Open Journal of Civil Engineering, 2011, 1, 7-12

doi:10.4236/ojce.2011.12002 Published Online December 2011 (http://www.SciRP.org/journal/ojce)

Flexible Pavement Analysis Considering Temperature

Profile and Anisotropy Behavior in Hot Mix Ashalt Layer
Joonho Choi1, Youngguk Seo2, Sung-Hee Kim3*, Samuel Beadles4
1
Postdoctoral Research Fellow, Georgia Institute of Technology, Atlanta, USA
2
Pavement Research Group, Korea Expressway Corporation, Seongnam-si, South Korea
3
Assistant Professor, Civil and Construction Engineering, Southern Polytechnic State University, Marietta, USA
4
Civil and Construction Engineering, Southern Polytechnic State University, Marietta, USA
E-mail: Joonho@gatech.edu, seoyg89@hotmail.com, {*skim4, sbeadles}@spsu.edu
Received September 20, 2011; revised October 31, 2011; accepted November 14, 2011

Abstract

A three Dimensional finite element model (FEM) incorporating the anisotropic properties and temperature
profile of hot mix asphalt (HMA) pavement was developed to predict the structural responses of HMA
pavement subject to heavy loads typically encountered in the field. In this study, ABAQUS was adopted to
model the stress and strain relationships within the pavement structure. The results of the model were veri-
fied using data collected from the Korean Highway Corporation Test Road (KHCTR). The results demon-
strated that both the base course and surface course layers follow the anisotropic behavior and the incorpora-
tion of the temperature profile throughout the pavement has a substantial effect on the pavement response
predictions that impact pavement design. The results also showed that the anisotropy level of HMA and base
material can be reduced to as low as 80% and 15% as a result of repeated loading, respectively.

Keywords: Anisotropic Behavior, Finite Element Method, Aggregate Base, HMA

1. Introduction and base layers with temperature profile variations through

the HMA layer. The approach was also validated by
Existing FEMs that model the behavior of layers in comparing measured pavement responses with the pre-
HMA pavement rely on simplifying assumptions about dictions of the three dimensional finite element analysis
the variations of the properties within the pavement. results.
Specifically, existing models assume linear and isotropic
variations within the layers of the pavement and use the 2. Anisotropic Behavior of Granular
annual maximum or minimum pavement temperature to Materials
recommend a suitable asphalt binder performance grade.
These assumptions are made in order to reduce the com- Inherent anisotropy in granular materials exists even be-
putational complexity involved in modeling the behavior fore the pavement is subjected to traffic due to the effects
of HMA pavement. A number of researchers, however, of compaction and gravity [6]. Stresses due to construc-
have shown that the pavement materials exhibit nonlin- tion operations and traffic are anisotropic and new parti-
ear, anisotropic behavior [1-4]. Researchers also men- cle contacts are formed due to breakage and slippage of
tioned that the structural or load-carrying capacity of the particles, which induces further anisotropy [6]. To more
pavement varies with temperature for HMA layer and it accurately investigate the effect of stress-dependency of
is necessary to predict the temperature distribution within granular materials on pavement response and surface
the HMA layers to accurately determine in situ strength deflection predictions, researchers developed a method to
characteristics of flexible pavement [5]. fully characterize the stress-sensitive and cross-anisot-
This paper deals with the predictions of the pavement ropic properties of unbound aggregate bases, which are
critical responses using three dimensional FEM analysis resilient moduli in the vertical and radial directions, Ey
technique. The FEM developed using ABAQUS in this and Ex; shear modulus in vertical direction, Gxy; Pois-
study incorporates the anisotropic properties of the HMA son’s ratio for strain in the vertical direction due to a

Copyright © 2011 SciRes. OJCE

8 J. CHOI ET AL.

horizontal direct stress, υxy; and Poisson’s ratio for 45. This two-lane, 7.7 km-long highway was composed
strain in any horizontal direction due to a horizontal di- of 25 concrete and 33 asphalt sections as illustrated in
rect stress, υxx [2,4]. Figure 1.
As expressed in Equations (1) through (3), the modulus With projected AADT (Average Annual Daily Traffic)
of the unbound aggregate base was modeled using an of 57,520 in year 2011 and a load distribution factor of
Uzan type stress-dependent model with a cross- anisot- 0.8, the traffic volume for a 10-year design life was esti-
ropic approach based on the recent studies that empha- mated to be 44.7 million 80-kN ESALs (Equivalent Single
sized the importance of accounting not only for stress Axle Loads) for all pavement sections. The AASHTO
dependency but also the anisotropy in order to properly interim design guide was adopted for the structural de-
model the unbound aggregate modulus properties and the sign of both pavement types. Detailed information on
stress state distributions in the layer [2,4]. design and construction of the KHCTR can be found
k2 k3 elsewhere [9].
 I  τ  Approximately 1900 sensors were installed at KHCTR
E y  k1    oct  (1)
 Pa   Pa  to obtain stress and strain responses and to monitor
moisture and temperature variations at different locations
Ex of sections during construction. Most of asphalt sections
n (2)
Ey were instrumented with 636 sensors that include strain
gauges, soil pressure cells and thermal couples. Figure 2
G xy
m (3) illustrates a sensor layout for one of asphalt sections, A5,
Ey where strain gauges were placed in longitudinal and
transverse directions to quantify the anisotropic levels in
In Equatipns (1), (2), and (3), Ey is the vertical
HMA layers. In Figure 2, ASTM 19 is the 19-mm dense
modulus, Ex is the horizontal modulus, Gxy is the shear
graded HMA used for surface layer, BB5 is the 25-mm
modulus, I is the first stress invariant (bulk stress), τoct is
dense graded HMA used for intermediate layer, and BB3
the octahedral shear stress, Pa is the atmospheric pres- is the 25-mm dense graded HMA used for base layer.
sure, and ki are material model parameters obtained from Thermocouples were also embedded from the surface to
regression analyses of the laboratory modulus test data. the bottom of asphalt layer and those measurements were
To consider the anisotropic behavior of HMA layer, considered as inputs for the FEM described in subse-
the horizontal and shear moduli ratio was approximated quent sections.
and used as an input for the FEM analysis. A series of moving load tests was performed at A5
with a dump truck. A three-axle dump truck (single tire
3. Field Response Test front and dual tire rear tandem) with a 12R22.5 type ra-
dial tire was utilized as a load source all along. This test
The KHCTR has been regarded as the most realistic re- vehicle was operated by the same highly trained driver to
search tool to evaluate the performance of pavements minimize natural sway in a moving vehicle. The planar
influenced by many complex variables such as construc- configuration of tire and axle is seen in Figure 3. During
tion, traffic, climatic, materials, etc. Test road in Korea testing, air pressures for tires of each axle were main-
was first constructed in December 2002 and opened to tained at levels: 1076 kPa (1st axle), 828 kpa (2nd axle),
traffic in March 2004. Road research institute at the Ko- and 1076 kpa (3rd axle). These pressure levels have been
rea Highway Corporation (KHC) played a leading role in determined by averaging field survey data collected at
the construction and operation of this test road, and con- countrywide weighing stations in Korea.
ducted a wide range of field tests to better characterize The test results conducted was selected for the imple-
the response and performance of highway pavements mentation of FEM developed in this study. The full-scale
[7,8]. The KHCTR is located between the Yeo-ju junc- pavement test results were shown in Table 1. First and
tion on Interstate 50 and the Gam-gok interchange on second values in the table were the minimum and maxi-
Interstate 45, and it goes almost parallel to the Interstate mum from the experiments, respectively.

Interstate 45 Gam-gok IC

Yeo-ju JC
KHC Test Road

25 PCC Sections 33 AC Sections

Figure 1. Plan view of KHCTR.

Copyright © 2011 SciRes. OJCE

 J. CHOI ET AL. 9

Soil pressure cell

Longitudinal strain gauge
Transverse strain gauge
Thermocouple

A5
ASTM19
.5
0.5
BB5
12

20 BB3

30
Depth, cm

40 Subbase

50

60

70
Anti-frost Figure 4. Finite Element (FE) model with mesh and loading
80 area.
90
27.2 kN, 21.5 kN, and 21.9 kN were applied for front,
Figure 2. Cross-section and sensor layout of A5 at KHCTR.
middle, and back tire areas respectively. A fixed bound-
ary condition was also assumed at the bottom of the sec-
tion. A regression equation that fits the temperature pro-
file data was developed as shown in Equation (4) and
implemented into the FEM model.
Temperature  20.411(depth) 0.175 (4)
In order to develop values for vertical and horizontal
resilient modulus ratios, twelve sensitive analyses were
performed with different level of anisotropy. All materi-
als were assumed as anisotropic material except anti-
frost layer and Table 2 shows the material properties
Figure 3. Plan view of the passing lane at A5 asphalt section. used in the FEM models.
Table 3 shows the comparisons between the measured
Table 1. Performance response summary of pavement test data collected from the KHCTR and FEM results for the
sections. simulations. Based on comparisons, the higher value of
Top Anti-frost Top Subbase Bottom AC the vertical strain at the top of anti-frost layer was ob-
tained when the unbound aggregate base and HMA layer
Vertical Vertical Vertical Vertical Horizontal
Stress (kPa) Strain (10-6) Stress (kPa) Strain (10-6) Strain (10-6) were modeled as 15% level of anisotropy and 70% level
of anisotropy. Similarly, the tensile strains at the bottom
19.3 - 24.8 N/A 36.4 - 81.9 N/A 30.6 - 56.5 transitioned from the linear isotropic case without con-
sidering the temperature profile variation to the anisot-
4. Analysis of the Results ropic analyses with the consideration of temperature pro-
file variation throughout the HMA layer. These are the
A pavement comprised of a 120-mm HMA layer, a critical pavement responses that are directly related to rut-
180-mm unstabilized base course, a 300-mm thick stress ting and asphalt fatigue cracking that significantly con-
softening subbase layer, and a 300-mm thick anti-frost tribute to the overall pavement performance and overlay
layer was used for the KHCTR section model. As shown design.
in Figures 1 and 2, a three-axle truck was passing Table 3 also compares the measured pavement responses
through the lane and the half of the area was adapted to with the three dimensional FEM predictions from the
the FE simulation. The red area shows the FEM model different analyses. In general, the predicted pavement
part and FEM model was generated. Figure 4 showed responses from FEM analysis with the anisotropic model
the three dimensional finite element mesh model repre- for HMA and aggregate base are in reasonably good
senting loading area and the red area on the top repre- agreement with the measured pavement responses from
sented the loading location. The perimeter boundary KHCRT when the HMA layer and Base layer is consid-
conditions were assumed as simply supported. Loads of ered as 80% and 15% level of anisotropy, separately,

Copyright © 2011 SciRes. OJCE

10 J. CHOI ET AL.

Table 2. Anisotropic material properties.

Simulation Layer Thickness Vertical Resilient Vertical Horizontal Resilient Horizontal Shear Stress
Number Type (m) Modulus (MPa) Poisson’s Ratio Modulus (MPa) Poisson's Ratio (MPa)

HMA 0.12 2760 0.35 1930 0.245 1930

HMA: 70% Base 0.18 517 0.35 78 0.05 78

Base: 15% Subbase 0.30 241 0.35 36 0.05 36
Anti-frost 0.30 69 0.40 - - -
HMA 0.12 2760 0.35 1930 0.245 1930

HMA: 70% Base 0.18 517 0.35 155 0.11 155

Base: 30% Subbase 0.30 241 0.35 72 0.11 72
Anti-frost 0.30 69 0.40 - - -
HMA 0.12 2760 0.35 1930 0.245 1930

HMA: 70% Base 0.18 517 0.35 259 0.18 259

Base: 50% Subbase 0.30 241 0.35 121 0.18 121
Anti-frost 0.30 69 0.40 - - -
HMA 0.12 2760 0.35 2210 0.28 2210

HMA: 80% Base 0.18 517 0.35 78 0.05 78

Base: 15% Subbase 0.30 241 0.35 36 0.05 36
Anti-frost 0.30 69 0.40 - - -
HMA 0.12 2760 0.35 2210 0.28 2210

HMA: 80% Base 0.18 517 0.35 155 0.11 155

Base: 30% Subbase 0.30 241 0.35 72 0.11 72
Anti-frost 0.30 69 0.40 - - -
HMA 0.12 2760 0.35 2210 0.28 2210

HMA: 80% Base 0.18 517 0.35 259 0.18 259

Base: 50% Subbase 0.30 241 0.35 121 0.18 121
Anti-frost 0.30 69 0.40 - - -
HMA 0.12 2760 0.35 2480 0.32 2480

HMA: 90% Base 0.18 517 0.35 78 0.05 78

Base: 15% Subbase 0.30 241 0.35 36 0.05 36
Anti-frost 0.30 69 0.40 - - -
HMA 0.12 2760 0.35 2480 0.32 2480

HMA: 90% Base 0.18 517 0.35 155 0.11 155

Base: 30% Subbase 0.30 241 0.35 72 0.11 72
Anti-frost 0.30 69 0.40 - - -
HMA 0.12 2760 0.35 2480 0.32 2480
Base 0.18 517 0.35 259 0.18 259
HMA: 80%
Subbase 0.30 241 0.35 121 0.18 121
Base: 50%
Anti-frost 0.30 69 0.40 - - -
Anti-frost 0.30 69 0.40 - - -

Copyright © 2011 SciRes. OJCE

 J. CHOI ET AL. 11

Table 3. FEM results with different anisotropic level.

Top Anti-frost Bottom Subbase Top Subbase Bottom Base Top Base Bottom AC
Anisotropy Vertical Vertical Vertical Vertical Vertical Vertical Vertical Vertical Vertical Vertical Horizontal
Level Stress Strain Stress Strain Stress Strain Stress Strain Stress Strain Strain
(kPa) (10–6) (kPa) (10–6) (kPa) (10–6) (kPa) (10–6) (kPa) (10–6) (10–6)

HMA: 70%
23.5 300 26.5 120 40.2 180 46.8 100 68.5 140 59
Base: 15%
HMA: 70%
17.8 240 20.4 100 36.9 166 45.9 109 76.6 155 36
Base: 30%
HMA: 70%
14.2 190 16.1 95 33.4 154 44.0 110 81.5 159 19
Base: 50%
HMA: 80%
22.8 285 25.6 116 38.3 168 44.4 98 64.3 132 54
Base: 15%
HMA: 80%
17.4 230 19.8 102 35.4 159 43.8 104 72.3 147 34
Base: 30%
HMA: 80%
13.9 185 15.8 92 32.2 148 42.1 106 77.3 152 18
Base: 50%
HMA: 90%
22.1 276 24.8 112 36.7 160 42.3 93 60.7 125 50
Base: 15%
HMA: 90%
17.1 225 19.4 99 34.1 153 42.0 100 68.6 140 32
Base: 30%
HMA: 90%
13.7 181 15.5 90 31.1 144 40.5 102 73.7 146 18
Base: 50%
Isotropic,
(Temperature
21.8 272 24.4 111 35.9 157 41.4 92 59.3 122 90
Variation Not
Considered)

19.3 36.4 30.6

Exp
–24.8 –81.9 –56.5

with the temperature profile consideration throughout the performance, and the level of anisotropy of HMA and
pavement. This indirectly shows that anisotropic model- aggregate base impacts the stress and strain distributions.
ing and temperature variation consideration of HMA and With better and more accurate predictions of these re-
base layers provide a much more realistic approximation sponses, a structurally more adequate pavement can be
of the measured responses. With better and more accu- designed.
rate predictions of these responses, a more structurally
adequate pavement can be designed. 6. References

5. Conclusions [1] S. H. Kim, D. N. Little and E. Tutumluer, “Validated

Model for Predicting Field Performance of Aggregate
The synergistic effect of anisotropic behavior of HMA Base Courses,” Transportation Research Record 1873,
2003.
and base layers considering the temperature variation
throughout the HMA layer was investigated by perform- [2] S. H. Kim, D. N. Little and E. Masad, “Simple Methods
to Estimate Inherent and Stress-Induced Anisotropic
ing a three dimensional finite element analyses. The re-
Level of Aggregate Base,” Transportation Research Re-
sults from the ABAQUS finite element models clearly cord 1913, TRB, National Research Council, Washington
showed that anisotropic model of both the asphalt and DC, 2005, pp. 24-31.
aggregate base layers gave the most realistic predictions [3] S. H. Kim, D. N. Little and E. Tutumluer, “Effect of
when compared to measured values for pavement re- Gradation on Nonlinear Stress-Dependent Behavior of a
sponses. Substantially higher critical pavement responses Sandy Flexible Pavement Subgrade,” Journal of Trans-
were predicted with the increased anisotropic level of portation Engineering, Vol. 133, No. 10, 2007, pp. 582-
HMA and base layer. Horizontal strain is a critical 589. doi:10.1061/(ASCE)0733-947X(2007)133:10(582)
pavement response, which is directly related to fatigue [4] E. Tutumluer, D. N. Little and S. H. Kim, “Validated

Copyright © 2011 SciRes. OJCE

12 J. CHOI ET AL.

Model for Predicting Field Performance of Aggregate doi:10.1002/(SICI)1096-9853(199909)23:11<1075::AID-

Base Courses,” Transportation Research Record 1837, NAG24>3.0.CO;2-S
TRB, National Research Council, Washington DC, 2003, [7] Y. Seo, “Distress Evolution in Highway Flexible Pave-
pp. 41-49. ments: A 5-Year Study at the KHC Test Road,” ASTM
[5] B. K. Diefenderfer, I. L. Al-Qadi and S. D. Diefenderfer, Journal of Testing and Evaluation, Vol. 38, No. 1, 2010,
“Model to Predict Pavement Temperature Profile: De- pp. 32-41.
velopment and Validation,” Journal of Transportation [8] Y. Seo, “A Full Scale in Situ Evaluation of Strain Char-
Engineering, Vol. 132, No. 2, 2006, pp. 162-167. acteristics at Highway Flexible Pavement Sections,”
doi:10.1061/(ASCE)0733-947X(2006)132:2(162) ASTM Journal of Testing and Evaluation, Vol. 38, No. 4,
[6] J. J. Dong and Y. W. Pan, “Micromechanics Model for 2010, pp. 390-399.
Elastic Stiffness of Non-Spherical Granular Assembly,” [9] Korea Highway Corporation, “A Study on the Construc-
International Journal for Numerical and Analytical tion and Management of Korea Highway Test Road,”
Methods in Geomechanics, Vol. 23, 1999, pp. 1075-1100. Final Report, 2002.

Copyright © 2011 SciRes. OJCE