MODELING RESILIENT MODULUS AND TEMPERATURE CORRECTION

FOR SAUDI ROADS

By Hamad I. Al-Abdul Wahhab,1 Ibrahim M. Asi,2 and Rezqallah H. Ramadhan3

ABSTRACT: Temperature is one of the most important factors affecting the design and performance of pave-
ment. Temperature variations within the pavement structure contribute in many different ways to distress and
possible failure of that structure. The structural performance of pavements is highly dependent on temperatures
to which these pavements are exposed. Under loading conditions, the pavement temperature is a major factor
affecting the deformation response of bituminous structures. The load-spreading characteristics depend on the
Downloaded from ascelibrary.org by Universidad Politecnica De Valencia on 05/14/15. Copyright ASCE. For personal use only; all rights reserved.

moduli of temperature-sensitive layers whose strength is significantly reduced by increased temperature. Varia-
tions of temperature across pavement depth must be considered in the back-calculation of flexible pavement
layer moduli. This paper presents the results of a comprehensive study that was carried out to explore trends of
temperature variation in an arid environment and their implications on the moduli of flexible pavements. Tem-
perature correction factors and resilient modulus estimation equations from basic material physical properties
were developed using statistical procedures to a high degree of reliability.

INTRODUCTION lating pavement layer moduli in nondestructive testing.

Objective pavement design and evaluation require the de-
termination of the resilient modulus of pavement layers. The
functional and structural performance of flexible pavements is
highly dependent on the temperature regime to which these
pavements are exposed. The strength and deformation char-
acteristics of asphalt concrete mixes have been shown to
change significantly with temperature, which affects the resil-
ient modulus of the asphalt layers. The resilient modulus is a
basic input for most of the mechanistic designs and is an ob-
jective for the mechanistic evaluation procedures. The Amer-
ican Association of State Highway and Transportation Officials
1986 structural pavement design takes the resilient modulus
of the different pavement layers into account when designing
new pavement structures. The determination of the resilient
modulus requires an expensive setup of testing equipment that
might not be readily available for most of the field engineers.
The development of a regression formula that can be used to
correlate the resilient properties of local asphalt mixes to basic
material properties, with acceptable accuracy, is of great ben-
efit, especially since only two types of mixes are used all over
the kingdom of Saudi Arabia. One is for the wearing course
and the other is for the base course.
Temperature variation affects both the functional perfor-
mance and the structural performance of asphalt concrete
pavements. Cold temperatures accelerate cracking of the as-
phalt-bound layers due to shrinkage, or cause fracture of these
layers due to frost heaving of the underlying soils. High tem-
peratures, on the other hand, can cause distortion of the as-
phalt-bound layers (such as rutting) or may produce slippery
surfaces due to bleeding of the asphalt.
All mechanistic methods of pavement structural design at-
tempt to take into account the seasonal variation in the strength
and deformation characteristics of the pavement layers. Pave-
ment temperature is of primary importance when back-calcu-
1
Prof., Dept. of Civ. Engrg., King Fahd Univ. of Petr. and Minerals,
Dhahran 31261, Saudi Arabia.
2
Asst. Prof., Dept. of Civ. Engrg., King Fahd Univ. of Petr. and Min-
erals, Dhahran 31261, Saudi Arabia.
3
Res. Engr. III, Res. Inst., King Fahd Univ. of Petr. and Minerals, Box
501, Dhahran 31261, Saudi Arabia.
Note. Associate Editor: Dallas Little. Discussion open until January 1,
2002. To extend the closing date one month, a written request must be
filed with the ASCE Manager of Journals. The manuscript for this paper
was submitted for review and possible publication on November 3, 1998;
revised September 22, 2000. This paper is part of the Journal of Mate-
rials in Civil Engineering, Vol. 13, No. 4, July/August, 2001. 䉷ASCE,
ISSN 0899-1561/01/0004-0298–0305/$8.00 ⫹ $.50 per page. Paper No.
19587.
298 / JOURNAL OF MATERIALS IN CIVIL ENGINEERING / JULY/AUGUST 2001
Knowledge of seven-day maximum and minimum pavement
temperatures is a prerequisite for the application of the new
performance-based Strategic Highway Research Program
binder specification. It is therefore vital to know the range of
temperatures to which the pavement will be subjected. Such
knowledge will undoubtedly assist in improving the mix and
structural designs of asphalt pavements.
Extensive research on temperature distribution in asphalt
pavements has been carried out in many different climatic ar-
eas of the world, such as the United States, Australia, Kuwait,
and South Africa (Kallas 1966; Bissada 1972; Williamson
1977). Several researchers have developed mathematical mod-
els to simulate pavement temperature at and beneath the pave-
ment surface to a reasonable degree of accuracy (Venkatara-
man and Venkatasubramanian 1977; Williamson 1972;
Dickinson 1978; Thompson et al. 1988). In Saudi Arabia, no
single study has attempted to develop mathematical models to
simulate the temperature to which local pavements are sub-
jected.
Extensive laboratory testing was carried out as part of the
national study of pavement rutting in Saudi Arabia (Fatani et
al. 1990). Wearing course as well as base course cores and
slab samples were extracted from 19 major highways across
the country. The resilient modulus of the core samples was
evaluated for a range of in-service temperatures, while phys-
ical properties of asphalt and aggregates were determined from
materials recovered from slab samples. The collected infor-
mation was used to come up with a regression equation to
correlate the resilient modulus of field mixes to the basic phys-
ical properties of asphalt mix ingredients. The developed equa-
tion may be used to estimate the resilient modulus of asphalt
mixes for design purposes or to estimate a seed value or
boundary limit for the back-calculated resilient modulus in
common pavement evaluation techniques.
Pavement temperature variation with the pavement depth
has been monitored across the country for a period of two
consecutive years (Al-Abdul Wahhab and Balghunaim 1994)
and a comprehensive database has been developed. Statistical
analysis of the database yielded a statistically reliable corre-
lation between air temperature and pavement temperature at
any depth in the local environment. The developed relations
are also presented in the form of nomographs that can be used
to calculate the pavement temperature at any depth if the sur-
face or air temperature is known. These relations are essential
for the back-calculation of pavement layer moduli.
The back-calculation process is an inverse problem for es-
tablishing pavement materials properties given the in situ re-
sponse data (Stolle and Hein 1989). From pavement deflection

J. Mater. Civ. Eng. 2001.13:298-305.

 data, and utilizing multilayer linear elastic modeling, the load-
carrying capacity and the overlay design of the pavement
structure can be established. One important step in the back-
calculation of pavement layer moduli is to correct the modulus
value for the temperature effect. This can be achieved by con-
verting the back-calculated moduli to the same base temper-
ature. The correction factor was developed utilizing the resil-
ient modulus testing results on core samples in the national
rutting study. A mathematical equation and nomograph are de-
veloped for this purpose. The basic objective of this paper is
to outline procedures followed to develop empirical relations
for the estimation of the resilient modulus, temperature cor-
Downloaded from ascelibrary.org by Universidad Politecnica De Valencia on 05/14/15. Copyright ASCE. For personal use only; all rights reserved.

rection factor, and pavement temperature at any depth. These

correlations are essential for pavement design and evaluation.

STUDY SECTIONS
Selection of Test Sections
Data required for this study were collected as part of the
national research project for the evaluation of the permanent
deformation of asphalt concrete pavement in Saudi Arabia (Fa-
tani et al. 1990). The asphalt concrete characteristics for sev-
eral in-service highways were evaluated for asphalt concrete
cores and slabs extracted from test sections. The test sections
were selected to include traffic volume greater than 1,000 FIG. 1. Sketch of Sampling Scheme
equivalent single axle loads (ESAL)/day, old and new con-
struction, and different performance. Table 1 lists the selected lanes as well as the slow lanes of the test sections in divided
sections in the different regions of the kingdom. The test sec- highways. For these sections, a total of 12 cores were col-
tions were selected to include six nonrutted sections on five lected, six from the slow lane (outer lane, cores 1–6) and six
old highways that used to carry traffic prior to the new high- from the fast lane (inner lane, cores 7–12). In addition, four
way network, in addition to 13 rutted and nonrutted sections, slabs (100 cm ⫻ 100 cm) were obtained from each section.
which were also selected on six new expressways. On each For two-lane undivided highways, the sampling scheme is sim-
road, two adjacent sections (rutted and nonrutted) were se- ilar to that of the fast lane of divided highways. The detailed
lected so that they would be carrying similar traffic and would sampling scheme is shown in Fig. 1.
be subjected to similar external factors; they were built by the
same contractor and had the same construction materials. Laboratory Evaluation
Sampling Scheme Collected pavement samples were divided into two groups.
Each group was subjected to different laboratory testing, as
The collection of field samples was intended to build up a
follows:
database about the material characteristics of asphaltic con-
crete mixes. This required the testing of pavement samples
including cores and slabs. Samples were obtained from the fast • Group I. Cores (numbers 1, 4, 5, 8, 9, and 12) were
collected from fast lanes and slow lanes and
TABLE 1. Selected Test Sections were subjected to resilient modulus testing at
25⬚C, followed by 40⬚C, then 50⬚C, in addition
Construc- Rut to specific gravity determination.
tion Study depth Trafficb
• Group II. Cores (numbers 2, 3, 6, 7, 10, and 11) and slabs
Road name date section (mm) (ESAL/day)
(S1, S2, S3, and S4) were collected from fast
Abuhadriah-Dammam section A 1982 E1N a 12 6,300 lanes and slow lanes and were subjected to an
Abuhadriah-Dammam section A 1982 E2R 44 6,300 asphalt mix property analysis such as void anal-
Abuhadriah-Dammam section B 1982 E3N 8 6,300
Riyadh-Dammam section 7 1983 E4R 24 8,170
ysis, bitumen content and viscosity testing, and
Riyadh-Dammam section 7 1983 E5N 10 8,170 aggregate gradation and properties determina-
Abqaiq-Dhahran old road 1962 E6N 5 1,580 tion, as shown in Fig. 2.
Riyadh-Dammam section 4 1985 C1R 33 7,680
Riyadh-Dammam section 4 1985 C2N 9 7,680
Riyadh-Dammam old road 1958 C3N 7 1,700 The detailed laboratory testing procedures were as follows.
Riyadh-Kharj 1979 C4R 30 5,400
Riyadh-Kharj 1979 C5N 8 5,400 Resilient Modulus
Riyadh-Kharj old road 1958 C6N 5 2,180
Makkah-Jeddah old road 1960 W1N 6 3,900 When cores of the first group were received at the labora-
Makkah-Jeddah old road 1960 W2R 25 3,900
Makkah-Taif Expressway 1984 W3N 7 5,700
tory, the thickness of the cores was measured. The wearing
Makkah-Taif Expressway 1984 W4R 50 5,700 course layer was separated from the base course to make 63.5
Makkah-Madinah section 8A 1984 W5N 8.5 4,500 mm specimens (or less, if layers were not thick enough). Spec-
Makkah-Madinah section 8A 1984 W6R 60 4,500 imens were heated to 40⬚C, then desiccated under a vacuum
Jeddah-Madinah old road 1956 W7N 6.7 3,600 for the following tests to be made:
a
E = eastern region, C = central region, W = western region, R = rutted,
and N = nonrutted.
b
For the slow lane.
• Bulk specific gravity (ASTM D 2726)
• Modulus of resilience (ASTM D 4123)
JOURNAL OF MATERIALS IN CIVIL ENGINEERING / JULY/AUGUST 2001 / 299

J. Mater. Civ. Eng. 2001.13:298-305.

Downloaded from ascelibrary.org by Universidad Politecnica De Valencia on 05/14/15. Copyright ASCE. For personal use only; all rights reserved.
FIG. 2. Flowchart of Laboratory Testing of Group II Samples
The resilient modulus property was determined by means
of a diametral modulus device. This test is basically a repeti-
tive load test using the stress distribution principle of the in-
direct tensile test. Loads were applied and horizontal elastic
tensile deformations were recorded to determine the modulus,
using the following equation:
MR = [P(␮ ⫹ 0.2734)]/[t(⌬h)] (1)
where MR = resilient modulus; P = dynamic load; t = thickness
of specimens; ⌬h = elastic horizontal deformation; and ␮ =
Poisson’s ratio (assumed 0.35).
The resilient modulus was determined for all test sections
at three testing temperatures, 25⬚C, followed by 40⬚C, then
50⬚C. Higher temperatures were not used because at higher
temperatures, the test would be a destructive test.
Asphalt Mix Properties
When group II samples were received at the laboratory,
cores were subjected to the following process (Fig. 2):
1. The total core thickness and the thickness of each layer
were measured, and a concrete saw was used to cut the
core into the different composition layers.
2. The bulk specific gravity was then determined for all
samples according to ASTM D 2726.
3. Each sample was then heated in the oven at 110⬚C for
about 30 min in order to loosen the mix. Quantitative
extraction (ASTM D 2171, procedure B), using the re-
flux extractor, was then performed on each sample in
order to determine its asphalt content. Sieve analysis was
also performed on the aggregate of each sample (ASTM
C 136).
Slabs were cut to manageable sizes and stored for process-
ing. The first step was to heat the slabs in the oven at 80⬚C
300 / JOURNAL OF MATERIALS IN CIVIL ENGINEERING / JULY/AUGUST 2001
J. Mater. Civ. Eng. 2001.13:298-305.
for about 4 h. This was followed by separation of the different
asphalt-bound layers. These layers were placed separately in
22-L tin containers with tight covers. Each container was iden-
tified by its content and stored for further processing.
Extensive extraction and binder recovery were performed
on mixes obtained from the different layers of the slabs. These
were done in order to recover the binders and aggregates for
testing.
The following points were taken into consideration through-
out slab testing:
1. Materials from slabs obtained from one section are com-
bined for testing.
2. The different layers of the base course in a section should
be mixed together if they have the same mix design.
3. For maximum specific gravity determination, each slab
was tested individually and the results were averaged.
For the base course, the different layers should be com-
bined.
4. Percent asphalt content (AC ) is determined for each layer
individually. Base course layers of slabs from one section
are combined to determine the asphalt content for the
combined material. Also, the surface course is similarly
treated. For determination of the asphalt content, the re-
flux extractor is used.
5. For separating the asphalt cement and the aggregate for
their subsequent testing, the centrifuge extractor is used.
Aggregates were subjected to the following tests:
• Particle shape and texture (ASTM D 3398)
• Soundness (ASTM C 88)
• Specific gravity (ASTM C 128)
• Gradation (ASTM C 136)
 • Los Angeles abrasion (ASTM C 131)
Extracted asphalt was subjected to the following tests:
• Viscosity at 25⬚C (ASTM D 3570), 60⬚C (ASTM
D 2171), and 135⬚C (ASTM D 2170)
• Penetration (ASTM D 545)
• Softening point (ASTM D 36)
• Composition analysis
Evaluation Parameters
Downloaded from ascelibrary.org by Universidad Politecnica De Valencia on 05/14/15. Copyright ASCE. For personal use only; all rights reserved.
The following parameters were considered in the analysis.
Surface Area
The gradation is known to have a significant effect on the
mechanical properties of the mix. The surface area will give
a good index of the fineness versus the coarseness of the gra-
dation. The Asphalt Institute (TAI) approach described in
Manual Series No. 2 (TAI 1984) was used to calculate the
surface area.
Filler/Asphalt Ratio (F/A)
Many researchers have reported the effect of filler percent-
age on the thermoplastic properties of the asphalt mix (TAI
1984). Filler is defined herein as the material passing the #200
sieve.
Film Thickness
Asphalt content has a pronounced effect on the mix perfor-
mance. However, this depends on the aggregate gradation and
absorption. An index, such as the film thickness, that combines
the effect of the asphalt content, the aggregate gradation, and
absorption was analyzed. The following equation was used to
calculate the film thickness (Bolk et al. 1982):
film thickness = [(A ⫺ B)/C ] 10⫺6 (2)
where A = (100 * AC )/(100 ⫺ AC ); B = Pbe /100 [100 ⫺
(100AC/100 ⫺ AC )]; C = Gb (total aggregate surface
area) ⫻ 62.4 ⫻ 3.2808 ⫻ 100; AC = asphalt content; Pbe =
effective asphalt percentage; and Gb = bitumen specific gravity.
Gradation Curve
Many researchers have found that a hump on the gradation
curve at sieve #30 results in a mix that is prone to rutting or
tenderness. A tender mix will have a hump exceeding 3% up-
ward from a straight line connecting the origin of the 0.45
power gradation chart to the point where the gradation curve
intersects the #4 line.
Temperature Measurement
The national rutting study (Fatani et al. 1990) aimed to col-
lect data for pavement temperatures in typical Saudi and inland
and coastal environments. The locations were chosen for easy
accessibility as well as close proximity to research laborato-
ries; Riyadh, the capital of Saudi Arabia, and Dhahran, in east-
ern Saudi Arabia, were selected for this research program. Ri-
yadh is located at latitude 24⬚ and 42 min north and longitude
46⬚ and 44 min east at an elevation of 611 m. Dhahran is
located at latitude 26⬚ and 16 min north and longitude 50⬚ and
10 min east at an elevation of 17 m. For Dhahran, the location
was selected on the entrance road to King Fahd University of
Petroleum and Minerals (KFUPM), and the test site for Riyadh
was chosen within the premises of the Ministry of Commu-
nications (MOC) research laboratory.
JOURNAL OF MATERIALS IN CIVIL ENGINEERING / JULY/AUGUST 2001 / 301
J. Mater. Civ. Eng. 2001.13:298-305.
In an attempt to evaluate the effect of the thickness of the
asphalt-bound layers on the temperature distribution at various
pavement depths, asphalt concrete slabs of 15, 20, and 30 cm
thickness were built at the test site at MOC in Riyadh. The
Dhahran test location was composed of a 25-cm asphalt-bound
layer on top of a compacted aggregate subbase. The asphalt
slabs were constructed using a 5-cm dense graded wearing
course mix and a dense graded base course mix with variable
thickness.
The mixes met the Ministry of Communications’ specifi-
cations for wearing courses (1.9-cm maximum size) and base
courses (2.5-cm maximum size). The asphalt cement was a
locally produced 60/70-penetration grade.
Temperature may vary considerably throughout the pave-
ment depth, depending on the climatological conditions, pave-
ment depth and surface color, and the number of layers and
properties of each layer. To have a clear picture of the tem-
perature profile through the pavement, it is necessary to record
the temperature at various depths in the pavement structure.
Temperature measurements were taken at the pavement surface
and at depths of 2, 4, 8, and 16 cm, and at the bottom of the
pavement; the air temperature at a height of 1.5 m above the
pavement surface was also measured.
Field instrumentation was initiated by removing 10-cm di-
ameter cores from the pavement slabs, which were then taken
to the laboratory to install thermocouples at the selected
depths. For each core, a groove 3-mm wide and 10-mm deep
was cut out with an electric saw along its entire length. A drill
was then used to drill holes of 2-mm diameter and 5-mm depth
in the base of each groove at depths of 2, 4, 8, and 16 cm,
and near the bottom of the core. These holes were used to
accommodate the tips of the thermocouples.
The thermocouple tips were secured in each hole, using hot
asphalt cement. The lead wires of the thermocouples were all
placed within the groove. A mixture of fine sand and asphalt
was then used to seal the groove completely, leaving the lead
wires emerging from the top of the core. Hot asphalt cement
was used to coat the bottom and the sides of the holes from
where the cores were obtained. After each core was placed in
the corresponding hole, hot sand-asphalt mix was used to fill
the gap between the core and the sides of the hole. Thermo-
couple wires coming out of the core were then placed in a
previously prepared groove leading from the core location to
the edge of the pavement. The groove was about 3-mm wide
and 2-cm deep. Hot sand-asphalt mix was also used to fill this
groove. When temperatures were about to be measured, lead
wires from the thermocouples were plugged into the temper-
ature-reading unit. The temperature of the surface was mea-
sured using a surface probe, while air temperature was mea-
sured using a penetration probe.
RESULTS AND DISCUSSION
Table 2 shows a summary of the obtained test results for
the wearing course layer for both the rutted and the nonrutted
sections. The notations used in Table 2 are as follows:
• Study site: study site identification
• F/A: filler to asphalt ratio
• VFB: percent voids filled with bitumen
• Gmb: mix bulk specific gravity
• AV: percent air voids
• AC: percent asphalt content
• VMA: percent voids in mineral aggregate
• MR 25, 40, and 50⬚C: resilient modulus at the different
specified temperatures (106 kPa)
• Film: thickness of the asphalt film around the aggregate
(␮m)
• Surface: surface area of the aggregate (m2/kg)
 302 / JOURNAL OF MATERIALS IN CIVIL ENGINEERING / JULY/AUGUST 2001
Downloaded from ascelibrary.org by Universidad Politecnica De Valencia on 05/14/15. Copyright ASCE. For personal use only; all rights reserved.

TABLE 2. Summary of Test Results for Wearing Course Layer of Different Asphalt Mixes
Sound. Sound. Sof.
Study VFB AV AC VMA MR 25⬚C MR 40⬚C MR 50⬚C Film Surface Hump CA FA % ⫺ #4 A. vis K. vis Pnt.
site F/Aa (%) Gmb (%) (%) (%) (106 kPa) (106 kPa) (106 kPa) (␮) (m2/kg) value (%) (%) Gsb % #4 % #10 % #40 % #200 ⫹ #100 (poise) (CS) Pen (⬚C)
C1R 0.79b 80.9 2.397 2.0 4.89 10.45 4.37 1.48 0.83 6.4 5.77 6 7.6 7.4 2.546 56.2 41.3 22.8 4.1 47.9 20,601 930 30 60.5
C2N 1.02 67.1 2.353 4.2 5.08 12.64 8.09 3.06 1.40 5.5 6.89 5 4.1 5.5 2.556 58.5 42.5 23.5 5.9 47.0 45,278 1,264 —c 65.3
C3N 1.09 93.9 2.447 0.8 5.07 13.08 8.12 2.58 1.11 5.6 7.05 0 0.7 5.2 2.583 62.2 44.2 22.4 5.5 48.8 — — — —
C4R 1.00 84.4 2.409 1.5 4.87 9.61 6.43 1.88 0.80 6.0 5.93 2 3.3 8.7 2.536 54.0 35.4 19.8 5.1 44.5 33.826 1,352 25 63.2
C5N 1.88 54.9 2.361 5.7 3.93 12.64 7.72 1.70 0.79 4.2 7.45 3 9.0 14.8 2.596 56.9 39.2 22.3 7.7 44.4 2E ⫹ 05 2,344 20 74.2
C6N 1.26 54.8 2.315 7.5 3.96 16.57 10.49 5.76 2.63 4.6 6.43 3 1.1 4.6 2.588 54.7 40.7 22.1 5.0 43.3 — — — —
E1N 1.76 68.5 2.348 4.1 4.97 13.14 6.82 2.44 0.96 4.7 8.64 9 3.9 8.9 2.586 53.9 40.2 27.5 9.2 39.2 64,795 1.843 17 66.2
E2R 1.48 75.2 2.375 3.0 5.05 12.08 6.54 1.90 0.96 4.1 8.17 11 1.6 8.2 2.565 57.4 43.0 28.7 7.9 44.2 20,765 1.258 25 62.1
E3N 1.78 49.1 2.286 7.3 4.41 14.36 6.54 2.03 1.04 4.0 8.14 8 0.6 6.7 2.551 54.2 39.5 26.3 8.2 40.6 48,228 1,559 21 67.6
E4R 0.99 60.6 2.312 5.2 4.58 12.74 5.21 1.29 0.57 5.5 6.58 8 32.7 33.8 2.502 53.0 38.1 25.2 5.9 43.7 42,365 1,646 18 64.1
E5N 2.46 50.4 2.312 4.8 4.43 9.70 6.50 1.76 0.79 2.2 9.84 13 30.0 31.8 2.447 56.0 43.3 31.5 11.2 40.8 66,252 1,953 19 68.9
E6N 1.61 58.4 2.254 8.2 5.17 19.77 9.72 4.63 2.45 3.8 7.90 4 1.3 10.0 2.503 60.2 42.8 24.0 8.3 47.3 — — — —
W1N 1.43 63.4 2.423 4.5 5.63 12.20 8.15 2.86 1.43 3.9 8.73 6 10.2 15.6 2.605 63.5 48.8 28.2 8.5 48.7 93,732 2,128 16 70.0
W2R 2.02 51.4 2.431 4.8 5.31 9.80 7.87 3.53 1.87 2.4 9.10 5 10.3 19.7 2.552 67.5 48.7 27.9 11.3 50.1 65,862 1,920 19 68.6
W3N 2.42 52.9 2.438 7.0 3.70 14.80 8.15 2.82 1.41 4.7 7.17 3 2.3 7.1 2.757 51.0 35.7 20.7 9.3 38.0 2E ⫹ 05 2,278 15 72.2
W4R 2.05 62.8 2.506 4.4 4.60 11.71 7.12 2.43 1.01 4.4 7.12 4 3.4 7.0 2.708 57.4 41.9 23.2 9.9 43.3 43,668 1,420 24 64.6
W5N 1.76 83.0 2.575 2.8 4.77 12.21 3.26 1.04 0.45 5.1 8.06 3 1.6 6.6 2.793 55.6 39.8 22.7 8.8 41.6 11,351 893 34 58.8
W6R 1.71 78.5 2.587 2.0 4.62 9.20 3.01 1.00 0.40 3.8 7.78 4 1.0 5.6 2.719 55.4 39.6 22.9 8.3 41.8 7,552 769 46 54.1
W7N 2.50 70.6 2.408 4.3 4.36 14.72 10.09 5.92 3.16 4.6 8.83 0 2.8 9.9 2.667 66.8 44.1 23.0 10.9 53.2 — — — —
a
Abbreviations are shown in the main text.
b
All individual values are averages of at least three numbers.
c
No readings are obtained.

J. Mater. Civ. Eng. 2001.13:298-305.

 • Hump: percent hump of the gradation curve at sieve #30
• Sound. CA: soundness value of the coarse aggregate (%)
• Sound. FA: soundness value of the fine aggregate (%)
• Gsb: binder specific gravity
• % #4, #10, #40, and #200: percent of the aggregate pass-
ing the specified sieves
• % ⫺ #4 ⫹ #100: percent of the aggregate passing sieve
#4 retained on sieve #100
• A.vis: asphalt absolute viscosity at 60⬚C (Pa ⭈ s)
• K. vis: asphalt kinematic viscosity at 60⬚C (m2/s)
• Pen: asphalt penetration at 25⬚C (0.1 mm)
• Sof. Pnt.: asphalt softening point (⬚C)
Downloaded from ascelibrary.org by Universidad Politecnica De Valencia on 05/14/15. Copyright ASCE. For personal use only; all rights reserved.
show that their rutted and nonrutted means are signifi-
cantly different, which implies also that these parameters
have a direct effect on rutting. These parameters are
VFB, Gmb and AV of the mix, MR at 25, 40, and 50⬚C of
the mix, Surface, Sound. FA, and Gsb of the aggregate
particles.
3. From the above observation, it can be concluded that Gmb
and MR at 25⬚C of the mix are the only common param-
eters having a direct effect on rutting for both wearing
and base course cores. Therefore, these parameters can
be used as indicators for rutting potential in asphalt pave-
ments.

The above parameters were also evaluated for the base course
layers of the selected sections.
In an attempt to find if there are differences between rutted
(R) and nonrutted (N) sections with respect to the performed
tests, an analysis was made of the data from the fast lanes of
both rutted and nonrutted sections. The two sections (R and
N) from the same road were paired and analyzed using a
paired two-sample t-test for means. A separate analysis was
performed for the wearing and base courses. The hypothesis
tested was
H0: ␮N = ␮R versus Ha : ␮N ≠ ␮R
Table 3 shows the statistical analysis results for both wearing
and base courses. If the P-value is less than a specified alpha
(say, ␣ = 0.05), the two means for the specific test are signif-
icantly different, implying that there is a relationship between
this particular parameter and rutting. From Table 3, the fol-
lowing remarks can be made:
MR-Material Correlation
From the previous section, it was found that the resilient
modulus test at 25⬚C is capable of predicting the rutting po-
tential of a pavement section. The resilient modulus test is an
expensive test, and requires special equipment and specialized
personnel. To predict the resilient modulus of an asphalt mix
from the different mix and materials physical properties, step-
wise forward-selection regression was run to relate MR values
with the different mix and materials physical properties. The
purpose of using stepwise regression is to help in selecting a
(3) smaller subset of the variables that affect the MR values to be
included in the model. The criterion for including any factor
in the regression model was set to be 2.0; i.e., the variable
will be included in the model if it adds to the significance
level of the F-ratio a value of 2.0 or greater. In the develop-
ment of the model, individual test values, not averages, were
utilized. The generated MR at any temperature model is

MR at Temp. = 5.354 ⫺ 0.212 Temp. ⫹ 0.111 Sof. Pnt.

1. For wearing course cores, seven parameters show that
their rutted and nonrutted means are significantly differ- ⫺ 0.170 Surface (4)
ent, which implies that these parameters have a direct
effect on rutting. These parameters are Gmb, MR at 25⬚C where all shown variables were previously defined.
of the mix, passing #4 and retained on #100 sieves of The coefficient of determination of this model is 0.85, which
the aggregates, absolute viscosity, kinematic viscosity, indicates that 85% of the variability in MR is explained by the
penetration, and softening point of the asphalt binder. All model. The standard error of the estimate for the suggested
of these parameters have a probability (P-value) of less model is 1.0047. The variables shown in (4) are arranged ac-
than the specified value of ␣ = 0.05. cording to effect; i.e., Temp. has more effect on MR values
2. For the base course cores, another seven parameters than does Sof. Pnt., and Sof. Pnt. has more effect than does
Surface. Other variables that are not included in the model
TABLE 3. Paired Two-Sample t-Test Analysis for Tested Samples indicate that they have a minimal effect on MR or their effect
is explained by the other variables. Fig. 3 shows a plot be-
Wearing course Base course tween the measured MR values and predicted values using the
Test P-value P-value suggested model. It can be noticed that most of the points in
F/A 0.172 0.149 this figure are scattered around the 45⬚ line, which is an in-
VFB 0.179 0.000 dication of the model’s power in predicting the MR value at
Gmb 0.019 0.001
any temperature from the asphalt mix parameters.
AV 0.057 0.000
AC 0.342 0.929
VMA 0.090 0.180
MR 25⬚C 0.046 0.000
MR 40⬚C 0.288 0.000
MR 50⬚C 0.394 0.000
Film 0.639 0.360
Surface 0.097 0.016
Hump 0.760 0.294
Sound. CA 0.889 0.135
Sound. FA 0.991 0.033
Gsb 0.114 0.000
% #4 0.605 0.719
% #10 0.892 0.543
% #40 0.465 0.198
% #200 0.273 0.956
% ⫺ #4 ⫹ #100 0.034 0.458
A. vis 0.045 0.411
K. vis 0.008 0.157
Pen 0.009 0.763
Sof. Pnt. 0.003 0.828
FIG. 3. Measured versus Predicted MR Values

JOURNAL OF MATERIALS IN CIVIL ENGINEERING / JULY/AUGUST 2001 / 303

J. Mater. Civ. Eng. 2001.13:298-305.

 Temperature Correction Factor
When performing back-calculation of the resilient modulus
as part of nondestructive testing, obtained readings are highly
affected by the temperature at which the test is performed.
Therefore, all calculated readings have to be adjusted for the
temperature effect by the use of correction factors. In this re-
search, resilient modulus testing was performed for group I
samples at three different temperatures (25, 40, and 50⬚C). It
was performed for the cores that were extracted from pave-
ment test sections from the three regions of the kingdom (east-
ern, central, and western) and from both the wearing course
and base course layers. To check the possibility of grouping
Downloaded from ascelibrary.org by Universidad Politecnica De Valencia on 05/14/15. Copyright ASCE. For personal use only; all rights reserved.
sured at a depth of 2 cm, while surface temperatures, which
were measured using a surface probe, were equal to those at
2 cm during the winter months only. For the Riyadh slabs,
temperatures increased gradually, reaching their maximum of
65⬚C during the month of July and decreasing thereafter. The
Dhahran slabs reached their maximum of 72⬚C during the
month of August.
Air temperature reaches its lowest level during the month
of January. The minimum recorded air temperatures in the
early morning hours during the test period were 3⬚C and 5⬚C
for the Riyadh and Dhahran test locations, respectively. The
maximum average recorded air temperature for Riyadh was

43⬚C during the month of July, with an absolute maximum of

the test samples from the three regions in addition to samples 47⬚C recorded at 1300 h. The Dhahran site had a maximum
from both asphalt layers, analysis of variance was performed. recorded air temperature of 50⬚C at 1300 h during the month
Test results indicated that at ␣ = 0.05, there is no evidence to of August.
reject the hypothesis of equal means between the different
groups, and that data from the different regions and different Correlation of Air Temperature with Pavement
layers can be grouped. The correction factor is the ratio be- Temperature
tween the MR value at 25⬚C and the MR value at the test tem-
perature. Actual pavement temperature must be considered in design
The obtained correction factor model, which has a coeffi- standards for flexible pavements. However, pavement temper-
cient of determination of 89%, is atures, at different depths in the pavement, are difficult to mea-
sure. Therefore, prediction models for pavement temperature
MR (25) estimation from climatological data, like air temperature, were
correction factor = = 0.1732e(0.0699 Test Temp.)
(5)
MR (T ) tried at different parts of the world. Fatani et al. (1990), in
their study on Saudi Arabian asphalt pavement temperatures,
Temperature Data Analysis found that the single most important factor that affects pave-
Pavement temperature data collection started in April 1988 ment temperature was the air temperature, which is directly
and ended in April 1990. Two complete years of temperature affected by cloud cover and solar radiation.
data are available from the test location at the MOC research A correlation is established between the recorded air tem-
laboratory in Riyadh, and from the test location at KFUPM, perature and the pavement temperature at the surface and at
the eastern province. depths of 4, 8, and 16 cm. Results show a good correlation
Data were analyzed to establish between pavement temperature and air temperature, with a co-
efficient of determination value (R 2) of 93%. Individual test
values were used in developing the correlation lines, which
1. The daily, monthly, and yearly distribution of pavement
have the following form:
temperatures at different pavement depths
2. The correlation between pavement temperatures and air T (d ) = 3.714 ⫹ 1.006T (a) ⫺ 0.146d (6)
temperatures
3. The maximum and minimum pavement temperatures at where T (d ) = pavement temperature at depth d (⬚C); T (a) =
different pavement depths for each month air temperature (⬚C); and d = depth from pavement surface
(cm).
Daily temperature variations indicated that the temperature Tables 4 and 5 show the model fitting results and the anal-
cycle is directly related to the sunshine. The pavement surface ysis of variance for the generated model. In the model fitting
temperature increases steadily, reaching its peak at 1300 h, and results, the t-value statistics and the significance level for the
then decreasing due to the reduction of solar radiation. variables used in the model are given. It is worth noting that
Average monthly temperatures at 1300 h for the 25-cm as- all of the variables used are significant at a level of at least
phalt concrete slab are shown in Fig. 4, where pavement tem- 99.7%. In the analysis of variance table, it is shown that the
peratures at all depths and the air temperature increase grad-
ually, reaching their maximum during the month of August TABLE 4. Model Fitting Results for Pavement Temperature
Prediction Model
and decreasing thereafter. The highest temperatures were mea-
Independent Standard Significance
variable Coefficient error t-value level
Constant 3.713569 1.369743 2.7111 0.0090
Air temperature 1.006466 0.037808 26.6202 0.0000
Depth ⫺0.146248 0.048302 ⫺3.0278 0.0038

TABLE 5. Analysis of Variance for Full Regression for Pavement

Temperature Prediction Model
Degrees
Sum of of Mean
Source squares freedom squared F-ratio P-value
Model 9,261.08 2 4,630.54 358.902 0.0000
Error 683.805 53 12.9020 — —
[Total (correlation)] 9,944.89 55 — — —
Note: R 2 = 0.931241, standard error of estimate = 3.59193, and Dur-
FIG. 4. Average Monthly Temperature Variation for Flexible Pavement bin-Watson statistic = 1.723.
for Riyadh City

304 / JOURNAL OF MATERIALS IN CIVIL ENGINEERING / JULY/AUGUST 2001

J. Mater. Civ. Eng. 2001.13:298-305.

Downloaded from ascelibrary.org by Universidad Politecnica De Valencia on 05/14/15. Copyright ASCE. For personal use only; all rights reserved.
FIG. 5. Resilient Modulus-Temperature Correction Factors Chart for Saudi Arabia
confidence level of the fitted model is 93.12%. A design chart
(Fig. 5) was generated that combines the correction factor cal-
culation and the correlation lines between the air temperature
and asphalt temperature at various depths. Fig. 5 can be used
to determine the temperature at different pavement depths if
the air temperature or surface temperature is known; at the
same time, it can determine the correction factor to predict the
resilient modulus at 25⬚C from the back-calculated resilient
modulus at that given depth. For example, if the asphalt con-
crete layer depth is 8 cm and the measured air temperature is
42⬚C, then the temperature at the surface can be determined
to be 49⬚C, and 47.5⬚C at a depth of 4 cm and 46⬚C at a depth
of 8 cm. The back-calculated resilient modulus at the middle
of the layer (i.e., MR at 47.5⬚C) can be corrected for a tem-
perature of 25⬚C by multiplying by a shift factor of 4.7, as
shown in Fig. 5.
CONCLUSIONS
A comprehensive testing program was adopted to determine
the asphalt concrete mix properties, from which a statistical
model for the asphalt concrete resilient modulus was devel-
oped. The second model was developed to predict pavement
temperature at different depths as a function of air temperature,
while the third developed model was for the resilient modu-
lus–temperature correction factor correlation. These models
can be used in pavement design and analysis for Saudi roads.
ACKNOWLEDGMENTS
The writers wish to acknowledge the Ministry of Communications and
King Abdulaziz City for Science and Technology for providing data re-
quired for this research, and King Fahd University of Petroleum and
Minerals for supporting this research.
JOURNAL OF MATERIALS IN CIVIL ENGINEERING / JULY/AUGUST 2001 / 305
J. Mater. Civ. Eng. 2001.13:298-305.
REFERENCES
Al-Abdul Wahhab, H. I., and Balghunaim, F. A. (1994). ‘‘Asphalt pave-
ment temperature related to arid Saudi environment.’’ J. Mat. in Civ.
Engrg., ASCE, 6(1), 1–14.
The Asphalt Institute (TAI). (1984). ‘‘Mix design methods for asphalt
concrete and other hot-mix types.’’ MS-2, Md.
Bissada, F. (1972). ‘‘Asphalt pavement temperature related to Kuwait
climate.’’ Hwy. Res. Rec. 404, Transportation Research Board, Wash-
ington, D.C., 71–85.
Bolk, N. J., Van der Heide, J. P., and Zantvliet, M. C. (1982). ‘‘Basic
research into the effect of filler on the mechanical properties of dense
asphalt concrete.’’ Proc., Assn. of Asphalt Paving Technologists, Kansas
City, Mo., 51, 398–453.
Dickinson, E. J. (1978). ‘‘A method of calculating temperature gradients
in asphalt concrete pavement structures based on climatic data.’’ Aus-
tralian Road Res., Victoria, Australia, 8(4), 16–34.
Fatani, M. N., Al-Abdul Wahhab, H. I., Balghunaim, F. A., Bubshait, A.,
and Al-Dhubeeb, I. (1990). ‘‘Evaluation of permanent deformation of
asphalt concrete pavement in Saudi Arabia.’’ Progress Rep. No. 6 on
Nat. Res. Proj., King Abdulaziz City for Science and Technology, Ri-
yadh, Saudi Arabia.
Kallas, B. F. (1966). ‘‘Asphalt pavement temperature.’’ Hwy. Res. Rec.
150, Transportation Research Board, Washington, D.C., 1–11.
Stolle, D., and Hein, D. E. (1989). ‘‘Parameter estimate of pavement
structure layers and uniqueness of the solution.’’ Nondestructive testing
of pavements and backcalculation of moduli, STP 1026, A. J. Bush III
and G. Y. Baladi, eds., ASTM, West Conshohocken, Pa., 313–322.
Thompson, M. R., Dempsey, B. J., Hill, H., and Vogel, J. (1988). ‘‘Char-
acterizing temperature effect for pavement analysis and design.’’
Transp. Res. Rec. 1121, Transportation Research Board, Washington,
D.C., 14–22.
Venkataraman, T. S., and Venkatasubramanian, V. (1977). ‘‘Temperature
studies on bituminous pavements in tropics.’’ Australian Road Res.,
Victoria, Australia, 7(1), 32–37.
Williamson, R. H. (1972). ‘‘Effect of environment on pavement temper-
ature.’’ Proc., 3rd Int. Conf. on Struct. Des. of Asphalt Pavements,
University of Michigan, Ann Arbor, Mich., 1, 144–157.