Dilatancy For Cohesionless Soils DAFALIAS PDF

Download as pdf or txt
Download as pdf or txt
You are on page 1of 12
At a glance
Powered by AI
The key takeaways are that dilatancy needs to be modelled as a state-dependent quantity rather than a unique function of stress ratio, and a constitutive framework is presented that treats dilatancy as state-dependent within the framework of critical state soil mechanics.

Dilatancy is the ratio of plastic volumetric to deviatoric strain increments, and describes the tendency of a soil to expand or contract during shear. It is important for modelling soil behavior because it is fundamentally related to stress ratio and affects the direction of plastic flow. Treating it as a unique function of stress ratio does not capture behavior over a range of densities.

The paper proposes to model dilatancy as a state-dependent quantity within the framework of critical state soil mechanics. Micromechanical analysis is used to motivate a simple macroscopic constitutive framework, and a rudimentary model is presented.

Li , X. S. & Dafalias, Y. F. (2000).

Gc!U/(!Chllicjl{(!

50, No.4. 449-460

Dilatancy for cohesionless soils


X . S. L1 * and Y. F. DAFALIASt

Dilatancy is often considered a unique function of the stress


ratio 1/ = q / pi, in terms of the triaxial stress variables q
and p'. With this assumption, the direction of plastic flow is
uniquely related to 1/, irrespective of the material internal
state. This obviously contradicts the facts. Consider two
specimens of the same sand, one is in a loose state and the
other in a dense state. Subjected to a loading from the same
1/, the loose specimen contracts and the dense one dilates.
These two distinctly different responses are associated with a
single 1/ but two different values of dilatancy, one positive
and the other negative. Treating the dilatancy as a unique
function of 1/ has developed into a major obstacle to unified
modelling of the response of a cohesionless material over a
full range of densities and stress levels (before particle
crushing). A theory is presented that treats the dilatancy as
a state-dependent quantity within the framework of critical
state soil mechanics. Micromechanical analysis is used to
justify and motivate a simple macroscopic constitutive framework. A rudimentary model is presented, and its simulative capability shown by comparison with experimental data
of the response of a sand under various initial state and
loading conditions.

KEYWORDS: constitutive relations; plasticity; sands.

INTRODUCTION

The concept of critical state (Roscoe et al. , \958) has been


successfully applied to modelling the behaviour of cohesive
soils. However, sand modelling has not always been formulated
within the critical state framework, and this is because the
behaviour of sand is somehow different from clay. Consider the
stress-strain response of a sand along a path of constant stress
ratio 17 = q/ pi, where pi = (01 + 20':.)/3 is the effective confining pressure and q = 0 I - 03 is the deviatoric stress in a
triaxial setting. First, unlike clay, sand does not possess a
unique relationship between the void ratio e and pi for a
particular '7. In fact, the density of a typical sand in the pressure
range before particle crushing cannot be altered considerably by
a constant t7 compression, either isotropic (17 = 0) or anisotropic
(17 #- 0). Secondly, when the '7 of a sand reaches its limiting
valuc M (the critical stress ratio) during plastic loading, it does
not necessarily follow that the sa nd is at a critical state. The
stress path can actually move along the '7 = M line, as for
example in an undrained dilative shear path up to ultimate
failure. These differences suggest that the well-established
framework for clay modelling should not be directly transplanted to sand without a careful examination.
One of the fundamental issues in modelling the stress- strain
behaviour of a soil is to correctly desc ribe its dilatancy d, the
ratio of plastic volumetric strain increment to plastic deviatoric
strain increment in the triaxial space : d = dE~ / ld E l~ l, wh ere
dE\" = dEl + 2dE3, dEq = 2(dfl - dE, )/3, and the superscript' p '
stands for ' plastic' (Roscoe & Burland, 1968; Nova & Wood,
1979; Wood, 1990; Wood el al., 1994; Vardoulakis & Sulem ,
1995). The second law of thermodynamics shows that '7 and d
are interrelated at a very fundamental level (Vardoulaki s &

La dilatance est souvent consideree comme une fonction


unique du rapport d'effort 1] = q/ pi, en termes de variantes
de contrainte triaxiales q et p'. Avec cette hypothese, la
direction de l'ecoulement plastique est Iiee uniquement it '1),
quel que soit I'etat interne du materiau. Ceci contredit les
faits de maniere evidente. Prenons deux specimens d'un
meme sable, Pun meuble et I'autre dense. Soumis it un
chargement du meme '1), Ie specimen meuble se contracte et
Ie specimen dense se dilate. Ces deux reponses bien differentes sont associees it un seul '1) mais it deux valeurs de
dilatance differentes, une positive et I'autre negative. Le fait
de traiter la dilatance comme fonction unique de '1) est
devenu un obstacle majeur it la creation de modeles unifies
de la reponse d'un materiau non cohesif sur toute une
gamme de den sites et de niveaux de contraintes (avant Ie
broyage des particules). Cet expose presente une theorie qui
traite la dilatance comme une quantite dependante de Petat
dans Ie cadre de travail de la mecanique de sol it Petat
critique. Nous utilisons une analyse micromecanique pour
justifier et motiver un cadre de travail constitutif macroscopique simple. Nous presentons ensuite un modele rudimentaire et nous montrons sa faculte simulative par une
comparaison avec les donnees experimentales de la reponse
d'un sable dans divers etats initiaux et sous diverses conditions de charge.

Sulem , 1995). Taylor (1948) proposed 17 + d = constant, based


on the hypothesis that a constant ' effective' friction coefficient
exists. Rowe (1962) showed, based on the theory of least rate of
internal work, that d could be expressed as a function of the
stress ratio and the true angle of friction between the mineral
surfaces of the particles. Although the particular forms proposed
by Taylor & Rowe were different, both of them suggested that
the dilatancy d was a unique function of the stress ratio I]:
d

= d(17 , C)

(I)

where C is a se t of intrinsic material constants.


Equation (I) worked quite satisfactorily for cohesive soils.
For example, d in two versions of Camclay models (Roscoe &
Schofield , 1963; Roscoe & Burland, 1968) were given by
d = M - '1 and d = (M 2 - 17 2 )/ 217, respectively, where M is
the critical stress ratio, an intrinsic material constant. In agreement with the concept of critical state, these models make sure
that the soil yielding at '1 = M is coincident with d = 0; that
is, the material being modelled reaches its ultimate failure
whenever a plastic deformation takes place at '1 = M.
In contrast, it was soon found, based on experimental
evidence, that the applicability of equation (I) to granular soils
depends on the density. Observing the divergence between the
proposed theory and the test results, Rowe (1962) pointed out
that a variable depending on the sample density and the stress
history should be added to the stress - dilatancy relationship that
he had derived earlier. Rowe attributed the divergence to a
rearrangement of particle packing, a fact that was ignored when
his stress - dilatancy relationship was derived .
Rowe 's \Nork was followed in many later inves tigations (e.g.
Nova & Wood, 1979; Pastor el al. , 1990; Wood, 1990; Jefferies,
1993 ; Wood el al. , 1994) on sand modelling, in which, however,
the dependence of d on the material internal state was considered insignificant and thus dropp ed. Thi s simplification leads to
the cOl11l11on practice that treats a sa nd with different initial
den sities as different material s and results in multiple sets or

Manuscript received 18 October Il)l)l); revised manuscript accepted 28


January 2000
Di scussion on thi s paper closes 26 November 2000.
* The Hong Kong University of Science and Technology.
"I" University of California at Davis.

449

450

Ll AND DAFALlAS

parameters for a single sand and thus does not have a good
control over changes in the material state during loading.
Attempts have been made in recent years to tackle this issue
from the perspective of dilatancy. With the concept of critical
state as basis, Been & Jefferies (1985) introduced a scalar
quantity Vi called the state parameter, which measures the
difference between the current and critical void ratios at the
same p' . Kabilamany & Ishihara (1990) provided experimental
evidence showing that d + 'q increases as shear deformation
increases. Manzari & Dafalias (1997) presented a sand model in
which a linear dependence of the phase transformation, or
dilatancy stress ratio (the stress ratio at which the response
changes from contractive to dilative) on Vi was introduced. Li
(1997) investigated the response of sand at the ultimate stress
ratio and explicitly pointed out that the dilatancy d is not
related only to the stress ratio but is also a function of plastic
volumetric strain. More recently, Wan & Guo (1998) proposed a
model with its dilatancy modified from Rowe's stress-dilatancy
equation . The modified dilatancy equation includes the density
dependence with the critical void ratio as a reference.
Cubrinovski & Ishihara (1998) also showed a dilatancy relationship that depends on the material state represented by cumulative plastic shear strain. Li et 01 . (1999) introduced a statedependent dilatancy .into an existing hypoplasticity sand model
(Wang et al., 1990), resulting in a successful simulation of the
responses of Toyoura sand to both drained and undrained
triaxial loading over a wide range of densities and pressures.
In the present paper a number of issues on this subject are
discussed, starting from some microscopic analytical considerations and ending with the presentation of a simple macroscopic
constitutive framework and modelling, the simulative capability
of which is shown by comparison with experimental data of the
response of a sand under various initial state and loading
conditions.

MICROSCOPIC OBSERVATION ON DILATANCY

Consider a pack of uniform rigid rods, packing A, as shown


in Fig. I. Rowe showed that

0;

tan(fJ

+ (3)

particles; and f3 is the deviation of the tangent at the contact


points from the major principal direction. The quantity
D = I + d V/ VEl is clearly a form of dilatancy measure.
As shown in Appendi x I, resulting from the equilibrium
condition at rod contacts, the stress ratio R = 0; / 02 is uniquely
related to f3 as follows:
R

0;
= -02 = tan(q')". . + (3) tanf3

(3)

and resulting from kinematical compatibility condition, the void


ratio
e

= 8 sin f3 cos f3 _ 1

(4)

:rr

By combining equations (2)-(4) with f3 as an implicit variable,


a unique relation between Rand D is established, with e or f3
as an implicit dependent variable. That is, Rand D are uniquely
related, but the value of R (or D) depends on the volume of the
packing. This dependence is due to the equilibrium and kinematic constraints imposed by the given packing.
Now consider a different packing B, as shown in Fig. 2. For
this packing equation (2) is still valid but equations (3) and (4)
are not. As shown in Appendix I, for this packing

0;
2sinf3
--,
= tan(fJ + (3) I + 2 cos f3
02

(5)

and

e=

8 sin f3( I + 2 cos (3)

(6)

3:rr

It shows again that there exists a unique relationship between R


and D, with e or (3 as an implicit variable. However, the
relationship for this packing is different from that for the
packing A.
Rowe applied equation (2) to random mass of irregular
particles based on the hypothesis that the rate of internal work
done is a minimum. This hypothesis yielded f3 = 45 - fJ/2.
Thus,

(2)

(7)

where (J I and 02 are the major and minor principal stresses,


respectively; EI is the strain in 0; direction (tak ing compression
as positive for both stress and strain); V is the volume of the
pack ; p is the angle of friction between the surfaces of the

Equation (7) is the well-known stress - dilatancy equation. It can


be see n that the minimi zation procedure makes the stress ratio
R uniquely related to D and independent of the packing of the
particles and the volume of the mass, in contradiction to the

02(\

+ dV / VEt}

tanf3

i. \

1\ =60

L~. --II~

L, --II~

L
I
L .

Fig. I. Regular packing A of a uniform rod

L,

DILATANCY FOR COHESIONLESS SOILS

451

I.. .

L?

L2

.Lt.

L2

Lz

L1

L1

Fig. 2. Regular packing B of a uniform rod

exact analytical conclusion reached by the sets of equations


(2)-(4) and (2), (5), and (6), for the two examples of different
packing arrangements.
Figure 3 shows the relationships between (a; - aD/
(a; +02) and -dV/Vfl (a form of plots similar to 1] versus d
plot in the triaxial setting) for the packings A and B, and the
stress- dilatancy equation (equation (7) , respectively. Fig. 3
indicates that the relationships between the stress ratio and the
dilatancy depend on microscopic constraints. At a given stress
ratio, a particular packing is associated with a particular void
ratio, reflecting the internal microscopic constraints. Therefore,
the dilatancy depends not only on the stress ratio, but also on
the void ratio.
This density dependence is not reflected in equation (7), this
being the result of an unconstrained minimization of the rate of
internal work. In the derivation, the rate of work was obtained
from microscopic observations on regular packings of particles
sliding in a given direction, but the rate minimization was done
by zeroing its derivative with respect to the sliding direction.
This approach implicitly treated a particulate system as a
continuum without considering the static and kinematical constraints at the particle contacts. This treatment captured the
main feature of dilatancy and led to a unique relationship
between the dilatancy and the stress ratio . However, as it does
not take into account the microconstraints. which vary as the

10 r - -- - - -- - -- - - - -- -- - - - . -- - - - ,

O~;

............

+.. 0'6

~, 0-4
't;

--

02

F'ael--in9 A

C' not specHI

- - Packing B
<!

0:

'cl
(1 ,261

Equation Ci!
2

dVi V~;

Fig. 3. Deviation from the stress-dilatancy equation due to constraints at contacts

material state changes, the theory also shows notable deviations


from experimental observations. It is commonly observed that
at low deviatoric strains both dense and loose samples show
contractive behaviour and the dilatancy at that stage is not so
related to the material state . However, as shearing increases, the
deviation between the dilatancy of loose and dense specimens
becomes increasingly pronounced. This phenomenon can also
be seen in Fig. 3. As any microscopic constraints resulting in a
deviation from the hypothetical sliding direction f3 = 45 cjJfJ/2 will increase the rate of internal work, these constraints
tend to increase d. In Fig. 3, at the same (a; - 02)/(0; + aD,
the stress-dilatancy equation (equation (7)) yields the lowest
value of -dV / VEl'

PROBLEMS WITH UNIQUE RELATIONSHIP BETWE EN d AND 1/

Although the dependence of dilatancy on the internal state of


the material was noted decades ago, it has been largely ignored
in subsequent developments. This fact is, however. not trivial,
as a major obstacle to unified modelling of sand behaviour
arises from this ignorance.
When subjected to shear, loose sand contracts and dense sand
dilates. According to critical state soil mechanics, a loose or
dense state is defined not only in terms of density but also of
the confining pressure. This is because such a definition is
relative to the critical state line in the e versus p' space. For a
given e, for example, the sand will behave like dense for a
sufficiently low and like loose for a sufficient ly high p'.
Furthermore, for a sand that initially is either in the loose or
dense state, there is an ultimate state of failure at which the
volumetric strain rate is zero. This ultimate state is the wellknown critical state (Roscoe el al. , 1958) characterized by a
unique combination of p', q and critical void ratio ec in a
triaxial setting .
Consider two specimens of the same sand. One is in a loose
state and the other in a dense state, accounting for both density
and pressure . Subjected to a shear loading increment from the
same '] , the loose specimen contracts and the dense specimen
dilates, as shown in Fig 4 (data from Verdugo & Ishihara
(\996 in terms of undrained stress path in q- p' space. These
two distinctly different responses are associated with a single rl
but two different values of dilatancy, one positive and the other
negative. However, if d were a unique function of 17. the
direction of plastic flow, and hence of the undrained stress path,
would be uniquely related to 17, irrespective of the material

452

Ll AND DAFALIAS
1GOO

'.000

r-------------------,,-----,
e = 083:3

Dr = 3'7 '9%

1400

1200

1500
0:;

0:;

2;: 1000
0-

~
0
ro;;

800

" "0" ,,,"1

2;:

')0/0//0/

'0'
ui

<fl
Q)

~ 1000
0

ro;;

Q)

<ll

600

...........
400

500
Dense state, d < 0

200

o .""
o

200

400

600

800

1000

1200

Effective mean normal stress . p ': kPa

1500
Effecti ve mean normal stre ss. p' : kPa

(a )

(b )

2000

Fig. 4. Variation in dilatancy with material state (data from Verdugo & Ishihara (1996 . Undrained response of a sand with (a) different
densities and (b) the same density but under different confining pressures

state. This assumption obviously contradicts the fact as described.


Now consider a sand in a dense state subjected to an
undrained shear. As shear proceeds, '7 passes a so-called 'phase
transformation state' at which '7 = Md and d = 0 (Ishihara et
at., 1975) and then approaches the critical state at which
'7 = M and d = O. If eq uation ( I) held true , Md wou ld be equal
to M , because d is equal to zero at both the phase transformation state and the critical state . As the critical stress ratio M is
considered an intrinsic material property, independent of the
in itial material state, the logical outcome M" = M from equation ( I) would render the phase transformation intrinsic too ,
resulting in a uniqu e phase transformation lin e for a particu lar
sa nd at which the response of the sa nd wou ld change from
contractive to dilative , irrespective of its density and stress
level. However, tests show that the phase transformation phenome non can be seen only w hen the material is in a dense
state, and M" is in general a variab le quantity not eq ual to M.
As sand becomes 'looser' , Md becomes hi gher. This is clearly
co rroborated by test data such as those show n in Fig. 5
(Verdugo & Ishihara, 1996), where MrI is identified by the dark
. circles on the q- pI path where the tangent is parallel to the q

4000 , - - - - - - - - - - - - - - - - - - - - - - --,

~ ~IOOO
Va ri eci pll ase
tidflsior lTIatiC! n

Ul

~ "O()(1
tn
L.

C>

axis. Eventually, when the sand becomes too loose, the phase
transformation phenomenon totally disappears. The assumption
that d is uniquely related to 1J again contradicts the observation .
Furthermore, undrained tests on dense sand often show that
the q- pI stress path eventually converges with a line of more
or less constant 1J = M towards an ultimate state (Figs 5 and
6). At the ultimate state, both the stresses and the plastic
volumetric strain stop changi ng, as shown in Fig. 6 (see also
Figs 10- 12). Since the stresses do not change, neither does the
elastic volumetric strain . By definition , this is a critical state
where dpl = dq = dE" = 0 while dEq =f. O. Along the approximately '7 = M path and before the critical state is reached, the
fact that Ee tends towards a constant value implies that the
dilatancy d tends towards a zero va lue . If d were a unique
function of 'I , however, along thi s path d wou ld be esse nti ally a
constant, which means that, as shear proceeds, pI would increase contin uou sly as a result of the undrained constraint of
zero total vo lum etric change, and the criti cal state wou ld never
be reached. One m ay argue that along this path the stress ratio
'I only approaches M asymptotica ll y, and correspondingly the
evo lu tion of d from a no n-zero value towards zero would be a
result of the tiny deviations in '7 frolll M . However, this
argum en t on ly facilitates a mathematical description that barely
make s equation ( I) not violate the concept of a critical state.
Cons id ering the un certaint ies involved in soil testing, one would
not be able to physically identify and quantify sllch tiny
deviations , if any, in a mean ingful manner. On the other hand,
wit h the hypothesis that d and '] are not uniquely re lated. such
an argume nt becomes unnecessary.
The above observations lead to the concl usion that a sa nd
model with its dilatancy fo ll owing equatio n (I) works well on ly
when the change in the material internal state is minor.

(ii

;;

2;

1000

GOO

1000
1500
200\)
2500
Efieclive mean narrn,]1 sir ' 5S, p ' kP,]

3000

fig. 5. Variation in the phase transformation stress ratio with


material state (data from Verdugo & Ishihara (1996

GENERAL EXPRESSION FOR STATE-DEPEND ENT DILATANCY


Based on th e aforementioned observat ions and accounting for
the critical state constitutive framework , one Illay propose a
general expression for the dilatancy:
d = dUI , e, Q, C )

(8)

where Q and C, as coll ective terms, denote internal state


variables oth er than the vo id ratio (! (e.g. the evolving tensor of

DILATANCY FOR CO HESION LESS SOILS


i ,1 -~

453

0)

dL P

Ultimate/critica l slate

.. .

(d = 0)

....

... . .. .. .. .......... - ... ... .... .. .. .... .. .. .. ...... .. ...... .. - .. _. _._. -... -. -- .._. _. _. - .:.;
-

"-;

Ultimate/critical
states

(d;i 0)

Phase tr8ns forrnation

(b)

(8)

Fig. 6. Illustration of the dilative shear on the failure surface: (a) stress path; (b) stress-strain response

anisotropy (Dafalias, 1986)) and intrinsic material constants,


respectively. Equation (8) expresses the dependence of d on the
state variables, which consist of the external variable '7 and the
internal variables e and Q. Hence, equation (8) defines a statedependent dilatancy. Strictly speaking, a dilatancy expressed by
equation (I) is also related to material state via 1]. However, as
discussed above, such a relationship is not unique because it is
not complete. The term 'state-dependent dilatancy ' introduced
here signifies and emphasizes the need to define state dependence on both '7 and e and Q. With this additional dependence ,
d is now uniquely related to an existing state, a combination of
the external stress state expressed via '1] , and th e internal
material state ex pressed via e and Q. A subtle point here is
that, although it may appear that no explicit dependence of d
o n p' is introduced in equation (8) , the dependence on e and Q
may in fac t introduce indirectly such p' depe ndence.
There are certain requirements to be satisfied in formulating
d within the framework of equation (8) . First, the dilatancy
must be zero at a critical state; that is, d = 0 when II = tv! and
e = e" (the vo id ratio at the c ritical state) si multaneou sly:
dUI

M, e

= ee, Q, C) = 0

(9)

PARTICULAR EXPRESSION OF STATE-DEPENDENT DILATANCY


To obtain d within the framework of equation (8) subjected

to the requirement of equation (9) for the critical state response ,


one needs to quanti fy the dependence on e and the variables Q.
As the state of a material depends not only on its density (void
ratio e) but also on p', the aforementioned quantification should
be able to describe adequately the physical conditions of a
material, including both its density and its confining pressure .
Attempts have been made to describe the state, on which d
depends, with a single scalar quantity, which of course implies
the assumption of isotropy, since otherwise the use of tensorvalued quantities is necessary. Bee n & Jefferi es (19 85) defin ed
a state parameter 1/' = e - ee, where e is the current void ratio
and ee is the critical void ratio on the critical state line in the
e-- p' plane corresponding to the current p', as shown in Fig. 7.
Here V, is a measure of how far the material state is from the
critical state in terms of density. Bolton (1986) proposed a
scalar parameter h , called the 'relative dilatancy inde x', that
also combined the influence of density and confining pressure.
Ishihara (1993) introduced a scalar quantity Is, called the 'state
index', that takes some characteristic states other than critical
state in the e- p' plane as references .

In other words, the condition '7 = M alone does not guarantee


= O. It is important to emphasize in relation to equation (9)
that when 17 and e attain the ir critical values M and ec it is not
necessary for Q to reach a corresponding critical value. It is
entirely possible to reach a critical state with different values of

ClllTen l sl<l te ;0

Q.
Secondly, it is possible for sand to hav e a so-called ' phase
transformation state' at which d = 0 but 17 i- M and e i- ee, as
discussed earlier. Analytically thi s means that the equation

Il' > 0

(contractive)
u

>

Criticl:Il stale jin '

( 10)
can be lI sed to specify th e combination of 'I , e , and Q that
defin es a phase transformation state . Co nverse ly, one may use
an ([ priori experimental knowledge of phase transformation
states to specify via equation (10), together with equation (9),
app ropri ate forms of the dilatancy function in equation (8).

(pip ):

Fig. 7. Critical state line and state parameter IfJ

Ll AND DAFALIAS

454

In the present study, 1/.' was chosen to be th e state va ri able


that, in conjunction with '7, affects d. For improving the fitting
with experimental data of certain sands, V.' is represented by
(II)

where er, Ac and are the material constants determining the


critical state line in the e- p' plane (Li & Wang, 1998), and Pu
is the atmospheric pressure for norm alization. The dependence
of don e occurs via its dependence on V, = e - ec( p') (observe
introduction of explicit p' dependence via ec), while any other
dependence on Q is suppressed. Hence , an equation of the
following form is proposed:

= d(rJ , l/J, C)

(12)

which, according to equation (9), must satisfy the condition


d(17 = M, 1/J = 0, C) = O.
To illustrate analytically the effect of dilatancy on the stressstrain relationship, consider undrained triaxial loading during
which dev = de~ + de~ = O. With deq = de~ + de~ and the elastic relations de~ = dp'IK , de~ = dql3G in terms of the elastic
bulk and shear moduli K and G, respectively, the condition
dev = 0 yields deq/dq = (l/3G)-(dp'/dq)IKd , recalling that
d = dee! de~. When the stress path reaches and moves along a
line of more or less constant stress ratio 17 = M (failure surface ;
see Fig. 6(a, one has dp' /dq = II M, and hence the above
relationship becomes (Li , 1997)
de q )
( dq

_~
3G

I
KMd

_ _

11 =114-

( 13)

Equation (13) portrays an analytical conclusion for a general


elastoplasticity class of constitutive setting, not related to a
specific model. This equation states that the deq/dq, while
17 = M, is controlled primarily by the dilatancy d = deC/de~,
which defines the direction of plastic strain increment in the
volumetric-deviatoric plane. Equation (13) also shows that
(de q /dq)'7=M is independent of plastic hardening within the
approximation implied by setting dp'/dq = II M .
Figure 6(b) illustrates a typical response described by equation (13 ) for medium-to-dense sand. The shear stress - strain
curve is characterized by a slope dq I df q which, for lower q
values, keeps increasing, but as q reaches higher values begins
to dec rease as the stress-strain curve bends over, eventually
leading to the critical state a t which dp = dq = dE" = 0 while
dEq =I O. A long the path to the critical state, while 17 is constant,
d changes te nding towards a zero value. Recalling that the state
parameter 1/' enters equation (12) for d, and that d is the main
variable, which according to equation (13) can be used to
describe the curve of Fig . 6(b), one may propose a form of
equation (12) for 17 = M:

d = d o((' 111 '/'

I)

( 14)

in which do and 111 are two po sitive modelling parameters.


Observe that d and 1/' have the same sign a nd d = 0 when
II' = 0, which sati sfies the relation ship between the dilatancy
and the internal material state at the critical state.
While equation (14) may describe the dilatancy for I} = M,
one needs to generalize it for 17 f M . In this inve stigation , th e
specific form of the lJI dependent dilatancy according to equ ation ( 12) subjected to y ielding (equation (14 w hen '7 = M is
obtained from a modification to the dilat ancy function in the
original Camclay model , as follo ws:
( 15)
It can be seen that the Camclay dilatancy d = M - 1} is a
special case of equation (15) (111 = 0 and d o = M). Note that at
I} = M equation (15) is reduced to equation (14). More importantly, at a critical state, VI = 0 and '7 = M s imultaneously, and
hcnce equation (15) yield s a zero dilatancy, satisfying th e
require me nt set by equa tion (9).

Applying now equation (10 ) to the particular forlll of d,


(equation ( 15)). one obtains the phase transformation stress ratio
17 = M" = Me ""1'. This yields the following interesting interpretation to the last member of equation (15) for d. It postulates
that the dilatancy d depends on the difference of the current
stress ratio 1} from a reference stress ratio Me 1111/', which is
simil ar to Rowe's stress - dilatancy theory but with the reference
stress ratio varying with 1/J instead of being fixed. Based on
equation ( 10) this reference stress ratio represents the variable
with 1/J phase transformation line. This is exactly the concept
described by Manzari & Dafalias (1997), who used the linear
dependence Md = M + m1/!. The same concept was recently
used by Li et al. (1999) to improve the performance of the
hypoplasticity model proposed by Wang et al. (1990). In other
words, one could have started with the hypothesis that
Md = Me 1111/', corroborated by data, such as those shown in Fig.
5 where the variation in the phase transformation line can be
clearly seen, and then define d by d = (dol M)(M d - rJ), according to the classical stress- dilatancy framework . It follows
now that 1/' < 0 (dense states) implies Md < M, 1/' > 0 (loose
states) implies Md> M, and l/J = 0 (phase transformation
states) implies Md = M.

A SIMPLE MODEL FOR TRIAXIAL COMPRESSION


Within the scope of this paper, it is assumed that plastic
deformation occurs whenever 17 exceeds its historic maximum
and a constant rJ path induces no plastic deformation. This is,
of course , only approximately true, but it is still a good
approximation in many cases since under nonnal levels of
confining pressures of interest a constant 17 path induces only a
relatively small plastic volume change in sands, before graincrushing levels of pressures are reached as corroborated experimentally by Poorooshasb et al . ( 1966, 1967). However, for a
fully fledged model where the plastic deformations under constant rJ are to be considered, additional mechanisms, such as a
pi controlling cap , can be added with ease (Wang et al., 1990).
Under the above assumption , the yield criterion can be
written as

r = q -lIP' = 0

(16)

By the theory of pl asticity Wafalia s, 1986). a loading index


L can be defined as

(Of ,

I
of )
L =-,dp + -::----dq
Kp up
()q

dq - '7 dp'

Kp

P'd17
Kp

(17)

where Kp is a plastic hardening modulus. With the dilatancy


d = d E~)/ d t~;, the plastic strain inc re ment can then be written as

d f~

{ dE~

} _

L{

1 } _ { p'
d

Therefore. for L

df q

_
c
- d Ep

( 18 )

> 0:

p _

+ d Eq

dl}1Kp

d p ' dlJ l Kp

-;z;:- (-3GI+Kp-1) dq - -Kp'7d p'

dq
p' elI7 _
3G +

( 19)

dp'

d
Kp

d e\'= dE \, + d E~ =- +d d f ~=-dq+

( I

d17)

~--

1\

Kp

dp '
(20)

Equ ation s (19) and (20) establish the relationship between the
stress and strain increments. They can be inverted by a straightforwa rd algebraic manipulation , and expressed in a matri x form
as:

DILATANCY FOR COHESIONLESS SOILS


{:;, } =

[(

30G

0)

h(L)
( 9C
Kp + 3C - K1}d 3KCd
(21 )

where h(L) is a Heaviside function with heLl = I for L > 0 and


h(L) = 0 otherwise . Note that the extension of the foregoing
relationship to account for reverse loading requires some additional mechanism such as a back-stress (Manzari & Dafalias,
1997), or a memory of the reversal stress ratio point (Wang el
01., 1990). These aspects are not addressed within the restricted
scope of this paper.
For the model to be completed, in addition to the dilatancy d,
which is defined in equation (15), the moduli C, K, and Kp
must also be defined. The elastic shear modulus G can be
expressed by the following empirical equation (Richart el 01.,
1970):

G = Go

(2.97 - e)2

I+ e

V P Pa

G _2_(1_+_v_)
3(1 - 2v)

automatically satisfied by equation (24). Condition (e) will be


met at a particular material state 1/1 < 0 (a dense state), for
which 1} = Mb = M e - lIll', depending on the parameter 11 . In
addition to the above conditions, it also follows from equation
(18) that dEel P' dry = d l Kp. As dEe/d17 = 0 at 17 = 0 is normally observed, it is necessary to have d I Kp = 0 at '/1 = O. By
combining equations ( 15) and (24), it can be seen that this
condition is also satisfied automatically, since Kp = 00 at
I} = O.
In equation (24), G serves as a normalizing factor of h. It
was found that a variable h with density fitted the experimental
data better. In the present investigation, the simple linear
dependence
(25)

was used, where hI and h2 are two material constants, and e is


the current void ratio.
As shown in equation (19), the shear stress-strain response
is controlled by Kp. Substitution of equation (24) into equation
(19) with a constant p' (a drained condition) and 11 = 0 for
simplicity, yields

(22)
dEq

where Go is a material constant, and e is used instead of initial


void ratio employed in Richart et af. (1970). Based on elasticity
theory, the elastic bulk modulus K is equal to

(3~ +

q
- =

where hand 11 are two positive model parameters; the state


parameter 1/1 is calculated from equation (II) . Equation (24) is
a modified version of the plastic modulus in a bounding surface
hypoplasticity model (Wang et aI., 1990). The modification is
intended to model the peak stress ratio response and softening
of dense sands, the lack of such a response for loose sands, and
the failure at a residual stress ratio M at the critical state for all
densities. Based on the last member in equation (24), it follows
that Kp depends on the difference of the current stress ratio 'I}
from a ' virtual' peak stress ratio MO = Me- llIl ' attainable at the
current state defined by 1/'. Such peak stress ratio is variable
with 1/' in a way that yields MO > M for 1/' < 0 (dense states),
MO < MfaI' 1/' > 0 (loose states), and Mb = M for l/J = 0
(critical states) . The idea of having a virtual peak stress ratio
varying with 1/' in order to address the issue of peak stress and
subsequent softening of dense sand in drained conditions was
proposed by Wood ef al. (1994). The idea was followed by
Manzari & Dafalias (1997), where a slightly modified version
of the Wood ef 01. (1994) linear relation Mb = M - m/' was
introduced instead of the present M h = Me - 111/' .
Observe that it follow s from equation (18) that Kp =
p' dll l df~ , which requires the Kp function (equation (24 to
satisfy the following conditions:
Kp = 00 at 17 = 0, because dE~ = 0 in response to a nonzero dl7 at J7 = 0 (the material IS assumed isotropic without
previous loading);
(h) Kp = 0 at the critical states (I) = M and 1/, = 0) because at
a critical state dl7 / dE~ = 0;
( e ) Kp = 0 at drained peak stresses M b , because at those peaks
d}}ldE~ = O.

(a)

For the hardening and softening responses before and after the
peaks , Kp is positive and negative as 17 < Mb and 17 > MO,
respectively. It can be seen that conditions (a) and (b) are

dq =

C~ + hG(q~ _ q) dq

(I I I

- - - - - I n ( 1 - r)
3G hG hGr .

Eq

(24)

;J

(26)

where qr = Mp' is the value of q at failure . Integration of both


sides of equation (26) yields

(23)

where v is the Poisson's ratio. In this model, 11 is considered as


a material constant independent of pressure and density.
For the plastic modulus Kp the following constitutive relation
is proposed:

455

)-1

(27)

where r = q I qr. Equation (27) can be converted into a normalized modulus reduction curve (secant shear modulus normalized
to its maximum value G max versus shear strain) with h as a
parameter. Fig. 8 shows a family of such curves together with
the curve based on the hyperbolic stress-strain relationship
ql Eq = 3G ma x /( I + EIEqr) , where G max and Eqr are two material
constants. It can be seen that the Kp function used here allows
more flexibility in calibrating the shear stress-strain response
than does using the hyperbolic stress-strain response.

CALIBRATION OF MODEL CONSTANTS

There are eleven material constants in the model, shown


according to their functions in separate columns in Table I. A
systematic procedure can be followed to calibrate all these
parameters, based on triaxial data, as follows.
The critical state constants consist of M, the critical state
stress ratio, and the parameters er , ..1. c , and'; of equation (II).
These four constants can be determined by directly fitting the

12 ,------,---

--,---

--,---,----,-----,.----,------,

11 :: 6

(5

~ (Hi

-+-- h :::3

~ 0 4 - --- h::: 1 2
::;

~ 02 - ... 11 =06

.......

h :~ 0 3

0 03

0 1

03

10

Fig. 8. Modulus reduction as a function of the hardening parameter

"

LI AND DAFALIAS

456
Table I. Model parameters calibntted for Toyoura sand

Critical state
parameters

Elastic
parameters
Gn = 125
v = 005

Hardening
parameters

Dilatancy
parameters

125

do

er = 0934
Xe = 00 19

III

= 088

hi

= 3.5

3 15

h2 = 305
n

= 11

,; = 07

test data for the critical stress ratio and the critical state line in
the e- p' plane.
The parameter 111 can be determined by equation (15) at a
phase transformation state, at which d = O. Hence,
I
Md
m=-In-

(28)

ljJd

where l/Jd and Md are the values of 1/) and 17 at the phase
transformation state, measured from drained or undrained test
results .
The parameter 11 can be determined by equation (24) at a
drained peak stress state, at which Kp = O. Hence,
I

n =-In l/) b
Mb

(29)

where l/Jb and Mb are the values of l/J and 17 at the drained peak
stress state, measured from test results.
Next, consider the drained tests. Ignoring the smal1 elastic
deformations,
dEy

~ dE~

dE~

dEq

d= do (e

1111/' _

!l..)
M

(30)

The parameter do can then be calibrated based on the Ev- Eq


curves.
By combining equations (19) and (24) for the drained tests,
with either the conventional test (dp' = dq/3) or the constant
p' test (dp' = 0), one has
dq ~ dq _ Kp
dEq ~ dE~ - 1 - aJ7
=

hGo{

(2.97 - e )2 '; p' Pa[(M /'ll- ell./'J}


(I

+ e)( 1- m71

dp'

Kd

2Go( 1 + v)

dq
dEq

I I)
(-+
3G

Kp -

-I

1] Kd

dq

~- -

dE~ -

K -17Kd

(33)

As all the model parameters have already been determined, the


simulated undrained q-Eq curves can be used against their
experimental counterparts to examine the quality of the calibration based on equation (33). If the fit is not satisfactory, one or
more parameters are to be fine tuned and the calibration can be
repeated until an optimal result is obtained. Note that, after the
undrained stress path approximately converges with 1] = M, the
Kp value is still not zero as long as ljJ i= 0 (equation (24 .
Hence, the second member of equation (33) does not yield
equation (13) until Kp becomes very small and is neglected.
This shows the approximate character of equation (13) in
reference to an actual model such as the one presented here.
It should be pointed out that the set of model parameters
calibrated is for one material over a wide range of densities and
pressures. There is no need to do the calibration again for the
same material when the initial state changes.

SIMULATION BY THE MODEL

(31 )

in which the parameter (J is either equal to 1/ 3 (for conventional tests) or to zero (for constant p' tests). As al l the material
constants in the brackets have been predetermined, the COl1lbined parameter hG o ca n be calibrated independently based on
the experimental q - ELI curves. It may be found during calibration that the quantity hGo varies with density. Fitting these
values of hGn into equation (25) yie lds the constants hi and h2
(aftcr Go has been determincd).
Now let us turn to undrained (constant volume) tests . For
dE" = 0, equation (20) yields :

~=l - !.J..=1 _ 3(1 - 2v) {hGLl[(M/lll - ell'/'J}

value of 2Go( I + 1')/ 3(1 - 2v) , and could be negative. A micromechanics study (Chang & Misra, 1990) has shown that the
Poisson 's ratio of an assembly of particles is predominantly
control1ed by the ratio of the shear stiffness to the normal
stiffness at particle contacts. The value of the Poisson's ratio of
the assembly could be much lower than that for the particle
material itself. If this stiffness ratio is high, v could be negative.
Even though a negative v is theoretically justifiable and affects
nothing but the volumetric strain at extremely low strain ievel
(pure elastic range), if it is encountered and disliked, as an
alternative one may pick a v value first and then calculate Go.
As this alternative approach does not guarantee the accuracy of
the elastic shear response, it should be used only when the
shear stiffness in the elastic range is unimportant or when
accurate values of G are unavailable. Once Go has been
determined, h, and therefore hI and h2, can finally be found
from equations (25) and (31).
Last, but not least, one can obtain the undrained deviatoric
stress-strain response by substituting dp' = K dE~ = - K dE~
= -Kd dE~ into equation (32) and accounting for dq = 3G dE~.
The relationship is as follows:

(32)

As al1 the material co nstants in the brackets have been predetermined, the combined parameter 2Go( I + v)/3( I - 2v) becomes the only means at this stage of adjusting the undrained
p' - q responses of the model. By matching these responses with
their experimental counterparts, the value of 2G o( I + v) /
3( I - 2v) can be determined.
Finally, one needs to separate the parameter Go from hand
v . If shear stiffness at small strains is important , Go should be
determined by independent small strain tests, such as resonant
column tests or bendcr element tests, through fitting the test
data into equation (22). However, if Go is high, the value of
Poisson 's ratio ]I is reduced based on the already calibrated

Verdugo & Ishihara ( 1996) presented a sequence of triaxial


test results on Toyoura sand whic h are particularly suited to
demonstrating the simulative capability of a critical-state-based
sand model, since some of the data provide a definite identification of the critical state line in (! - p' space, which is of cardinal
importance for the determination of 1/.'. The sand is described as
uniform fine sand consisting of subrounded to subangular
particles. The maximum and minimum void ratios are 0977
and 0597. respectively. Verdugo & Ishihara reported a total of
17 shear loading tests in their paper. The tests include both
drained and undrained triaxial compression tests. The density
ranged from e = 0735 (relative density Dr = 637%) to
e = 0996 (Dr ~ 0%). The initial confining pressure p ' for the
tests ranged from 100 kPa to 3000 kPa. This set of test results
covers comprehens ively the behaviour of the Toyoura sand
under monotonic triaxial compression loading conditions.
All 17 tests were simulated using the simple model described
earlier with the unified set of model parameters listed in Table
I. Figs 9- 14 show the experimental results for all the 17 tests,
as well as the results of the simulations obtained with the
model. It can be seen that the model simulations broadly match
the experimental results, indicating the effectiveness of the
critical-state framework in conj unction with the state-dependent
dilatancy.
To compare with others, the discrepancies between the model
simulation and the test results for drained responses at very low
densities (initial void ratios eo = 096 and eo = 0996; or relative densities Dr = 45% and Dr ~ 0%) are more notable. A
simple modification of equatio n (25) by replacing the void ratio

457

DILATANCY FOR COHESIONLESS SOILS

2500 r---------------------------------------------.

OY5 r----------------------------------------------,

- - Mod ":! simulation

0-90

.......... Tes t resulls

2000

ro

2z
'C7

u;

~ 085

(f,)

,9

Ui

ro

'D

g 080

'~

Expllrirn ental

075
500

1000

1500

2000

2500

3000

25

30

Effective mean normal stress, p' : kPa


(a)

Effective mean stress , p': MPa

2500

Fig. 9. Experimental data and theoretical fitting of the critical state


line in e- p' space for Toyoura sand

co 2000

2z

Po" 3000 kPa

'0-

~ 1500

e:

5000r---------------------------------------------~

co 4000

---::::::::-;:::-.:.-...

Ui

- - Model simulation

ro

.......... Test results

':;;
Q)

2z

Po': 1000 kPa

'0~

3000

Po' : 2000 kPa

~
E
2000
co

10

Q)

20

15
Axial strain , 1'1: a::n
(b)

:;;

1000

Fig. 11. Comparison between undrained triaxial compression test


results and model simulations for Dr = 379% (e = 0833)
1500

1000

500

2000

2500

3000

Effective rne oln normal stress. p': kPa


(a )

1200

5000 r---------------------------------------------~

- - Model simulation

co

ro

4000

2z
~
.u;.

E
2000
cr

2z

.. ~ .....--.. - .--.

u; JOO()

./
.--

~
' " Pu'

..Plj' : i

--...----. Test results

1000

:.:1- SOO

e:

00 kPa

"th

)00

(5

1000 kPa

n,;

:;

.~

400

Q;

200

p '- : 3000 kPa

10

15

20

2[;

2500

30

:lOOO

Effectiv rm,,(ln norrn81 siress. [J ' kF\ :l

/-\xial strcli n, . ',' (::~

(8)

(tJ)
12ClO

Fig. 10. Comparison between undrained triaxial compression test


results and model simulations for Dr = 63'7 (e = 0'735)

1000
ro

c:
C'"

2000 kPa

800

UJ

e with the initial void ratios eo (i.e. h = hi - 112eo instead of


h = hi - h2 e) brought the model simulations much closer to

the experimental counterparts (Fig. 15). This modification also


slightly improved other drained simulations, but has no effect
on the undrained response because I:' = eo under undrained
conditions. This 1:'0 dependence may be due to the influence of
material fabric at very low densities, since thc void ratio of a
sand without a significant shearing history eo is strongly
correlated to the packing structure of the sand (see Figs I and
2). However, since the stress and material state corresponding to
a given eo are not always clearly and objectively defined,
introducing eo into a constitutive equation as a general parameter needs further investigation ,

if;

7T:

.9

\"1:.;;:

CJ

[)

10

15
P.Xi 81 ,;trail1 , I'

20

30

. ~:i,

(Il)

Fig. 12. Comparison between undrained triaxial compression test


results and model simulations for Dr = 185% (e = 0907)

LI AND DAFALIAS

458
3'.in

CONCLUSIONS
- - IvlocJ el sirnulil lion

300

Tes t results

ru

~ 250
.&

~ 200

150

g
in

0:;

100

Q)

50

08

095

09

085

Void ratio.

10

(a )

350
300
m

22 250
0-

m
sQ)
0

10

15

20

25

30

A xi al strain , {' 1: %,

The classical stress dilatancy theory in its exact form ignored


the extra energy loss due to the static and kinematic constraints
at particle contacts. While this hypothesis leads to a unique
relationship between the stress ratio and dilatancy, it obstructs
unified modelling of the behaviour of cohesionless soils over a
full range of densities and stress levels .
To remove this obstacle, additional dependence of dilatancy
on the internal state of a material is needed. T he concept of
state-dependent dilatancy was introduced, in conjunction with
the basic concepts of critical-state soil mechanics. The general
expression and basic requirements for dilatancy were addressed .
It has been shown that the state parameter, the difference
between the current void ratio and the critical state void ratio
corresponding to the current confining pressure, is an effective
means of measuring how far the material state is from the
critical state. With the state parameter as the state variable, a
particular form of state-dependent dilatancy was proposed, and
was shown to be equivalent to an interpretation whereby the
phase transformation stress ratio is variable with the state
parameter, an idea introduced by Manzari & Dafalias (1997).
With this fonn of dilatancy and a state-parameter-dependent
plastic modulus , for which an interpretation is again possible
whereby the peak stress ratio depends on the state parameter as
proposed by Wood et al. (1994), a simple model in the triaxial
space as well as a systematic calibration procedure was introduced. It was shown that this simple model has the ability to
simulate data successfully, with a single set of model constants,
for a suite of 17 triaxial tests , both drained and undrained, of
Toyoura sand over a relative density range of around 0-64%
subjected to a confining pressure range of 100- 3000 kPa.

(b)

Fig. 13. Comparison between drained triaxial compression test


results and model simulations for Pn = 100 kPa

1600
Model sim ulation

ACKNOWLEDGEMENTS

The financial support provided to X. S. L. by the Research


Grants Council (RGC) of Hong Kong through Grant
HKUST721 /96E is gratefully acknowledged. Y. F. D. acknowledges the support from the National Science Foundation, Grant
No. CMS-9800330, of the programme directed by Dr Cliff
Astill.

Test results
~ 1 200
.;,L

"c

APPENDIX. DILATANCY OF TWO REGULAR PACKINGS OF


UNIFORM RIGID RODS
Packing II (Fig. I)
Following Rowe (1962). when sliding takes place, one has the ratio of the
loads

t~

BOO

1]
>
(!)

-100

(34)

where L; is the load per rod in direction i, 1)" is the angle of friction
between the surfaces of the rods in contact, and (J is the deviation of the
tangent at the contact points from the direction I. The size of a basic cell
in the packing is characteri zed by

Of)

09

() 85
Void

095

"li'o, e
la)

[35a)

I, = 4rsin/j

1600

and
(35b)

I" = 4reos/i

Therefore, the ratio of the major principal st ress to the minor principal
stress is

~ 1200

0-

0; /01 = L,I, / L" I" = tan/3tan( " +/1 )

~:

rr,

'>J)

800

(36)

The ratio of the strain rates in direction s I and 2 is

m
;;

E"
-=

r.J

-dii/ v-f,

E,

400

E,

=-

( 1+
dV)
VEl

i"l,

')

=~ = -tan - II

(37)

I,I?

Therefore, the ratio of the work done per unit volume by the major
principal stress to the work done on the minor principal stress is
0

10

15
A xial strain
(t))

20
I,

25

30

0'
: \;

Fig. 14. Comparison between drained triaxial


results and model simulations for Pn = 500 kPa

aiEl

0;

OlE?

02(1 -I- d/>/VE ,)

tan(<j.l,,+13)
tan 13

(38)

The void ratio of the packing is


compression

test

I:'

= 1,1" - 2m'?'
2:71'~

= 16r~ sin/3cosj) _ I = 8sinl3cos(J _


2;r r ~

::r

(39)

459

DILATANCY FOR COHES ION LESS SOILS


350

iGOO
Moclel sim ula ti on

- - MO{jE:1 sirnu l3tio [l

300

Tes t re sults

Test results

~ 1200

t:i-

Q'

~ 250

~ 200

~
0
rn',>

vi

<1)

(jj

150

800

rn

'>
Q)

100

Q)

400

50
0
08

0,88

084

0,96

092
Void ralio,

10

084

08

088

092

096

Void ratio, e

1600

350
300

~ 250

-.-..-.

t:i-

t:i-

~ 200

~
(jj

e!

(jj

rn
',>
Q]

-_._.._----.._..- ....- ...

~ 1200

ru

150

800

e Q =0'810

rn
',>

100

Q)

400

50
0

0
20

10

30

10

Fig. 15. Simulations of drained tests with h varying with eo instead of e: (a) pu

Packing B (Fig. 2)

At each contact point, sliding takes place when the ratio of thc loads
LI
'
(3'
Ld2 = tan(if>J1 + , )

(40)

Thc size of a basic cell in this packing is characterized by


(41 a)

I I = 4rsin/~

and
(4Ib)

12 = 21' + 4,.cos/3 = 2,.0 + 2cos(J)

Thereforc, the ratio of the major principal stress to the minor principal
stress is

0;

2L I/ I
J 2sin/3
L212 = tan(r/J p +/) 1+ 2cos(J

(42)

The ratio of the strains in directions I and 2 is

h
(
di; )
~= - I+~

2sin2/J

i2/1

=- - (I

= il/2

+ 2 cos (3)cos/J

(43)

Thercfore, the ratio of the work done per unit voluille by the major
principal strcss to the work done on the minor principal stress is equal to
tJ; EI
tJ2E2

0;
02(1 +dil / vEI)

tan(if>f l + /-1)
tan/3

The void ratio of the packing is


11/2 -3m 2 8,.2 sin /J( I + 2 cos /3)
e==
3m 2

REFERENCES
Becn, K. & Jefferies, M.
Geofechnique 35, No , 2,
Bolton, M. D. (\ 986). The
niqlle 36. No. I , 65 - 78.
Chang. C. S. & Misra. A.

30

E 1: ~/O

(b)

(a)

01 =

20
Axial strain.

Axial strain , E;: %

(44)

1=

8sin/3( 1 + 2 cos (3)


- I
3;r
(45)

G. (1985). A statc parameter for sands.


99 - 112.
strcngth and dilatancy of sands. Geofech
(1990). Packing structure and mechani ca l

= 100 kPa,

It

hI - "zeo; (b) Po

= 500 kPa,

hI - "zen

properties of granulates. J Engng Mech, Div., ASCE 116, No.5,


1077-1093 .
Cubrinovski , M. & Ishihara, K. (1998). Modelling of sand behaviour
based on state concept. Soils Foundalions 38, No.3, 115-127.
Dafalias, Y. F (1986). An anisotropic critical state soi l plasticity model.
Meck Res. Commun. 13, No.6, 341-347.
Ishihara. K. (1993). Liquefaction and flow failure during earthquakes.
33rd Rankine lecture. Geofecilllhjlle 43, No.3, 351-415 .
Ishihara, K .. Tatsuoka, F & Yasuda, S. (1975). Undrained deformation
and liquefaction of sand under cyclic stresses. Soils Foundations 15,
No. I. 29-44.
Jefferies, M. G. (1993) . NorSand: a simple critical state model of sand.
Geolechnique 43. No. I. 91-103.
Kabilamany, K. & Ishihara. K. (1990). Stress dilatancy and hardening
laws for rigid granular model of sand. Soil D,1'nl/1l. Earthquake
Engng. 9, No. 2. 66- 77.
Li, X. S. (1997). Modelling of dilative shear fai lure. J Geofech, Engng
Dill , ASCE 123. No.7, 609 - 6 16.
Li. X. S. & Wang, Y. (1998). Linear representation of steady-state line
for sand. J Geutech. Geoenl'iron. Engng 124, No. 12, 12 15 - 1217.
Li, X. S., Dafalias, Y. F. & Wang, Z. L. (1999). State dependent
dilatancy in criti cal state constitutive modelling of sand. Cln.
Geo/ech. J 36, No 4. 599- 6 11 .
Manzari. M. T. & Dafalias. Y. F. (1997). A critical state two-surface
plasticity model for sands. Geofechnique 47, No. 2, 255 - 272.
Nova, R. & Wood, D. M. ( 1979). A constitutive modcl for sand in triaxial
compression. Inll J NUI1I. Anal. Methods Geol1lech. 3, 255 - 27)3.
Pastor, M ., Zienkiewicz, O. C. & Chan, H. C. (1990) . Generalized
plasticity and the modelling of soi l behaviour. Inll J NUl1lerical
Anl/. Me fh ods Geomech . 14, 151-190.
Poorooshasb, H. 8.. Hoilibec, 1. & Sherbourne, A. N. (1966). Yielding
and flow of sand in triaxial compression, Part I. Can. Geotech. J 3,
No.5, 179 - 190.
Poorooshasb, H. B.. Hoilibec, I. & Sherbourne, A. N. (1967). Yielding
and tlow of sand in triaxial compression, Part 11. Can . Geotech. 1. 4,
No.4 , 376 - 397.
Roscoe, K. H. & Burland, 1. B. (1968). On the generalized stress - strain
behav iour of 'wet' clay. Engineering plls/icil1~ pp. 535 - 609. Cambridge: Cambridge University Press.
Roscoe, K. H. & Schofield. A. N. (1963). Mechanical behaviour of an

460

LI AN D DAFALIAS

idealized 'wet' clay. Pmc-. EII/: COllI Sail Mech. FOllnd Ellgllg.
Wiesbaden 1,47 - 54.
Roscoe, K. H., Schofield, A. N. & Wroth , C. P. (1 958). On the yielding
of soi ls. Geatechnique 8, No. I , 22 - 53.
Rowe, P. W. (1962). The stress - dilatancy relation for stnti c equilibrium
of an assembly of particles in contact. Pmc. R. Soc. , LOlldon, SeJ: A
269, 500-527.
Ri chart , F. E. Jr., Hall , J. R . & Woods. R. D. (1970). Vibrations of" soils
and foundations. Englewood Cliffs, NJ: Prentice-Hall.
Taylor, D. W. ( 1948). Fllndam el1lals of soil mechanics. New York:
Wiley.
Vardoulakis, I. & Sulem. J. (1995). BijiJrcatian all a lysis in geomeehanics. Glasgow : Blackie Academic & Professional.

Verdugo. R. & Ishihara, K. ( 1996). The stea dy state of sand y so il s. Soils


FOllndlltions 36. No . 2. 8 1- 9 1.
Wan , R. G. & Guo, R. G. (1998). A simple constituti ve model for
granul ar soils: modified stress - dilatancy approach. Compl/t. Geotech. 22 , No.2, 109 - 133 .
Wang, Z. L. , Dafalias, Y. F. & Shen, C. K. (1 990) . Bounding surface
hypopl asticity model for sand. J. Engng Meeh. Div.. ASCE 116, No.
5,983 - 1001.
Wood, D. M. ( 1990). Soil behaviour and critical stale soil mechallics.
New York : Cambridge Uni versity Press.
Wood, M. D. , Belkhcir, K. & Liu, D. F. (1994). Strain softening and
state parameter for sand modelling . Geolechniqlle, 44, No . 2, 335 339.

You might also like