Beruflich Dokumente
Kultur Dokumente
net/publication/328126754
CITATIONS READS
2 94
1 author:
Senthilnath G T
18 PUBLICATIONS 12 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Senthilnath G T on 06 October 2018.
Senthilnath G T
Endorsed by
December 2014
Face Stability of Closed TBMs in Urban
Tunnels
Senthilnath G T
December 2014
Acknowledgements
This work was carried out during the on-the job training / Internship period from
July 2014 to November 2014 under the guidance of Prof. Daniele Peila with the help of
my employer Dr. Oskar Sigl.
I am very grateful to my supervisors who gave me the freedom and the trust to select
the research area myself and let me develop the things, which I considered important
from engineering practice point of view. The discussions, comments and the way of
communication of Dr. Oskar Sigl has been very motivating and inspiring.
Last, but not least, I would like to express my deep gratitude to my parents and parents-
in-law for their unconditional love. Of all people, I am most grateful to my wife, Sujitha
for all her patience, encouragement and love.
ii
Theory and calculation are not substitute for judgement, but are the basis for
sounder judgement.
- Ralph B. Peck
iii
Abstract
In recent years, several large size shield tunnels have been constructed in urban
areas at shallow depth with the increasing practical demands such as accessibility,
serviceability and economy. One of the main objectives of such tunnels is to adequately
support the soil and to minimize deformations during and after construction, because in
such conditions, tunnel collapse or extensive deformations can be catastrophic, and even
limited soil deformations may damage buildings. To prevent this it is necessary to
continuously support the excavated face. The analysis of face stability of any pressurized
shield machines are performed to avoid both, collapse (active failure) and blow-out
(passive failure) of the soil mass near the tunnel face. A number of studies have been
dedicated to tunnel face stability [1] and recent developments (like "multi-block
mechanism", "method of slices" etc.) have updated the widely used methods of analysis.
Here, in this study, around 30 face stability models (analytical, empirical and semi-
empirical) are studied and a comparison of minimum face pressure values computed
using recently developed methods are made.
iv
Contents
Acknowledgements
Abstract
List of Figures
1 Introduction ........................................................................................... 1
v
3.1 Uncertainties ................................................................................. 61
Bibliography
Curriculum Vitae
Appendix A
vi
List of Figures
Figure 1. Failure mechanism observed in sand and clay [16].......................................... 5
Figure 2. Broms and Bennemark face stability model [17] ............................................. 6
Figure 3. Davis' face stability model (a) [11] .................................................................. 7
Figure 4. Davis' face stability model (b) [11] .................................................................. 7
Figure 5. Stability ratio design line proposed by Kumara & Mair [18] .......................... 8
Figure 6. Ellstein's face stability model [19] ................................................................... 9
Figure 7. Layout of plane strain heading problem [22] ................................................. 10
Figure 8. Failure mechanism proposed by [23].............................................................. 12
Figure 9. Stability number for 2D case [23] .................................................................. 13
Figure 10. Stability model of the method Atkinson and Potts [26] .............................. 14
Figure 11. Failure bulbs comparison with theoretical mechanism [27] ......................... 15
Figure 12. Improved 3D wedge prism model [28].......................................................... 16
Figure 13. Mechanical Equilibrium of the wedge [28] ................................................... 17
Figure 14. Distribution of the vertical stress σz along side slip surfaces [28] ................ 18
Figure 15. Design chart by Chen et al for cohesionless soils......................................... 19
Figure 16. Failure mechanism by Murayama [8]........................................................... 21
Figure 17. Statically admissible solution by Muehlhaus [12] ........................................ 22
Figure 18. Failure mechanics proposed by Krause [8]................................................... 22
Figure 19. Failure mechanics proposed by Mohkam [1] ................................................ 23
Figure 20. Conical blocks and kinematic conditions used in MI, MII and MIII [13] .... 24
Figure 21. 3D visualization of failure mechanism by Leca and Dormieux [37] ............. 24
Figure 22. Failure mechanics proposed by Leca and Dormieux [13] ............................. 25
Figure 23. Geometry of Mechanism MII [13] ................................................................ 26
Figure 24. Geometry of Mechanism MIII [13] ............................................................... 27
Figure 25. Stability model by Jancsecz & Steiner [33].................................................. 29
Figure 26. Seepage force f and effective support pressure s’ [2] .................................... 30
Figure 27. Stability model of Anagnostou & Kovari based on Horn [2] ....................... 30
vii
Figure 28. Nomograms for non-dimensional factors F0, F1 [2] ....................................... 31
Figure 29. Nomograms for non-dimensional factors F2, F3 [2] ...................................... 31
Figure 30. Stability model of Belter and Katzenbach [8] .............................................. 32
Figure 31. Multilayer wedge model by Broere [8] ......................................................... 32
Figure 32. Multilayer wedge model by Broere [8] ......................................................... 33
Figure 33. Extended Caquot's model [25] ..................................................................... 34
Figure 34. Excavation face as hemisphere [36].............................................................. 35
Figure 35. Multi-block failure mechanism [14] .............................................................. 36
Figure 36. Non-dimensional coefficients for face pressure estimation [14] .................... 37
Figure 37. Design chart for critical collapse pressure for C-Phi soil [14] ...................... 38
Figure 38. Cross section and horizontal section of a tunnel [9] ................................... 38
Figure 39. Force acting on infinitesimal slice [10] ......................................................... 39
Figure 40. Stability model used for analysis [10] .......................................................... 40
Figure 41. Design diagrams for effective support pressure, ∆h/H = 0-2 [10] .............. 43
Figure 42. Design diagrams for effective support pressure, ∆h/H = 3-5 [10] ................ 44
Figure 43. Comparison of face pressure for different cohesive soil model ..................... 45
Figure 44. Comparison of face pressure for different Cu values.................................... 46
Figure 45. Comparison of face pressure for Cohesionless soils ...................................... 47
Figure 46. Comparison of face pressure for different ϕ values ...................................... 48
Figure 47. Comparison of face pressure for c-ϕ soil ...................................................... 49
Figure 48. Comparison of face pressure for c-ϕ soil - influence of cohesion .................. 50
Figure 49. Comparison of effective face pressure as a function of ϕ’ ............................ 52
Figure 50. Comparison of effective face pressure as a function of c’ ............................. 53
Figure 51. Comparison of face pressure as a function of hydraulic gradient ................ 54
Figure 52. Comparison of Blowout pressure ................................................................. 55
Figure 53. Geological profile and tunnel alignment – Case study ................................ 56
Figure 54. Estimated TBM operating face pressure – Case study................................ 58
Figure 55. General stochastic problem [45] ................................................................... 61
Figure 56. COV of geotechnical parameters [45]........................................................... 61
Figure 57. Input values for MC simulation using Anagnostou, 2012 ............................ 64
viii
Figure 58. Snapshot of Random number generation ..................................................... 65
Figure 59. Histogram of MC simulation using Anagnostou, 2012................................. 66
Figure 60. CDF of MC simulation using Anagnostou, 2012 ......................................... 66
Figure 61. Summary of MC simulation using Anagnostou, 2012 in Excel .................... 67
Figure 62. Output of MC simulation using Chen et al 2014 model .............................. 68
Figure 63. Output of MC simulation using Mollon et al 2010 model ........................... 69
Figure 64. Distribution of input parameters for MC simulation ................................... 70
Figure 65. Comparison of PDF using the two methods ................................................ 70
ix
1 Introduction
1.1 Background
The problems posed by urban development and the challenges of traffic, utilities etc. are
often solved by digging tunnels. More and more tunnels are being excavated and there
are a host of new technologies to assist in excavation. Most urban tunnels are now shield
driven. This increases the importance of controlling the tunnel face pressure during its
advance. In the case of shallow tunnels, this pressure influences significantly the surface
settlements. This pressure must be kept above a certain threshold, or the face is unstable
and there is risk of tunnel collapse.
To prevent this it is necessary to support soft and non-cohesive soils from the time they
are excavated to the moment the final support is installed. Where groundwater is present
it is also necessary to prevent a flow towards the tunnel face, as this flow may have an
eroding effect on the tunnel face. In a tunnel boring machine (TBM) drive, the radial
support and water tightness is initially ensured by the shield and then by the tunnel
lining. At the face, a mixture of the excavated soil mass and varying additives (or
pressurized slurry in case of a slurry TBM) are used to provide necessary face pressure.
Though many metro projects have been completed, due to the complicated geological
conditions tunnel face collapse still takes place occasionally in recent tunnelling project
[3]. This calls for a better understanding of the problem and the uncertainties involved
in it.
Although a number of studies have been dedicated to tunnel face stability, only the
collapse failure mode of the ultimate limit stated is studied. The deterministic model is
either based on the upper bound or lower bound methods. The inherent uncertainty
(spatial and in determination of properties) are not considered in the deterministic
models.
1
1.2 Objective and Scope of the Study
A number of studies have been dedicated to tunnel face stability[1] and recent
developments (like “multi-block mechanism”, “method of slices” etc.) have updated the
widely used methods of analysis. Most analytical results are based on the limit
equilibrium method and the limit analysis method. This study is focused to discuss the
basis and use of different methods for different practical conditions from a design office
point of view.
Numerical methods were also frequently used to investigate stability of the tunnel face.
However, because of the level of attention required to select proper constitutive model
and parameters, numerical analyses could be time consuming. Moreover, the validity of
numerical analyses should also be checked, either by in situ measurements or by
laboratory model tests. Hence numerical analysis are generally difficult to be adopted
for routine design [4] and hence not discussed extensively in this report.
More importantly, the implementation of probabilistic analysis for computing
probability of face collapse using the analytical method is explored. This will allow to
incorporate uncertainty and variability of geotechnical parameters considered for the
analysis by performing probability analysis of failure. Similar techniques are already a
standard procedure for geotechnical problems like slope stability etc. Simplified model
for implementing probabilistic analysis with various face stability analytical models are
explored and presented in this study for quick implementation in design routines.
Chapter 2 deals with different tunnel stability models reported in the literature. Within
the chapter, the stability models are classified based on the soil-type, briefly illustrated
and compared. Developments and recent trends in the stability models are explained
and compared with each other. Chapter 3 deals with the Probabilistic estimate of face
support pressure. Methods to compute probability of failure and statistical reliability
quantification for the design are discussed in this chapter. Chapter 4 presents the
observations and summarizes the work presented in this report.
2
2 Tunnel Face Stability
One of the main objectives of the tunnel boring process is to adequately support the soil
and to minimize deformations during and after construction. The required face pressure
is maintained to ensure safe working conditions. It should not be as low as to allow
uncontrolled collapse of the soil into the working chamber, nor as high as to lead to large
deformations of the soil or to a blow-out or in some cases, change in pore water pressure.
The actual support pressure would mainly depends on encountered soil and groundwater
conditions, the excavation method and the size and overburden of the tunnel. To find
the minimal and maximal allowable support pressures, a number of models has been
proposed in literature over the years to describe different possible failure mechanisms of
the tunnel face and to calculate the properties of the support medium necessary to
prevent collapse. In this chapter around 30 of such stability models are discussed and
briefly illustrated. Section 2.3 presents a comparison of limit support pressure values
obtained using some of the discussed stability models.
For the estimation of support pressure, there are mainly two types of methods which are
commonly used to assess the proper support pressure:
• Limit Equilibrium Analysis
• Limit State Concept – Stress method based on plasticity
The limit equilibrium solutions including the Silo theory [5] are improved by other
researchers, for example,. Anagnostou and Kovari [6], [7] for homogenous soil and later
for heterogeneous soils by Broere [8], Anagnostou [9], [10] etc.
Using the limit state concept, Davis et al. [11] investigated the support pressure for
tunnelling in cohesive soil by a 2D plane analysis. Afterwards, Muelhaus [12] found a
lower bound solution for the tunnel stability in frictional soil characterized by the Mohr–
Coulomb criterion. Leca and Dormieux [13] improved the former lower bound solution
and carried out an upper bound analysis for the tunnel face stability in low- cohesive
3
soil in 3D. It is further developed recently to multi-block failure mechanism by Mollon
[14]. Tang et al [3] extended the limit state concept for layered soils.
It is evident from above summary that the oldest stability model dates back to 1961 and
many recent models are being developed to address the heterogeneity in the soil and
actual difficult geological conditions. Following sections discusses the stability models,
classified based material model and results are compared.
When discussing previous research on tunnel heading stability, one has to distinguish
between drained and undrained conditions. For undrained conditions, as dominant in
clays, practical design curves have been derived on the basis of model tests [15] and these
curves have been largely confirmed by theoretical studies [11]. The question whether a
drained or undrained stability analysis should be carried out can be answered by
considering the type of ground and the advance rate of the tunnel face. According to a
parametric study by Anagnostou and Kovári [2], drained conditions tend to apply when
the ground permeability is higher than 10-7 to 10-6 m/s and the net excavation advance
rate is 0.1 to 1.0 m/hr or less. In a predominately sandy soil, therefore, drained stability
conditions should be considered. In a clayey, low-permeability soil the undrained analysis
is valid during excavation, but the drained analysis applies in case of a standstill. Hence,
even for excavations in clay it is important to investigate drained soil conditions.
It is also observed that from the laboratory tests that the geometry of the failure
mechanism in sand and in clay is notably different. Tests in sand show a chimney-like
failure mechanism, described by the wedge stability models. Centrifuge tests on clay
show a much larger and widening zone influenced by the instability Figure 1.
Hence, this chapter briefly presents the stability models classified based on the soil model
for which they are applicable. List of all stability models studied in this report is enclosed
in Appendix A. This classification can be useful to select appropriate model for a given
design consideration.
4
Figure 1. Failure mechanism observed in sand and clay [16]
This section discusses the stability models available in literature specifically for
undrained cohesive soft soils. Summary of the stability models discussed in this section
are presented in table below.
Table 1. Summary of stability models for cohesive soil
Name of the Model Year Soil Type Analysis Method Reference
Global Equilibrium /
Ellstein 1986 Homogeneous [19]
Limit Equilibrium
5
Stress Methods are analytical methods, which are based on the construction of statically,
and kinematically admissible solutions to take advantage of what are known the lower
and upper bound theorems of plasticity.
Broms and Bennermark [17] was one of the first stability models for cohesive soils. This
method provides a relation for the stability analysis of an unsupported opening in
cohesive undrained material - Tresca criteria,[24]. With this relationship they define the
stability ratio N to be equal to the difference between total overburden stress and
support pressure divided by the undrained shear strength
where,
γ is soil density, cu is undrained cohesion, C is overburden, R is tunnel radius, qs is surface
load. From observations of collapses in both building pits and tunnel constructions, and
from laboratory extrusion tests, they find that an opening in such conditions will become
unstable N ≥ 6.
6
D avis et al (1980)
In 1980 Davis et al. [11] investigate the stability of an idealised partially unlined tunnel
heading in a Tresca material and introduce the distance P between the face and the
point where a stiff support is provided. They use the vertical opening as presented by
Broms & Bennermark as one of three limit cases for their stability analysis. They were
perhaps the first to propose that a statically admissible solution for a spherical opening,
could be used to analyse the stability of the front of a lined tunnel [25]. They further
derive upper and lower bound plasticity solutions for the stability of a plane strain
unlined cavity and a plain strain ‘long wall mining’ problem, obtained by taking P =
Infinity.
7
Of particular relevance to the face stability of a tunnel excavated using a shield machine
is the case P = 0. Davis et al. derive two lower bound solutions, using a cylindrical and
a spherical stress field.
𝐶𝐶
𝑁𝑁 = 2 + 2 ln � + 1� [𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐]
𝑅𝑅
𝐶𝐶
𝑁𝑁 = 4 . ln � + 1� [𝑠𝑠𝑠𝑠ℎ𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒]
𝑅𝑅
In 1981, Kimura and Mair [18] conducted a number of Cambridge centrifuge and
obtained a stability ratio design line for tunnels in undrained conditions. When plotted,
the results form neat hyperbolic patterns which are easily analysed.
Figure 5. Stability ratio design line proposed by Kumara & Mair [18]
Ellstein (1986)
Ellstein [19] gave an analytical expression of N for homogeneous cohesive soils based on
a limit equilibrium analytical approach. His results are in good agreement with those by
Kimura and Mair [18].
8
Figure 6. Ellstein's face stability model [19]
4 + √2
𝑁𝑁𝑐𝑐 = 𝐷𝐷
𝐾𝐾° + 1.5𝐻𝐻
For the tunnels excavated using fluid pressure against heading, Nc is modified as:
4 + √2
𝑁𝑁𝑐𝑐 = 𝐷𝐷 𝑝𝑝
𝐾𝐾° + 1.5𝐻𝐻 − 𝛾𝛾𝛾𝛾
Where p being the applied fluid pressure, H being the overburden above crown, D is
outer diameter of the tunnel, h is H+D/2 and K0 is coefficient of earth pressure at rest.
Mori [20] and Bezuijen [21] have come up with an expression for fracturing model to
estimate blow-out pressure.
Mori [20] defines the pressure at which (vertical) fracturing must occur for a normally
consolidated soil as
𝑠𝑠𝑓𝑓 = 𝐾𝐾0 𝜎𝜎𝜗𝜗′ + 𝑝𝑝 + 𝑞𝑞𝑢𝑢
with qu the unconfined compressive strength. In general fracturing will occur somewhat
earlier, resulting in an unsafe estimate of the maximal support pressure. For highly over
9
consolidated soils with K0 ≥ 1, the horizon
tal effective stress may be greater than the
vertical, as a result of which horizontal fracturing will occur at a stress significantly
lower than determined above.
which defaults to Mori's expressed for ηf = 2 and normally consolidated soil. Bezuijen
holds that when taking ηf = 1, this expression will yield safe estimates of the maximum
allowable support pressure.
A ugarde et al (2003)
Augrade [22] presented the stability of an idealised heading in undrained soil conditions.
They presented load parameters, for a wide range of heading configurations and ground
conditions using Finite element limit analysis methods, based on classical plasticity
theory. The approach presented by Augarde et al uses load parameter (σs - σT) / cu0
instead of N (stability ratio or overload factor) and also accounted for cu varying with
the depth. The results of the FE bound analysis are presented as dimensionless stability
charts and tables for easy implementation in design routine. The load parameter table
is included here for quick reference.
10
Table 2. Load parameter table based on [22]
K lar et al (2007)
Klar et al [23] suggest an upper bound solution for cohesive soils. They adopted elasticity
theory to estimate an admissible velocity field over which differential equation of
continuity are integrated. The velocity field for the upper bound solution is assumed to
be proportional to the displacement field.
11
Figure 8. Failure mechanism proposed by [23]
where C is the tunnel cover depth and D is the tunnel diameter. If an upper bound value
is evaluated for σs - σT, then
̇
∫ 𝑉𝑉2�∈|𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚 + ∫ 𝑆𝑆3 𝑣𝑣𝑡𝑡 𝑑𝑑𝑑𝑑 − (𝛽𝛽/𝐷𝐷) ∫ 𝑉𝑉 𝑣𝑣𝑣𝑣 𝑑𝑑𝑑𝑑 𝐶𝐶 + 𝐷𝐷/2
𝑁𝑁 = + 𝛽𝛽
𝑉𝑉 . 𝐿𝐿 𝐷𝐷
where β= γ D / Cu and the variation with respect to C/D ratio is presented in figure
below.
12
Figure 9. Stability number for 2D case [23]
Summary of the stability models discussed in this section are presented in Table 3.
Although models available for Mohr-coulomb material could be easily used for pure
cohesionless soil, this classification is made to capture the development of stability
models specifically for cohesionless soils.
In 1977, Atkinson & Potts [26] derived the minimal support pressure for an unlined
cavity in a dry cohesionless material. They differentiate between two limit cases. The
13
first case is a tunnel in a weightless medium with a surface load, the second a tunnel in
a medium with γ>0 but without surface loading. For the second case, two lower limit
solutions are furnished. The solution, which is independent of the overburden, provides,
in general, the result associated with the greater safety:
Where,
𝑘𝑘𝑝𝑝 = (1 + sin 𝜑𝜑)/(1 − sin 𝜑𝜑)
As this represents a statically admissible stress field, it is a safe estimate for the minimal
support pressure. The kinematic upper bound solution on the other hand, which is the
inherently unsafe estimate, yields a lower value for the minimal support pressure. It
should be noted that both upper and lower bound solutions found by Atkinson & Potts
are independent of the relative overburden C/D.
Figure 10. Stability model of the method Atkinson and Potts [26]
Chambon & Corté [27] used a model tunnel sealed by a membrane and embedded in a
homogeneous sand layer. They gradually reduced the support pressure and found wedge
shaped failure planes loaded by soil silos. Their investigation focussed on the influence
of the overburden on the silo formation and shows that a soil silo will form even if only
a small overburden is present but that its influence will be limited by the proximity of
14
the soil surface. They further show that a soil silo can develop fully at an overburden
equal to the tunnel diameter or larger. Chambon and Corte also found that the cover-
to-diameter ratio C/D (i.e., relative depth; C is the cover depth and D is the diameter
of the tunnel) has significant effect on the failure mechanism. That means, when relative
depth C/D is low (e.g., C/D = 0.5), the failure zone in the limit state has extended to
the ground surface. When C/D is high (e.g., C/D = 1 or 2), the failure zone in the limit
state will be still in the interior of the ground.
The test results have provided better understanding of the behaviour of tunnel face in
pure cohesion less soils and have suggested that the process leading to failure of face
could be (p is pressure applied, pc is the pressure at which the first movements of the
face could occur and pf is the pressure at which failure would occur:
For p> pc, no movement of face would occur
For pc > p > pf, small displacements of the face and surface settlements would occur
For p = pf, there would be a sudden but localized collapse
For p < pf, there would be a flow of soil in the tunnel
There is no time effect in any of the stages; at constant internal pressures, the tunnel
face is stabilized. In a dry sand, in all cases, including in the presence of a load on the
15
surface, a hydrostatic pressure is sufficient to provide stability to the face. However, the
face is not self-stabilizing and must be supported. Small uniform pressures, of the order
of about 10 kPa, are sufficient. The pressure at failure is little affected by changes of
geometry at the relative depths considered (C/D = 0.5, 1, 2, 4) and by the density of
the soil.
Chen et al (2014)
Chen et al. [28] proposed a 3D wedge prism model which is based on limit equilibrium
method to determine the limiting the support pressure at the tunnel face. This model
considers the influence of relative depth on the failure mechanism as observed by
Chambon and Corte [27]. The other aspect of the improvement is properly considering
soil arching effects (Broere, [8]) to calculate the vertical force applied on the wedge’s
upper surface more accurately, as it has significant influence on the limit support
pressure.
16
Figure 13. Mechanical Equilibrium of the wedge [28]
4𝑆𝑆
𝑠𝑠𝑙𝑙𝑙𝑙𝑙𝑙 = 𝑚𝑚𝑚𝑚𝑚𝑚 � � = 𝑚𝑚𝑚𝑚𝑚𝑚{𝑓𝑓1 𝜎𝜎𝑎𝑎𝑎𝑎 + 𝑓𝑓2 𝛾𝛾𝛾𝛾}
𝜋𝜋𝐷𝐷2
Where,
tan(𝛽𝛽 − 𝜙𝜙)
𝑓𝑓1 = − 𝜆𝜆ℰ tan 𝜙𝜙
tan 𝛽𝛽
tan(𝛽𝛽 − 𝜙𝜙) 1
𝑓𝑓2 = − 𝜆𝜆ℰ tan 𝜙𝜙
2 tan 𝛽𝛽 3
4
𝜀𝜀 = [cot 𝛽𝛽 + tan(𝛽𝛽 − 𝜙𝜙)] cos 𝛽𝛽
𝜋𝜋
𝜋𝜋 𝜋𝜋
𝛽𝛽𝛽𝛽 � , �
4 2
Where, λ is the ratio of horizontal to vertical stress in the wedge given by:
𝜆𝜆 = 1 − sin Φ
𝛾𝛾𝛾𝛾 𝑈𝑈
𝑠𝑠 tan 𝜙𝜙𝜙𝜙
−𝐾𝐾 𝐴𝐴
𝜎𝜎𝑎𝑎𝑎𝑎 = �1 − 𝑒𝑒 �
𝑈𝑈𝐾𝐾 𝑠𝑠 tan 𝜙𝜙
17
with,
𝜋𝜋𝜋𝜋 𝐷𝐷
𝐴𝐴 = 𝐵𝐵 × 𝐿𝐿 = ×
4 tan 𝛽𝛽
𝜋𝜋𝜋𝜋 𝐷𝐷
𝑈𝑈 = 2 × (𝐵𝐵 + 𝐿𝐿) = 2 × � + �
4 tan 𝛽𝛽
2𝐷𝐷
𝐻𝐻 = 𝑚𝑚𝑚𝑚𝑚𝑚 �𝐶𝐶, �
tan 𝛽𝛽
2𝐷𝐷
𝑞𝑞𝑜𝑜 = 𝛾𝛾 × �𝐶𝐶 − 𝑚𝑚𝑚𝑚𝑚𝑚 �𝐶𝐶, ��
tan 𝛽𝛽
Figure 14. Distribution of the vertical stress σz along side slip surfaces [28]
where
𝜋𝜋 𝜙𝜙 𝜋𝜋 𝜙𝜙
𝜃𝜃𝑜𝑜 = + , 𝐾𝐾𝑎𝑎 = 𝑡𝑡𝑡𝑡𝑡𝑡2 � − �
4 2 4 2
This model has sufficient accuracy as the limit support pressure obtained via this model
agrees well with that obtained via Vermeer et al.’s method [29] which was proposed
based on a series of rigorous FEM calculations. In addition, the improved wedge-prism
model is also a relatively safe estimation of the limit support pressure for the practical
application.
18
0.25
C/D = 1
C/D = 0.5
0.2 C/D = 2
C/D = 3
C/D = 4
0.15
Slim/γ.D
0.1
0.05
0
25 30 35 40 45
Phi [degrees]
Figure 15. Design chart by Chen et al for cohesionless soils
Chen et al also presented a design chart (Figure 15) for easy implementation in design
office.
19
2.1.3 Models for C-Phi Soil (Mohr Coulomb)
M urayam a (1966)
Murayama (as cited in [30]) calculated the minimal support pressure using a two-
dimensional log-spiral shaped sliding plane in 1966. The soil weight acting on the
pressure wedge is calculated in accordance with the Terzaghi (1943) theory. The face
stability requires the equilibrium between the moment of the acting weight forces and
the resistant forces i.e, the force applied on the tunnel face (P) and shear strength along
the failure surface. The method contemplates the iterative search for the solid-load width
20
that determines the more unfavourable loading condition and, therefore, the maximum
stabilisation pressure, P.
1 𝜔𝜔1 2
𝑟𝑟𝑎𝑎2
𝑠𝑠𝑚𝑚𝑚𝑚𝑚𝑚 = (𝐺𝐺𝐼𝐼 )
𝐺𝐺 + 𝑞𝑞𝑤𝑤 𝜔𝜔1 �𝑙𝑙𝜔𝜔 + � − 𝑐𝑐 �𝑟𝑟𝑑𝑑 − �
2𝑅𝑅𝑙𝑙𝑝𝑝 2 2 tan 𝜑𝜑
M ühlhaus (1985)
Mühlhaus [12] was first to propose a complete statically admissible solution for tunnel
stability. The solution could be used to analyse the stability of the unsupported span in
the vicinity of the tunnel front. Since in urban TBM drive, unsupported heading is
almost zero, this model is not relevant for face pressure calculation and hence not
discussed further.
21
Figure 17. Statically admissible solution by Muehlhaus [12]
K rause (1987)
Krause [31] proposed minimal support pressures needed for a semi-circular and spherical
limit equilibrium mechanism using the shear stresses on the sliding planes. Of the three
mechanisms proposed, the quarter circle (Figure 18 b) will always yield the highest
minimal support pressure. However, this may not always be a realistic representation of
the actual failure body. In many cases the half-spherical body (Figure 18c) will be a
better representation.
22
For Half circle,
1 1 1
𝑠𝑠𝑚𝑚𝑚𝑚𝑚𝑚 = 1 � 𝐷𝐷𝛾𝛾 ′ − 𝜋𝜋𝜋𝜋�
2
+ 𝑡𝑡𝑡𝑡𝑡𝑡 𝜑𝜑 6 2
M ohkam (1989)
Mohkam [32] described a limit equilibrium model using a roughly log-spiral shaped wedge
which has to be obtained from a variational analysis over the unknown position angle
and total stress of the failure plane. The model also includes the effect of the reduced
effectiveness of the slurry pressure due to infiltration using a non-linear relation between
infiltration distance and effective slurry pressure.
Taking into account the support-free length before the installation of a stiff support, two
failure mechanisms are assumed: one involves the face excavation (Figure 19 a & b) and
the other involves the tunnel wall (Figure 19c), along the failure surface, respectively,
logarithmic spiral and cylindrical. The load acting on the wedge is based on Terzaghi’s
arch effect. The large number of unknowns present in the model leads to a highly
iterative solution procedure and a rather unwieldy model and hence not discussed further
in this report.
23
Leca and D orm ieux (1990)
Leca and Dormieux [13] propose a series of conical bodies in 1990 (Figure 22). Combined
with different stress states, similar to those proposed by Davis et al., they derive lower
and upper bound limits for both the minimal and maximal support pressure of a limed
tunnel in a dry Mohr-Coulomb material. Three failure mechanisms which involve the
movement of solid conical blocks with circular cross sections, are shown in Figure 21.
MI and MII failure mechanisms are single-cone and dual-cone systems, respectively;
where the cones move into the excavation. An MIII failure mechanism is a single-cone,
passive mechanism, where the cone moves outward to the surface.
Figure 20. Conical blocks and kinematic conditions used in MI, MII and MIII [13]
24
Figure 22. Failure mechanics proposed by Leca and Dormieux [13]
Expression for support pressure in MII collapse failure mechanism, which is most
common failure pattern in shield tunnelling [27], is presented below:
1 sin(𝛽𝛽 − 𝜙𝜙 ′ )𝑅𝑅𝐸𝐸2
𝑁𝑁𝑠𝑠 =
cos 𝛼𝛼 𝑐𝑐𝑐𝑐𝑐𝑐 2 𝜙𝜙 ′ sin(𝛽𝛽 + 𝜙𝜙 ′ )𝑅𝑅𝐴𝐴
25
1 𝑐𝑐𝑐𝑐𝑐𝑐𝜙𝜙 ′ cos(𝛽𝛽 + 𝜙𝜙 ′ ) 𝑅𝑅𝑐𝑐3 1 sin(𝛽𝛽 − 𝜙𝜙 ′ ) 𝑅𝑅𝐸𝐸3
𝑁𝑁𝛾𝛾 = [𝑡𝑡𝑡𝑡𝑡𝑡𝑅𝑅𝐵𝐵 + − ×
3 2𝑠𝑠𝑠𝑠𝑠𝑠𝜙𝜙 ′ cos(𝛽𝛽 + 𝜙𝜙 ′ ) 𝑅𝑅𝐴𝐴 2𝑠𝑠𝑠𝑠𝑠𝑠𝜙𝜙 ′ cos 𝛼𝛼𝑐𝑐𝑐𝑐𝑐𝑐 2 𝜙𝜙 ′ sin(𝛽𝛽 + 𝜙𝜙 ′ ) 𝑅𝑅𝐴𝐴
�[cos(𝛼𝛼 − 𝜙𝜙 ′ ) cos(𝛼𝛼 + 𝜙𝜙 ′ )]
𝑅𝑅𝐴𝐴 =
cos 𝜙𝜙 ′
[cos(𝛼𝛼 − 𝜙𝜙 ′ ) cos(𝛼𝛼 + 𝜙𝜙 ′ )]
𝑅𝑅𝐵𝐵 =
sin(2𝜙𝜙 ′ )
1⁄2
cos(𝛼𝛼 + 𝜙𝜙 ′ ) sin(β − ϕ′ )
𝑅𝑅𝐶𝐶 = � �
cos 𝜙𝜙 sin(𝛽𝛽 + 𝜙𝜙 ′ )
sin 𝛽𝛽
𝑅𝑅𝐷𝐷 =
sin 𝜙𝜙 ′ sin(𝛽𝛽 + 𝜙𝜙 ′ )
𝑐𝑐𝑐𝑐𝑐𝑐 2 𝜙𝜙 ′ 2𝐶𝐶
𝑅𝑅𝐸𝐸 = 𝑅𝑅𝑐𝑐 − sin 𝜙𝜙 ′
cos(𝛼𝛼 + 𝜙𝜙 ′ ) 𝐷𝐷
26
Above relation provides the best upper bound association with MII when α chosen such
that Ns , and Nγ are at minimum. The above results only apply when the ground surface
is reached by the failure mechanism. that is to say when:
𝐶𝐶 cos(𝛼𝛼 + 𝜙𝜙 ′ ) sin(𝛽𝛽 − 𝜙𝜙 ′ )
≤
𝐷𝐷 2 sin 𝜙𝜙 ′ sin(𝛽𝛽 + 𝜙𝜙 ′ )
Recently, this method was updated by Lee et al. [38]to account for seepage forces acting
on the tunnel face under steady-state flow conditions. However, general solutions which
combine the depth-dependence of the effective cohesion of NC clays and the influence of
seepage have not been reported.
Expression for support pressure in MIII blow-out failure mechanism is presented below:
27
𝑅𝑅𝐴𝐴 = cos 𝛼𝛼 �[cos(𝛼𝛼 + 𝜙𝜙 ′ ) cos(𝛼𝛼 − 𝜙𝜙 ′ )]
𝑅𝑅𝐵𝐵 𝑅𝑅𝑐𝑐2
𝑁𝑁𝑠𝑠 =
𝑅𝑅𝐴𝐴
sin 𝛼𝛼 𝑅𝑅𝐵𝐵 𝑅𝑅𝐶𝐶 3 𝑅𝑅𝐴𝐴 3
𝑁𝑁𝛾𝛾 = �� � −� � �
3𝑅𝑅𝐴𝐴 sin 2𝜙𝜙1 sin 𝛼𝛼 cos 𝛼𝛼
The best upper bound associated with MIII is found by choosing a such that
Ns , and Nγ, are at minimum. Since failure always reaches the ground surface, above
expression is valid for all values of C/D.
Jancsecz & Steiner [33] proposed a soil wedge model similar to Horn [5], Method. The
three-dimensional failure scheme shown in Figure 25 consists of a soil wedge (lower part)
and a soil silo (upper part). The vertical pressure resulting from the silo and acting on
the soil wedge is calculated according to Terzaghi’s solution.
28
Figure 25. Stability model by Jancsecz & Steiner [33]
𝑘𝑘𝑎𝑎3 = (sin 𝛽𝛽 . cos. −𝑐𝑐𝑐𝑐𝑐𝑐 2 𝛽𝛽. tan 𝜙𝜙 − 𝐾𝐾. 𝛼𝛼. cos 𝛽𝛽. tan 𝜙𝜙⁄1.5)⁄(sin 𝛽𝛽. cos 𝛽𝛽 + 𝑠𝑠𝑠𝑠𝑠𝑠2 𝛽𝛽. tan 𝜙𝜙)
where,
Κ ≈ [1 − sin ϕ + tan2 (45 + ϕ⁄2)]⁄2 ; α = (1 + 3 . t⁄D)⁄(1 + 2 . t⁄D).
Anagnostou & Kovári [2] method, is based on the silo theory [33] and to the three-
dimensional model of sliding mechanism proposed by Horn [5]. The analysis is performed
in drained condition, and a difference between the stabilizing water pressure and effective
pressure in the plenum of a EPBs is presented. If there is a difference between the water
pressure in the plenum and that in the ground, destabilizing seepage forces occur and a
higher effective pressure is required at the face.
29
Figure 26. Seepage force f and effective support pressure s’ [2]
However, accepting this flow, the total stabilizing pressure is lower than the pressure
required in the case of an imposed hydro-geological balance. The effective stabilizing
pressure is:
where F0, F1, F2, F3are non-dimensional factors derived from nomograms, which are
function of H/D and friction angle.
Figure 27. Stability model of Anagnostou & Kovari based on Horn [2]
30
Figure 28. Nomograms for non-dimensional factors F0, F1 [2]
Belter and Katzenbach introduced for the first time, a two-dimensional wedge model
that can include layered soils above the TBM, but not in front of the TBM [30]. It
remains unclear from the brief description, however, whether this model includes arching
effects of the soil above the tunnel or to what extent these heterogeneities influence the
required support pressure. The model sketched by Belter and Katzenbach (shown in
Figure 30) could also be obtained by including layered soil above the face in the wedge
31
models described by Jancsecz & Steiner or Anagnostou & Kovári and hence not discussed
in detail here.
Broere [8] pointed out some important shortcomings of the existing edge stability models
and developed a model to address the following aspects:
32
Figure 32. Multilayer wedge model by Broere [8]
For the layer “i” with top z = t (1) the following formulation is proposed for a stratified
soil, in the range t (i) < z < t (i+1):
′(𝑖𝑖) 𝑎𝑎𝛾𝛾 ′(𝑖𝑖) − 𝑐𝑐 ′(𝑖𝑖) ′(𝑖𝑖)
−𝐾𝐾(𝑖𝑖) 𝑡𝑡𝑡𝑡𝑡𝑡𝑓𝑓 𝑎𝑎
𝑍𝑍
𝜎𝜎𝜈𝜈,𝑎𝑎 = (𝑖𝑖) �1 − 𝑒𝑒 �
𝐾𝐾 tan 𝜙𝜙 ′(𝑖𝑖)
It is found that for the simplified case of a single slide wedge in homogeneous soil, the
resultant formulation corresponds to that of Jancsecz [33].
Vermeer and Ruse [34] conducted numerical analysis using FE-code PLAXIS and
deduced the following approximation for the limit support pressure for the case ϕ> 20°
𝑐𝑐 1
𝑝𝑝 ≈ − + 2𝛾𝛾𝛾𝛾 � − 0.05�
tan 𝜑𝜑 9 tan 𝜑𝜑
The underlying results are obtained with an elastic-ideal plastic constitutive law
assuming Mohr-Coulomb yield surface and associated plasticity. The consequences of
friction dependent arching are considerable. The stress arch carries the ground cover
independent of the magnitude of its thickness. Authors point that once the friction angle
is larger than about twenty degrees, stability is completely independent of the ground
cover. Compared to the soil weight stability number Nγ and the cohesion stability
number Nc, Nq is very low and it is disregarded by the authors.
33
For the case ϕ= 0, instead, they obtain a linear relation between p and h. For this
relation no analytic expression is given.
Caquot’s model considers the equilibrium condition for material undergoing failure above
the crown of a shallow circular (cylindrical or spherical) cavity. The material has a unit
weight γand a shear strength defined by Mohr-Coulomb parameters c (cohesion) and ϕ
(friction angle). A support pressure ps can be applied inside the tunnel, while a surcharge
qs acts on the ground surface. The Caquot generalised solution for dry conditions, can
be represented by the following equation developed by Carranza-Torres [25]
Where,
1 + sin 𝜙𝜙 𝜋𝜋 𝜙𝜙
𝑁𝑁𝜙𝜙 = = 𝑡𝑡𝑡𝑡𝑡𝑡2 � + �
1 − sin 𝜙𝜙 4 2
34
K olym bas (2005)
Kolymbas [35], [36] considered the distribution of the vertical stress between the ground
surface and the crown of a spherical cavity and approximated this distribution by a
quadratic parabola. Material strength is assumed to be fully mobilised at the crown.
Support pressure is given by the expression:
𝑐𝑐 cos 𝜑𝜑
𝛾𝛾 − 𝑟𝑟 1−sin 𝜑𝜑
𝑓𝑓
𝑝𝑝𝑐𝑐 = ℎ ℎ 2 sin 𝜑𝜑
1 + 𝑟𝑟 1−sin 𝜑𝜑
𝑓𝑓
35
δ is effective wall friction angle of the sliding surface
D is the diameter of the excavation area
K is the earth pressure coefficient
σv(t) is vertical stress at depth
M ollon et al (2010)
Mollon et al [14] improved the existing limit analysis mechanisms by taking into account
the entire circular tunnel face and not only the inscribed ellipse. This was made using
spatial discretization technique and hence it is possible to generate multi-block
mechanism and not restricted to standard geometric shapes such as cones or cylinders.
Mollon et al [14] also presented design charts for easy implementation in design routines.
The critical values of Nγ, Nc and Ns presented in Figure 35 are using the superposition
method. It was shown that the error induced by the superposition principle is quite small
and is always conservative.
This method estimates the three dimensional surface by defining the contours at several
different vertical places parallel to the tunnel face. The failure mechanism respects the
36
normality condition required by limit analysis since the three dimensional failure surface
generated in such a manner that the velocity vector makes an angle ϕ with the velocity
discontinuity surfaces anywhere along these surfaces. Another advantage of this method
being, it can be applied to even tunnels of other cross sections.
The proposed failure mechanism always outcrops in the case of purely cohesive soil. It
means that in this case the parameter C/D is of major importance. This is not the case
for c-phi soil which high to moderate friction angle (40 to 20 degree) since the critical
tunnel pressure is independent of the tunnel cover in these cases. Numerical results have
shown that multi-block mechanism composed of three blocks is a good compromise
between computation time and results accuracy.
37
0.4
Phi = 15
0.35 Phi = 20
Phi = 25
0.3
Phi = 30
0.25 Phi = 35
σc/γD
Phi = 40
0.2
0.15
0.1
0.05
0
0 0.02 0.04 0.06 0.08 0.1
c/γD
Figure 37. Design chart for critical collapse pressure for C-Phi soil [14]
A nagnostou (2012)
Anagnostou [9] revisits their computational model (Anagnostou and Kovári [7]) and an
alternative model is presented which is based on the method of slices. This method does
not need an a priori assumption as to the distribution of horizontal stress. This method
could accounts for horizontal arching around the tunnel face as the shear stresses τs
developing at the two vertical slip surfaces (as shown in Figure 38) contribute to the
stability of the wedge.
The method of slices is adopted form the stability assessment of slurry walls. In analogy
to the silo theory, the method of slices assumes proportionality between the horizontal
stress and vertical stress
38
𝜎𝜎𝑦𝑦′ = 𝜆𝜆𝜎𝜎𝑧𝑧′
where the coefficient of lateral stress λis assumed to be constant. In order to calculate
the distribution of the vertical stresses inside the wedge, the equilibrium of an
infinitesimally thin slice is considered (Figure 39). This makes it possible to analyse cases
with non-uniform face support, heterogeneous ground consisting of horizontal layers or
non-uniform distribution of the seepage forces along the height of the face.
The method of slices also makes it possible to estimate on a more consistent basis
(similarly to silo theory) the vertical stresses within the wedge. The method of slices
leads to support pressures which are much closer to the numerical predictions of Vermeer
et al. [39]. These results indicate that the reason for the differences from the numerical
results is the simplified way of considering horizontal arching in the model of Anagnostou
and Kovári [6], [7]. The author also presented a simplified formula for estimation of
support pressure when λ = 1.0 (Terzaghi’s initial assumptions)
In this model, Perazzeli et al [10], [40] combines the model for face stability under drained
condition by considering wedge and prism mechanism [6] with the method of method of
slices presented above [9]. They considered limit equilibrium model and the mechanism
fails if the load exerted by the prism upon the wedge exceeds the force which can be
39
sustained by the wedge at its upper boundary. At the limit equilibrium the prism load
is equal to the bearing capacity of the wedge. The prism load is calculated based on the
silo theory, while the bearing capacity of the wedge is calculated by considering the
equilibrium of the infinitesimal slice (as in [8]).
Where, s1' is the minimum support pressure required to ensure that tensile effective
stress does not exceed c'/tanϕ' (tensile strength exhibited by weak rocks) at any point
in the sliding surface and is given by:
where,
1 − 𝑒𝑒 −𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 Λ𝐶𝐶Δℎ (1)
𝑃𝑃1 = +
Λ𝐶𝐶𝑠𝑠 (1)
�1 + (𝑃𝑃 +cot 𝜔𝜔)� 𝑃𝑃𝑠𝑠 + cot 𝜔𝜔 + Λ𝐶𝐶𝑠𝑠 (1)
𝑠𝑠
Λ𝐶𝐶𝛾𝛾 (1)𝐵𝐵
𝑃𝑃2 =
(𝑃𝑃𝑠𝑠 + cot 𝜔𝜔 + Λ𝐶𝐶𝑠𝑠 (1))𝐻𝐻
40
𝑀𝑀𝑐𝑐 𝐻𝐻
𝐵𝐵
+ 𝑃𝑃𝑐𝑐 + Λ𝐶𝐶𝑐𝑐 (1) 𝑃𝑃𝑠𝑠 + cot 𝜔𝜔
𝑃𝑃3 = +
𝑃𝑃𝑠𝑠 + cot 𝜔𝜔 + Λ𝐶𝐶𝑠𝑠 (1) (𝑃𝑃𝑠𝑠 sin 𝜔𝜔 + cos 𝜔𝜔)(𝑃𝑃𝑠𝑠 + cot 𝜔𝜔 + Λ𝐶𝐶𝑠𝑠 (1)) tan 𝜙𝜙
s2' is the minimum support pressure required for soft soil without any tensile strength
and hence effective normal stress σn' must be higher than 0.
𝑠𝑠2′ = 𝑃𝑃1 𝛾𝛾𝑤𝑤 Δℎ + 𝑃𝑃2 𝛾𝛾 ′ 𝐻𝐻 − 𝑃𝑃4 𝑐𝑐 ′
where,
𝑀𝑀𝑐𝑐 𝐻𝐻
𝐵𝐵
+ 𝑃𝑃𝑐𝑐 + ΛCc (1)
𝑃𝑃4 =
𝑃𝑃𝑠𝑠 + cot 𝜔𝜔 + Λ𝐶𝐶𝑠𝑠 (1)
s3' is the minimum support pressure required so that the bearing capacity of the wedge
is higher than the vertical load exerted by the prism and is given by:
𝑡𝑡∗
−𝜆𝜆 tan 𝜙𝜙′
1 𝐶𝐶𝛾𝛾 (1)𝐵𝐵 tan 𝜔𝜔𝜔𝜔(1 − 𝑒𝑒 𝑅𝑅 )
𝐹𝐹2 = � + �
𝐶𝐶𝑠𝑠 (1) 𝐻𝐻 𝐵𝐵𝐵𝐵 tan 𝜙𝜙 ′
∗
′ 𝑡𝑡
1 tan 𝜔𝜔𝜔𝜔(1 − 𝑒𝑒 −𝜆𝜆 tan 𝜙𝜙 𝑅𝑅 )
𝐹𝐹3 = (𝐶𝐶 (1) +
𝐶𝐶𝑠𝑠 (1) 𝑐𝑐 𝐵𝐵𝐵𝐵 tan 𝜙𝜙 ′
2λ tan ϕ′
Λ=
cos ω − sin ω tan ϕ′
𝑀𝑀 = 𝑀𝑀𝑐𝑐 𝐵𝐵 2 𝑐𝑐 ′ − 𝑀𝑀𝛾𝛾 𝐵𝐵 3 𝛾𝛾 ′
𝑃𝑃 = 𝑃𝑃𝑐𝑐 𝐵𝐵 2 𝑐𝑐 ′ + 𝑃𝑃𝑠𝑠 𝐵𝐵 2 𝑠𝑠 ′
41
𝑃𝑃𝑠𝑠 = tan(𝜙𝜙 ′ + 𝜔𝜔)
where,
𝑀𝑀𝛾𝛾 = tan 𝜔𝜔
Λ tan 𝜔𝜔
𝑀𝑀𝑐𝑐 =
𝜆𝜆 tan 𝜙𝜙 ′
Λ
𝑃𝑃𝑐𝑐 =
2𝜆𝜆 tan 𝜙𝜙 ′ cos 𝜔𝜔
𝐶𝐶𝜐𝜐 (1) − 1
𝐶𝐶𝑠𝑠 (1) = 𝑃𝑃𝑠𝑠
Λ
𝐹𝐹(1)
𝐶𝐶𝛾𝛾 (1) = 2 𝑀𝑀𝛾𝛾
Λ
𝐶𝐶𝜐𝜐 (1) − 1 𝐹𝐹(1)
𝐶𝐶𝑐𝑐 (1) = 𝑃𝑃𝑐𝑐 + 2 𝑀𝑀𝑐𝑐
Λ Λ
(Λ𝐻𝐻⁄𝐵𝐵 )
𝑐𝑐𝑣𝑣 = 𝑒𝑒
Λ𝐻𝐻
𝐹𝐹(1) = 𝐶𝐶𝑣𝑣 (1) − 1 −
𝐵𝐵
−𝑏𝑏 tan 𝜔𝜔 )
𝑎𝑎(1 − 𝑒𝑒 𝑅𝑅 𝑡𝑡∗ 𝐻𝐻𝐻𝐻 tan 𝜙𝜙′ +𝑅𝑅𝑅𝑅
𝛼𝛼� = (1 − 𝑒𝑒 𝐻𝐻 𝑅𝑅 )
𝑏𝑏 tan 𝜔𝜔 𝐻𝐻𝐻𝐻 tan 𝜙𝜙 ′ + 𝑅𝑅𝑅𝑅
Where R is the ratio of the area to the circumference of a horizontal cross-section. The
constants a and b are determined for curve fitting. Value of varies from 4.496 to 2.850
and b varies from 1.935 to 1.638.
The significant improvement over the existing method is that the single wedge method
proposed by Anagnostou and Kovári [2], [7] considers only the average horizontal seepage
force over the height of the wedge and only the resultant normal force (but not the
distribution of the normal stress) at the inclined shear plane. Consequently, the tensile
stresses that may develop locally at the upper part of the wedge remain undetected as
long as the resultant normal force is compressive.
42
Based on the calculations discussed, the authors proposed a dimensionless design
diagrams for necessary support pressure for the condition Overburden/ Equivalent width
= 1. The authors propose that these results are approximately accurate for higher
overburden also. Further authors produced nomograms for two condition: γw/γ’ = 1.0
and for γw/γ’ = 0.8. Design charts for the conservative case (γw/γ’ = 1.0) for various head
difference are produced below for reference.
Figure 41. Design diagrams for effective support pressure, ∆h/H = 0-2 [10]
43
Figure 42. Design diagrams for effective support pressure, ∆h/H = 3-5 [10]
Most of the models for soft cohesive soils are derived for C/D less than 5 and hence our
comparison of different models are restricted to that ratio. Face pressure values are
compared for a surcharge value of 20 kPa in a soil with unit weight γ = 20 kN/m3 and
cu = 50 kPa. The results of the comparison are shown below in Figure 43.
44
450
400
350
Face Pressure in kPa
300
250
200
150
100
50
0
0 1 2 3 4 5 6
C/D Ratio
Broms and Bennermark method Davis method - Cylindrical
Davis method - Spherical Kimura & Mair
Mori - Blowout Augarde et al - Lower
Klar et al
Figure 43. Comparison of face pressure for different cohesive soil model
The results indicate that face pressure calculated using Davis method result in a
conservative face pressure values compared to the recent model by Klar et al 2007 [23].
The face pressure using Mori's model are high because it relates to blowout condition.
To check the sensitivity of undrained shear strength on the estimated face pressure
values, a comparison is made by changing the cu values while retaining C/D ratio of 2
(Overburden = 10m, Diameter = 5m) with a surcharge of 20 kPa. Unit weight and
undrained shear strength are considered constant with depth. Comparison are presented
in Figure 44. It is observed that face pressure estimation by Davis method [11] is higher
as noted in previous plot. Compared to other methods shown in the plot, Davis method
[11] is less sensitive to increase in the undrained shear strength. Minimum face pressure
estimated by Augarde et al [22] is underestimated for lower undrained shear strength
values.
45
300
250
200
Face Pressure in kPa
150
100
50
0
0 20 40 60 80 100 120
Undrained Shear Stength Cu [kPa]
Davis method - Spherical Kimura & Mair
Ellstein Augarde et al - Upper
Klar et al
Figure 44. Comparison of face pressure for different Cu values
Minimum face stability is compared for Cohesionless soil and presented for different C/D
values. For the case of comparison, even Mohr-Coulomb models are considered but with
very less drained cohesion (5 kPa). Surcharge (20 kPa), unit weight of soil (20 kN/m3)
and ϕ' (30°) are kept constant for this comparison. The results are plotted in Figure 45.
The face pressure values using Atkinson and Potts method [26] are very conservative
and are beyond the extents of this plot. An evident observation in the comparison is
that the face pressure for cohesionless soils are independent of the overburden.
Anagnostou & Kovari's method is used without considering the seepage forces.
46
45
40
35
Face Pressure in kPa
30
25
20
15
10
0
0 1 2 3 4 5
C/D Ratio
Krause - Quarter Circle (c' = 5) Leca & Dormieux - MII (c' = 5)
It can be observed that when the relative depth C/D is less than 1, there is a slight
increase in support pressure with the overburden. Above C/D of 1, there is no change
in the minimum support pressure. Support pressure calculated using Mollon's method
[14], Leca & Dormieus method [13], Anagnostou's method [9], Vermeer's empirical [29]
method and Anagnostou & Kovari's method [6] are comparable. Face pressure values
calculated using other methods are conservative. Face pressure values calculated using
Jancsecz and Steiner [33] are linearly increasing with the overburden as it is based on
Terzaghi's load theory for shallow tunnels.
47
200
180
160
140
Face Pressure in kPa
120
100
80
60
40
20
0
15 20 25 30 35 40 45 50
Phi [°]
Krause - Quarter Circle (c' = 5)
Leca & Dormieux - MII (c' = 5)
Caquot/ C.Carranza-Torres (c' = 5)
Kolymbas Method (c' = 5)
Anagnostou 2012 (c' = 5)
Chen, Tang, Yinm et al
Vermeer' Empirical Method (c' = 5)
Mollon (c' = 5)
Atkinson and Potts
Jancsecz & Steiner (c' = 5)
Anagnostou & Kovari 1996 (c' = 5, no seepage)
To understand the influence of ϕ on the minimum face pressure, ϕ' angle is varied for
the above models for the condition C/D = 1.5 (Overburden = 7.5m and Tunnel Diameter
= 5.0m), with surcharge of 20 kPa and unit weight of 20kN/m3. For simplicity, ground
water is not considered and hence no seepage forces are included. The results are
presented in Figure 46. It can be observed that the face pressure estimates by latest
48
models (Anagnostou [9], Chen et al [28] and Mollon [14]) are closer. Face pressure
estimation by Caquot's model [25] are most sensitive to friction angle.
Similar to the behaviour in cohesionless soil, even face pressure in C-Phi soil stay
constant or mildly influenced with the change in overburden.
160
140
120
Face Pressure in kPa
100
80
60
40
20
0
0 1 2 3 4 5
C/D Ratio
Jancsecz & Steiner Krause - Quarter Circle
Figure 47 presents the comparison of face pressure values using models available for c-ϕ
soils. Anagnostou and Kovari's method [6] is compared without considering the seepage
forces. The face pressure estimated in Figure 47 are based on effective cohesion of 25
kPa and ϕ' of 20°. Surcharge and unit weight of soil is considered at 50 kPa and 20
49
kN/m3. Face pressure values estimated by Anagnostou 2012 [9] is constant with the
change in C/D as it is independent of overburden. Anagnostou and Kovari's method [6]
is slightly dependent on overburden depth. It can also be observed that face pressure
estimated using this method are conservative in nature.
700
600
500
Face Pressure in kPa
400
300
200
100
0
20 30 40 50 60 70 80 90 100
c' [kPa]
Jancsecz & Steiner Krause - Quarter Circle
Anagnostou & Kovari 1996 (no seepage) Caquot/ C.Carranza-Torres - Spherical
Kolymbas Method Vermeer' Empirical Method
Anagnostou 2012 Mollon 2009
Figure 48. Comparison of face pressure for c-ϕ soil - influence of cohesion
Figure 48 compares the influence of value of cohesion on the face pressure estimation. It
can be observed that most of the models have equal effect on the change of cohesion.
Kolumbas' method [35] seems to be less sensitive to the change in cohesion. Its again
noticed that the face pressure values estimated using Anagnostou 2012 [9], and Mollon
[14] are comparably similar. Face pressure estimation by Jancsecz's method [33] is
independent of cohesion and the values estimated are beyond the bounds of the graph,
hence it is not shown in the figure.
50
Above comparison is made for a C/D ratio of 1.5m (Overburden = 15m and Diameter
= 10m). Constant surcharge (50 kPa), soil unit weight (20 kN/m3) and ϕ (20°) is
considered for this comparison.
The effect of seepage flow on the face stability was reported by Anagnostou and Kovari
[2], [7] and presented nomograms for the assessment of the required effective support
pressure for different hydraulic boundary conditions. Later, Lee et al. [38] considered
seepage forces in the upper bound solution of Leca and Dormieux [13]. While Lee et al.
[38] investigated only the case of purely frictional soils, Park et. al [37] investigated the
same in a cohesive frictional soil. It was only recently, Perazzelliet. al 2014 [10] presented
dimensionless design normalized diagrams (shown in Figure 41 and Figure 42) for
computation of effective support pressure using method of slices for seepage conditions.
In this section, face pressure values are calculated considering seepage pressure using
two different models - Anagnostou and Kovari 1996 [2] and Perazzelli et. al 2014 [10]
are compared and discussed.
Effective face pressure is estimated for 10m dia tunnel with the condition: Overburden
/ Diameter = 1, c’ = 0 and hydraulic head ∆h = 30m for various ϕ’ values and presented
in Figure 49.
51
350
300
Effective Face Pressure in kPa
250
200
150
100
50
0
15 17.5 20 22.5 25 27.5 30 32.5 35
ϕ' [°]
Perazzelli, Leone, Anagnostou 2014 Anagnostou & Kovari 1996
It is observed that the effective face pressure estimated using Perazzelli et al nomograms
are constantly lower than that of the Anagnostou 1996. Constant difference is
maintained even with the increase of ϕ’ values. This suggests that the method of slices
leads to lower effective support pressure values (for equilibrium condition). The same
comparison is repeated with a constant ϕ’ (= 25°) but now varying the effective cohesion
instead, and results are presented in Figure 50. However, face pressure calculated using
Anagnostou 1996’s nomograms are considerably lower than the one Perazzelli et al’s
nomograms for higher c’ values. This is because, Anagnostou et al 1996 [2] considers
only equilibrium of the prism and does not check the tensile failure.
52
250
200
Effective Face Pressure in kPa
150
100
50
0
0 20 40 60 80 100 120
c' [kPa]
Perazzelli, Leone, Anagnostou 2014 Anagnostou & Kovari 1996
Thus, in case of high hydraulic gradient and if the cohesion of the ground is high (which
may be true for weak rocks), the necessary effective face support pressure may be much
higher than the pressure required for the stability of the wedge. Because, in this case,
tensile failure rather than sliding becomes the critical mode for the determination of
support pressure [10]. This means that, in such situations nomograms of Anagnostou et
al. 1996 [2] may underestimate the necessary support pressure and thus may be unsafe.
This effect is further studied by comparing effective face pressure with varying hydraulic
gradient for two different cohesion value (0 and 100 kPa), using both the methods and
is presented in Figure 51. Results indicate that, as observed above, the results from
Anagnostou et al 1996 [2] are underestimating the support pressure at higher cohesion.
Another important observation from Figure 51 is, as the hydraulic gradient increases,
the estimate by Anagnostou et al 1996 [2] is approaching the values estimated using
Perazzelli et al. (i.e, the governing mechanism is changing from tensile failure back to
limit equilibrium failure).
53
600
500
Effective Face Pressure in kPa
400
300
200
100
Two models to estimate blowout pressure are discussed in this study. One of them is for
Soft cohesionless soil (Mori 1991 [20]) and the second model is for C-Phi soils (MIII
mechanism of Leca and Dormieux [13]). In order to compare, the blowout pressure is
estimated for a shallow tunnel (C/D of 1, with C = 5m) using both the methods for
different values of undrained cohesion. ϕ value of 1 degree is used in Leca and Dormieux's
model. A constant surcharge and unit weight of 20kPa and 20kN/m3 is considered.
54
700
600
Blow Out Pressure in kPa
500
400
300
200
100
0
0 10 20 30 40 50
Cu [kPa]
From Figure 52 it is evident that blowout pressure estimation using Leca and Dormieux
[13] method is very sensitive to the Cu value. Mori's method [20] is based on fracturing
model in cohesive soils whereas Leca's model is primarily derived for frictional soil and
then extended to general C-Phi soil. This can be the reason for wide difference in the
blowout pressure estimation at higher Cu value. Hence model selection for a particular
site needs to selected with due care.
In order to apply the theory discussed so far, a real life case study is presented in this
section. Project under consideration is part of a Mass Rapid Transit (MRT) metro
project. The project includes twin bored tunnels with outer diameter of 6.35m. The bored
tunnels are being constructed using an Earth Pressure Balanced (EPB) Tunnel Boring
Machine (TBM). Pre-cast concrete segmental lining is installed within the tail shield of
55
the TBM, designed to provide final support for the tunnel. The length of the tunnel
considered for this case study is 0.7km between two metro stations.
The TBM is excavated to encounter weathered Granite and soil during its drive. The
soil profile typically consists of fill followed by fluvial sands, fluvial clays, estuarine and
Granite in its various degree of decomposition. The soil profile along the tunnel
alignment is shown in Figure 53. Dark line represents the interface between the Granite
and soil. Based on the geological profile, it can be observed that there are many mixed
face conditions expected along the tunnel drive.
TBM operating pressure is estimated using the theory discussed in section 2.1. Stability
model by Anagnostou et al 1996 [2] and their nomograms are used to calculate effective
support pressure while considering the seepage forces. The total face support pressure is
the sum of effective face support pressure and relevant pore water pressure in the TBM
cutter head chamber.
The imposed loads comprise of surface surcharge of 20 kPa and gravity loading of soil
above the sliding wedge. Ground water table at 1m below the ground surface is
56
considered in calculations. Both drained and undrained behaviour of relevant soil
materials are examined.
TBM face support pressure required to maintain stable face conditions at the cutter
head during TBM driving has been performed at analysis sections based on every 10m
interval. A total of 70 number of analysis sections have been calculated for the alignment
shown in Figure 53. Both serviceability (SLS) and ultimate (ULS) limit states have been
considered. For ULS calculations, minimum partial factors have been applied to soil
parameters -ϕ’ and c’ divided by 1.25 (for ϕ’, the reduction is applied to tanϕ’).
It is observed that when the tunnel is passing through ground consisting of interbedded
sands and clays, it is not possible, using the simple charts, to account for the combination
of the different types of soil. Averaging the parameters with different weights for crown,
upper face, invert etc could be done to reach a pseudo-realistic behaviour. However, use
of the Horn model, or its derivatives, is not appropriate where there are substantial
thicknesses of clay, as the failure surface in clay does not correspond to that used for the
Horn model.
This limitation is overcome by using separate analyses for the beds of clay or sand in
and over the face of the tunnel. The highest calculated face pressure can then be adopted
as the Target / Recommended pressure. At this point, understanding the limitations /
behaviour of different stability models gives a better judgement in selecting appropriate
stability models. This approach should be conservative, in relation to the actual
conditions. Using it, a variety of possible ground conditions are allowed for, giving a
significant degree of robustness to the calculations. Based on these considerations, the
minimum total face pressure, Recommended / Target face pressure and maximum face
pressure is calculated and presented in Figure 54. In addition, other relevant details like
hydrostatic water pressure and octahedral stress are also presented in the chart.
57
Figure 54. Estimated TBM operating face pressure – Case study
58
3 Probability Analysis
The most widely accepted and successful way to deal with the uncertainties inherent in
dealing with geological materials came to be known as the observational method. It is
also an essential part of the New Austrian Tunnelling Method [41]. The observational
method grew out of the fact that it is not feasible in many geotechnical applications to
assume very conservative values of the loads and material properties and design for those
conditions. The resulting design is often physically or financially impossible to build.
Instead the engineer makes reasonable estimates of the parameters and of the amounts
by which they could deviate from the expected values. Then the design is based on
expected values - or on some conservative but feasible extension of the expected values
- but provision is made for action to deal with the occurrence of loads or resistances that
fall outside the design range. During construction and operation of the facility,
observations of its performance are made so that appropriate corrective action can be
made. This is not simply a matter of designing for an expected set of conditions and
doing something to fix any troubles that arise. It involves considering the effects of the
possible range of values of the parameters and having in place a plan to deal with
occurrences that fall outside the expected range. It requires the ongoing involvement of
the designers during the construction and operation of the facility.
As discussed above, one of the traditional ways used to address uncertainty is to use
factor of safety design approach. More recent alternatives are the load and resistance
factor design (LRFD) approach and the characteristic values and partial factors used in
the limit state design approach in Eurocode 7. Yet another approach can play at least a
useful complementary role to LRFD and Eurocode 7, namely the design based on a
target reliability index that explicitly reflects the uncertainty of the parameters and their
correlation structure. It must be emphasized at the outset that reliability approaches do
not remove uncertainty and do not alleviate the need for judgment in dealing with the
world. They do provide a way of quantifying those uncertainties and handling them
consistently [42].
59
The simplest definition of the term reliability is the probability of a desirable and
sustainable function of a system. Engineering reliability, as in mathematics, is a basic
science which may be used in tunnelling as well. Since years ago, many researchers have
been dealing with the issue of reliability of structures. Also, some researchers have tended
to work on reliability in geotechnical structures. Various methods for application of
reliability analysis in engineering problems can be found out in Dai and Wang [43]. There
are also some references that indicate the application of reliability engineering specifically
on geotechnical engineering.
Reliability analyses, involving neither complex theory nor unfamiliar terms, can be used
in routine geotechnical engineering practice. These reliability analyses require little effort
beyond that involved in conventional geotechnical analyses. Baecher and Christian [42]
reported many fundamental approaches for reliability practices in geotechnical field.
More recently, Duncan [44] did a comprehensive study on the factors of safety and
reliability in geotechnical engineering. Reliability evaluation can provide a means of
evaluating the combined effects of uncertainties in the parameters involved in the
calculations, and offer a useful supplement to conventional structural analyses.
Engineering reliability is based on the principles of probability and mathematics which
is now a useful tool for estimating the proper function of systems. Reliability based
design calls for a willingness to accept the fundamental philosophy that
• Absolute reliability is an unattainable goal in the presence of uncertainty
• Probability theory can provide a formal framework for developing design criteria
that would ensure that probability of failure is acceptably small.
This also provides a consistent method for propagation of uncertainties and a unifying
framework for risk assessment across disciplines and national boundaries [45].
60
Figure 55. General stochastic problem [45]
3.1 Uncertainties
Two main sources of geotechnical uncertainties can be distinguished. The first arises
from the evaluation of design soil properties, such as undrained shear strength and
effective stress friction angle. This source of geotechnical uncertainty is complex and
depends on inherent soil variability, degree of equipment and procedural control
maintained during site investigation, and precision of the correlation model used to relate
field measurement with design soil property. Based on Phoon [45] three ranges of soil
property variability (low, medium, high) were found to be sufficient to achieve
reasonably uniform reliability levels for simplified reliability based checks.
The second source arises from geotechnical calculation models. Although many
geotechnical calculation models are “simple,” reasonable predictions of fairly complex
61
soil–structure interaction behaviour still can be achieved through empirical calibrations.
Model factors, defined as the ratio of the measured response to the calculated response,
usually are used to correct or simplifications in the calculation models.
A wide range of engineering and scientific disciplines use simulation methods based on
randomized input, often called Monte Carlo methods. They have been employed to study
both stochastic and deterministic systems. Because developments are often bound to the
application disciplines, notation and nomenclature sometimes reflect the preferences of
a particular field. Monte Carlo methods can be divided into two broad areas. First, there
is the simulation of a process that is fundamentally stochastic. The second area involves
problems that are not inherently stochastic but can be solved by simulation with random
variables. Typically, Monte Carlo method require that the function be evaluated a large
number of points. This becomes easy with the present computing systems. The idea of
Monte Carlo methods is the generation of random events in a computer model and this
generation is repeated many times and count the occurrence number of specific
conditions. The basic procedure of Monte Carlo method is:
• Define a domain of possible events
• Generate events randomly
• Perform deterministic judgements of system based on the events
62
• Count the occurrence number of a specific system state among total observations
Having assigned PDF for input parameters considered as random variables, the
probabilistic assessment of the geotechnical parameters is accomplished. The Monte
Carlo simulation is applied to obtain minimum face stability pressure using the Microsoft
Excel program. In the Monte Carlo simulation the values of each input parameter is
generated randomly considering the PDF.
The advantage of the Monte Carlo technique compared to other probabilistic methods
is the possibility of getting complete PDF of output. The main disadvantage of the
method is the large number of simulations to be executed.
The problem can be idealized by considering a circular rigid tunnel of diameter D driven
under a depth of cover C. A surcharge qs is applied at the ground surface
Due to uncertainties in the soil shear strength parameters, the cohesion c, and the angle
of internal friction ϕ are considered as random variables. Probability assessment is
carried out for three models - Mollon et al 2010 [14], Anagnostou 2012[9] and Chen et al
2014 [28]. Results of the Monte Carlo simulation are presented in this section. Monte
Carlo simulation is performed using simple spreadsheet software. The pseudo random
number generator is used to simulate failure times of individual units in the system.
63
Logical expressions are then used to determine system success or failure. (PDF) and
cumulative distribution function (CDF) of the support pressure are presented.
In order to explain the procedure used for Monte Carlo (MC) simulation, the work flow
is explained using a simple model at first and then outputs using other models are
presented and compared.
Figure 57 shows the input required for performing Monte Carlo simulation for a simple
closed form model using Anagnostou, 2012 [9]. In addition to the four variables in the
equation, additional inputs to indicate the expected variation in the parameters are
required. This could be defined using statistical parameters, mean and coefficient of
variation. However, for simplicity, only absolute maximum and minimum values of each
parameters are used as input (the random generator considers equal probability for each
value in that range). No variation is considered in diameter.
Using the parameters as discussed above, Monte Carlo simulation is run with a sample
size of 5000. For each computation, the variables are randomly selected within the
bounds of expected variation.
64
Figure 58. Snapshot of Random number generation
Figure 59 presents the histogram of one such iteration. More number of iterations could
be performed to refine the results. It could be observed from the histogram plot that
with the expected variation in the input paramters, the most likely estimate of support
pressure (using Anagnostou, 2012 [9]) is around 42 kPa. However, there are chances that
the expected variablity in the parametes could result in support pressure as high as 135
kPa.
65
600
500
400
Count
300
200
100
0
17.7
25.6
33.4
41.3
49.2
57.1
65.0
72.9
80.7
88.6
96.5
1.9
9.8
104.4
112.3
120.2
128.0
135.9
143.8
151.7
159.6
Support Pressure [kPa]
Figure 59. Histogram of MC simulation using Anagnostou, 2012
1.00
0.90
0.80
Cumulative Probablity
0.70
0.60
0.50
0.40
0.30
0.20
0.10
0.00
104
112
120
128
136
144
151
159
10
17
25
33
41
49
57
65
73
81
88
96
2
66
Figure 61. Summary of MC simulation using Anagnostou, 2012 in Excel
Since the simulation generates the PDF and CDF of the required support pressure, we
can estimate the minimum support pressure for the required reliability percentage
instead of (over-conservatively) assuming certain factor of safety. Here in this case, the
factor of safety that would have resulted in 95 percentile to result is back calculated and
found to be around 1.8.
By varying the input parameters, it is observed that, for the required confidence on the
minimum support pressure, (back calculated) factor of safety varies with the physical
configuration and spread of input parameters. Hence, the minimum support pressure
calculated using probabilistic analysis are more reliable and relevant to site conditions /
uncertainties.
Similar probabilistic assessment is performed for cohesionless soils using Chen et al [28]
model (by varying γ and ϕ as assumed in Figure 57, with c = 0 and C/D = 1.5) and the
67
results are presented in Figure 62. It is observed that results using this model have higher
central tendency and follow an ideal normal distribution. It is to be noted that for such
probability distribution, higher factor of safety values are not required even for most
reliable estimate of face pressure.
700 1.00
0.90
600
0.80
Cumulative Probability
500 0.70
0.60
400
Count
0.50
300
0.40
200 0.30
0.20
100
0.10
0 0.00
13
15
16
18
19
21
22
24
25
27
28
30
31
33
34
36
37
39
40
41
43
68
350 1.00
0.90
300
0.80
Cumulative Probability
250 0.70
0.60
200
Count
0.50
150
0.40
100 0.30
0.20
50
0.10
0 0.00
11
14
17
19
22
25
28
31
34
37
40
43
46
49
52
55
58
2
5
8
-1
Based on this procedure, Monte Carlo simulations are performed for Face stability
models which consider seepage forces (Anagnostou and Kovari 1996 [2] and Perazzelli
et. al 2014 [10] ) and important observations are noted.
For this analysis, effective cohesion c’ and effective friction angle ϕ’ are assumed to
follow normal distribution. Parameters for c’ are: mean = 50 kPa, standard deviation =
20 kPa. Parameters for ϕ’ are: mean = 25°, standard deviation = 5°. The distribution
of c’ and ϕ’ considered in the analysis is presented in Figure 64. C/D ratio of 1 is
considered with a constant hydraulic gradient, ∆h/H of 3. Submerged unit weight of soil
is taken as 10 kN/m3.
69
.
Figure 64. Distribution of input parameters for MC simulation
Monte Carlo simulation is run with a sample size of 10,000 for the both the methods
and Probability Distribution Function of the effective face pressure are compared in
Figure 65. Green histogram are corresponding to the results from Anagnostou and Kovari
1996 and Red histogram are the values corresponding to Perazzelli et. al 2014 [10].
0.016
0.014
0.012
0.01
Frequency
0.008
0.006
0.004
0.002
0
-200
-150
-100
100
150
200
250
300
350
-50
50
0
Some of the important observations from this comparison are, effective face pressure
values estimated using Anagnostou and Kovari 1996 [2] have wider spread of possible
70
values. That means, higher factor of safety needs to be used using this method to account
for various uncertainties. Whereas, the estimation by the second method leads to values
with higher central tendency and factor of safety required is relatively less to achieve
the same reliability. In summary, a logical method to select the appropriate model for
face pressure estimation is discussed using simple analytical tools. More importantly, an
approach to consider the uncertainties in the design (rather than relying on over
conservative factor of safety) is discussed.
71
4 Results and Conclusion
In preparation of this report, the author has investigated different available models for
face pressure calculation for different types of soils. Recent development of stability
models were discussed and compared with widely used methods in the industry. With
the constant process of ‘pushing the envelope’, bigger, shallower (or deeper) tunnels, and
ever more stringent criteria, expansion of the existing design nomograms / curves, to
cover lower and higher C/D ratios than currently available, would also be useful.
Based on the sensitivity analysis presented in section 2.2 and the case study presented
in section 2.3, we can deduce that it is very important to understand the limitations of
each model. Although there has been efforts to formulate a generalized model for all
types of soils following Mohr-Coulomb criterion, some models are found to be sensitive
for certain range of input values.
Similar to the closed faced TBMs, this approach could be used in conventional tunnelling
/ SEM / NATM to determine probability of face failure / probability of tunnel face
72
deformation using different support mechanism and assess related risks or select most
reliable support mechanism.
Although the calculations are an essential part in the development of suitable operating
pressures, a significant degree of engineering judgement needs to be applied both before
and after the calculations. Before carrying out the calculations the ground and
groundwater model has to be developed, select critical sections etc. and after
calculations, we need to assess the risks of encountering more adverse conditions
(requiring significantly higher pressure), understand the influence of interfaces etc.
73
Bibliography
[1] V. Guglielmetti, P. Grasso, A. Mahtab, and S. Xu, Mechanized tunnelling in urban
areas: design methodology and construction control. CRC Press, 2008.
[3] X.-W. Tang, W. Liu, B. Albers, and S. Savidis, “Upper bound analysis of tunnel
face stability in layered soils,” Acta Geotechnica, vol. 9, no. 4, pp. 661–671, Jul.
2014.
[4] R. Chen, J. Li, L. Kong, and L. Tang, “Experimental study on face instability of
shield tunnel in sand,” Tunnelling and Underground Space Technology, vol. 33,
pp. 12–21, Jan. 2013.
[6] G. Anagnostou and K. Kovári, “Face stability in slurry and EPB shield
tunnelling,” in Geotechnical Aspects of Underground Construction in Soft Ground,
1996, pp. 453–458.
[8] W. Broere, “Tunnel Face Stability & New CPT Applications,” TU Delft, 2001.
[10] P. Perazzelli, T. Leone, and G. Anagnostou, “Tunnel face stability under seepage
flow conditions,” Tunnelling and Underground Space Technology, vol. 43, pp. 459–
469, Jul. 2014.
74
[11] E. H. Davis, R. J. Mair, H. N. Seneviratine, and M. J. Gunn, “The stability of
shallow tunnels and underground openings in cohesive material,” Géotechnique,
vol. 30, no. 4, pp. 397–416, 1980.
[12] H.-B. Mühlhaus, “Lower bound solutions for circular tunnels in two and three
dimensions,” Rock Mechanics and Rock Engineering, vol. 18, no. 1, pp. 37–52,
1985.
[13] E. Leca and L. Dormieux, “Upper and lower bound solutions for the face stability
of shallow circular tunnels in frictional material,” Géotechnique, vol. 40, no. 4, pp.
581–606, 1990.
[14] G. Mollon, D. Dias, and A. Soubra, “Face Stability Analysis of Circular Tunnels
Driven by a Pressurized Shield,” no. January, pp. 215–229, 2010.
[15] J. H. Atkinson and R. J. Mair, “Soil mechanics aspects of soft ground tunnelling,”
Ground Engineering, vol. 14, no. 5, pp. 20–38, 1981.
[16] R. J. Mair and R. N. Taylor, “Theme lecture: Bored tunneling in the urban
environment,” of XIV ICSMFE [131], pp. 2353–2385, 1999.
[18] T. Kimura and R. J. Mair, “Centrifugal testing of model tunnels in soft clay,” in
Proceedings of the 10th International Conference on Soil Mechanics and
Foundation Engineering, 1981, pp. 319–322.
[19] A. R. Ellstein, “Heading failure of lined tunnels in soft soil,” Tunnels and
Tunnelling, pp. 51–54, 1986.
75
[22] C. E. Augarde, A. V. Lyamin, and S. W. Sloan, “Stability of an undrained plane
strain heading revisited,” Computers and Geotechnics, vol. 30, no. 5, pp. 419–430,
Jul. 2003.
[23] A. Klar, A. S. Osman, and M. Bolton, “2D and 3D upper bound solutions for
tunnel excavation using ‘elastic’ flow fields,” International Journal for Numerical
and Analytical Methods in Geomechanics, vol. 31, no. 12, pp. 1367–1374, 2007.
[24] H. Tresca, “On the Flow of Solid Bodies Subjected to High Pressures,” Comptes
Rendus de l’Académie des Sciences, vol. 59, p. 754, 1864.
[30] S. Kanayasu, I. Kubota, and N. Shikibu, “Stability of face during shield tunneling
– A survey of Japanese shield tunneling,” Fujita and Kusakabe, vol. 71, pp. 337–
343, 1966.
[32] M. Mohkam and Y. . Wong, “Three dimensional stability analysis of the tunnel
face under fluid pressure,” in Numerical Methods in Geomechanics, 1989, pp.
2271–2278.
76
[33] S. Jancsecz and W. Steiner, “Face support for a large mix-shield in heterogeneous
ground conditions,” in Tunnelling’94, Springer, 1994, pp. 531–550.
[37] J. K. Park, J. T. Blackburn, and J. Ahn, “Upper bound solutions for tunnel face
stability considering seepage and strength increase with depth,” in Underground
Space – the 4th Dimension of Metropolises Barták, Hrdina, Romancov & Zlámal
(eds), 2007, no. c, pp. 1217–1222.
[38] I. M. Lee and S. W. Nam, “The study of seepage forces acting on the tunnel lining
and tunnel face in shallow tunnels,” Tunnelling and Underground Space
Technology, vol. 16, no. 1, pp. 31–40, 2001.
[40] P. Perazzelli, T. Leone, and G. Anagnostou, “A limit equilibrium method for the
assessment of the tunnel face stability taking into account seepage forces,” in
World Tunnel Conference 2013 Geneva, 2013, pp. 715–722.
[41] L. . Rabcewicz, “The New Austrian Tunnelling Method (Part 1 to 3),” Water
Power, 1964.
[43] S.-H. Dai and M.-O. Wang, Reliability analysis in engineering applications. Van
Nostrand Reinhold, 1992.
77
[45] K.-K. Phoon, Reliability-based design in geotechnical engineering: computations
and applications. CRC Press, 2008.
78
Curriculum Vitae
In Aug. 2009, he started working as Geotechnical design engineer and was mostly
involved in major infrastructure projects including Road and Rail Tunnels, Mining and
urban development projects.
In Jan. 2014, he started to pursue the 2nd level specializing master’s course at Politecnico
di Torino and after finishing the course work, he joined the consulting firm - Geoconsult
Asia Singapore as Executive Civil Engineer (Tunnels).
79
Appendix A
List of stability models studied for this report.
Sl. No. Name of the Model Year Soil Type Analysis Type Type of Model Constitutive Model
3D model with linear triangular
1 Horn Model 1961 Homogeneous Global Equilibrium -
wedge
2 Murayama Model 1966 Homogeneous Global Equilibrium 2D log-spiral wedge Mohr-Coulomb material
3 Broms & Bennermark 1967 Homogeneous Emperical Model 2D/3D Empirical model Tresca material
6 Kimura & Mair 1981 Homogeneous Emperical Model 2D/3D Empirical model Tresca material
13 Chambon & Corte 1994 Homogeneous Stress Method 2D Empirical model Cohesionless material
14 Jancsecz & Steiner 1994 Homogeneous Global Equilibrium 3D Wedge model - Earth pressure Mohr-Coulomb material
17 Belter / Katzenbach 1999 Heterogeneous Global Equilibrium 2D linear wedge model Mohr-Coulomb material
20 Vermeer' Empirical Method 2001 Homogeneous Semi-emperical Model 2D/3D empirical plasticity solution Mohr-Coulomb material
25 Klar et al 2007 Homogeneous Stress Method 2D/3D kinematic approach Tresca material
26 Mollon 2010 Homogeneous Stress Method 3D multi-block failure mechanism Mohr-Coulomb material
80