Rock Mech Rock Eng

DOI 10.1007/s00603-015-0860-5

ORIGINAL PAPER

Behaviour of Weak Shales in Underground Environments

C. Derek Martin1 • Silvio Giger2 • G. W. Lanyon3

Received: 30 September 2015 / Accepted: 30 September 2015

 Springer-Verlag Wien 2015

Abstract Predicting the ground response for tunnels in the laminations (fissile). While the geological classification
weak shales remains challenging. Predicting the ground of shale appears straightforward, the classification for
response is challenged by difficulties in characterising the engineering purposes is often confusing. Underwood
material, and our ability to predict deformations that are (1967) provided the first description of shale that made the
driven by coupled hydromechanical processes, when this distinction between ‘‘soil-like’’ compaction shales and
material yields. The techniques that are used for charac- ‘‘rock-like’’ cemented shales. He coined the term ‘clay
terising weak shales are reviewed, and three case histories shale’ to capture the characteristics of hard clay yet
are examined that demonstrate the behaviour of these recognising its origin as a shale. In this paper, weak shale is
weak rocks during tunnelling. A general framework is synonymous with clay shale and has strengths that typi-
provided for assessing the squeezing potential for weak cally range from R1 to R2 (1–25 MPa). Characterising
shales. these materials draws on methods from both soil and rock
mechanics.
Keywords Shales
Weak rocks
Squeezing Index
While tunnelling experience in weak rocks has
Tunnel strain increased significantly in the past 25 years, predicting
the ground response for tunnels in weak (clay) shales
remains the most challenging. While terms such as
1 Introduction ‘squeezing’ and ‘swelling’ are generally used to
describe the behaviour of the ground, predicting the
Geologically, shale is an argillaceous fine-grained sedi- ground response is explicitly linked to predicting the
mentary rock that forms from the compaction of silt and deformations. Predicting the ground response is chal-
clay-size mineral particles commonly referred to as lenged by (1) difficulties in characterising the material
‘‘mud’’. Because of this grain size, the material is generi- and (2) our ability to predict deformations that are
cally classed as Mudstone (Fig. 1). Shale is distinguished driven by coupled hydromechanical processes, when the
from other mudstones because it is made up of many thin matrix and bedding planes yield. In this paper, the
layers (laminated) and readily splits into thin pieces along techniques that are used for characterising weak shales
are reviewed, and three case histories are examined that
demonstrate the behaviour of these weak rocks com-
monly encountered when tunnelling.
& C. Derek Martin
derek.martin@ualberta.ca
1
Department of Civil and Environmental Engineering, 2 Characterising Weak Shales
University of Alberta, Edmonton, AB T6G 2W2, Canada
2
National Cooperative for the Disposal of Radioactive Waste The characteristics and behaviour of weak shale commonly
(Nagra), 5430 Wettingen, Switzerland encountered in laboratory test programmes are briefly
3
Fracture Systems Ltd., St. Ives, Cornwall TR26 1EQ, UK discussed below.

123
 C. D. Martin et al.

Fig. 1 General geomechanical

classification of argillaceous
soils and rocks (modified from
Bock 2010)

2.1 Percentage of Weak Clay Minerals montmorillonite and illite, common clay minerals found in
and Plasticity weak shales.

Weak shales are characterised by a high portion of clay 2.2 Porosity

minerals, often ranging from 30 % to greater than 50 %.
That portion and the type of clay minerals and their Unlike many rocks, weak shales have total porosities that
structure have a significant influence on its mechanical, typically range between 5 and 30 %. The porosity is an
hydromechanical and geochemical behaviour. One of the indicator of the depth of burial in geological time and
consequences of the high clay content is the very low past loading conditions, with the lower the porosity the
hydraulic conductivity typically \10-10 m/s. This poses greater the depth of burial and loading. Low porosity
challenges for strength testing when pore pressures should (\3 %) is also an indicator of the potential for true
be measured. chemical cohesion. Typically, the higher the porosity the
The Atterberg limits remain one of the first steps in weaker the rock, but the porosity can be artificially
identifying the potential behaviour of the weak shale and increased due to unloading and sample disturbance
also the clay mineralogy. Figure 2 illustrates the location (Fig. 3).
of the clay minerals on the Casagrande plasticity
chart based on the Atterberg limits. Note the location of the 2.3 Swelling and Slaking

The type of clay minerals (e.g. smectites) has a significant

impact on propensity for slaking and swelling. Figure 3
illustrates the general behaviour of weak shales when
immersed in water or exposed to moisture in the environ-
ment. Swelling pressures of undisturbed samples can
exceed 1 MPa when exposed to moisture.

2.4 Strength and Stiffness as a Function of Moisture

Content

The unconfined and confined strength and stiffness of these

materials are a function of water content. As illustrated in
Fig. 4, small increases in the water content can signifi-
Fig. 2 The Casagrande plasticity chart with common clay minerals
cantly decrease the unconfined compressive strength. This
(after Holtz et al. 2011). The abbreviations CH, CL, CM, etc., are the characteristic emphasises the need for high-quality samples
Unified Soils Classification symbols when undertaking a strength testing programme.

123
Behaviour of Weak Shales in Underground Environments

Fig. 3 Responses of weak shale

when exposed to moisture—
from total disintegration to mud,
to no reaction (left
figure adapted from Santi 2006)

Fig. 4 Effect of water content on the unconfined strength of several

weak shales

2.5 Anisotropic Strength and Stiffness Fig. 5 Example of the uniaxial compressive strength relative to the
direction of loading for the Opalinus clay

Because of the aligned clay particles, many weak shales strength is only a function of the frictional characteristics
exhibit distinct bedding at the mm scale. Consequently, of the particles. Consequently, pore pressures must be
strength and stiffness will be transversely isotropic and measured when establishing the strength. Because of the
testing programmes are usually carried out with this in low permeability, measuring the pore pressures is extre-
mind (Fig. 5). While the strength reduces when loading is mely difficult and time-consuming. Traditional soil
parallel or sub-parallel to bedding, the stiffness increases. mechanics labs often do not have the equipment with
sufficient capacity to handle weak rocks and rock
2.6 Failure Envelope mechanics labs are typically not set up to measure the pore
pressures in such tight materials. If pore pressures are not
Establishing a failure envelope for weak shale is particu- measured, the effective strength components (cohesion and
larly challenging. If the material has no true cohesion, the friction) cannot be known with confidence.

123
 C. D. Martin et al.

Fig. 7 Relationship between uniaxial compressive strength and

suction. Mean value and data range for each applied suction are
indicated (modified from Wild et al. 2014)

envelopes that do not take these water retention charac-

teristics into account can significantly over predict the
strength measured in laboratory samples.
Consequently, because of these characteristics, it is
challenging to obtain laboratory properties that are not dis-
turbed, and because of this disturbance, it is difficult to know
if the characteristics and properties observed in the labora-
Fig. 6 Possible interpretation of the failure envelope for a weak shale tory samples are relevant for predicting the in situ response.
when pore pressures are not measured and the measured failure plane
inclinations (modified from Amann et al. 2012)
In the following section, three case histories in weak shales
are reviewed and the properties of those shales described.
Figure 6 illustrates a bi-linear failure envelope for tri-
axial tests carried out using traditional rock mechanics
testing protocols (rapid loading and no pore pressure 3 The In Situ Performance of Tunnels Excavated
measurements). Notice the marked reduction in friction in Shale
angle, from 43 to 11 when the confining stress exceeds
1 MPa. Weak shales with their high clay content would be In this section, tunnelling case histories are examined in
expected to have friction angles between 18 to 25, various shales where the laboratory properties are charac-
depending on the mineralogy. A likely explanation for the terised, the in situ stresses were measured, and the defor-
bi-linear failure envelope can be seen in the inclination of mations during tunnel excavation were also measured.
the failure plane in Fig. 6. At low confining stresses, the Table 1 provides the laboratory properties and in situ stress
samples are not saturated and the increase in strength is magnitudes for 5 different shale formations that range in
dominated by dilation and suction, evidenced by the steep age from 29 to 443 Ma. The percentage of clay minerals
inclination of the failure plane relative to the direction of ranges from 25 to 60 % and the uniaxial strength ranges
loading. At confining stresses above 1 MPa, the sample from 2 to approximately 50 MPa.
saturation is now increased and strength is dominated by
the pore pressure response, evidenced by the 45 inclina- 3.1 Geotechnical Properties of Five Shales
tion of the failure plane suggesting a total stress response
with / = 0 (giving a failure plane oriented at 45). The range in uniaxial compressive strengths for the five
The important role of suction in establishing the strength shale formations is illustrated in Fig. 8a, which compares
and modulus of weak shales was highlighted by Wild et al. the strength using the ISRM strength classification chart for
(2014). Figure 7 shows the impact of full saturation and both rock and clays. Four of the shale formations can be
partial saturation (suction) on the uniaxial strength of the classed as weak rock, while the Queenston shale, with a
Opalinus Clay. As the suction increases, the strength UCS of approximately 48 MPa, is classed as R3 (medium
increases from 4 to 15 MPa. Similar results were observed strong rock). When the UCS values suggest a ‘‘weak rock’’,
for the tensile strength and stiffness. Hence, failure there is always the potential that the behaviour of the rock

123
Behaviour of Weak Shales in Underground Environments

Table 1 Comparison of the properties of shales with documented tunnel performance

Parameter Boom claya Opalinus claya Shaftsbury formationb Callovo-Oxfordian Queenston Shale
(Mol URF) (Mont Terri URL) (Site C test chamber) (Bure URL)a (Niagara Tunnel)c

Geologic age
Stratigraphic Oligocene Lower Dogger Mid Cretaceous Callovo-Oxfordian Upper Ordovician
Absolute (Ma) 32–29 176–172 99.6–112 163–158 *443.7
Tunnel depth (m) 186 240 90 490 140 (max)
SHmax/SHmin (MPa) 4.5/4.5 4.6/2.2 2.2/0.8 14.6/11.3 14.5/7.3
P
All clay minerals (%wt) 60 66 35–49 25–60 60
P
Swelling clay minerals (%wt) 40 45 34–52 10–40 –
Bulk density (saturated) (kg/m3) 2000 2430 2500 2430 2600
Water loss porosity (%) 36.3 13.7 – 18 10.2–11.4
Water content (%wt) 21.5 4–6.4 3–8 8 2.7
Atterberg limits (PL/LL) (%) 25/70 23/38 23/45 – 18/20
UCS (MPa) 2.0 11–15 5–13 21 ± 6.8 39–48
Young’s modulus (GPa) \/|| 3 4/10 4/6.5 4/10 11.5
Poisson’s ratio 0.43 0.29 0.24 0.29 0.36
Swelling pressure (MPa) 0.9 0.6–1.2 0.02–0.05 1 4–5

UCS uniaxial compressive strength, – implies no data

a
Bock et al. 2010
b
Cornish and Moore (1985), Sargent and Cornish (1985)
c
Tom Lam, pers. comm

will be similar to that of a hard clay (see Fig. 8a.) To check

if the behaviour of the weak shale is similar to that of clay,
the Atterberg Limits for the shales should be assessed using
the Casagrande Plasticity Chart. Figure 8b compares the
Atterberg Limits for the five shales using the Casagrande
Plasticity Chart. Note that Queenston shale with the highest
uniaxial strength has essentially zero plasticity, while the
weak Boom Clay has the highest plasticity and the lowest
UCS. It is clear that the combination of these two charts
illustrate the potential difference in the shale strength and
behaviour.
Three case histories are described in the following
sections, which highlight the range in tunnel performance
observed in these formations. These case histories are
simply named Site C Test Chamber, Mont Terri Mine-by,
and the Niagara Tunnel.

3.2 Site C Test Chamber

A hydroelectric development proposed at Site C on the

Peace River near Fort St. John, British Columbia, Canada
required the construction of three 9-m-diameter or two
12-m-diameter diversion tunnels. To aid in the decision, a
11-m-diameter Test Chamber was constructed along the
proposed centreline of one of the diversion tunnels, near
the midpoint where ground cover was greatest (Fig. 9).
Fig. 8 Comparison of the uniaxial compressive strength and plastic- Details of the Test Chamber construction are given in Little
ity for the five shales in Table 1
(1989), and only a brief summary is provided here.

123
 C. D. Martin et al.

Fig. 9 Plan and section of the

test chamber (modified from
Imrie 1991)

The Test Chamber was primarily excavated using a ahead of the face, convergence pins and load cells. These
roadheader with limited reach and hence a top heading instruments were installed in six circumferential rings at
and bench was required (Fig. 10). Each advance cycle 6-m intervals along the test chamber. The most effective
was 2 m long, followed by installation of support and instruments were the multipoint extensometers. The max-
instrumentation. To compare excavation methods, the imum total movement recorded on any extensometer was
final 15 m of the top heading was drilled and blasted, 2.2 mm. For most extensometers, the total recorded
using rounds from 2 to 3 m in length. Support and movement was 1 mm or less and most of this movement
instrumentation was installed after each round. Con- occurred within 4 m of the excavation profile (Little 1989,
struction of the test chamber took 4 months, with com- see Fig. 10).
pletion at the end of August 1981. One of the challenges when comparing measured dis-
Rock support for the test chamber consisted of an initial placements with those predicted using numerical approa-
50-mm layer of shotcrete, followed by a 2-m square pattern ches is the selection of the input parameters. Inspection of
of tensioned rockbolts that varied in length from 4.3 m the Test Chamber instrumentation and observations
(sidewalls) to 4.9 m (roof), followed by a second 50-mm revealed no evidence of overstressing or non-elastic
layer of shotcrete. Most shotcrete was steel fibre reinforced behaviour (Fig. 10). Hence, all evidence suggested the rock
except in the roadheader section, where plain shotcrete mass behaviour was essentially elastic (Little 1989). This
with welded wire mesh reinforcement was used in the would imply that the input parameters for the numerical
second layer. All support bolts were anchored with epoxy predictions should be based on the intact laboratory results.
resin cartridges and tensioned to 13.6 tonnes. Wire mesh It was recognised early during the site investigations for
was anchored with 0.6-m-long Split-Sets. At the time of its Site C that preserving core samples for laboratory testing
construction in 1981, the Test Chamber was one of the was a challenge. This was highlighted by the Young’s
largest underground openings in weak shale supported modulus values obtained from laboratory tests, carried out
using steel-fibre shotcrete (Little 1989). days after the samples were selected, compared to those
Performance of the rock mass and support system was values obtained from biaxial tests done on site within about
monitored using multipoint extensometers, stress cells 1 h of retrieving the samples. Figure 11 provides the

123
Behaviour of Weak Shales in Underground Environments

Fig. 10 Example of the radial displacement recorded for 3C and 3E during the upper heading and lower heading excavations, respectively
(modified from Little 1989)

Young’s modulus results parallel and perpendicular to the following: (1) a transversely isotropic model is needed
bedding from the biaxial samples. The Young’s modulus to replicate the measured elastic displacements and (2) the
values from traditional laboratory samples (tested days in situ modulus needed to obtain agreement with the
after selection) typically ranged from 1 to 2 GPa, while the measured displacements is greater than the values obtained
biaxial results ranged from 4.4 to 6.6 GPa. from traditional laboratory tests that do not account for the
Numerical analysis analyses were carried out using the sample disturbance. An additional point is related to the
two-dimensional finite element programme phase 2 (http:// weak bedding plane (BP25) in Fig. 12. This weak bedding
www.rocscience.com). Initially, the model input was based plane (approximately 10 mm thick) was explicitly mod-
on isotropic elasticity, while subsequent models used the elled using the measured laboratory properties (/ = 9,
transversely isotropic model to respect the Young’s mod- Kn = 10,000 MPa/m and Ks = 5000 MPa/m). Excluding
ulus difference perpendicular and parallel to the bedding, the bedding plane reduced the agreement with the mea-
shown in Fig. 11. A weak, bedding plane (BP 25) that sured values for region A and E (see location of A and E in
crossed the excavation was also included in the model. The Fig. 13).
model and the input parameters are illustrated in Fig. 12.
The results from the numerical analyses are provided in 3.3 Mont Terri Mine-by Tunnel
Fig. 13. The measured displacements fell between the
isotropic results with a Young’s modulus ranging between Tunnelling and rock mechanics research in the Opalinus
5.4 and 8 GPa (Fig. 13a). The analyses were repeated Clay has been carried out at the Mont Terri Laboratory
using the transversely isotropic model with E1 (horizon- (Switzerland) over the past 15 years. The main experiment
tal) = 8 GPa and E2 (vertical) = 5.4 GPa (Fig. 13b). The level is at a depth of approximately 240 m. The geotech-
measured displacements are in good agreement with the nical properties of the Opalinus Clay are given in Table 1.
elastic transversely isotropic model. These two-dimen- These properties have been taken from Bock (2010).
sional modelling results are limited but serve to highlight Because the experiments have been excavated at a single

123
 C. D. Martin et al.

level, the tunnels tend to be either aligned perpendicular to ground response for tunnels excavated parallel to bedding
the bedding or parallel to the bedding, which strikes was the Mine-by Tunnel (Fig. 14). Extensive instrumen-
approximately 080 and dips approximately 30–40 to the tation was used to monitor the hydromechanical response
Southeast. One of the recent experiments to establish the as the tunnel was excavated.
The Mine-by Tunnel is located near the centre of Gal-
lery08. It consists of a 24.6-m-long and 4.5-m-diameter
circular opening trending 242 clockwise from the north.
The direction of the excavation advance is sub-parallel to
the average strike of the bedding planes. The tunnel rises at
a slope of approximately 0.98 % from the entrance to its
face. The tunnel was advanced between 0.6 and 1.9 m per
day using a road header, and the tunnel face was cut ver-
tically. The tunnel was not advanced over the weekends.
Support consisted of six anchors installed about 0.8 m
behind the advancing face, 100-mm wire mesh reinforce-
ment affixed to the entire circumference of the Mine-by
Tunnel, and two layers of shotcrete. A 50-mm-thick shot-
crete layer was placed immediately following installation
of the wire mesh and followed by an additional 100-mm
final cover that was placed following completion of the
Mine-by Tunnel. The anchors were 22 mm in diameter,
2 m long and installed in 30-mm-diameter pilot holes.
Approximately 1.3 m of each anchor was fully bonded
using resin epoxy (radial dashed lines, Fig. 15).
Analysis of the excavation performance was based on
the total measured response, the instant and time-dependent
components of these responses, and the predicted elastic
behaviour. Macciotta (2012) defined the measured
response as follows:
• Total response—total response measured by the instru-
ment. It includes both the excavation-induced and time-
dependent response when there are no excavation
Fig. 11 Comparison of the modulus values from the USBM biaxial
tests activities.

Fig. 12 Geometry of test

chamber in phase 2 model used
to evaluate the transversely
isotropic model. The contours
illustrate the deformation
pattern, and the numerical
displacements at A, B, C, D and
E are shown in the small
rectangles in metres. The
measured displacements at A, B,
C, D and E are also shown in
small rectangles, e.g.
A = 0.4 mm

123
Behaviour of Weak Shales in Underground Environments

tunnel strain (defined as the ratio of tunnel convergence to

tunnel diameter) is an important indicator for identifying
potential tunnelling instability. Figure 17 presents the
guidelines, based on tunnel strain, developed by Chern
et al. (1998). The short-term maximum tunnel strains
measured at the end of Mine-by Tunnel excavation fall
between Level II and Level III of the empirical guidelines
proposed by Chern et al. (1998). These maximum strains
rise above Level III by the end of monitoring. This would
suggest the tunnel was marginally stable, which is sup-
ported by the initial support repairs required during the
advancement of the tunnel and shortly after completion of
the tunnel. Placement of the final 100 mm of shotcrete and
concreting of the invert provided additional support, which
limited the convergence.
In contrast with the tunnels driven perpendicular to the
strike of the bedding at Mont Terri, the measured response
for the Mine-by Tunnel clearly showed the development of
an extensive plastic zone around the 4-m-diameter tunnel.
The support requirements for these bedding-parallel tun-
nels are therefore different. A support system that reduces
the time-dependent deformations would increase the sta-
bility. Thicker initial shotcrete liners and/or applying a
final reinforcement ring near the face advance would
reduce these time-dependent deformations.

3.4 Niagara Tunnel

Fig. 13 Comparison of the results from the isotropic model and the The Niagara Tunnel Project (NTP) is a 10.1-km-long
transversely isotropic model with the measured extensometer dis- water-diversion tunnel in Niagara Falls, Ontario, which
placements from the test chamber was excavated by a 14.4-m-diameter tunnel-boring
machine. The tunnel intersects the Queenston formation at
• Instant response—measured during the period the a depth of approximately 120–140 m. The Queenston
excavation was being advanced. This response includes Formation is a mudstone and siltstone of Silurian Age,
both the elastic response and non-elastic response. which is generally massive, and as shown in Table 1 and
• Time-dependent response—measured when there was Fig. 8 is classed as R3, medium strong rock with essen-
no excavation advance. While there may be no tially no plasticity. A description of the project the site
excavation advance, there can be other construction/ conditions is provided in Gschnitzer and Goliasch (2009)
excavation-related activities, i.e. installing roof support and Perras et al. (2014).
or instrumentation. According to Gschnitzer and Goliasch (2009), the first
3.3 km of tunnel presented challenging mining conditions
Convergence was measured at five locations on the with overbreaks exceeding 3 m over the roof shield. Perras
tunnel perimeter (targets P1 through P5) and at five sec- et al. (2014), based on observations and measurements,
tions during the tunnel advance. The convergence points P1 divided the overbreak into four zones (Fig. 18). Three of
and P4 recorded the response approximately parallel to the four overbreak zones were within the Queenston
bedding, and convergence points P2 and P5 record the Formation.
response approximately perpendicular to bedding (Fig. 16). According to Gschnitzer and Goliasch (2009) and Perras
Inspection of Fig. 16 shows that the convergence varies et al. (2014), the overbreaks in zones 2 through 4 were
with the distance to the tunnel face and location around the associated with the stresses on the boundary of the tunnel
opening, and that the majority of the convergence is exceeding the rock strength. The high horizontal stresses in
associated with time-dependent deformations. the Queenston are well known in this region of Southern
Based on studies by Chern et al. (1998), the relationship Ontario and are discussed in Perras et al. (2014). Based on
between rock mass strength and the magnitude of the several stress measurement campaigns, Perras et al. (2014)

123
 C. D. Martin et al.

Fig. 14 Location of Mine-by

tunnel at Mont Terri. The tunnel
is oriented approximately
parallel to the strike of the
bedding (Azimuth 240), which
dips approximately 35 towards
the Southeast

et al. (2014), and Perras and Diederichs (2015) concluded

the overbreak in zones 2 through 4 was the result of stress-
induced spalling.
Measurements made in the various overbreak zones are
summarised in Fig. 19. Notice that zone 4 shows the greatest
amount of overbreak, ranging from 1.5- to 4-m depth with
chord closures ranging from 10 to 20 mm. According to
Perras et al. (2014), the orientation of the tunnel in zone 4 is
essentially perpendicular to the maximum horizontal stress
(see Fig. 19). The maximum elastic tangential stress on the
tunnel boundary, in zone 4, can be approximated using the
Kirsch solution. Using the lower bound for the maximum
horizontal stress of 14 MPa and a vertical stress of 4.8 MPa,
the maximum tangential stress is approximately 38 MPa. As
the orientation of the tunnel in zone 2 became more parallel
Fig. 15 Mine-by Niche cross section showing the initial 50 mm and
to the maximum horizontal stress, the maximum tangential
final 100-mm-thick shotcrete ring, schematic positioning of the stress is reduced to approximately 24 MPa. It appears that
anchors (dashed radial lines), and 300-mm concrete invert the reduced tangential stress may be sufficient to reduce the
amount of overbreak assuming the overbreak is primarily
caused by stress-induced spalling.
concluded that the maximum horizontal stress in the
Queenston Formation at the depth of the tunnel was
approximately 3–5 times the vertical stress (with magni- 4 Assessing Ground Behaviour
tudes ranging from approximately 14–24 MPa), while the
minimum horizontal stress (9.6 MPa) was approximately In the previous section, the ground responses for three
two times the vertical stress. The shape of the overbreak tunnels were examined. Two tunnels were excavated in
zones in zones 2 though 4 are illustrated in Fig. 19. Perras weak rocks that are classed as R2 and one tunnel was

123
Behaviour of Weak Shales in Underground Environments

Fig. 16 Convergence deformations measured at each monitoring target. The x-axis is the tunnel chainage with the tunnel face indicated by the
dashed line

Fig. 17 Empirical correlation between tunnel performance and rock range of rock mass strength found in the Mont Terri laboratory and
mass strength (yellow circles are unstable, red squares are stable, the range of convergence strains measured (colour figure online)
after Chern et al. 1998). The pink rectangular area represents the

123
 C. D. Martin et al.

Fig. 18 Longitudinal section of the tunnel showing major geological groups, the tunnel alignments (existing, old and final) and the overbreak
zones 1 through 4 (from Perras et al. 2014)

Fig. 19 Location of the overbreak zones from 2 through 4; the representative cross sections in those zones; and the chord closure and amount of
overbreak measured in those zones (modified from Perras and Diederichs 2015)

excavated in a medium strong rock classed as R3. While all conditions encountered driving the 5-m-diameter, 23.3-km-
the rocks are classed as ‘shales’, the ground behaviour was long Yacambu-Quibor tunnel in graphitic phyllite with a
clearly very different. Given the challenges with the lab- maximum depth of 1270 m. Based on measurements and
oratory characterisation of weak shale, it is no surprise that observations they concluded that there was general agree-
predicting the performance of tunnels excavated in shales ment between the tunnel strain, expressed as convergence
is also challenging. Predicting the performance of a tunnel normalised to the tunnel diameter, and the regions of high
means estimating if yielding of the ground will occur and stress caused by the mountainous terrain. They also noted
the deformations associated with the yielding. In weak the large plastic zone associated with this type of yielding
rocks, this yielding is often referred to as squeezing. Hoek and the challenges in determining the rock mass strength
and Guevara (2009) describe the severe squeezing and selecting an appropriate support system.

123
Behaviour of Weak Shales in Underground Environments

Hoek and Marinos (2000) using a numerical plasticity like behaviour/characteristics of the material. Hence, in
model and calibrating the model to field experience sug- order to use the general concept proposed in Fig. 20, the
gested that tunnel strain as a function of rock mass strength rock strength will be replaced with the laboratory uniaxial
could be used to predict tunnel performance (Fig. 20). The compressive strength, and the far-field maximum stress
primary purpose of this work was to identify support issues will be replaced with the maximum tangential stress on the
that could arise for tunnels excavated in rocks that have the boundary of the tunnel. Using the maximum tangential
potential to ‘squeeze’. In Fig. 20, the rock mass strength stress provides a simple method to incorporate the direc-
was determined using the Hoek–Brown Geological tionality of the tunnel relative to the stress tensor.
Strength Index (GSI) methodology and then normalised to The response of the tunnels excavated in the shales
the far-field stress. This Squeezing Index was related to the listed in Table 1 was compiled to assess the methodology
diametral strain (convergence/diameter 9 100 %) proposed by Hoek and Marinos (2000). The results are
observed in various tunnels and to the expected support. presented in Table 2 and plotted in Fig. 21. The tunnel
They suggested that when the diametral strains are less strain ranges from 0.05 to 2.67 % and the Squeezing Index
than 1 %, few support problems are encountered, but when ranges from 0.52 to 1.43. The tunnel strain in Table 2 is not
tunnel strains exceed 2.5 %, the support becomes chal- directly comparable to the tunnel strain given by Hoek and
lenging. Note that the tunnel strains in Fig. 20 assume that Marinos (2000). For the tunnels in Table 2, the tunnel
no support was installed. In shales, the Hoek–Brown GSI strain is influenced by the installed support and how the
methodology is not normally applied because of the soil- measurements were taken, i.e. when measuring conver-
gence from inside the tunnel, a portion of the total con-
vergence is not recorded, while Hoek and Marinos (2000)
reported the total tunnel strain without support. This is
most obvious for the tunnel excavated in the Boom Clay
where the tunnel liner used in conjunction with the tunnel-
mining machine is considered very stiff compared to the
support system used in the other tunnels in Table 2.
Nonetheless, the general agreement between Figs. 20 and
21 is evident.

5 Discussion and Conclusions

In the previous section, the Squeezing Index was used to

Fig. 20 Numerically predicted squeezing convergence normalised to
establish a relationship to tunnel strains. As noted by
the tunnel diameter versus rock mass strength normalised by the far-
field maximum stress (from Hoek and Marinos 2000) Einstein (1996), squeezing is the time-dependent shearing

Table 2 Compilation of tunnel strain and Squeezing Index for tunnels excavated in the shales in Table 1
Tunnel Dia (m) rmax (MPa) rmin (MPa) UCS (MPa) Squeezing Index Tunnel strain (%) References

Bure 5.2 11.27 11.27 21 0.93 0.83 Armand et al. (2013)

Bure 5.2 11.27 11.27 21 0.93 0.42 Armand et al. (2013)
Bure 4.6 14.651 11.27 21 0.64 1.39 Armand et al. (2013)
Bure 4.6 14.651 11.27 21 0.64 2.20 Armand et al. (2013)
MT 4.5 6.5 2.29 17 0.99 0.50 Lanyon et al. (2014)
MT 4.5 6.5 2.29 17 0.99 0.80 Lanyon et al. (2014)
MT 4.5 6.5 4.21 8 0.52 1.33 Lanyon et al. (2014)
MT 4.5 6.5 4.21 8 0.52 1.56 Lanyon et al. (2014)
MT 3 6.5 4.21 8 0.52 2.67 Lanyon et al. (2014)
Boom 4.8 4.5 4.5 2 0.22 1.88 Bernier et al. (2007)
Site C 11 2.6 0.8 10 1.43 0.05 Little (1989)
Niagara 14.4 14.5 5.3 48 1.26 0.10 Perras et al. (2014)
MT Mont Terri

123
 C. D. Martin et al.

the processes in shales. A key aspect of squeezing is the

associated time-dependent deformations. This is illustrated
in Fig. 22 for the Mine-by Tunnel excavated in the Opal-
inus Clay. In the case histories used to develop Fig. 21,
only the Site C tunnel and the Niagara Tunnel did not
display time-dependent deformations. The Site C tunnel,
despite its large diameter and medium plasticity (see
Fig. 8b), was not deep enough to cause overstressing of the
shales and induce squeezing and the associated time-de-
pendent deformations.
Inspection of Fig. 8b shows that all the shales had sig-
nificant plasticity except the Queenston shale. The
Queenston shale is also not considered a weak rock, as
Fig. 21 The response of tunnels excavated in various shales shown in Fig. 8a. The Niagara tunnel, excavated in the
Queenston shale, was included in Table 2 and Fig. 21 for
comparison. The tunnel strain provided by Perras and
Diederichs (2015) is also only a portion of the total tunnel
strain. However, given that the chord displacements were
taken in the area of significant overbreak, the amount of
displacements for the 14.4-m-diameter tunnel is relatively
small, suggesting essentially an elastic response.
It was clear from the description provided by Perras
et al. (2014) that the main ground response observed during
construction of the Niagara tunnel in zone 4 was spalling.
Spalling is typically considered a brittle response that
occurs immediately during tunnel excavation, and any
time-dependent deformations are associated with sec-
ondary processes. Martin et al. (1999) compiled the depth
Fig. 22 Illustration of the time-dependent convergence observed in of spalling observed in various tunnels in strong rocks
the Mine-by Tunnel (modified from Lanyon et al. 2014) using the approach given in Fig. 23. Also plotted in Fig. 23
are the overbreak results from the Niagara tunnel for the
high-stress and low-stress conditions given by Perras et al.
(2014). It is clear from Fig. 23 that the ground response
from the Niagara tunnels in zone 4 is in agreement with the
spalling experience observed in other strong rock tunnels.
Assessing the potential for squeezing around tunnels
remains a challenge. The case histories reviewed in this
paper highlight these challenges. While the Squeezing
index appears promising, additional case histories are
needed to build confidence in the approach.

Acknowledgments The author wishes to thank Tom Lam (NWMO)

for compiling the properties of the Queenston Shale, Dr. Matt Perras
for providing Fig. 19 and Dr. Evert Hoek for his review comments.

Fig. 23 Comparison of the spalling depth of failure observed at the

Niagara project (high stress and low stress) with other spalling References
measurements
Amann F, Kaiser PK, Button EA (2012) Experimental study of brittle
of the ground, leading to inward movement of the tunnel behavior of clay shale in rapid triaxial compression. Rock Mech
perimeter. It is generally accepted that squeezing and Rock Eng 45(1):21–33
swelling processes may occur simultaneously. For the Armand G, Noiret A, Zghondi J, Seyedi DM (2013) Short- and long-
term behaviors of drifts in the Callovo-Oxfordian claystone at
purposes of this discussion, no distinction is made between the Meuse/Haute-Marne underground research laboratory.
the two processes since it is often not practical to separate J Rock Mech Geotech Eng 5(3):221–230

123
Behaviour of Weak Shales in Underground Environments

Bernier F, Li XL, Bastiaens W (2007) Twenty-five years’ geotech- Opalinus clay—a synopsis of the state of knowledge from Mont
nical observation and testing in the tertiary boom clay formation. Terri. Nagra Report NAB, pp 14–87
Géotechnique 57(2):229–237 Little TE (1989) Construction and performance of a large diameter
Bock H, Dehandschutter B, Martin CD, Mazurek M, de Haller A, test chamber in shale. In: Lo KY (ed) Proceedings of interna-
Skoczylas F, Davy C (2010) Self-sealing of fractures in tional congress on progress and innovation in tunnelling, vol II,
argillaceous formations in the context of geological disposal of Toronto, pp 869–876
radioactive waste. NEA 6184. Nuclear Energy Agency Organ- Macciotta R, Martin CD, Elwood D, Lan H, Vietor T (2012)
isation for Economic Co-Operation and Development Measured convergence at a test tunnel in the Opalinus clayshale
Chern, JC, CW Yu, Shiao FY (1998) Tunnelling in squeezing ground formation. In: CD-ROM proceedings of 21st Canadian rock
and support estimation. In: Proceedings of the regional sympo- mechanics symposium, RockEng 2012, Edmonton
sium on sedimentary rock engineering, Taipei, pp 293–297 Martin CD, Kaiser PK, McCreath DR (1999) Hoek–Brown param-
Cornish LJ, Moore DP (1985) Dam foundation investigations for a eters for predicting the depth of brittle failure around tunnels.
project on soft shale. In: Proceedings of 38th Canadian Can Geotech J 36(1):136–151
geotechnical conference, Edmonton, pp 171–178 Perras MA, Diederichs MS (2015) Observations and numerical back
Einstein HH (1996) Tunnelling in difficult ground-swelling behaviour analysis of an excavation in the Queenston mudstone. In:
and identification of swelling rocks. Rock Mech Rock Eng Proceedings 13th ISRM congress, Montreal, pp 1–12
29(3):113–124 Perras MA, Diederichs MS, Besaw D (2014) Geological and
Gschnitzer E, Goliasch R (2009) TBM modification for challenging geotechnical observations from the Niagara Tunnel Project.
rock conditions—a progress report of the Niagara Tunnel Project Bulletin of Engineering Geology and the Environment, pp 1–21
(NTP). Geomech Tunn 2(2):68–78 Santi PM (2006) Field methods for characterizing weak rock for
Hoek E, Guevara R (2009) Overcoming squeezing in the Yacambú- engineering. Environ Eng Geosci 12(1):1–11
Quibor tunnel. Venezuela. Rock Mech Rock Eng 42(2):389–418 Sargent DW, Cornish LJ (1985) Water susceptibility of Shaftesbury
Hoek E, Marinos P (2000) Predicting tunnel squeezing problems in shale. In: Proceedings of 38th Canadian geotechnical conference,
weak heterogeneous rock masses—part 2: potential squeezing Edmonton, pp 197–206
problems in deep tunnels. Tunn Tunn Int 32(11):45–51 Underwood LB (1967) Classification and identification of shales.
Holtz RD, Kovacs WD, Sheahan TC (2011) An introduction to ASCE J Soil Mech Found Div 93(SM6):97–116
geotechnical engineering, 2nd edn. Pearson/Prentice Hall, Upper Wild KM, Wymann LP, Zimmer S, Thoeny R, Amann F (2014)
Saddle River, NJ Water retention characteristics and state-dependent mechanical
Imrie AS (1991) Stress-induced response from both natural and and petro-physical properties of a clay shale. Rock Mech Rock
construction-related processes in the deepening of the Peace Eng 48(2):427–439
River Valley, BC. Can Geotech J 28(5):719–728
Lanyon GW, Martin CD, Giger S, Marschall P (2014) Development
and evolution of the Excavation damaged zone (EDZ) in the

123