by the Bureau of Transport and Communications Economics in Australia 8

evaluated the effects of reducing the roughness of the National Highway System and the 9 Pacific Highway on greenhouse gas emissions and vehicle operating costs over the period 10 1996-2015 (12). During their analysis, they took into consideration GHG emissions 11 originating from the production. They also considered GHG emissions in transport and 12 application of road rehabilitation materials into their analysis. They estimated the total 13

emission reduction as the reduction in emissions resulting from reduced fuel consumption of 14 vehicles, minus emissions generated by the production, transport and application of the 15

materials used in the rehabilitation process. The results showed that there was a reduction in 16 CO2 equivalents emissions. The results were based on a comparison between a NAASRA 17 value of 110 (266 in/mi) and any IRI resulting from different types of pavement 18 rehabilitation. According to them, the most beneficial rehabilitation in terms of reducing 19

greenhouse gas emissions from vehicles was the concrete rehabilitation which was assumed 20 to have NAASRA value after rehabilitation of 40 (99 in/mi). 21

Using MDSHA rating system, this comparison was basically between a pavement in 22 fair condition (99 in/mi) and a pavement in poor condition (266 in/mi). A NAASRA value of 23 less than 70 is considered as very good, 70 to 110 is considered as good, 110 to 150 is 24

considered as fair, and greater than 150 is considered as poor (12). Not only their rating 25 system is different from MDSHA’s, they did not make any comparison with a pavement in 26

good condition or better as per MDSHA criteria (<95 in/mi), in their computations they did 27 not use any emissions simulation software either. 28

Another study on the effect of road works on global warming gave very useful but 29 brief insights on the relationship between roadway profile condition and gas emissions (23). 30 The unit of measurement used for the roadway profile was a specific indicator very similar to 31 IRI. This study showed that there is a relationship between the roadway profile and the fuel 32

consumption. However, this relationship was based on very limited information. The data 33 was collected from a single location. In addition, they neither measured vehicle emission 34 directly nor utilized emission modeling software to calculate the emission as an alternative 35 option. They assumed that gas emissions are correlated with fuel consumption. Although it 36 was concluded that the extra fuel consumption due to poor pavement could be significant, 37

there is lack of sufficient data to support this statement. 38 Both studies seem to agree that higher pavement roughness values are associated with 39

higher gas emissions. Although one of the studies even shows that a reduction in GHG 40 emissions is possible if the pavement is rehabilitated, the pavement roughness criteria are 41 different from the criteria used in MDSHA and in this paper. The other study has no 42 convincing evidence suggesting that texture induce increased fuel consumption and hence 43 greater emissions. In addition, both studies did not use any software to generate GHG 44

emissions. 45

Pavement Roughness and Fuel Consumption 46 Some studies have confirmed that a relationship exists between fuel consumption and 47

pavement roughness despite the fact that there is no numerical formula. 48



A change in pavement roughness from smooth to rough can increase fuel consumption 1

by as much as 12% (24). However, a numerical relationship cannot be determined due to 2 external factors that make it difficult to isolate the effect that pavement roughness alone has 3 on fuel consumption. 4

Similarly, the National Center for Asphalt Technology (NCAT) reported that a change 5

in IRI from 1.08 m/km (68 in/mi) to 1.18 m/km (75 in/mi), resulted in a 2% fuel consumption 6 increase (25). This change, however, did not account for the vehicle age (26); if this is 7 isolated from the calculation, fuel economy would likely be less sensitive to changes in 8 roughness than reported values (27). 9

A study performed by the University of Texas at Arlington investigated differences 10

that may exist in fuel consumption and CO2 emissions when operating an automobile on an 11 Asphalt Concrete versus a Portland Cement Concrete pavement under city driving conditions 12 (28). They considered pavement type, IRI, roadway grade, vehicle’s acceleration and vehicle 13 speed in their analysis. Although they considered IRI in their study, their objective was to 14

relate gas consumption rates to pavement types and provide considerations of costs and 15 savings in the life cycle analysis of alternative pavement designs for city streets. They did not 16 demonstrate whether IRI was related to fuel consumption or CO2 emissions. 17

An exact relationship correlating pavement roughness to fuel consumption may not 18 have been established but the few existing sources have been able to link the two (27). 19

Software 20 There are software programs such as Mobile6.2 and MOVES2010a that are used to estimate 21 the amount of gas emissions in the atmosphere. Mobile6.2 is an emission factor modeling 22 software for vehicles created by EPA. Mobile6.2 is from the 6th generation of Mobile 23

software series. The first generation of this model was released in 1979. Mobile models have 24 been used since then and MOVES2010a will be a substitute by spring 2012 (29). For state 25

implementation plans and conformity analyses, states and various agencies use the model for 26

developing vehicular emissions inventories as a mandatory requirement. 27

Mobile6.2 predicts emission factors for several pollutants such as several hydrocarbon 28 (HC) categories, including volatile organic compounds (VOC), carbon monoxide (CO), 29 oxides of nitrogen (NOx), sulfur dioxide (SO2), and 10-micron particulate matter (PM10). The 30

model accounts for various parameters, including vehicle types, temperature, vehicle speeds, 31

and inspection/maintenance (I/M) programs to generate current emission factors. 32 MOVES2010a was developed recently by the EPA to be a substitute for Mobile6.2. 33

MOVES2010a estimates emission factors and inventories of several pollutants produced by 34 cars, trucks, buses, and motorcycles. It is an improved version of Mobile6.2 in many different 35 ways. MOVES2010a is also used to estimate highway vehicles gas emissions. Many States 36

are currently reviewing the model for future applications (29). 37 Tang et al. utilized Mobile6.2 to estimate air toxic emissions on different road classes 38

and found that air toxic emission factors per vehicle miles traveled (VMT) reduces when the 39 percentage of VMT on a freeway increases (30). Typically, the freeway system is held at 40

higher standards compared to other types of roads. The pavement condition on freeways 41 usually meets or exceeds the acceptable standards. Knowing the quality of maintenance on 42 freeways is generally high and therefore the pavement roughness is low, it may be interpreted 43

that air toxic emissions are generally coming from roads with higher pavement roughness. 44 Although several studies have been conducted on pavement roughness, vehicle speed, and 45 gas emissions, none of these studies examined the relationship between pavement roughness 46 and gas emissions. This relationship is investigated in this study. 47

48



METHODOLOGY 1

There is a two-step approach involved in this study: data collection and data analysis. Data 2 collection involves obtaining IRI information for state roads in Baltimore County and 3 collecting average speeds on selected road segments. Data analysis includes determining the 4

significance of the relationship between the average vehicle speeds based on the IRI 5 categories, estimating emissions, and defining the correlation if any and its significance 6

between IRI and GHG emissions. 7

Data Collection 8 Data collection consists of obtaining data from MDSHA and Maryland Department of 9 Environment (MDE) records as well as from field. Stored MDSHA and MDE records 10 include roadway functional classification, IRI, Highway Location Reference (HLR), and 11

posted speed limits. Field data comprises information regarding the actual vehicle’s speeds 12

on a second by second basis as the vehicle is in motion. 13

International Roughness Index Data 14 The Maryland State Highway Administration utilizes two types of systems to store IRI data: 15

RoadCare and PMBase database. RoadCare was developed by Applied Research Associate 16 (ARA) (31). This software program provides a fast way to look for specific pavement criteria 17 such as IRI for pre-determined roadway segments. The PMBase is a data retrieval program 18 for performance and other highway related data maintained in the Pavement Management 19

Database. 20 We collected two sets of data. The first set of IRI data collected in the summer of 21

2011 was done using RoadCare. The second set of data collected in the fall of 2011was 22 performed using the PMBase database. Unlike RoadCare, the PMBase database allows to 23 input the beginning and ending limits for specific segments. The results of this study are only 24

based on the second set of data as the speed data from the first set of segments were affected 25

by additional factors making the data unreliable. More details on this are provided in the later 26 sections. The 2010 data for all roadway segments in Baltimore County including IRI and 27 roadway functional classifications were exported to an excel file. 28

In this study IRI was classified into three categories: poor (> 170), fair (≥ 95 and 29 ≤170), and good (< 95). For the purpose of this study, roadway segments including all IRI 30

categories were analyzed in order to complete a thorough study. The next section provides 31

details on the technique used to determine the sample size. 32

Roadway Selection Sampling Method 33

According to the 2010 Highway Location Reference, there are 75 roads in Baltimore County 34 maintained by MDSHA. Out of the 75 roads, 7 are “Interstates” roads, 65 are “Maryland” 35

roads and 3 are “United States” roads. 36

37



Equation 3 was used to compute the number of road samples, n, from which segments 1

were selected. The sample size, n = 12, used in the analysis corresponds to a margin of error 2

of 26% at the 95% confidence level as shown below: 3

N

1s1

sn ,

2

2

c

)p1)(p)(z(s (Equation 3) 4

where, 5

N = 75 (population size) 6 Z = 1.96 (standard score for 95% confidence level) 7 p = 0.5 (percentage of picking a choice) 8 c = 0.26 (margin of error) 9

10

Once we had the sample size, we stratified the population based on the road type and 11

then distributed the sample accordingly. The results are depicted in Table 1. 12

TABLE 1 Number of Road Samples 13

Baltimore County Roads Samples

Interstates 7 (9%) 2

Maryland Roads 65 (87%) 9

US Routes 3 (4%) 1

Total 75 (100%) 12

14

Following the stratification, a simple random sampling using Excel was performed to 15 select 12 samples. Once all roads were selected, we generated IRI values. IRI data was 16

produced for every 0.1 mile and for the entire length of the road. Our segment selection 17 process consisted of picking a segment that was at least 0.1 mile and located at least 1000 feet 18

away from a traffic signal, stop signs, or from major ramps. 19 Sites visits were scheduled as soon as all segments locations were finalized. We 20

collected speed data during peak period as well as off-peak period. However, the peak hour 21 data was very unstable due to heavy congestion on few selected segments. Therefore, we 22

excluded them from the study. The final segment sample size included 33 segments for the 23 off-peak period. Two or three iterations of data collection were performed for each segment 24

and the average was calculated. Data from off-peak period are presented in Table 3. 25

Speed Data Collection 26 Vehicle speeds are very crucial to this study. They are the intermediary between the 27 pavement roughness and GHG emissions. Vehicle speeds were collected using a GPS 28

device. The GPS device was recording longitude, latitude information, and spot speeds for 29 every 1-second interval. 30

During the summer data collection, we used five drivers. The data collection in the 31 fall involved only one driver. Unlike in the summer where drivers were not given any driving 32

instructions, in the fall, the driver was instructed to follow the flow of traffic. Not following 33 the flow of traffic was another reason of excluding the summer data from the study. We 34 collected all the data in a span of six days. This includes four days of collecting peak time 35 data and two days of off peak data. The weather condition was dry during all six days. The 36 average speed of each road segment was calculated and posted speed limits were observed. 37

38



Data Analysis & Results 1 Data analysis was undertaken in two steps: (1) determining the significance of the 2 relationship between the average vehicle speeds based on the IRI categories; (2) defining the 3 correlation if any and its significance, between IRI and GHG emissions. 4

Relationship between Average Vehicle Speeds based on IRI Categories 5 In order to investigate the net effect of IRI on speed, the posted speed limit should be 6 constant during the analysis. However, further classification of the roads into posted speed 7

limit would reduce the sample size in each category dramatically. To avoid such a situation, 8 the relative average speed which is average speed divided by posted speed limit was 9

introduced in the analysis (Table 2). 10

We conducted F-tests and t-tests to compare the variances and the means (µ) of the 11 relative speeds for the IRI categories. In order to test if the means of the observed relative 12 speeds are the same for the two IRI categories (poor vs. good) considered, a two-tail t-test is 13

required. However, since we expect the speed on roads with high IRI values to be less than or 14 equal to the speed on roads with low IRI values, one-tail t-tests are proper as well. The null 15

and alternative hypotheses for one-tail and two-tail tests are: 16

One-tail test 17

ji0 :H 18

ji1 :H , where i, j: poor, fair, good; i≠j 19

Two-tail test 20

ji0 :H 21

ji1 :H , where i, j: poor, fair, good; i≠j 22

Based on the F-test analysis on the speed variances for the poor and good IRI categories, a p-23 value of 0.09 was obtained, which is considered not significant at the 5% level of 24 significance. The F-test results for poor and good IRI segments are presented in Table 3. The 25 results from the t-test assuming two equal variances (based on the F-test results) are presented 26

in Table 4. 27

28



TABLE 2 Speed Data during off-peak period 1

Route Type Route No. BMP EMP

Length of

Segment

(miles)

Direction Functional Classification Speed Limit IRI Value IRI Category Average Speed Grade Relative Speed

MD 128 0.22 0.588 0.368 WB Urban OPA 30 199 Poor 42.3 -0.41% 1.41

MD 150 3.54 3.74 0.2 WB Urban OPA 30 73 Good 39.3 1.13% 1.31

MD 7 3 3.6 0.6 EB Urban Minor Arterial 35 206 Poor 40.9 1.02% 1.17

MD 695A 0.51 1.05 0.54 SB Urban OPA 35 176 Poor 41.6 -0.24% 1.19

MD 25 2.69 3.05 0.36 NB Urban Minor Arterial 40 200 Poor 47.9 -0.78% 1.20

MD 26 1.287 1.797 0.51 EB Rural Minor Arterial 40 199 Poor 44.3 0.47% 1.11

MD 26 4.197 4.847 0.65 EB Urban OPA 40 255 Poor 43.5 -0.96% 1.09

MD 45 4.63 4.96 0.33 SB Urban OPA 40 260 Poor 40.2 0.47% 1.00

MD 45 5.85 6.06 0.21 SB Urban OPA 40 241 Poor 43.2 1.40% 1.08

MD 695A 0.51 1.05 0.54 NB Urban OPA 40 195 Poor 45.4 -0.21% 1.13

MD 7 7.91 8.41 0.5 WB Urban Minor Arterial 40 141 Fair 48.2 -2.12% 1.20

MD 25 0.31 0.92 0.61 NB Urban Minor Arterial 40 97 Fair 44.6 4.05% 1.12

MD 26 0.163 1.083 0.92 EB Rural Minor Arterial 40 96 Fair 54.8 3.76% 1.37

MD 128 4.7 5.24 0.54 WB Rural Major Collector 40 139 Fair 49.7 -0.76% 1.24

MD 150 9.26 9.83 0.57 WB Urban Minor Arterial 40 109 Fair 45.3 0.30% 1.13

MD 695A 1.4 1.9 0.5 SB Urban OPA 40 133 Fair 45.2 0.00% 1.13

MD 25 4.97 5.65 0.68 NB Urban Minor Arterial 40 76 Good 48.5 0.69% 1.21

MD 25 8.73 9.53 0.8 NB Rural Major Collector 40 88 Good 48.8 -2.11% 1.22

MD 150 8.61 9.12 0.51 EB Urban Minor Arterial 40 65 Good 44.6 -0.19% 1.11

MD 128 4.12 4.35 0.23 WB Rural Major Collector 45 101 Fair 50.7 1.07% 1.13

MD 41 1.24 1.52 0.28 NB Urban Freeway Expressway 45 61 Good 48.0 0.19% 1.07

MD 45 15.03 15.436 0.406 NB Rural Major Collector 45 80 Good 45.0 1.32% 1.00

MD 128 2.6 2.74 0.14 WB Rural Major Collector 45 89 Good 50.3 1.61% 1.12

MD 128 4.42 4.62 0.2 WB Rural Major Collector 45 87 Good 50.7 3.24% 1.13

MD 700 1.41 1.62 0.21 NB Urban Minor Arterial 50 192 Poor 48.3 0.09% 0.97

US 40 14.7 14.93 0.23 WB Urban OPA 50 159 Fair 54.1 1.61% 1.08

US 40 15.42 15.76 0.34 WB Urban OPA 50 74 Good 54.1 0.49% 1.08

US 40 23.62 23.87 0.25 WB Urban OPA 55 186 Poor 58.1 5.41% 1.06

US 40 24.95 25.14 0.19 WB Urban OPA 55 232 Poor 53.2 0.27% 0.97

IS 83 0 2.27 2.27 SB Rural Interstate 55 117 Fair 66.1 -0.24% 1.20

IS 83 11.01 13.26 2.25 SB Urban Interstate 55 90 Good 66.8 0.78% 1.22

IS 95 19.7 20.47 0.77 NB Urban Interstate 65 102 Fair 64.2 -0.46% 0.99

IS 95 21.74 23.63 1.89 NB Urban Interstate 65 45 Good 68.1 1.27% 1.05 2



1

TABLE 3 F-Test Results of Comparing Variances of Poor IRI Roads and Good IRI Roads 2

Poor Good

Mean 1.1200 1.1429

Variance 0.0184 0.0113

Observations 34 30

Df 33 29

F 1.6312

P(F<=f) one-tail 0.0922

F Critical one-tail 1.8369

3 4

TABLE 4 t-Test Results of Comparing Mean of the Poor IRI Roads and Good IRI Roads 5

Assuming Equal Variances 6

Poor Good

Mean 1.12004 1.14286

Variance 0.01836 0.01125

Observations 34 30

Pooled Variance 0.01504

Hypothesized Mean Difference 0

Df 62

t Stat -0.74281

P(T<=t) one-tail 0.23020

t Critical one-tail 1.66980

P(T<=t) two-tail 0.46040

t Critical two-tail 1.99897

7

The t-test results indicate that at the 95 percent confidence level and for the range of 8 IRI values analyzed the mean relative speeds from poor IRI roads are not statistically 9 different compared to the mean from the good IRI roads. A p-value of 0.230 was obtained 10 from the two-tail test. Notwithstanding, the mean relative speed value for the poor IRI 11 category is slightly less than that for the good IRI category. 12

Correlation between IRI and Gas Emissions 13 The analysis in the previous section showed a slight increase (albeit statistically insignificant 14 at the 95 percent confidence level) in the relative speeds when going from poor IRI category 15

roads to good IRI category roads. Consequently, we proceeded with the second step and 16

calculated gas emissions quantities to analyze their relationship with IRI. 17

MOVES2010a Input Data 18

We computed gas emissions using MOVES2010a that provides three analysis scales: 19 national, county and project. All the analyses in this study were done at the project scale, as it 20

is suitable for the link level. The main inputs required were link source type, link length, link 21 volume, link average speed, link grade, vehicle age distribution, meteorology data, fuel 22 supply, fuel formulation, and inspection/maintenance. The last four inputs were defaults 23 values. As the roadways under study had different characteristics such as grade, the estimated 24 emissions would be affected by these variables, in addition to speed. Therefore, we combined 25



the roads based on their posted speed limit and IRI, and kept all other variables but average 1

speed fixed for all categories. 2 3 Link Average Speed: We computed the average speed for each segment based on the 4 collected second-by-second speed using the GPS device. Then, we grouped the data based on 5

posted speed limits and IRI categories, and calculated the average speeds for each group. 6 Only three groups of roads (40MPH, 50MPH and 55MPH) include speed information for 7 both poor and good IRI categories. As presented in Table 5, an increase in the average speed 8 was observed when going from roads in poor IRI category to those in good IRI category. 9 10

Link Length & Link Grade: Using the GPS coordinates, Google Earth, and the MDSHA 11 Highway Location Reference (HLR) (32), we were able to estimate the link length and grade 12 of each link. The HLR provides a method for identifying roadway locations by the use of 13 milepoints. The link length is the milepoint change between the starting point and the ending 14

point. We calculated the grade using the change between gain and loss in elevations from 15 Google Earth divided by the link length. We used an average value of 0.7 mile for link length 16 and 0.83% for link grade for both the poor category and the good category. 17

18 Link Volume & Link Source Type: We assumed a volume of one vehicle per hour and thus 19

only one vehicle type, in order to calculate the emission only for the vehicle that was driven 20 on the road segments. 21

22 Age Distribution: The vehicle’s age is 8 years. 23 24

Calculation Type: Inventory 25 26

Time Spans: 10am – 11am, weekend, November 2011 27

28

Geographic Bounds: Baltimore County in the state of Maryland 29

30 On Road Vehicle and Equipment Panel: Gasoline Passenger Car 31 32 Road Type : The 55MPH group was classified as urban restricted access. All other groups 33

were classified as urban unrestricted access. 34 35

Pollutants and Processes: Although we calculated the emission for all pollutants, we only 36 used CO2 Equivalent as it relates to GHG (17). Information on other pollutants not used in 37 this study could be used in future studies. 38

MOVES2010a Results 39

In general, the results we obtained were consistent with our expectation (see Table 5). With 40 the exception of the 55 mph group where increased speed associated with lower IRI values 41 resulted in increased gas emissions, validated by other studies indicating increase in 42

emissions when speed increases beyond 55 MPH (33), increased average speeds associated 43 lower IRI values generally translate into decreased gas emissions. The difference (Δ) in 44 emission quantity obtained in comparing roads from poor IRI category versus roads from 45 good IRI category is also included in Table 5. 46

47 48



TABLE 5 MOVES2010 Results Per Passenger Car 1

POSTED

SPEED

(MPH)

AVERAGE SPEED

(MPH)

CO2 Equivalent

EMISSIONS

(GRAMS)

CHANGE IN CO2

Equivalent

EMISSIONS

(GRAMS)

Poor Good Poor Good ΔPoor to Good

30 42.34 - 245 - -

35 41.22 - - - -

40 44.08 47.04 243 239 -4

45 - 48.53 - - -

50 48.28 54.10 238 233 -5

55 55.67 66.84 231 241 +10

65 - 68.10 - - -

2

CONCLUSION 3

In order to find the impact of road roughness on GHG emissions, several data such as 4

vehicle speeds and IRI values were collected on selected roads in Baltimore County, 5 Maryland. The t-test results indicate that the mean relative speeds from roads in poor IRI 6 roads are not statistically different from roads in good IRI roads. IRI values for all segments 7

analyzed are between 45 and 260. The results confirm those obtained by Paterson and 8 Symonds, who contended that operating speeds are usually not affected at less than 160 NRC 9

(IRI value of 386 in/mi)(18). Furthermore, Symonds et al. concluded that travel speeds are 10 usually not affected at less than 120 NRC (IRI value of 290 in/mi)(20). Our recommendation 11

for future studies is to consider road segments having IRI values greater than 290 and 12 collection the data during off-peak periods. Collecting data during peak requires considering 13

so many factors that make it very difficult to capture the effect of pavement roughness alone 14 on the speed. 15

Note that the emission values in Table 5 are for a single vehicle, and the total 16

quantities would be proportional to the traffic volume on the road. For example, improving 17 IRI from poor to good condition on a road with 40 mph speed limit and an average traffic 18

volume of 1000 vehicles per hour will result in a decrease of the GHG emissions by 4kg per 19 hour, 96 kg per day, or about 35,040 kg per year. The annual reduction can be estimated 20

similarly for all other roads and traffic conditions. IRI affects not only ride quality and 21 vehicle operating costs but also affects gas emissions. This paper has shown the importance 22 of road maintenance, associated with low IRI values, in possibly reducing GHG emissions. 23 Transportation agencies, particularly those located in the nonattainment areas, are encouraged 24

to upgrade their roads from poor condition to at least fair condition. 25 It is important to remember that the results discussed herein were derived under the 26

assumption that all other inputs but the average speed were held constant. In real world, this 27

is generally not the case. 28

29



REFERENCES 1

1) U.S. Energy Information Administration Greenhouse Gases, Climate change, and 2 Energy http://www.eia.gov/oiaf/1605/ggccebro/chapter1.html Accessed October 14, 3 2011. 4

2) Maryland Department of the Environment. Maryland Greenhouse Gas Emissions 5 Inventory 1990. 2001. 6

3) Browning, Louis, et al. Representing Freight in Air Quality and Greenhouse Gas 7 Models. Washington, DC: Transportation Research Board, 2010. p. 30, Project 8 NCFRP-16. 2010. 9

4) Environmental Protection Agency. Inventory Of U.S. Greenhouse Gas Emissions And 10 Sinks: 1990 - 2009 http://epa.gov/climatechange/emissions/downloads11/US-GHG-11

Inventory-2011-Complete_Report.pdf Accessed October 15, 2011. 12 5) Environmental Protection Agency. Basic Information Transportation and Climate US 13

EPA. US March 2006. http://www.epa.gov/oms/climate/basicinfo.htm. Accessed July 14

23, 2011. 15 6) Shepherd, Tim and Ways, Marcia. Vehicle Miles Travelled and Air Pollution in 16

Maryland. MDE. December 2009. 17

http://www.mde.state.md.us/programs/ResearchCenter/ReportsandPublications/Pages/18 ResearchCenter/publications/general/emde/vol4no2/vmt.aspx. Accessed July 30, 19 2011. 20

7) Environmental Protection Agency. Climate Change and Health Effects 21 http://www.epa.gov/climatechange/downloads/Climate_Change_Health.pdf Accessed 22

October 16, 2011. 23 8) International Energy Agency (IEA). CO2 Emissions from Fuel Combustion 24

HIGHLIGHTS (2010 Edition), Paris. www.iea.org/co2highlights/co2highlights.pdf. 25

Accessed October 15, 2011. 26

9) Environmental Protection Agency. Health and Environmental Effects. 27 http://www.epa.gov/climatechange/effects/index.html. Accessed October 16, 2011 28

10) Environmental Protection Agency. Frequent Questions – Climate Change 29

http://www.epa.gov/climatechange/fq/emissions.html. Accessed October 16, 2011 30 11) Fuel Economy. Many Factors affect MPG. 31

http://www.fueleconomy.gov/feg/factors.shtml. Accessed November 2010. 32 12) Bureau of Transport and Communications Economics, Roads, Vehicle Performance 33

And Greenhouse: Costs And Emission Benefits Of Smoother Highways, 1997. 34 www.bitre.gov.au/publications/05/Files/WP032.pdf. Accessed October 15, 2011. 35

13) Environmental Protection Agency. Metrics for Expressing Greenhouse Gas 36 Emissions: Carbon Equivalents and Carbon Dioxide Equivalents. 37 http://www.epa.gov/otaq/climate/420f05002.htm. Accessed October 16, 2011 38

14) Santero, N., Masanet, E., and Horvath, A., Life-Cycle Assessment of Pavements: A 39

Critical Review of Existing Literature and Research, SN3119a, Portland Cement 40 Association, Skokie, Illinois, USA, 2010, 81 pages. 41

15) Elvik, R., Hoye A., Vaa, T., and Sorensen M., The Handbook Of Road Safety 42

Measures Bingley, UK : Emerald, 2009,1124 pages 43 16) Shafizadeh, Kevan, Mannering, Fred and Pierce, Linda. A Statistical Analysis Of 44

Factors Associated With Driver-Perceived Road Roughness On Urban Highways. 45 Springfield: National Technical Information Service, 2002. 46

17) Maryland State Highway Administration Pavement and Geotechnical Division of the 47 Office of Materials Technology, 2010 Pavement System Preservation Report, 48

Maryland, July 2011 49



18) Paterson, W and Watanatada, T. Relationships Between Vehicle Speed, Ride Quality, 1

and Road Roughness. Measuring Road Roughness and Its Effects on User Cost and 2 Comfort. Philadelphia: American Society for Testing and Materials, 1985, pp. 89-110. 3

19) Austroads, Austroads Test Method Ag:Am/T001 Pavement Roughness Measurement 4 With An Inertial Laser Profilometer, 2007 5

http://www.austroads.com.au/pdf/TestMethod1/AGAMT001_Roughness_survey.pdf. 6 Accessed October 16, 2011. 7

20) Symonds Travers Morgan Pty Ltd and Cox J.B. , Effect of Road Surface 8 Condition on Vehicle Operating Costs in Australia : Literature Review and Fleet 9 Database Analysis, Report to Austroads, 1996 10

21) Zaniewski, J and Butler, B. Vehicle Operating Costs Related to Operating Mode, 11 Road Design, and Pavement Condition. Measuring Road Roughness and Its Effects on 12 User Cost and Comfort. Philadelphia: American Society for Testing and Materials, 13 1985, pp. 127-142. 14

22) Barnes, G and Langworthy, P. Per Mile Costs Of Operating Automobiles And Trucks. 15 2004, Journal of the Transportation Research Board, pp. 71-77. 16

23) Lepert, Philippe and Brillet, Francois. The overall effects of road works on global 17

warming gas emissions. Transportation Research Part D: Transport and 18 Environment. 2009, pp. 576-584. 19

24) Sandberg, U.S.I., Road Macro- and Megatexture Influence on Fuel Consumption, in 20 Surface Characteristics of Roadways: International Research and Technologies, W.E. 21

Meyer and J. Reichert, Editors. American Society for Testing and Materials: 22 Philadelphia, PA. pp. 460-479. 1990. 23

25) Hugo, F. and Martin, A.E., Significant Findings from Full-Scale Accelerated 24

Pavement Testing. National Cooperative Highway Research Program and 25 Transportation Resarch Board. NCHRP Synthesis 325. 2004. 26

26) Gillespie, J.S. and McGhee, K.K., Get In, Get Out, Come Back! Transportation 27

Research Record. Vol. 1990, pp. 32-39. 2007. 28

27) Santero, N., Pavements and the environment: A life-cycle assessment approach. Ph.D. 29 dissertation, University of California, Berkeley, In Dissertations & Theses: Full Text 30

[database on-line); available from http://www.proquest.com (publication number 31 AAT 3369133; Accessed September 18, 2011). 32

28) Ardekani, Siamak and Sumitsawan, Palinee. Effect of Pavement Type on Fuel 33

Consumption and Emissions in City Driving. Civil Engineering, University of Texas 34 at Arlington. Arlington, 2010. 35

29) Baker Jr., Michael. Air Quality Conformity Analysis for Fine Particulates (PM 2.5) 36 NAAQS. 2010. 37

30) Tang, Tianjia, et al. MOBILE6.2 Modeling of Exhaust Air Toxic Emission Factors. 38

2005, Journal of the Transportation Research Board, pp. 99-106. 39 31) Applied Research Associates. ARA News Articles. Undated. 40

http://www.ara.com/Newsroom_Whatsnew/press_releases/mdsha.html. Accessed July 41 17, 2011. 42

32) Maryland State Highway Administration. Highway Location Reference. Maryland 43 State Highway Administration. 2010. 44 http://sha.maryland.gov/pages/hlr.aspx?PageId=832. Accessed May 25, 2011. 45

33) Transportation Research Board. Expanding Metropolitan Highways: Implications for 46

Air Quality and Energy Use. TRB Special Report 245. 1995 47

48


Evaluation of the Impact of Pavement Roughness on …docs.trb.org/prp/12-2872.pdfEvaluation of the Impact of Pavement Roughness on Vehicle Gas Emissions in Baltimore County ... 12

Documents