You are on page 1of 8

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/258741670

Identifying and following particle-to-particle contacts in real granular media:


An experimental challenge

Article · June 2013


DOI: 10.1063/1.4811868

CITATIONS READS

11 264

4 authors:

Gioacchino Viggiani Edward Andò


Université Grenoble Alpes French National Centre for Scientific Research
170 PUBLICATIONS   4,152 CITATIONS    112 PUBLICATIONS   1,140 CITATIONS   

SEE PROFILE SEE PROFILE

Clara Jaquet Hugues Talbot


Institut Supérieur de BioSciences de Paris CentraleSupélec
4 PUBLICATIONS   29 CITATIONS    214 PUBLICATIONS   3,406 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Geomechanical Modelling of Inception and Propagation of Compaction Bands in Porous Rocks View project

Nanoparticles interaction with cell membrane View project

All content following this page was uploaded by Gioacchino Viggiani on 05 May 2014.

The user has requested enhancement of the downloaded file.


Identifying and following particle-to-particle contacts in real granular media:
An experimental challenge
Gioacchino Viggiani, Edward Andò, Clara Jaquet, and Hugues Talbot

Citation: AIP Conf. Proc. 1542, 60 (2013); doi: 10.1063/1.4811868


View online: http://dx.doi.org/10.1063/1.4811868
View Table of Contents: http://proceedings.aip.org/dbt/dbt.jsp?KEY=APCPCS&Volume=1542&Issue=1
Published by the AIP Publishing LLC.

Additional information on AIP Conf. Proc.


Journal Homepage: http://proceedings.aip.org/
Journal Information: http://proceedings.aip.org/about/about_the_proceedings
Top downloads: http://proceedings.aip.org/dbt/most_downloaded.jsp?KEY=APCPCS
Information for Authors: http://proceedings.aip.org/authors/information_for_authors

Downloaded 08 Jul 2013 to 111.118.216.99. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://proceedings.aip.org/about/rights_permissions
Identifying and following particle-to-particle contacts in real
granular media: an experimental challenge
Gioacchino Viggiani∗ , Edward Andò∗ , Clara Jaquet† and Hugues Talbot∗∗

Grenoble-INP / UJF-Grenoble 1 / CNRS UMR 5521, Laboratoire 3SR, Grenoble, France
† Université Paris Est Créteil, Institut Supérieur de BioSiences, Créteil, France
∗∗ Université Paris-Est, Laboratoire d’Informatique Gaspard-Monge, Equipe A3SI, ESIEE, Noisy-le-Grand, France

Abstract. In the experimental study of the mechanics of granular media, x-ray tomography combined with 3D image analysis
is finally opening the way for studies at the grain scale. Recent work shows that all the grains in a specimen containing several
tens of thousands of grains can be successfully followed throughout deformation, allowing micro-mechanical interpretation
of behaviour observed at the macro-scale. There are practical issues when trying to measure grain-to-grain contacts (i.e.,
identification of contacts and measurement of their orientation) from the same tomography images, however, these are an
essential ingredient for a more profound understanding of the micro-mechanics of deformation.
In this paper, we present a short selection of “mature” grain-scale measurements and then discuss the challenges associated
with grain-to-grain contacts. Results from ongoing work show that very advanced image analysis techniques can allow this
measurement to be made successfully, even when grains surfaces are not imaged at high resolution.
Keywords: experimental micro-mechanics, granular materials, in-situ x-ray tomography, image analysis, inter-particle contacts
PACS: 62.20.F, 07.05.Pj, 81.70.Tx

INTRODUCTION AND MOTIVATION Figure 2 illustrates the sorts of grain-based measure-


ments possible over different increments of a test on a
In previous work, x-ray tomography has been used to dense specimen Caicos ooids (a rounded carbonate sand
record 3D images of small specimens of sand during with a D50 of 420 μm). Grains are tracked over all in-
loading. Triaxial compression tests on dense specimens crements; in Figure 2, results are shown for increments
of dry sand are performed inside the x-ray scanner in 01-02 (beginning of test), 03-04 (before peak), 05-06
Laboratoire 3SR allowing specimens to be scanned at (up to the peak) and 16-17 (in the residual stress state).
key points during the test – see Figure 1. A 3D image of Slices through the 3D image of the specimen have been
x-ray attenuation in the scanned domain is obtained each extracted and are oriented in such a way that they con-
time the specimen is scanned. In order to be able to image tain both the axis of the specimen and the normal to the
each individual grain, specimen size is considerably less
than standard (22 mm height by 11 mm diameter). A
small specimen benefits from the zoom that the x-ray Detector
cone beam provides, resulting in images that contain
X-ray source
several tens of thousands of grains, with around 5000
voxels per grain at 15μm per pixel for grains with a D50 Triaxial cell

of around 400μm. Further details on the experimental


apparatus and procedures can be found in [3]. Specimen

A variety of image analysis techniques have been de- Rotation stage


veloped to identify and measure individual grains in Worm drive for
these 3D images. Very shortly, these require the reliable trolley translation

identification of the inside of each grain by a marker;


markers are then expanded to occupy all the grain vol-
Rails
ume by a watershed algorithm. When these procedures
are applied to several different images from a test, ID-
Track [1] is used to follow grains. This paves the way Trolley Loading system
for measurements of the 3D displacement (by following
each grain’s centre of mass). The measurement of each FIGURE 1. Labelled photograph of the Laboratoire 3SR x-
grain’s 3D rotation presents more of a challenge, which ray scanner, with background faded out for clarity. The blue
has been met by developing a hybrid ID-Track and Digi- edges visible left and right are the door frame of the cabin
tal Image Correlation method [2].

Powders and Grains 2013


AIP Conf. Proc. 1542, 60-65 (2013); doi: 10.1063/1.4811868
© 2013 AIP Publishing LLC 978-0-7354-1166-1/$30.00

60

Downloaded 08 Jul 2013 to 111.118.216.99. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://proceedings.aip.org/about/rights_permissions
shear band that eventually forms. Grains in Figure 2 are yond just grain kinematics. In particular, a natural evolu-
coloured by their vertical displacement measured over tion is to include the characterisation of contact kinemat-
the increment shown (top), and by the intensity of their ics (i.e., the gain or loss of contacts, and the orientation of
3D rotation as measured by the hybrid technique cited those contacts) during deformation – as suggested by [6]
above (bottom). for 2D experiments, and as is currently done in Discrete
0102 0304 0506 1617
Element simulations.
This paper aims to show some recent work in the direc-
Vertical Displacement (mm)

0
tion of achieving the ambitious goal of measuring contact
kinematics on x-ray images like those analysed in Fig-
ure 2. First the detection of contacts between particles
is discussed, and is followed by a discussion of the con-
siderable challenges in making precise measurements of
0.22
contact orientation. The errors associated with classical
6
7
4
5 7
8 particle-separation techniques are studied, and a promis-
Stress Ratio R

6
9
ing new technique allowing this measurement with high
(1 / 3 )

5 3 10
11 12
4
15 16
17 18 precision is detailed, tested on various types of data and
2
3 finally used to make measurements on “real” materials.
2
1
1
Volumetric
Strain (%)

-3

0
GRAIN-TO-GRAIN CONTACTS
0 5 10 15
Normalised Axial Shortening (%)
In the work described above, the material has been im-
30
aged at a resolution sufficient to resolve individual grains
Intensity of Rotation (°)

but not the details of the surface of each grain – this is


visible in Figure 3b, which shows the voxelated surface
of a single Caicos ooid. When characterising the contact
0
between two such objects, the lack of detail at each con-
0102 0304 0506 1617 tact surface (highlighted in the examples shown in Fig-
ures 3a and 3c) causes some measurement problems that
are discussed in this section.
FIGURE 2. Slices showing grains of a triaxial test on Caicos
ooids at 100 kPa confinement coloured by their vertical dis- a) Glass ballotini b) Caicos ooid c) Hostun sand
placement (top) and intensity of 3D rotation (bottom). The in- Contacting grains 2D A single 3D grain Contacting grains 2D
crements studied are highlighted on the stress-ratio and volu-
metric strain vs. axial shortening in the middle of the figure
The first and last increments show fully-developed
mechanisms of deformation: the first increment shows
no grain rotation and a symmetrical and uniform distri-
bution of grain displacements (note: zero displacement
is mechanically imposed at the top of the specimen). The
last increment shows a fully-developed shear band with a
thin band of around six grains with very intense rotations.
The middle increments show a transition mechanism be-
tween uniform deformation and fully localised strain: in Contacting grains 3D Contacting grains 3D

increment 03-04 the displacement field has started to tilt,


FIGURE 3. a) 2D and 3D views of two Glass ballotini in
and some very widely distributed rotations can be seen in contact (these particles are close to spheres); b) 3D view of
the vicinity of the zone where the shear band finally ap- a single Caicos ooid (courtesy of J. Andrade); c) 2D and 3D
pears. In increment 05-06 both fields show signs of con- views of two Hostun sand grains in contact
centration, and the rotations show some chains of grains
rotating together aligned normal to the shear band.
These grain-based measurement tools are extremely
powerful in that they allow grain-scale explanations of Contact identification
macro-scale responses (e.g., [2, 4]). The complex pat-
terns of rotating grains visible in Figure 2 in the incre- When the interconnected solid phase of a 3D image
ments before the peak call for a richer analysis going be- is to be split into individual grains, a separation surface

61

Downloaded 08 Jul 2013 to 111.118.216.99. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://proceedings.aip.org/about/rights_permissions
between each grain must in some way be defined. This is Contact orientation
typically obtained using a classical watershed algorithm
(see, for example, [12, 16]). If two objects are split by Identifying the presence of a contact and measuring
such a procedure, this is an indication that they can be its “area” is not sufficient to fully characterise the way
said to be in contact to a degree of accuracy which is in which two particles touch; what is missing is the ori-
a function of the resolution of observation (taking into entation of the contact plane. The most natural approach
account the partial volume effect in the voxels concerned to obtain this is simply to fit a plane through the voxels
by the contact). It is thus tempting to consider the voxels defining the separation surface coming from the water-
defining a separation surface directly as a volumetric shed procedure.
representation (of unit thickness) of a contact. The computed orientation as an (x, y, z) unit vector can
Figures 3a and 3c underline how small the contact ar- be projected onto the unit z-positive half-sphere, since we
eas can be in these images (i.e., how few voxels are used cannot distinguish between either normals to the plane.
in the definition of the separation surface). Although To represent this projection two numbers are sufficient,
small, contacts can effectively be identified, and their equivalent to a longitude and latitude. To render these
size measured. As an example, Figure 4 shows the dis- projections in print, the Lambert polar equal-area projec-
tribution of contact “areas”, for four different specimens tion [13] is used.
(each comprised of around fifty thousand grains, and im-
aged under an isotropic stress of 100 kPa). The four dif- 90°
ferent materials imaged are: Hostun sand (very angular),
135° 45°
Ottawa sand (sub-rounded), Caicos ooids (very rounded)
and glass ballotini (almost spherical).

1.00
Hostun sand 0° 45° 90°
180° 0°
Ottawa sand

0.75 Caicos ooids


Relative Frequency

Glass ballotini

0.50
225° 315°

270°
0.25

FIGURE 5. Equal area stereographic projection of contact


orientations for a subvolume of around one thousand Caicos
0.00
0 50 100 150 200 ooids as measured from a standard topological watershed [8]
Contact area (number of voxels) (projected using a Lambert equal-area projection)
Figure 5 shows the distribution of orientations of
FIGURE 4. Histogram showing the distribution of contact
sizes in specimens of Hostun sand, Ottawa sand, Caicos ooids around four thousand contacts calculated for a subvol-
and glass ballotini, calculated with a bin width of 5 voxels ume of around a thousand Caicos ooids. If orientations
are identically and independently distributed on the unit
As suggested by the images in Figure 3, and confirmed half-sphere, this projection yields a point distribution of
in Figure 4, the size of the most frequent “area” of uniform density. In the case of Figure 5, however, it is
contact is strongly correlated with roundness: Hostun clear that there is a striking amount of bias in the 45◦ di-
sand tends to have small areas – likely to represent point- rections. As shown in [2] for this same subvolume, differ-
to-plane contacts. As grain shapes get more rounded, ent implementations of the watershed procedure give dif-
this configuration gets increasingly less probable and ferently biassed results. Since there is no a-priori knowl-
faces rather than points are more likely to be in contact; edge of the actual distribution of contact orientations in
these appear to be larger contact areas at the resolution this granular assembly, it is impossible to assess the ac-
of observation. Furthermore, when the “tail-end” of the curacy of these measurements.
distributions is studied, large contacts are slightly more In the following, this problem is studied more closely
probable for Hostun than the other materials, since face- using synthetic data rather than images from x-ray to-
to-face contacts are possible (but not likely) between mography. A series of around four thousand binary 3D
flat faces, yielding large areas. In summary, Figure 4 spheres (radius 10 pixels) are placed in random orienta-
shows that the areas in contact between grains are better tions with their centres 20 voxels apart. Spheres are used
distributed for more angular materials. because they have the unique property that the vector

62

Downloaded 08 Jul 2013 to 111.118.216.99. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://proceedings.aip.org/about/rights_permissions
90° 90° tershed), has been developed to better define the sepa-
135° 45° 135° 45° ration plane between two binary objects. The technical
details are outside the scope of this paper, and can be
0° 45° 90° 0° 45° 90°
found in [14]. In essence, starting from markers well in-
180° 180°
0° 0°
side the grains, the Power watershed algorithm (as intro-
duced by [9]) assigns for each voxel of the solid phase
225° 315° 225° 315°
the probability of belonging to either marker, which for
270° 270°
the settings used corresponds to the energy of a Ran-
a) Branch Vectors b) Standard Watershed 1 dom Walker. The contact plane is then oriented using the
(Ground Truth) (Topological watershed) voxels representing the “contested plateaus” between the
90° 90°
two markers (i.e., the voxels which are either side of the
135° 45° 135° 45°
0.5 probability isosurface). More precisely, the contact
plane is defined with sub-voxel precision by estimating
the equal-probability isosurface between the two parti-
180°
0° 45° 90°
0° 180°
0° 45° 90°
0° cles in contact. The drastic improvement in the measure-
ment of orientations for the set of simulated spheres is
clearly visible in Figure 6d. These high-quality measure-
225° 315° 225° 315°
ments, however, have a considerable computational cost.
270° 270° In fact, the technique is currently implemented on a per-
c) Standard Watershed 2 d) Proposed Method
(Meyer Algorithm) (Power Watershed)
contact basis (i.e., treating one pair of contacting grains
at a time); the calculation of the Power watershed, which
FIGURE 6. Equal area stereographic projection of contact is the bottleneck in this computation, causes the calcu-
orientations from simulated spheres: a) orientation of branch lation to take a few seconds per contact on a powerful
vectors linking centres of spheres; b) contact orientations ob- desktop machine.
tained with a topological watershed; c) contact orientations ob- Given the very satisfactory performance of the new
tained with Meyer’s algorithm; d) contact orientations obtained
with the Power watershed based technique proposed (after [14])
measurement tool based on Power watershed on syn-
thetic images, its performance was then evaluated on
an x-ray tomography of a real granular assembly made
of glass ballotini, many of which are close to perfect
connecting their centres (“branch vector”) has the same spheres. This allows the same verification as above to
orientation as the normal to the contact surface between be performed. In this case, the branch vector provides a
the spheres, which provides a “ground truth” contact ori- good estimation of the expected contact orientation. In
entation. Figure 6a shows the distribution of orientations order to exclude the “defective” ballotini, 4.7% of those
for this series of spheres obtained from their branch vec- whose surface area to volume ratio is furthest from a
tors, which is uniform overall (note that a small discrete sphere are excluded. Figure 7 shows the distribution of
bias is introduced because spheres are only simulated
with centres located on a voxel, therefore there is only
a finite number of possible orientations). 1.00
Figure 6b shows the orientations of the separation
planes obtained with a standard topological watershed.
0.75
The distribution of the orientations is clearly different
Relative frequency

from the ground truth: the topological watershed intro-


duces extreme artefacts into the measured orientations, 0.50

aligned at 45◦ to the coordinate system, as already no-


ticed on Caicos ooids in Figure 5. Figure 6c shows the
0.25
orientations of the separation planes obtained with an-
other watershed procedure (the Meyer algorithm [15]),
in which the separation plane between the particles is in- 0.00
0 10 20 30 40 50 60 70 80 90
terpixel (i.e., it does not have unit thickness). Although Absolute relative error (degrees)
the measurements are less biassed than with the topolog-
ical watershed, the accuracy remains poor. FIGURE 7. Glass ballotini: frequency distribution of the ab-
Having confirmed that the watershed procedure is re- solute angular deviation between the orientation of the branch
sponsible for introducing the bias in the measured con- vectors and the orientation of the contact as measured by the
tact orientations, a new technique (based on Power wa- Power watershed technique

63

Downloaded 08 Jul 2013 to 111.118.216.99. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://proceedings.aip.org/about/rights_permissions
the absolute angular deviation between branch vector and from the centre of the circle, and the number of angular
contact orientation, as measured by the Power watershed bins in θ at a given bin distance is: 4(2n + 1). The
technique. The most frequent difference between these data shown in the top of Figure 8 is binned using 5
angles is only 4.04◦ . This relatively small error confirms radial bins, giving the segments visible in the bottom
that Power watershed is a reliable method for the mea- of Figure 8. The number of projected points that falls
surement of contact orientations. inside each bin is recorded, and bin totals are normalised
by the median bin value. The colourmap is centred on
1.0 so that bins in white are those in which the median
Contact kinematics number of points has been counted, while red and blue
indicate bins in which more or less than the median
The measurement of contact orientations with the has been counted, respectively. This figure reveals that
topological watershed presented in Figure 5 is repeated there is a small amount of preferential alignment of the
with this new tool in the top left plot of Figure 8. Al- orientations towards the cardinal directions (up, down,
though, as stated above, there is no a-priori knowledge left, right, front, back); this is likely due to extremely
of the orientations expected, the fact that there are no ev- small contact areas where the orientation can only be
ident artefacts is encouraging. The same measurement is poorly defined.
repeated for the subvolume of grains at the end of the The binned images in Figure 8 clearly indicate the
test (in state 17 – see Figure 2), and is shown in the top way in which the distribution of contacts evolves: many
right plot in Figure 8; some indications of a change in the contacts are gained in the x-direction (left and right of
distribution can be seen in this projection. plot) and many are lost in the z-direction (middle of plot)
between configurations COEA01-GL-01 and COEA01-
90° 90°
GL-17. Given that the sub-volume of grains on which
135° 45° 135° 45°
this is calculated becomes part of the shear band that
forms (whose normal is approximately [-1, 0, -1]), this
seems to indicate that grain-to-grain contacts are lost in
180°
0° 45° 90°
0° 180°
0° 45° 90°
0° the direction of principal stress applied to the specimen
(along the axis of compression). Conversely, contacts
are gained in the direction in which the shear band is
225° 315° 225° 315°
advancing.
270° 270°

COEA01-GL-01 COEA01-GL-17
90° 90° CONCLUSION
135° 45° 135° 45°

This paper has shown the type of grain-scale informa-


0° 45° 90° 0° 45° 90°
tion that can be obtained from x-ray tomography images
180° 180°
0° 0°
of real mechanical tests on a representative number of
grains. As seen in previous work (in particular [1, 2]),
225° 315° 225° 315°
high-quality kinematical measurements (although chal-
270° 270°
lenging) are now possible on the vast majority of the
50,000 grains (imaged to have around thirty pixels for a
grain diameter) that make up the specimens tested herein.
0 1 2
The measurements of grain kinematics show complex
patterns of rotating grains in increments up to the peak
FIGURE 8. Equal area stereographic projection of contact
orientations in a subvolume of Caicos ooids. Top: “raw” pro- of the specimen’s global axial stress response; the ulti-
jected orientations for the subvolume at the beginning and at the mate explanation for these patterns must lie in the spatial
end of the test; bottom: same data, but binned (colour represents distribution of inter-granular forces throughout the spec-
the number of projected points per bin divided by the median imen, which is not yet accessible experimentally. How-
number of points in all bins) ever, the characterisation of contacts may shed light on
This change is more visible in the bottom two plots how grains transmit forces.
of Figure 8, where the projected orientations are divided The presence and size of contacting areas has been
into equal-area bins on the plane onto which they have shown to be measurable within certain tolerances,
been projected. Using cylindrical coordinates (r and θ ), whereas the orientation of contacts is much more chal-
the circle in the plane is divided into equally spaced lenging to measure. This is particularly true when grain
bins according to r. Radial bins are labelled n = 0, 1, 2... surfaces are not described at very high resolution, which
is necessarily the case when a significant number of

64

Downloaded 08 Jul 2013 to 111.118.216.99. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://proceedings.aip.org/about/rights_permissions
grains has been imaged. ASTM STP, Vol. 977, pp. 290–310
The “ordinary” approach to the measurement of the 8. Couprie, M., Najman, L., and Bertrand, G. (2005),
orientation of a contact between two grains, imaged at “Quasi-linear algorithms for the topological watershed”,
Journal of Mathematical Imaging and Vision, Vol. 22, No.
this scale, has been shown to introduce extreme bias into
2–3, pp. 231–249
the measurements such that they are no longer usable. 9. Couprie, C., Grady, L., Najman, L., and Talbot, H.
Close collaboration with experts in mathematical mor- (2011), “Power watersheds: A Unifying Graph-Based
phology (co-authors of this paper) has allowed this mea- Optimization Framework”, IEEE Transactions on
surement to be made successfully with a newly devel- Pattern Analysis and Machine Intelligence, Vol. 33, pp.
oped technique based on very advanced algorithms. 1384–1399
The kind of measurements now possible with the tool 10. Fu, P., and Dafalias, Y. F. (2011), “Study of anisotropic
shear strength of granular materials using DEM
presented in this paper are able to detect changes in con- simulation”, International Journal for Numerical and
tact orientation of a subvolume of grains during devi- Analytical Methods in Geomechanics, Vol. 35, No. 10,
atoric loading. The next step is to look individually at pp. 1098–1126
contacts between key grains (i.e., those participating in a 11. Fu, P., and Dafalias, Y. F. (2011), “Fabric evolution within
shear band, or at the heart of the complex patterns pre- shear bands of granular materials and its relation to
ceding the formation of the shear band), and how they critical state theory”, International Journal for Numerical
and Analytical Methods in Geomechanics, Vol. 35. No.
evolve. The ability to supplement grain-based measure- 18, pp. 1918–1948
ments with measurements on contacts between grains is 12. Gonzalez, R. C. and Woods, R. E. (2007), “Digital Image
ideal for the kind of abstract analysis of sand behaviour Processing”, Prentice-Hall, 3rd edition
using techniques from complex networks recently pre- 13. Hinks, A. R. (1921), “Map Projections, 2nd rev. ed.”,
sented in [17]. Cambridge University Press, Cambridge, England
Furthermore, full characterisation of the evolution of 14. Jaquet, C., Andò, E., Viggiani, G., and Talbot, H. (2013),
“Estimation of separating planes between touching 3D
contacts within a granular assembly under load might objects using power watershed”, International Symposium
support and encourage the development of more micro- on Mathematical Morphology, vol. 11, in print
mechanically-based constitutive models for granular ma- 15. Meyer, F. (1991), “Un algorithme optimal de ligne de
terials (in the style of [10, 11]), and help feed these mod- partage des eaux”, Proc. 8ème Congrès Reconnaissance
els with data coming from experiments on real materials. des Formes et Intelligence Artificielle, pp. 847–857
16. Russ, J. C. (1999), “The image processing handbook,
third edition”, CRC Press
17. Tordesillas, A., Walker, D. M., Andò, E., and Viggiani,
REFERENCES G. (2013), “Revisiting localised deformation in sand with
complex systems”, Proceedings of the Royal Society A:
1. Andò, E., Hall, S. A., Viggiani, G., Desrues, J., Mathematical, Physical and Engineering Sciences, Vol.
and Bésuelle, P. (2012a), “Grain-scale experimental 469, No. 2152
investigation of localised deformation in sand: a discrete
particle tracking approach”, Acta Geotechnica, Vol. 7,
No. 1, pp. 1–13
2. Andò, E., Hall, S. A., Viggiani, G., Desrues, J., and
Bésuelle, P. (2012b), “Experimental micromechanics:
grain-scale observation of sand deformation”,
Géotechnique Letters, Vol. 2, No. 3, pp. 107–112
3. Andò, E. (2013), “Experimental investigation of micro-
structural changes in deforming granular media using
x-ray tomography”, Phd Thesis, Université de Grenoble
4. Andò, E., Hall, S. A., Viggiani, G., Desrues, J. (2013),
“Experimental micro-mechanics of granular media
studied by x-ray tomography: recent results and
challenges”, Géotechnique Letters, Submitted
5. Bagi, K. (1996), “Stress and strain in granular
assemblies”, Mechanics of Materials, Vol. 20, pp.
165–177
6. Calvetti, F., Combe, G., and Lanier, J. (1997),
“Experimental micromechanical analysis of a 2D
granular material: relation between structure evolution
and loading path”, Mechanics of Cohesive-Frictional
Materials, Vol. 2, pp. 121–163
7. Colliat-Dangus, J.-L., Desrues, J., and Foray, P. (1988),
“Triaxial testing of granular soil under elevated cell
pressure”, Advanced Triaxial Testing of Soil and Rock,

65
View publication08
stats
Downloaded Jul 2013 to 111.118.216.99. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://proceedings.aip.org/about/rights_permissions

You might also like