Professional Documents
Culture Documents
164]
This paper explains why the critical state of sand is non-unique when expressed in terms of stress
and void ratio only. For this purpose, a thermodynamically consistent, micromechanically inspired
constitutive modelling framework with competing grain crushing and dilation is developed. While
grain crushing is described through the theory of breakage mechanics, dilation is modelled in a
novel way by acknowledging its negative contribution to the overall positive rate of dissipation. The
competition between dilation and grain crushing underpinned by this framework yields a unique
critical state in a space of stress, void ratio and breakage, in agreement with recent experiments. As
an example, a simple constitutive model with only five mechanical parameters is proposed, which
not only predicts the critical state but also quantitatively connects the full constitutive behaviour to
key index properties related to grading- and breakage-dependent minimum and maximum densities.
695
Void ratio, e
Void ratio, e
08
075
070 e0 = 0664
04
e0 = 075 076
065
e0 = 065 067
emin = 060 0
0 200 400 600 800 100 1000 10 000
Mean effective stress: kPa Mean effective stress, p: kPa
(a) (b)
Fig. 1. Experiments revealing the non-uniqueness of critical state in a space defined solely in terms of stress and void ratio. (a) Dependence on the
initial void ratio e0, using drained plane strain compression tests on masonry sand (adapted from Finno & Rechenmacher, 2003). (b) Dependence
on grading (through relative breakage (Br, Hardin, 1985) using triaxial compression tests on Dogs Bay sand (adapted from Bandini & Coop,
2012); the latter results will be predicted using the new model in the section entitled Comparison with experimental data
60 Current VS
distribution, F
p
40
Ultimate 2c
distribution, Fu
pv 2d
20 1
In the following, this latter approach is followed, which
implies that is a purely dissipative internal variable (see
Effective grain size, xg (log scale) equation (2d)).
X gm X gM Appendix 1 presents a more general expression that ex-
plicitly relates the porosity rate to strain rate for grains with
Fig. 2. Definition of breakage B (Einav, 2007a) based on a universal any grain mineral Poisson ratio, g. It recovers the above
initial distribution as a Heaviside step function about the maximum relations for g = 0 as well as the
more traditional
assumption
grain size X M
g
in soil mechanics that v =1 for (in fact, unrea-
listic) incompressible grains (i.e. g = 05).
Ds Mpps 11c
Dissipation rates of dilatant crushable sand The first term DB in equation (11a) is the one proposed by
In general, the total rate of dissipation depends on Einav (2007b), which physically expresses the rate of loss of
contributions from the rates of the three dissipative internal residual breakage energy; that is, the rate of elastic energy
variables (i.e. rates of breakage, porosity and plastic shear loss needed to move a material at a given state towards an
strain). Assuming first-order homogeneity in these rates and ultimate GSD. The only change for DB from the original
consulting Appendix 2 formulation of Einav (2007b) is the use of the Macaulay
@ @ @ brackets function h i (i.e. hxi = 0 for x , 0, or x otherwise) in
B p ps 2
B =hBi. This is important for the current model, to ensure
@ B @ @ s 7
B 0 even during dilation, such that crushing is strictly
EB B E qps connected to grain size reduction. Therefore, the use of the
q
Mp
Rates of inelasticity and conditions for critical state
Combining equation (10) with equation (11) the following Dilation regime
(F < 0)
yield function can be derived directly from the explicit form
Critical state (F = 0)
of the dissipation rate function 1
r 2 2
EB q
y 1 B 1 0 13
EC Mp Crushing regime (F > 0)
Also using equations (10) and (11) the following flow rules
are obtained, which determine the evolution of the dissipa-
tive internal variables
@y* EB
B (1 B)
@EB EC
r
EB 1 B
2 1 B p cos2 14a Fig. 4. The yield surface in a generalised stress-space, which
EC EB EC highlights the various flow regimes and a unique critical state
750 300 =0
B=0
B = 03 = 03
B = 05 = 07
600 240
B = 08 = 10
450 p
180 p
M M
q: kPa
q: kPa
=
q
= q
300 120
150 60
0 0
150 300 450 600 750 60 120 180 240 300
p: kPa p: kPa
(a) (b)
Fig. 5. The dependence of the yield surface on (a) breakage B for = 05 and (b) the relative porosity for B = 0, when linear elasticity is employed
within breakage mechanics
2500 400 =0
B=0
B = 03 = 03
B = 05 = 07
2000 320
B = 08 = 10
1500 240 p
p M
M
q: kPa
q: kPa
= =
q q
1000 160
500 80
0 0
500 1000 1500 2000 2500 80 160 240 320 400
p: kPa p: kPa
(a) (b)
Fig. 6. The dependence of the yield surface on (a) breakage B for = 05 and (b) the relative porosity for B = 0 when non-linear
pressure-dependent elasticity is employed within breakage mechanics
16 16
B=0
B=0
Dr = 025
12 B = 03
Dr = 05
Dr = 075
Dr = 025 e 08 B = 07
12
Dr=
B = 03
1
04
Dr = 05
B = 095
e 08 Dr = 075
102 103 104
B = 07
p: kPa
Dr
=1
04
B = 095
0
3000 6000 9000 12 000
p: kPa
Fig. 7. Predicted family of critical state lines when linear elasticity is employed within breakage mechanics (equation (18)). Dashed lines show
constant B critical state lines; solid lines show constant Dr critical state lines. Darker lines show the upper bound envelope. Inset is a semi-log
representation
16 16
B=0
B=0 Dr = 025
12 B = 03
Dr = 05
Dr = 075
Dr = 025 e 08 B = 07
12 D
B = 03 r =
1
04
Dr = 05
B = 095
Dr = 075
e 08 B = 07 102 103 104
p: kPa
D
r =1
04
B = 095
0
3000 6000 9000 12 000
p: kPa
Fig. 8. Predicted family of critical state lines when non-linear, pressure-dependent elasticity is employed within breakage mechanics
(equation (20)). Dashed lines show constant B critical state lines; solid lines show constant DR critical state lines. Darker lines show the upper
bound envelope. Inset is a semi-log representation
express the critical state lines as an implicit relation between under increasing breakage. The figure shows that the grains
pcs, cs and Bcs, as shown in Fig. 8. strength and their potential to dilate indeed increase the
The exact definition of critical state in terms of F = Fcs = 0 critical state porosity for a given pressure, as one might expect
is expressed in terms of a number of material properties. intuitively.
Therefore, the effect of these parameters on the critical state
can be explored within the context of the proposed frame-
work. For example, the effect of grains strength (by way Comparison with experimental data
of parameter EC) and their potential to dilate (by way of Reconsider the experimental results in Fig. 1(b) by Bandini
parameter ) on critical state can be seen by the upper bound & Coop (2012). Rather than fitting lines to those experimen-
critical state envelope in Fig. 9. As mentioned before, this tal data points, it is now possible to use the new framework to
envelope comprises two lines: the zero breakage critical state attempt an actual prediction. For that purpose, first the
line, and a line defined by the highest relative density, Dr = 1, parameters controlling the exact form of the critical state of
e e
08 08
04 04
Fig. 9. Dependence of the upper bound critical state envelope on (a) the materials potential to dilate through and (b) the grains strength through
the critical breakage energy, EC
K
G M EC: kPa u l u l
Dogs Bay sand (a crushable carbonate sand) are determined. using a single critical state point in stressbreakageporosity
These are grouped in terms of mechanical parameters from space. As discussed above, this parameter can be deduced
stressstrain curves, and index properties from geometrical independently from critical state using the volumetric response
tests such as sieve analysis and density measurements. of the material and can then be used to predict the critical
state. Alternatively, it was shown that could be determined by
anchoring the critical state family of lines to pass through a
single critical state point/experiment. The rest of the critical
Index properties. The grading index 2/(5 )
state relationship is truly predicted. This therefore disting-
(Buscarnera & Einav, 2012, using zero minimum grain size)
uishes the current framework from previous methodologies
was evaluated assuming an ultimate fractal GSD with a
within critical state soil mechanics for crushable sand that
fractal dimension = 26, in agreement with the proposed
require curve-fitting many experimental critical state points
theoretical values in Sammis et al. (1986) and experimental
using material-dependent empirical relationships. The high
results from large ring shear by Coop et al. (2004). In the
value of = 095 obtained (closer to 1 than 0) appears to reflect
absence of data on Dogs Bay sand, the power coefficients
the high angularity and strongly dilative behaviour of Dogs
u and l determining the progression of relative porosity
Bay sand observed in Coop (1990).
with B in equation (3) were assumed to be those deduced in
The above parameters (mechanical and index properties)
Fig. 3 for the data by Youd (1973). The limiting porosities at
are tabulated in Table 1. The corresponding theoretical
zero breakage, l and u, were evaluated from the minimum
prediction of the family of critical state lines is shown in
and maximum porosities measured in Golightly (1989) for
Fig. 10, which the authors believe to be very encouraging,
the corresponding grading.
especially considering the agreement is not a result of curve-
fitting. As can be observed in that figure, the proposed model
predicts non-linear critical state curves, unlike the linear ones
Mechanical parameters. The non-dimensional bulk and drawn ad hoc in Bandini & Coop (2012) and Muir Wood &
shear stiffness coefficients K and G were deduced from the Maeda (2007). Furthermore, the new theory predicts the
isotropic compression and bender element tests in Jovicic & reduced inclination of the critical state lines with increasing
Coop (1997), respectively. The critical state friction coeffi- breakage, in agreement with the experimental conclusion of
cient, M = 165, was reported in Coop (1990). The critical Bandini & Coop (2012) and Fig. 1(b).
breakage energy EC can be expressed in terms of the yield
pressure in isotropic compression, pcr, by inserting equation
(22a) into equation (13) and solving for q = 0 CONSTITUTIVE RESPONSE
" #2=3 To complete the constitutive model and evaluate its
3 12 Ec K 1 B response, all that is left is to specify . As shown above,
pcr pr 25
2pr 1 B1=3 this does not control the form of the critical state. The
parameter represents the sands tendency to undergo irre-
where can be calculated using equation (4), using the versible volume changes through fragment rearrangement.
reported initial porosity. Then, EC is found by employing pcr Previously, this parameter was taken as a material constant
from the data in Coop (1990). Finally, the parameter (Einav, 2007b), but now can be expressed in terms of the
controlling the sands potential to dilate, , was determined relative porosity. More specifically, in the compacting
K
G M EC: kPa u l u l
075 012
v
q/p
050 006
p0 = 5% pcr
025 0
p0 = 40% pcr
p0 = 75% pcr
0 006
006 012 018 024 030 006 012 018 024 030
a a
(a) (b)
100
p0 = 5% pcr p0 = 5% pcr
p0 = 40% pcr p0 = 40% pcr
12
p0 = 75% pcr 075 p0 = 75% pcr
B=0 Critical state
09
e B 050
06
B = 087 025
03 B = 093
0
0
300 600 900 1200 1500 006 012 018 024 030
p: kPa a
(c) (d)
Fig. 11. Drained triaxial test responses for 0 = min, for different initial pressures p0 = 5%, 40% and 75% of pcr: (a) stress ratio q/p with vertical
strain; (b) strain response; (c) evolution of void ratio from initial value to critical state; (d) evolution of breakage, B
Breakage and pressure at critical state in triaxial tests (Rubin & Einav, 2011). A novel decomposition of the vol-
A number of simulations were run up to critical state for umetric strain rate was proposed, with which the traditional
different initial relative porosities under the drained and the simplifying assumption of grain incompressibility was re-
undrained triaxial conditions presented above for increasing moved. This, in return, enabled the role of the solid skeleton
confinements. This has been used to obtain Fig. 15 relating deformation to be considered in the mechanism of elastic
critical state breakage Bcs and mean stress pcs. The role of the energy storage.
initial relative porosity 0 is apparent at low confinements, but A novel thermodynamics framework was developed to
vanishes at higher confinements, where breakage asympto- address the role of dilation in controlling energy dissipation,
tically approaches unity. The proposed model is in agreement which reflects its negative contribution to the rate of work.
with previous breakage mechanics models (Einav, 2007c; From the proposed formulation, two competing regimes of
Nguyen & Einav, 2009) at the high-pressure regime they were response have been identified: a dilative regime and a com-
aimed at, and extends the scope of the original formulation pacting breakage regime. The transition between the two
to lower pressures. Similar relations between pcs and Bcs were regimes is marked by the critical state, as a unique stable
observed experimentally by Lade et al. (1996), where relative equilibrium point towards which the material flows in a
breakage (Hardin, 1985) was measured in both drained and generalised thermodynamics stress space.
undrained triaxial tests on Cambria sand. Motivated by the This competition between dilation and crushing produces
model, future experimental research is recommended to a family of critical state lines in a stressporosity plane, as
explore the effect of the initial density on the pcsBcs a function of the extent of breakage undergone by the
relationship. material. The critical state emerges as an explicit function
from the formulation itself, rather than being imposed
a priori. The family of the critical state lines is shown to
CONCLUSIONS be a function of mechanical parameters, including the
This paper has dealt with the physical origins of the friction angle, and the grains strength and stiffness; and
experimentally observed dependence of critical state in sand further depends on index properties, including the grading-
on grain crushing by means of a micro-mechanics based dependent minimum and maximum porosities and the
constitutive model, developed under the theory of breakage grading index.
mechanics (Einav, 2007a, 2007b). The new breakage mechanics model depends on only
Porosity was introduced as a frame-independent internal five mechanical parameters: two stiffness parameters, two
state variable, and its value was ensured to vary between strength parameters and a parameter controlling the trans-
grading-dependent minimum and maximum porosities ition between the dilation and the breakage-dominant
075 018
v
q/p
050 012
p0 = 5% pcr
025 006
p0 = 40% pcr
p0 = 75% pcr
0 0
006 012 018 024 030 006 012 018 024 030
a a
(a) (b)
e B
06
B = 088 015
B = 094
03
0
0
250 500 750 1000 006 012 018 024 030
p: kPa a
(c) (d)
Fig. 12. Drained triaxial test responses for 0 = max, for different initial pressures p0 = 5%, 40% and 75% of pcr: (a) stress ratio q/p with vertical
strain; (b) strain response; (c) evolution of void ratio from initial value to critical state; (d) evolution of breakage, B
regimes. The model also makes explicit use of index prop- For the elastic component (i.e. ignoring grain rearrangement),
erties, to be evaluated from the GSD and corresponding the difference between the solid volume after (VII S ) and before
minimum and maximum porosities. This is a powerful fea- deformations (VIS) over time (dt) allows the rate of change of the
ture of this model, which enables the framework to avoid the solid fraction to be specified in terms of an elastic expansion
coefficient (c)
use of ad-hoc mechanical fitting parameters while connect-
ing the constitutive behaviour to the evolving fabric. VS
V S V T 1 c 28
Finally, a parametric study was carried out that illustrated VT
that the proposed model qualitatively captures previous
Therefore, since the above applies only for the elastic component
observations of critical state and constitutive responses. It
is proposed to carry out further experimental investigations V S 1
into the effect of grading on the minimum and maximum ev 29
VS 1 c
porosities, and to explore the full parametric space for further e p
quantitative assessment of the model. The porosity rate involves both reversible ( ) and irreversible ( )
e p
contributions. Then, assuming and together with
equations (27) and (29)
APPENDIX 1. GENERAL STRAIN DECOMPOSITION ! !
e p
IN POROUS MEDIA
Recalling the definition of the total volumetric strain rate c v
e
v
p
0 30
1 1
(v V T =VT ) and its decomposition into elastic and plastic
strain rates (i.e. v ev pv ) equation (1) becomes Therefore, the relations between the rates can be grouped in terms of
the reversible and irreversible components
V S
pv ev 27
VS 1 e
c1 ve 31a
Here two distinct phenomena accommodating volumetric strain
p
in granular materials can be identified: grain rearrangement and
inter-granular strain. Grain rearrangement is considered as a pv 31b
1
dissipative, irreversible process, and therefore it is connected to the
plastic strain rate. The inter-granular strain is associated with What is left is to determine the expansion coefficient, c. Before
the elastic energy stored in the system, and is therefore related to addressing porous granular materials, a bulk isotropic elastic solid
the elastic strain. Fig. 16 illustrates kinematical aspects of these two made out of the mineral constituting the grains will first be con-
mechanisms. sidered. From the theory of elasticity, c must depend on the grain-
600
q: kPa
04
q/pcr
400
02
200
0 0
200 400 600 800 1000 006 012 018 024 030
p: kPa a
(a) (b)
100
16 p0 = 5% pcr p0 = 5% pcr
p0 = 40% pcr p0 = 40% pcr
p0 = 75% pcr 075 p0 = 75% pcr
12 Critical state
B=0
e B 050
08
B = 072
B = 077 025
04
0
0
200 400 600 800 1000 006 012 018 024 030
p: kPa a
(c) (d)
Fig. 13. Undrained triaxial test responses for 0 = min, for different initial pressures p0 = 5%, 40% and 75% of pcr: (a) stress paths in qp plane;
(b) stress ratio q/pcr with vertical strain; (c) evolution of void ratio from initial value to critical state; (d) evolution of breakage, B
mineral Poisson ratio g and the loading conditions. Under APPENDIX 2. MODEL DERIVATION
unconfined uniaxial compression c = 2g (one compressing principal Assuming rate independence, three distinct dissipative mechan-
strain); under biaxial compression c g/(1 g) (two compressing isms are expressed as homogeneous first order in the rate of the three
principal strains); whereas under isotropic triaxial compression c = 0 dissipative internal variables. Therefore, according to Eulers
(three compressing principal strains). theorem for first-order functions
For a granular assembly, which is intrinsically porous, it is further
expected that c will depend on the porosity . For ! 0 the material @ @B
approaches the behaviour of a bulk isotropic elastic solid, where B ; B B EB B 33a
@ B @ B
grains can no longer expand and c = 0. For highly loose materials c
should approach the idealisation of the unconfined uniaxial
compression, as most of the grain surfaces are free to expand. @ @
; E 33b
Therefore, the following expansion coefficient may be considered @ @
(for g 0)
@ p @s p
c 2vg 1= 32 s ; p s qps 33c
@ps s @s
which reflects the above idealisations with: (a) c = 0 when = 0 (for The stress conjugates to the dissipative internal variables can then be
g , 05); (b) c = 2g when = 1; and (c) c g/(1 g) when 06; specified using equation (9)
furthermore, note that (d ) c = 1 for g = 05, irrespective of , which
represents the limit of an incompressible solid. 0 1
As the hypothesis of incompressibility is often taken in soil @ @DB B DB C @DB
mechanics to relate porosity with volumetric strain, the special case EB @q rB A 34a
@DB @ B D2 D2 D2 @ B
of g = 05 (c = 1) will now be considered. In the current formulation, B s
indeed, it follows from equation (28) that the solid fraction does 0 1
not change its volume. Therefore, using equation (1b), it is found
that v =1 , which recovers the usual approximation @ @D B D C @D
E @q r A 34b
in soil mechanics. However, taking quartz and silica as typical @D @ D D D2
2 2 @
B s
sand minerals with g = 017, and for typical porosities , 05, it is
estimated that c , 01, much smaller than 1 and closer to 0. 0 1
Therefore, it follows that equation (2) (following the assumption
that ev V S =VS ) is more realistic than the traditional relation in @ @Ds B Ds C @Ds
q @q rs A p 34c
equation (27) (following the grain incompressibility assumption that @Ds @ps D D D2
2 2 @s
B s
V S 0).
q/pcr
030
200
100 015
0 0
100 200 300 400 006 012 018 024 030
p: kPa a
(a) (b)
16 045 p0 = 5% pcr
p0 = 40% pcr
p0 = 75% pcr
12 B = 001
B = 007 030
B = 012
e B
08
015
p0 = 5% pcr
04 p0 = 40% pcr
p0 = 75% pcr
Critical state 0
0
100 200 300 400 006 012 018 024 030
p: kPa a
(c) (d)
Fig. 14. Undrained triaxial test responses for 0 = max, for different initial pressures p0 = 5%, 40% and 75% of pcr: (a) stress paths in qp plane;
(b) stress ratio q/pcr with vertical strain; (c) evolution of void ratio from initial value to critical state; (d) evolution of breakage, B
10 10
0 = 0 0 = 0
08 08
0 = 03 0 = 03
0 = 07 0 = 07
06 0 = 1 06 0 = 1
Bcs Bcs
04 04
q: kPa
q: kPa
02 02
p: kPa p: kPa
0 0
400 800 1200 1600 200 400 600 800
pcs: kPa pcs: kPa
(a) (b)
Fig. 15. The dependence of breakage on the pressure at critical state for different initial relative porosities in (a) drained triaxial compression and
(b) undrained triaxial compression. The inset of (a) shows the drained stress paths until critical state. The inset of (b) shows indicative undrained
paths for 0 = 0 until critical state
The general form of the yield function is then found by rearranging Also, using equation (34) and Eulers theorem for first-order
equation (34) functions
2 !2 @ @ @
EB E B p ps
y* rB r @ B @ @ s 36
@DB =@ B @D =@
35 EB B E qs
2
q which confirms the use of equation (9) as a valid rate of dissipation
rs 1 0
@Ds =@ps potential.
El
ic
as
st
dt tic
a
Pl
dt
d pv After (II) d ev V IIs = Va + Vb
V IIs = V sI V
Va = V sI + VT s dt
VT
V
Vb = cVT s dt
VT
Fig. 16. Plastic and elastic deformations from grain rearrangement and inter-granular elastic strains, respectively. The volume of solids after
S can always be decomposed into a self-similar component Va and a superposed expansion component Vb
elastic deformation VII
060
100
045
075
q/pcr
q/p
030
050
s=1 s=1
025 s=2 015 s=2
s=4 s=4
s = 100 s = 100
0 0
002 004 006 008 010 002 004 006 008 010
a a
Fig. 17. Effect of the seismicity parameter, s, in equation (38) on the stressstrain responses of the model: (a) stress ratio q/p plotted against
vertical strain during drained triaxial tests for = min and p0 = 5% of pcr; (b) normalised deviatoric stress q/pcr plotted against vertical strain
during undrained triaxial tests for = min and p0 = 40% of pcr
Then, using the above, the first and second laws of thermo- follows
dynamics in equation (5), and the expression for the rate of work in
equation (6) @y*
B hi1 ys 2hi1 ys
@EB
r 38a
@ p @ EB 1 B
EB B E p q s
p 1 B p cos2
@B 1 @s EC EB EC
37
@ @
p e
q es 0 @y*
@ev v
@es hi1 ys 2hi1 ys
@E
Finally, identifying p and q with the corresponding elastic strains r 38b
EB 1 B EB 2
and employing Zieglers orthogonality conditions, equation (17) 1 B p sin
EC EB EC E
follows.
@y* q
sp hi1 ys 2hi1 ys 38c
@q Mp2
APPENDIX 3. A POSSIBLE MODEL EXTENSION
In the following it is demonstrated that the model can be extended where the seismicity parameter s (Einav, 2012) controls the gradual
within the proposed framework to predict more realistic stressstrain development of inelasticity observed through continuous acoustic
responses, if desired, without affecting the prediction of the critical emission (e.g. Fernandes et al., 2010). These flow rules are activated
state. Here, a notable limitation of the simple model proposed is continuously, irrespective of the value of the function y or the
addressed the issue of the non-smooth transition from elastic to sign of . It could also be demonstrated that using the above flow
inelastic responses. rules automatically enhances the models ability to simulate cyclic
To resolve this issue, the recent theory of Einav (2012) is used; stressstrain responses. Most importantly, the flow rules for the
this enables a simple extension of the hyper-plastic type of models porosity and breakage rates vanish under the same condition in
(that are incrementally bilinear) into a more general hypo- equation (15), that determines the critical state. Therefore, this
plastic form (that is incrementally non-linear) that obeys thermo- extension does not alter the predictions of critical state. Also,
dynamics laws. In Einav (2012) such hypohyper thermodynamics since the difference between the above flow rules and those
models were termed h 2 plastic models. In using h 2 plasticity in equation (14) is given by strictly positive terms, the rate of
the constitutive behaviour is no longer split bilinearly using the dissipation is maintained non-negative by virtue of equation (7).
yield surface, but the yield function in equation (13) and the Along the lines of h 2 plasticity, it is also possible to introduce rate
plasticity multiplier are still used when defining the flow rules as dependence as recently proposed by Hollenstein et al. (2013).
075 012
v
q/p
050 006
p0 = 5% pcr
025 p0 = 40% pcr 0
p0 = 75% pcr
0 006
006 012 018 024 030 006 012 018 024 030
a a
(a) (b)
10
p0 = 5% pcr
p0 = 40% pcr
12 08
p0 = 75% pcr
Critical state
09 06
e B
06 04
p0 = 5% pcr
B = 087
03 B = 093 02 p0 = 40% pcr
p0 = 75% pcr
0 0
300 600 900 1200 1500 006 012 018 024 030
p: kPa a
(c) (d)
Fig. 18. Drained triaxial test responses with s = 1 for 0 = min, for different initial pressures p0 = 5%, 40% and 75% of pcr: (a) stress ratio q/p with
vertical strain; (b) strain response; (c) evolution of void ratio from initial value to critical state; (d) evolution of breakage, B
The effect of s on the response of the model is highlighted in G shear elastic stiffness
Fig. 17, showing (for both drained and undrained triaxial con-
G non-dimensional material constant representing the shear
ditions) that under s ! the model recovers the response of the elastic stiffness in non-linear elasticity
model shown in the main body of the paper. Under finite s values the H Heaviside step function
model predicts smoother and more realistic responses. K bulk elastic stiffness
As a further illustration, the drained triaxial tests in Fig. 11 K non-dimensional material constant representing the bulk
for s = 1 will also be repeated. As shown in Fig. 18 the constitutive elastic stiffness in non-linear elasticity
response of the model is now more realistic (in terms of smoothness). l exponent controlling the evolution of limiting minimum
Note that in this model yielding is continuous and gradual, and thus porosity with breakage
pcr only indicates the pressure at which the rate of comminution M critical state friction coefficient
suddenly begins to increase. This is why breakage develops p mean stress
even prior to attaining this pressure level during isotropic com- pcr critical pressure: yield pressure in isotropic loading
pression, and prior to the start of shearing, as shown in Fig. 18(d) pr reference pressure for pressure-dependent model
for p0 = 75% pcr. q deviatoric stress
rB multiplicative term of the rate of dissipation related to
breakage
rs multiplicative term of the rate of dissipation related to shear
NOTATION r multiplicative term of the rate of dissipation related to
A symbol linked to the pressure-dependent model, see equation porosity evolution
(21) s seismicity parameter
B breakage variable: measure of the advancement of grain u exponent controlling the evolution of limiting maximum
crushing porosity with breakage
C combination of model parameters, used in equation (24) VS volume of solids
c elastic expansion coefficient VT total volume
D contribution to rate of dissipation in coupled model W rate of mechanical work on the boundaries of the RVE
Dr relative density (representative volume element)
g
EB breakage energy, energy conjugate to B XM maximum effective grain size
EC critical breakage energy X gm minimum effective grain size
E porosity energy, energy conjugate to xg effective grain size
e void ratio y yield surface
F function controlling the transition from dilative to compac- y* general form of the yield function
tive regime, expressed in equation (15) l minimum porosity at breakage B = 0
Fcs locus of F = 0, corresponding to critical state u maximum porosity at breakage B = 0