Determination of Parameters For Hyperbolic Model of Soils
Determination of Parameters For Hyperbolic Model of Soils
Determination of Parameters For Hyperbolic Model of Soils
Civil Engineers
Geotechnical Engineering 156
April 2003 Issue GE2
Pages 105–117 N. Al-Shayea S. Abduljauwad R. Bashir H. Al-Ghamedy I. Asi
Associate Professor, Professor, Department of Research Assistant, Associate Professor, Assistant Professor,
Paper 13008 Department of Civil Civil Engineering, King Fahd Department of Civil Department of Civil Department of Civil
Engineering, King Fahd University of Petroleum & Engineering, King Fahd Engineering, King Fahd Engineering, King Fahd
Received 29/07/2002 University of Petroleum & Minerals, Saudi Arabia University of Petroleum & University of Petroleum & University of Petroleum &
Accepted 25/11/2002 Minerals, Saudi Arabia Minerals, Saudi Arabia Minerals, Saudi Arabia Minerals, Saudi Arabia
Keywords: geotechnical
engineering/mathematical
modelling/strength & testing of
materials
Analysis of some geotechnical problems using finite- ˜åvol change in volumetric strain
element methods requires the implementation of a non- ˜óm change in mean stress
linear model for soil materials, to better represent their ˜ö reduction in angle of internal friction between
actual behaviour. Constitutive modelling of soil mass ó3 ¼ Pa and ó3 ¼ 10Pa
behaviour and material interfaces is an essential å axial strain
component of the solution of boundary and initial value åu ultimate volumetric strain at large stress
problems. The hyperbolic model is one of the most åvol volumetric strain
frequently used non-linear models for predicting the rd dry density
behaviour of soils in boundary value problems. The rd,max maximum dry density
parameters of this model for specific soils need to be rfield field density
determined experimentally. This paper presents the rmax maximum density
results of extensive laboratory testing carried out on rmin minimum density
three soils from the Eastern Province of Saudi Arabia: ó normal stress on failure plane
sand, marl and sabkha. The tests used to obtain these óm mean stress (confining pressure for hydrostatic
results were the triaxial compression test, the load)
hydrostatic (isotropic compression) test, and the direct ón normal stress
shear test. Additionally, other tests were used to obtain ó1 major principal stress
various physical properties needed for the complete ó3 minor principal stress, confining pressure (in
characterisation of these soils. Parameters of the triaxial test)
hyperbolic models for non-linear tangent Young’s and ó1 ó3 deviator stress
bulk moduli are presented. These parameters compare (ó1 ó3 )f actual deviator stress at failure
well with those reported in the literature. They are (ó1 ó3 )u ultimate deviator stress at large strain
incorporated in the hyperbolic model and used to back- (theoretical)
predict the stress–strain behaviour of the investigated ô shear strength of soil
soils. The calibrated models are found to predict soil ô shear stress
behaviour very well. ö angle of internal friction
öo angle of internal friction at ó3 ¼ Pa
NOTATION
B bulk modulus
Bi initial tangent bulk modulus 1. INTRODUCTION
Bt tangent bulk modulus Soils are very complicated engineering materials, whose
C cohesion constitutive response depends on many compositional and
Cc coefficient of curvature environmental factors. The problem manifests itself in the non-
Cu coefficient of uniformity linear deformation of soil under mechanical loads. The
Dr relative density availability of high-speed computers and powerful numerical
Ei initial tangent Young’s modulus techniques (such as the finite-element method) makes it
Et tangent Young’s modulus possible to incorporate the non-linear behaviour of materials
Gs specific gravity into the analysis of soil systems and soil–structure interaction
K hyperbolic parameter 1 for Young’s modulus problems. Some advanced soil models have been proposed for
n hyperbolic parameter 2 for Young’s modulus the non-linear stress–strain behaviour of soils, including the
1 2,3
Pa atmospheric pressure hypoelastic models, the hyperelastic models, and the
4
Rf failure ratio plasticity models. However, these models require the
w water content determination of many parameters for the investigated soils.
wn natural water content
wopt optimum water content Non-linear elastic or piecewise linear elastic models have been
˜H horizontal displacement developed to account for the influence of stress or strain on
Geotechnical Engineering 156 Issue 2 Hyperbolic model of soils Al-Shayea et al. 105
Delivered by ICEVirtualLibrary.com to:
IP: 148.212.219.1
On: Wed, 21 Sep 2011 18:25:51
material behaviour. These include models based on the
functional representation of one or more observed stress–strain
and volumetric response curves. Hyperbolic representation has
(σ1 σ3)u
been used for static and quasi-static behaviour, and can
provide a satisfactory prediction of load–displacement
Deviatoric stress, σ1 σ3
5
behaviour under monotonic loading. (σ1 σ3)f A
Hyperbola
In this paper, the hyperbolic model was employed for 1
simulating the stress–strain response of soils, which is needed Ei
for the analysis and design of important geotechnical projects. Actual
The theoretical background of such a model is presented. Three
soil types were used in this investigation, and they have been
thoroughly characterised. Parameters for the hyperbolic model
were determined experimentally for these local soils from the
Eastern Province of Saudi Arabia.
Axial strain, ε
2. BACKGROUND
A brief overview of the hyperbolic model for soil is presented. Fig. 1. Comparison of typical stress–strain curve with
In this work, emphasis was placed on the parameters of the hyperbola
model and their determination, from laboratory
characterisation of representative soil samples. The hyperbolic
model is a variable-parameter model used to simulate the non-
linear stress–strain response of soils, and is often used with
numerical solution techniques such as the finite-element
4,6
method. The use of hyperbolae was proposed to represent the å
7 ó1 ó3 ¼ 1 å
stress–strain behaviour of cohesive soils and cohesionless 1 þ
8 9
soils. The model was modified by Hansen. Since the stress– Ei (ó1 ó3 )u
strain behaviour of soil depends on confining stress, Duncan
5
and Chang incorporated that effect and used the hyperbola in
conjunction with the relationship between the initial modulus where å is the axial strain, Ei is the initial tangent modulus,
10
and confining pressure proposed by Janbu. (ó1 ó3 )u is the ultimate deviatoric stress at large strain, and
ó1 and ó3 are the major and minor principal stresses
5 11
Duncan and Chang and Duncan et al. presented a non- respectively.
linear stress–strain model based on a hyperbolic type of
relation between the deviator stress (ó1 ó3 ) and the axial The hyperbola is considered valid up to the actual soil failure
strain (å), which is determined from triaxial test results. This (point A, Fig 1). Thus the ultimate deviatoric stress is defined in
model is defined by a variable Young’s modulus and a variable terms of the actual failure deviatoric stress, (ó1 ó3 )f , by the
bulk modulus. Young’s modulus increases with increasing failure ratio, Rf , as
confining stress and decreases with increasing shear stress. The
bulk modulus also increases with confining pressure, and is (ó1 ó3 )f
related to the power of the confining stress. The model uses 2 Rf ¼
(ó1 ó3 )u
isotropic linear-elastic stress–strain relationships but with the
elastic parameters varied according to the stress state. A
tangent, rather than secant, formulation is used for Young’s
modulus, making this model particularly suitable for The parameters Ei and (ó1 ó3 ) u can be found by plotting the
incremental simulation. actual test data in a linearised hyperbolic form. The appropriate
straight line used to represent the transformed equation (1) is
12 11
Selig extended the above model of Duncan et al. by using
the same Young’s modulus formulation but with an alternative å 1 å
bulk modulus having a hyperbolic formulation—that is, a 3 ¼ þ
ó1 ó3 Ei (ó1 ó3 )u
tangent bulk modulus—which is a function of the mean normal
stress state. It was found to represent hydrostatic compression
better than the power law form proposed by Duncan and
5 10
Chang. The hyperbolic form was also found capable of According to Janbu, the initial tangent modulus is assumed
representing uniaxial (unconfined) and triaxial compression. to increase with the confining pressure, ó3 , as follows:
106 Geotechnical Engineering 156 Issue 2 Hyperbolic model of soils Al-Shayea et al.
Delivered by ICEVirtualLibrary.com to:
IP: 148.212.219.1
On: Wed, 21 Sep 2011 18:25:51
plot of Ei =Pa against ó3 =Pa . Because specimens in the present
Bi åvol
study were drained and not saturated, the stress ó3 can be 9 óm ¼
1 (åvol =åu )
taken as total or effective.
The actual envelope is often curved. Thus either the best-fit To determine the parameters Bi and åu , the hydrostatic test
straight line can be used, or ö may be varied with ó3 . The 13
data are plotted in a linearised hyperbolic form. Yang and
latter case was used in this investigation, and ö is represented 14
Lin have shown that the hyperbolic formulation for the
by tangent bulk modulus, equation (10), represents soil behaviour
in a hydrostatic compression test better than the formulation
11
proposed by Duncan et al.
6 ö ¼ öo ˜ö log10 (ó3 =Pa )
3. GEOLOGY OF THE AREA
The geology of the Arabian Peninsula broadly consists of the
where öo is the value of ö for ó3 ¼ Pa , and ˜ö is the Arabian shield in the west and the Arabian shelf in the east.
reduction in ö for a tenfold increase in ó3 . The Arabian shelf comprises a sedimentary succession of
Cambrian to Pliocene layers covering the eastern and northern
The tangent Young’s modulus for any stress state may be part of the peninsula. In eastern Saudi Arabia, the strata dip
determined by differentiating equation (1), and then using gently east and north-east, reflecting the buried basement
equations (2), (4) and (5). The resulting equation for the tangent configuration. The surface rocks of the Eastern Province of
modulus is Saudi Arabia include formations of consolidated sediments
ranging from the Palaeocene to the Middle Eocene and the
15
Miocene to the Pliocene. Unconsolidated materials include
Rf (1 sin ö)(ó1 ó3 ) 2
7 Et ¼ 1 KPa (ó3 =Pa )n various sediments of Quaternary age, comprising wind-blown
2C cos ö þ 2ó3 sin ö sand dunes, beach sand and gravel, basin deposits of gravel
16, 17
and silt, sabkha sediments, shale, marl and claystone. The
generalised geological map of the Eastern Province region is
where ö is as expressed in equation (6). shown in Fig. 2. Recent studies have focused on the
geotechnical properties of various geological deposits in the
18
2.2. Bulk modulus area. Abduljauwad and Al-Amoudi studied the behaviour of
19
The bulk modulus, B, is defined as saline sabkha soils. Azam et al. investigated the expansive
characteristics of gypsiferous/anhydritic soil formations.
20
Abduljauwad et al. performed laboratory and field
˜óm measurements to study the response of structures to the heave
8 B¼
˜åvol 21
of expansive clay. Al-Shayea presented a case study of the
inherent heterogeneity of sediments in the area.
where ˜óm is the change in mean stress, and ˜åvol is the 4. SITE SELECTION FOR SOIL MATERIALS
change in volumetric strain. Because this research was supported by the pipeline industry,
sampling sites were selected in areas where a pipeline network
11
Duncan et al. proposed a formulation for B based on data encountered a specific soil type in the Eastern Province of
from triaxial tests. An alternative method for obtaining the Saudi Arabia. Emphasis was placed on the surface geology of
bulk modulus is from a hydrostatic (isotropic) compression test. the area to determine the types of natural soil associated with
In this test, the soil specimen is compressed under an local pipeline networks, by combining geological maps with
increasing confining pressure applied equally in all directions. pipeline network maps. Three different types of soil were
According to equation (8), the tangent bulk modulus, Bt , is the considered: sand, marl and sabkha. The geological
12
slope of the hydrostatic stress–strain curve. Selig observed distributions of these soils in the area have been studied by
22
that the curve relating hydrostatic confining pressure, óm , and different investigators. Ahmad presented a map showing
23
the volumetric strain, åvol , can be reasonably represented by locations of marl, Al-Ayedi presented a similar map for
24
the hyperbolic equation sabkha, and Al-Gunaiyan presented a similar map for the
Geotechnical Engineering 156 Issue 2 Hyperbolic model of soils Al-Shayea et al. 107
Delivered by ICEVirtualLibrary.com to:
IP: 148.212.219.1
On: Wed, 21 Sep 2011 18:25:51
20 and 30 cm below the
Eolin sand Dam formation ground surface. Extensive
(including sabkha deposits)
laboratory testing took place
Dammam formation Hadrukh formation
in order to characterise all
three different soil materials.
Neogene formation Umm Er Radhuma formation
5. EXPERIMENTAL
Hofuf formation Rus formation
PROGRAMME
50°00E
51°00E
48°00E
49°00E
different soil types. These
28° include the following:
Saffaniyah
108 Geotechnical Engineering 156 Issue 2 Hyperbolic model of soils Al-Shayea et al.
Delivered by ICEVirtualLibrary.com to:
IP: 148.212.219.1
On: Wed, 21 Sep 2011 18:25:51
using a nuclear gauge. The marl was tested at two different dry vertical stresses applied were 100, 200, 300 and 1000 kPa.
densities of 1·64 and 1·87 g=cm3 , corresponding to relative Measurements included vertical displacement, horizontal
compactions (RC) of 81·4% and 92·8% respectively. For the displacement and horizontal pressure. Tests were performed
marl, these conditions are labelled as low density and high according to ASTM D 3080. Tests were made at the very slow
density respectively. The marl samples were prepared with a rate of loading of 0·048 mm=min. For the marl and sabkha, the
water content equal to the optimum moisture content. The conventional apparatus was modified to allow for a large
sabkha soil was tested at two different dry densities of 1·64 and horizontal displacement, up to 26 mm instead of only 6 mm for
1·86 g=cm3 , corresponding to relative compactions of 85·2% the standard set-up.
and 96·8% respectively. For the sabkha, these conditions are
labelled as low density and high density respectively. As the
sabkha samples were collected from a location below the 6. RESULTS AND DISCUSSION
groundwater table, they were prepared with a water content
equal to the optimum moisture content corresponding to the 6.1. Physical properties
respective density from the compaction curve. Table 1 provides Particle size distribution (PSD) curves for all samples are shown
a summary of the properties of the tested soils. in Fig. 3. The coefficients of uniformity, C u , for the sand and
marl are 1·744 and 3·285 respectively. The coefficients of
The three soil types at the densities stated above were subjected curvature, C c , for the sand and marl are 0·924 and 0·992
to the following tests. respectively. The sand was classified as SP (poorly graded sand)
according to the Unified Soil Classification System, whereas the
5·2·1. Triaxial tests. A set of at least four unconsolidated marl was classified as SP (gravelly sand with fines). The fine
drained (UD) triaxial compression tests were conducted for fraction of the marl was found to be non-plastic. For the
each soil type at loose (low density) and dense (high density) sabkha, the 15% coarser than sieve no. 200 was visually
conditions. These tests were done at confining pressures of 100, inspected and found to contain fine sand and remains of
200, 300 and 1000 kPa. Measurements included axial strain, marine organisms. The clay fraction (, 2 ìm) was about 50%.
deviatoric stress and volume change. The tests were performed
according to ASTM Standards D 2850 and D 4767. They were
performed at the very slow rate of loading of 0·0466 mm=min.
Samples were loaded on the testing apparatus to a very high
Sieve No.
strain level, up to about 50%. Each test took about one day.
# 200
# 140
# 100
# 60
# 40
# 30
# 20
# 10
#4
5·2·2. Hydrostatic (isotropic) compression tests. One test was 100
conducted for each soil type at loose (low density) and dense
(high density) conditions at confining pressures of 25, 50, 100, 80
200, 400 and 1200 kPa. The volume change and confining
Percentage passing
tests was calibrated in order to indicate the cell expansion Sabkha (sieve)
Sabkha (hydrometer)
following the application of pressure. A solid steel cylinder 0
0.00 0.00 0.01 0.10 1.00 10.00
with a volume equal to that of the soil sample was used in this Grain size: mm
calibration.
Fig. 3. Particle size distribution curve for the sand, marl and
5·2·3. Direct shear test. A set of four direct shear tests were sabkha tested
conducted for each soil type at loose and dense conditions. The
Geotechnical Engineering 156 Issue 2 Hyperbolic model of soils Al-Shayea et al. 109
Delivered by ICEVirtualLibrary.com to:
IP: 148.212.219.1
On: Wed, 21 Sep 2011 18:25:51
found to be 25·6%. The liquid limit (LL) and the plastic limit å . 15%, the value of (ó1 ó3 )f is taken to be the value of the
(PL) of the sabkha soil were found to be 28·3% and 22·9% deviatoric stress at å ¼ 15%, (ó1 ó3 )å¼15% , as suggested by
27
respectively. The plasticity index (PI) was 5·4. This sabkha can Bowles.
be classified as CL-ML (silty clay).
Figure 5 presents the results of triaxial tests on the low-density
The maximum and minimum densities (rmax and rmin ) obtained sand. Similar results were obtained for the high-density sand.
in the laboratory for the sand were found to be 1·787 and The dense sand fails at a much lower strain and at a much
1·515 g=cm3 respectively. The in-situ density (rfield ), as higher deviatoric stress than loose sand. Also, the dense sand
measured by the nuclear gauge, was found to be 1·641, 1·667, exhibits a smaller reduction in volume initially, and greater
1·690 and 1·712 g=cm3 at depths of 5, 10, 20 and 30 cm expansion at higher strain levels, compared with the loose
respectively. The relative densities, Dr , of the loose and dense sand. By increasing the confining pressure, the deviatoric stress
sands were 30% and 80%, respectively. Therefore the sand increases and the volumetric strain decreases. This behaviour is
28
samples were prepared for loose and dense conditions at dry similar to that reported in the literature.
densities of 1·587 and 1·725 g=cm3 respectively. These limits
cover the range of in-situ results of the densities found in the For the low-density sand (Dr ¼ 30%), and by using the Mohr–
field. Coulomb failure criterion, the Mohr circles and the Coulomb
failure envelope were drawn in Fig. 6. For this sand, the
The compaction curve of the marl from the modified Proctor cohesion, C, was zero, and the angle of internal friction, öo ,
test (ASTM Standard D 1557) is shown in Fig. 4, which gives a was taken as the slope of the tangent of the first three circles,
maximum dry density (rd,max ) of 2·022 g=cm3 and an optimum and found to be 38·308. The reduction in the angle of internal
moisture content, wopt , of 9·1%. Fig. 4 also shows the friction, ˜ö, was taken to be the difference between the slope
compaction curves of the sabkha from the modified Proctor of the tangents of Mohr circles at confining pressures, ó3 , of
test, which gave rd,max as 1·925 g=cm3 and wopt as 13·55%. The 100 and 1000 kPa, and found to be 3·168.
optimum moisture contents for marl at low density
(rd ¼ 1:64 g=cm3 ) and at high density (rd ¼ 1:86 g=cm3 ) are The parameters of the hyperbolic model were obtained from the
23·25% and 17·8% respectively. triaxial data, according to the explanation given in section 2·1.
Triaxial tests at ó3 ¼ 100, 200 and 300 kPa were considered in
6.2. Results of triaxial tests obtaining these parameters. The first step involves finding the
initial tangent modulus, Ei , and the failure ratio, Rf , by
6·2·1. Soil parameters. This section presents the soil plotting å=(ó1 ó3 ) against å, as defined earlier. Plots of
parameters obtained from triaxial tests performed on the å=(ó1 ó3 ) against å are presented in Fig. 7 for the low-
various types of soil (sand, marl and sabkha) for both loose and density sand. Fig. 7 represents a set of straight lines, each of
dense conditions. The results of the triaxial tests are presented which can be expressed by the equation of a straight line given
in terms of deviatoric stress, ó1 ó3 , against axial strain, å, in equation (3). The slope of each line gives 1=(ó1 ó3 )u and
and volumetric strain against axial strain. The deviatoric stress
at failure, (ó1 ó3 )f , is taken as the value of deviator stress at
the peak, provided that the peak occurs at a strain (å) less than
15%. In the cases where the peak is not evident, or it occurs at 3000
2500
1.95
1000
Dry density: g/cm3
1.90 500
1.85 0
2.6
0 5 10 15
1.80 2.0
Volumetric strain: %
1.4
1.75
0.8
0.2
1.70
0.4
1.65 1.0
0 5 10 15 20 25 0 5 10 15
Moisture content: % Axial strain: %
Fig. 4. Compaction curves for the marl and sabkha Fig. 5. Results of triaxial tests on the sand at low density
110 Geotechnical Engineering 156 Issue 2 Hyperbolic model of soils Al-Shayea et al.
Delivered by ICEVirtualLibrary.com to:
IP: 148.212.219.1
On: Wed, 21 Sep 2011 18:25:51
log(ó3 =Pa ), where Pa is the atmospheric pressure (Fig. 8).
4000
Equation (4) can be written as
φo
3500
n
φo ∆φ Ei ó3
11 ¼K
3000 C 0 kPa Pa Pa
φo 38.30°
∆φ 3.316°
Shear stress: kPa
2500
or
2000
Ei ó3
1500 12 log10 ¼ log10 (K) þ n log10
Pa Pa
1000
0.00020
Confining pressure The results of triaxial tests on the marl at low and high
100 kPa densities are presented in Fig. 10. For the low-density marl, no
200 kPa
300 kPa increase in volumetric strain was experienced. For the high-
0.00015
density marl under a confining pressure of ó3 ¼ 1000 kPa the
sample continued to experience a reduction in volume across
the entire strain range. Triaxial test results for the marl at low
ε/(σ1 σ3)
Geotechnical Engineering 156 Issue 2 Hyperbolic model of soils Al-Shayea et al. 111
Delivered by ICEVirtualLibrary.com to:
IP: 148.212.219.1
On: Wed, 21 Sep 2011 18:25:51
Confining Peak stress, Slope: Intercept: (ó1 ó3 )u ¼ Ei ¼ 1=intercept : Rf ¼ (ó1 ó3 )f
pressure: (ó1 ó3 )f : 3103 3106 1=slope 3105 kPa 3 slope
kPa kPa 3103 kPa
2
n 6·3·1. Soil parameters. Soil parameters were obtained from
hydrostatic (isotropic compression) tests performed on the
1000 various types of soil (sand, marl and sabkha) for both loose and
9
8
7
K
dense conditions. The results of the hydrostatic compression
6
5 tests for all three soil types are presented in Figs 12, 14, 15 and
4 16 in terms of confining pressure, óm , against volumetric
3 strain, åvol . The volumetric strain was corrected to account for
2
the expansion of the chamber upon applying the pressure. This
expansion was considered as part of the calibration of the cells
used for the hydrostatic compression tests.
100
5 6 7 8 9 2 3 4 5 6 7 8 9 2
1 10 Figure 12 presents the results of hydrostatic tests performed on
Confining pressure: σ3/Pa
the sand at low and high densities, in terms of hydrostatic
stress, óm , against volumetric strain, åvol . The loose sand
Fig. 8. Determination of model parameters K and n for sand at experienced more volumetric strain than the dense sand,
low density especially at low values of hydrostatic stress. As the test
progressed, the loose sand started to densify and the rate of
reduction in volume reduced. The plot of óm =åvol against óm
that the soil parameters (K, n and Rf ) were determined based on for the low-density sand is shown in Fig. 13. Equation (9) can
the results of triaxial tests at ó3 ¼ 100, 200 and 300 kPa only; be written as
the experimental results for ó3 ¼ 1000 kPa were not included
in the determination of these parameters. This was done to óm 1
adhere to the standard procedure, and to avoid contaminating 14 ¼ Bi þ óm
åvol åu
these parameters with non-linear effects at high confining
pressure. Similar results were obtained for the high-density
sand, as shown in Fig. 9(b). It is important to note that the
hyperbolic model is not valid beyond the peak of the stress– Equation (14) is the equation of a straight line whose intercept
strain diagram (point A, Fig. 1). This is because the hyperbola represents the initial bulk modulus, Bi , and whose slope
cannot capture the post-peak strain softening behaviour. In the represents the reciprocal of the ultimate volumetric strain at
case of soil materials exhibiting a sharp peak, such as dense large stress, 1=åu . For the loose sand, the values for Bi and åu
soils, the peak usually indicates failure. are 3836·95 kPa and 0·0251 respectively.
112 Geotechnical Engineering 156 Issue 2 Hyperbolic model of soils Al-Shayea et al.
Delivered by ICEVirtualLibrary.com to:
IP: 148.212.219.1
On: Wed, 21 Sep 2011 18:25:51
3000 2500
Back-prediction
Experimental curve
2500
2000
Back-prediction
Confining pressure
Experimental curve
2000 σ3 100 kPa
σ1 σ3: kPa
Confining pressure
σ3 300 kPa
σ3 100 kPa
1500 σ3 1000 kPa
σ3 200 kPa
σ3 300 kPa
σ3 1000 kPa 1000
1000
500
500
0 0
0 5 10 15 0 5 10 15 20 25
Axial strain: % Axial strain: %
(a) (a)
5000 4000
Back-prediction
4500 Experimental curve
3500
4000
3000
3500
σ1 σ3: kPa
σ3 100 kPa
σ3 100 kPa
σ3 200 kPa
2500 Back-prediction 2000 σ3 200 kPa
σ3 300 kPa
Experimental curve σ3 300 kPa
σ3 1000 kPa
σ3 1000 kPa
2000
1500
1500
1000
1000
500
500
0 0
0 2 4 6 8 10 0 5 10 15
Axial strain: % Axial strain: %
(b) (b)
Fig. 9. Back-prediction of behaviour of sand under triaxial Fig. 10. Back-prediction of behaviour of marl under triaxial
compression, at: (a) low density; (b) high density compression, at: (a) low density; (b) high density
sabkha at low and high densities, with water contents back-predicted óm –åvol curves for the low- and high-density
corresponding to the respective values of optimum water sand. Similar results were obtained for the marl at low and
content. The low-density sabkha experienced more volumetric high densities (Fig. 15), and for the sabkha (Fig. 16). The
strain reduction than the high-density sample because of the experimental and back-predicted curves are in very close
amount of water lost during consolidation. Water was observed agreement.
emerging from the sabkha soil samples during testing.
6.4. Results of direct shear tests
6·3·2. Back-prediction. The soil parameters obtained from The results of the direct shear tests for all soil types are
hydrostatic tests on the various types of soil under different presented in terms of shear stress and normal displacement
conditions (Table 4) were used to back-predict the behaviour of against horizontal displacement.
these soils under hydrostatic stress conditions. The predicted
behaviour for each soil type is compared with the actual curve Figure 17 presents the results of direct shear tests on the low-
obtained from the experimental tests. The hyperbolic form, density sand. Similar results were obtained and analysed for
given in equation (9), was used to back-predict the behaviour the high-density sand, the marl and the sabkha (Table 5). The
of soils during hydrostatic loading. shear stress and the dilation (positive normal displacement) of
dense sand are more, and the compression is less, than for
Figure 14 presents comparisons between the experimental and loose sand. The shear stress, ô, is taken as the value of ô at the
Geotechnical Engineering 156 Issue 2 Hyperbolic model of soils Al-Shayea et al. 113
Delivered by ICEVirtualLibrary.com to:
IP: 148.212.219.1
On: Wed, 21 Sep 2011 18:25:51
3000 50000
Confining pressure
σ3 100 kPa
σ3 200 kPa
2500 σ3 300 kPa 40000
σ3 1000 kPa
2000
30000
σ1 σ3: kPa
σm/εvol
Back-prediction
1500 Experimental curve
1/εu
20000
1
1000
10000
500
Bi
0
0 0 200 400 600 800 1000
0 5 10 15 20 25 σm: kPa
Axial strain: %
The results of direct shear tests on the marl at low and high
densities showed an extremely large horizontal displacement,
which exceeds the 6 mm limit for conventional results. The
1200
Low density low-density marl did not experience any dilation, and its shear
High density
stress did not reach a peak value.
1000
The results of direct shear tests were obtained for the sabkha
soil at low and high densities, with water contents
800 corresponding to the respective values of the optimum water
contents.
σm: kPa
600
The low-density sabkha soil did not experience any dilation.
114 Geotechnical Engineering 156 Issue 2 Hyperbolic model of soils Al-Shayea et al.
Delivered by ICEVirtualLibrary.com to:
IP: 148.212.219.1
On: Wed, 21 Sep 2011 18:25:51
1200 1200
Experimental curve Experimental curve
Back-prediction
Back-prediction
1000 1000
800 800
σm: kPa
σm: kPa
600 600
400 400
200 200
0 0
0.00 0.01 0.01 0.02 0.02 0.03 0.000 0.020 0.040 0.060 0.080
εvol εvol
(a) (a)
1200 1200
Experimental curve Experimental curve
Back-prediction Back-prediction
1000 1000
800 800
σm: kPa
σm: kPa
600 600
400 400
200 200
0 0
0.000 0.005 0.010 0.015 0.020 0.000 0.010 0.020 0.030 0.040
εvol εvol
(b) (b)
Fig. 14. Back-prediction of behaviour of sand under Fig. 15. Back-prediction of behaviour of marl under
hydrostatic compression, at: (a) low density; (b) high density hydrostatic compression, at: (a) low density; (b) high density
For the results from triaxial compression tests with a sharp simulate the response of deep soil strata under high overburden
peak of the stress–strain diagram, it is important to note pressure using results of conventional triaxial tests.
that the hyperbolic model is not valid beyond the peak.
This is because the hyperbola cannot capture the post-peak For hydrostatic (isotropic) compression, the hyperbolic model
strain softening behaviour. The peak value is considered to based on tangent bulk modulus shows strong capabilities to
be the failure value, beyond which the soil sample is predict the behaviour of different soils.
generally not of interest in non-linear elastic problems. For
soil with a flat peak, the model can predict the soil
behaviour up to extremely large strains. This shows the 8. ACKNOWLEDGEMENTS
ability of this model to simulate excessive deformation of The authors acknowledge the support of King Fahd University
low-density soils. of Petroleum and Minerals in general, and the Civil
Engineering Department and the Research Institute in
At high confining pressure (1000 kPa), the triaxial results were particular, for providing computing, laboratory and editing
predicted by the hyperbolic model with parameters obtained at facilities. They would also like to acknowledge Saudi-ARAMCO
relatively low confining pressure (100–300 kPa). This is for supporting this research. Thanks also go to Mr Hasan
considered to be another strength of this model, which can Zakariya from the geotechnical laboratory.
Geotechnical Engineering 156 Issue 2 Hyperbolic model of soils Al-Shayea et al. 115
Delivered by ICEVirtualLibrary.com to:
IP: 148.212.219.1
On: Wed, 21 Sep 2011 18:25:51
1000 1000
Experimental curve Experimental curve
Back-prediction
Back-prediction
800 800
600 600
σm: kPa
σm: kPa
400 400
200 200
0 0
0.000 0.040 0.080 0.120 0.160 0.000 0.020 0.040 0.060
εvol εvol
(a) (b)
Fig. 16. Back-prediction of behaviour of sabkha under hydrostatic compression, at: (a) low density; (b) high density
Sand (low density) 3 836·95 0·0251 38·37 Sand (low density) 0·00 41·40 2·05
Sand (high density) 23 275·90 0·0337 232·76 Sand (high density) 0·00 52·04 5·64
Marl (low density) 6 243·87 0·0915 62·44 Marl (low density) 4·19 32·10 0·00
Marl (high density) 5 356·65 0·0388 53·57 Marl (high density) 43·69 38·58 4·12
Sabkha (low density) 620·84 0·1420 6·21 Sabkha (low density) 42·29 36·78 0·65
Sabkha (high density) 2 608·77 0·0618 26·09 Sabkha (high density) 53·84 42·52 4·71
Table 4. Soil parameters obtained from hydrostatic tests Table 5. Soil parameters obtained from direct shear tests
900
Normal pressure
800 100 kPa
200 kPa
700
300 kPa 1000
φ
1000 kPa
Shear stress: kPa
600
φ ∆φ
500
800
400
300
φ 39.35
Shear stress: kPa
200 600
100 φ 41.40
0
0 1 2 3 400
Normal displacement: mm
0.20
200
0.05
0
0.10 0 200 400 600 800 1000 1200
0 1 2 3 Normal stress: kPa
Horizontal displacement: mm
Fig. 18. Shear stress plotted against normal stress for sand at
Fig. 17. Results of direct shear tests on low-density sand low density
116 Geotechnical Engineering 156 Issue 2 Hyperbolic model of soils Al-Shayea et al.
Delivered by ICEVirtualLibrary.com to:
IP: 148.212.219.1
On: Wed, 21 Sep 2011 18:25:51
REFERENCES 14. LIN R.-S. D. Direct Determination of Bulk Modulus of
1. COON M. D. and EVANS R. J. Recoverable deformation of Partially Saturated Soils. Project Report ACP87–341P for
cohesionless soils. Journal of the Soil Mechanics and MS degree, University of Massachusetts, Amherst, MA, 1987.
Foundations Division, ASCE, 1971, 97, No. SM2, 375–391. 15. POWERS L. F., RAMIREZ L. F., REDMOND C. D. and ELBERG E. L.
2. COROTIS R. B., FARZIN M. H. and KRIZEK R. J. Non-linear Jr. Geology of the Arabian Peninsula: Sedimentary Geology
stress–strain formulation for soils. Journal of the of Saudi Arabia. Arabian American Oil Company/United
Geotechnical Engineering Division, ASCE, 1974, 10, No. States Geological Survey (ARAMCO-USGS), Washington,
GT9, 993–1008. 1963.
3. KO H.-Y. and MASSON R. M. Nonlinear characterization and 16. AL-SAYARI S. J. and ZÖTL J. G. Quaternary Period in Saudi
analysis of sand. Proceedings of the 2nd International Arabia. Springer-Verlag, Wien, 1978, vol. 1.
Conference on Numerical Methods in Geomechanics, 17. ABDULJAUWAD S. N. Swelling behavior of calcareous clays
Virginia, 1976, 294–305. from the Eastern Province of Saudi Arabia. Quarterly
4. DESAI C. S. and SIRIWARDANE H. J. Constitutive Laws for Journal of Engineering Geology, 1994, 27, 333–351.
Engineering Materials, with Emphasis on Geologic 18. ABDULJAUWAD S. N. and AL-AMOUDI O. S. B. Geotechnical
Materials. Prentice-Hall, Englewood Cliffs, NJ, 1984. behaviour of saline sabkha soils. Géotechnique, 1995, 45,
5. DUNCAN J. M. and CHANG C. Y. Nonlinear analysis of stress No. 3, 425–445.
and strain in soils. Journal of the Soil Mechanics and 19. AZAM S., AL-ABDULJAUWAD S. N., AL-SHAYEA N. A. and
Foundations Division, ASCE, 1970, 96, No. SM5, AL-AMOUDI O. S. B. Expansive characteristics of
1629–1653. gypsiferous/anhydritic soil formations. Engineering Geology
6. WONG K. S. and DUNCAN J. M. Hyperbolic Stress–Strain Journal, 1998, 51, No. 2, 89–107.
Parameters for Non-linear Finite Element Analyses of 20. ABDULJAUWAD S. N., AL-SULAIMANI G. J., BASUNBUL I. A. and
Stresses and Movements in Soil Masses. Department of AL-BURAIM I. Laboratory and field studies of response of
Civil Engineering, University of California, Berkeley, 1974, structures to heave of expansive clay. Géotechnique, 1998,
Report No. TE-74-3 to National Science Foundation. 48, No. 1, 103–121.
7. KONDNER R. L. Hyperbolic stress–strain response: cohesive 21. AL-SHAYEA N. A. Inherent heterogeneity of sediments in
soils. Journal of the Soil Mechanics and Foundations Dhahran, Saudi Arabia: a case study. Engineering Geology
Division, ASCE, 1963, 89, No. SM1, 115–143. Journal, 2000, 56, No. 3–4, 305–323.
8. KONDNER R. L. and ZELASKO J. S. A hyperbolic stress–strain 22. AHMAD H. Characterization and Stabilization of Eastern
formulation of sands. Proceedings of the 2nd Pan- Saudi Marls. MS thesis, Civil Engineering Department,
American Conference on Soil Mechanics and Foundation King Fahd University of Petroleum & Minerals, Saudi
Engineering, Sao Paulo, 1963, 1, 289–324. Arabia, 1995.
9. HANSEN J. B. Discussion of ‘Hyperbolic stress–strain 23. AL-AYEDI E. S. Chemical Stabilization of Al-Qurayyah
response: cohesive soils’ by R. L. Kondner. Journal of the Eastern Saudi Sabkha Soil. MEng report, Civil Engineering
Soil Mechanics and Foundations Division, ASCE, 1963, 89, Department, King Fahd University of Petroleum &
No. SM4, 241–242. Minerals, Saudi Arabia, 1996.
10. JANBU N. Soil compressibility as determined by oedometer 24. AL-GUNAIYAN K. A. Assessment of the Geotechnical
and triaxial test. Proceedings of the European Conference Properties of Sand–Marl Stabilized Mixtures for
on Soil Mechanics and Foundation Engineering, Wiesbaden, Constructional Purposes. MS thesis, Civil Engineering
1963, Vol. 1, pp. 19–25. Department, King Fahd University of Petroleum &
11. DUNCAN J. M., BYRNE P., WONG K. S. and MABRY P. Minerals, Saudi Arabia, 1999.
Strength, Stress–Strain and Bulk Modulus Parameters for 25. AMERICAN SOCIETY FOR TESTING AND MATERIALS. Annual Book
Finite Element Analysis of Stresses and Movements in Soil of American Society for Testing and Materials (ASTM)
Masses. University of California, College of Engineering, Standards. ASTM, Philadelphia, PA, 1993, vol. 04·08.
Berkeley, CA, 1980, Report No. UCB/GT/80-01. 26. LEE I. K., WHITE W. and INGLES O. G. Geotechnical
12. SELIG E. T. Soil parameters for design of buried pipelines. Engineering. Pitman, Boston, MA, 1983.
Proceedings of the Pipeline Infrastructure Conference, 27. BOWLES J. E. Engineering Properties of Soils and their
ASCE, Boston, 1988, pp. 99–116. Measurement. McGraw-Hill, New York, 1992, 4th edn.
13. YANG G.-R. Hyperbolic Young’s Modulus Parameters for 28. HOLTZ R. D. and KOVACS W. D. An Introduction to
Compacted Soils. Project Report ACP87-Pe for MS degree, Geotechnical Engineering. Prentice Hall, Englewood Cliffs,
University of Massachusetts, Amherst, MA, 1987. NJ, 1981.
Please email, fax or post your discussion contributions to the secretary by 1 October 2003: email: [email protected];
fax: þ44 (0)20 7799 1325; or post to Mary Henderson, Journals Department, Institution of Civil Engineers, 1–7 Great George Street,
London SW1P 3AA.
Geotechnical Engineering 156 Issue 2 Hyperbolic model of soils Al-Shayea et al. 117
Delivered by ICEVirtualLibrary.com to:
IP: 148.212.219.1
On: Wed, 21 Sep 2011 18:25:51