SOILS AND FOUNDATIONS Vol. 48, No.

3, 319–332, June 2008

Japanese Geotechnical Society

CYCLIC TRIAXIAL TESTS ON ASPHALT CONCRETE AS A

WATER BARRIER FOR EMBANKMENT DAMS

SIAMAK FEIZI-KHANKANDIi), ALI ASGHAR MIRGHASEMIii),

ABBAS GHALANDARZADEHiii) and KAARE HOEGiv)

ABSTRACT
The seismic behavior of asphaltic concrete used in embankment dams subjected earthquake loads has been studied.
In order to evaluate the dynamic behavior, an extensive series of monotonic and cyclic tests were carried out on triaxial
specimens of asphalt concrete used in hydraulic structures. The MTS-dynamic equipment at the Norwegian Geotechni-
cal Institute (NGI) was used for this purpose. Temperature and frequency eŠects on specimen behavior and on speci-
men degradation have been studied under the cyclic loads in both isotropic and anisotropic initial stress conditions.
For investigation of the fatigue behavior, thousands of cyclic loads were imposed on some of the specimens.
Moreover, to study any sign of material degradation due to the cyclic loading, the post-cyclic monotonic stress-strain
curve was compared with the corresponding curve for specimens that were not ˆrst subjected to cyclic loading. Geo-
technical parameters to be used in dynamic numerical analysis models are also presented.

Key words: asphaltic concrete core dams, cyclic tests, monotonic tests, seismic behavior (IGC: H4/M3)

dam and showed that relatively large shear strains may

INTRODUCTION occur in the top of the core if the dam slopes are very
The sealing of earth and embankment dams by means steep. However, he concluded that rockˆll dams with
of asphalt concrete cores has attained importance asphaltic concrete core in general have a favorable seis-
throughout the world. This kind of material is virtually mic protection.
impervious, ‰exible, and resistant to erosion and aging Meintjes and Jones (1999) analyzed the Ceres dam lo-
and exhibits visco-elastoplastic behavior (ICOLD, 1992; cated in South Africa. They also used the Newmark
Hoeg et al., 2007). In regions with cold and rainy method to estimate permanent shear displacements. The
weather, construction of this kind of dam is easier than predicted behavior of the dam was satisfactory.
that of clay core dams. For many years, monitoring of Gurdil (1999) performed seismic analyses the Kopru
these dams has indicated their suitable behavior during dam in Turkey. His analyses were based on the equivalent
construction and operation. However, little information linear method. He concluded that some cracking may oc-
exists on the behaviour of asphalt concrete core dams cur in the core, near the crest level. However, the self
subjected to seismic loads. There are only a few published healing behavior of asphaltic concrete will solve this
documents providing information on the behavior of problem.
asphalt concrete used as impervious water barriers in Ghanooni and Mahin-roosta (2002) performed dynam-
dams during and after earthquake shaking. ic analyses on a typical 115 m high asphaltic concrete core
rockˆll dam. They concluded that, in nonlinear analyses,
Previous Numerical Studies the top section of the core experiences small tensile stress-
Valstad et al. (1991) analyzed the Storvatn dam located es which are less than asphalt material strength.
in Norway using a Newmark approach to compute the Feizi-Khankandi et al. (2004) performed a 3-D analysis
earthquake induced permanent displacements along the on a typical 60 m high asphaltic concrete core dam. They
critical sliding surfaces. They studied whether the perma- concluded that as in the case of 2-D analysis, the top sec-
nent shear displacements of a dam due to severe shaking tion of the core experiences some tensile stresses, but
may be so great that a thin core may be sheared oŠ somewhat more than in the 2-D analysis. Furthermore,
toward the dam crest. they concluded that although there is a possibility of
Hoeg (1993, 2005) presented the results of Storvatn some cracking in the top of the asphaltic core, the dam

i)
Ph.D. Student, School of Civil Eng., College of Engineering, University of Tehran, Iran (sfeizi＠ut.ac.ir).
ii)
Associate Professor, ditto (aghasemi＠ut.ac.ir).
iii)
Assistant Professor, ditto (aghaland＠ut.ac.ir).
iv)
Professor, Norwegian Geotechnical Institute (NGI) and University Of Oslo Norway (kaare.hoeg＠ngi.no).
The manuscript for this paper was received for review on March 5, 2007; approved on February 14, 2008.
Written discussions on this paper should be submitted before January 1, 2009 to the Japanese Geotechnical Society, 4-38-2, Sengoku,
Bunkyo-ku, Tokyo 112-0011, Japan. Upon request the closing date may be extended one month.

319

This is an Open Access article under the CC-BY-NC-ND license.

320 SIAMAK ET AL.

could be designed to behave safely. crete on the road and airˆeld pavements. However, on
the water barrier for hydraulic structures such asphaltic
Previous Experimental Work concrete core or face dams, there is not much research in
The ˆrst experimental research in the ˆeld of seismic the literature about the behavior of asphalt concrete, es-
behavior of asphaltic concrete core dams was performed pecially during an earthquake occurrence.
by Breth and Schwab (1973). Their study was based on To investigate the stress-strain behavior of asphalt con-
ˆnite element analysis of a dam with a height of 180 m. crete under static and dynamic loads and for determina-
They devised an interesting set-up to impose computed tion of geotechnical parameters of this material, mono-
cyclic horizontal shear stresses on representative elements tonic and cyclic tests were performed. At least 50 and at
of the asphaltic concrete core. They concluded that the most 10000 cycles were applied to the samples in diŠerent
cyclic loads did not change the structural strength of the conˆning pressures, temperatures and frequencies. The
asphalt concrete which behaved like an elastic body. tests were done in both isotropic and anisotropic condi-
Ohne et al. (2002) performed one-way uniaxial cyclic tions. In addition, some monotonic tests were carried out
tests on the specimens drilled out from Higashifuji dam before and after cyclic tests to investigate post-cyclic be-
in Japan (the asphaltic concrete face dam was damaged havior of asphalt concrete and loading eŠect on the
by an earthquake in 1996). Twenty stress cycles were ap- material strength.
plied at each static stress level. They deˆned the dynamic In this research, the laboratory investigations are divid-
yield strain for the asphalt material. The authors con- ed into three sections. In the ˆrst, monotonic tests were
cluded that the observed cracks that opened in the facing performed to determine stress-strain behavior of the
of the dam were caused by cyclic compression stresses. asphalt concrete before application of the cyclic loads.
Wang (2005) reported a series of cyclic loading tests on Triaxial cyclic tests in both isotropic and anisotropic con-
triaxial specimens of asphaltic concrete. He showed that ditions were carried out in the second part of this study.
there was no sign of cracking or degradation of the speci- Performing monotonic tests after application of the cy-
mens. clic loads, to compare the results with ˆrst section was the
Salemi (2005) performed some numerical and ex- third part of the study.
perimental tests for Meyjaran dam in Iran with a height Brie‰y, the following topics form the main scope of the
of 60 m. Small scale physical models of asphalt core dams present experiment:
were also tested in a centrifuge under impact loads. She Material degradation due to cyclic loading (specially
concluded that her numerical analysis corresponded well after 50 cycles)
with data recorded in the model and mentioned that the EŠects of diŠerent parameters on the behaviour of
asphalt-concrete core behaves safely, even under a very specimens
severe earthquake. Possible cracking of samples due to cyclic loads
It is important to determine the level of tensile stress Behaviour of specimens under thousands of cyclic
and the amount of tensile strain that asphalt concrete in a loads
dam core can sustain before it cracks. This strain level is Determination of geotechnical parameters to be used in
clearly a function of temperature and rate of loading. In the numerical analysis
earthquake prone regions, the asphalt concrete mix is Post-cyclic behaviour of asphalt concrete
usually made with a soft grade bitumen and/or an added
(0.5–1)z bitumen content to increase the ‰exibility and
ductility and the tensile cracking strain (Hoeg et al., PREPARATION OF THE ASPHALT CONCRETE
2007). SPECIMENTS
The tensile or breaking strength of asphalt concrete All specimens were prepared in the asphalt laboratory
decreases with the time of loading or with the increase in of Kolo-Veidekke in Norway. Firstly, small size speci-
temperature. The tensile strength of an asphalt mix is of mens were prepared based on the standardized Marshal
the order of 10z of the compression strength of the method. The size distribution of the sand and gravel in
asphalt concrete (Creegan and Monismith, 1996). The the asphalt concrete mix complied with the Fuller's equa-
most recent paper discussing the tensile strength and ten- tion (Hoeg, 1993; Creegan and Monismith, 1996):
sile cracking strain is the one presented by Nakamura et
Ø dd »
0.41
i
al. (2004). The main goal of their study was to compare Pi＝100 z (1)
max
the engineering properties of conventional asphalt con-
crete with a special admixture (called Super‰ex-phalt). Where: Pi is the percent by weight smaller than the
The Super‰ex-phalt has a much lower tensile strength and equivalent grain size dimension di. These initial tests were
a higher tensile cracking strain than conventional asphalt accomplished to reach the optimum percent of the bitu-
concrete used in hydraulic structures. men value to mix with the aggregates. The used bitumen
was of grade B60 and the tests were done with the bitu-
men percentages by weight of 5, 6, 6.5, 7 and 7.5. The
PURPOSE AND SCOPE OF THE PRESENT type of asphalt binder used is important, because the
RESEARCH shear modulus and the damping ratio are dependent on
Many researches have been done on the asphalt con- the properties of asphalt binders as well. There is a wide
 CYCLIC TRIAXIAL TESTS ON ASPHALT CONCRETE 321

range of bitumen grades to be chosen for hydraulic the tests. Figure 1 shows prepared samples for an exam-
asphalt concrete. The asphalt binder type selection de- ple. Flatness, roughness and parallelism for the specimen
pends on speciˆc project conditions and behavior re- ends had the ability to satisfy the suggested criteria by
quirements of asphalt concrete including its degree of Japanese Geotechnical Society, 2000 and ASTM D
penetration. 3999–91.
Based on the Marshal's test results, the value of 7.0
percent by weight was selected for the mix. The laborato-
ry triaxial specimens were prepared in a mould with a di- MONOTONIC TRIAXIAL TESTS
ameter of 100 mm and a height of 200 mm. Dry ag- Monotonic triaxial compression tests were used to
gregates and the added ˆller, in accordance with the cal- study the stress-strain-strength behavior of asphalt con-
culated weight based on Fuller's equation were put inside crete. Six monotonic tests were performed in diŠerent
the oven to reach a temperature of 1529 C. Besides, bitu- conˆning pressures (Table 1). All prepared specimens
men was put inside another oven for preheating and to were put inside the cooling room with temperature of 59 C
reach a temperature of 1459C (Baron et al., 1955). Both before starting the monotonic tests. Specimens were
of these materials remained inside the oven for ten hours. taken from the cooling room and were set up within a
After this period, the calculated weight for bitumen was triaxial cell. In all tests, there were membranes used
added to the aggregates and then placed in a mixer for around the specimens. The triaxial cell was ˆlled with de-
nearly three minutes. A standard Marshall tamping ham- aerated water and then the equipment was put inside a big
mer was used with 30 blows per layer for compaction of cell used to retain a constant temperature during the test.
the samples (one blow per second). This hammer has a All monotonic tests were performed by use of strain-con-
weight of 4.5 kg and the height of its drop is 45 cm. The trolled compression loading system. After applying the
specimens were built in four layers of equal thickness. Air predeˆned conˆning pressure and reaching to a constant
void of all samples was obtained to be less than 1z. temperature, the axial load was applied up to the failure
Wang and Hoeg (2002) showed that this compaction point. The imposed axial strain rate was 2z per hour. All
procedure gives specimens that have the same compres- tests were continued to a very large axial strain (about
sive strength, but a somewhat higher compression modu- 20z). At that stage, the specimens had a pronounced
lus than ˆeld specimens drilled out of a dam core com- barrel shape and were seriously cracked. During the test,
pacted by a light vibratory roller. the amount of axial stress, axial strain and volumetric
Later on, the samples were trimmed with a special saw strain were recorded by the electronic sensors used for
to a length of 200 mm. During trimming of specimens, a this purpose.
very high sensitive trimmer was used and the surface of
the samples were cut and polished with very high degree Presentation of the Results
of precision to decrease the bedding error eŠect during Table 1 presents the summary of results for monotonic
triaxial tests. Figures 2 and 3 show the values of deviator
stress and volumetric strain versus axial strain for im-
posed conˆning pressures (250, 500 and 1000 kPa). For
all conˆning pressures, the same stress-strain trend is seen
in Fig. 2. As expected, the higher is the value of conˆning
pressure, the more is the amount of failure axial strain.
Values of axial strain at failure point for s3＝1000, 500
and 250 kPa are 15, 6 and 5 percent respectively. Also,
the curves show a good harmony between two repeated
tests in the same conˆning pressures. This similarity is
even more evident in higher conˆning pressures. This is
because, with the increasing conˆning pressure, the sam-
ples, to some extent, behave like rigid materials.
Equivalent Young modulus ( E ) was derived from the
initial stage of the curve, up to an axial strain of 1z:

Table 1. Results of monotonic triaxial tests

Test s3 E s1-s3 (kPa) Temperature Axial strain

No. (kPa) (MPa) at failure (9C) at failure (z)

T1 250 135 2522 5.5

T2 250 110 2197 5.5
T3 500 150 3129 6
5
T4 500 150 3332 6
T5 1000 160 3879 15
Fig. 1. Picture of prepared samples T6 1000 150 3793 15
322 SIAMAK ET AL.

of the aggregate alone, which show a modulus increasing

markedly with increasing conˆning pressures. However,
for the strain values more than 1z, a signiˆcant increase
in shear strength is observed while increasing conˆning
pressures.
To study the eŠect of reduction of bitumen viscosity,
supplementary triaxial tests results were reported by
Hoeg, 1993. The results showed that the same geotechni-
cal parameters are observed with 5.9z B180 and 8z
B60.
Figure 3 shows the relation between volumetric and ax-
ial strains. The results show that with the increase in the
conˆning pressure, the value of dilatancy decreases. Up
to an axial strain of 3z, however, the amount of dilatan-
Fig. 2. Deviator stress- axial strain curves
cy is small, e.g., less than 0.5z. Moreover, this ˆgure
shows a very little volumetric compression at the initial
stage of the tests. It is quite common, as in the ˆrst sec-
onds after the load application; little spaces existing in-
side the specimens raze. After a few seconds, the volume
shows expansion. This important phenomenon is due to
the opening of small ˆssures. Although no visible cracks
may appear, the dilatation may lead to an increase in
permeability. However, the increase in permeability only
occurs when ˆssures get opened in consequence of shear
deformations at a stress level which causes specimen
failure.

TRIAXIAL TESTS WITH CYCLIC LOADING

Twenty-four cyclic triaxial tests were carried out in this
Fig. 3. Degree of dilatancy as a function of conˆning pressure research (Table 2). The specimens were loaded under ini-
tial isotropic condition (kc＝s1/s3＝1.0) and anisotropic
conditions (kc＝2.0 and 3.0). During the tests, the
amount of anisotropy coe‹cient was ˆxed by applying
sd
E＝ (2) the desired loads from the load cell of triaxial equipment.
ea The conˆning pressure was varied from 85 kPa to 500
For conˆning pressures of 250 to 1000 kPa, the range of kPa. As the behavior of the asphaltic concrete core near
the secant modulus is between 110 MPa and 150 MPa the top of the dam is of most concern when it is subjected
(Table 1). to the cyclic loads of earthquake, this range of conˆning
As previously known, the value of elasticity modulus is stress was selected. The cell pressure was generated by the
a function of many parameters like porosity and conˆn- pressurized de-aired water where the cell was fully under
ing pressure (Kramer, 1996). In other words: computer controlled data acquisition system.
The specimens with rubber membranes were placed in
E1z＝f (e, s?0, . . .)＝K×f (e)×f (s?0) (3)
the triaxial cell and then subjected to a conˆning pres-
where: K is a constant parameter and e is the specimen sure. Moreover, the triaxial cell was put inside a bigger
void ratio. cell connected to a water pumping system. This system
In all specimens, the amount of void ratio is less than had the capability of applying any temperature to the
1z. Therefore: pumped water. In the duration of 12 hours, the tempera-
ture gradually reached a predeˆned value. All tests were
E1z＝A×s y0 (4)
performed at two diŠerent temperatures, T＝59 C and T
where: A is a constant parameter. ＝189 C. These temperatures were chosen from the tem-
Based on the monotonic tests results (Fig. 2), the value perature monitoring of the embankment dams (Dan-
of y in the above relation is calculated and the following nicliŠ, 1996). Conditions inside a dam will be rather con-
equation is deˆned for asphalt concrete materials: stant, and selected temperatures to cover some typical
variation, were set to T＝59 C and T＝189 C; T＝59 C, as
E1z＝A×s 0.18
0 (5)
an assumed year-around temperature inside a typical dam
Young modulus for asphalt concrete does not show a in sun-arctic climate and T＝189 C, as an assumed year-
substantial increase while increasing the conˆning stress. around temperature inside dam in countries with tropical
This is in contrast with the results from triaxial samples or sub-tropical climate. On the other hand, in embank-
 CYCLIC TRIAXIAL TESTS ON ASPHALT CONCRETE 323

Table 2. Summarized information of cyclic triaxial tests

Test No. s3 (kPa) Kc T 9C Frequency (Hz) Number of cycles G (GPa) Type of loading

F1 500 1 5 2 50 1.80 A

F2 500 1 5 2 50 1.75 A
F3 500 2 5 2 50 2.50 B

F4 500 2 5 2 50 2.30 B

T5 250 1 5 2 50 1.40 A

T6 250 2 5 2 50 1.67 B

T7 250 3 5 2 50 1.70 B
T8 250 2 5 2 50 1.72 B and C

F9 500 1 5 2 50, 400, 50 2.00 A

F11 500 3 5 2 50, 10000 3.75 B

E12 85 1 5 2 and 10 50, 2000, 50 1.30 A

E13 85 2 5 2 and 10 50, 1000, 1000, 50 1.33 B

F14 500 2 5 2 and 10 50, 1000, 1000, 50 3.20 B

F15 500 3 5 2 and 10 50, 1000, 1000, 5000 4.00 B

T16 250 2 5 2 and 10 50, 1000, 1000, 50 1.6 B

F17 500 3 18 2, 5 and 10 50, 50, 50, 50 1.75 B and C

T18 250 3 18 2, 5 and 10 50, 50, 50, 50 1.25 B

E19 85 1 18 2, 5 and 10 50, 50, 50, 50 0.85 A

T20 250 1 18 2, 5 and 10 50, 50, 50, 50 0.92 A

F21 500 1 18 2, 5 and 10 50, 50, 50, 50 1.00 A

T22 250 3 18 2, 5 and 10 50, 50, 50, 50 1.80 B

E23 85 1 18 2, 5 and 10 50, 50, 50, 50 0.75 A

F24 500 3 18 2, 5 and 10 50, 50, 50, 50 1.90 B

ment dams with clay cores, the eŠect of reservoir temper-

ature is not an important factor, while in asphaltic con-
crete core dams, it would be. As known, nearly six to
eight percent of bitumen is used in the mix design of
asphalt concrete. Since the asphalt binder is a tempera-
ture dependent material, the temperature has a signiˆcant
eŠect on the specimens' behavior. Therefore, the selec-
tion of a suitable asphalt binder to mix with aggregates is
of great importance. The recommendation of selecting
type B180 in cold regions like Norway and type B60 in Fig. 4. EŠective axial stress-loading time, s3＝500 KPa and Kc＝1.0
(Type A)
tropical areas such as Iran are examples of this fact.
Although the results of numerical analysis show that the
greater part of the earthquake energy is in the frequency
range of 2 to 5 Hz and all prepared specimens were im- the starting point is 1500 kPa, reaching 3000 kPa and
posed to loadings of 2 Hz frequency, there were also then decreasing to nearly 0.0 kPa (cyclic load＝±1500
some extra tests carried out in higher frequencies of 5 and kPa).
10 Hz. In most of previous works, regarding the conˆnements
of setting up the loading equipment, there was no pos-
Deˆnition of Loading sibility of applying the two-ways loads to asphalt con-
Figure 4 shows for instance, the applied cyclic loads on crete. Consequently, the upper part of the cyclic loading
the samples subjected to a conˆning pressure of 500 kPa. records was applied to specimens (as indicated in Fig.
The value of axial stress starts from 500 kPa, reaching 5(b)). This type of loading was also used for some tests in
1000 kPa and then decreasing to nearly 0.0 kPa. In the present study.
anisotropic conditions (Kc＝3.0 as an example, Fig. 5(a)), In brief, the following types of cyclic loading used in
324 SIAMAK ET AL.

Fig. 6. Cyclic stress–strain hysteresis loop, Test T5

Fig. 5. EŠective axial stress-loading time, s3＝500 KPa, Kc＝3.0

this research can be noted:

Type A: Isotropic condition with symmetric cyclic load-
ing (Kc＝1.0)
Type B: Anisotropic condition with non-symmetric cyclic
loading (Kc＝2.0, 3.0)
Type C: Anisotropic condition with one-side cyclic load-
ing

Presentation of the Results Fig. 7. Cyclic stress-strain hysteresis loop, Test T20
The MTS system was scheduled to control the number
of cycles. Table 2 summarizes the diŠerent parameters of
these performed cyclic tests. For all specimens, the num-
ber of cycles applied at a given load level was set to 50.
This number of cycles corresponds to a loading induced
by an earthquake with the magnitude of 7.5 in Richter
scale (Kramer, 1996). However, for tests F9 to F24 in
Table 2, there were staged cycles planned and applied;
for example, in test F9, 50 cycles, 400 cycles and then 50
cycles were applied. Moreover, there were small intervals
(about ˆve seconds) between each stage and the next for
better observation of the asphalt concrete degradation
behaviour during the cyclic loading.
Figures 6 to 11 show the hysteresis loop of the cyclic
loading as examples. The hysteresis loops were plotted
for the ˆrst, ˆftieth, hundredths and thousandths cycles. Fig. 8. Cyclic stress-strain hysteresis loop, Test F4
The value of shear stress versus axial strain has been
presented in these ˆgures. The initial value of the loops
was obtained from the value of axial and conˆning pres- the values of Kc＝2.0 and Kc＝3.0.
sures. The starting point position in the hysteresis loop is First, for the isotropic condition (Tests T5 and T20);
deˆned as follows (Figs. 6 to 11): Figs. 6 and 7 have been plotted for instance, at two diŠer-
s －s3 ent temperatures of T＝59 C and T＝189C.
Y position＝Shear Stress in the loop＝ 1


2 Figures 8 and 9 (Tests F4 and E13) are indicated for an
anisotropic state with anisotropy coe‹cient which has the


X position＝Shear Strain in the loop＝0.0

value of 2.0 (Kc＝2.0). In these tests, the temperature
The results are described as follows in two categories; remained constant at T＝59 C.
isotropic condition (Kc＝1.0) and anisotropic states with Figures 10 and 11 (Tests F15 and T18) are shown for an
 CYCLIC TRIAXIAL TESTS ON ASPHALT CONCRETE 325

tained results are aŠected by two sources of error; com-

pliance and bedding error. The compliance of the loading
system, consisting of all parts (top and bottom platens
and connections) between where the specimen deforma-
tion is monitored and the specimen shall be determined
(Tatsuoka and Shibuya, Kohata, 1992, 1995). In the
present study, errors due to apparatus compliance were
evaluated with reasonable certainty by careful calibration
in the laboratory. For this purpose, a cylindrical steel
dummy of a similar size and length to asphalt concrete
specimens was placed into the location normally occupied
by the specimen where some calibration tests were per-
formed. The Young Modulus of the dummy specimen
had a minimum of ten times the modulus of the asphalt
Fig. 9. Cyclic stress-strain hysteresis loop, Test E13 concrete (ASTM D 3999–91). Based on the obtained
results, the correction coe‹cient was deˆned and used in
the main tests.
On the other hand, to decrease the eŠect of bedding er-
ror, the surface of specimens was cut and polished with
very high accuracy (Fig. 1). In addition, bedding error
could cause lower stiŠness in initial cycles and higher stiŠ-
ness in later cycles that was not observed in hysteresis
loops (Figs. 6 to 11). Therefore, the eŠect of bedding er-
ror can be ignored in these tests with the above considera-
tions.

Dynamic Properties for Asphalt Concrete

In the cyclic triaxial tests, the axial stiŠness and damp-
ing parameters can be directly obtained by analyzing the
deviator stress-axial strain loops (Fig. 12). Cyclic triaxial
tests are traditionally oriented to analyses of cyclic behav-
Fig. 10. Cyclic stress-strain hysteresis loop, Test F15 ior, described by the relationship between the dev-
iator/radial eŠective stress ratio (q/s?r) and the number of
cyclic loads. The upper part of the shear stress-axial
strain hysteresis curves is used to calculate the shear
modulus. The following relations would then be used for
this purpose:

t
E＝ , g＝(1＋n)ea
2e a
E
G＝ (6)
2(1＋n)

Where: t＝shear stress, ea＝axial strain, g＝shear strain

and n＝Poisson ratio.
Table 2 presents the complete information for all per-
formed tests. The shear modulus increases from 1.5 GPa
in the isotropic condition to nearly 4.0 GPa for
anisotropic state with anisotropy coe‹cient of Kc＝3.0,
Fig. 11. Cyclic stress-strain hysteresis loop, Test T18 at a low temperature of T＝59 C. For the higher tempera-
ture (T＝189C), the value of shear modulus decreased to
half or less as much as the mentioned values. Although
anisotropic state with anisotropy coe‹cient of 3.0 (Kc＝ the amount of shear modulus can change during cyclic
3.0) for two diŠerent temperatures of T＝59C and T＝ loading, the 10th cycle was chosen to calculate the shear
189 C. modulus. While it is observed that in the isotropic condi-
tion and low temperature, the conˆning stress does not
Accuracy of Measurements have signiˆcant eŠect on the shear modulus, its eŠect in-
The ˆdelity of the results depends on the accuracy of creases with increase of anisotropy and temperature. By
the measurements of both stresses and strains. The ob- the following equation, the damping ratio (D) is calculat-
326 SIAMAK ET AL.

Table 3. Comparison of dynamic parameters between asphalt con-

crete and the other materials

Material Gsec (MPa) Damping (z)

Asphalt concrete 700¿4000 5¿30

Crushed rock 200¿500 2¿35

Round rock 150¿300 2¿20

Sandy-gravel 100¿200 5¿20

Sand º100 2¿15

Plastic concrete 500¿5000 2¿30

Fig. 12. Cyclic triaxial test

(a) Loads on asphalt concrete specimen Investigation of DiŠerent Parameters on Dynamic Prop-
(b) Interpretation erties
Figures 13, 14 and 15 present the eŠect of diŠerent
parameters such as conˆning stress, anisotropy, loading
ed as: frequency, temperature and hysteresis loop shapes on the
1 WD shear modulus and damping ratio. In the following para-
D＝ ×100(z) (7) graphs, the eŠects of the above-mentioned factors are de-
4p W S
scribed in detail.
where, WD is the damping energy in a single loading cycle a) EŠect of anisotropy
and WS is the equivalent elastic energy Based on the In each part of Fig. 13, the temperature and anisotropy
presented curves, the areas of the hysteresis loops and the coe‹cient were ˆxed while the conˆning pressure was
indicated triangle were calculated. The values of the varied. It can be seen that the dynamic shear modulus of
damping ratio range between 0.066 and 0.35. In the same asphalt concrete is strongly dependent upon the shear
asphalt binder percent, binder type B180 causes the strain. At low strain amplitudes, the shear modulus is
decrease of the shear modulus while increasing the damp- high, but it decreases while the strain amplitude increas-
ing ratio compared to type B60 (Hoeg, 1993). ing. Moreover, by comparing Figs. 13(a) with (d) (at T＝
To show the characteristics of asphalt concrete relative 59C) or Figs. 13(b) with (e) (at T＝189C), it can be seen
to the other geotechnical materials, dynamic properties that the higher the value of anisotropy coe‹cient, the
of asphalt concrete were compared with those of other ge- higher the amount of shear modulus that is obtained.
otechnical materials (Table 3). Based on the stiŠness This is because at higher values of anisotropy state, the
values, asphalt concrete can be laid between soft rocks amount of smean(＝s1＋s3/2) increased and this was the
and soils. In comparison with crushed-rock (G0§2000 to main reason for the shear modulus augmentation.
500 MPa), round-gravel (G0§150 to 300 MPa), sandy- b) EŠect of temperature
gravel (G0§100 to 200 MPa) and sand (G0Ã100 MPa) Comparison of shear modulus presented at the same
(Kokusho and Esashi, 1981), asphalt concrete is a stiŠer anisotropy coe‹cients in Figs. 13 ((a) and (b) at Kc＝1.0)
material. It should be emphasized that the presented or Figs. 13 ((d) and (e) at Kc＝3.0) stands for decreasing
range of shear modulus is aŠected by the value of conˆn- the shear modulus as the temperature increases. In addi-
ing stress and void ratio. The results in Table 2 show that tion, the temperature has an important eŠect on the
the amount of shear modulus for asphalt concrete ranges threshold point position in G-g curves. It is clear that
between 700 MPa to 4 GPa. These values correspond well with the increase in the temperature, the amount of shear
to sedimentary soft rocks (Tatsuoka and Shibuya, 1992) modulus falls faster.
and plastic concrete used as a cut-oŠ wall in the embank- Figure 14 shows the damping ratio versus axial strain
ment dams (Mahab-Ghodss, 2007). In addition, low-am- in diŠerent conˆning stresses at two diŠerent tempera-
plitude shear modulus for undistributed gravelly-sand tures of T＝59C and T＝189 C. Figure 14(a) shows that in
(Kokusho and Tanaka, 1994) ranges between 1 GPa to 5 comparison with shear modulus, damping ratio is not sig-
GPa which, can be considered to be similar to that of niˆcantly dependent on the shear strain amplitude.
asphalt concrete. However, its value increases gradually with increase of
On the other hand, it should be noted that the damping the shear strain level. It is observed that dependency on
ratio percentage obtained in present study for the asphalt the strain is more distinct at the higher temperature of T
concrete is high, (especially in higher temperatures which ＝189 C than that at a lower temperature of T＝59C (Fig.
is about 5z to 35z). Coarse material and plastic con- 14(b)). Furthermore, by increasing the temperature, the
crete can be put in this damping range. damping ratio increases. That is because in high tempera-
tures, viscosity of asphalt concrete causes the material to
easily absorb the applied energy. Therefore, the dissipat-
ed energy in a single loading cycle and consequently the
 CYCLIC TRIAXIAL TESTS ON ASPHALT CONCRETE 327

Fig. 13. EŠects of s3 and Kc on strain-dependent modulus at two diŠerent temperatures of 59C and 189C

damping ratio increases. types of modulus. It shows that the value of shear modu-
c) EŠect of conˆning stress lus in compression region is more than two times of the
The eŠect of conˆning stress on shear modulus is also value in extension side in the same temperature and con-
illustrated in Fig. 13. For an example in Fig. 13(a), at the ˆning stress (GcÆ2Ge). It is clearly seen in Fig. 15(c) that
constant temperature and anisotropy coe‹cient, the with increasing conˆning stress and/or anisotropy
higher values of conˆning stress, the higher amount of coe‹cient at the constant temperature, the values of Gc
shear modulus is obtained. Comparison between Fig. and Ge increase. Moreover, Fig. 15(b) shows that the
13(a) and (d) (or Fig. 13(b) and (e)) shows that at a high values of Gc and Ge decrease at the higher temperature.
level of anisotropy such as Kc＝3.0, the eŠect of conˆning e) EŠect of reversal coe‹cient (rc)
stress is more distinct. In addition the threshold point oc- Reversal Ratio is introduced by the reversal coe‹cient
curs at higher amplitudes of shear strain in comparison (rc), which is the proportion of the positive portion in ap-
with a low conˆning stress. This point is observed in Figs. plied cyclic shear stress to the whole domain of the shear
13(a), (c) and (d) as examples. Figure 14(a) shows that by stress. The eŠect of stress reversal ratio in the G-log N di-
decreasing the conˆning pressure from 500 kPa to 85 agram has been also studied, (where N is the number of
kPa, the damping ratio increases from 5z to 35z respec- applied cycles.)
tively. Figure 16 shows the G-log N curves for diŠerent values
d) EŠect of hysteresis loops shapes of rc. It is observed that the larger the value of rc, the
Because of the diŠerence between the curve inclinations higher is the curve in the G-log N diagram. This means
in compression and extension regions, two types of shear the potential for degradation increases when the exten-
modulus, Gc and Ge, were calculated from the hysteresis sion mode has become more predominant. In addition,
loops (Figs. 6 to 11 and 12). The eŠects of conˆning stress this ˆgure shows that in lower values of rc (such as rc＝
(Fig. 15(a)), anisotropy coe‹cient and the temperature 0.5), the amount of degradation for shear modulus is less
(Figs. 15(b) and (c)) were plotted separately for these two than that of higher values (rc＝0.85). In addition, this
328 SIAMAK ET AL.

Fig. 16. G-Log N curves for diŠerent values of rc

Fig. 14. EŠects of conˆning stress and temperature on strain-depen-

dent damping ratio

Fig. 17. EŠect of cycles number on shear modulus

ˆgure shows that increasing the number of cycles caused

decrease of shear modulus.
f) EŠect of number of cycles
Figure 17 presents the eŠect of the number of cycles on
modulus reduction behavior at diŠerent conˆning stress-
es, anisotropy coe‹cients and temperatures. The value of
shear modulus is plotted in the ˆrst and ˆftieth cycles. It
is seen that holding constant the values of conˆning
stress, anisotropy coe‹cient and temperature, by increas-
ing the number of cycles, the amount of shear modulus
Fig. 15. Comparison between shear modulus in extension and com-
decreases (Figs. 17(a), (b) and (c)). This reduction behav-
pression states at two temperatures of 5 9C and 189C with diŠerent ior is more distinct at a low level of shear strain. In addi-
conˆning stresses tion, the shear modulus reduction behavior is more
 CYCLIC TRIAXIAL TESTS ON ASPHALT CONCRETE 329

prevailing in a low conˆning stress (s3＝250 kPa) than

higher conˆning stresses (Fig. 17(a) for an example). At
the same temperature and conˆning stress, comparison
between Figs. 17(a) and (b) shows that with increase of
anisotropy coe‹cient, the eŠect of the number of cycles
decreases. In other words, in the same cycle number, the
specimens degradation with a higher value of anisotropy
coe‹cient (Kc＝3.0) is less than that of Kc＝2.0. Figures
17(b) and (c) shows that the value of temperature has a Fig. 18. s3-De curves for diŠerent values of frequency
signiˆcant in‰uence on the dynamic properties of asphalt
concrete between the ˆrst and ˆftieth cycles. Indeed, in
high temperatures, the threshold point for G-g curves oc-
curs at a low level of shear strain.

Strain Values and Specimens Cracking

In the present research, because of using ASTM stan-
dards at the laboratory, the external transducers were
used to measure deformations. Though the research
developments in advanced triaxial equipments (e.g., Tat-
suoka et al., 1992, 1995) have shown that shear strains as
low as 0.001z can be resolved in static (e.g., Goto et al.,
1991) and cyclic triaxial tests by using local displacement
transducers, however, with proper mounting as well as
carefully calibrating the high resolution transducer and
by considering the eŠects of equipment compliance and
bedding error (ASTM D 3999–91), the reliability of the
results can be put in an acceptable range. Fig. 19. Cross setion of the specimen after 10000 cycles
As expected, the value of strain was very small. Hence,
a very high precision electronic displacement transducer
was used to record the value of displacement during the against cyclic loads.
cyclic loads. According to ASTM D 5311–92 and
3999–91, displacement measuring devices such as LVDT Degradation of the Asphalt Concrete Specimens
may be used if they have an accuracy of ±0.02z of the The shape and inclination of the hysteresis curves
initial specimen height. Since specimens height in the would be good criteria to investigate the material degra-
present study was 200 mm, the accuracy of ±0.04 mm dation. Hence, the average inclination of the ˆrst hystere-
had to be the minimum required accuracy for the used sis loop was calculated and compared with the slopes of
LVDT. The LVDT used at the laboratory satisˆed this other loops. This comparison showed that by increasing
condition very well and other criteria suggested in the the number of cycles, the amount of curves inclination
ASTM standards. decreased (Figs. 6 to 11 and 17). It was also observed that
Figure 18 summarizes the axial strain (De) for per- the value of shear modulus was gradually decreasing dur-
formed tests at the end of the loading in diŠerent frequen- ing the cyclic loading. Some tests were run with thou-
cies. In high speeds of cyclic loading, the strain values sands of load cycles to study whether there is a long-term
decrease ( F＝5 Hz and 10 Hz). Since, in higher loading degradation (fatigue) phenomenon which was not found
speeds, asphalt concrete cannot show its ‰exibility and noticeable to be the case up to 10000 cycles (Fig. 19).
viscosity behavior very well, consequently the axial strain Figures 6 to 11 show that the banana shape was seen in
values are less than those of low speeds. Good compac- the extension mode. In the presented ˆgures (Figs. 6 to
tion of the samples was one of the reasons of rather small 11), the curves' inclination in extension mode is less than
displacements. In addition, the value of axial strain in- that of the compression mode. Therefore during the cy-
creases with the increase of conˆning pressure. With clic triaxial tests, specimen strength reaches the failure
reference to Figs. 13 and 15, it is obvious that the temper- line in extension mode. Consequently the values of axial
ature has the largest eŠect on the strain values. As expect- strain increases. With the application of compression
ed, the higher the value of temperature, the greater is the loads, the specimens' behaviour is changed. However,
amount of axial strain. some residual strain remains. It is one of the explications
After the cyclic tests, specimen surfaces were well in- to describe the banana shape in the extension region.
spected. In addition, some specimens were cut horizon- Anisotropy in asphalt concrete; because of the direction
tally and vertically to investigate the cracking in the in- of compaction, is another reason to make the banana
terior surfaces (Fig. 19). There was not any sign of looking shape in the extension mode.
cracks, even after 10000 cycles. This shows a good
response of the asphalt concrete specimens resisting
330 SIAMAK ET AL.

Description of DiŠerent Behaviors of the Asphalt Con- ＝189 C, in all conˆning pressures, the extension behavior
crete was observed while at the temperature of T＝59 C, this
During the cyclic triaxial tests of asphalt concrete, two behavior was just seen at s3＝85 kPa. In other words, in a
diŠerent behaviors were observed; Extension and Com- low level of conˆning pressure (s3＝85 kPa), only the ex-
pression, which are summarized in Table 4. According to tension behavior is observed and the temperature eŠect
this table, the extension behavior may occur in a low level can be negligible.
of conˆning stress such as near the dam crest and hence Near the dam crest the amount of conˆning pressure is
this part of dam is more vulnerable during the cyclic load- not considerable. It is also well established that this part
ing. of the core is very vulnerable during the earthquake.
Table 4 shows that during the cyclic tests with higher Therefore, special control of the dam during the con-
values of Kc, the compression behavior occurs. Although struction on this region would be necessary.
the amount of strain is very small, the loads compress the
material. For anisotropic condition, changing the tem-
perature, conˆning pressure and loading frequency aŠect POST-CYCLIC BEHAVIOUR OF THE ASPHALT
the strain values only and do not alter the general behav- CONCRETE SAMPLES
ior (compression) of the specimens. After an earthquake shocking, the structures should
As mentioned above, the extension behavior is only retain their e‹ciency and operate normally. The men-
seen in the isotropic state (Kc＝1.0). Increasing the tem- tioned period is titled the ``post-cyclic operation time''.
perature or decreasing the conˆning pressure are the main To simulate this occurrence, after completing the cyclic
factors causing extension. In this state, compression be- tests, some specimens were selected to be imposed by
havior was just seen at a low temperature (T＝59 C) and monotonic loading. The post-cyclic monotonic stress-
high conˆning pressures (s3＝250,500 kPa) while exten- strain curve would be compared to the corresponding
sion behavior was seen in other cases. curve for the specimens not ˆrst subjected to cyclic load-
Since the extension behavior is just observed at Kc＝
1.0, the eŠects of temperature and conˆning stress on the
general behavior of asphalt concrete is explained more in
the following paragraphs:
a) EŠect of temperature (in isotropic condition)
The cyclic tests were performed at two diŠerent tem-
peratures, T＝59 C and T＝189C. As presented in Table
4, in the constant conˆning stress, extension behavior oc-
curred in a higher temperature. For the reason that in a
higher temperature, the eŠect of material viscosity is
more distinct and the increment of conˆning pressures
can not substantially aŠect the compression behavior of
the material. It should be noted that the prepared speci-
mens for the cyclic tests are bitumen rich. The so called
``Rich'' is used for the specimens with the high percen-
tage of the bitumen. In a low temperature (T＝59 C), the
sample behaves rigidly. With increasing temperature, the Fig. 20. Post cyclic behaviour (stress-strain curve), s3＝500 KPa, T＝
compaction, mixture quality of aggregate with bitumen 59C
and the particles interlocking are more in‰uential.
b) EŠect of conˆning stress (in isotropic condition)
The cyclic tests were performed in three diŠerent con-
ˆning pressures, s3＝85, 250 and 500 kPa. By increasing
the conˆning pressure, the compression behavior is more
distinct than that of extension; this is speciˆcally more
remarkable at low temperatures. At the temperature of T

Table 4. DiŠerent types of asphalt concrete behavior during the cyclic

loading

s3 Kc＝1.0 Kc＝2.0 Kc＝3.0

(kPa) T＝59C T＝189C T＝59C T＝59C T＝189C

85 Extension Extension

250 Compression Extension Compression Compression Compression

500 Compression Extension Fig. 21. Post cyclic behaviour (stress-strain curve), s3＝250 KPa, T＝
189C
 CYCLIC TRIAXIAL TESTS ON ASPHALT CONCRETE 331

ing, to study any sign of material degradation due to the After the completion of cyclic loading, monotonic tests
cyclic loading. Figure 20 shows the comparison of the were carried out on the samples to investigate the post-
stress-strain curve for one of the specimens before and cyclic behaviour. The results show that the asphalt con-
after the cyclic loading at a low temperature of T＝59C. crete behaves much the same way as prior to cyclic
The cyclic tests were performed at diŠerent anisotropy loading. However, by increasing the temperature, the
states (Kc＝1.0 and Kc＝3.0). The amount of degradation amount of degradation increases. Post-cyclic behav-
is nearly 15 percent at the pick point of the curve. In addi- iour shows that the reduction in shear strength after cy-
tion, the ˆgure shows a similar overall behavior for the clic loading is insigniˆcant. The increase of permeabil-
samples before and after subjecting to cyclic loading. ity only occurs when ˆssures get opened near the failure
Figure 21 shows the monotonic test results for the level in monotonic loading.
specimen after cyclic loading in a high temperature (T＝ This study shows that asphaltic concrete is resistant to
189 C). The ˆgure shows the same behavior trend for the earthquake excitations. The earthquake has to be very
tests before and after the cyclic loading in a high tempera- strong to cause any detrimental cracking or material
ture of T＝189C like T＝59C. degradation of the properties of a ductile asphaltic
It can be concluded that the asphalt concrete retains its concrete core in embankment dam.
e‹ciency after cyclic loads application and the post-cyclic
behaviour of this material is still suitable.
ACKNOWLEDGMENT
The present research was supported by the Iran Water
SUMMARY AND CONCLUSIONS Resources Management Organization (IWRMO) and
Lack of high quality experimental data on the asphalt Mahab-Ghodss consulting engineers in Iran and the con-
concrete subjected to earthquake loading was the main tractor Kolo-Veidekke in Norway. The authors appreci-
incentive to perform this research. The outcome of the ate the assistance of laboratory employees at the Nor-
present study shows the behavior of asphalt concrete un- wegian Geotechnical Institute and Kolo-Veidekke during
der cyclic loading. During the design procedure, it is the experimental work.
necessary to have exact material properties to control the
dam stability.
The results obtained from the present study can be NATATION
summarized as follows: The following terms are utilized in this research:
Triaxial monotonic tests were performed to study the e: void ratio of specimens
stress-strain behavior of asphalt concrete material. n: porosity of specimens
Strength and stiŠness increased with higher conˆning y: Poisson ratio
stress, s3. Based on the monotonic tests results, the s3: conˆning stress, is a pressure applied into the triaxial
Young's secant modulus at 1z axial strain is proposed cell
to be E1z＝A×s0.18 0 . In addition, higher conˆning s1: axial stress, is applied in the axial direction of the
stresses imposed on the specimens caused lower specimen, while lateral stress is applied in the radial direc-
dilatancy and volume expansion during shearing. tion of the specimen
In the cyclic triaxial tests, ˆfty cycles were imposed on sd: deviator stress, is the diŠerence between major and
the specimens to simulate earthquake excitations. In minor principal stresses in a triaxial test
some cases, the cyclic loads were continued to thou- Kc: anisotropic stress ratio, is calculated by dividing the
sands of cycles. However, there was no signiˆcant axial stress by lateral stress (Kc＝s1/s3)
degradation detected on the specimen behavior. No Reversal ratio, is introduced by the reversal coe‹cient
cracks on the specimen surfaces were detected, even af- (rc), which is the relative value of positive portion of ap-
ter 10000 cycles. plied cyclic shear stress to the whole domain of shear
Many factors in‰uence the dynamic properties of the stress
asphalt concrete, such as conˆning stress, stress G: shear modulus, is calculated from hysteresis loops. Gc
anisotropy, loading frequency and temperature. The and Ge are deˆned for the compression and extension
eŠects of the mentioned factors are more distinct on regions inclination, respectively
the shear modulus than on the damping ratio. The dy- D: damping ratio, is carried out from hysteresis loops
namic shear modulus of asphalt concrete is strongly de- Degradation, is the reduction amount of material
pendent on the shear strain. strength
The damping ratio increases with the increase of dy- Threshold point, is the point that separates the constant
namic strain at lower stress ratios (Kc), while being con- and falling parts of the G-g curve
stant at higher stress ratios. Compression behavior, is the shortening of the
The cyclic strain values were less than 0.20z for the specimen's height under the cyclic loading
performed tests. The cyclic amplitude remains constant Extension behavior, is the elongation of the specimen's
even for a large number of cycles. The smallest values height under the cyclic loading
of strain occur for low temperatures and high fre-
quency loading.
332 SIAMAK ET AL.

15) Kokusho, T. and Esashi, Y. (1981): Cyclic triaxial test on sands and
REFERENCES coarse materials, Proc. 10th ICSMFE, (Quoted by Ishihara 1986),
1) Baron, W. F. and Van Asbeck (1955): Bitumen in Hydraulic En- Stockholm, Sweden.
gineering, Shell Petroleum Co., Ltd., 1, London, England. 16) Kokusho, T. and Tanaka, Y. (1994): Dynamic properties of gravel
2) Breth, H. and Schawab, H. H. (1973): Zur Eignung des asphaltbe- layers investigated by in-situ freezing sampling, Proc. Ground
tons fur die Innendictung von Staudammen, Wassewirtschaft, 69, Failure under Seismic Conditions, ASCE Annual Convention, At-
Heft 11, 348–351, Stuttgart, Germany. lanta, USA.
3) Creegan, P. and Monismith, C. (1996): Asphaltic Concrete Water 17) Kramer, S. (2007): Geotechnical earthquake engineering, Cut of
Barriers for Embankment Dams, ASCE Press. Wall for Gotvand Dam, Prentice Hall, Inc, USA, 1996–Mahab-
4) DunnicliŠ, J. (1996): Geotechnical Instrumentation for Monitoring Ghodss report, Iran.
Filed Performance, 2nd Edition. 18) Meintjes, H. A. C. and Jones, G. A. (1999): Dynamic analyses of
5) Feizi-Khankandi, S., Mirghasemi, A. A. and Ghanooni, S. (2004): the new cores dam, Proc. 12th Regional Conference for Africa on
3-D seismic analysis of asphaltic concrete core rockˆll dams, ICGE SMGE, Durban, South Africa.
Conference, 220–225, UAE. 19) Nakamura, Y., Okumura, T., Narita, K. and Ohne, Y. (2004): Im-
6) Ghanooni, S. and Mahin-roosta, R. (2002): Seismic analysis and provement of impervious asphalt mixture for high ductility against
design of asphaltic concrete core embankment dams, Journal of earthquake, Proc. 4th International Conference on Dam Engineer-
Hydropower and Dams, 6, 75–78. ing, 18–20, China.
7) Goto, S., Tatsuoka, F., Shibuya, S., Kim, Y. S. and Soto, T. 20) Ohne, Y., Nakamura, Y., Okumura, T. and Narita, K. (2002):
(1991): A simple gauge for local small strain measurements in Earthquake damage and its remedial measure for earth dams with
laboratory, Soils and Foundations, 31(11), 169–180. asphalt facing, Proc. 3rd US-Japan Workshop on Earthquake En-
8) Gurdil, A. F. (1999): Seismic behavior an Asphaltic Concrete core gineering for Dams, 15–26, Japan.
dams, Proc. 1st Symposium on Dam Foundation, Antalya, Turkey. 21) Salemi, S. (2005): Dynamic behavior investigation of asphaltic con-
9) Hoeg, K. (1993): Asphaltic Concrete Cores for Embankment crete core Rockˆll Dams, IUST University, PhD Dissertation, Iran.
Dams, Norwegian Geotechnical Institute, Oslo, Norway. 22) Tatsuoka, F. and Shibuya, S. (1992): Deformation characteristics
10) Hoeg, K. (2005): Earthquake resistance of Asphaltic concrete core, of soils and rocks from ˆeld and laboratory tests, Report of the In-
Report No. 20051031–1, Oslo, Norway. stitute of Industrial Science, University of Tokyo.
11) Hoeg, K., Valstad, T., Kjaernsli, B. and Ruud, A. M. (2007): 23) Tatsuoka, F. and Kohata, Y. (1995): StiŠness of hard soils and soft
Asphalt core embankment dams: recent case studies and researches, rocks in engineering applications, Report of the Institute of Indus-
Journal of Hydropower and Dams, 13(5), 112–119. trial Science, University of Tokyo.
12) ICOLD Press (1982, 1992): Bituminous Cores for Earth and Rock- 24) Valstad, T., Selness, P. B., Nadim, F. and Aspen, B. (1991): Seis-
ˆll Dams, Bulletin 42 and 84. mic response of a rockˆll dam with an asphaltic concrete core,
13) International Navigation Association Press (1997): Seismic Design Journal of Water Power and Dam Construction, 43, 1–6.
Guidelines for Port Structure, working group No. 34 of the Mari- 25) Wang, W. (2005): Cyclic Tests on Asphalt Concrete, Xi'an Univer-
time Commission International Navigation Association. sity Press, China.
14) Japanese Geotechnical Society (2000): Standards of Japanese 26) Wang, W. and Hoeg, K. (2002): EŠects of compaction method on
Geotechnical Society for Laboratory Shear Test, Japan. the properties of asphalt concrete for hydraulic structures, Interna-
tional Journal on Hydropower and Dams, 9(6), 63–71.