Development and Influence of Statewide

Axle Load Spectra on Flexible

Pavement Performance
Nam H. Tran and Kevin D. Hall
traffic data are converted into a number of ESALs on the basis of
load equivalency factors (LEF). Typically, LEF values are different
for exible and rigid pavement design.
The MEPDG requires identical traffic data for different pavement
types (flexible or rigid) and design types (new or rehabilitated).
The typical traffic data required for the MEPDG are categorized as
follows (2):

The MechanisticEmpirical Pavement Design Guide (MEPDG) developed

under NCHRP Project 1-37A requires new inputs for traffic characterization. One important traffic input is axle load distribution factors, or
axle load spectra. These spectra represent the percentage of the total axle
applications within each load interval for single, tandem, tridem, and
quad axles. The Arkansas State Highway and Transportation Department sponsored research to develop statewide axle load spectra and
evaluate the significance of the developed inputs in the MEPDG. Of 25
weigh-in-motion sites selected for this study, only 10 stations provided
good weight data for development of statewide axle load spectra. On the
basis of the available weight data, statewide axle load spectra for single,
tandem, and tridem axles were developed. However, the data contained
few quad axles; therefore, statewide quad-axle load spectra were not
generated. A sensitivity analysis related to the axle load spectra showed
a significant difference in predicted pavement performance resulting
from the statewide and MEPDG default axle load spectra. Therefore, the
state-specific axle load spectra are recommended for implementation of
the MEPDG in Arkansas and updated periodically unless no significant
changes are observed in the future.

Truck traffic volume (base-year information),

Truck traffic volume adjustment factors,
Axle load distribution factors, and
General traffic inputs.

The Arkansas State Highway and Transportation Department

(AHTD) sponsored research to develop statewide axle load spectra
and evaluate the signicance of the developed inputs in the MEPDG.
This paper presents the results of the study and a recommended
guidance for using the developed traffic inputs.

INPUTS REQUIRED FOR

TRAFFIC CHARACTERIZATION

The 1993 edition of the AASHTO Guide for Design of Pavement

Structures (hereinafter called the 1993 Guide) (1) has served well as
a design guide for many state highway agencies in designing new
and rehabilitated pavements for several decades. Design procedures
in the 1993 Guide were developed on the basis of AASHO road tests
in the 1960s. Since then, traffic, materials, construction practices,
and other parameters affecting pavement performance have changed
signicantly. A new MechanisticEmpirical Pavement Design Guide
(MEPDG) has been developed under NCHRP Project 1-37A (2).
Pavement design and analysis procedures in the MEPDG represent
a significant departure from the empirical procedures in the 1993
Guide. One change relates to traffic inputs; the MEPDG requires the
use of axle load distribution factors, or axle load spectra, rather than
the traditional equivalent single-axle load (ESAL) input in current
procedures. State agencies currently collect several types of traffic
data by means of various traffic data collection devices, such as static
weigh stations, automatic traffic recorders (ATRs), automatic vehicle
classiers (AVCs), and weigh-in-motion (WIM) scales. The mixed

The MEPDG requires four basic categories of traffic inputs for the
structural pavement design (2). First is base year traffic volume.
One important input in this category is annual average daily truck
traffic (AADTT) of Vehicle Classes 4 through 13. This information
can be derived from WIM, AVC, or vehicle count data and is typically available within a state highway agency. The use of the average AADTT for the 3 years immediately before the design base year
is recommended (2).
Second, the base year AADTT must be adjusted by using traffic
volume adjustment factors, including monthly distribution, hourly
distribution, class distribution, and traffic growth factors. These factors can be determined on the basis of the traffic counts obtained from
WIM, AVC, or vehicle count data. NCHRP Project 1-37 researchers
found that roadways within the same functional classication had
signicant variability in truck distribution. Thus, a new grouping
system based on 17 truck traffic classification (TTC) groups was
introduced to define roadway groups in MEPDG (2). The TTC system is based on the distribution of four truck groups: buses, singleunit trucks, single trailers, and multitrailers. Class distribution
factors for a pavement design should be selected on the basis of the
roadway truck traffic classification.
The third input category is axle load distribution factors (axle load
spectra). The axle load distribution factors represent the percentage
of the total axle applications within each load interval for a specic

Department of Civil Engineering, University of Arkansas, 4190 Bell Engineering

Center, Fayetteville, AR 72701. Corresponding author: N. H. Tran, nhtran@uark.edu.
Transportation Research Record: Journal of the Transportation Research Board,
No. 2037, Transportation Research Board of the National Academies, Washington,
D.C., 2007, pp. 106114.
DOI: 10.3141/2037-10

106

Tran and Hall

axle type (single, tandem, tridem, and quad) and truck class. The axle
load spectra can be determined only from WIM data. Analyses of
WIM data in the Long-Term Pavement Performance (LTPP) program database showed that the differences between year-to-year and
month-to-month load spectra were not signicant. Thus, for MEPDG,
axle load spectra can be normalized on an annual basis (2).
The last traffic input category includes general traffic inputs, such
as number of axles per truck, axle conguration, and wheel base.
These data are used in the calculation of traffic loading for determining pavement responses (2). The default values provided for the general traffic inputs are recommended if more accurate data are not
available.
A traffic data collection plan for the MEPDG should follow the
practices outlined in the FHWAs Traffic Monitoring Guide (TMG)
(3). State agencies that currently collect traffic data in accordance
with the TMG would meet the traffic characterization requirements
for the MEPDG. However, in some situations, agencies may not
properly collect detailed traffic data for accurate characterization
of historical and future traffic for design purposes. Thus, the level of
detail of available traffic data is considered in determination of the
reliability of the pavement design by using a hierarchical approach.
Three levels of traffic inputs are defined in MEPDG on the basis
of the detail of traffic data available. It is anticipated that the last
two levels are the most commonly used for both new pavement and
rehabilitation designs (2).

Level 1
Level 1 is considered the most accurate, and thus most reliable, trafc data available, requiring detailed knowledge of historical axle
load spectra, classication, and volume data at or near the project
site. Typically, this level refers to a roadway segment near the design
location with no inuencing intersecting roadways.

107

program and a short-duration count program. The continuous-count

program includes 79 automated traffic data collection sites. Of the
79 automated sites, 55 feature WIM collection, which is used for
continuous collection of traffic volume, vehicle classication, and
vehicle weight. All WIM sites in Arkansas use piezoelectric sensors
and are calibrated every 3 years. The calibration is performed in
accordance with the guidelines in the TMG (3).
Traffic data collected from the 55 WIM sites from 2003 through
2005 were provided for this study. The data le format used in this
study is the vehicle weight record (W-record), ssyy.WGT. The format for this record can be found in the TMG (3). Each vehicle weight
le contains a 1-month record of passing vehicles with axle weight
and spacing data.
Selection of WIM sites for development of traffic inputs in MEPDG
is based on availability of data collected at each station. A preliminary check of the data showed that for many stations sufficient data
were not available for several months in a given year. As a result,
only 25 sites provided sufficient data for the evaluation of monthly
traffic variation; these 25 sites included 18 in rural areas and seven
in urban areas.
Several publications have reported that traffic data collected by
means of automated traffic collection sites often contain errors,
particularly data collected from WIM sites featuring temperaturedependent piezoelectric sensors (4, 5). Recently, the University of
Washington reported that only 11 of 52 WIM stations operated in
the University of Washington could provide data suitable for the
development of axle load spectra in the MEPDG (6 ). Therefore, the
traffic records identied for use in this studywhich had not been
checked for data qualitywere evaluated according to the quality
control procedure recommended in two LTPP and FHWA publications (3, 7 ). Among the 25 WIM sites previously selected, only
10 provided weight data suitable for the development of axle load
spectra for the MEPDG (8). Five sites exhibited uctuations in the
data, a situation indicating that the WIM scales had failed. Nine sites
showed signs suggesting that the WIM scale was out of calibration.
One site had very few trucks.

Level 2
Level 2 is the intermediate traffic input level. It requires substantial
knowledge of design traffic volume and vehicle classication. Level 2
uses statewide or regional axle load spectra, instead of the sitespecific axle load spectra required for Level 1. The analyses of
regional axle load spectra for each truck class must be completed
externally by design agencies.

Level 3
Level 3 is the least accurate or reliable input level in the MEPDG.
It requires only estimates of annual average daily truck traffic
(AADTT) and truck percentages with no site-specic knowledge of
traffic characteristics at the design location. An estimate of traffic
inputs based on local experience is also considered in Level 3.

TRAFFIC DATA AVAILABLE FOR DEVELOPMENT

OF TRAFFIC INPUTS
The traffic-monitoring program in Arkansas is performed and managed by the Technical Services section of AHTD, generally in accordance with the guidelines in the TMG (3). Currently, Technical
Services performs two traffic count programs: a continuous-count

DEVELOPMENT OF AXLE LOAD SPECTRA

Methodology
On the basis of the requirements for traffic inputs in MEPDG and
the available traffic data in Arkansas, it was decided to develop sitespecic AADTT and traffic growth factors (Level 1). Statewide trafc inputs that can be developed on the basis of available traffic data
include monthly, hourly, class, and axle load distribution factors.
Default values for other general traffic inputs should be used unless
more accurate data are available.
Development of traffic inputs is an intensive, analytical data reduction process. The procedure divides a massive traffic data set into
groups. Each group is manipulated to provide one input, or a set of
traffic inputs. Automation of the process is highly desirable. Traoad
(9), a computer program developed under NCHRP Project 1-39 for
generating traffic inputs for the MEPDG, was tried first. The program was able to import the data stored in the vehicle classication
(C-records) file format. However, the software could not read the
W-records provided by AHTD. The format of the W-records was
checked carefully, and no error was found. The software error is
unknown. Thus, the Traoad software was not used in this project.
For this study, a computer program named WEIGHT.xls was developed to reduce vehicle weight data and generate axle load spectra for

108

Transportation Research Record 2037

between 8,000 and 12,000 lb, which is the expected range for VC 9
single axles (3, 7). Extensive analyses of single-axle load spectra for
other vehicle classes showed similar results. Thus, it is reasonable to
group the single-axle load spectra for each vehicle class from all WIM
stations for developing statewide single-axle load spectra. Table 1
presents the statewide single-axle load spectra for Arkansas.

MEPDG. The WEIGHT.xls program used for development of axle

load spectra was programmed through the following general steps:
1. Open a W-record (one le contained weight data from all active
WIM stations throughout the state in 1 month).
2. Find the range containing all weight data from a specic WIM
station.
3. Count the number of axles in each weight bin for different
vehicle classes using the following load intervals:
Single axles (3,000 to 40,000 lb at 1,000-lb intervals),
Tandem axles (6,000 to 80,000 lb at 2,000-lb intervals), and
Tridem and quad axles (12,000 to 102,000 lb at 3,000-lb
intervals).
4. Summarize the monthly axle load distribution in Step 3 and
determine the axle load spectra for the site.
5. Save the axle load spectra to a file which could be read automatically into the MEPDG software.

Tandem-Axle Load Spectra

Figure 2 shows VC 9 tandem-axle load from the 10 WIM sites. The
distributions seem to have several clusters. The TTC system was
successfully used for grouping vehicle class distributions (8). However, additional analyses revealed that the tandem-axle load spectra
in the same TTC group did not exhibit a high degree of agreement.
An example is shown in Figure 3. The tandem-axle load spectra of
VC 9 within TTC Group 7 were considerably different.
Another cluster analysis attempted to group tandem-axle load
spectra. This grouping method is based on an observation that a distribution curve for VC 9 tandem-axle load spectra should have two
peaks:

The program can be used to develop axle load spectra for a specic
site. The developed axle load spectra can be used for Level 1 design
in the MEPDG software. The program was used to develop axle load
spectra on the basis of suitable weight data from 10 stations. Cluster
analyses were then performed to develop statewide axle load spectra
for Arkansas. These analyses are presented in the following section.

Tandem axles of unloaded VC 9 trucks and

Tandem axles of loaded VC 9 trucks.
Thus, tandem-axle load spectra could be divided into three groups:

Development of Statewide Axle Load Spectra

Load Spectrum Group 1, in which the numbers of unloaded and

loaded trucks are similar;
Load Spectrum Group 2, in which the number of unloaded
trucks is much greater than that of loaded trucks, and
Load Spectrum Group 3, in which the number of loaded trucks
is much greater than that of unloaded trucks.

The MEPDG software requires axle load spectra for four axle types,
including single, tandem, tridem, and quad axles. As noted earlier,
only 10 WIM sites in Arkansas provided suitable weight data for
developing axle load spectra. These represented TTC Groups 3, 6,
7, and 10. For Interstates and four-lane highways in Arkansas, the
primary truck class is vehicle class (VC) 9single trailer trucks
which comprise up to 70% of truck traffic (8). Thus, analyses for
developing Arkansas statewide axle load spectra were based on VC 9.

Figure 4 shows the distribution curves of VC 9 tandem-axle

load spectra for Load Spectrum Group 2. Compared with the VC 9
tandem-axle load spectra within the same TTC group shown in
Figure 3, the tandem-axle load spectra grouped in Figure 4 are
less variable. A similar cluster analysis was presented by Lu and
Harvey (10).
Even though the cluster analysis described previously appeared
to dene tandem-axle load spectrum groups well, dividing a small-

Single-Axle Load Spectra

Normalized Distribution (%)

Figure 1 compares VC 9 single-axle load spectra on the basis of

weight data from the 10 WIM sites. The curves are similar and peak

40.0
35.0
30.0
25.0
20.0
15.0
10.0
5.0
0.0
0

10

15

20

25

30

35

40

Axle Load (kips)

FIGURE 1

580236

71813

720034

740035

480037

170064

460006

230001

600870

430038

Single-axle load spectra for VC 9.

45

Tran and Hall

TABLE 1

109

Statewide Single-Axle Load Spectra

Vehicle Class

Axle Load (lb)

10

11

12

13

3,000
4,000
5,000
6,000
7,000
8,000
9,000
10,000
11,000
12,000
13,000
14,000
15,000
16,000
17,000
18,000
19,000
20,000
21,000
22,000
23,000
24,000
25,000
26,000
27,000
28,000
29,000
30,000
31,000
32,000
33,000
34,000
35,000
36,000
37,000
38,000
39,000
40,000
41,000

2.641
6.976
4.826
5.517
6.473
6.186
10.021
8.636
11.153
7.798
8.202
5.464
4.888
2.802
2.414
1.422
1.126
0.679
0.651
0.542
0.288
0.325
0.189
0.126
0.107
0.120
0.054
0.062
0.062
0.035
0.059
0.027
0.035
0.039
0.011
0.010
0.035
0.004
0.001

11.089
29.213
16.709
11.963
7.451
4.667
4.711
2.970
2.868
1.725
1.639
1.014
1.010
0.619
0.617
0.376
0.360
0.221
0.210
0.153
0.091
0.089
0.051
0.051
0.030
0.026
0.015
0.017
0.010
0.009
0.005
0.005
0.004
0.004
0.003
0.002
0.002
0.002
0.001

1.601
3.982
3.307
4.627
6.378
8.287
15.724
14.774
15.561
8.505
6.580
3.130
2.552
1.346
1.172
0.629
0.546
0.312
0.297
0.191
0.114
0.103
0.060
0.058
0.027
0.030
0.024
0.016
0.014
0.011
0.008
0.009
0.004
0.009
0.005
0.003
0.004
0.002
0.001

0.119
0.566
0.648
1.542
1.971
2.792
7.054
8.952
14.333
13.091
14.594
10.271
8.441
5.271
3.989
1.807
1.928
0.686
0.654
0.380
0.213
0.170
0.084
0.187
0.081
0.048
0.024
0.031
0.017
0.015
0.004
0.004
0.000
0.017
0.000
0.000
0.013
0.000
0.002

8.354
19.658
11.548
9.931
7.976
6.091
7.187
5.042
5.166
3.440
3.519
2.271
2.248
1.430
1.447
0.907
0.896
0.569
0.563
0.436
0.277
0.255
0.160
0.155
0.088
0.090
0.057
0.057
0.034
0.040
0.020
0.022
0.016
0.016
0.012
0.006
0.011
0.005
0.003

0.294
1.313
1.407
2.202
4.240
7.592
18.825
21.041
22.301
10.116
5.909
2.105
1.263
0.529
0.370
0.170
0.124
0.064
0.047
0.030
0.015
0.014
0.008
0.007
0.004
0.004
0.002
0.002
0.001
0.001
0.001
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000

1.058
3.705
3.216
4.082
6.172
8.285
15.948
14.930
16.425
9.422
7.215
3.269
2.389
1.109
0.881
0.438
0.380
0.228
0.197
0.149
0.101
0.089
0.060
0.070
0.030
0.020
0.034
0.037
0.003
0.004
0.014
0.017
0.007
0.001
0.003
0.002
0.008
0.000
0.001

0.361
1.545
2.874
6.806
9.282
9.211
14.091
10.435
10.618
7.004
7.571
5.018
4.886
2.967
2.774
1.517
1.233
0.614
0.462
0.274
0.148
0.104
0.050
0.048
0.016
0.016
0.008
0.008
0.024
0.011
0.010
0.009
0.001
0.001
0.001
0.001
0.001
0.000
0.000

0.481
2.602
4.130
10.036
13.239
12.182
16.428
11.665
10.666
5.755
5.254
2.471
1.960
0.999
0.834
0.396
0.337
0.155
0.161
0.069
0.040
0.042
0.017
0.012
0.017
0.008
0.007
0.008
0.003
0.003
0.003
0.010
0.010
0.001
0.001
0.000
0.001
0.000
0.000

6.155
10.304
5.735
6.908
6.701
6.324
8.673
5.102
7.266
4.097
6.628
3.535
3.578
3.367
3.468
2.493
2.658
1.250
1.205
1.262
0.527
0.588
0.427
0.508
0.181
0.202
0.193
0.149
0.086
0.096
0.024
0.091
0.053
0.051
0.042
0.025
0.009
0.037
0.004

sample size of 10 WIM sites into smaller groups for analysis may
not significantly improve the design accuracy. In addition, it could
complicate any future MEPDG design process, as the process is
already much more complex than the empirical design procedures
in the 1993 Guide. For this effort, it was most desirable to develop
only one set of statewide axle-load spectra for the initial implementation of the MEPDG. Thus, it was decided to group the tandem-axle
load spectra from the 10 WIM stations for development of statewide
tandem-axle load spectra. The cluster analysis can be performed
later when more data are available. Table 2 presents the statewide
axle load spectra for Arkansas. Some vehicle classes, including VCs

5, 7, and 11, did not have tandem axles, so axle load spectra for these
vehicle classes were not available (and are therefore shown as 0.000
in Table 2).

Tridem- and Quad-Axle Load Spectra

Statewide tridem-axle load spectra were developed in the same manner as for the statewide tandem axle. Since few quad axles were found
in the WIM data in Arkansas, statewide quad-axle load spectra were
not developed. Default quad-axle load spectra can be used, if required.

Transportation Research Record 2037

Normalized Distribution (%)

110

14.0
12.0
10.0
8.0
6.0
4.0
2.0
0.0
0

5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90
Axle Load (kips)

Normalized Distribution (%)

FIGURE 2

580236

71813

720034

740035

480037

170064

460006

230001

600870

430038

Tandem-axle load spectra for VC 9.

12.0
10.0
8.0
6.0
4.0
2.0
0.0
0

5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90
Axle Load (kips)
170064

Normalized Distribution (%)

FIGURE 3

460006

230001

600870

VC 9 tandem-axle load spectra for TTC 7.

12.0
10.0
8.0
6.0
4.0
2.0
0.0
0

5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90
Axle Load (kips)
720034

FIGURE 4

740035

230001

600870

430038

VC 9 tandem-axle load spectra for Load Spectrum Group 2.

Tran and Hall

111

TABLE 2

Statewide Tandem-Axle Load Spectra

Vehicle Class

Axle Load (lb)

10

11

12

13

6,000

1.048

0.000

2.788

0.000

7.168

1.186

1.267

0.000

0.416

4.488

8,000

2.336

0.000

9.477

0.000

13.251

3.940

2.549

0.000

1.800

8.243

10,000

1.836

0.000

10.963

0.000

11.831

6.237

2.716

0.000

4.297

6.281

12,000

2.538

0.000

10.010

0.000

11.627

8.563

3.814

0.000

10.014

6.847

14,000

3.965

0.000

10.456

0.000

10.441

8.924

6.530

0.000

10.447

7.326

16,000

3.542

0.000

7.766

0.000

6.841

6.826

7.607

0.000

9.835

6.278

18,000

3.973

0.000

6.678

0.000

5.173

5.962

8.315

0.000

10.847

7.332

20,000

5.528

0.000

5.739

0.000

4.010

5.439

8.499

0.000

13.733

6.117

22,000

7.661

0.000

4.943

0.000

3.314

5.334

7.700

0.000

12.420

5.012

24,000

9.697

0.000

4.266

0.000

3.055

5.625

7.933

0.000

9.828

5.152

26,000

11.085

0.000

4.004

0.000

3.004

6.098

7.479

0.000

6.669

4.730

28,000

11.073

0.000

3.804

0.000

2.908

6.578

6.661

0.000

3.854

5.120

30,000

9.557

0.000

3.530

0.000

2.798

6.783

5.882

0.000

2.644

4.538

32,000

7.768

0.000

3.042

0.000

2.451

6.253

4.803

0.000

1.366

4.305

34,000

6.433

0.000

2.520

0.000

2.095

5.073

4.166

0.000

0.626

3.216

36,000

4.286

0.000

2.096

0.000

1.786

3.684

3.360

0.000

0.375

2.648

38,000

2.434

0.000

1.756

0.000

1.553

2.489

2.655

0.000

0.228

2.182

40,000

1.456

0.000

1.436

0.000

1.281

1.643

2.031

0.000

0.196

2.090

42,000

1.043

0.000

1.215

0.000

1.158

1.174

1.684

0.000

0.119

1.819

44,000

0.756

0.000

0.815

0.000

0.860

0.680

1.217

0.000

0.073

1.511

46,000

0.499

0.000

0.664

0.000

0.705

0.452

0.747

0.000

0.044

1.511

48,000

0.282

0.000

0.480

0.000

0.540

0.308

0.562

0.000

0.062

0.729

50,000

0.184

0.000

0.365

0.000

0.442

0.213

0.456

0.000

0.045

0.497

52,000

0.240

0.000

0.280

0.000

0.361

0.149

0.287

0.000

0.013

0.395

54,000

0.081

0.000

0.213

0.000

0.300

0.105

0.217

0.000

0.011

0.513

56,000

0.102

0.000

0.150

0.000

0.226

0.078

0.235

0.000

0.008

0.368

58,000

0.156

0.000

0.127

0.000

0.174

0.057

0.159

0.000

0.009

0.135

60,000

0.074

0.000

0.092

0.000

0.132

0.041

0.143

0.000

0.004

0.120

62,000

0.076

0.000

0.073

0.000

0.112

0.030

0.105

0.000

0.002

0.068

64,000

0.090

0.000

0.057

0.000

0.092

0.021

0.045

0.000

0.003

0.066

66,000

0.130

0.000

0.045

0.000

0.055

0.016

0.045

0.000

0.005

0.066

68,000

0.007

0.000

0.033

0.000

0.050

0.012

0.023

0.000

0.005

0.023

70,000

0.028

0.000

0.030

0.000

0.041

0.009

0.028

0.000

0.001

0.039

72,000

0.002

0.000

0.024

0.000

0.047

0.008

0.017

0.000

0.001

0.017

74,000

0.030

0.000

0.023

0.000

0.033

0.005

0.016

0.000

0.000

0.048

76,000

0.000

0.000

0.017

0.000

0.035

0.004

0.022

0.000

0.000

0.027

78,000

0.007

0.000

0.014

0.000

0.024

0.003

0.020

0.000

0.003

0.016

80,000

0.000

0.000

0.008

0.000

0.020

0.002

0.005

0.000

0.000

0.019

82,000

0.001

0.000

0.004

0.000

0.009

0.001

0.001

0.000

0.000

0.111

SIGNIFICANCE OF STATEWIDE TRAFFIC INPUTS

Evaluation Approach
The statewide axle load spectra were developed on the basis of vehicle weight data collected at WIM stations in Arkansas, while the corresponding default values provided in the MEPDG software were
developed on the basis of LTPP data collected throughout the United
States. It is reasonable to expect the statewide inputs to be different

from the default values. Analyses were performed in this study to

quantify the differences between the developed statewide axle load
spectra and the MEPDG default values.
The MEPDG software (Version 0.8) was used for these analyses.
All analyses were conducted with an assumed 20-year design life.
National calibration parameters for pavement performance prediction models, as provided in the design software, were used without
adjustment. A new exible pavement with AADDT of 10,000 was
assumed for the analyses. Daily truck traffic of more than 10,000 is

112

Transportation Research Record 2037

TABLE 3 Inputs for Sensitivity Analyses of Statewide

Traffic Inputs
Description

ND =

x default xstatewide
100
x default

(1)

Input

where

General information
Type of design
Reliability
Traffic volume
Two-way AADTT
Lanes in design direction
Monthly distribution factors
Vehicle class distribution factors
Hourly distribution factors
Axle load distribution factors
Climate
Asphalt layer
Thickness
Mix
Modulus
Granular base
Thickness
Modulus
Subgrade
Classication
Modulus

ND = normalized difference,
xdefault = predicted distress based on default inputs, and
xstatewide = predicted distress based on statewide inputs.

Flexible
50%
10,000
2
Default
Default
Default
Variable
Fayetteville, AR

Significance of Statewide Axle Load Spectra

11 in.
12.5 mm and PG 70-22
Level 3
14 in.
40,000 psi
A-7-6
5,000 psi

often observed on Interstates and several four-lane highways in

Arkansas. Important information related to the analyses is summarized in Table 3. The inputs are typical for materials and climate
conditions in Arkansas. The pavement structure shown in Table 3 was
designed through use of the 1993 Guide.
Two predicted distresses of interest in this study are total rutting
and fatigue cracking. Default design limits for these distresses in the
MEPDG software are 0.75 in. of rutting and 25% of fatigue cracking.
These design limits are used in this study because they are currently
used in most studies relative to the MEPDG. Differences between
the predicted distresses that are based on statewide and default trafc inputs were normalized for analyses through use of Equation 1.
Detailed analyses are presented later.

The developed statewide axle load spectra were compared with the
default values with a graphical method. Figure 5 shows a comparison of the statewide and default single-axle load spectra for VC 9
trucks. The statewide distribution is more concentrated between
8,000 and 12,000 lb than the default one. Figure 6 compares the
statewide tandem-axle load spectra for Class 9 trucks with the default
values. The peaks representing unloaded and loaded trucks on the
statewide and default distribution curves were different. For other
vehicle classes, the statewide axle load spectra were also different
from the default values. Such differences in axle load spectra produced questions about the effect on predicted pavement performance
from the use of statewide versus default values.
Figure 7 shows comparisons of the predicted distresses from the
statewide and default axle load spectra. The predicted rutting after
20 years was higher than the design limit of 0.75 in., and the predicted fatigue cracking was lower than the design limit of 25%. Thus,
predicted pavement lives were determined on the basis of the predicted rutting graphs. The average predicted pavement life was about
13 years. Figure 8 shows the normalized differences in predicted
pavement life and predicted pavement distresses after 13 years.
The difference in predicted pavement life was more than 25%.
Because the effect of the differences between the statewide and
default axle load spectra appears significant, it is recommended
that the statewide axle load spectra be used for future design purposes. However, the MEPDG has not been calibrated specically for
Arkansas. While it is not expected that results of sensitivity analyses
such as those described here will be signicantly affected, additional
studies will be performed subsequent to local calibration efforts.

Normalized Distribution (%)

40.0
35.0
30.0
25.0
20.0
15.0
10.0
5.0
0.0
0

10

15

20
25
30
Axle Load (kips)

Default

FIGURE 5

35

Statewide

Statewide and default single-axle load spectra for VC 9.

40

45

Tran and Hall

113

Normalized Distribution (%)

16.0
14.0
12.0
10.0
8.0
6.0
4.0
2.0
0.0
0

5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90
Axle Load (kips)
Default

FIGURE 6

Statewide

Statewide and default tandem-axle load spectra for VC 9.

1.2

Rutting (in.)

1
0.8
Default
0.6

Statewide
Design Limit

0.4
0.2
0
0

50

100
150
200
Pavement Age (mon.)

250

(a)
5

Fatigue Cracking (%)

3
Default
Statewide
2

0
0

50

100
150
Pavement Age (mon.)

200

250

(b)
FIGURE 7 Predicted distresses based on statewide and default axle load
spectra: (a) predicted rutting and (b) predicted fatigue cracking.

114

Transportation Research Record 2037

Normalized Difference (%)

30
25
20
15
10
5
0
Pavement Life

FIGURE 8

Rutting
Predicted Parameters

Cracking

Normalized differences in predicted parameters.

CONCLUSIONS
The following observations and conclusions are based on the development and analysis of statewide axle load spectra for Arkansas:
On the basis of quality control checks on weight data collected
at 25 WIM sites in Arkansas, only 10 stations provided suitable data
for developing statewide axle load spectra. The other sites showed
evidence indicating that scales had failed or were out of calibration.
The Traoad software was unable to import the W-records provided by AHTD. The format of the records was checked carefully,
and the error was unknown. Instead of using the Traoad program,
the researchers developed a computer program for this study.
Statewide single-, tandem-, and tridem-axle load spectra were
developed in this study, but statewide quad-axle load spectra were
not available because few quad axles were found in the data.
Statewide axle load spectra were different from the default
values in the MEPDG. The inuence of the differences between the
statewide and default axle load spectra on pavement performance
predictions provided by the MEPDG software was signicant. The
difference in predicted pavement life can be more than 25%.
RECOMMENDATIONS
The following recommendations are based on the findings in this
study:
Weight data collected at WIM stations should be checked carefully before it is used for design purposes.
The WEIGHT.xls program developed in this study can be used
to develop axle load spectra for a specic site in the future.
Developed statewide axle load spectra should be used instead
of default values for future design applications.
Statewide axle load spectra should be updated periodically unless
no signicant changes are observed in the future.
ACKNOWLEDGMENTS
This paper was prepared under the research project Projected Traffic
Loading for MechanisticEmpirical Pavement Design Guide, sponsored by the Arkansas State Highway and Transportation Depart-

ment and the U.S. Department of Transportation, Federal Highway

Administration.
REFERENCES
1. Guide for Design of Pavement Structures. AASHTO, Washington,
D.C., 1993.
2. ARA, Inc., ERES Consultants Division. Guide for MechanisticEmpirical
Design of New and Rehabilitated Pavement Structures. NCHRP Project 1-37A. Final report. Transportation Research Board of the National
Academies, Washington, D.C., 2004.
3. Traffic Monitoring Guide. FHWA, U.S. Department of Transportation,
Washington, D.C., 2001.
4. Hallenbeck, M., and H. Weinblatt. NCHRP Report 509: Equipment for
Collecting Traffic Load Data. Transportation Research Board of the
National Academies, Washington, D.C., 2004.
5. Tam, W. O., and H. Von Quintus. Use of Long-Term Pavement Performance Data to Develop Traffic Defaults in Support of Mechanistic
Empirical Pavement Design Procedures. In Transportation Research
Record: Journal of the Transportation Research Board, No. 1855, Transportation Research Board of the National Academies, Washington, D.C.,
2003, pp. 176182.
6. Al-Yagout, M. A., J. Mahoney, L Pierce, and M. Hallenbeck. Improving Traffic Characterization to Enhance Pavement Design and Performance: Load Spectra Development. WA-RD 600.1, Washington State
Transportation Center, University of Washington, Seattle, 2005.
7. Guide to LTPP Traffic Data Collection and Processing. FHWA, U.S.
Department of Transportation, Washington, D.C., 2001.
8. Tran, N. H., and K. D. Hall. Projected Traffic Loading for Mechanistic
Empirical Pavement Design Guide. Draft final report, University of
Arkansas, Fayetteville, 2006.
9. Cambridge Systematics, Inc., Washington State Transportation Center,
and Chaparral Systems, Inc. NCHRP Report 538: Traffic Data Collection, Analysis, and Forecasting for Mechanistic Pavement Design.
Transportation Research Board of the National Academies, Washington, D.C., 2005.
10. Lu, Q., and J. T. Harvey. Characterization of Truck Traffic in California
for MechanisticEmpirical Design. In Transportation Research Record:
Journal of the Transportation Research Board, No. 1945, Transportation Research Board of the National Academies, Washington, D.C.,
2006, pp. 6172.
The views expressed in this paper and the accuracy of the data and facts are the
responsibility of the authors and do not necessarily represent the official views of
the sponsoring agencies. This paper does not constitute a standard, specification,
or regulation.
The Flexible Pavement Design Committee sponsored publication of this paper.