Professional Documents
Culture Documents
Alonso Pineda 2006
Alonso Pineda 2006
net/publication/266618329
CITATIONS READS
5 1,131
2 authors, including:
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Jubert Andres Pineda on 09 October 2014.
Abstract
Mechanisms leading to degradation of shales are first described from a qualitative perspective. Stress changes and
suction cycles are stressed as important contributors. The effect of these two effects on the degradation of clay
rocks around tunnels are then described, following recent results of a few large scale experiments performed in
underground rock mechanic laboratories. Also, examples of claystone degradation in excavations exposed to
atmospheric action are reviewed. A number of laboratory experiments, which involve, in general, the application of
loading and suction cycles are presented. They highlight the relevance of extreme suction cycles and the importance
of rock mineral constituents. Constitutive models for bonded soils are aoutlined. The most recent developments
describe the clay rock as a composite of an elastoplastic clay matrix and a brittle damage model for the bond
constituents. The paper ends with the description of a model in which the clay matrix model expands and shrinks
under suction cycles. Then, realistic predictions of weathering effect can be made.
Resumen
Se describen inicialmente, de forma cualitativa, los mecanismos que contribuyen a la degradación de rocas
arcillosas. Se destacan dos mecanismos: los cambios de tensión y los ciclos de succión. Su efecto sobre la
degradación de la roca en el contorno de túneles se ha descrito a partir de unos pocos ensayos a gran escala
realizados en laboratorios subterráneos de mecánica de rocas. Se resumen también algunos casos de degradación de
arcillitas expuestas a la acción atmosférica en excavaciones. Se describen algunos ensayos de laboratorio que
someten a la roca a ciclos de carga descarga y de cambios de succión. Los ciclos extremos de succión y la
composición mineralógica son dos aspectos fundamentales para explicar la degradación. Se presentan varios
modelos constitutivos recientes que describen la roca como un material compuesto. La matriz se representa
mediante un modelo elastoplástico y la cementación mediante un modelo de daño. El artículo termina con la
descripción de un modelo constitutivo que permite la expansión y el hinchamiento de la matriz bajo cambios de
succión, lo que permite hacer predicciones realistas de los efectos de la alteración.
Figure 10.Layout of the Mont Terri laboratory (Thury & Bossart, 1999)
length), (ii) a road header, and, (iii) a pneumatic
Three different excavation methods were used hammer. In all cases, water was not used during
in the new gallery: (i) conventional blasting with excavation to avoid swelling of the rock. The final
electronic detonators (around 70% of tunnel cross section had a horseshoe shape section with
an area of 15m2. In the central part of the new
gallery, a stretch 35m long was excavated using a
road header (about 15%) resulting in a circular
section with an area of 11m2. 15% of the niches
were excavated by diesel-fuel-powered pneumatic
hammer excavators. All excavated sections were
lined using shotcrete (15-20cm of thickness),
reinforced with steel or plastic fibres (Thury &
Bossart, 1999).
Expansion of the tunnel during and after
excavation resulted in the formation of an EDZ.
Oening cracks developed during the niches
excavation (unloading) with parallel orientation to Figure 12.Gypsum spots formed on unloading
the tunnel wall. Martin et al. (2002) reported four joint surface in Opalinus clay (Bossart et al, 2004)
types of stability problems during and after
construction of the tunnels in Mont Terri Based on field measurements at different scales
laboratory: (i) tensile brittle cracking, (ii) stress- Bossat et al (2002) presented a structural model to
induced breakouts, (iii) sliding along the bedding describe the development of the EDZ in galleries
planes, and, (iv) swelling and softening. However, directed NW-SE in the Mont Terri Laboratory.
during construction stage only tensile brittle Two zones are identified in this model (Figure
cracking was observed due to stress concentration 13): an inner zone with an extent ranged from 0.1
developing extension cracks (Figure 11). In some to 1.25m within the tunnel wall. The inner zone
cases cracking zones are accompanied by small consists of interconnected fracture network in
gypsum spots. This zone corresponds to the first unsaturated condition. The outer zone presents an
metre of the tunnel wall in which oxidation takes extent of 2m approximately where fractures are
place due to the air presence (see Figure 12) not interconnected (generally). On the side walls
(Bossart et al. (2004). The fractures observed both unloading fractures are parallel to the tunnel wall
in floor and roof were interpreted as a reactivation (oblique to the bedding planes), whereas at the
of bedding planes or tectonic fault planes. Bossart bottom and top of the galleries fractures are
et al. (2002) reported that fractures extended to oblique to the tunnel wall (parallel to the bedding
depths up to 1.5m from the wall of the planes).
reconnaissance gallery.
Figure 14.Permeability variation around the new gallery (Bossart et al, 2004)
Effect of wetting and drying cycles: VE instrumentation phase, a total of about 100
Ventilation test in Mont Terri Laboratory on boreholes were drilled wherein 24 mini-
Opalinus Clay piezometers, 4 piezometers, 16 capacitive
The test consists in the ventilation of 10m long hygrometers, 16 psychrometric hygrometers, 10
section of a non-lined micro-tunnel, 1.3m in TDR sensors (they measure the rock water
diameter, in the Mont Terri laboratory, (see content via wave travel times) and 8 mini-
Figure 16). The micro-tunnel was excavated in extensometers were placed (see Figure
February of 1999 but the ventilation experiment 19)(Garitte, 2007). A schematic layout of the
(VE) did not started until July of 2002, after a 3 ventilation test section is presented in Figure 20
year long instrumentation phase. During the where two ventilation pipes can be distinguished.
The vapor inflow is driven towards the right part Figure 19 shows the variation in RH,
of the micro-tunnel (section A1) whereas the measured by hygrometers, due to ventilation of
vapor outflow exits at the left part of the the tunnel. Temperature measurements are also
microtunnel (section A4). This scheme shown in the plot. The desaturation stage started
guarantees that the vapor will flow along of the at day 314 when RH decreased in steps from
overall section. Instrumentation was used to 90% to 47% and finally to a limiting value 15%.
evaluate the rock response along the controlled This process lasted for 250 days. Afterwards a
ventilation phases of the test, which lasted from saturation stage followed in which a RH value
July the 18th of 2002 to June the 30th of 2004 close to 95% was applied (for 150 days).
(phase 1) and from July the 1rst of 2005 to June However, the RH measured in the hygrometers
the 30th of 2006 (phase 2). was close to 90%. Then, the vapor inflow was
stopped until an equilibrium condition was
reached inside the micro-tunnel (for one year
approximately). Finally, a desaturation stage was
applied, in which a RH of 15% was again
applied. During all stages, temperature in the
micro-tunnel presents very little variation,
ranging between 15º and 16º (Garitte, 2007).
The layout of the mini-piezometers installed in
sections SA1-4 is shown in Figure 22a. Figure
22b shows the variation of pore water pressure in
section A1 during the 2 stages of the VE test. It
is interesting to note that in most cases pore
water pressure remain at atmospheric condition
during the desaturation stages. This behaviour
was observed in sensors located within a distance
Figure 16.General layout of the Ventilation test of 1.8m of the micro-tunnel wall. The measured
(VE) in Mont Terri Laboratory (Mayor et al, pore pressure at t = 314 days gives a good idea of
2005) the pressure state of the test area at the beginning
of the VE. The pressure state at this time is
mainly governed by the drainage occurred since
the excavation of the micro-tunnel in February
1999 (Garitte, 2007).
Measurements of RH suggest that the
unsaturated area due to ventilation was not
deeper than 25 to 35cm.
Displacements were recorded in mini-
extensometers located at a distance of 2m from
the micro-tunnel wall The displacement
behaviour is consistent with shrinkage and
dilating mechanisms recorded when a soil is
Figure 17.Instrumented section in (VE) dried or wetted, respectively.
experiment in Mont Terri Laboratory (Mayor et 314 560 1079
100
al, 2005) 90
RH-HyV-In
Relative Humidity [%] and
RH-HyCnt-L
80 T-HyCnt-L
Temperature [ºC]
70
60
50
40
30
20
10
0
0 500 1000 1500
Time [days from VE start]
Figure 18.Schematic layout of the Ventilation
system (Mayor et al, 2005) Figure 19.Variation of the RH with time. RH
input vs RH measured (Garitte, 2007)
3 (Figure 21). This mudrock shows an irreversible
Section A1 anisotropic shrinkage due to the application of
2 BVE-78 variable air drying. Swelling was also observed
BVE-81 due to the hydration of the rock minerals.
1
0.5m
Combination of these two factors leads to a fast
disintegration of Beaufort mudrock.
0 MT Engineering geological problems have been
1m reported in sedimentary rocks containing pyrite.
-1 Presence of pyrite has a great influence on the
BVE-80 BVE-79 weathering mechanism leading in some cases to
-2 a rapid instability of rock slopes. (Chigira &
1m
Oyama, 1999). Oyama & Chigira (1999)
-3 presented one case in which chemical weathering
-3 -2 -1 0 1 2 3
due to pyrite oxidation developed in a Japanese
1200
105 314 350 560 713 1079 1534 old tunnel.
252 413 P-B78/2.10
1000 P-B79/2.10
P-B80/1.80
Pressure [kPa]
800 P-B81/1.80
Series13
600 Series12
400
200
0
0
Section A1
200 400 600 800 1000
Time [days from VE start]
1200 1400
Figure 21.Strain behaviour of Beaufort mudrock
under wetting/drying cycles (Olivier, 1990)
Figure 20.Variation of pore water pressure with
time. (a) location of mini-piezometers (b) pore Similar mechanisms of chemical weathering of
water pressures changes (Garitte, 2007) natural slopes have been presented by Chigira
(1990) and Oyama et al. (1992).
2.1.3 Other examples
Some examples of weathering processes in 2.2 Excavations
tunnels have been reported in the literature. The influence of weathering on the mechanical
However, many of them involve a combination behaviour of soils has great importance in many
of physical and chemical weathering processes in aspects of geotechnical engineering. Such
which sometimes is difficult to separate the two properties as shear strength, compressibility,
processes. permeability and coefficient of earth pressure at
Olivier (1990) has shown the influence of the rest are likely to be modified where the
moisture redistribution and the variations of magnitude of the modification depends on the
environmental relative humidity on the type of change of the soil structure produced by
development of weathered zones (EDZ zones) in the weathering process. Some examples of in-
tunnel excavations. He stated that the parameters situ clay shales in which mechanical behaviour
that control weathering and disintegration of has been affected by weathering processes are
mudrocks in underground excavations are: described below.
Variations in the environmental humidity (e.g,
Van Eeckhout, 1976; Harper et al, 1979; Olivier, 2.2.1 Pamplona marls (Spain)
1979; Olivier, 1990; Mayor et al, 2005; Garitte et
al., 2006; Rejeb & Cabrera, 2006) Pamplona marl is an argillaceous rock of North-
Wetting and drying cycles (e.g. Lempp, 1978; ern Spain. This material can be defined as “prob-
Olivier, 1990; Mayor et al, 2005; Garitte, 2007) lematic” rock because of the extreme changes in
Intensive air drying (e.g. Olivier, 1979; behaviour when it is subjected to seasonal varia-
Olivier, 1990). tions. In nature, a well defined weathering front
Olivier presented the strain behaviour of is usually observed. Weathered materials, located
Beaufort mudrock due to the alternating cycles above the front show an average fine content of
of water absorption (immersion of samples in 90%. They are low plasticity clays (CL) with
water) and intensive air drying (RH=55%) liquid limit values ranging between 27% and
50% and a plastic limit between 10.5% and 0
w (%)
Depth (m)
LL (unaltered)
-15 LP (weathered)
30% to 50%.
Pineda et al. (2006) (this Conference) have
-30
presented some experimental results describing Figure 22.Variation in water content and
the degradation of the mechanical properties of Atterberg limits with depth. Pamplona marls.
the Pamplona marls and their influence on the (Mapa Geotécnico del Área de Pamplona, 2002,
foundation stability of the Monreal Dam from Pineda et al, 2006).
(Spain).Typically the depth of the weathering Unconfined shear strenght, qu (Kp/cm )
2
Depth (m)
mainly formed by Illite). Also mica and chlorite -15
80
Sample submerged in distilled
70 water by 5 min prior the shearing of the movement in this zone is characterized by
60
important deformations of the valley floor
50
(swelling) and a process of squeezing alongside
40
the slopes of the hill. In Figure 28b and 28c
water content and pressures profiles obtained
30
20
10
from a borehole drilled in the valley are
0 presented. According to Picarelli et al. (2000) the
0 1 2
400
300
' v =42kPa
200
100
0
0 5 10 15 20 25 30 35 40
2.2.2 Tectonized clay shales of Italy Figure 28.Bisaccia hill: (a) cross section, (b)
water content profile, (c) water pressure profile
The tectonized clay shales of Southern Apenni- (from Picarelli et al, 2000; after Picarelli &
nes are particular materials which present weak Urciuoli, 1993 and from Di Nocera et al, 1995)
Climatic processes may cause permanent
modifications of the geotechnical properties of
hard soils and soft rocks. Cafaro & Cotecchia
(2001) described the mechanical decaying of an
Italian overconsolidated clay due to weathering.
Unweathered clay at Montemesola site is a
highly overconsolidated grey stiff clay, mainly
composed by illite, deposited in a marine
environment during the Pleistocene in the
Montemesola Basin near Taranto. After
consolidation this clay was cemented in the
presence of carbonates during diagenesis
(Cotecchia & Chandler, 1995). Cafaro &
Figure 29.Face of the quarry and samples
Cotecchia investigated the weathering process in
location (Cafaro & Cotecchia, 2001)
a quarry wall where two weathering fronts (one
grey front and one oxidised yellow front) are
The stress sensitivity (S='y/*e) is defined as
identified (Figure 31). Oxidation of the upper
the ratio between the gross-yield stress ('y) and
clay (yellow) down to a depth of 20-25m is
the vertical stress on the ICL corresponding to
attributed to a regional denudation and erosion
the specific volume of the natural clay at gross-
during long time. Weathering of yellow clays
yield stress (*e). For yellow and grey clays the
exposed on the excavation front occurred in the
stress sensitivity values were1.5 and 2.4,
last 10 to 15 years. Mineralogical changes were
respectively. Figure 30 shows the Ko curves for
evaluated on the yellow formations where
natural yellow and grey clays in the void index
calcium and iron sulfates (due to oxidation) were
Iv-log p’ plane (Burland, 1990) in which the
found. In the case of London Clay (a highly
variation in Sis clear. Considering that the
overconsolidated high plasticity clay) Chandler
structure of both clays was defined as originally
& Apted (1988) have shown that mineralogy
similar, the S values for these clays should be
does not change significantly due to weathering.
similar. Nevertheless, the lower Svalue
In Montemesola, a mineralogical comparison
obtained for the yellow clay is due to structural
indicated similar compositions of the yellow and
weakening as a result of weathering whose
grey clays. The only significant change in
mechanical effects are related to the reduction in
composition due to weathering was the partial
the stress sensitivity for yellow clay (from 2.4 to
oxidation of pyrite, which explains the change in
1.5) (Cafaro & Cotecchia, 2001).
color. Cotecchia & Chandler (1995) reported that
the long-term weathering in the Montesemola
Basin was affected by the occurrence of wetting-
drying cycles, due to variations of the table water
level. For the yellow clay, weathering caused
changes in their specific weight (it decreased
from 1.65 for the grey clay to 1.55 for the yellow
clay). Liquidity index changes from LI=-0.05 for
grey clays to LI=-0.19 for yellow clays, This
reduction on LI in the yellow clay is consistent
with a loading process affecting strongly the
upper clays. This loading was attributed to
drying following a previous idea by Cotecchia &
Chandler (1995).
Figure 30.K0 curves for natural yellow and grey
Results of oedometer tests on natural yellow
clay. Influence of the stress sensitivity (Cafaro &
and grey clays shows a gross-yield vertical stress
Cotecchia, 2001)
(Cotecchia & Chandler, 2000) of about 2.0MPa
for the natural yellow clay, whereas for grey clay
Using the idea that yellow clay derives from
the gross-yield stress was about of 2.6MPa. The
the grey clay by a drying process, Cafaro &
influence of the structure on the compression
Cotecchia (2001) performed wetting-drying
behaviour was studied in terms of the stress
cycles on samples of grey clay to cause a
sensitivity Sσ (Burland et al, 1996; Cotecchia &
progressive contraction of the clay and a
Chandler, 2000).
permanent reduction in the specific volume
together with structure degradation. The assumed
drying wetting history under natural conditions is
schematized in Figure 33. During wetting-drying
cycles suction varied between 0kPa (initial
saturated condition) and 9.2MPa. Results show a
permanent reduction of the specific volume at
the same suction value after the second drying
cycle. The evolution of the swelling index
obtained from the oedometer test (Cs) and the
value measured during swelling after drying
(v/log(pk)) were compared to confirm this
hypothesis. Figure 34 shows the variation of Cs
in terms of the maximum effective vertical stress Figure 34.Variation of the swelling index Cs of
’vmax (from oedometer test), and maximum the grey clay obtained from oedometer test and
value of suction reached during drying, pk after drying cycles (Cafaro & Cotecchia, 2001)
(wetting-drying test). It can be seen that the
swelling index obtained from the oedometer test
increases initially faster than the value obtained 2.2.3 Other examples
from drying, which is still increasing once the
stress yield have been reached. Based on this Chandler (1969) examined the influence of the
result, Cafaro & Cotecchia (2001) stated that the weathering processes on the mechanical behav-
evolution from grey into yellow clay (reducing iour of the Keuper Marl, a heavily overconsoli-
the specific volume, degrading the diagenetic dated Triasic deposit (UK). In undisturbed state
bonding and increasing the swell potential) has this material is a mudstone mainly composed of
been induced by in-situ cyclic drying processes. silt sized aggregations with clay minerals parti-
cles. Chandler grouped the Keuper marls in four
groups depending of their degree of weathering:
zone I is the unweathered marl; in zone II weath-
ering begins to form fissures; in zone III a
weathering profile is developed and the weath-
ered matrix remains predominantly silty. Finally,
in zone IV the marl has lost much of the silty
texture of the matrix of less weathered marl, and
as much as 50% of the soil may consist of dis-
crete clay-sized particles.
Figure 33 shows the variation of the index
properties within the weathering profile. Near the
ground surface the liquid limit and natural water
content both increase while the bulk density
decreases. For zones I, II and III the liquid limit,
plastic limit and water content fall slightly with
depth, and the plasticity index remains almost
constant.
Changes in the stress-strain behaviour were
described by Chandler based on drained triaxial
tests on samples with different degrees of
weathering (Figure 34). In Zone I the material
Figure 33.Schematic stress history of natural clay exhibits a brittle failure. However, the curve for
at Montesemola: (1) geological stress state, (2) the marl in Zone IV has no peak, the failure
unloading due to erosion, (3) current stress state being entirely plastic. The suppression of the
for grey clay (lower), (4) drying/wetting cycles, peak in the stress-strain curve is largely
(5) current stress state for yellow clay (Cafaro & responsible for the reduction of the effective
Cotecchia, 2001) strength parameters recorded. The internal
friction angle (’) varies from 40º in zone I, to
42º to 32º for zone III and to 32º - 25º for zone included). Different unloading curves were
IV. obtained for the Bearpaw shale and this was
attributed by Bjerrum to different degrees of
weathering (alteration). Alteration produces
breaking of diagenetic bonds generated during
the geological history of the shale. During
unloading swelling can take place if diagenetic
bonds are broken, releasing the locked-in
recoverable strain energy. If diagenetic bonds are
weak, most of the strain energy will be dissipated
during unloading and shale disintegration will be
low. However, when strong diagenetic bonds are
present, a small part of the strain energy will be
lost during unloading, and the swelling within
the disintegration zone will be large (Bjerrum,
1967).
A similar behaviour was reported by Chandler
(1972) for the Upper Lias Clay from
Northamptonshire (UK). Upper Lias clay is a
Figure 33.Variation of the index properties in highly overconsolidated clay which has been
weathering zones for Keuper marl (Chandler, subjected to natural weathering processes. Based
1969) on in-situ boreholes Chandler reported strength
drops from 200kPa to 63kPa accompanied by
water content increase from 16% to 30%,
depending of the sample level within the
weathering profile. Chandler presented a
schematic relationship between water content
and shear strength in terms of the failure
envelopes (Figure 36).
3. DEGRADATION OBSERVATIONS IN
LABORATORY
v 3 2 q
p 2 2
p c ( p c ) i exp
Bp
where ξ and Bp are model parameters and (pc)I
is the value of pc when plastic strain is zero. The
plastic potential g was assumed to be similar to
the yield function:
p `* 3 2
g 9 3 ln *2 1 * 0
p co 2 9 Figure 60.Simulation of an oedometric test with
Finally, a hypoelastic law is used to describe different bonding values (Gens & Nova, 1993)
the behaviour within the yield surface,
ve kp ' p' and qe L 4.2 Lagioia & Nova model (1993)
An advantage of this formulation is that the Using the same mathematical framework devel-
bond component is independent of the oped by Gens & Nova (1993), Lagioia & Nova
constitutive law for the unbonded soil so that (1993) presented a constitutive model to describe
different bonding models can be used. the mechanical behaviour of soft rocks.
Figure 60 shows the simulation of triaxial Lagioia & Nova (1995) introduced two yield
compression test with different initial confining surfaces (for the structured and de-structured
pressures at the same degree of initial bonding, states (Figure 62). Hardening was controlled by
bo=1. volumetric and deviatoric plastic strains. No
One dimensional tests were also simulated and anisotropic-induced anisotropy was introduced in
results are presented in Figure 61 for bo=1 and these models.
bo=2. An initial value of pc=50kPa was used in
using a kinematic hardening rule. Plasticity
theory can be used to describe the inelastic
behaviour using a classical flow rule and a yield
surface. When the current stress state moves
beyond the initial boundary of the bubble it will
move it until it touches the structured yield
surface. The model was shown to reproduce
undrained triaxial tests on an overconsolidated
clay
ub
p b pb 0 vb qb qb 0 qb
The associated volumetric strains are given by:
de 2
d v is the total volumetric strain
1 e The evolution of L can be obtained by drawing
de the change in elastic moduli during
d vb b is the volumetric strain of the
1 e unloading/reloading cycles under different
bonds, applied energies. Using this damage variable,
de strains in the damaged bonds and the damage
d vM M is the volumetric strain of the modulus can be re-written as:
1 e
clay matrix. Taking in account that eM=eb+e, the
vb e L vb and qb e L qb
volumetric strains follow the same relationship:
d vM d vb d v and d qM d qb d q K b e L K b 0 and Gb e L Gb 0
Another feature of this model is that structure
effects are applied to the bonds so that the Stress equilibrium between applied external
argillaceous matrix behaviour can be defined stresses and internal stresses (distributed between
using a conventional law for destructured bonds and matrix) during any strain increment,
material. This means that the argillaceous matrix leads to:
will have the same local behaviour d d
independently of the degree of bonding. p p M 1 vb pb p b 0 vb
d v d v
Experimental test have shown that bonds
exhibit typically a brittle behaviour. Using the d qb d
q q M 1 qb qb 0 qb
theory of quasi-brittle materials, Vaunat & Gens d q d q
applied an elastic damage law to model the
behaviour of bonds. They used the general The main assumption made is that the ratio
anisotropic damage model proposed by Carol et between volumetric strains inside the bonds and
al. (2001). However, only the isotropic part was the overall volumetric strain is constant and
used in this model for simplicity. The scalar equal to a new variable χ (χ0 for undamaged
damage theory assumes a linear elastic behaviour state). A similar criterion was assumed for the
of the material but during loading the cross deviatoric strains. Mathematically this can be
sectional area implied in the definition of stress written as:
is reduced due to the development of micro- d vb d qb
and
cracks. d v d q
If D is defined as the ratio between micro-cracks
When damage is developed inside the bonds
area and overall area D Amicro cracks / Aoverall , and (L>0), the expressions for the mean and
K b 0 and Gb 0 are the bulk and shear modulus of deviatoric stresses are given by
undamaged material, the mean and deviatoric
stresses are given by: p 1 p M pb pb 0
pb pb 0 1 D K b 0 vb K b 0 vb K b vb q 1 q M qb qb 0
vb e L where it can be observed that this model can
where 0e L . reproduce the delayed damage of the clay.
v
Vaunat & Gens (2003) applied this model to
simulate the behaviour of two materials. First,
they simulated an isotropic compression test on a
structured synthetic clay. Then, they applied
their model to reproduce the mechanical
behaviour of a tertiary sedimentary mudstone of
low porosity with a content of calcium content
close to 20% (Callovo-Oxfordian clay).
In this case, the failure surface of the intact
material was taken as:
q qt 2 m
q qt m p pt Rc 0
Rc 3
where Rc RcM 1 . For this case, the
variation of the Young modulus is given by
(from the isotropic part of the elastic law):
E 1 0 E M 1 0 0 Eb 0 0
Results of simulation on Callovo-Oxfordian
clay are shown in Figure 64 in which simulated Figure 65.Simulation and experimental result of
shear strength values are compared with a multi stage triaxial creep test (Vaunat & Gens,
experimental data. Experimental results lie inside 2003)
the zone defined by the structured and
destructured material. This model was used to reproduce the thermo-
hydro-mechanical (THM) behaviour of Opalinus
clay in the Mont Terri Underground Laboratory
during the heating test (HE test, see section
2.1.2) by Muñoz & Alonso (2005a). A horizontal
niche of 5.40m wide and 4.70m high was
excavated and then a vertical borehole was
drilled. Subsequently, a central cylindrical heater
and bentonite buffer between heater and rock
was emplaced into the borehole. Then, the
bentonite was hydrated. The test involved a
history of heating and cooling which was
introduce to investigate the rock response. Pore
water pressures, stresses and temperatures were
monitored at specific points during the test. This
test was modelled by Muñoz & Alonso using the
finite element code CODE_BRIGHT (Olivella et
al, 1996) developed at the Department of
Geotechnical Engineering and Geosciences at
the UPC in Barcelona. Some results of
Figure 64.Experimental and simulated peak modelling were compared with in-situ
shear strength for Callovo-Oxfordian clay (Vau- measurements on Opalinus clay as described
nat & Gens, 2003) below. Simulation was divided in two parts: (i)
an excavation of the HE niche (420 days), and
In addition, this model was used to simulate a drilling the vertical borehole BHE-0 during 10
multi-stage triaxial creep test. To take into additional days (pore water pressures,
account the creep phenomena, a viscous temperature and stress state were calculated
component was introduced in the constitutive during this process), and, (ii) the bentonite buffer
law for the bonds. The simulation and the emplacement followed by hydration (982 days),
experimental results are shown in Figure 65 heating (554 days) and cooling (365 days)
phases.
Due to the low porosity of Opalinus clay During heating, liquid pressure increments are
(close to 14%) the yield surface followed the induced by the thermal expansion of the water.
Hoek & Brown criterion, a similar choice of The magnitude of the increments of liquid
Vaunat & Gens (2003) for the Callovo- pressures is controlled by several factors: the rate
Oxfordian clay. Bond and clay matrix were of temperature increase, the rock permeability,
modelled following the framework previously the rock porosity, the rock stiffness and
described (bonds by means of a logarithmic geometry and boundary conditions of the
damage law and clay matrix by means of a experiment (Muñoz & Alonso, 2005). The model
conventional model for clay). On the other hand, captured the liquid pressure tendency during the
bentonite was modelled using a thermoplastic whole test (see Figure 69).
model for unsaturated soils but it will not be
described in this paper (to more details see Lloret
et al, 2002). A schematic representation of the
yield surfaces used for the Opalinus clay model
is presented in Figure 66
.
coordinates.
Cooling stage
First yield
7
0 (1 S ( ))
P
Figure 71.Schematic representation of an elasto- c c0 1 Sc ( ) 0 tan 0 S Sc
viscoplastic material (Muñoz & Alonso, 2005b) c0
X, Y, Z: Global coordinate system
tan tan 0 (tan 0 tan res ) S ( )
X', Y', Z': Local coordinate system
: Angle of strike (0º<360º) // Y
The viscoplastic potential Q is similar to the
with respect to Y axis hyperbolic yield function F , but it depends of
: Angle of dip (0º90º)
Z' the tensile strength, cohesion (c) and
dilatancy angle The evolution of the
Isotropic plane
X' viscoplastic potential function Q is given by the
Y'
reduction of parameters (, c and ). The
ip
fd
co
n
t ou
L in
0
Mohr - Coulomb
Hyperbolic
yield surface
res
0 0
Mean stress (p)
Hyperbolic
yield surface
0
Alonso, 2005b)
10
A viscoplastic stress state outside of the yield
surface can exist. In this case, there is no
condition forcing stress state to stay on the yield 5
Vertical stress - Dip
condition).
Dip () [º]
Stress [MPa]
Circumferential stress (Borehole wall X direction)
Vertical stress (Borehole wall X direction)
Radial stress (Borehole wall Y direction)
Circumferential stress (Borehole wall Y direction)
Vertical stress (Borehole wall Y direction)
Radial stress(Isotropic model)
Borehole BHE-0 Circumferential stress (Isotropic model)
Vertical stress (Isotropic model)
0
Pw = Hydrostatic pore water
20.00 m
5
0 200 400 600 800 1000 1200 1400 1600 1800 2000
Time [days]
0.6
Liquid pressure [MPa]
Plastic zone
e = 0.08 m
0.4
0.2 Borehole
(2)
-0.2
0 200 400 600 800 1000 1200 1400 1600 1800 2000
Time [days] Figure 80.Plastic zones developed in a horizontal
Figure 78.Evolution of liquid pressure with time section at 5.0 m depth: (top) Isotropic model,
in points located in rock during the all phases of (bottom) anisotropic model (Muñoz & Alonso,
the HE test (Muñoz & Alonso, 2005b) 2005b)
Despite of its increased complexity and the
added difficulty to find material parameters, it is
believed that a realistic simulation of shales
requires the development of anisotropic models.
Engineering works such as excavations have Figure 81.A conceptual representation of the
two distinct effects on the exposed materials: a bonded clayey model.
reduction in confining stress (and possibly an
increase in deviatoric stress components) and a 5.1 Matrix behaviour
cyclic change in humidity due to atmospheric
The matrix reacts to stresses Mij and suction
interaction. A new constitutive model inspired in
the observed features of clayey rocks, outlined changes. For expansive materials, an explicit
before has been developed by Pinyol et al. consideration of the two structural levels defined
(2007). Specific attention is given to the for the matrix (micro and macro) was used to
simulation of weathering effects and, in describe the features of expansive soil behaviour
particular, to two important mechanisms that are according to Gens & Alonso (1992) and Alonso
believed to control, in practice, the loss of et al (1999). Since clay expansion is interpreted
mechanical properties of these materials: the as local hydration phenomena in which clay
straining associated with stress changes and the aggregates “capture” the water available in
wetting-drying cycles imposed by atmospheric macropores, it is reasonable to assume that, away
action. from equilibrium, two different suction values
In this model, the clayey soft rock is exist in the soil: a microstructural suction, smicro,
conceptualised as a composite material made of a prevailing inside the aggregates and a
clay matrix and a distributed network of bonds, macrostructural suction existing in macropores.
which occupy partially the macropores of the The microstructural behaviour is assumed to be
clay matrix (Vaunat & Gens, 2003). The clay volumetric and non-linear elastic. Mean stress
matrix has its own structure: clay platelets are and smicro control their volumetric response,
stacked in aggregates that may swell or shrink. where the mean stress in the aggregates is in
Some examples of clay stacks in shaly rocks and equilibrium with the externally applied mean
clay shales are given by Pye & Krinsley (1983) stress on the matrix, pM . Suction in macropores,
and Seedsman (1987). smacro, will not be necessarily in equilibrium with
Volume changes of these stacks (which may smicro. Local suction gradients can result in local
integrate thousands of individual clay sheets) are fluid transfer. However, they only present the
thought to disrupt the bonded structure of the drained response of the material (smacro = smicro=
rock. The structure of the clay matrix is s)
conveniently described as a micro-macro Since clay aggregates are supposed to remain
structure. “Micro” refers to the porosity inside saturated, in view of the very small spacing
the clay aggregates. “Macro” describes the void between clay platelets, the effective stress,
space between clay aggregates. The “macro” defined by the sum (pM + smicro) holds and a
space includes open voids available for fluid simple constitutive law may be assumed for the
transfer and the volume occupied by the volume change of aggregates:
cementation minerals (bonds). Two types of
d m
m d p s
M
d p
M
s
pores coexist in the medium: the small pores vol
1 em pM s Km
inside the aggregates and the relatively large
pores between aggregates (see Figure 81). where m is a (constant) compressibility
coefficient and Km is the bulk modulus of clay
aggregates.
Macrostructural behaviour includes the
competition of two mechanisms: the response of
the matrix skeleton (organized in aggregates) to
load and suction, as it is considered in a non
expansive material, and the macrostructural
arrangement due to microstructural swelling and
shrinkage. The first mechanism is modelled
through the Barcelona Basic Model (Alonso et
al., 1990). The second mechanism is introduced
by means of an additional component of
volumetric plastic strain which depends on the
volumetric change experienced by the
microstructure and the distance of the current
stress state to the collapse yield loci. The onset Figure 82.Elastic zone and yield loci for the
of this coupled plastic strain is defined by two matrix expansive model (volumetric behaviour)
additional yield loci: The SD (Suction Decrease) (Pinyol et al, 2007)
yield surface which marks the development of
irreversible swelling strains, and the SI (Suction 5.2 Bond behaviour
Increase) surface, which defines the development It may be expected that, for high bond
of irreversible shrinkage. concentrations, the nature of the bond material
The elastic response of the arrangement of will dominate the bond constitutive behaviour.
aggregates (macro) is defined by an expression For low to moderate concentrations, which is the
similar to the last equation, but in this case the case of hard and cemented clay formations, the
effective stress principle does not hold and two nature of bond material has probably a more
separate contributions associated with the matrix limited effect. Experimental results have
stress and the (macro) suction lead to the elastic suggested that bonding introduces a stiff
response before yielding and that it also provides
strain:
a distinct brittleness. The idea favoured in the
model is that bonding becomes damaged when it
dpM s ds dpM ds
dMvol
e
is subjected to a given amount of strain energy,
1 eM* pM 1 eM* s patm KM K s irrespective of the stress path applied. This
hypothesis is also consistent with the idea that
where and s are the elastic macro bonding is established once the matrix is
stressed. Bond strain energy takes as a reference
stiffness parameters for changes in mean matrix zero state the initial stress of the investigated
stress and in macro suction, respectively. KM and stress path. Bond behaviour was modelled using
Ks are the bulk elastic moduli of the clay an isotropic damage theory (Carol et al, 2001) in
aggregate arrangement against changes in matrix terms of a scalar damage parameter D which
stress and (macro) suction. patm indicates the varies between 0 and 1. Figure 83 shows that the
atmospheric pressure. A representation of amount of bond damage must control the
yielding states of the model (for isotropic stress intensity of stresses resisted by the bond.
states) is given in Figure 82 in a plane defined by Carol et al. (2001) proposed that damage
the mean matrix stress (pM) and the suction s. should be linked to the increments of energy
When yield loci SI or SD are activated, stored in the bonds per unit of volume. This
additional irreversible strain component are energy is defined as the elastic secant energy
generated. In the formulation developed the which would be recovered upon unloading
plastic straining associated with basic (elastic) 1
mineral expansion or contraction is controlled by ( u b bij bij ). With this formulation, current
2
the current overconsolidation ratio. bond damage locus is defined by a threshold (r0)
The formulation of the plastic straining of the secant elastic energy that can be
associated with suction changes in terms of two represented by an ellipse in pb-qb space. Inside
yield limits, SI and SD helps to reproduce the this ellipse, the bond stiffness remains constant.
evolving deformation of clays when several When changes in the tensional state of the bond
wetting-drying cycles are applied. lead to the secant elastic energy to reach the
threshold, damage takes place (D increases) and
stiffness reduces. The damage evolution is 5.4 Validation of the model
determined by means of the function suggested In order to illustrate the capabilities of this
by Carol et al. (2001): model, Pinyol et al. (2007) performed synthetic
r L r0 e r1L u b oedometer and triaxial tests to evaluate the effect
Where L=ln(1/(1-D)) and r1 determines the of varying the initial damage state and the
damage rate. The consistency condition damage rate. One example is provided in Figure
associated with this energy locus allows the 84 which represents simulated oedometer results.
determination of L. The initial stiff response of the material
corresponds to the undisturbed state. The slope
of the first part of the compression curve is
identical for all the simulated tests. The initiation
of damage is controlled by the initial size of the
bond energy locus (r0). As load is increased, the
induced damage results in a sharp transition
towards a softer response (Figure 84). As strain
accumulates the material evolves toward the
Figure 83.A damaged bond will not be locally matrix behaviour. The reason is that, as bonding
strained by a macro-deformation degrades, the stress is progressively transferred
to the matrix. The elastoplastic matrix response
(a continuous line in the figure) has been
5.3 Strain and stress partition computed with the same parameters of the
The bond strain have been defined as bij ijv undisturbed bonded material but imposing Cb =0.
0.
where (o , L) is assumed to depend on an ini-
tial reference value 0, and on the logarithmic
damage variable, L. This strain relationship ex-
plains that in a highly damaged material ( is
small) the local bond deformation will be small
when the macrovoids are deformed (see Figure
83). On the contrary, the local bond strain will
increase when damage is reduced because the
bond stiffness is high in comparative terms.