2006:76

LICENTIATE T H E S I S

Behaviour and Stability of Shallow

Underground Constructions

Jimmy Töyrä

Luleå University of Technology

Department of Civil and Environmental Engineering
Division of Mining and Geotechnical Engineering

2006:76|: -1757|: -lic -- 06 ⁄76 -- 

 LICENTIATE THESIS

Behaviour and Stability of Shallow Underground Constructions

Jimmy Töyrä
Division of Mining and Geotechnical Engineering
Luleå University of Technology
SE - 971 87 Luleå
 i

PREFACE

The research presented in this thesis was carried out at the Division of Mining and
Geotechnical Engineering at Luleå University of Technology. It was performed during the
years 2003-2006, under the supervision of Professor Erling Nordlund. The financial support
for the project has been provided by SveBeFo, SBUF and Banverket.

First of all I would like to thank Prof. Erling Nordlund for giving me the chance to do this
work here at the university and for making time for me in his tight schedule. I also want to
thank my reference group, Johnny Sjöberg at Vattenfall Power Consultant AB, Tomas
Franzén Managing Director of SveBeFo, Olle Olofsson at Banverket, Anders Fredriksson at
Golder Associates AB, Beatrice Lindström at WSP Group and Bengt Niklasson and Björn
Stille at Skanska for pointing me in the right direction.

All the people who helped me along the way deserve thanks. Lars Malmgren, David Saiang,
Tomas Villegas, Alexander Bondarchuk and Fredrik Perman have helped me with numerical
modelling. Catrin Edelbro, Kristina Larsson and Andreas Eitzenberger who got me through
all the bureaucracy, Magnus Westblom for playing pool with me, and for all the co-workers at
the institution for the interesting discussions during coffee breaks.

I would like to thank the Trailer Park Boys, Mattias, Jonas and Arvid for the adventures in the
mountains. Kloot has been moved, long live Kloot.

Hotel Tellus, for staying put. It feels like home. The Mattsson family (Inger, Pär, Ida and
Otto) for their support.

Haluan myös kiittää tätejäni Marjattaa, Kirstiä ja Riittaa heidän loppumattomasta avustaan.

My family for giving me the opportunities in life, for the never ending-support, care and love.

Finally, I want to thank my beloved Sara. I’m looking forward to all the adventures that lie
ahead of us.

Luleå, December 2006

Jimmy Töyrä
ii
 iii

SUMMARY

Due to lack of space on the surface in urban areas there is an increased need for underground
constructions. These constructions are often situated at shallow depth. With a better
understanding of which parameters that control the behaviour and stability of shallow
underground constructions the cost efficiency of pre-investigations and the tunnel
construction will be improved.

The objective of this thesis is to identify and describe the important factors that control the
behaviour and the stability of shallow seated tunnels. This knowledge will make it possible to
determine data which needs to be collected and the accuracy necessary to be able to make
reliable analyses. Shallow tunnels are in this work defined as tunnels that have an overburden
of less than 0.5 times the tunnel diameter. This thesis consists of a literature review that
contains studies of the mechanics and design of shallow tunnels. Furthermore, it contains
conceptual numerical analyses as well as numerical analyses of a real case.

The conceptual analyses included factors such as rock mass strength, virgin state of stress, the
location of the tunnel in the rock mass and geological structures. The results show that the
most important factor concerning the stability is large geological structures. Other factors that
can be considered as important are the virgin state of stress and rock mass strength especially
the tensile strength, since failure was primarily in the form of tensile yield. From these results,
a check list was compiled. It is based on how sensitive the behaviour and stability are to
variations of the different factors.

The case study in this thesis is a section of Arlandabanan, Shuttle station 2. It is a shallow
tunnel that is seated underneath Terminal 5 of Arlanda airport. The geology of the rock mass
was characterised by a clay filled structure, and a mica schist with weaker layers. The
comparison between measured deformations and results from the numerical analysis was
partly contradictory. Probable reasons are overestimation of the impact of mica schist
orientation, the clay filled structure, the load from Terminal 5 as well as the conclusions of the
virgin stress drawn from the stress measurements. It also shows that steep structures may be
vulnerable combined with surface loads and low horizontal stresses.
iv
 v

SAMMANFATTNING

I takt med ökad urbanisering minskar möjligheten att nyttja markytan och behoven av
konstruktioner under jord ökar. Genom bättre förståelse av vilka parametrar som kontrollerar
beteendet och stabiliteten för en underjordskonstruktion går det att förbättra
kostnadseffektiviteten i förundersökningar och under tunnelbyggnationen.

Målet med detta projekt är att identifiera och beskriva de faktorer som styr beteendet och
stabiliteten för en ytlig tunnel. Denna kunskap ska göra det möjligt att bestämma vilka indata
som behöver samlas in och vilken noggrannhet som behövs för att kunna göra en relevant
analys. Ytliga tunnlar har i denna rapport definierats som tunnlar vars bergtäckning är 0.5
gånger tunnelns diameter. Denna rapport består av en litteraturstudie som behandlar det
mekaniska beteendet och design av ytligt belägna tunnlar. Rapporten innehåller dessutom
konceptuella numeriska analyser samt numeriska analyser av ett verkligt fall.

I de konceptuella analyserna studerades faktorer såsom bergmassans hållfasthet, det primära

spänningstillståndet, tunnelns placering i bergmassan samt geologiska strukturer. Resultaten
visade att den viktigaste faktorn att ta hänsyn till vid en stabilitetsanalys är stora geologiska
strukturer. Andra faktorer som anses vara viktiga är det primära spänningstillståndet och
bergmassans hållfasthet. Eftersom de brott som förekom i bergmassan var dragbrott, är
bergmassans draghållfasthet en viktig faktor. Baserat på dessa analyser utformades en lista där
faktorernas påverkan på beteendet och stabiliteten beskrevs.

Det verkliga fallet som analyserades var en sektion av Arlandabanan, Shuttle station 2.
Stationen är en ytlig tunnel som ligger under Arlandas Terminal 5. Andra faktorer som fanns i
detta fall var ett lerslag samt skifferplanen i glimmerskiffern som utgjorde bergmassan.
Jämförelsen mellan uppmätta deformationer och resultat från de numeriska analyserna var
delvis motstridiga. Troliga förklaringar till detta är överskattning av betydelsen av
svaghetsplanen i glimmerskiffern, lerslaget och/eller underskattade horisontalspänningar.
Analyserna visade också att branta stupningsvinklar kan vara kritiska när det finns ytlaster
och horisontalspänningen är låg.
vi
 vii

TABLE OF CONTENTS

1 INTRODUCTION............................................................................................................... 1
1.1 Shallow seated tunnels............................................................................................... 1
1.2 Objectives and approach............................................................................................ 2
1.3 Outline of thesis......................................................................................................... 3
2 STABILITY AND PERFORMANCE OF SHALLOW SEATED TUNNELS.................. 5
2.1 Mechanics of shallow seated tunnels......................................................................... 5
2.1.1 Virgin stresses at shallow depth ....................................................................... 5
2.1.2 Induced stresses at shallow depth................................................................... 10
2.1.3 Failure modes and failure mechanisms .......................................................... 14
2.2 Design of tunnels at shallow depth.......................................................................... 15
2.2.1 Stress and deformation analysis ..................................................................... 15
2.2.2 Stability analysis ............................................................................................ 18
2.2.3 Empirical design............................................................................................. 24
3 CONCEPTUAL NUMERICAL ANALYSES.................................................................. 33
3.1 Introduction.............................................................................................................. 33
3.2 Conceptual model factors ........................................................................................ 34
3.3 Input data – Typical Swedish rock mass conditions................................................ 37
3.4 Model development ................................................................................................. 39
3.4.1 Model setup and input data ............................................................................ 40
3.4.2 Comparison of analytical and numerical solution.......................................... 45
3.4.3 Effect of floor inclination ............................................................................... 46
3.5 Parameter study ....................................................................................................... 49
3.5.1 Model setup and input data ............................................................................ 49
3.5.2 Parameter variation ........................................................................................ 53
3.5.3 Instability indicators....................................................................................... 57
3.5.4 Results – Continuum models.......................................................................... 60
3.5.5 Results – Discontinuum models..................................................................... 79
4 IDENTIFICATION OF IMPORTANT FACTORS ......................................................... 85
5 CASE STUDY – ARLANDABANAN ............................................................................ 91
5.1 Introduction.............................................................................................................. 91
5.2 Model setup and input data...................................................................................... 93
5.3 Focus areas of the analysis ...................................................................................... 99
5.4 Results.................................................................................................................... 101
5.4.1 Tangential stress around the tunnel boundary.............................................. 101
5.4.2 Deformation of the tunnel boundary ............................................................ 107
5.4.3 Deformation of the ground surface .............................................................. 111
 viii

5.4.4 The area of plasticity .................................................................................... 115

5.4.5 Extensometer measurements ........................................................................ 119
6 DISCUSSION ................................................................................................................. 121
6.1 Conceptual analysis ............................................................................................... 121
6.2 Case Study ............................................................................................................. 123
6.3 Summary................................................................................................................ 125
7 CONCLUSIONS............................................................................................................. 127
8 RECOMMENDATIONS FOR FUTURE WORK.......................................................... 129
9 REFERENCES................................................................................................................ 131
 1

1 INTRODUCTION

1.1 Shallow seated tunnels

Urbanization leads to an increased need for underground space, for railroads, roads, sewers,
telecommunication and high voltage cables. These constructions are in many cases situated in
urban areas and at shallow depth. Increased environmental demands have resulted in an
increased use of conventional tunnelling technique at shallow depths and a decreased use of
the so-called cut-and-cover technique. Examples of recent and on-going tunnelling projects in
urban areas in Sweden are Arlandabanan, Södra Länken in Stockholm, Götatunneln in
Gothenburg and Citytunneln in Malmö.

The mechanical properties and the state of stress in the shallow portions of the earth’s crust
are much more sensitive to anomalies in the rock mass than at depth. This makes it harder to
estimate and measure these properties. Since it is common to make excavations at relatively
shallow depth, it is important to increase the understanding of the specific problems and the
important factors governing the performance of such constructions in order to improve the
design.

Many factors influence the behaviour and the stability of shallow underground constructions.
Some factors, among others, are the depth at which the excavation is located, the topography
of the ground surface, the reduction of strength and stiffness of the rock mass caused by
weathering, geological structures (forming wedges and blocks or structures that are
continuous and gently dipping which can lead to large destressed volumes of rock), damage
from blasting, loads from buildings at the ground surface and the state of stress in the shallow
portions of the bedrock.

Since the stress magnitude at shallow depth is low around the excavation, the risk of
destressing of the rock mass in the vicinity of the excavation is large. This implies that
gravitational sliding of wedges and blocks are the most probable stability problem. However,
if the overburden is heavily weathered and consequently has a low strength, even a low
absolute stress magnitude may induce failure and stability problems.

The excavation at shallow depth can lead to subsidence on the surface and thereby damage to
buildings in the area. It is very important to minimize the subsidence in urban areas. The load
induced in the rock mass by external loads from buildings above the excavation can also
affect the excavation in an unfavourable way. The optimum dimension of the underground
opening has a large economical and technical importance. For example, if a bearing arch is
 2

formed, in some cases it may not be necessary to redistribute the load from overlying
buildings. The need for reinforcements and the choice of excavation method is also affected
by the behaviour of the rock mass around the tunnel. Some pertinent questions are:
- How accurately must the virgin state of stress be known to be able to perform realistic
analyses?
- How are the results from a stability analysis affected by the precision/uncertainty of
strength, stiffness, geological variations, blast damage, and loads from buildings on
the ground surface?
- How are the results from an analysis of magnitude and location of subsidence affected
by the precision/uncertainty of strength, stiffness, geological variations, blast damage
and load from buildings on the ground surface?
- In what way will the uncertainty of the virgin state of stress, stiffness, strength, and the
other factors affect the design of tunnels and rock caverns and the reinforcement of
these?
- How should the pre-investigations be conducted to decrease the total costs of a
project? What parameters need to be determined with highest precision.

1.2 Objectives and approach

The objective of this thesis is to identify the factors that have most impact on the stability of
shallow tunnels and constructions in hard rock. This will aid in determining which data needs
to be collected and the accuracy needed for a given case so that the design of the tunnel
becomes as cost efficient as possible. This work will also provide understanding of which
factors are important and which combinations are favourable for the stability.

These objectives were accomplished with conceptual numerical analyses, where the different
factors are varied one at a time to find out the impact of each factor on the stability and
behaviour of the tunnel. A case study was also carried out. A section of Shuttle station 2 of
the Arlandabanan was chosen. It is a well known and well documented shallow tunnel in
Sweden that includes many of the factors analysed in the conceptual modelling.
 3

1.3 Outline of thesis

Following this introduction, a literature study is presented in Chapter 2. This will serve as a
basis for the following chapters.

To gain information on how the different factors influence the stability and behaviour of
shallow tunnels conceptual analyses were performed. In Chapter 3, these conceptual
numerical analyses are presented. It starts with a short introduction, followed by a
presentation of the input data used for the reference, or base case. The program FLAC 5.0
(Itasca 2005a) was used for all continuum models. UDEC 4.0 (Itasca 2005b) was used for
some complementary analyses of discontinuum problems. Examine2D (RocScience, 1996) was
used for model development only. A model was developed in FLAC to optimise the
computational time and minimize boundary effects. To do this, results from models with
different sizes and zone sizes were examined and compared to other programs as well as an
analytical solution. This is followed by the actual parameter variation analysis and results
from this.

Based on the preceding chapter, the identification of important factors is presented in Chapter
4. The impact that the different factors have on the stability of a shallow seated tunnel is
discussed.

The analysis of Arlandabanan tunnel section, Shuttle station 2, is presented in Chapter 5.

After a short introduction, the model setup and input data is given. The results from the
analysis as well as measurements conducted at the site while excavating the tunnel section
ends the chapter.

Discussions on both the conceptual analyses and the case study are presented in Chapter 6 and
concluded in Chapter 7. Finally, in Chapter 8, recommendations to future work are given.
4
 5

2 STABILITY AND PERFORMANCE OF SHALLOW SEATED TUNNELS

2.1 Mechanics of shallow seated tunnels

Behaviour and stability of a shallow tunnel does not only include the tunnel itself, it also
includes the behaviour and stability of the ground surface. Deformations on the ground
surface may be of great importance for the infrastructure and buildings in the vicinity of the
tunnel.

2.1.1 Virgin stresses at shallow depth

The state of stress at shallow depth is often complex and seldom measured and reported. This
means that stress measurements reported in the literature are often from a depth greater than
50 m and the stresses for the shallower portions of the bedrock are often estimated by
extrapolation.

The material properties of a rock mass are highly variable parameters. According to Leijon
(1989) the random measuring error in the data obtained with overcoring corresponds to a
standard deviation of r2 MPa in the average normal stress. Considering that the stresses may
be of the same magnitude as the error near the surface it is obvious that the uncertainty in the
measured stresses at shallow depths is great.

The undisturbed state of stress (virgin) at shallow depth is much more sensitive to factors like
weathering, irregular topography, residual stresses, erosion and melting of land ice than at
greater depth. At shallow depth irregular topography may alter both the stress directions and
the stress magnitude, weathering may reduce the already low stresses, geological structures
may create large destressed blocks and zones, and residual stresses can create large anomalies
in the state of stress.

The ratio between the horizontal and the vertical stresses are probably greater than at greater
depth. This is partly because the tectonic stresses at shallow depth represent a larger part of
the total state of stress than the gravitational stresses. Furthermore, the fact that erosion and
melting of the land ice affects the ratio at shallow depths contributes more to a higher
horizontal to vertical stress ratio.

The stresses that exist in the rock mass are related to the weight of the overburden but also to
its geological history. Knowledge of the in situ state of stress in a rock mass is important in
civil and mining engineering. The stress magnitudes in the rock mass generally increase with
 6

depth. Consequently, stress-related problems such as failure due to high stress magnitude also
increase with depth. However, excavations at shallow depth may also be challenging, either
because of high horizontal stresses or due to the lack of horizontal stresses (Amadei and
Stephansson, 1997).

Rock stress can be divided into virgin stress and induced stress. The virgin (or in situ) stresses
are those existing before any artificial disturbance, while induced stresses are created by
artificial disturbances like excavation and drilling or induced by changes in natural conditions
such as drying and swelling.

The virgin state of stress is normally described with one vertical stress and two horizontal
stresses that are denoted Vv, Vh and VH. The vertical principal stress is usually a result of the
weight of the overburden per unit area above a specific point in the rock mass and is normally
assumed to be a function of depth and is defined as

Vv Ugz (2.1)

where U is the density of the rock mass (kg/m3), g is the gravity acceleration (m/s2), and z is
the depth below ground surface (m).

The horizontal component due to gravitational loads depends on the rock mass properties. If
the material can be considered linear-elastic and isotropic and a one-dimensional state of
strain is assumed, the average horizontal stress is defined according to the theory of basic
elasticity as

Q
VH Vv (2.2)
1 Q

where Q is Poisson’s ratio ranging from 0.15 to 0.35 for most rock types, with a common
value of 0.25.

Equation (2.2) assumes that the rock mass is isotropic, but this is seldom true on a larger
scale. Anisotropy is common in rock and can be a result of bedding, stratification, schistose
planes, or jointing. Anisotropy is often divided into transversely isotropy and orthotropic
anisotropy. A transversely isotropic material is a material that has different properties in two
directions, i.e., there are isotropic planes, while an orthotropic material has different
properties in all three directions, i.e. there are three orthogonal planes of elastic symmetry.
 7

Factors that affect the horizontal stresses

A contribution mainly to the horizontal stresses are the tectonic forces that come from the
interaction of the tectonic plates, with each other and with the earth’s mantle resulting in
boundary forces between the plates. Two groups of forces are responsible for the tectonic
stresses, namely broad-scale tectonic forces and local tectonic stresses. The broad-scale
tectonic forces are forces acting in lithospheric plate boundaries such as shear tractions at the
base of the lithosphere, slab pull at subduction zones, ridge push from oceanic ridges and
trench suction. Local tectonic stresses are related to bending of the lithosphere due to surface
loads, isostatic compensation and downbending of oceanic lithosphere. The tectonic stresses
are constant in areas where the length and width are several times the thickness of the elastic
part of the lithosphere (Zoback et al., 1989). Tectonic forces will give a constant contribution
which means that a horizontal stress will exist that is non-zero at ground level and that the
stress will increase with the depth due to the gravitational stresses. In Scandinavia it is
principally the driving forces from the accretion zone (the Mid-Atlantic ridge) that contribute
to the horizontal stresses.

Another form of difference in elastic constants is anisotropy in which the values of

deformation modulus and Poisson’s ratio may vary with direction of a homogeneous rock
mass. Anisotropy can influence both the magnitude and orientation of local principal in situ
stresses and should not be ignored in the evaluation of stress measurement data (Amadei,
1983). This is not dealt with further in this work

Under periods of time the crust has been loaded by land ice, layers of sedimentary rock and
sediments, all of them several kilometres thick. The crust has been deformed plastically from
the load of these layers and adjusted to the present state of stress. When the rock and the
sediments erode and when the ice melts the vertical and horizontal stresses decrease.It is
considered that erosion and melting of land ice results in an increase the magnitude of VH
relative to the magnitude of Vv.

An irregular topography affects the magnitude and direction of the virgin stresses, see Figure
2.1. The influence of the topography is large near the surface and will decrease with depth and
a rough estimation can be obtained by examining hills and valleys with compressive
respectively tension loads on a flat surface. The stresses that arise when a half space is loaded
with a uniformly distributed or a linear varying load can be solved through integration of the
solution of a linear load.
 8

A)

B)

Figure 2.1 The effect of the primary stresses due to an irregular topography can be
estimated by replacing hills and valleys with linear varying loads.

Rock masses are rarely uniform and variations in geology and the existence of geologic
structures and heterogeneities may affect the distribution and magnitude of in situ stresses and
contribute to the scatter often observed in field measurements (Fairhurst, 1986). Hudson and
Cooling (1988) identified three cases depending on the relative stiffness of the material in the
discontinuity versus the material in the surrounding rock; (1) if the discontinuity is open, the
major principal stress is diverted parallel to the discontinuity, (2) if the discontinuity have
similar properties as the surrounding rock, the principal stresses are unaffected, and (3) if the
material of the discontinuity is rigid, the major principal stress is diverted perpendicular to the
discontinuity. Geological structures are significant for the state of stress at shallow depths. An
open discontinuity above a tunnel can lead to a destressed zone in the roof so that a bearing
arch above the tunnel cannot be formed.

State of stress in Scandinavia

In most cases in Scandinavia, the stress field corresponds to a thrust faulting (ıH> ıh< ıv) or
to a strike-slip faulting (ıH> ıv> ıh) with the general direction of the major horizontal stress
of NW-SE. This is considered to originate from the Mid-Atlantic Ridge and from the
interacton between the African and the European plate (Berg, 2005).
 9

Stress measurements by hydraulic fracturing and overcoring in Scandinavia analyzed by

linear regression of the principal horizontal stress versus depth have been summarized by
Stephansson (1993) and can be seen in Table 2.1 and Figure 2.2. VH is the maximum
horizontal stress, Vh is the minimum horizontal stress and z is the depth.

Table 2.1 Variation of horizontal stress components with depth in Fennoscandia,

Stephansson (1993)

Variation of ıH and ıh (MPa) with depth (m) Type of measurement and depth range (m)
1. V H 10.4  0.0446 z Leeman-Hiltscher overcoring (0-700)
V h 5  0.0286 z
2. VH 6.7  0.0444 z Leeman-type overcoring (0-1000)
Vh 0.8  0.0399 z
3. VH 2.8  0.04 z Hydraulic fracturing (0-1000)
Vh 2.2  0.024 z

VH (MPa) Vh (MPa)
0 4 8 12 16 0 4 8 12 16
0
Base case
VH=6.7+0.0444z
Vh=0.8+0.0329z
VH=10.4+0.0446z
20 Vh=5+0.0286z
VH=2.8+0.0399z
Depth (m)

Vh=2.2+0.0024z

40

60

Figure 2.2 Principal states of stress for different measurement methods, compiled by
Stephansson (1993).

Berg (2005) has studied results from stress measurements conducted in Sweden at shallow
depth (10 – 50 m) with the overcoring technique from four different sites. The results showed
that stress measurements at shallow depth give a relatively large scatter compared to stress
 10

measurements at greater depth, both within a measurement level, between different levels and
at different sites.

Furthermore, Berg (2005) compared the results from the study with the regression line
presented by Stephansson (1993), see Table 2.1 (3) and Figure 2.2. This comparison showed
that the regression line did not fit the shallow stress measurements very well, although it
cannot be said whether it overestimates or underestimates the stress conditions at shallow
depth.

2.1.2 Induced stresses at shallow depth

When a tunnel is excavated, the rock that is left around the open space has to carry more load
since the support from the rock that existed earlier in the now excavated cavity has been
removed. The redistributed stresses are called secondary or induced stresses.
If the horizontal stresses are assumed to be greater than the vertical stresses, the redistributed
stresses will form major principal stress trajectories that will have the shape as shown in
Figure 2.3.

Shape of tunnel

The low stress levels at shallow depths are particularly critical if the shape of the tunnel is
badly chosen, which could lead to unnecessarily low stress levels in the tunnel roof.
“Streamlined” sections in the direction of the major principal virgin stress are preferred to
minimize the risk of a destressed roof, while a tunnel section with a flat roof will redistribute
the stresses away from the tunnel boundary into the rock, see Figure 2.3.

a) b)

Figure 2.3 Stress trajectories around a) an arched roof and b) a flat roof.
 11

When the virgin horizontal stress is greater than the virgin vertical stress, like it usually is at
shallow depth in Scandinavia, the stress situation can lead to destressed tunnel walls. A
rounded tunnel shape will decrease the risk of destressed tunnel walls.

Less overburden will in most cases result in lower stresses around the tunnel. Thus, the risk of
slip along pre-existing geological structures and fallouts in the roof and walls of the
excavation increases. A thin overburden could also mean that concentrations of horizontal
stresses can occur above the excavation, leading to stress-induced failure.

Weathering, topography and geological structures

If the overburden is weathered and has decreased strength and stiffness, the horizontal stress
trajectories will most likely be redistributed from the ground surface and be concentrated
beneath the tunnel instead of in the tunnel roof. This leads to a destressed tunnel roof, see
Figure 2.4.

a) b)
Weathered rock Weathered rock

Fresh rock
Fresh rock

Figure 2.4 The stress situation for a shallow seated tunnel in weathered rock, a) the stress
trajectories and b) the distribution of the virgin horizontal stresses.

As discussed earlier, an irregular topography affects the magnitude and direction of the virgin
stresses. For example, a valley above a tunnel may result in stress concentrations above the
tunnel, which can lead to stress-induced failure in the roof of the tunnel, see Figure 2.5. A hill
above the tunnel may lead to larger vertical stresses and unchanged horizontal stresses
compared to a flat overburden. An irregular topography can also lead to an extremely thin
overburden.
 12

Figure 2.5 The stress trajectories of a tunnel beneath a valley.

Geological structures such as a destressed block in the form of a nappe or an open

discontinuity can create a stress shadow and if larger volumes are isolated it can lead to large
volumes of destressed rock, see Figure 2.6. These areas of destressed rock are very sensitive
to gravitational fallouts of blocks and wedges.

Destressed block

a)

Fresh rock

b)
Discontinuity
Fresh rock

Figure 2.6 Geological structures that can lead to large volumes of destressed rock, a) a
destressed block in form of a nappe, b) an open discontinuity.

Loading from foundations

The stresses in the rock caused by a load at the ground surface can be derived using linear
elastic theory and the solution by Boussinesq (1883) for a point load applied to a semi-infinite
body. A number of solutions for different loading situations are presented in the rock
mechanics literature for isotropic as well as anisotropic rock masses (see for instance Poulos
and Davis, 1974, and Gaziev and Erlikhman, 1971).
 13

The effect of the load located on the ground surface on the stability of shallow seated
underground excavations depends on whether a bearing arch can form above the roof of the
tunnel or not. The effect of the surface load on the stability of the tunnel and subsidence of the
ground surface, therefore, decreases with increasing virgin horizontal stress.

Damaged rock around the tunnel

The stability of an underground structure is very much dependent on the integrity of the rock
immediately surrounding the excavation. Gravity driven fallouts from the roof of the tunnel
are especially related to the interlocking of the blocks formed in the roof. Blast damage from
unrestricted blasting can extend several meters into the rock mass and the zone of loosened
rock can lead to instability problems in the rock surrounding the excavation, especially in
shallow seated tunnels where the overburden is limited.

When excavating a tunnel, the rock in the immediate surrounding of the tunnel will have, in
addition to the naturally existing cracks and joints, cracks that are caused by blasting and re-
distribution of stresses. These new cracks and joints will probably decrease the stiffness and
the strength of the rock which suggests that the ability to attract stresses will also be reduced,
see Figure 2.7.

Blast damage

Figure 2.7 The stress and the strength of damage caused by the excavation.

The excavation damaged zone at the Underground Research Laboratory (URL) in Canada was
estimated by Martino (2002) through seismic and permeability measurements. The
measurements were made on two tunnels at different depths (240 and 420 m). Both tunnels
showed similar patterns of excavation damage, with both tunnels having increased damage
 14

immediately surrounding the tunnels. The most highly damaged inner zone (defined as the
zone where sharp changes in the rock mass properties and visible cracks have occurred) was
relatively narrow, 0.1 to 0.3 m in width. A less damaged outer zone (defined as the zone
where gradual changes in the rock mass properties have occurred), where the measured values
return towards background values, surrounds the inner zone. In the tunnel located at a depth
of 240 m, the outer zone was smaller than for the tunnel located at a depth of 420 m. The
difference is believed to be caused by the difference in the magnitude of the in situ stresses
surrounding the two tunnels, which increases the damage around the deeper tunnel. Martino
(2002) assumed that, for the tunnel at greater depth the damage is caused by both the blasting
and the altered stress conditions while the damage for the shallower tunnel is only caused by
blasting since the stress level is relatively low. The total damaged zone around the shallower
tunnel varies between 0.2 and 0.5 m.

The typical background S-wave velocity for the upper tunnel was 3200 m/s, with a decrease
of 400 m/s in the inner damaged zone. The background velocity of the P-wave around the
same tunnel was 5400 - 5600 m/s and decreased from 400 – 1300 m/s in the inner damaged
zone. The reductions in the seismic velocities would correspond to a reduction of 35 – 40% in
Young’s modulus (if the density is assumed to be constant).

Tests were also performed at the Äspö Hard Rock Laboratory (HLR) in Sweden by Emsley et
al. (1997) to estimate the damage and disturbance caused by tunnel excavation. Several
seismic techniques where used to measure the rock properties in the damaged zone. All
seismic methods showed a significant decrease in velocity close to the tunnel excavated by
drilling and blasting. The zone with reduced velocities was 0.3 m up to 1 m occasionally.

2.1.3 Failure modes and failure mechanisms

The rock mass has a complex behaviour due to the existence of natural discontinuities. The
mechanical behaviour of the rock mass, therefore, depends on the mechanical behaviour of
blocks and joints. Depending on the induced state of stress around the excavation different
kinds of stability problems can occur.

Stability problems in jointed rock masses with low stress magnitudes are generally associated
with gravity driven fallouts of wedges and blocks from the roof and sidewalls. This failure
process is controlled by the three-dimensional geometry of the tunnel and the rock structure
(Hoek and Brown, 1980). There must exist at least three joint surfaces that separate the wedge
from the surrounding rock mass to form a wedge in an excavation, see Figure 2.8.
 15

Figure 2.8 Gravity driven wedges and blocks that can form in the tunnel roof and wall.

Stress concentrations above the tunnel due to, for example, irregular topography could cause
stress-induced failures. If the rock contains few joints and is brittle, spalling, and in extreme
cases, rock bursts may occur.

For intermediate stress conditions, a bearing arch can be formed but the stresses may not be
large enough for stress induced failure to occur. In Table 2.2 failure mechanisms for different
stress magnitudes in shallow tunnelling are listed.

Table 2.2 Failure mechanisms for different states of stress in shallow tunnelling.

Low stresses and jointed Intermediate stresses High stresses, few joints
rock and brittle rock
- Fallouts of wedges and Stresses are enough for a - Stress induced failure
blocks bearing arch to form. - Spalling
- Beam failure - Rock burst
- Ravelling

2.2 Design of tunnels at shallow depth

2.2.1 Stress and deformation analysis

Analytical stress analysis

 16

General closed-form solutions for tunnels with circular cross-section at shallow depth were
presented by Mindlin (1939, 1948). The solutions presented gave the expressions for the
tangential stresses at the boundary of the circular opening and at the free surface. Mindlin
presented solutions for the following loading cases

a) a tunnel at shallow depth in a uniaxial state of virgin stress (1948),

b) a tunnel excavated in a rock mass with a gravitational virgin stress state (1939).

The solutions are, however, very complex and the tangential stress is expressed in terms of
infinite series and is therefore not presented here.

In 1939 Mindlin presented a solution for the tangential stress at the boundary of a tunnel with
a circular cross-section excavated in a rock mass with a gravitational state of stress. He
considered three different cases,
- hydrostatic state of stress
- uniaxial state of strain
- no horizontal restraints, that is, no stresses in the horizontal direction.

No numerical examples are presented here. Although the solutions are complex, they may still
be used practically to calculate the tangential stresses around a circular opening. Closed-form
solutions for other cross-sections have not been found in the literature.

Numerical methods

The rock mass comprising the Earth’s upper crust is a discrete system and even though
analytical methods may be useful for evaluating the effect of many factors, closed-form
solutions do not exist for all geometries. Numerical methods must, therefore, be used to study
practical problems. Even though the results obtained by numerical analyses are not exact, the
accuracy is sufficient for practical design applications, in particular when considering other
uncertainties of the rock mass. Due to differences in the underlying material assumptions,
different numerical methods have been developed for continuous and discontinuous problems,
see Table 2.3 (Jing, 2003).
 17

Table 2.3 Different numerical methods, after Jing (2003).

Continuous methods Discontinuous methods

The finite difference method (FDM) The discrete element method (DEM)
The finite element method (FEM) The discrete fracture network method (DFN)
The boundary element method (BEM)

A continuum approach is relevant for intact rock conditions as well as for cases when the
discontinuities are so pervasive and closely spaced relative to the size of the problem domain
that the rock mass can be represented as a continuum with equivalent rock mass properties,
see Figure 2.9. This can be a rock mass property as well as a scale effect. If the rock mass
contains either a few joints or a large number of closely spaced joints in several directions, the
behaviour will be more or less continuous. If, for a given rock mass, the opening of interest is
small or large (with the joint oriented in all directions) compared to the spacing of the joints
the behaviour can also be expected to be continuous.

A discontinuum problem is, for example, when the rock mass of interest consists of a number
of discrete, interacting blocks. The intact rock and the discontinuities are described separately;
see Figure 2.9. In these models the rock mass movements are described with deformation of
the intact rock, slips along joint surfaces, separation, and rotation.

Figure 2.9 Examples of continuous and discontinuous rock masses (Edelbro, 2003).

In the continuum approach normally linear elastic behaviour is assumed for stress levels
below strength or yield limit. The result of such analyses can be expressed in terms of
stresses, infinitesimal strains and displacements. When the load on the rock exceeds the
strength or yield limit, linear elastic behaviour will no longer give relevant stresses and
deformations. Constitutive models simulating plastic behaviour can then be used. Because
plastic deformations mean permanent displacement changes within the material, plastic strains
cannot be defined uniquely in terms of the current state of stress. Plastic strain depends on the
 18

loading history so the theory of plasticity must use an incremental loading approach in which
incremental deformations are summed to obtain the total plastic deformation.

The continuum methods can also be divided into groups depending on the way the problem is
solved. There are mainly two different approaches, the boundary element approach and the
finite element/finite difference approach. In the boundary element method, all boundaries are
discretised. The result is the exact solution on the boundary to the used discretisation of the
problem. Finer elements will give a result which better represents the solution of the original
problem. The boundary element method has limitations. The use of different material
behaviour in different parts of a model is often not possible. Plastic models are not always
available. For the group of methods based on finite element and finite difference formulations,
the whole model is discretised, i.e., all material. The size of the elements with respect to the
scale of the problem is crucial for the accuracy of the results. All kinds of constitutive models,
such as elastic, plastic, isotropic and anisotropic, are generally available for these methods.

The numerical methods and programs used for the study of shallow seated tunnels must have
the ability to consider the ground surface and to model the factors of interest. Continuous
models will be used to study the behaviour of a closely jointed rock mass, the effect of
weathering of the rock mass, and the damaged zone around the tunnel. The discontinuous
models can be used in order to be able to study the effect of individual large- and small-scale
discontinuities. Both types of models should also have the ability to model a varying
topography, different virgin states of stress, different overburden, and loading applied to the
ground surface. Since the stresses may exceed the strength in some cases the models should
also be able to simulate non-elastic behaviour.

2.2.2 Stability analysis

Assessing failure mode

Modes of structurally controlled failure can be analysed by the means of the hemisspherical
projection technique. For a wedge to form in the roof or the walls of a tunnel, at least three
joint planes must exist that separate the wedge from the rock mass. This will be visible on the
stereonet by the great circles of three joint planes intersecting each other and forming a closed
area in the stereo net.

The wedge analysis using hemispherical projection is a method that only considers if the
kinematic conditions for a block/wedge are such that sliding can occur. It also considers if the
sliding resistance is greater or less than the driving force, under the assumptions that the joint
 19

strength is purely frictional. This methodology can, however, be very conservative, since a
small contribution from the stresses in the vicinity of the opening to the normal stresses acting
on a joint, may increase the stability considerably.

Block analysis

While hemispherical graphic projections may locate and to some extent judge whether the
wedge is stable or not, there are also methods to calculate the safety factor against wedge
failure. The factor of safety and an estimation of the weight of the formed wedges can be
obtained through an analysis with, for example, the program UNWEDGE (Hoek et al., 1995).
The calculation used to determine the wedges assumes that the discontinuities are ubiquitous,
in other words that the wedges can occur anywhere in the rock mass. Furthermore, the
program assumes that the joints, bedding planes and other structural features included in the
analysis are planar and continuous. This means that the program will always detect the largest
wedges, which can seem conservative since the size of the wedges will be limited by the
persistence and spacing of the structural features. However, the program will allow wedges to
be scaled down to more realistic sizes if desired. By typing in the density of the rock, the
friction angle, and the cohesion of the joints, the safety factor against wedge failure for the
different wedges is obtained.

Compressed arch action - Voussoir Beam Theory

Since the absolute stress magnitudes at shallow depth are generally low, the stability of the
underground openings is governed by the possibility of blocks and wedges sliding or falling
into the opening. The Voussoir beam theory or compressed arch theory is a method which
considers the state of stress around the opening in the equilibrium analysis of blocks and
wedges.

Compressed arch action

An arch is a construction which mainly transfer load as compressive force. The classical arch
is a number of blocks arranged in such a way that the joints transfer only compressive forces
(Stille et al., 2004).
 20

Figure 2.10 Transfer of load in an arch.

An arch can collapse in three ways

- sliding along joints,
- crushing of the joint surfaces or the block material, or
- rotation of blocks.

Sliding along a joint means that the shear stress exceeds the strength of the joint. Crushing or
spalling occur when the compressive stress exceeds the compressive strength. Rotation of the
blocks takes place if the load induces a bending moment causing a tensile stress in a joint with
no tensile strength.

The compressed arch for a certain load q(x), is obtained (Stille et al., 2004) by integration of

q ( x)
y cc  (2.3)
H

where H is the horizontal support reaction according to Figure 2.11.

 21

q(x)

f Compressed
y
arch

x
H L

Figure 2.11 Compressed arch.

The expression for the compressed arch can be described as,

ª 2 x 2 º
y f «1  » (2.4)
¬ L ¼

and maximum load, q, which a compressed arch can carry, is determined by

8Hf
q max (2.5)
L2

which depends on the geometry of the arch (L and f) and the lateral pressure (horizontal
support reaction, H).

Voussoir beam theory

The Voussoir beam theory also deals with the development of a compression arch. There are
several versions of this model but we will refer here to the one presented by Diederichs
(1999). The model considers deflection due to self weight, external loads such as load from
the rock above, water pressure and support, and the deformability of the beam.

Rock masses characterized by parallel laminations are often encountered in underground

excavations. Laminated rock can be a result of sedimentary layering, extensile jointing, and
rock masses created through metamorphic or igneous flow processes or through stress
 22

fracturing parallel to the excavation of massive ground. This structure can be the main factor
controlling the stability of roofs of civil excavations or mines.

In most cases the rock mass is not only represented by the lamination partings but also by
joint sets cutting through the laminations (Figure 2.12). These joint sets reduce the ability to
sustain boundary parallel tensile stresses such as those assumed in conventional beam theory.
However, if these joints cut through the laminations at steep angles or if reinforcement has
been installed, one can assume that a compression arch can be generated within the beam
which will transmit the beam loads to the abutments (Diedrichs, 1999).

Figure 2.12 a) Jointed rock beam b) Voussoir beam analogue (Diederichs, 1999).

The primary modes of failure in the Voussoir beam model are buckling or snap-through
failure, lateral compressive failure (crushing) at the midspan and abutments, abutment slip
and diagonal fracturing, see Figure 2.13. Abutment slip or shear failure (Figure 2.13c) is
observed when the ratio between span and thickness is low (i.e., thick beams), while crushing
(Figure 2.13b) and snap-through (Figure 2.13a) are observed in thinner beams. Ran et al.
(1994) showed that if the angle between the cross cutting joints and the normal to the
lamination plane is less than 30 to 50 % of the effective friction angle of the joints then the
application of the Voussoir beam theory is valid. If the angle between the cross cutting joints
and the normal of the lamination plane is larger than 30 to 50 % of the effective friction, then
slip along the cross cutting joints and premature shear failure of the beam is likely to occur.

Stimpson and Ahmed (1992) have shown in a physical model of thick laminations that
external loading can produce diagonal tensile fracturing (Figure 2.13d) that propagates
parallel to the compression arch, from the upper midspan to the lower abutments.
 23

Figure 2.13 Failure modes of the Voussoir beam. a) snap-through, b) crushing at the
midspan and abutments, c) abutment slip and d) diagonal fracturing (Diederichs,
1999).
 24

2.2.3 Empirical design

Classification and characterisation

The rock mass is a complex composition of rock and discontinuities and it may be difficult to
determine whether a certain rock mass is suitable for construction work or not. Analytical and
numerical analyses of rock mechanical problems often needs simplifications or a large
quantity of input that might be hard to determine.

Numerous classification/characterisation systems have been developed to judge the usefulness

of the rock mass for different constructions. These systems combine the defined properties of
the rock mass with practical experience from different kinds of constructions in rock.

In practice, the difference between the process of classification and characterisation of the
rock mass is not that large. The difference as such that the rock mass characterisation is
describes the rock with emphasis on colour, shape, weight, properties etc, while rock mass
classification is when one arranges and combines different features of a rock mass into
different groups or classes following a specific system or principle (Edelbro, 2003). It is the
descriptive terms that constitute the main difference between characterisation and
classification. For detailed information, see for instance Palmström (1995).

Rock mass classification is today the most commonly used tool in preliminary design of rock
excavations and in assessment of reinforcement. Rock mass classification can also be used for
estimating deformation properties and strength of the rock mass, either directly (such as the
Q-system) or indirectly (such as Hoek-Brown-RMR). Some of the classification systems have
been modified to provide input data that describes the rock mass conditions in the Hoek-
Brown and Mohr-Coulomb failure criteria.

RMR (Bieniawski 1974) and the Q-system (Barton et. al., 1974) are probably the most
commonly used rock mass classification systems in Sweden today. Both were developed for
designing tunnels and rock caverns, but their area of application has expanded to both mining
applications and slopes. There are also systems, such as GSI (Hoek et al., 1995) and RMS
(Stille et al. 1982). For more information see for example Edelbro et al. (2007).

The classification systems are made up of a number of basic parameters that are estimated in
the field. These parameters are rated using tables. The parameters used for RMR and Q are
presented in Table 2.4. In RMR, the ratings of the parameters are summed to obtain the RMR
rating of the rock mass while the Q-system uses a multiplicative function
 25

RQD J r J w
Q . (2.6)
J n J a SRF

Table 2.4 The basic parameters used for the RMR and Q systems.

RMR Q
1. Unaxial compressive strength of intact rock 1. RQD: Rock Quality Designation
material 2. Jn: joint set number
2. Rock quality designation (RQD) 3. Jr: joint roughness number
3. Ground water conditions 4. Ja: joint alteration number
4. Joint or discontinuity spacing 5. Jw: joint water and pressure reduction factor
5. Joint characteristics 6. SRF: stress reduction factor-rating for faulting,
strength/stress ratios in hard massive rocks, and
squeezing and swelling rock

The achieved classification value is then used to roughly determine the quality of the rock
mass, see Table 2.5..

Table 2.5 Geomechanical classification (RMR) of rock masses (Bieniawski, 1974, Hoek
and Brown, 1980) and Q-value and rock mass quality (Barton et al., 1974).

Sum of Class Description of rock mass Q-value Rock mass quality for tunnelling
rating
increments
81 - 100 I Very good rock 0.001 - 0.01 Exceptionally poor
61 - 80 II Good rock 0.01 - 0.1 Extremely poor
41 – 60 III Fair rock 0.1 - 1.0 Very poor
21 - 40 IV Poor rock 1-4 Poor
<20 V Very poor rock 4 - 10 Fair
10 - 40 Good
40 - 100 Very good
100 - 400 Extremely good
> 400 Exceptionally good
 26

The rock mass classification systems can also be used to estimate rock mass parameters for
further analysis of constructions in rock. For example, the in situ deformation modulus of a
rock mass is an important parameter in any form of numerical analysis. Bienawski (1976) and
Serafim and Pereira (1983) studied case histories and both proposed a relationship for
estimating the in situ deformation modulus, Em, from RMR, see Table 2.6 (1) and (2), and
Figure 2.14.

Barton et al. (1980), Barton et al. (1992) and Grimstad and Barton (1993) proposed a
relationship between the Q-system and the deformation modulus, which found good
agreement between measured displacements and predictions from numerical analyses, see
Table 2.6 (3) and Figure 2.14.

Table 2.6 Relationships for estimation of the in situ deformation modulus from RMR and
Q.

( RMR 10 )
1 Em 10 40 Serafim and Pereira (1983)
2 Em 2 RMR  100 Bienawski (1978)

E m 25 log10 Q Barton et al. (1980), Barton et al. (1992) and

3
Grimstad and Barton (1993)
 27

Figure 2.14 Relationships for estimation of the in situ deformation modulus from RMR and
Q. (Hoek et al., 1995)

Rocscience Inc. has developed a program, RocLab (RocScience, 2006), for determining rock
mass strength parameters based on the Hoek-Brown failure criterion in conjuction with GSI
(Hoek et al. 2002). RocLab uses the Geological Strength Index (GSI), the unconfined
compressive strength of intact rock, ıci, the intact rock parameter mi and the disturbance factor
D to determine the generalized Hoek-Brown strength parameters. In this report, this method
was used to determine material parameters for the Mohr-Coulomb failure criterion.

Hoek et al., (1995) introduced the Geological Strength Index, as a complement to overcome
the deficiencies in RMR for very poor quality rock masses. GSI estimates the reduction in
rock mass strength for different geological conditions (Edelbro, 2004).

There are three ways of calculating the GSI:

1. By using the rock mass rating for good quality rock masses (GSI > 25)

For RMR76' > 18

GSI = RMR76' (2.7)

 28

For RMR89' > 23

GSI = RMR89' -5 (2.8)

For both versions, dry conditions should be assumed — i.e., assigning a rating of 10 in
RMR76' and a rating of 15 in RMR89' for the groundwater. In addition, no adjustments for
joint orientation (very favourable) should be made, since the water condition and joint
orientation should be assessed during the rock mass analysis (Hoek et al., 1995). Since it
is difficult to estimate RMR for very poor quality rock masses, Hoek et al., (1995)
suggested that the Q-system (Barton et al., 1974) should be used for these circumstances,
see below.

2. By using the Q-system

For all Q-values:

GSI 9 ln Q'  44 . (2.9)

In doing this, both the joint water reduction factor (Jw) and the stress reduction factor
(SRF) should be set to 1.

3. By using GSI-classification.

Hoek and Brown (1997) did not specifically recommend the use of the Q-system; rather
they recommended using GSI-classification, see Table 2.7 (Hoek et al., 1995). The aim of
the GSI system is to determine the properties of the undisturbed rock mass; otherwise,
compensation must be made for the lower GSI-values obtained from such locations.
 29

Table 2.7 Estimation of Geological Strength Index (GSI) (Hoek et al, 1997).

Guidelines for excavation and support have been developed for both RMR (see Bienawski,
1989) and the Q-system (see Grimstad and Barton, 1993). An example of how RMR is used to
 30

estimate the needed support is presented in Table 2.8, where the excavation and support is
presented for a 10 m span for different RMR-values.

Table 2.8 Guidelines for tunnelling and support of 10 m span following the RMR system
(adapted from Bienawski, 1989)

Rock Mass Class Excavation Rock bolts Shotcrete Steel sets

(20 mm diameter, fully
grouted)
RMR:81-100 Full face, 3 m advance Generally no support required except spot bolting
RMR: 61-80 Full face, 1-1.5 m advance. Complete Locally, bolts in crown 3 m 50 mm in None
support 20 m from face. long, spaced 2.5 m with crown where
occasional wire mesh required
RMR: 41-60 Top heading and bench, 1.5-3 m Systematic bolts, 4m long, 50 mm in None
advance in top heading. Commence spaced 1.5-2 m with crown and 30
support after each blast. Complete occasional wire mesh. mm in sides.
support 10 m from face.
RMR: 21-40 Top heading and bench, 1-1.5 m Systematic bolts 4-5 m long, 100-150 mm in Light to mediu ribs
advance in top heading. Install support spaced 1-1.5 m in crown and crown and 100 spaced 1.5 m where
concurrently with excavation, 10m walls with wire mesh. mm in sides. required.
from face.
RMR: <20 Multiple drifts 0.5-1.5 m advance in Systematic bolts 5-6 m long, 150-200 mm in Medium to heavy
top heading. Install support spaced 1-1.5 m in crown and crown, 150 mm ribs spaced 0.75 m
concurrently with excavation. walls with wire mesh. Bolt in sides and 50 with steel lagging
Shotcrete as soon as possible after invert. mm on face. and forepoling if
blasting. required. Close
invert.

The Q-system uses a ratio between the tunnel span and the Excavation Support Ratio (ESR)
which is a value related to the intended use of the excavation and ranges between 3
(temporary mine opening) and 0.8 (underground nuclear power stations etc.) and the Q-value
to determine the support needed for the excavation, see Figure 2.15.
 31

Reinforcement categories
1. Unsupported 6. Fibre reinforced shotcrete, 90-120mm, bolting
2. Spot bolting 7. Fibre reinforced shotcrete, 120-150mm, bolting
3. Systematic bolning 8. Fibre reinforced shotcrete, >150mm, with
reinforced ribs of shotcrete, bolting
4. Systematic bolting with 40-100mm 9. Cast concrete lining.
unreinforced shotcrete
5. Fibre reinforced shotcrete, 50-90mm, bolting

Figure 2.15 Support categories based on the Q-system (after Grimstad and Barton 1993).
32
 33

3 CONCEPTUAL NUMERICAL ANALYSES

3.1 Introduction

To gain a better understanding of the performance of shallow seated underground

constructions, conceptual numerical analyses were conducted. These analyses were carried
out using the programs FLAC, UDEC and Examine2D. FLAC (Fast Lagrangian Analysis of
Continua) is a two-dimensional explicit finite difference program for engineering mechanics
computation and UDEC (Universal Distinct Element Code) is a two-dimensional numerical
program based on the distinct element method for discontinuum modelling. Both programs
simulate both elastic and plastic behaviour of rock and soil. In FLAC the materials are
represented by zones which form a grid that can be adjusted to fit the shape of the structure to
be modelled. The elements behave according to a prescribed linear or nonlinear stress/strain
law in response to the applied forces or boundary restraints. The material can yield and flow,
and the grid is deformable (Itasca, 2005a). FLAC is used in this project when the rock mass is
considered to be homogenous and isotropic.

UDEC simulates the response of discontinuous materials, such as jointed rock masses,
subjected to either static or dynamic loading. The discontinuous materials are represented as a
number of discrete blocks. The discontinuities are treated as boundary conditions between
blocks. Individual blocks behave as either rigid or deformable. The deformable blocks are
divided into a mesh of finite-difference zones, and each zone responds according to a
prescribed linear or nonlinear stress-strain law (Itasca, 2005b). UDEC will be used in this
project to study the impact on the stability given from large geological structures such as a
large single joint transecting the overburden.

Examine2D is a two-dimensional indirect boundary element and displacement discontinuity

program for calculation of stresses and displacements around underground and surface
excavations in rock (Rock Engineering Group, 1996). Since the material cannot yield in
Examine2D, this program has only been used in the development of a satisfying FLAC model.

The impact of the different factors is analyzed using a conceptual model where the behaviour
of the tunnel is studied. This involves a series of numerical simulations in which each
parameter is varied. The design of the models and the used input data for all analyses has
aimed at as much as possible resemble real case parameters and the design of a shallow
tunnel. However, it is important to remember that the results should only be seen as trends
showing the impact of different parameters.
 34

3.2 Conceptual model factors

There are a large number of factors that influence the stability of a tunnel at shallow depth.
Stability problems occur when the stress exceeds the strength of the rock mass (both in
compression and tension) and/or fallouts and sliding along geological structures occur. When
the stability of a shallow tunnel is considered, the factors that control the stability can be
divided into five groups of factors, namely the strength of the rock mass, tunnel size and
location of the tunnel in the rock mass, surface loads, geological structures and the state of
stress, Figure 3.1.

Weathering and blast damage decrease the strength of the rock mass and are therefore
important factors. Tunnel size and shape is of course of importance, but can seldom be
altered. At shallow depth a small tunnel with an arched roof is preferable. The tunnel location
is also an obvious factor to consider.

Since shallow tunnels are more frequent in urban areas, loading from buildings and bridges on
the surface is common. This affects the tunnel in many ways. Firstly, the weight of the
structure on the surface will affect the tunnel stability negatively. There is also a risk that
subsidence due to tunnelling will jeopardize the stability of the structure of the surface.

Geological structures such as faults, crushed zones and joints are a part of the rock mass.
These are of great importance, especially at shallow depth, when the in situ stresses are
relatively low. This means that the state of stress also is a factor to consider. It will affect both
failure controlled by geological structures to some degree, as well as stress induced failure.

The factors discussed above are visualized in Figure 3.2.

 35

Rock Mass
Strength

Tunnel
State of
Size and
Stress
Location
Tunnel
Stability

Geological Surface
Structures Loads

Figure 3.1 Groups of factors that influence the tunnel stability.

 36

Weathered rock
Fresh rock

Fresh Rock

Rock Mass
Strength

Blast damage

Tunnel size
Location of the
Tunnel
ground surcace
Size and
Location

Loads from
infrastructure

Surface
Loads

Geological
Structures

State of
Stress ? MPa

Figure 3.2 Examples of factors that influence the behaviour and stability of shallow
tunnels.
 37

In this work conceptual models have been used to study the effect of different factors on the
stability and behaviour of a tunnel. A base case was established. The different factors were
varied to study the impact of the rock mass parameters on the behaviour and stability of the
tunnel. The factors have been varied in such a way to resemble the uncertiany and scatter of
measurements and estimations of such parameters in real preliminary tunnel investigations.
This was done with respect to the precision and uncertainty of methods used to
estimate/calculate the parameters. The parameters that are studied here are the following:

— Virgin stresses
— Intact uniaxial compressive strength
— GSI-value
— Rock mass tensile strength
— Overburden
— Young’s modulus
— Cohesion/Friction of the rock mass
— Large geological structures
— Excavation damaged zone
— Weathering of the ground surface

Small scale geological structures such as joints that cut through the rock mass, forming
wedges and weaknesses have not been studied in this work even though it is considered an
important factor. The reason for this is that there are an infinite number of variations of joint
lengths, orientations and positions that can occur in a rock mass.

Loads from the ground surface have also been excluded even though it is an important factor.
The loads are site specific, and differ greatly from site to site. It was decided to exclude this
factor from the conceptual analysis. However, in the case study, Arlandabanan, loads on the
rock surface will be present.

3.3 Input data – Typical Swedish rock mass conditions

The input data in this work has been chosen to resemble a typical Swedish rock mass.
Obviously there are no typical rock masses, since they differ greatly from site to site, but the
chosen data are considered representative of a hard Swedish rock mass that is not
exceptionally competent or exceptionally poor.
 38

In the first part of the work conducted for this thesis a preliminary set of data was used. Later
a new set of parameters have been used. This was done from a conclusion that the earlier base
case somewhat overestimated the rock mass strength. This does not mean that the values used
earlier were unrealistic, but that some of the data had to be slightly changed, otherwise more
or less linear elastic conditions were obtained. The parameters that have been used in the
majority of this work are called the main data set. Both sets can be seen in Table 3.1. The
values that are marked in red are the same for both sets and the purple indicates where they
are interacting.

The intervals of the parameters have been chosen in reference meetings for this project, and
they where chosen to resemble possible variations Swedish rock masses. The variation of the
virgin state of stress has been done by using a number of stress versus depth expressions
presented in the literature.

Table 3.1 Model data for the preliminary data set and the main data set.

Storhet Preliminary data set Main data set

RMR (GSI) 60 48 60 72
Q 0.25 0.25
Vci [MPa] 0.5 Vci 250 1.5 Vci 140 180 220
mi 33 33
VH [MPa] 1.5Vv 2.8 + 0.04z 2.5Vv 2.8 + 0.04z 6.7 + 0.0444z 10.4 + 0.0446z
Vh [MPa] 2.2 + 0.024z 2.2 + 0.024z 0.8 + 0.0329z 5 + 0.0286z

Two types of cross-sections have been used in the analyses, the Banverket maximum single
track tunnel (cross-section A) and the standard double track tunnel (cross-section B). Cross-
section A was used in the first part, and is used in this report only to analyze the effect of size
of the model and zones and the geometry of the floor. Since it is more likely that stability
problems will occur in openings with a larger span, the double track cross-section was chosen
for all further analyses. The double track cross-section is the Banverket 2002 standard double
track cross-section, with an extra 0.25 m around the theoretical cross-section which is
considered to be closer to the real cross-section after excavation.
 39

R 8.05
(m) 4.5 R 3.05
54o

9.7
72o

6.7
0.32
2.0
9.0 0.32

Cross-section A Cross-section B

Figure 3.3 Cross-sections used in this work, after Banverket (2002).

3.4 Model development

A finer grid gives more accurate results but too many zones give long calculation times. To
achieve accurate results and acceptable computational time, the model consists of an inner
grid with finer zones and an outer grid with coarser zones, see Figure 3.4 The ratio of the zone
length of the outer and the inner grids was set to 4, since this will give a smooth transition of
displacements across the boundaries between the grids according to Itasca (2005a). A
sensitivity analysis was conducted to determine how large the zone sizes should be, if the size
of the finer grid was significant for the results of the model, and how large the outer model
boundaries should be to avoid boundary effects. The cross-section that was used in these
analyses is A from Figure 3.3. A circular cross-section was also analyzed in order to compare
the model results to an analytical solution.
 40

Finer grid

Coarser grid

Figure 3.4 Simplified model setup for the analyses made in FLAC.

According to the standard cross-section there should be a ditch on one side where the wall
meets the floor, however this is not done when the tunnel is excavated due to technical
reasons. Instead an inclined floor is used, see Figure 3.3.

A model was also constructed in the program Examine2D in order to compare the results from
the FLAC model with results from another program. The Examine2D model used the same
cross-section as the model in FLAC.

3.4.1 Model setup and input data

An overburden of 5 m was used for all models in this preliminary study. The model was fixed
in the y-direction at the bottom and in the x-direction at the vertical boundaries. The virgin
state of stress was defined by Table 2.1 ((3) after Stephansson, 1993). The material properties
of the rock mass were chosen to resemble a normal Swedish rock mass, with a RMR-value of
60, from which Young’s modulus was calculated, according to Serafim and Pereira (1983), to
17.8 GPa and a Poisson’s ratio of 0.25 was assumed. From a defined base case, variation of
the three factors was studied by comparing the base case with two other models according to
Figure 3.5, Table 3.2 and Table 3.3.
 41

a) B A C

b) D A E

c) F A G

Figure 3.5 The different factors studied in the analysis, a) zone size, b) the size of the inner
grid and c) model size.

Table 3.2 Models used in the sensitivity analysis.

Model Model size Inner grid Outer grid

Width x Height Width x Height Zone size Zone size
[m x m] [m x m] (Width x height) (Width x height)
[m x m] [m x m]
A: Base case 0.25 x 0.25 0.5 x 0.5
27 x 25
B: Fine grid 0.125 x 0.125 0.25 x 0.25
70 x 50
C: Coarse grid 0.5 x 0.5 1x1
D: Small inner grid 18 x 20
E: Large inner grid 45 x 34
0.25 x 0.25 0.5 x 0.5
F: Small model 50 x 40 27 x 25
G: Large model 100 x 70
 42

Table 3.3 Input data for the models.

Parameter Value
Em 17.8 GPa
Q 0.25
Vv Ugz
VH 2.8 + 0.04z
Vh 2.2 + 0.024z

The input data used for the Examine2D model is according to Table 3.3. The boundary element
size used in this model was 0.071 m.

The results show that the greatest differences among the models occur for zone size and
model size. Very small differences could be observed for different sizes of the inner grid, see
Table 3.4. It could also be observed that a model with smaller zones gives a smoother curve
over the arch shaped roof, see Figure 3.6b. The results show that different zone sizes and
different model sizes are the most important factors to consider and that the model size seems
to give the greatest difference. Therefore, model B: Fine grid and model G: Large model,
should be combined, if possible.
 43

a) 6
Base Case (A)
Courser Grid (B)
5 Finer Grid (C)
Examine

4
Tunnel wall (m)

0
-3,5 -3,0 -2,5 -2,0 -1,5 -1,0 -0,5 0,0
Tangentiall stress (MPa)

b)

14

Base Case (A)

12
D Courser Grid (B)
Finer Grid (C)
10
Examine
Tangetial stress (MPa)

-2

-4
0 20 40 60 80 100 120 140 160 180
D (degrees)

Figure 3.6 The tangential stress on the boundary of the a) tunnel wall and b) roof for the
cases with different zone sizes.
 44

a)
6

Base Case (A)

Large Model (G)
5 Small Model (F)
Examine

4
Tunnel wall (m)

0
-3,5 -3,0 -2,5 -2,0 -1,5 -1,0 -0,5 0,0
Tangential stress (MPa)

b)

14

Base Case (A)

12
D Large Model (G)
Small Model (F)
10
Examine
Tangential stress (MPa)

-2

-4
0 20 40 60 80 100 120 140 160 180
D(degrees)

Figure 3.7. The tangential stress on the boundary of the a) tunnel wall and b) roof for the
cases with different model sizes.
 45

Table 3.4 The results from the sensitivity analysis.

Zone size B: Coarse A: Base case C:Fine

Vertical displacement of ground surface
(mm) 0.429 0.438 0.439
Vertical displacement of tunnel roof (mm) 0.508 0.517 0.519
Tangential stress in tunnel roof (MPa) 10.47 10.90 11.03
Tangential stress in tunnel wall (MPa) -2.84 -2.88 -2.89
Horizontal displacement of tunnel wall
(mm) -0.551 -0.557 -0.559
Inner grid size D: Small A: Base case E: Large
Vertical displacement of ground surface
(mm) 0.437 0.438 0.438
Vertical displacement of tunnel roof (mm) 0.0517 0.0517 0.0517
Tangential stress in tunnel roof (MPa) 10.90 10.90 10.90
Tangential stress in tunnel wall (MPa) -2.88 -2.88 -2.88
Horizontal displacement of tunnel wall
(mm) -0.557 -0.557 -0.557
Model Size F: Small A: Base case G: Large
Vertical displacement of ground surface
(mm) 0.454 0.438 0.411
Vertical displacement of tunnel roof (mm) 0.525 0.517 0.497
Tangential stress in tunnel roof (MPa) 10.22 10.90 11.39
Tangential stress in tunnel wall (MPa) -2.75 -2.88 -2.98
Horizontal displacement of tunnel wall
(mm) -0.516 -0.557 -0.579

3.4.2 Comparison of analytical and numerical solution

Tests were also performed to compare a FLAC model and an Examine2D model with an
analytical solution from Mindlin (1939). To be able to do this, some new input data had to be
used. A circular cross-section had to be used in this model, along with constant horizontal
stresses and no vertical stresses, see Table 3.5. The model size and grid size for the FLAC
model in this study were the same as for the base case, A, in the earlier studies.

Table 3.5 Input data for the models that were compared to an analytical solution.

Parameter Value
Em 17.8 GPa
Q 0.25
Vv 0 MPa
VH 3.2 MPa
 46

The results showed that the analytical solution and the Examine2D model are identical, while
the FLAC model shows some discrepancy. One explanation for the discrepancy can be that
some zones are distorted in an unfavourable way when the grid is adjusted to fit a circular
excavation.

14

12

10

8
Tangential stress (MPa)

0 T

-2 Analytical
Examine
-4 Base Case

-6
-90 -70 -50 -30 -10 10 30 50 70 90
T(degrees)

Figure 3.8 Comparison of the tangential stress distribution around a circular tunnel
calculated by an analytical solution, Examine2D and FLAC.

3.4.3 Effect of floor inclination

It is shown in the sensitivity analysis that a larger model gives more accurate results, see
Figure 3.6, while a finer grid gives smoother results, see Figure 3.7. Since the computer
capacity limits the number of total zones, symmetry with respect to the y-axis has to be
assumed for the model. However, the cross-sections according to Banverket (2002) have an
inclined floor. To be able to use symmetry the cross-section has to be somewhat altered, from
the inclined floor to a horizontal floor, see Figure 3.9. The behaviour of the original and the
symmetric model was therefore compared. The model size and grid size for the FLAC model
in this test were the same as the base case, A, in the earlier tests. The model setup and the
input data can be seen in Table 3.6 and Figure 3.10.
 47

(m) 4.5

9.7

0.32
9.0

Figure 3.9 Alternative cross-section of a Banverket (2002) single track tunnel.

Table 3.6 Parameters used in symmetry analysis.

Parameter Value
Em 17.8 GPa
ıv ȡgz
ıH 2.8 + 0.0399z
ıh 2.2 + 0.0286z
ȡ 2700 kg/m3
 48

(m) Zone size: 0.25 x 0.25

50
Zone size: 0.5 x 0.5

35

70

Figure 3.10 Model dimensions used in symmetry analysis.

When examining the roofs of the two different cross-sections, no significant differences can
be detected, while small differences can be seen in the lower abutments, see Figure 3.11.
Since the floor and lower abutments are not of great importance for the stability of a tunnel,
this difference is considered acceptable and a cross-section with a horizontal floor will be
used in all further analyses.
 49

Figure 3.11 The tangential stress in the walls and lower abutments for the two different
cross-sections.

3.5 Parameter study

3.5.1 Model setup and input data

From previous analyses the model was set up. It is shown in Chapter 3.4 that the zone size
and the model size were of importance to get a reliable model. It was also shown that it is
possible to assume vertical symmetry. With this in mind, and with the availability of a
computer with more capacity, the model was set up as in Figure 3.12, with a model width of
70 m (when symmetry is assumed) and a height of 70 m. Furthermore, the finer inner grid size
is three times the tunnel radius with a zone size of 0.1 x 0.1 m while the coarser grid zone size
is 0.2 x 0.2 m.
 50

Finer gridt 0.1 x 0.1

(m)
Coarser grid 0.2 x 0.2

Vh z
VH
70
Vv

70

Figure 3.12 Model size, zone size and stress directions for the continuum model

However, when the case of discontinuities was analyzed, a symmetric model could not be
used. For these models, UDEC 4.0 (Itasca, 2005b) was used. The model setup can be seen in
Figure 3.13. This model has the same model size, 140 x 70 m as well as the same cross-
section (Figure 3.3b) as the FLAC model. The maximum edge length near the tunnel was 0.5
m, and then increased in two steps, to 1.0 m and 1.5 m, as seen in Figure 3.13. Some extra
cracks were used to obtain an even grid. These cracks were given high strength parameters to
inhibit any slip on these “fictitious” discontinuities. Otherwise, the discontinuum model uses
the same strength data, except for additional data for the discontinuity. A variation of
discontinuity parameters, as well as a variation of dip angles, D, were used in the
discontinuum analysis. The discontinuity, marked in blue in Figure 3.13, was oriented so that
it intersected the tunnel in the roof centre point for all dip angles.
 51

Figure 3.13 Model size and maximum edge length of the disconinuum model.

The cross-section used is the Banverket (2002) normal double track cross-section as described
in Figure 3.3, but with a horizontal floor, see Figure 3.14.

Figure 3.14 The cross-section used in the parameter study.

The input data for the base case was chosen to correspond to normal hard bedrock in Sweden.
As described earlier, the program RocLab (RocScience, 2006) can be used to estimate rock
mass parameters according to the generalized Hoek and Brown failure criterion, (Hoek et. al.
2002). The elasto-plastic material properties are determined through fitting a Mohr-Coulomb
failure envelope to the Hoek-Brown criterion, using the programme RocLab, see Figure 3.15.
The Hoek-Brown criterion is based on the intact compressive strength, Vci, the intact rock
parameter, mi, the geological strength index, GSI, and the disturbance factor, D, along with a
failure envelope range, V3max. The ı3max-value is determined by examining a ı3 plot of an
elastic model. A maximum value of ı3 from the area of a tunnel width around the tunnel was
chosen from the base case, see Figure 3.16 (the stress concentrations in the corners were not
considered). This gave V3max = 2.5 MPa.
 52

80

60

Major principle stress (MPa)

40

20 Hoek-Brown
Mohr-Coulomb

0
-1 0 1 2 3
Minor principle stress (MPa)

Figure 3.15 Hoek-Brown criterion with a Mohr-Coulomb fit.

VV3


W W W

Figure 3.16 Area from where ı3 was chosen.

A GSI-value of 60 corresponds to a blocky to intact rock mass, with good to fair surface
conditions. The intact uniaxial compressive strength was chosen to 180 MPa, which is
approximately that of a gabbro or gneiss. A disturbance factor of 0, corresponds to high
quality controlled blasting (which can be assumed for very shallow tunnels), and a mi factor of
33, which is normal for a granite or a gneiss was chosen. The density of the rock mass was
assumed to be 2700 kg/m3 and the overburden 3 m. The parameters are listed in Table 3.7.
 53

Table 3.7 Input data for the base case in the parameter study.

Parameter Value
GSI 60
ıv ȡgz
ıH 6.7 + 0.0399z
ıh 0.8 + 0.0329z
ȡ 2700 kg/m3
ıc 180 MPa
Overburden 3m
mi 33
D 0
ı3max 2.5 MPa
c 2.4 MPa
ø 64q
ıt -0.27 MPa

It should be emphasized that the water pressure has been considered negligible and not been
used in this work. Furthermore, a solve command was used to automatically terminate the
model when equilibrium has reached, which was considered to occur at an out-of-balance
force limit, f = 0.5.

3.5.2 Parameter variation

For the conceptual analyses, the sensitivity to variations in the following parameters was
studied:

1. Virgin state of stress.

The significance of the state of stress was examined by using the different stress
relations presented in Table 2.1. The direction of the stresses can be seen in Figure
3.12. A fourth model, where VH and Vh changed places, i.e., the horizontal stresses
were rotated 90q from the base case, so that the major horizontal stress was
parallel to the tunnel axis and the minor horizontal stress was perpendicular to the
tunnel axis.

2. Rock mass strength.

The variation of the rock mass strength was studied by changing the rock mass
 54

strength in the whole model. The impact of uniaxial compressive strength as well
as the GSI-value were examined, which are input parameters in RocLab to obtain
the rock mass parameters cohesion, friction angle, tensile strength and Young’s
modulus. The impact of these rock mass parameters was studied as well. The
uniaxial compressive strength and the GSI-value intervals were chosen to resemble
“typical” Swedish rock masses.

Even though cohesion, friction angle, tensile strength and Young’s modulus all are
functions of GSI and the uniaxial compressive strength (in RocLab), it is still
interesting to conduct a study of the influence of a variations of each of these
parameters. When the Mohr-Coulomb failure criterion is used in FLAC, the
residual tensile strength, Vtm,res, is zero. One extra model with a residual tensile
strength just below Vtm was analyzed.

To study how the cohesion and friction angle of the rock mass affected the
stability of the tunnel, the cohesion was reduced while the friction angle was
increased so that the rock mass would still have the same strength. In the other
case the cohesion was increased while the friction angle was reduced. In this way,
one can observe if the cohesion or the friction angle are more significant for the
stability of a shallow tunnel.

3. Overburden.
Analyses with different overburden thicknesses, with a maximum of 5 m, were
conducted to find out how important the overburden is for the stability.

4. Weathered and damaged rock.

Two cases of weathered rock masses and one case of an excavation damaged zone
(EDZ) were analyzed to investigate the impact they have on the tunnel stability.
The weathered rock has a reduction of 50 % in both strength and stiffness. Two
different depths were also used. The damaged rock was simulated using a
disturbance factor equivalent for poor quality blasting and a zone size
corresponding to Malmgren (2005).

5. Discontinuities.
The impact that a discontinuity, cutting through the tunnel roof, has on the stability
and behaviour of a tunnel was tested. Two different kinds of discontinuities, one
considered as soft (data from Fredriksson, 2006) and one considered as relatively
 55

stiff (data from Malmgren, 2005) were used, The dip angle of the stiff
discontinuity was varied.

Parameters 1 to 4 above were analysed using FLAC 5.0. The input data for these models are
listed in Table 3.8. Factor 5, discontinuities, was analyzed with the discontinuum program,
UDEC 4.0. The input data for these models are listed in Table 3.9.
 56

Table 3.8 Input data for the models ran in FLAC

Parameter Virgin state of stress Vci GSI Vtm Over- E, GPa Cohesion
MPa MPa MPa burden, m MPa
Model Friction
o

1 VH = 6.7 + 0.0444z 180 60 From 3 From From

Basecase Vh = 0.8 + 0.0329z Roclab Roclab Roclab
0.267 17.8 2.4
From
RocLab
64
2 VH = 10.4 + 0.0446z 180 60 0.267 3 17.8 2.4
Vh = 5 + 0.0286z 64

3 VH = 2.8 + 0.0399z 180 60 0.267 3 17.8 2.4

Vh = 2.2 + 0.0024z 64

4 Vh = 0.8 + 0.0329z* 180 60 0.267 3 17.8 2.4

VH = 6.7 + 0.0444z*
64

5 VH = 6.7 + 0.0444z 140 60 0.208 3 17.8 2.1

Vh = 0.8 + 0.0329z 63
6 VH = 6.7 + 0.0444z 220 60 0.267 3 17.8 2.7
Vh = 0.8 + 0.0329z 65
7 VH = 6.7 + 0.0444z 180 48 0.108 3 8.9 1.8
Vh = 0.8 + 0.0329z 62
8 VH = 6.7 + 0.0444z 180 72 0.661 3 35.4 3.8
Vh = 0.8 + 0.0329z 66
9 VH = 6.7 + 0.0444z 180 60 0 3 17.8 2.4
Vh = 0.8 + 0.0329z 64
10 VH = 6.7 + 0.0444z 180 60 2*0.267 3 17.8 2.4
Vh = 0.8 + 0.0329z 64
11 VH = 6.7 + 0.0444z 180 60 Vtm,res = 3 17.8 2.4
Vh = 0.8 + 0.0329z 0.26 64
MPa**
12 VH = 6.7 + 0.0444z 180 60 0.267 2 17.8 2.4
Vh = 0.8 + 0.0329z 64
13 VH = 6.7 + 0.0444z 180 60 0.267 5 17.8 2.4
Vh = 0.8 + 0.0329z 64
14 VH = 6.7 + 0.0444z 180 60 0.267 3 14.2 2.4
Vh = 0.8 + 0.0329z 64
15 VH = 6.7 + 0.0444z 180 60 0.267 3 21.4 2.4
Vh = 0.8 + 0.0329z 64
 57

Table 3.8 (continued.)

16 VH = 6.7 + 0.0444Z 180 60 0.267 3 17.8 1.9
VH = 0.8 + 0.0329Z 70
17 VH = 6.7 + 0.0444z 180 60 0.267 3 17.8 2.9
Vh = 0.8 + 0.0329z 59
18 VH = 6.7 + 0.0444z 90*** 60*** 0.134*** Weathered 8.9*** 1.75**
Vh = 0.8 + 0.0329z thickness
60**
3m

19 VH = 6.7 + 0.0444z 90*** 60*** 0.134*** Weathered 8.9*** 1.75***

Vh = 0.8 + 0.0329z thickness
5m 60***
20 VH = 6.7 + 0.0444z 180**** 60**** 0.149**** EDZ 11.6**** 1.7****
Vh = 0.8 + 0.0329z thickness
0.5 m 60****
*ıH is parallel to the tunnel axis, ıh is perpendicular to the tunnel axis.
**Vtm is the same as for the base case, i.e. 0.267 MPa.
***These are the parameters for the weathered rock mass. The fresh rock mass has the same input data as the base
case, model 1. The overburdens of the models are 3 m.
****These are the parameters for the EDZ zone. The fresh rock mass has the same input data as the base case,
model 1. The overburden of the model is 3 m.

Table 3.9 Properties of discontinuities used in the analysis.

Parameter Unit Stiff* Soft**

Friction angle [ƕ] 35 25
Cohesion MPa 0 0.05
Tensile strength MPa 0 0
Normal stiffness GPa/m 110 0.2
Shear stiffness GPa/m 9 8
15
Discontinuity angle [ƕ] 30 30
60
* Malmgren (2005)
** Fredriksson (2006)

3.5.3 Instability indicators

The aim of this project was to investigate the importance of a number of factors on the
stability and performance of shallow excavations in hard rock. Although this has proven to be
more difficult than first expected, some instability indicators were chosen in order to
determine the sensitivity to variations in the examined parameters. Because different
programs were used for the continuum and discontinuum models, slightly different instability
 58

indicators were used. The instability indicators for continuum models are (i) tangential stress
around the tunnel boundary, (ii) deformation of the tunnel boundary, (iii) deformation of the
ground surface, (iv) area of plasticity, and (v) extent of a complete tensile stress state (all
principal stresses are tensile in an elastic model).

For the discontinuum models, (I) deformation of the ground surface, (II) the zone of plasticity,
(III) roof stability (velocity vectors), and (IV) opening in and/or slip in the discontinuity, were
the primary indicators of instability. A short explanation of all indicators follows below.

(i) Tangential stress around the tunnel boundary

One of the greatest risks of shallow tunnelling is considered to be fallouts of wedge

formations due to low confining stresses. Low tangential stresses (risk of opening of pre-
existing joints or tensile failure) as well as very high stresses (risk of compressive failure) are
therefore a threat to the stability of the tunnel. The tangential stress has been examined at the
boundary and 0.5 m from the boundary.

(ii) Deformations of the tunnel boundary

Deformations of the tunnel boundary are an important quantity to examine. Too large
deformations might be an indication of instability. Furthermore, large deformations may
affect rock reinforcement and installations such as electricity, water and ventilation.

(iii) Deformations of the ground surface

Subsidence or heaving of the ground surface is mostly important when infrastructure or

buildings are located above the tunnel. It is therefore important to control the deformations
when tunnelling in, for example urban areas.

(iv) Area of plasticity

An important parameter for the stability and the rock support of a tunnel is the zone of
plasticity. In FLAC, such a zone indicates that plastic flow is occurring and may provide a
pattern that tells if a mechanism has developed. It was observed that the zone of plasticity was
dependent on when the model is terminated since the state hovered between “yield in tension”
and “yield in past”. The area in which the tensile strength has been changed from peak to
residual illustrates areas where tensile failure has occurred. The areas in the model where
 59

“yield in past” is indicated and the tensile strength is still equal to the peak value represents
areas where shear failure has occurred.

By comparing the model state plot and the tensile strength plot, the area that has yielded in
tension can be identified in the zone labelled “yield in past” see Figure 3.17. This does also
mean that if the peak tensile strength is indicated in the area labelled “yield in past”, this
portion has yielded in shear.

Area of plasticity Tensile strength of the model

Figure 3.17 Comparison of area if plasticity and tensile strength.

(v) Extent of tensile stresses

Many stability problems at shallow depth are a result of low compressive or tensile stresses.
When all principal stresses are tensile in an elastic model, this can be considered as a potential
risk of failure. It is therefore something that needs to be examined.

(I) Deformation of the ground surface

This is basically the same as (iii) subsidence of the ground surface.

(II) Area of plasticity

This is basically the same as (iv) area of plasticity.

 60

(III) Roof stability (velocity vectors)

Some of the discontinuum models could not to be run to equilibrium without having a part of
the roof deforming unrealistically and had to therefore be stopped prematurely. Examination
of the velocity vectors of the models assisted in the detection of unstable areas.

(IV) Opening of and/or slip in the discontinuity

The stability of a discontinuity is basically defined by the occurrence of slip and/or opening.
Discontinuities with zero normal stress are defined as open in UDEC and discontinuities at
shear limit are considered to be in state of slip.

3.5.4 Results – Continuum models

(i) Tangential stress around the tunnel boundary

Figure 3.18, Figure 3.19 and Figure 3.20 show that the tangential stress at the tunnel boundary
is tensile or zero at the wall and the abutment, while it is compressive in the roof. At a
distance of 0.5 m from the boundary, the behaviour seems to be similar, with the difference
that the compressive stresses start a few metres closer to the abutment.

Figure 3.18 shows that the cases with lower horizontal virgin stress perpendicular to the
tunnel axis show smoother curves and lower values than the cases with higher horizontal
stresses. Furthermore, figure 3.18 shows higher values for Vci and GSI, since greater strength
can carry greater stresses.

In Figure 3.19, it is shown that with a greater overburden the compressive tangential stresses
start closer to the abutment of the tunnel and vice versa. In the cases of weathering and EDZ
(Figure 3.20), the rock mass strength and the stiffness (E) have been reduced, which gives
lower stresses in the tunnel roof. It is also seen that the tangential stress is only weakly
dependent on the variation of cohesion, Young’s modulus and the tensile strength.
 61

Stress tangential to the tunnel boundary, Stress tangential to the tunnel boundary,
on the tunnel boundary 0.5 m into the rock mass
(MPa) (MPa)
Basecase
20 VH = 6.7 + 0.0444z
20 Basecase
VH = 6.7 + 0.0444z 10 Vh = 0.8 + 0.0329z
10
Vh = 0.8 + 0.0329z 0 VH = 10.4 + 0.0446z
0
Virgin state of stress

VH = 10.4 + 0.0446z -10 Vh = 5 + 0.0286z

-10
Vh = 5 + 0.0286z VH = 2.8 + 0.0399z
-20
-20 Vh = 0.8 + 0.0024z
VH = 2.8 + 0.0399z
Vh = 0.8 + 0.0024z Vh = 0.8 + 0.0329z
Vh = 0.8 + 0.0329z VH = 6.7 + 0.0444z
VH = 6.7 + 0.0444z (Vh perpendicular to
tunnel axis)
(Vh perpendicular
to tunnel axis)

(MPa)
-20 -10 0 10 20 -20 -10 0 10 20 (MPa)

(MPa) (MPa)
Uniaxial compressive strength, Vci

20
20
Basecase 10 Basecase
10 Vci = 180 MPa Vci = 180 MPa
0
0 Vci = 140 MPa Vci = 140 MPa
-10
-10 Vci = 220 MPa Vci = 220 MPa
-20
-20

(MPa)
-20 -10 0 10 20 -20 -10 0 10 20 (MPa)

(MPa) (MPa)

20
20
Basecase
10 Basecase
10 GSI = 60
GSI = 60
GSI = 48 0
0 GSI = 48
GSI = 72 -10
-10 GSI = 72
-20
-20
GSI

(MPa)
-20 -10 0 10 20 -20 -10 0 10 20 (MPa)

Figure 3.18 The tangential stress and its variation due to variations in virgin state of stress,
uniaxial compressive strength and GSI.
 62

Stress tangential to the tunnel boundary, Stress tangential to the tunnel boundary,
on the tunnel boundary 0.5 m into the rock mass
(MPa)
(MPa)
20 Basecase
20 Vtm = 0.267 MPa
Basecase 10
10 Vtm = 0.267 MPa Vtm = 0 MPa
0
Vtm = 0 MPa Vtm = 0.534 MPa
0
Tensile strength, Vtm

Vtm = 0.534 MPa

-10 Vtm,res = 0.26 MPa
-10
Vtm,res = 0.26 MPa -20
-20

(MPa)
-20 -10 0 10 20
-20 -10 0 10 20 (MPa)

(MPa) (MPa)

20
20
10 Basecase
10 Basecase Overburden = 3m
Overburden = 3m 0 Overburden = 2m
0
Overburden = 2m -10 Overburden = 5m
-10 Overburden = 5m
-20
-20
Overburden

(MPa)
-20 -10 0 10 20 -20 -10 0 10 20 (MPa)

(MPa) (MPa)

20
20
Basecase 10 Basecase
10 E = 17.8 GPa E = 17.8 GPa
0
0 E = 14.2 GPa E = 14.2GPa
Young’s modulus, E

E = 21.4 GPa -10 E = 21.4GPa

-10
-20
-20

(MPa)
-20 -10 0 10 20 -20 -10 0 10 20 (MPa)

Figure 3.19 The tangential stress and its variation due to variations in tensile strength,
overburden and Young’s modulus.
 63

Stress tangential to the tunnel boundary, Stress tangential to the tunnel boundary,
on the tunnel boundary 0.5 m into the rock mass
(MPa) (MPa)

20
20
Basecase 10
10 c = 2.4 MPa
0 Basecase
0 c = 1.9 MPa c = 2.4 MPa
c = 2.9 MPa -10 c = 1.9 MPa
-10
-20 c = 2.9 MPa
Cohesion, c

-20

(MPa)
-20 -10 0 10 20 -20 -10 0 10 20 (MPa)

(MPa)

20 20
10 10
0 0
Basecase
-10 -10
Weathered rock 3m Basecase
-20 Weathered rock 5m -20 Weathered rock 3m
Weathering

Weathered rock 5m

-20 -10 0 10 20
-20 -10 0 10 20 (MPa)

(MPa)

20 20
10 10
Basecase Basecase
0 0
0.5 m EDZ 0.5m EDZ
-10 -10
-20 -20
EDZ

-20 -10 0 10 20
-20 -10 0 10 20 (MPa)

Figure 3.20 The tangential stress and its variation due to variations in cohesion, weathering
and EDZ.
 64

(ii) Deformation of the tunnel boundary

The impact of the variation of rock mass properties on the deformation of the tunnel boundary
is presented in Figure 3.21 and Figure 3.22. A similar pattern of the deformation can be seen
for all factors, except the case where the residual tensile strength is not set to zero. The tunnel
wall and the abutment converge while the roof heaves. Variation of the virgin state of stress
gave the largest differences in deformation. With low stresses, the deformations are very
small, while greater stresses show larger deformations, see Figure 3.21. Other factors that
seem to have impact on the deformations are GSI and the overburden. The variation of
weathering (Figure 3.22) might seem strange, but the reason for this behaviour is that there is
a separation between the weathered rock mass and the fresh rock, as seen in Figure 3.23,
because the overburden and the depth of the weathered rock mass are the same.
 65

(cm) (cm)

Uniaxial compressive strength,Vci

4 Basecase
4
VH = 6.7 + 0.0444z
Vh = 0.8 + 0.0329z Basecase
2 2 Vci = 180 MPa
Virgin state of stress

VH = 10.4 + 0.0446z
1 1 Vci = 140 MPa
Vh = 5 + 0.0286z
0 0 Vci = 220 MPa
VH = 2.8 + 0.0399z
-2 -2
Vh = 0.8 + 0.0024z
-3 -3
VH = 0.8 + 0.0329z
Vh = 6.7 + 0.0444z
(Vh perpendicular to
tunnel axis)

-2 -1 0 1 2 4 (cm) -2 -1 0 1 2 4 (cm)
(cm) (cm)

4 4
Basecase Basecase
GSI = 60 Vtm = 0.267 MPa
2 2
Tensile strength, Vtm

GSI = 48 1 Vtm = 0 MPa

1 GSI = 72 0 Vtm = 0.537 MPa
0
-2 Vtm,res = 0.26 MPa
-2
-3
-3
GSI

-2 -1 0 1 2 4 (cm)
-2 -1 0 1 2 4 (cm)
(cm) (cm)

4 4
Basecase Basecase
2 Overburden = 3 m 2 E = 17.8 GPa
Young’s Modulus, E

1 Overburden = 2 m 1 E = 14.2 GPa

Overburden = 5 m E = 21.4 GPa
0 0
Overburden

-2 -2
-3 -3

-2 -1 0 1 2 4 (cm) -2 -1 0 1 2 4 (cm)

Figure 3.21 The sensitivity of variations in virgin state of stress, intact uniaxial compressive
strength, GSI, tensile strength, overburden and Young’s modulus on
deformation on the tunnel boundary.
 66

(cm) (cm)

4 4
Basecase
2 c = 2.4 MPa 2
1 c = 1.9 MPa 1
0 c = 2.9 MPa 0 Basecase
Cohesion, c

Weathering
-2 -2 Weathered rock 3 m
-3 Weathered rock 5 m
-3

-2 -1 0 1 2 4 (cm) -2 -1 0 1 2 4 (cm)
(cm)

4
Basecase
0.5m EDZ
2
1
0
-2
-3
EDZ

-2 -1 0 1 2 4 (cm)

Figure 3.22 The sensitivity of variations in cohesion, weathering and EDZ on deformation
on the tunnel boundary.

Figure 3.23 Separation between the weathered rock mass and the fresh rock.
 67

(iii) Deformaion of the ground surface

The results of the analysis are visualized in Figure 3.24. For all cases the ground surface is
heaving between 0.1 mm (very low horizontal stresses) and about 5 mm (for high horizontal
stresses and low GSI value). Besides GSI and virgin stresses,variations of Young’s modulus
and overburden results in a ground deformation that deviates significantly from that of the
base case.

There is a negligible difference in ground surface deformation if Vtm is 0.27 MPa or 0.53 MPa,
while there is a noticeable difference if Vtm is 0 MPa. The cases when Vtm = 0.27 MPa with a
residual Vtm,res of 0.26 MPa shows considerably less heaving than the case with a greater peak
value.

The variations of the intact uniaxial compressive strength, weathering of the overburden and
the cohesion have no affect on the behaviour of the overburden.
.
 68

6
Base case Basecase 6 Base case Base case
Vci = 180 MPa VH = 6.7 + 0.0444z Vtm = 0.267 MPa GSI = 60
Vci = 140 MPa Vh = 0.8 + 0.0329z Vtm = 0 MPa GSI = 48
VH = 10.4 + 0.0446z Vtm = 0.53 MPa GSI = 72
Vci = 220 MPa
Vh = 5 + 0.0286z Vtm,res = 0.26 MPa
4
VH = 2.8 + 0.0399z
4
Deformation (cm)

Vh = 0.8 + 0.0024z

Deformation (cm)
Vh = 0.8 + 0.0329z
VH = 6.7 + 0.0444z
(Vh perpendicular to
tunnel axis)
2

-80 -40 0 40 80 0
Ground Surface (m)
-80 -40 0 40 80
Ground Surface (m)
Variation of the uniaxial compressive Variation of the tensile strength, Vtm, and
strength, Vci, and virgin state of stress. GSI value.
6 6

Base case Base case Base case Base case

E = 17.8 GPa Overburden = 3m No weathering or EDZ c = 2.4 MPa
E = 14.2 GPa Overburden = 2m Weathered rock, 3m c = 1.9 MPa
E = 21.4 GPa Overburden = 5m Weathered rock, 5m c = 2.9 MPa
EDZ, 0.5m
4 4
Deformation (cm)
Defromation (cm)

2 2

0 0
-80 -40 0 40 80 -80 -40 0 40 80
Ground Surface (m) Ground Surface (m)
Variation of Young’s modulus, E, and Variation of weathering, EDZ and
overburden. cohesion, c.

Figure 3.24 The sensitivity of variations in rock mass parameters on ground surface
deformation.
 69

(iv) Area of plasticity

The results from the study of the impact of variations of the rock mass properties on the area
of plasticity are presented in Figure 3.25 - Figure 3.33. A similar pattern of zones which have
experienced yielding can be seen for all factors. There is an area, primarily in the tunnel wall
and the abutment where tensile failure has occurred. Almost all plastic flow is due to tensile
stresses. A very small area where the rock is at yield in shear next to the tunnel roof centre
can also be detected in most cases.

The different virgin states of stress analyzed in this study show the greatest differences in the
area of plasticity (Figure 3.25). Low stresses give a small area, while the case with the highest
stresses show a large area of plasticity that reaches the ground surface. Comparison of the
area of plasticity plot with the plot of the tensile strength in the case of high stress, shows that
yield in shear can be detected in the middle of the tunnel roof and at the ground surface over
the abutment of the tunnel.

Furthermore, the area of plasticity is strongly dependent on both the GSI-value (Figure 3.27)
and tensile strength (Figure 3.28). In the case of a low GSI-value, the area of plasticity reaches
ground surface, and some shear failure can be observed above the tunnel centre and above the
abutment at ground level. Only a small difference in area of plasticity can be observed when
Vtm is 0.267 MPa and 0.537 MPa. Vtm,res = 0.26 MPa shows similar size of the area of
plasticity, but a smoother pattern. The model with Vtm = 0 MPa, on the other hand, shows a
great difference in the area of plasticity compared to the base case. The area of plasticity is
larger, and extends over a large portion of the ground surface, see Figure 3.28. Since Vtm is
equal to Vt,res for this case, tensile failure cannot be detected in the tensile strength plot in
Figure 3.28.

The variation of the overburden shows that it is a significant factor for the area of plasticity,
where it is larger and closer to the ground surface when the tunnel is shallower (Figure 3.29).
Variations in Young’s modulus (Figure 3.30), cohesion (Figure 3.31), weathering (Figure
3.32) and EDZ (Figure 3.33) seem to have a small effect on the area of plasticity. However,
the case of the EDZ shows more shear failure, under the tunnel floor and in the middle of the
tunnel roof, than the base case.
 Intermediate stress
<- Low stress Intermediate stress High Stress ->
VH = 2.8 + 0.0399z
Vh = 0.8 + 0.0329z1 VH = 10.4 + 0.0446z
Vh = 2.2 + 0.0024z Base case
VH = 6.7 + 0.0444z1 Vh = 5 + 0.0286z
VH = 6.7 + 0.0444z
ıH perpendicular to tunnel axis
ıH parallel to tunnel axis. Vh = 0.8 + 0.0329z ıH perpendicular to tunnel axis
ıH perpendicular to tunnel axis

A)

(A) and tensile strength (B).

70

B)

Figure 3.25 The sensitivity of variations of the primary state of stress in area of plasticity
 Intermediate compressive strength
<- High compressive strength Low compressive strength ->
ıci = 220 MPa Basecase
ıci = 140 MPa
ıci = 180 MPa

A)

(A) and tensile strength (B).

71

B)

Figure 3.26 The sensitivity of variations of the compressive strength, ıc, in area of plasticity
 72

Low GSI-value ->

GSI = 48
Intermediate GSI-value
Basecase
GSI = 60
<- High GSI-value
GSI = 72

A)

B)

Figure 3.27 The sensitivity of variations of the GSI-value on area of plasticity (A) and
tensile strength (B).
 <- High tensile strength Residual tensile strength Intermediate tensile strength Low tensile strength ->
ıtm = 0.534 MPa ıtm = 0.267 MPa Basecase ıtm = 0 MPa
ıtm,res = 0 MPa ıtm,res = 0.26 MPa ıtm = 0.267 MPa ıtm,res = 0 MPa
ıtm,res = 0 MPa

A)

and tensile strength (B).

73

B)

Figure 3.28 The sensitivity of variations of the tensile strength, ıtm, on area of plasticity (A)
 74

Less overburden ->

2m
Intermediate overburden
Basecase
3m
<- More overburden
5m

A)

B)

Figure 3.29 The sensitivity of variations of the overburden thickness on area of plasticity (A)
and tensile strength (B).
 75
Low stiffness->
E = 14.2GPa
Intermediate stiffness

E = 17.8GPa
Basecase
<- High stiffness
E = 21.4GPa

A)

B)

Figure 3.30 The sensitivity of variations of Young’s modulus, E, on area of plasticity (A)
and tensile strength (B).
 76

Low cohesion ->

c = 1.9 MPa
Intermediate cohesion

c = 2.4 MPa
Basecase
<- High cohesion
c = 2.9 MPa

A)

B)

Figure 3.31 The sensitivity of variations of the cohesion,c, on area of plasticity (A) and
tensile strength (B).
 77
5 m of weathered rock
3 m of weathered rock
<- No weathering
Basecase

A)

B)

Figure 3.32 The sensitivity of variation of weathering on area of plasticity (A) and tensile
strength (B).
 78

<- No EDZ EDZ ->

Basecase EDZ = 0.5 m
A)

B)

Figure 3.33 The sensitivity of variations of EDZ on area of plasticity (A) and tensile strength
(B).

(v) Extent of tensile stresses

The extent of the tensile stresses (of the linear elastic analysis), i.e., portions of the rock mass
where all principal stresses are tensile see figure 3.34. The varied parameters are the virgin
state of stress, overburden and Young’s modulus. The results show that a greater horizontal
stress perpendicular to the tunnel axis results in a greater area, and greater tensile stresses. The
extent of a 3D tensile stress state is also sensitive to the variation of overburden. A shallow
tunnel results in a larger area and greater tensile stresses than a tunnel at greater depth.
Young’s modulus does not affect the extent of tensile stresses at all.
 79

Basecase
V H 6.7  0.0444 z VH 2.8  0.0399 z V H 6.7  0.0444 z V H 10.4  0.0444 z
V h 0.8  0.0399 z Vh 2.2  0.0024 z V h 0.8  0.0399 z V h 5  0.0286 z
ıH parallel to tunnel axis. ıH perpendicular to tunnel axis ıH perpendicular to tunnel axis ıH perpendicular to tunnel axis

Basecase
Overburden = 2 m Overburden = 3 m Overburden = 5 m
V1

Figure 3.34 The sensitivity to variations of elastic rock mass parameters on the extent of
tensile stresses. In FLAC, tensile stresses are positive and compressive stresses
negative, hence a positive scale for tensile stresses.

3.5.5 Results – Discontinuum models

(I) Deformaion of the ground level

The sensitivity of variation of discontinuity parameters on the deformation of the ground

surface is presented in Figure 3.35. The deformation depends on the angle of the
discontinuity. Although the behaviour is similar with the discontinuity angles of 15º and 30º,
the amplitude differs. The deformation of the ground surface in the case with D = 60º is very
similar to the response of the base case without a discontinuity. The subsidence/heaving of the
cases with a stiff discontinuity and the soft discontinuity are similar in both behaviour and
amplitude.
 80

Base case,
No discontinuity
R
Stiff, D 
2 R
Stiff, D 
R
Stiff, D 

Deformation (cm)
R
Soft, D 

-1
-80 -40 0 40 80
Ground surface (m)
Figure 3.35 The subsidence (negative values) and/or heaving (positive values) of the ground
surface for the different cases of discontinuity.

(II ) Area of plasticity

Figure 3.36 show that a discontinuity does affect the area of plasticity. Furthermore it shows
that the area of plasticity is more developed for a low angle (D), and the area where yield has
occurred decreases when the angle gets steeper. No significant difference can be observed
between a stiff and a soft discontinuity.
 81

Model type The area of plasticity Tensile strength

x = yield in past Red = Original value
o = yield in tension Purple = Residual value
* = yield in shear
Base case
No discontinuity

Disc. type: Stiff

D = 15o
ø = 35o
c = 0 MPa

Disc. type: Stiff

D = 30o
ø = 35o
c = 0 MPa

Disc. type: Stiff

D = 60o
ø = 35o
c = 0 MPa

Disc. type: Stiff

D = 30o
ø = 25o
c = 0.05 MPa

Figure 3.36 The sensitivity of varying discontinuity parameters on the area of plasticity and
the tensile strength.
 82

(III) Opening of and/or slip in the discontinuity

The analysis of the sensitivity of discontinuity parameters on the opening of and slip in the
discontinuity is presented in Figure 3.37. It shows that discontinuities with small dip angles
tend to open and the risk of slip increases. This is due to the fact that the major principal stress
in the overburden is horizontal. The normal stress of the discontinuity will decrease with a
decreasing dip angle. No significant difference can be seen between the soft and the stiff
discontinuities.

(IV) Roof stability (velocity vectors)

The sensitivity to variation in discontinuity parameters on the roof stability, i.e. velocity
vectors is presented in Figure 3.38. It is shown that steeper discontinuities results in more
stable conditions. Moreover it is shown that the stiff discontinuity is less stable than the soft
discontinuity. This can be explained with the low normal forces present in the discontinuity.
This means that friction is less important than cohesion for the stability when the normal
stresses are relatively small.
 83

Model type Open Slip

Joints with zero normal force or Joints at shear limit
stress
Base case
No discontinuity No discontinuity No discontinuity

Disc. type: Stiff

D = 15o
ø = 35o
c = 0 MPa

Disc. type: Stiff

D = 30o
ø = 35o
c = 0 MPa

Disc. type: Stiff

D = 60o
ø = 35o
c = 0 MPa

Disc. type: Stiff

D = 30o
ø = 25o
c = 0.05 MPa

Figure 3.37 The sensitivity to variations in discontinuity parameters in opening and slip in
discontinuities.
 84

Model type Roof stability

Velocity vectors
Base case
No discontinuity

Disc. type: Stiff

D = 15o
ø = 35o
c = 0 MPa

Disc. type: Stiff

D = 30o
ø = 35o
c = 0 MPa

Disc. type: Stiff

D = 60o
ø = 35o
c = 0 MPa

Disc. type: Stiff

D = 30o
ø = 25o
c = 0.05 MPa

Figure 3.38 The sensitivity to variations of discontinuity parameters on roof stability

 85

4 IDENTIFICATION OF IMPORTANT FACTORS

The main objective of this project was to identify the factors that are most important for the
stability and behaviour of shallow constructions in hard rock. The parameter values chosen
for the base case, were chosen to represent typical Swedish rock conditions. In order to study
how uncertainties in the input parameters affect the behaviour of shallow underground
constructions, intervals for all parameters (representing a factor) were chosen, see Chapter
3.5.

The conceptual analysis showed that the most important factor was the presence of
discontinuities. This was actually the only time that instability problems could be confirmed
in the conceptual analyses. Furthermore, the most important parameter of the discontinuity
was the dip angle. A steep angle is more favourable than a shallow angle. Equally important is
to know the location of the tunnel in relation to discontinuities and the ground surface. A
larger overburden is preferable, if there is a possibility to choose.

The sensitivity of the tangential stress in the walls, abutments and roof of the tunnel and at the
ground surface and the deformation of the ground surface and the tunnel boundary to
variations in the rock mass parameters are presented in Figure 4.1, Figure 4.2 and Figure 4.3,
where they are presented relative the base case. The value of the base case is 100 %. The
figures show that the tangential stress and the deformations are highly sensitive to variations
in the virgin state of stress and GSI. Other factors that give significant deviations in tangential
stresses and deformations are the size of the overburden and the tensile strength. However, the
only parameter that definitely indicated stability problems was the presence of discontinuities.
When the discontinuity parameters were examined, the discontinuity angle was the most
important factor for the stability. Furthermore, it showed that with low normal stresses on the
discontinuity, cohesion contributes more than friction to the stability.

The failures that have occurred in the conceptual analyses have mainly been tensile, which
means that the tensile strength is an important factor. However, the analyses show that the
main issue is not to know whether the peak tensile strength is 0.26 MPa or 0.52 MPa. It is
more important to know the post failure properties of the tensile strength, i.e. the residual
tensile strength. Variations of the residual tensile strength were shown to have much more
impact on the instability indicators than the peak strength.
 Tangential stress in the tunnel roof and on ground surface, relative the base case.

Heaving of the ground surface. Vertical deformation measured in the centre of

(20) EDZ (20) EDZ
Ground Surface

Ground Surface

the tunnel roof boundary and on ground surface. The value of the base case
Tunnel Roof

Tunnel Roof

(19) Weathered rock, 5m (19) Weathered rock, 5m

(18) Weathered rock, 3m (18) Weathered rock, 3m
(17) c = 2.9 MPa (17) c = 2.9 MPa
(16) c = 1.9 MPa (16) c = 1.9 MPa
(15) E = 21.4 GPa (15) E = 21.4 GPa
(14) E = 14.2 GPa (14) E = 14.2 GPa
(13) Overburden = 5m (13) Overburden = 5m
The value of the base case represents 100%.
(12) Overburden = 2m (12) Overburden = 2m
(11) Sigma t res = 0.26 MPa (11) Sigma t res = 0.26 MPa
86

(10) Sigma t =0.534 MPa (10) Sigma t =0.534 MPa

(9) Sigma t =0 MPa (9) Sigma t =0 MPa
(8) GSI = 72 (8) GSI = 72
(7) GSI = 48 (7) GSI = 48

represents 100%.
(6) Sigma c =220 MPa (6) Sigma c =220 MPa
(5) Sigma c =140 MPa (5) Sigma c =140 MPa
(4) Sigma h = 0.8 … (4) Sigma h = 0.8 …
(3) Sigma H = 2.8 … (3) Sigma H = 2.8 …
(2) Sigma H = 10.4 … (2) Sigma H = 10.4 …

Figure 4.1

Figure 4.2
250%

200%

150%

100%

50%

0%
200%
180%
160%
140%
120%
100%
80%
60%
40%
20%
0%
 87

250% 200% 150% 100% 50% 0%

(2) Sigma H = 10.4 …

(3) Sigma H = 2.8 …
(4) Sigma h = 0.8 …
(5) Sigma c =140 MPa
(6) Sigma c =220 MPa
(7) GSI = 48
(8) GSI = 72
(9) Sigma t =0 MPa
(10) Sigma t =0.534 MPa
(11) Sigma t res = 0.26
(12) Overburden = 2m
(13) Overburden = 5m
(14) E = 14.2 GPa
(15) E = 21.4 GPa
(16) c = 1.9 MPa
(17) c = 2.9 MPa
(18) Weathered rock, 3m
(19) Weathered rock, 5m
(20) EDZ

Figure 4.3 Deformation in the tunnel wall relative to the base case. The value of the base
case represents 100%.

The sensitivity to variations in the state of stress was large for all instability indicators. This
causes a dilemma. Stress measurements conducted at shallow depths are often considered as
unreliable since the stresses at shallow depth are more sensitive to disturbances affecting the
magnitude and direction and at the same time have a lower magnitude than stresses at greater
depth. This means that the standard deviation of the measurements may be of the same
magnitude as the stress. This fact has resulted in few stress measurements at shallow depth,
and when the stress is measured, it is not considered to be reliable. Berg (2005), however,
suggests an easy, fast and cheap stress measurement method where the closure of a slit made
by a saw cut is measured. The theory of elasticity can be used to evaluate the state of stress.
Saw cuts in different directions may give information about the stress orientation. The studies
made by Berg (2005) show that the stresses at the rock surface are good indicators of the
stress situation at shallow depth.

The conceptual analyses show that all instability indicators are sensitive to variations in GSI,
which makes GSI a very important factor for the analysis of stability and behaviour. The GSI-
value was varied 12 points. Edelbro (2004) showed that many of the classification systems are
very subjective, and GSI can easily be varied more than 12 points on the same rock mass, due
to subjective interpretations.
 88

Moreover it is of interest to know whether the rock mass or regions of the rock mass have
been subjected to weathering or other strength reducing mechanisms. If areas of the roof and
abutments lose 50 % of strength and stiffness, the rock mass will loose its ability to carry
stresses. This can lead to destressed areas and fallouts of wedge formations.

A general overview of how sensitive the instability indicators are to variations of the different
parameters is presented in Table 4.1. The sensitivity to the variation of a specific parameter on
the stability indicators was divided into three categories, high, medium and low. The intervals
of the categories for tangential stress,VT, in the tunnel walls, abutments and roof as well as
deformation of the tunnel boundary and the ground surface are presented in Table 4.2. The
sensitivity is estimated by comparing the results obtained for the maximum and minimum
values of the varied parameters with the results from the base case.

Area of plasticity and extent of tensile stresses could not be valued in the same way. For these
indicators, the ratings have been set by a subjective estimation.

Table 4.1 The impact that the rock parameters have on the instability indicators.

Vtan, Deformation Extent of

Subsidence Area of
tunnel of tunnel tensile
/ heaving Plasticity
boundary boundary stresses
Virgin state of
High High High High High
stress
Vci Medium Low Low Low -
GSI High High High High -
Vt High Medium High High -
Over burden High High High Medium Medium
E Low Medium Medium Low -
c/I Low Low Low Low -
Weathering High Medium Low Low -
EDZ High Low Low Low -
Discontinuities High High High High -
 89

Table 4.2 Rating intervals for Table 4.1.

Vtan on tunnel Deformation of Deformation of

boundary (MPa) tunnel boundary (cm) ground surface (cm)
High > 10 >1 >1
Medium >5 > 0.5 > 0.5
Low <5 < 0.5 < 0.5

Deformations on the tunnel boundary and ground surface are very important, although in this
work, the deformations cannot be directly translated into how they affect the stability of the
tunnel. If the deformation is 200 % larger than for the base case, it does not necessarily mean
that it is 2 times as unstable as the base case. This is still interesting information since it
affects reinforcement, media (water, electricity, ventilation etc.) and buildings and structures
located above the tunnel.

The tangential stress around the tunnel acts to hold blocks and wedges in place. Very low
compressive or tensile stresses increase the risk of fallouts and progressive failure, while too
high stresses can lead to compressive failure. It was shown in Chapter 3 that the tangential
stresses are low in the walls and abutments of the tunnel. This increases the risk of fallouts
defined by pre-existing structures, but the risk cannot be quantified in this work.
90
 91

5 CASE STUDY – ARLANDABANAN

5.1 Introduction

To be able to study if the conceptual models are compatible to a case study, and to gain
further knowledge of stability of shallow tunnels, a real case was analysed. A tunnel section
of Shuttle station 2 in Arlandabanan, a railroad tunnel under Arlanda airport was chosen.

The Arlanadabanan tunnel project was unique in Sweden especially since the design work
was done simultaneously with the rock excavation, so-called “active design”. For every blast,
geological mapping was carried out by the shift working geologist. Weekly follow-ups were
performed where the deformations and drilling rates were measured and joint mapping was
conducted. These follow-ups were then used along with weekly visits at the excavation site to
determine the amount of reinforcement that was needed (Chang and Hellstadius, 1998).

Shuttle station 2 is 155 m long and has a span of 23 m. The overburden varies between 8 and
13 m. Terminal 5 is founded on the rock surface above the station. The rock mass consists of
mica schist and mica gneiss. The structures of the rock mass have a general strike of 10 to 20q
to the tunnel axis and a dip of approximately 70q, see Figure 5.1. Two larger structures have
been encountered, a weaker zone consisting of a pegmatite dyke, and a clay gouge, Chang et
al. (1998). They can be seen in Figure 5.2.
 92

N
A
600
700 Shuttle station 2

10-200

Section A-A
Figure 5.1 The position of the station in relation to the strike and dip of the mica schist,
Chang et al. (1998).

The cross-section used for this case is 39/317, because both extensometer and convergence
measurements have been conducted in this cross-section. Also, there are three foundation
loads from Terminal 5 located in this cross-section. The location of extensometers,
convergence pins and surface loads is shown in Figure 5.2.

The tunnel is excavated with two smaller pilot tunnels, labelled U2 and N2 in Figure 5.2,
followed by pillar removal. However, in section 39/317, a small part of the pillar was left for
an elevator shaft (excavated after the pilot tunnels). The shaft will not be taken into
consideration in this work. This will probably result in larger deformations in the numerical
analysis than the real case. The excavation of the pilot tunnels are referred to as excavation
stage 1, or stage 1. The removal of the pillar is referred to as stage 2 of the excavation.
 93

F1 F2 F3 F4

Extensometer 2

Extensometer 3
Extensometer 1
Convergence
10 m

pins
10 m

Pilot tunnel Pilot tunnel

U2 Pillar N2

Clay filled
Pegmatite structure

23 m

Figure 5.2 Schematic sketch for the excavation of Shuttle station 2.

5.2 Model setup and input data

FLAC 5.0 was used to model Shuttle station 2. In this case, no symmetry could be used due to
the geological conditions (anisotropy and large structures) and surface loads. The effect of the
pegmatite dyke was considered negligible and the clay filled structure and the mica schist
would serve as weaknesses. Since the cross-section of Shuttle station 2 is wider than the
cross-section in the conceptual analysis, the model size had to be somewhat altered. The
width of the model of Shuttle station 2 was set to 210 m, and the height to 80 m, see Figure
5.3, to avoid boundary effects. The inner, finer grid size was set to at least one tunnel diameter
in each direction, i.e. 70 m wide and 30 m high. Since this model is much larger than previous
models, the zone size could not bee kept the same. The finer grid size in this model is 0.25 x
0.25 m and the courser grid 0.5 x 0.5 m. An interface was used in FLAC to simulate the clay
filled structure, while the structure of the mica schist was simulated with ubiquitous joints.
 94

Finer grid
Zone size: 0.25 x 0.25
(m)

30
70

80
Coarser grid
Zone size: 0.5 x 0.5

210

Figure 5.3 Model and grid size for the Shuttle station 2 analysis.

The virgin state of stress at Shuttle station 2 was estimated from rock stress measurements
conducted by Bergsten et al. (1995) close to Shuttle station 2, see Figure 5.4. The
measurements were interpreted and gave the following stress relations,

z
VH [MPa] (5.1)
5.27

z
Vh [MPa] (5.2)
10.0

Vv Ugz [MPa] (5.3)

where z is the depth, U is the density of the material, g is the gravitational force, VH is the
major horizontal stress, Vh is the minor horizontal stress and Vv is the vertical stress. In Figure
5.4 it can be seen that the direction of VH is around 120 to 150q from north, which makes VH
perpendicular to the tunnel axis.
 95

VH [MPa] Vh [MPa]
-5 0 5 10 15 -10 -5 0 5 10
0 0

5 5

10 10
Depth [m]

Depth [m]

15 15

20 20

25 25

30 30

Vv [MPa] Orientation of VH [°]

-5 0 5 10 0 30 60 90 120 150 180
0 0

5
5

10

10
15
Depth [m]

Depth [m]

15 20

25
20

30

25
35

30 40

Figure 5.4 Stress measurements conducted close to Shuttle station 2, modified from
Bergsten et al., (1995).
 96

The strength of the rock mass at Shuttle station 2 was determined from mapping protocols.
The Q-system was used to classify the rock mass of the station. The Q-value was converted to
a GSI-value according to Equation (2.9), suggested by Hoek et al., (1995).

The result of the classification of the station is presented in Figure 5.5. The cross-section
examined in this work is at co-ordinate 39317. The GSI-value chosen for this analysis was 58,
illustrated by the dashed line in Figure 5.5, and the straight vertical line represents the cross-
section.

100

80

60
GSI-value

40

20

0
39240 39280 39320 39360 39400
x co-ordinate (m)

Figure 5.5 GSI-values for the x-coordinates of Shuttle station 2.

The intact uniaxial compressive strength was estimated using the values of the R-scale
(Brown, 1981). In the area of the cross-section 39/317, the R-values were R3 and R4. R3 has a
uniaxial compressive strength span of 25-50 MPa and R4, 50-75 MPa. According to RocLab,
both groups could include schist. The uniaxial compressive strength for the base case, or Case
1 was assumed to be 75 MPa, and another case (Case 2) was examined with an uniaxial
strength of 50 MPa, see Table 5.2.

There are three foundations that transfers load to the rock surface. A fourth, more widely
distributed load is located approximately 20 m away from the investigated cross-section, see
Figure 5.2 and Figure 5.6. These foundations transfers permanent loads of 7150 kN (F1),
11000kN (F2), 1300 kN (F3) and 10000 kN (F4). The load that the foundation transfers to the
rock surface has in this two-dimensional model been simulated as a distributed load in the
 97

cross-section with an infinite extension length in the tunnel direction. The foundation loads
have in this study been divided with the area of the foundation to get the load per meter
tunnel, see Table 5.1. Six different cases have been analysed, see Figure 5.7.

Figure 5.6 Location of surface loads

Table 5.1 Surface loads.

Load Area Effective load

2
F1 7150 kN 2.5 x 2.5 m 1.14 MPa
2
F2 1100 kN 2.5 x 2.5 m 1.76 MPa
F3 1300 kN 2.5 x 2.5 m2 0.58 MPa
2
F4 10000kN 5.5 x 9.2 m 0.20 MPa
 98

Case 1 Case 2
F1 F2 F3 F1 F2 F3
Considers the Sig_c = 75 MPa
Same as case 1, Sig_c = 50 MPa

mica schist but with lower

orientation, clay Vci.
filled structure
and loads 1 - 3

Case 3 Case 4
F1 F2 F3 F1 F2 F3
Same as case 1, Sig_c = 75 MPa
Same as case 1, Sig_c = 75 MPa

except no except no
consideration to consideration to
mica schist clay filled
orientation. structure.

Case 5 Case 6
Sig_c = 75 MPa
Same as case 1,
F1 F2 F3 F4
Sig_c = 75 MPa Same as case 1,
except no
except load F4 surface loads.
is considered.

Case 7
F1 F2 F3
Same as case 1, Sig_c = 75 MPa

except virgin Sig_H = 6.7 + 0.0444z

state of stress
same as base
case in
conceptual
analysis.

Figure 5.7 Description of the different cases of Shuttle station 2.

 99

Table 5.2 Data used for the analysis of Shuttle station 2, cases 1 to 7.
Case 1

Case 2

Case 3

Case 4

Case 5

Case 6

Case 7
Vci (MPa) 75 50 75 75 75 75 75
58 58 58 58 58 58 58
Rock mass

GSI
E (GPa) 17.6 17.6 17.6 17.6 17.6 17.6 17.6
c (MPa) 1.1 0.8 1.1 1.1 1.1 1.1 1.1
I (o ) 55 52.5 55 55 55 55 55
Vt (MPa) 0.32 0.21 0.32 0.32 0.32 0.32 0.32
c (MPa) 0.05 0.05 0.05 - 0.05 0.05 0.05
structure

I (o )
filled

25 25 25 - 25 25 25
Clay

Vt (MPa 0 0 0 - 0 0 0

c (MPa) 0.3 0.3 - 0.3 0.3 0.3 0.3

Ubiquitous

I (o ) 40 40 - 40 40 40 40
Joints

Vt (MPa 0 0 - 0 0 0 0

F1 (MPa) 1.14 1.14 1.14 1.14 1.14 - 1.14

Surface
loads

F2 (MPa) 1.76 1.76 1.76 1.76 1.76 - 1.76

F3 (MPa) 0.58 0.58 0.58 0.58 0.58 - 0.58
F4 (MPa) - - - - 0.20 - -
VH z/5.27 z/5.27 z/5.27 z/5.27 z/5.27 z/5.27 6.7+0.044z
state of
Virgin

stress

Vh z/10.0 z/10.0 z/10.0 z/10.0 z/10.0 z/10.0 0.8+0.033z

Vv Ugz Ugz Ugz Ugz Ugz Ugz Ugz

5.3 Focus areas of the analysis

Since Arlandabanan and Shuttle station 2 already are excavated and fully operational, it is
known that Shuttle station 2 is stable. It is also known that no major instability problems
occurred during the excavation of this tunnel. To be able to study the difference in stability
and behaviour of the different cases defined above, and to be able to compare data from
measurements conducted during the construction of the station, four different indicators were
studied. They are (i) tangential stress around the tunnel boundary, (ii) deformation of the
tunnel boundary, (iii) the area of plasticity and (iv) extensometer measurements. All analyses
have been divided into stage 1 (the two pilot tunnels), and stage 2 (removal of the pillar).
Short explanations of the analyses follow below.
 100

The analysis of deformations of the tunnel boundary (ii) includes a comparison with the
convergence measurements conducted when excavating the station. The same applies for
extensometer measurements, where the extensometer measurements are compared to the
expansion of the overburden.

(i) Tangential stress around the tunnel boundary

The tangential stress around the tunnel boundary is an important factor for the stability of
wedges. Low compressive or tensile tangential stress increases the risk of opening of pre-
existing joints and tensile failure, while high compressive stresses can lead to compressive
failure. The tangential stress was determined in 6 different points in each pilot tunnel and in 9
boundary points in the fully excavated station.

(ii) Deformation of the tunnel boundary

The deformation of the tunnel boundary can be an indicator of instability, but might also
affect rock reinforcement and installations such as electricity, water and ventilation. The
deformations of the models are compared with the results from convergence measurements
conducted during the excavation of the tunnel.

(iii) Deformation of the ground surface

During the construction of the Arlandabanan, the deformation of the ground surface was a
main concern. Terminal 5 has its foundations on the rock surface above Shuttle station 2. Not
only does the foundations distribute load on the rock mass, but the deformation of the ground
surface must also be controlled to avoid damage to the terminal.

(iv) Extensometer measurements

The expansion and/or contraction of the overburden were measured with three sets of
extensometer measurements. The expansion and/or contraction over the overburden is highly
connected to the subsidence or heaving of the ground surface, but not necessarily equivalent.
The vertical displacement of the overburden was measured for the different cases, and
compared to the measurements conducted when excavating the station.
 101

5.4 Results

It should be noted that in Case 2 (the uniaxial compressive strength, Vci, was 50 MPa instead
of 75 MPa) the model was unable to reach equilibrium in Stage 2 of the analysis (removal of
the pillar). The tunnel suffers chimney caving between load F2 and the clay filled structure,
see Figure 5.8. Therefore Case 2 will not be present in the presentation of results for Stage 2.
However, The results for Stage 1 will be presented since the model was able to reach
equilibrium in this part of the analysis.

Figure 5.8 The displacement directions for case 2, where chimney caving occurs when
pillar is removed.

5.4.1 Tangential stress around the tunnel boundary

Stage 1

Generally the tangential stress is low, yet compressive. Higher values are generally obtained
at measurement points 3 (~ 6-8 MPa) and 10 (~ 4-8 MPa), see Figure 5.9, Figure 5.10 and
Figure 5.11. Case 3 (does not include the orientation of the mica schist) and case 4 (does not
include the clay filled structure) show similar behavior of the tangential stress with very low
stresses in the tunnel wall of tunnel N2 while the abutment of tunnel N2 has higher stresses
than the other cases.

Case 6 (no surface loads) has generally lower tangential stresses around the tunnel boundary,
especially in the pillar which seems destressed. Case 7 (higher virgin state of stress) does also
show a destressed pillar, but with high stresses at measurement point 3 and 10.
 102

16

Tangential stress (MPa)

12 F1 F2 F3
Case 1 Sig_c = 75 MPa

0
1 2 3 4 5 6 12 11 10 9 8 7
U2 N2
16
Tangential stress (MPa)

12 F1 F2 F3
Sig_c = 50 MPa
Case 2

0
1 2 3 4 5 6 12 11 10 9 8 7
U2 N2
16
Tangential stress (MPa)

12 F1 F2 F3
Sig_c = 75 MPa
Case 3

0
1 2 3 4 5 6 12 11 10 9 8 7
U2 N2

Figure 5.9 Tangential stress around the pilot tunnels U2 and N2 for cases 1, 2 and 3.
 103

16
Tangential stress (MPa)

12 F1 F2 F3
Sig_c = 75 MPa
Case 4

0
1 2 3 4 5 6 12 11 10 9 8 7
U2 N2
16
Tangential stress (MPa)

12 F1 F2 F3 F4
Sig_c = 75 MPa
Case 5

0
1 2 3 4 5 6 12 11 10 9 8 7
U2 N2
16
Tangential stress (MPa)

12 Sig_c = 75 MPa
Case 6

0
1 2 3 4 5 6 12 11 10 9 8 7
U2 N2

Figure 5.10 Tangential stress around the pilot tunnels U2 and N2 for cases 4, 5 and 6.
 104

16

Tangential stress (MPa)

12 F1 F2 F3
Sig_c = 75 MPa

Case 7 Sig_H = 6.7 + 0.0444z

0
1 2 3 4 5 6 12 11 10 9 8 7
U2 N2

Figure 5.11 Tangential stress around the pilot tunnels U2 and N2 for case 7.

Stage 2

The results from the pillar extraction show that the stresses increase in magnitude compared
Stage 1 of the excavation, but the stresses can still be considered as fairly low. However, at
measurement point 3, which is located at the left abutment of the tunnel, the tangential stress
is higher than the other measurement points, while the tangential stress in measurement point
4, on the left side of the tunnel roof, is considerably than in the others points. This applies for
all cases except case 6 (no surface loads) and case 7 (higher virgin state of stress). In case 6,
the tangential stress is more evenly distributed over the tunnel boundary. However, the
tangential stress at the right abutment and wall of case 6 is lower than in the other points. Case
7 show high tangential stress over the tunnel roof. The results of this analysis are presented in
Figure 5.12 and Figure 5.13.
 105

16
Tangential stress (MPa)

12 F1 F2 F3
Sig_c = 75 MPa
Case 1

0
1 2 3 4 5 6 7 8 9
16
Tangential stress (MPa)

12 F1 F2 F3
Sig_c = 75 MPa
Case 3

0
1 2 3 4 5 6 7 8 9

16
Tangential stress (MPa)

12 F1 F2 F3
Sig_c = 75 MPa
Case 4

0
1 2 3 4 5 6 7 8 9

Figure 5.12 Tangential stress around the tunnel after stage 2, for cases 1, 3 and 4.
 106

16

Tangential stress (MPa)

F1 F2 F3 F4
12 Sig_c = 75 MPa

Case 5
8

0
1 2 3 4 5 6 7 8 9
16
Tangential stress (MPa)

Sig_c = 75 MPa
12
Case 6

0
1 2 3 4 5 6 7 8 9
16
Tangential stress (MPa)

12 F1 F2 F3
Sig_c = 75 MPa

Sig_H = 6.7 + 0.0444z

Case 7

0
1 2 3 4 5 6 7 8 9

Figure 5.13 Tangential stress around the tunnel after stage 2, for cases 5, 6 and 7.
 107

5.4.2 Deformation of the tunnel boundary

Stage 1

The results from the analyses of the deformation of the tunnel boundary of stage 1 are
presented in Figure 5.14 and Table 5.3. The deformations from the numerical analyses are
multiplied with a factor of 300 to become visible in Figure 5.14 while the measured
convergences are only showing the general direction of the deformations. Cases 1 to 6 show
that the tunnel walls deform inwards and limited, if any, deformation is seen in the pillar. The
tunnel roof is deforming inward in all cases except case 6, where no consideration is taken to
surface loads. A shear displacement can be seen in the clay filled structure for all cases except
case 4, where the structure is not present. Case 7 (higher virgin state of stress) show greater
deformation than the other cases. The multiplication factor had to be reduced to 200 to get an
understandable image of the deformation in Figure 5.14. It can also be seen that some heaving
of the floor of tunnel N2 (right) occur.

The convergence measurements at Shuttle station 2, show that most of the deformation is in
the horizontal plane. The walls converge, while the pillar contracts.

A comparison of the tunnel measurements and the calculated displacements for the numerical
analyses, show that the measurement points in the tunnel walls (1 and 4) show good
agreement. The measurement point in the abutments (2 and 3) show good agreement in the x-
direction, while the deformations in the y-directions are contradictory. The measured
displacements in the pillar and the calculated displacements have moved in opposite
directions.
 108

Case 1 Case 2 Case 3

Maximum deformation: 2.0 mm Maximum deformation: 1.8 mm Maximum deformation: 1.8 mm

Case 4 Case 5 Case 6

Maximum deformation: 1.4 mm Maximum deformation: 2.1 mm Maximum deformation: 1.9 mm

Case 7 Deformation direction of measured data

Maximum deformation: 7.5 mm Maximum deformation: 1.3 mm
2 3

1 4
5 6

Theoretical boundary
Deformation direction

Figure 5.14 Deformation of the tunnel boundaries for Stage 1 calculated for cases 1 to 7 and
deformations from convergence measurements of the pilot tunnels. The
deformations from numerical analyses are multiplied with a factor of 300 for all
cases except case 7, which is multiplied with 200. The measured convergences
only show general direction.
 109

Table 5.3 Results from convergence measurements and values calculated for the different
cases for stage 1.

Case 1 (mm)

Case 2 (mm)

Case 3 (mm)

Case 4 (mm)

Case 5 (mm)

Case 6 (mm)

Case 7 (mm)
values (mm)
Measured

2 3

1 5 6 4

1 xdisp 0,9 1,8 1,8 1,2 1,2 1,8 1,9 6,9

ydisp -0,2 -0,5 -0,5 -0,4 -0,4 -0,4 -0,0 -0,7
2 xdisp 0,9 1,5 1,5 1,2 1,2 1,4 1,3 7,5
ydisp 0,3 -1,8 -1,8 -1,1 -1,1 -1,7 0,2 -1,7
3 xdisp -1,3 -0,3 -0,5 -1,0 -1,0 -0,3 -1,3 -4,3
ydisp -0,1 0,0 0,2 0,4 0,4 -0,4 0,6 -2,1
4 xdisp -0,7 -1,8 -1,8 -1,6 -1,6 -1,8 -1,7 -6,0
ydisp -0,2 -0,2 -0,1 -0,1 -0,1 -0,2 -0,4 -2,4
5 xdisp 0,8 -0,5 -0,5 -0,4 -0,4 -0,5 -0,5 -0,7
ydisp -0,2 0,2 0,2 0,0 0,0 0,2 -0,1 -0,4
6 xdisp -1,3 1,1 1,2 0,5 0,5 1,1 0,7 5,0
ydisp -0,6 -0,9 -0,9 -0,5 -0,5 -1,3 0,1 -3,9

Stage 2

The results from the pillar extraction (stage 2) are shown in Figure 5.15 and Table 5.4. The
deformations calculated in the numerical analyses shown in Figure 5.15 are multiplied with a
factor of 150, to become visible, while the measured convergences of Shuttle station 2 only
show the general direction of the deformations.

Cases 1 to 7 show that the tunnel roof moves downwards. Case 1, 3 and 5 show similar
behavior with a maximum total deformation at the clay filled structure of 23 to 26 mm. Cases
4 (does not include the clay filled structure) and 6 (no surface loads) show considerably less
deformation of the tunnel boundary. Case 7 (higher virgin state of stress) shows more
deformation of the tunnel walls, and heaving of the tunnel floor. It can also be seen that the
effect of the clay filled structure is reduced, when the stresses are higher.

Convergence measurements of the tunnel show subsidence of the left part of the tunnel
(measurement points 1 and 2) while the middle and right part (measurement points 3, 4 and 5)
of the tunnel heaves. The numerical analyses and the measured values show better agreement
in the left side of the tunnel than at the right side.
 110

Case 3
Case 1 Case 2
Maximum deformation: 26.0
Maximum deformation: 22.7 mm Maximum deformation: f mm
mm

The tunnel fails

Case 5
Case 4 Case 6
Maximum deformation: 23.0
Maximum deformation: 7.5 mm Maximum deformation: 2.9 mm
mm

Case 7 Deformation direction of measured data

Maximum deformation: 14.8 mm Maximum deformation: 1.0 mm
3
4
2

5
1

Theoretical boundary
Deformation direction

Figure 5.15 Deformation of the tunnel boundaries for Stage 2 calculated for cases 1, 3, 4, 5,
6 and 7 and deformations from convergence measurements of the pilot tunnels.
The deformations from numerical analyses are multiplied with a factor of 150.
The measured convergences only show general direction.
 111

Table 5.4 Results from convergence measurements and values calculated for the different
cases for stage 2.

Case 1 (mm)

Case 3 (mm)

Case 4 (mm)

Case 5 (mm)

Case 6 (mm)

Case 7 (mm)
values (mm)
3
Measured

2 4

1 5

1 xdisp 1,0 2,0 1,0 1,5 1,7 1,86 7,9

ydisp -0,6 -1,4 -1,7 -0,9 -1,2 0,01 -1,7
2 xdisp 0,5 2,0 1,8 1,0 1,4 1,35 8,9
ydisp -0,6 -8,7 -8,5 -5,0 -8,8 0,04 -5,2
3 xdisp 0,1 3,2 3,2 1,3 2,4 0,30 7,3
ydisp 0,5 -16,3 -17,6 -8,6 -16,5 -0,23 -9,8
4 xdisp -0,1 -1,0 -1,6 1,2 -0,2 -1,33 -6,0
ydisp 1,0 -3,5 0,1 -8,4 -8,8 0,55 -5,0
5 xdisp 0,1 -2,0 -1,9 -1,7 -1,9 -1,73 -7,8
ydisp 0,5 -0,3 -0,1 0,1 -0,2 -0,33 -3,0

5.4.3 Deformation of the ground surface

Stage 1

The results from the analysis of stage 1 are presented in Figure 5.16. It is shown that the left
part subsides, while the right part, over pilot tunnel N2 heaves. This is partly due to a shear
movement along the structure and the direction of the mica schist. Case 3 (does not include
mica schist orientation) and Case 4 (does not include the clay filled structure) show less
subsidence. The removal of the surface loads (Case 6) reduces the subsidence greatly and
higher horizontal stresses (Case 7) increased the heaving greatly. In the latter case, a smaller
peak in the heaving can be seen. This is because the rock mass is consolidated at ground
surface near the clay filled structure, see Figure 5.17.
 112

12 12
Case 1 Case 3
Case 2 Case 4

8 8
Deformation (mm)

Deformation (mm)
4 4

0 0

-4 -4
0 50 100 150 200 0 50 100 150 200
Ground surface (m) Ground surface (m)

12 12
Case 5 Case 7
Case 6

8 8
Deformation (mm)

Deformation (mm)

4 4

0 0

-4 -4
0 50 100 150 200 0 50 100 150 200
Ground surface (m) Ground surface (m)

Figure 5.16 Deformation of the ground surface in Stage 1 for cases 1 to 6.

 113

Figure 5.17 Displacement vectors of Case 7, higher virgin state of stress.

Stage 2

The behaviour of the ground surface after stage 2 is similar to that of Stage 1, with the
difference that the subsidence is greater, see Figure 5.18. Again, especially case 4 (no clay
filled structure) shows less subsidence. Case 6 (no surface loads) shows a tendency of
heaving. A notch can be seen in Case 1, Case 4 and Case 6. This is a small part of the rock
surface that heaves due to punching of surface load 2 into the ground surface, see Figure 5.19.
Case 7 (higher virgin state of stress) shows similar patterns as for the pilot tunnels.
 114

Case 1 Case 3
Case 4
10 10

Deformation (mm)
Deformation (mm)

0 0

-10 -10

-20 -20
0 50 100 150 200 0 50 100 150 200
Ground surface (m) Ground surface (m)

Case 5 Case 7
Case 6
10 10
Deformation (mm)

Deformation (mm)

0 0

-10 -10

-20 -20
0 50 100 150 200 0 50 100 150 200
Ground surface (m) Ground surface (m)

Figure 5.18 Deformations of the ground surface in Stage 2 for all cases for stage 2, except
case 2, which fails. Observe the difference in the scale to stage 1.
 115

Figure 5.19 Close up on surface load F2 (blue arrows) and displacement vectors (green
arrows) of Case 1.

5.4.4 The area of plasticity

Stage 1

The area of plasticity developed for the different cases of Shuttle station 2 consists mostly of
tensile failure, and slip along ubiquitous joints (i.e. slip along the mica schist). Figure 5.20
shows that tensile failure occurs in the wall and abutment of pilot tunnel N2 for all cases. In
case 6 (no surface loads) tensile failure can be observed at the wall of pilot tunnel U2 as well.
Ubiquitous joint failure can be observed in the pillar and in the wall of tunnel U2 for all cases
except case 3 (without ubiquitous joints) and case 4 (no clay filled structure). When the virgin
stresses were increased (Case 7) the zone of plasticity increases greatly. The pillar and the
tunnel walls have experienced tensile failure, and slip along ubiquitous joints has occurred in
the pillar, the left tunnel wall and over the right abutment. A band of shear failure extends
from the intersection of the tunnel roof and the structure and up to the ground surface. This
can be explained with the displacement vectors in Figure 5.17.
 116

Area of plasticity Tensile strength (MPa)

Case 1

Case 2

Case 3

Case 4

Case 5

Case 6

Case 7

Figure 5.20 Area of plasticity and tensile strength of case 1 to 7 in Stage 1.

 117

Stage 2

The area of plasticity grows after excavating the pillar, see Figure 5.21. Tensile failure can
bee seen in a region at the right tunnel abutment/roof. This is more or less obvious in all
cases, except for case 6 (no surface loads) Furthermore, it shows that the failure consists of
tensile failure, shear failure underneath the surface load F2, slip and tensile failure along
ubiquitous joints.

Just left of surface load F2, an area of tensile failure can be seen for all cases with a clay filled
structure, surface loads and low stresses, i.e. Case 1, Case 2, Case 4 and Case 5. This area
originates from a bending motion of the overburden created by slip in the clay filled structure
and the surface load, see Figure 5.15 and Figure 5.21. When the overburden bends down
along the clay filled structure, tensile failure takes place above the left abutment of the tunnel.

Furthermore, it can be seen that slips along ubiquitous joints occur in the tunnel walls and at
the right abutment.
 118

Area of plasticity Tensile strength (MPa)

Case 1

Case 2

Case 3

Case 4

Case 5

Case 6

Case 7

Figure 5.21 Area of plasticity and tensile strength of case 1 to 6 in Stage 2.

 119

5.4.5 Extensometer measurements

Stage 1

Extensometer measurements were conducted when the station was constructed. The
expansion and/or compaction of the overburden was calculated from the results of the
analyses and compared with real deformations.

Extensometer 1 and the corresponding values from the analyses show similar results, see
Figure 5.22. Extensometer 2 shows an expansion of 3 mm above the pillar, where the models
show values less than 0.5 mm, except for Case 7 (high virgin state of stress) that has an
extension of about 6 mm. Extensometer 3 returns a negative value, i.e. contraction of the
overburden. This is not seen in any of the cases. Again, Case 7 shows a great extension, this
time almost 9 mm.

12
Measured extension
10 Case 1
Case 2
8 Case 3
Extension (mm)

Case 4
6 Case 5
Case 6
Case 7
4

0
1 2 3
-2
Extensometer

1 2 3

Extensometers

Figure 5.22 Results from extensometer measurements of Shuttle station 2, and calculated
values from cases 1 to 7.
 120

Stage 2

In stage 2, extensometer 2 is missing due to excavation of an elevator shaft. Extensometer 1

shows an expansion of approximately 0.5 mm, which is comparable to the modeled data, see
Figure 5.23. Extensometer 3 shows an extension slightly greater than 0.5 mm. Cases 1, 4, 5
and most notably Case 7 show much larger expansions of the overburden. Case 3, (no
ubiquitous joints) shows a similar extension of the overburden as the one measured in the case
study.

14

Measured extension
12
Case 1
Case 3
10 Case 4
Extension (mm)

Case 5
8 Case 6
Case 7
6

0
1 2 3
Extensometer

1 2 3

Extensometers
Figure 5.23 Results from extensometer measurements (only extensometer 1 and 3) of Shuttle
station 2, and calculated values from all cases except case 2.
 121

6 DISCUSSION

To achieve the objectives of this project, conceptual analyses and analyses of an existing case
have been conducted. Important questions are, how can the results be used to describe the
behaviour and is it possible to translate the results into an absolute stability prognosis of a real
case?

6.1 Conceptual analysis

The procedure of estimating the rock mass parameters for the conceptual models have
resembled the established routines for data collection of real cases, i.e. the strength parameters
such as cohesion, friction angle, tensile strength and compressive strength and global strength
and Young’s modulus have been obtained from the intact uniaxial compressive strength and
classification systems, in this case, GSI and the computer program RocLab (RocScience,
2005). The factors have been varied in such a way as to resemble the measurements and
estimations of such parameters in real preliminary tunnel investigations. This was done with
respect to the precision and uncertainty of methods used to estimate/calculate the parameters.
The variation of virgin stress state has been done by using a number of stress versus depth
expressions presented in the literature

In the conceptual analysis it was shown that the interval of GSI has a great impact on
deformations, tangential stresses at and close to the boundary and the extent of the area of
plasticity. As shown in Edelbro (2007) the tables, used for GSI, may be experienced as
inaccurate as they are very basic. This means that the results from a certain site will be
subjective and may vary depending on the persons who classify the rock mass.

The area of plasticity in the models showed a fairly similar pattern for all varied factors.
Mainly tensile yielding occurred, and took place in the walls and abutments of the tunnel, and
propagated into the rock mass. Although, it is impossible to estimate exactly how this affects
the stability of the tunnel, it can be assumed that a larger area of plasticity is less favourable
than a small area, especially if it reaches to the ground surface. When the area where tensile
stresses, or very low compressive stresses, increases, the risk of fallouts may increase.
Furthermore, failure of the upper parts of the walls (areas where all stresses are tensile in the
elastic analyses) may undermine the abutment, and may lead to larger fallouts.

Since the confining stress is very low, and failure occurs mainly in the form of tensile
yielding, the Mohr-Coulomb criteria is vulnerable in accurately capturing the yield process,
since it is based on simultaneous mobilization of friction and cohesion (Saiang and Nordlund,
 122

2007). The strength of the rock mass will therefore depend more on the tensile and cohesive
strength, than on friction, at least in a continuum model and assuming that intact rock bridges
are present.

Since the tensile strength of the rock mass is calculated from the intact uniaxial compressive
strength, (ıci), and GSI, variations in GSI and ıci will give variations in the tensile strength,
which leads to variations in the area of plasticity. This is a rough simplification and the true
tensile strength of the rock mass is probably not equal to the calculated tensile strength. The
calculated value is an average for the whole rock mass. The true tensile strength is a
combination of the intact tensile strength for all points where no natural weaknesses are
present. If a discontinuity is present, the true tensile strength is very low or zero depending on
filling, intact rock bridges and length of the discontinuity. Moreover, the tensile strength is
direction dependent; this includes also the residual tensile strength, which was discussed
earlier in Chapter 4. Perpendicular to a discontinuity, the tensile strength is zero, while it may
be close to the intact tensile strength parallel to the discontinuity. However, to get this kind of
behaviour in numerical models, all discontinuities must be present in the model. Since the
structural geology of a rock mass is unique at every cross-section this kind of behaviour could
not be studied in the conceptual analysis.

For the conceptual analyses, the only factor that truly indicated instability was the existence of
large geological structures. This was so for relatively flatly dipping discontinuities (dip d 30o),
although no analyses were conducted to determine whether there is a relation between the dip
angle and the friction angle. If this is the case, it may be possible to judge the stability by
comparing the dip angle and the friction angle. For all other factors in the conceptual analysis,
it could only be said that the factors have more or less impact on the stability. It seems that a
shallow tunnel in hard rock without larger discontinuities is rather stable, and if excavated and
reinforced properly, there should not be any major instability problems.

At shallow depths the induced stresses can be very complex due to the geological conditions
and the highly anisotropic state of stress. Anisotropy has not been studied in the conceptual
analysis, but can probably affect the stability.

The deformation of the rock mass that can be considered as a continuum surrounding a
shallow tunnel appears mainly as heaving of the ground surface and tunnel roof and
convergence of the tunnel walls. However, the analyses in this thesis does not show when the
deformations occur relative to the tunnel face since the models are 2-dimensional. The
reinforcement of the tunnel will be affected differently depending on when it is installed. If
shotcrete, for example, is installed immediately after excavation, there is a risk that tensile
 123

stresses will be transferred through the rock-shotcrete interface in the walls and abutments of
the tunnel. Compressive stresses might be induced in the tunnel roof. Since this work did not
consider the face advance, no recommendation on when reinforcement should be installed can
be made.

6.2 Case Study

The data for the Shuttle station 2 analyses comes from mapping, measurements and
estimations conducted before and during the construction and excavation of the tunnel. Some
of the data was less reliable, like the value for the intact uniaxial compressive strength, Vci,
which comes from a simple estimation of R-values (Brown, 1981) along the tunnel. This is a
very rough way of estimating Vci and data from uniaxial compressive tests or point load tests
would have been preferred. Two values of Vci were chosen for the analyses. The lower value
led to collapse of the tunnel. The uniaxial compressive strength did not show great importance
in the conceptual analysis but showed to have a major effect on Shuttle station 2. A probable
reason why Case 2 fails is that the surface loads induce shear failure in the clay filled structure
and a bending failure of the left side of the overburden which leads to tensile yielding over the
left abutment, see Figure 6.1. The same kind of behaviour can be seen in the models that
include the clay filled structure, surface loads and a low state of stress but with a more limited
zone of bending failure in the left part of the overburden. When the intact uniaxial
compressive strength is reduced, the tensile strength of the rock mass is also reduced, and this
area of tensile yielding reaches all the way from the ground surface to the tunnel roof. The
cohesion and friction of the rock mass are also reduced when the intact uniaxial tensile
strength is reduced. This, along with the weakened rock mass due to tensile yielding leads to a
shear failure, induced by the surface load F2, from the load to the left abutment. The whole
roof is then pushed down by gravity and the surface load.

Figure 6.1 A simplification of the failure mechanism of Case 2.

 124

The reason why the intact uniaxial compressive strength has such an impact on the stability of
Shuttle station 2, while it has little impact on the conceptual analysis may be due to a variety
of many reasons. Firstly, the two values of the uniaxial compressive strength chosen for
Arlandabanan differed 50% while they only differed 20% in the conceptual analysis.
Moreover, the Arlandabanan values were considerably lower than the values used in the
conceptual analyses. The accuracy needed in estimation of data is probably more important
for low strength values. Finally, Arlandabanan contained surface loads and anisotropy and a
discontinuity while the conceptual analysis did not.

The cross-section used in the analysis was chosen because it included both convergence
measurements and extensometer measurements. However, a smaller part of the pillar was left
for an elevator shaft in the cross-section, which makes it a 3D problem. Because all earlier
analyses were conducted with 2D, this case was also analysed in 2D, and the elevator shaft
was neglected. This is a source of error for the pillar extraction stage (stage 2).

Case 3 (no mica schist) and Case 4 (clay filled structure) showed similar behaviour. From this
it can be concluded that geological structures and anisotropy can be dealt with in the same
way, and that the angle of the weaknesses are the most important parameter.

As for the conceptual analysis, it can be seen that low compressive and/or tensile stresses are
present in the walls and in some cases, in the abutments of the tunnel, while they are higher in
the roof. The area of plasticity in the conceptual models as well as in the models of
Arlandabanan show that tensile failure occurs mostly in the walls and abutments of the tunnel.
This can especially be noticed in the right pilot tunnel, N2. When the theoretical and
measured cross-sections are studied, tendencies can be seen that support these analysis results.
There is a tendency that the theoretical and measured tunnel section differ most in the right
side of the tunnel, see Figure 6.2.

Figure 6.2 Theoretical and measured cross-sections 39/289, 39/317 and 39/324 from
Shuttle station 2.
 125

There is a difference between the measured values of the deformations from the tunnel
excavation and those calculated in the analyses of the tunnel. The deformations measured by
the extensometers and the corresponding deformations obtained in the analyses shows the
largest difference, see Figure 5.22. When the pilot tunnels were excavated, there was a
difference between the reading of extensometer 2 and the corresponding deformation obtained
by numerical analysis. Case 7, with higher horizontal virgin stresses than those measures
close to the studied cross-section shows deformations similar to those measured by
extensometer 2. Heaving of the tunnel roof inducing tensile failure of the pillar was also
observed when the pilot tunnels were excavated. This is also seen in Case 7, see Figure 5.20.
One explanation is that the interpretation of the stress measurements was too conservative and
that higher stresses are present in addition to the fact that the surface loads may have been
overestimated when they were calculated for 2D conditions.

The extensometer measurements of stage 2, the pillar extraction, see Figure 5.23, resemble
Case 3 (no mica schist orientation). This shows that the significance of the mica schist might
have been over-exaggerated.

6.3 Summary

Yielding in points of the rock mass does not have to mean a stability problem for a tunnel.
The behaviour of the tunnel is well described by the tangential state of stress and the
deformations. However, it is difficult to translate the results from the conceptual analyses and
the Arlandabanan case into statements about the stability. It is known that no major instability
problems occurred while constructing Shuttle station 2.

Several well known and frequently used tools in rock mechanics have also been tested in this
work. First of all, GRC (Ground reaction curve) was planned to be used as a way to analyse
the stability and behaviour. However, this proved to be impossible, since all models in the
conceptual analysis experienced heaving of the overburden and GRC is founded on the idea of
tunnel roof convergence. Moreover the Factor of safety (FOS) function in FLAC was also
tested. This function calculates a relation between the rock mass strength of the existing
tunnel, and the rock mass strength when the tunnel collapses. This proved not to work since
the strength had to be reduced to such a degree that the rock mass failed in compressive
failure due to the virgin state of stress before the tunnel was excavated.
126
 127

7 CONCLUSIONS

Based on the conceptual analysis and the case study it can be concluded that:

- The failure that occurred in the conceptual analyses and the case study is tensile which
means that the tensile strength of the rock mass is an important factor. Furthermore, the
behaviour and stability of a shallow tunnel is more sensitive to variations in the residual
tensile strength than the peak tensile strength.
- The structural geology of the rock mass is the most important factor to consider when
analysing the behaviour and stability of a shallow tunnel. Moreover, the behaviour and
stability of a shallow tunnel is most sensitive to variations in the orientation of the
structures. Since the tensile strength of the rock mass depends strongly on the tensile
strength of discontinuities, the spacing and the number of joint sets are of great
importance.
- The behaviour and stability of a shallow tunnel is highly sensitive to variations of the
virgin state of stress. Intermediate stresses are favourable. Low stresses give lower
tangential stresses in the boundary of the tunnel, which increases the risk of fallouts. Very
high horizontal stresses gives larger deformations and greater tensile stresses in the area of
the abutments of the tunnel, under the assumption that the rock mass can be considered as
a continuum.
- Since the rock mass strength (compressive as well as tensile) is determined with the help
of GSI and the intact uniaxial compressive strength, Vci, it is important to be able to
determine these parameters as accurate as possible. GSI is more uncertain than the intact
uniaxial compressive strength, at least when the latter has been measured with unaxial
compressive tests or point load tests. This means that GSI is a more critical parameter in
stability analysis.
- It is equally important to know the extent and reduction of strength of the weaknesses of
the rock mass (weathering, blast damage etc.) and the rock mass strength.
- The greatest risk of instability of a shallow tunnel, in a rock mass without larger
geological structures, is fallouts from the walls and abutments of the tunnel.
- It is recommended that parameter studies are used to investigate the sensitivity of the
factors that have larger impact on the behaviour and stability and/or are considered to be
more uncertain when conducting a stability analysis of a shallow tunnel.

A general overview over how sensitive the behaviour and stability are to variations of the
different factors based on the results from the conceptual analyses and the case study is
presented in Table 7.1. The impact of the variations of the different factors has on the
instability indicators are divided into three categories, high, medium and low impact.
 128

Table 7.1 The significance level that the different parameter groups have on the different
instability indicators.

Extent of
Deformation of Subsidence / Area of
VI, tunnel boundary tensile
tunnel boundary heaving Plasticity
stresses
Virgin state of stress High High High High High

Vci Medium Medium Medium Medium -

GSI High High High High -

Vt High Medium High High -

Over burden High High High Medium Medium

E Low Medium Medium Low -

c/I Low Low Low Low -

Weathering High Medium Low Low -

EDZ High Low Low Low -

Structural geology High High High High -

Surface loads Medium High High Medium -

Tunnel size and
Medium High High High -
geometry
 129

8 RECOMMENDATIONS FOR FUTURE WORK

This project has shown that stability analyses using software based on the assumption of
continuous conditions cannot tell whether shallow underground constructions in hard rock are
stable or not. Furthermore, the project revealed some issues which need to be further studied
in order to improve the understanding of stability analyses of shallow tunnels. It is
recommended that a methodology for geomechanical modelling of shallow tunnels is
developed, consisting of
- recommendations for collecting of data,
- a suitable failure criteria,
- proper instability indicators.

Since using GSI and Vci together with RocLab was proven to be important, and at the same
time fairly uncertain, it is recommended that more research is done in this area. Moreover, the
Mohr-Coulomb failure criteria might not be suitable for analysis of shallow tunnels, since
tensile failure is the main type of failure of shallow tunnels and the Mohr-Coulomb failure
criteria is constructed for mainly compressive failure. This goes hand in hand with the
development of more suitable instability indicators for shallow tunnels. The indicators in this
work could not quantify the stability of the tunnel in a satisfactory manner.
130
 131

9 REFERENCES

Amadei, B., Stephansson, O. (1997) Rock stress and its measurement. London: Chapman and
Hall.

Banverket (2002) BV tunnel. Standard BVS 585.40.

Barton, N., By, T.L., Chryssanthakis, L., Tunbridge, L:, Kristiansen, J., Löset, F., Bhasin,
R.K., Westerdahl, H., Vik, G. (1992) Comparison of prediction and performance for a 62 m
span sports hall in jointed gneiss. Proc. 4th. in. rock mechanics and rock engineering conf.,
Torino. Paper 17.

Barton, N., Löset, F., Lien, R., Lunde, J. (1980) Application of the Q-system in design
decisions. In Subsurface space, (ed. M., Bergman), 2, pp. 553-561. New York: Pergamon.

Barton, N.R., Lien, R., Lunde, J. (1974) Engineering classification of rock masses for the
design of tunnel support. Rock Mech., 6, pp. 189-239.

Berg, S. (2005) Bergspänningsmätningar på litet djup. Master thesis 2005:426, Luleå

University of Technology, Luleå, ISSN:1402-1617(in Swedish).

Bergsten, K.,Å., Ljungren, C., Söder, P.E., Wikman, A. (1995) Mätrapport,

Bergspänningsmätningar på Arlanda, Arlandabanan, flygstationsområdet. Vattenfall
Hydropower, Internal report (in Swedish).

Bieniawski, Z.T. (1974) Geomechanics classification of rock masses and its application in
tunnelling. In: Proceedings of the 3rd International Congress on Rock Mechanics, Denver, pp.
27-32.

Bieniawski, Z.T. (1976) Engineering classification in rock engineering. In: Proceedings of

the Symposium on Exploration for Rock Engineering, Johannesburg, pp. 97-106.

Bieniawski, Z.T. (1989) Engineering rock mass classifications. John Wiley & Sons, New
York.

Boussinesq, J. (1883) Application des potentiels á l’étude de l’équilibre et du mouvement des

solides élastices. Paris, Gauthier – Villars.
 132

Brown, E.T., Editor (1981) Rock Characterization, Testing and Monitoring. ISRM Suggested
Methods. Pergamon Press, Oxford, 211 p.

Chang, Y., Hansen, L., Söder, P.E., Hässler, L. (1996) Arlandabanan, Projektunderlag för
permanent bergförstärkning. Internal report, Linkprojektörerna, Vattenfall Hydropower,
NCC Teknik, SIAB Teknik, Tyréns Infrakonsult, (in Swedish).

Chang, Y., Hellstadius, K. (1998) Shuttle station 2, Arlandbanan – bergprojektering och

uppföljning. In: Bergmekanikdagen 1998, SveBeFo, Stockholm (in Swedish).

Diederichs, M.S. (1999) Instability of hard rock masses: The role of tensile damage and
relaxation. Doctoral Thesis, University of Waterloo, Waterloo, Ontario, Canada.

Edelbro, C. (2004) Evaluation of rock mass strength criteria. Licentiate Thesis 2004:72,
Luleå University of Technology. ISSN:1402-1757

Edelbro, C., Sjöberg, J., Nordlund, E. (2007) A quantitive comparison of strength criteria for
hard rock masses. Tunneling and Underground Space Technology 22 (2007), pp. 57-68.

Edelbro, E. (2003) Rock mass strength – a review. Technical report 2003:16, Department of
Civil and Mining Engineering, Division of Rock Mechanics, Luleå university of Technology.
ISSN:1402-1536

Emsley, S., Olsson, O., Stenberg, L., Alheid, A-J., Falls, S. (1997) ZEDEX – A study of
damage and disturbance from tunnel excavation by blasting and tunnel boring. Stockholm
(Swedish Nuclear and Waste Management CO).

Fairhurst, C. (1986) In-situ stress determination – an appraisal of its significance in rock

mechanics. In Proc. Int. Symposium on Rock Stress and Rock Stress Measurements,
Stockholm, Centek Publ., Luleå, pp 3-17.

Fredriksson, A. (2006) Personal communication

Gaziev, E., Erlikhman, S. (1971) Stresses and strains in anisotropic foundations. In:
Proceedings Symposium on Rock Fracture, ISRM (Nancy), Paper II-1.
 133

Grimstad, E., Barton, N. (1993) Updating the Q-system for NMT. In: Proceedings of the First
International Symposium on Sprayed Concrete, Fagernes, Norway 1993, Norwegian
Concrete Association, Oslo.

Hoek, E., Brown, E.T. (1980) Underground excavations in rock. Revised first edition.
London: Institution of Mining and Metallurgy.

Hoek, E., Kaiser, P.K., Bawden, W.F. (1995) Support of underground excavations in hard
rock. A.A. Balkema, Rotterdam, Brookfield.

Hoek,. E., Carranza-Torres, C., Corkum, B. (2002) Hoek-Brown failure criterion – 2002
edition. In: Proceedings of the 5th North American Rock Mechanics Symposium and 17th
Tunnelling aAssociation of Canada Conference: NARMS-TAC 2002, July 7-10, University
of Toronto, pp. 267-271.

Hudson, J.A., Cooling, C.M. (1988) In situ rock stresses and their measurement in the UK –
Part I. The current state of Knowledge. In: Int. J. Rock Mech. Min. Sci. & Geomech. Abstr.,
Vol 25, pp. 363-70.

Itasca (2005a) FLAC version 4.0. Manual. Minneapolis, ICG.

Itasca (2005b) UDEC version 4.0. Manual. Minneapolis, ICG.

Jing, L. (2003) A review of techniques, advances and outstanding issues in numerical

modelling for rock mechanics and rock engineering. In: Int. J. Rock Mech. & Min. Sci., Vol
40, pp. 283-353.

Leijon, B.A. (1989) Relevance of point wise rock stress measurements – an analysis of
overcoring data. In: Int. J. Rock Mech. Min. Sci & Geomech. Abstr., vol 26, pp. 61-8.

Malmgren, L. (2005) Interaction between shotcrete and rock – experimental and numerical
study, Doctoral thesis 2005:48,Luleå University of Technology. ISSN:1402-1544.
 134

Martino, J.B. (2002) The excavation damage zone in recent studies at the URL, The 2002
international EDZ workshop, The excavation damage zone – causes and effects. Edited by
J.B. Martino, Atomic Energy of Canada Limited.

Mindlin, R.D. (1939) Stress distribution around a tunnel. Transactions American Society of
Civil Engineers, reprinted from Proceedings of the American Society of Civil Engineers,
Paper no. 2082.

Mindlin, R.D. (1948) Stress distribution around a hole near the edge of a plate under
tension. Proceedings of the Society for Experimental Stress Analysis V, No. 2, pp. 56-68.

Palmström, A. (1995) RMi – A rock mass characterisation system for rock engineering
purposes. Ph.D. Thesis, University of Oslo, Norway.

Poulos, H.G., Davis, E.H. (1974) Elastic solutions for soil and rock mechanics. New York,
John Wiley & Sons.

Ran, J.Q., Passaris, E.K.S., Mottahead, P. (1994) Shear sliding failure of the jointed roof in
laminated rock mass. Rock mechanics and Rock Engineering, 27 (4), pp. 235-51.

Rock Engineering group (1996) Examine2D version 6.0, User’s guide. University of Toronto.

RocScience (2006). Program: “RocLab”. URL:

http://www.rocscience.com/products/roclab.asp

Saiang, D., Nordlund, E. (2006) Continuum Analysis of Shallow Tunnels in Brittle Rock with
Damaged Rock Zone Around Their Boundaries – A Parameter Study, Luleå University of
Technology, Sweden, To be published.

Serafim, J.L., Pereira, J.P. (1983) Considerations of the geomechanical classification of

Bieniawski. Proceedings Int. Symposium in Engineering Geology and Underground
Construction (Lisbon), 1(II), pp. 33-42. Lisbon, SPG, LNEC.

Stephansson, O. (1993) Rock stress in the Fennoscandian shield, In: Comprehensive Rock
Engineering (ed. J.A. Hudson), Pergamon Press, Oxford, Chapter 17, Vol 3, pp. 445-59.
 135

Stille, H., Eriksson, M., Nord, G. (2004) Kompendium i bergmekanik. The Royal Institute of
Technology, Stockholm (in Swedish).

Stille, H., Groth, T., Fredriksson, A. (1982) FEM-analysis of rock mechanical problems with
JOBFEM, Stiftelsen Bergteknisk Forskning – BeFo 307:1, Stockholm.

Stimpson, B., Ahmed, M. (1992) Failure of a linear Voussoir Arch: a laboratory and
numerical study. Canada Geotech. J., 29, pp. 188-194.

Zoback, M., et al. (1989) Global patterns of tectonic stress. Nature, vol 341, pp. 291-8.