Relationships between desiccation cracking

Geomechanics from Micro to Macro – Soga et al. (Eds)
© 2015 Taylor & Francis Group, London, ISBN 978-1-138-02707-7
Relationships between desiccation cracking behavior and microstructure
of the Tournemire clay-rock by coupling DIC and SEM methods
A-L. Fauchille, S. Hedan, D. Prêt & P. Cosenza
IC2MP–HydrASA, CNRS, Université de Poitiers, ENSIP Poitiers, France
V. Valle
Institut PPRIME CNRS, Université de Poitiers, Futuroscope Chasseneuil, France
J. Cabrera
DEI-SARG Department, Institute for Radiological Protection and Nuclear Safety,
Fontenay-aux-Roses, France
ABSTRACT: The study presented here aims to identify the microstructural factors which govern the cracking
of Tournemire clay-rock caused by humidity changes. A new experimental setup was designed in laboratory
to compare the location of desiccation cracks, induced by humidity changes, to the microstructure of a sample
(20 × 20 × 20 mm3 ) of Tournemire clay-rock. The location of desiccation cracks was measured on a surface
of 5.6 × 4.2 mm2 by Digital Image Correlation (DIC) and the microstructure was quantitatively estimated by
Scanning Electron Microscopy on the same surface. First results show that at millimeter scale the desiccation
cracks are divided in two parallel networks: the first network is composed of continuous macrocracks located
next to local mineralogical heterogeneities such as the proportion and the mean size of grains, and the second
network of discontinuous microcracks at the boundaries between grain/grain, grain/clay matrix or in clay matrix.
1
INTRODUCTION
Thanks to mechanical and textural properties, the clayrocks are an interesting medium to nuclear waste
disposal at great depth. An important cracking due
to a desaturation process of the argillaceous medium
was observed on the gallery walls of the Underground
Laboratory of Tournemire (Aveyron, France). When
the galleries are excavated and operated, air supplying
implies humidity changes and participates to the extension of the Excavation Damaged Zone (EDZ), in which
the hydric and mechanical properties of clay-rocks are
modified. On the front and the walls of galleries, the
cracks are organized in two networks : the principal
network is sub-horizontal in the same direction that
bedding planes and the second is vertical opposed to
bedding planes. Their opening and their aperture are
principally driven by relative humidity in the gallery
(Hedan et al. 2014).
At present, the mechanical and geological factors which control this fracturing are poorly understood. The study presented here aims to identify
the relationship between the location of desiccation
cracks and the microstructure of the rock. A sample (20 × 20 × 20 mm3 ) was submitted a cycle of
shrinkage/swelling/shrinkage under controlled humidity changes (RH = 98-33-98-33%). One face was
filmed by a camera to follow the evolution of the cracks
by DIC. After humidity variations, the sample was analyzed by SEM to obtain the mineral map on the same
surface to compare it with the location of desiccation
cracks.
2
GEOLOGICAL SETTING AND SAMPLING
The Tournemire Underground Laboratory (URL) of
the French Institute for Radioprotection and Nuclear
Safety (IRSN) is located in a Mesozoic marine basin on
the southern limit of the Causse du Larzac. The sample used in this study is a cub of 20 × 20 × 20 mm3
extracted from the drill FD90 five meters deep in
the East96 gallery. It comes from decimeter samples (d = 80 mm, L = 200 mm) between 4.20 and 4.40
meter, where the rock is considered as saturated and
kept away outside from EDZ. The mineralogical composition of the rock shows that clay minerals represent about 20–50 wt% (kaolinite, illite, interstratified
illite/smectite,), quartz about 10–20 wt %, carbonates about 10–30 wt% (calcite, dolomite, siderite), and
sulphides 2–7 wt% (pyrites) (Cabrera et al. 2001).
One face of the sample is finely polished to avoid
topographic artefacts for SEM imaging and to have a
good contrast for DIC.
1421
Figure 2. Extract from a DIC image (on the left) and the
corresponding mineral map (on the right) for a region of
interest.
Figure 1. Experimental setup.
Table 1. Humidity steps applied
to the sample.
3
3.1
Desiccation 1
98 → 33%
Hydration
Desiccation 2
33 → 98%
98 → 33%
EXPERIMENTAL SETUP AND METHODS
Experimental setup
physical transformation as mechanical or hydric loadings. The images are composed of pixels and each
one is encoded by a grey level between 0 and 255
(8 bits). The method is based on the comparison of
small domains (subsets) of grey levels between two
images, in order to obtain the local displacement fields
U which are estimated thanks to the positional changes
of the grey level pattern in the subsets. The speckle
pattern is usually artificial to have good variations of
contrast but in this study, the speckle is the natural
surface of the sample. The software used was XCorrel (H-DIC) developed by Pprime Institute of Poitiers
(V. Valle et al. 2014). It allows the detection of displacement jumps which characterize the location of
desiccation cracks on the sample surface. The domains
are 32 × 32 pixels (70.4 × 70.4 µm2 ) with a precision
of 0.1 pixel (0.22 µm).
The experimental setup is composed of a plastic
waterproof box in which humidity and temperature
conditions are controlled. Relative humidity is controlled by saline solutions and temperature by an
air-conditioning unit at 23◦ C. The box contains a RH/T
(relative humidity/temperature) sensor, a precision
balance and a stirred saline solution. The solution stirring is controlled by a programmable electric plug and
imposed for 2 min over 15 min to preserve a constant
temperature inside. A camera of 5 Mp (2560 × 1920
pixels) takes images (0.5image.min−1 ) with a spatial
resolution of 2.2 µm.pixel−1 . A spot of 400 W lights
up the sample while the camera taking an image. The
sample was saturated for two months with a relative
humidity of 98%. Then, it was submitted to a cycle
of desiccation/hydration cycle between 98 and 33%
of humidity during 55 days (Table 1). After the desiccation 2, the sample was analyzed under SEM with
back-scattered electron to obtain the mineral map on
the same surface.
3.2.2 Scanning Electron Microscopy (SEM)
After the desiccation 2 (table 2), the surface which
was recorded by DIC was analyzed under SEM with
back-scattered electrons (BSE). This type of analysis is sensitive to the mean atomic number encoded
by grey levels. A mineral map of the surface of
5.6 × 4.2 mm2 was investigated by acquiring a mosaic
image of 8806 × 6606 pixels with a resolution of
0.625 µm.pixel−1 . The drift of the electron beam was
estimated by measuring displacement, rotation, deformation and grey levels changes of BSE images. The
displacement and the grey levels of the SEM images
were corrected and the deformation and rotation were
considered as insignificant. The precision of the superimposition between displacement fields from DIC and
the SEM mosaic image is ±2 pixels (4.4 µm).The software µPHASEmap developed in IC2MP laboratory in
Poitiers allows to threshold clay matrix and non-clay
grains (Prêt et al. 2010 a,b) (Figure 2).
3.2
4
Methods
3.2.1 Digital Image Correlation (DIC)
Optical full field measurement methods are more and
more used in experimental mechanics. These techniques have many advantages like the non-destruction
of the samples and the full-field data. DIC method
was chosen in this study among these methods. It
allows the measurement of displacement and strain
fields, the detection of heterogeneities like cracks or
strain gradients due to contrasts of different mechanical properties. It consists in recording images during a
4.1
RESULTS AND COMMENTS
Cracking organization
On the mineral map, there are many cracks corresponded to real desiccation cracks and cracks formed
after or before the hydric loading. DIC method allows
the selection of only working desiccation cracks during
the experiment.
The desiccation cracks of the sample are divided
in two networks. By DIC mapping of local displacements between the initialstate and the final state of
1422
Figure 3. Ux and Uy displacement fieldsat the end of the
desaturation 1.
Figure 4. a) Uy displacement field at the end of desaturation 1; b) mineralogy map from SEM, corresponding to the
black rectangle zone on figure 3.
the sample, a first network (network I) of cracks is
composed of three principal macrocracks in the same
direction of bedding planes (Figure 3). There is no
vertical macrocracks, in comparison with the gallery
fractures (Hedan et al. 2012; 2014). The cracks I are
continuous and spaced out of 1 mm.
The second network (network II) is composed of
microcracks between the macrocracks I at the interfaces of grain/grain, grain/clay matrix or in clay
matrix (Figure 4). By BSE imaging, they seem to be
discontinuous.
These observations about the network II confirm
previous works (Montes et al. 2004) on argillite
samples from the Meuse Haute Marne Underground
Research Laboratory (MHM-URL) in Eastern France
obtain only at small scale by ESEM.
4.2 Relationships between macrocracks
and microstructure
So as to compare the crack location to the microstructure, some textural parameters were calculated from
the mineral map, with the softwares ImageJ and Visual
Studio C++ in same local subsets that DIC domains
(70.4 × 70.4 µm2 ) used previously.
The parameters are: the local proportion of grains
Pg (example: Pg = 60%: there are 60 percent of nonclay grains in the domain), the local mean size of nonclay grains S (pixels), the local mean orientation of
grains α (degrees), the local mean ratio length/width
of grains L/h (≥1) and the local number of grains N.
The textural parameters are presented as 2D maps,
which are compared to the crack location from DIC
(lines on figure 5). The local textural heterogeneities
are surrounded on the maps.
Figure 5. Superimposition between crack I location and
textural parameter maps for Pg, S, α, L/h and N.
The first results show that the location of cracks I
is influenced by mineralogical heterogeneities especially by Pg and S. The cracks are located at the
interfaces between a local heterogeneity and a more
homogeneous zone (zones with Pg < 0.3 and S < 700
pixels, Figure 5).
1423
The sole parameters Pg and S do not explain the
complete location of desiccation cracks. It is necessary to consider α, L/h and N to understand it. When
all the parameters are combined on a same map, a
close relationship between location crack and mineral
heterogeneities is displayed.
5
ACKNOWLEDGMENTS
The authors wish to thank especially Pascal Touvenet,
Pascal Rogeon, Claude Laforest and FrédéricLimousin
in ENSIP, and NEEDS (Nucléaire: Énergie, Environnement, Déchets, Société) project for supporting this
research work.
CONCLUSION
REFERENCES
The new experimental setup investigated in this study
allowed the coupling between DIC and SEM methods in order to compare the location of desiccation
cracks with the microstructure of the argillaceous
rock of Tournemire at millimeter scale. The previous
research works on the comparison between mineralogy and fracturing were generally done on small areas
(<400 × 400 µm2 ), due to the classical field of view
analyzed by SEM and ESEM. In this study, the comparison was done on a surface of 5.6 × 4.2 mm2 , nearly
20 times as much as wider than the previous areas
with high resolution under SEM. During a dry/wet
cycle, the desiccation cracks were organized in two
networks at two different scales. The network I is
composed of continuous macro-cracks in the same
direction that bedding planes. The second network is
composed of discontinuous microcracks at the interfaces grain/grain, grain/matrix or in the clay matrix,
between macrocracks I. There is a close relationship
between the location of desiccation cracks I and mineralogical heterogeneities, including local proportion
of grains and clay matric or mean size of grains,
which constitute local mechanical incompatibilities.
The study of the desiccation cracking was completed
at the meso-scale (millimeter scale) and that permits
to shed light on the detection of local mineralogical
heterogeneities which are very difficult to identify at
microscale (micrometer scale) or global scale (meter
scale) due to resolution limitations.
At the end, one of the objectives of this current work
is the comparison between the evolution of the aperture/closure of cracks I to the relative humidity during
the hydric loading. This comparison should provide
physical insights into the understanding of the transient desiccation cracking behaviour observed by DIC
in the galleries at the Tournemire experimental station.
(Hedanet al. 2014).
Cabrera J., Beaucaire C., Bruno G., De Windt L., Genty A.,
Ramambasoa N., Rejeb A., Savoye S., Volant P. 2001.
Le projet Tournemire comme support de l’expertise sur
le stockage profound en milieu argileux : synthèse des
programmes de recherche, rapport IRSN/SERGD 01-19
Hedan, S., Cosenza, P., Valle, V., Dudoignon, P., Fauchille,
A. L., Cabrera, J. 2012. Investigation of the damage
induced by desiccation and heating of tournemire argillite
using digital image correlation. Int. J. of Rock Mech. and
Min. Sciences 51: 64–75.
Hedan S., Fauchille AL., Valle V., Cabrera J., Cosenza P., 2014
One-year monitoring of desiccation cracks in Tournemireargillite using Digital Image Correlation, accepted in
Int. J. of Rock Mech and Min Sciences.
Montes H.G., Duplay J., Martinez L., Escoffier S.,
Rousset D. 2004. Structural modifications of CallovoOxfordianargilite under hydration/dehydration conditions, Applied Clay Science 25: 187–194
Prêt D., Sammartino S., Beaufort D., Meunier A., Michot L.
2010. A new method for quantitative petrographybased
on image processing of chemical element maps : I. Mineral mapping applied to compacted bentonites, American
Mineralogist, 95: 1379–1388.
Prêt D., Sammartino S., Beaufort D., Fialin M., Sardini P.,
Cosenza P., Meunier A. 2010. A new method for quantitative petrography based on image processing of chemical
element maps : II. Semi-quantitative porosity mapping
superimposed on mineral map, American Mineralogist,
95: 1389–1398.
Valle V., HedanS., Cosenza P., Fauchille AL. 2014. Digital
Image Correlation development for the study of materials
including multiple crossing cracks, in review in ExpMech
Wang L. 2012. Micromechanical experimental investigation and modeling of strain and damage of argillaceous
rocks under combined hydric and mechanicals loads, PhD
thesis, EcolePolytechnique.
1424