Professional Documents
Culture Documents
DOI 10.1007/s11440-007-0041-0
RESEARCH PAPER
Received: 1 April 2007 / Accepted: 7 September 2007 / Published online: 27 October 2007
Springer-Verlag 2007
1 Introduction
Laboratory testing of soils and development of constitutive models for soils tend to assume that the soil that is
being prepared or studied at the start of a test is the same
as the soil that emerges at the end of the test and that, in
principle, the emerging soil could be used to repeat the
test and obtain essentially identical response. Such a
description is of course a caricature: it has long been the
custom in research laboratories to discard the soil at the
end of a test because of potential damage to individual
particles; and over recent years there has been increasing
study of soils such as carbonate sands in which particle
breakage is an inescapable feature even at relatively
modest stress levels. But if the particles are breaking at
lower or higher stresses, then there is a change of material
during the test and we could certainly not expect the same
response if the damaged particles were used to create a
new sample. From a modelling point of view too we
should recognise that the material is changing: particle
breakage represents an irreversible change to the material
with the proportion of fine material in the soil monotonically increasing.
The opposite transition is also possible and this will be
briefly described in the context of a dam in British
Columbia where there is evidence that fine particles have
been slowly washed out of the originally well-graded core
material. Again the soil is changing during its life and the
mechanical response of the original material may not be
the same as that of the material today.
From the point of view of constitutive modelling, there
is a need for a simple parameter to provide some indication of the nature of the current particle size
distribution of a soil and then assumptions are needed
concerning the way in which variation of this parameter
actually influences the mechanical behaviour of the soil.
This paper presents some numerical results concerning
one aspect of this influence: the nature of critical states.
For many soils, constitutive models which incorporate
some description of asymptotic volumetric states
123
123
123
123
7
100
90
before removal
after removal
for RD=10
80
W (%)
70
60
50
RD(=Dmax /Dmin)=20
40
30
RD=10
RD=5
20
10
Fig. 5 Particle size distributions for natural well-graded soils compared with fractal distributions
0
5
10
50
RD=2
100
123
generation
generationofof
particles
particles
stress
stress
regulation
regulation
change
changeinin
interparticle
interparticlefriction
friction
initial
initial
packing
packing
emax
ng=30-40%
tang=1.0
m=0.0
e
tan=0.25
e
tang=0.1
ng=30-40%
m=0.0
tan=0.25
emin
tan=0.25
ng=10%
tang=0.0
m
m
Isotropic comp.
up to 0.05(MPa)
Isotropic comp.
& shearing
0.05 (MPa)
m0
Fig. 7 Method of preparation of DEM samples: varying generation porosity, ng, and interparticle friction, tan /l to produce different eventual
packings
123
Friction during
initial compression
tan /lg
Friction during
shearing tan /l
Final void
ratio
emax
0.30.4
0.25
0.30.4
0.1
0.25
0.1
0.25
emin
0.4
-5
0.2
c=0.1MPa
-0.2
0
RD
2
5
10
20
4
6
8
10 12
Normal strain, yy (%)
Stress ratio, m / m
0.6
0.4
-5
0.2
-0.2
0
4
6
8
Normal strain, yy (%)
1.25
RD=Dmax/Dmin=20
isotropic compression
shearing
origin of shearing
critical state
1.20
1.15
0
1.0
2.0
3.0
Mean normal stress, m (MPa)
v=a0+a1m
RD=Dmax/Dmin
-10
RD
2
5
10
20
5
14
m C k ln p0
5
10
-10
Stress ratio, m / m
0.6
Having created and compressed the assembly of particles it is then sheared to large straintypically around 10%
which seems to be sufficient to reach reasonably asymptotic conditions which might be interpreted as critical states
(Fig. 8). (The inevitable scatter in the response in these
numerical experiments can be smoothed visually when
numerical limiting values are extracted.)
Here, the mobilised friction angle at the critical state
/cv & 16 (sin /cv = sm/rm = 0.275) is only slightly larger than the interparticle friction angle /l = 14
(tan /l = 0.25). Some results indicate that the mobilised
friction angle at the critical state /cv should always be
larger than /l (e.g., [3, 4]). The contrasting result found
here suggests that there is little energy dissipation in the
granular assemblies arising from the geometric arrangement of the particles. Other DEM simulations have been
performed to examine the influence of /l in the wide range
from 0 to 90 on the mobilised friction angles at peak /f
and critical state /cv (e.g., [17, 18, 24]). Maeda and
Hirabayashi observed that /f was greater than /l when /l
was less than about 30 for samples formed of circular
particles. However, when /l was greater than 30, the
value of /f was found to be less than /l. Further, /f
converged to a limit value of about 30 for values of /l
greater than 1520, even for /l close to 90. The value of
the critical state angle /cv showed the same tendencies as
/f, as /l was varied and was, if anything, less sensitive to
the value of /l. As a further group of numerical
1.25
RD= 2
RD= 5
RD=10
RD=20
:
:
:
:
a0
1.239
1.230
1.211
1.189
a1
-0.015
-0.015
-0.013
-0.011
|r|
0.968
0.976
0.986
0.993
1.20
1.15
0
1.0
2.0
3.0
Mean normal stress, m (MPa)
123
10
Coordination Number, Nc
counter-intuitional: the broader the grading the more efficient the particle packing, the lower the limiting values of
specific volume or void ratio (lower C), and the higher the
stiffness (lower k). Thus this accords with the experimental
result of Lade et al. [13] that the presence of fine particles
lowers the void ratio for reference states such as the critical
state line.
The coordination number provides some indication of
the current fabric of the material. The spread of coordination number in samples with different gradings at the
start of shearing is shown in Fig. 10. While the infrequent
occurrence of coordination numbers as high as 16 is
observed for the broadest grading studied (RD = 20) the
dominant elements of the distribution of number of contacts are (surprisingly?) insensitive to the grading. The
most frequent number of contacts is generally four. There
is a hint of a reduction towards three for the broadest
grading and a slight increase in the number of particles
with one contact or fewer. A partial view of the evolution
of fabric during the tests is then shown in a plot of average
coordination number with mean stress (Fig. 11). A reasonably unique relationship emerges corresponding to the
findings of Thornton and Zhang [30] and Nougier-Lehon
et al. [20]. Thus, with coordination number and a measure
of grading together we now have two extra dimensions to
feed into the description of the asymptotic state of the
granular material.
The next series of tests were started as before. A sample
with a given grading was compressed isotropically and
sheared to a certain mobilised friction. The process of
changing the grading by removal of particles was then
performed as follows. From a state of equilibrium, the
smallest disc in the sample was located and removed from
the assembly. The removal of this particle leaves some
unequilibrated interparticle forces in the assembly: external
stresses were controlled and kept constant until deformations induced by these unequilibrated forces had
RD=Dmax/Dmin=2
4.0
isotropic compression
shearing
origin of shearing
critical state
Critical State Line
3.5
3.0
1.0
2.0
3.0
0.5
RD= 2
RD= 5
RD=10
RD=20
0.3
0.350
0.300
V o id ra tio , e
Frequency
0.4
0.2
max. coordination number 16
for RD=20
0.1
0.250
0.200
emin
0.150
0
0
10
15
20
123
removal underm/m
0 (i.c.)
: start point for shearing (i.c.)
0.27
: critical state point
0.32
points
emax critical
estimated by C.S.lines
RD=Dmax / Dmin
Fig. 12 Paths followed in void ratio: grading plane in particle
removal analyses
11
123
12
a surface of indefinite extent along the mean stress direction: at least for the stress range of relevance to a particular
application, mean stress is assumed not to lead to particle
crushing.
Where the dominant phenomenon leading to change of
grading is particle crushing then the surface is limited by
the interaction of mean stress and grading index. As the
stress level increases then this will induce crushing and
hence change in grading. We may still suppose (for simplicity and in the absence of experimental data) that the
change in slope of the critical state line is small as IG
increases. However, as mean stress increases (beyond some
threshold) the smaller particles will crush and IG will
increase. The critical state surface thus cannot exist for
stress levels above some limit which will depend on the
current value of IG. An attempt to show the three-dimensional nature of this critical state surface is shown in
Fig. 14b. The three-part critical state line in Fig. 13 is then
understood to be a track across this surface. The distinction
is important because the presentation of a three-stage single
123
6 Conclusion
The recognition that in some circumstances the soils that
we are using or encountering are changing in their nature
during their use is obviously important for the development
of constitutive models. An analogy could be drawn with the
modelling of the effects of cementation or bonding
between soil particles which may have been created by
chemical effects over the geological history of the hard
soil/weak rock but which can be destroyed by mechanical
or chemical means. A number of models have been
developed to incorporate these destructuring phenomena
13
References
1. Been K, Jefferies MG (1985) A state parameter for sands. Geotechnique 35(2):99112
2. Been K, Jefferies MG (1986) Discussion: a state parameter for
sands. Geotechnique 36(1):127132
3. Calvetti F, Tamagnini C, Viggiani G (2002) On the incremental
behaviour of granular soils. In: Pande G, Pietruszczak S (eds)
Proceedings of NUMOG. Swets and Zeitlinger, Lisse VIII,
pp 39
4. Calvetti F, Combe G, Lanier J (1997) Experimental micromechanical analysis of a 2D granular material: relation between
structure evolution and loading path. Mech Cohesive-Frictional
Mater 2(1):121163
5. Cheng YP, Bolton MD, Nakata Y (2005) Grain crushing and
critical states observed in DEM simulations. In: Garca-Rojo R,
Herrmann HJ, McNamara S (eds) Powders and grains, vol 2.
Taylor & Francis Group, London, pp13931397
6. Cheng YP, White DJ, Bowman ET, Bolton MD, Soga K (2001)
The observation of soil microstructure under load. In: Kishino Y
(ed) Powders & Grains. Balkema, pp 6972
7. Coop MR, Sorensen KK, Bodas Freitas T, Georgoutsos G (2004)
Particle breakage during shearing of a carbonate sand. Geotechnique 54(3):157163
8. Gajo A, Muir Wood D (1999) A kinematic hardening constitutive
model for sands: the multiaxial formulation. Int J Numer Anal
Methods Geomech 23:925965
123
14
9. Gajo A, Muir Wood D (1999) Severn-Trent sand: a kinematic
hardening constitutive model for sands: the q-p formulation.
Geotechnique 49(5):595614
10. Hardin BO (1985) Crushing of soil particles. J Geotech Eng Proc
ASCE 111(10):11771192
11. Hardin BO (1987) 1-D strain in normally consolidated cohesionless soils. J Geotech Eng ASCE 113(12):14491467
12. Jefferies MJ (1993) Nor-sand: a simple critical state model for
sand. Geotechnique 43(1):91103
13. Lade PV, Liggio CD, Yamamuro JA (1998) Effects of non-plastic
fines on minimum and maximum void ratios of sand. Geotech
Test J 21(4):336347
14. Luzzani L, Coop MR (2002) On the relationship between particle
breakage and the critical state of sands. Soils and Found
42(2):7182
15. McDowell GR, Bolton MD, Robertson D (1996) The fractal
crushing of granular materials. J Mech Phys Solids 44(12):2079
2102
16. Maeda K, Hara Y, Ohno R (2003) Interaction between pile with
different skin roughness and granular ground by DEM. Computational Plasticity 7, 2003, CD-ROM
17. Maeda K, Hirabayashi H (2006) Influence of grain properties on
macro mechanical behaviours of granular media by DEM. J Appl
Mech JSCE 9:623630
18. Maeda K, Hirabayashi H (2006) Micromechanical influence of
grain properties on deformation-failure of granular media by
DEM. In: Proceedings of International symposium on geotechnics of particulate media (IS-Yamaguchi), pp 173179
19. Muir Wood D (2006) 20th Bjerrum Lecture: the magic of sands.
Norsk Geoteknisk Forening, Oslo
20. Nougier-Lehon C, Vincens E, Cambou B (2005) Structural
changes in granular materials: the case of irregular polygonal
particles. Int J Solids Struct 42(2425):63566375
123