Service Life Prediction of Metro Tunnel Lining Based On Water Ingress

See discussions, stats, and author profiles for this publication at: https://www.researchgate.
net/publication/343236053
Service Life Prediction of Metro Tunnel Lining Based on Water Ingress
Thesis · December 2017

DOI: 10.13140/RG.2.2.20356.76160
CITATIONS READS
0 32
1 author:
Yuqi Tan
ATC Williams
5 PUBLICATIONS 14 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
Service life prediction of a metro tunnel View project
All content following this page was uploaded by Yuqi Tan on 27 July 2020.
The user has requested enhancement of the downloaded file.

Service Life Prediction of Metro Tunnel Lining Based on Water Ingress
A thesis submitted in fulfilment of the requirements for the degree of

Doctor of Philosophy
Yuqi Tan
B. Eng (University of South China), M. Eng (University of Wollongong)
School of Engineering
College of Science, Engineering and Health
RMIT University
December 2017
Declaration
I certify that except where due acknowledgement has been made, the work is that of the
author alone; the work has not been submitted previously, in whole or in part, to qualify for
any other academic award; the content of the thesis is the result of work which has been
carried out since the official commencement date of the approved research program; any
editorial work, paid or unpaid, carried out by a third party is acknowledged; and ethics
procedures and guidelines have been followed.
Yuqi Tan
19th December 2017
i
Abstract
The reliability of a tunnel is an ever-present concern for both tunnel managers and daily
commuters. Water ingress of a tunnel was identified as a hazardous element for tunnels by
previous work. An example metro tunnel was investigated in this study, in which water
ingress is considered to be the major failure criterion.
This thesis aims to provide a time-dependent reliability prediction for the selected tunnel
based on water ingress. Water ingress rate was measured during site inspection in order to
understand the water ingress condition of the tunnel. An analytical method was developed as
a tool to achieve a more efficient onsite water ingress rate estimation. Using this analytical
method, the measurements of the water ingress rate were extrapolated to the entire tunnel.
It was found from tunnel inspections that the cracking condition of the concrete lining is a
major contributing factor for water ingress. Water ingress through cracked concrete was
studied as a combination of the mechanisms of water seeping through a porous material and
water inflow through cracked material. The crack connectivity was defined to describe the
crack interconnection within a concrete. The crack connectivity of the selected tunnel was
determined from the inspected crack patterns and generalised using Monte Carlo simulation.
A relationship between the crack connectivity and the hydraulic conductivity of a tunnel
lining was established analytically.
The hydraulic conductivity of the tunnel lining is a key parameter for tunnel water ingress
calculations. Previous studies of water ingress into a tunnel are mainly based the assumption
that the hydraulic conductivity of the lining is homogeneous. The hydraulic conductivity of a
cracked concrete is mainly determined by the crack width according to previous research.
The crack width in concrete is difficult to measure onsite when the crack is covered by
precipitated mineral matters. An analytical method was applied to evaluate the overall crack
ii
width of the tunnel lining under the assumption that it is fully homogeneous. The
probabilistic distribution of crack width was calculated when the hydraulic gradient and the
rock mass properties are known.
The tunnel lining investigated in this study has a thickness of more than 550 mm and the
crack pattern through the tunnel lining appears to be variable (based on existing ground
penetrating radar data) and could cause inhomogeneous hydraulic conductivity. An analytical
solution for water ingress was developed for a tunnel with inhomogeneous lining under a
steady state assumption. The results show that the inhomogeneity of the lining affects the rate
of water ingress significantly when the inhomogeneity degree is high.
The cracking condition of the selected tunnel has been monitored for three years with eight
inspections by the tunnel operator. The crack inspection data was evaluated in this study in a
time-dependent way. Based on the cracking data and the newly developed water ingress
analytical model, a time-dependent reliability calculation for the selected tunnel was
developed. The hydraulic parameters including water table, rock mass property and crack
formation were considered probabilistically. The service failure probability of each section of
the selected tunnel was calculated time-dependently. The results show that the remaining
service life of the selected tunnel varies by sections. The most critical section is predicted to
possibly fail in eight years while some other sections do not show signs of increased water
inflow.
iii
Acknowledgement
It has been a challenging and interesting experience for my PhD study. There are many
people who have helped me with this research. I would not be able to complete this thesis
without their support.
I would like to thank my parents Weidong Tan and Xiangfen Zhu for their care and love.
They encouraged and supported me both mentally and financially in the past twenties years. I
also would like to thank my partner, Ms Xin Ma for her mental support which is invaluable
for me.
I would like to thank my supervisory team Associate Professor John Smith and Professor
Chun-Qing Li for their guidance through my candidature. Thank you for their constant
support and mentoring. I have learnt how to be a good researcher, a good engineer, and more
importantly, how to learn and think in a more effective way. It was truly an honour and
privilege to be their student.
I would like to thank Professor Yufei Wu, Dr Gang Ren, Dr Jie Li, Dr Guoyang Fu, Dr
Matthew Currell and Dr Hassan Baji for their invaluable advice on my research. Also many
thanks go to Mr David Wood and Mr Jordan Kalyvas for their help on English of this thesis.
Financial support from Metro Trains Melbourne is also gratefully acknowledged here.
Last but not least, I would like to thank all my colleagues and friends. Thank you all for their
company through every moment of my life.
iv
List of Publication
The following are the research papers that were published through this research candicature.
TAN, Y., SMITH, J. V., LI, C.-Q., CURRELL, M. & WU, Y. 2018. Predicting external water
pressure and cracking of a tunnel lining by measuring water inflow rate. Tunnelling
and Underground Space Technology, 71, 115-125.
TAN, Y., SMITH, J. V., LI, C.-Q. & DAUTH, J. 2017. Calcium leaching of a concrete tunnel
lining under aggressive groundwater conditions. World Tunnel Congress 2017, 2017
Bergen, Norway. Norwegian Tunnelling Society, 1-5.
TAN, Y., SMITH, J. V., LI, C. Q. & DAUTH, J. 2016. Water infiltration estimation in
tunnels: with reference to the Melbourne underground. Transforming the Future of
Infrastructure through Smarter Information: Proceedings of the International
Conference on Smart Infrastructure and Construction. Cambridge, UK, 597-602.
REN, G., LI, C. Q. & TAN, Y. 2015. Numerical analysis of effects of water leakage with loss
of fines on concrete tunnel lining. In: TATIANA ROTONDA, M. C., FRANCESCO
SILVESTRI, PAOLO TOMMASI (ed.) Volcanic Rocks and Soils. The Netherlands:
CRC Press/Balkema, 191-193.
v
Table of Contents
Declaration .................................................................................................................................. i
Abstract ......................................................................................................................................ii
Acknowledgement .................................................................................................................... iv
List of Publication ...................................................................................................................... v
List of Figures ........................................................................................................................... xi
List of Tables .........................................................................................................................xvii
1 Introduction ........................................................................................................................ 1
1.1 Background ................................................................................................................. 1
1.2 Significance ................................................................................................................. 3
1.3 Aim and objectives ...................................................................................................... 5
1.4 Scope of research ........................................................................................................ 6
1.5 Outline of the thesis..................................................................................................... 7
2 Literature Review ............................................................................................................. 10
2.1 Introduction ............................................................................................................... 10
2.2 Construction of tunnels ............................................................................................. 11
2.2.1 Design of tunnels ............................................................................................... 11
2.2.2 Construction methods of tunnels ....................................................................... 14
2.2.3 Tunnel behaviour ............................................................................................... 17
2.2.4 Tunnel maintenance methods ............................................................................ 23
2.3 Concrete properties ................................................................................................... 27
2.3.1 Concrete composition ........................................................................................ 28

vi
2.3.2 Concrete strength ............................................................................................... 33
2.3.3 Permeability of concrete .................................................................................... 33
2.3.4 Factors affecting water ingress in concrete ........................................................ 37
2.4 Engineering hydraulics .............................................................................................. 38
2.4.1 Basics of hydraulics ........................................................................................... 38
2.4.2 Mechanisms of water movement in concrete..................................................... 41
2.4.3 Hydraulic conductivity of cracked material ....................................................... 43
2.4.4 Homogeneity of cracked material ...................................................................... 49
2.4.5 Water inflow into tunnel .................................................................................... 50
2.5 Reliability theory ....................................................................................................... 63
2.5.1 Basics of reliability theory ................................................................................. 63
2.5.2 Outcrossing method ........................................................................................... 68
2.5.3 Monte Carlo simulation ..................................................................................... 72
2.5.4 System reliability ............................................................................................... 75
2.6 Summary ................................................................................................................... 79
3 Water Ingress Into Concrete Lining ................................................................................. 81
3.1 Introduction ............................................................................................................... 81
3.2 Hydraulic conductivity of different categories of concrete ....................................... 84
3.2.1 Different categories of concrete ......................................................................... 85
3.2.2 Development of hydraulic conductivity............................................................. 87
3.2.3 Crack connectivity coefficient ........................................................................... 92
vii
3.2.4 Worked example and discussion ...................................................................... 100
3.3 Crack width evaluation of a homogeneous tunnel lining ........................................ 108
3.3.1 Profile of tunnel for effective crack width evaluation ..................................... 108
3.3.2 Development of effective crack width model .................................................. 110
3.4 Water inflow into a tunnel with inhomogeneous lining .......................................... 116
3.4.1 Development of water inflow solution of a inhomogeneous lining ................. 119
3.4.2 Hydraulic pressure and water inflow rate of lining ......................................... 125
3.5 Summary ................................................................................................................. 142
4 Site Investigation ............................................................................................................ 144
4.1 Introduction ............................................................................................................. 144
4.2 Techniques for site investigation ............................................................................ 146
4.2.1 Technique of water inflow measurement ......................................................... 147
4.2.2 Technique of concrete compressive strength test ............................................ 148
4.2.3 Technique of crack feature investigation ......................................................... 149
4.2.4 Technique of chemical analysis of samples ..................................................... 150
4.3 Water inflow rate of lining ...................................................................................... 151
4.3.1 Development of local water inflow rate calculation ........................................ 152
4.3.2 Ratio between measure flow rate and total flow rate ....................................... 158
4.3.3 Water inflow rate measured on site ................................................................. 162
viii
4.4 Compressive strength of lining concrete ................................................................. 163
4.4.1 Investigation in CCL ........................................................................................ 164
4.4.2 Investigation in BLP ........................................................................................ 166
4.4.3 Summary of compressive strength of lining concrete...................................... 167
4.5 Crack features on the lining .................................................................................... 168
4.5.1 Crack mapping ................................................................................................. 169
4.5.2 Changing of number of wet crack .................................................................... 173
4.5.3 Cracking data ranking ...................................................................................... 181
4.6 Chemical analysis of samples ................................................................................. 183
4.6.1 Sample list........................................................................................................ 183
4.6.2 Properties of solid samples .............................................................................. 186
4.6.3 Properties of water samples ............................................................................. 188
4.7 Summary ................................................................................................................. 190
5 Tunnel Service Life Prediction ....................................................................................... 193
5.1 Introduction ............................................................................................................. 193
5.2 Reliability of tunnel lining subjected to water inflow ............................................. 195
5.2.1 Model of environmental effect......................................................................... 197
5.2.2 Function of acceptance .................................................................................... 200
5.2.3 Reliability of tunnel due to water inflow ......................................................... 201
5.3 Reliability of tunnel lining due to water seeping .................................................... 214
5.3.1 Model of environmental effect......................................................................... 215
ix
5.3.2 Function for acceptance ................................................................................... 217
5.3.3 Reliability of water seeping failure mode ........................................................ 218
5.4 System failure probability of tunnel lining ............................................................. 220
5.4.1 System of tunnel .............................................................................................. 220
5.4.2 Reliability of tunnel lining ............................................................................... 221
5.4.3 Service life of the tunnel lining........................................................................ 226
5.4.4 Risk rating ........................................................................................................ 227
5.5 Summary ................................................................................................................. 230
6 Water Inflow Treatment of Tunnel Lining ..................................................................... 232
6.1 Introduction ............................................................................................................. 232
6.2 Injection material..................................................................................................... 232
6.2.1 Epoxy injection resins ...................................................................................... 233
6.2.2 Polyurethane injection resins ........................................................................... 233
6.2.3 Silicate injection resins .................................................................................... 233
6.3 Crack treatment in tunnel lining .............................................................................. 234
6.4 Failed treatments ..................................................................................................... 237
6.5 Summary ................................................................................................................. 242
7 Conclusions and Recommendations ............................................................................... 244
8 Reference ........................................................................................................................ 248
Appendix A ............................................................................................................................ 256
x
List of Figures
Figure 2.1 Water permeability testing device (Reinhardt and Gaber, 1990) ........................... 34
Figure 2.2 Experimental set-up of water inflow through sand column (Darcy, 1856) ............ 41
Figure 2.3 Water flow through an impermeable material with one aperture ........................... 44
Figure 2.4 Tunnel profile of classical method ......................................................................... 51
Figure 2.5 Conformal mapping of physical plane on circular ring .......................................... 56
Figure 2.6 Series system. ......................................................................................................... 75
Figure 2.7 Parallel system. ....................................................................................................... 76
Figure 3.1 Schematic illustration of a seepage path of concrete .............................................. 87
Figure 3.2 Two-dimensional crack and the relevant parameters ............................................. 93
Figure 3.3 Illustration of the seepage path when cracks are crossing ...................................... 96
Figure 3.4 Flow chart of Monte Carlo simulation for the connectivity coefficient of critical
seepage path ............................................................................................................................. 98
Figure 3.5 Three-dimensional crack and the relevant parameters ......................................... 100
Figure 3.6 Seepage path hydraulic conductivity increase when connectivity coefficient
increases from 0 to 0.99 (a) and 0.99999 to 1 (b), the other parameters are fixed determined
value ....................................................................................................................................... 101
Figure 3.7 Increase of water flow rate because of connectivity coefficient increase from 0 to
0.99, the other parameters are fixed determined value .......................................................... 102
Figure 3.8 Lowest and highest calculated connectivity coefficient based on the site measured
data ......................................................................................................................................... 103
Figure 3.9 Water ingress through concrete causes (a) water dripping and (b) efflorescence 104
Figure 3.10 Water seeping through time under different crack connectivity coefficient ...... 105
Figure 3.11 Time for water to seep through lining relative to thickness ............................... 106
Figure 3.12 Illustration of tunnel geometry parameters......................................................... 109

xi
Figure 3.13 Examples of a crack difficult to access for crack width measurement (a) and
crack difficult to measure width due to the deposit on crack surface (b) .............................. 110
Figure 3.14 Relation between water inflow rate and crack width when the other parameters
are given ................................................................................................................................. 114
Figure 3.15 Multiple-dimensional design chart for crack width evaluation .......................... 115
Figure 3.16 Schematic illustration of hydraulic conductivity of three types of homogeneous
tunnel...................................................................................................................................... 118
Figure 3.17 Head loss under both homogeneous (right) and inhomogeneous (left) conditions
................................................................................................................................................ 120
Figure 3.18 Illustration of the example concrete lining with steel reinforcement elements (a)
transverse section (b) longitudinal section ............................................................................. 126
Figure 3.19 Relation of the water head at different lining position r-a with different value
when =10. The horizontal axis is the distance outward from the inner surface
of the lining, equal to r-a. Parameters derived from selected tunnel ..................................... 128
Figure 3.20 The relationship of water head and inhomogeneous hydraulic conductivity
coefficient at outer surface of the lining. Parameters derived from selected tunnel ......... 130
Figure 3.21 The relationship of water inflow rate and inhomogeneous hydraulic conductivity
coefficient at the outer surface of the lining when varies from 0.01 to 100. Parameters
derived from the selected tunnel ............................................................................................ 131
Figure 3.22 The water inflow factor chart for parameters of the selected tunnel. An example
of collected data is shown ...................................................................................................... 133
Figure 3.23 Water flow rate is calculated compared with the water inflow rate according to
the water inflow factor chart .................................................................................................. 134
Figure 3.24 Relation between crack features and inhomogeneous hydraulic conductivity
coefficient ( ). Parameters for the selected tunnel and collected data are marked .............. 136
xii
Figure 3.25 Illustrations of types of lining and rock mass hydraulic conductivity conditions.
(a) Lining hydraulic conductivity homogeneous; (b) Lining hydraulic conductivity
constrictive; (c) Lining hydraulic conductivity expansive..................................................... 138
Figure 3.26 Illustrations of types of hydraulic conductivity conditions of a tunnel with a
drainage layer. (a) homogeneous lining hydraulic conductivity; (b) constrictive lining
hydraulic conductivity; (c) expansive lining hydraulic conductivity .................................... 139
Figure 4.1 Device used for surface water collection.............................................................. 148
Figure 4.2 Illustration of plinth, walk way and tunnel lining in a tunnel .............................. 149
Figure 4.3 A transverse wall crack with a fan shaped flow is observed in the centre of the
photograph (as arrow). A complex crack comprising diagonal and horizontal segments is
observed from top centre to the right hand side of the photograph. The latter crack is
currently dry but previously flowing evidenced by the white mineral staining .................... 153
Figure 4.4 Illustration of measuring water inflow rate of a fan shaped flow caused by
transverse cracks .................................................................................................................... 154
Figure 4.5 Illustration of the representative point in a Cartesian coordinate ......................... 155
Figure 4.6 Flow velocity variation through given elevation within the flow area, when (a)
, (b) (b) ......................................................................................................... 159
Figure 4.7 Flow velocity comparison on the water source along elevation direction ........... 159
Figure 4.8 Relation between flow rate ratio and the collector dimension ( )................ 160
Figure 4.9 Water flow rate ratio chart when and .......................... 161
Figure 4.10 Calculated effective crack width under a given water inflow rate when the other
parameters are given .............................................................................................................. 163
Figure 4.11 Statistical comparison of concrete compressive strength of plinth, walkway and
tunnel lining in CCL .............................................................................................................. 165
xiii
Figure 4.12 Statistical comparison of concrete compressive strength of plinth, walkway and
tunnel lining in BLP ............................................................................................................... 167
Figure 4.13 Statistical comparison of concrete strength of plinth walk way and tunnel lining
(overall condition) .................................................................................................................. 168
Figure 4.14 Crack mapping on site ........................................................................................ 169
Figure 4.15 Crack mapping from chainage 5550 to 5595 in BLP ......................................... 170
Figure 4.16 Crack mapping from chainage 5550 to 5595 in CLP ......................................... 170
Figure 4.17 Crack mapping from chainage 5550 to 5595 in NLP ......................................... 171
Figure 4.18 Crack mapping from chainage 5550 to 5595 in CCL......................................... 171
Figure 4.19 Representative section of GPR data ................................................................... 173
Figure 4.20 Average number of wet cracks per 100 m length of tunnel, (a) CLP, (b) CCL, (c)
BLP and (d) NLP. .................................................................................................................. 174
Figure 4.21 Number of wet cracks per 100 m length of tunnels from portal to FGS, (a) CLP,
(b) CCL, (c) BLP and (d) NLP .............................................................................................. 176
Figure 4.22 Number of wet cracks per 100 m length of tunnels from FGS to MCS, (a) CLP,
(b) CCL, (c) BLP and (d) NLP .............................................................................................. 177
Figure 4.23 Number of wet cracks per 100 m length of tunnels from MCS to PS, (a) CLP, (b)
CCL, (c) BLP and (d) NLP .................................................................................................... 178
Figure 4.24 Number of wet cracks per 100 m length of tunnels from PS to portal, (a) CLP, (b)
CCL, (c) BLP and (d) NLP .................................................................................................... 179
Figure 4.25 Increasing rate of number of wet cracks by tunnels and sections ...................... 181
Figure 4.26 Tunnel section ranking based on increasing rate of number of wet cracks, (a)
CLP, (b) CCL, (c) BLP and (d) NLP ..................................................................................... 183
Figure 4.27 Site data collection ............................................................................................. 184
xiv
Figure 4.28 Comparison of water samples chemistry from tunnel and local ground water
property (a) pH comparison, (b) Salinity comparison, (c) Calcium ion content comparison
and (d) Iron ion content comparison ...................................................................................... 189
Figure 5.1 Illustration of a location where the water inflow rate greater than 1 L/min ......... 201
Figure 5.2 Failure probability of one metre tunnel component from portal to FGS section, (a)
CLP, (b) CCL and (c) BLP and (d) NLP ............................................................................... 203
Figure 5.3 Failure probability of one metre tunnel component from FGS to MCS section, (a)
Figure 5.4 Failure probability of one metre tunnel component from MCS to PS section, (a)
Figure 5.5 Failure probability of one metre tunnel component from PS to portal section, (a)
Figure 5.6 Failure probability of component system from portal to FGS sections, (a) CLP, (b)
CCL, (c) BLP and (d) NLP .................................................................................................... 209
Figure 5.7 Failure probability of component system from FGS to MCS sections, (a) CLP, (b)
CCL and (c) BLP and (d) NLP .............................................................................................. 210
Figure 5.8 Failure probability of component system from MCS to PS sections, (a) CLP, (b)
Figure 5.9 Failure probability of component system from PS to portal sections, (a) CLP, (b)
Figure 5.10 Illustration of a location where the white deposit was seen on the lining surface
................................................................................................................................................ 218
Figure 5.11 Failure probability of (a) BLP and NLP and (b) CLP and CCL due to water
seeping ................................................................................................................................... 219
xv
Figure 5.12 Probability of system failure of tunnel due to water ingress from Portal to FGS, (a)
Figure 5.13 Probability of system failure of tunnel due to water ingress from FGS to MCS, (a)
Figure 5.14 Probability of system failure of tunnel due to water ingress from MCS to PS, (a)
Figure 5.15 Probability of system failure of tunnel due to water ingress from PS to portal, (a)
Figure 6.1 Relation of increasing rate of number of wet cracks and number of treated crack,
BLP, CCL, CLP and NLP refer to different tunnels .............................................................. 236
Figure 6.2 Relation of increasing rate of number of wet cracks and number of treated crack in
section between MCS and PS, BLP, CCL, CLP and NLP refer to different tunnels ............ 237
Figure 6.3 Number of failed treatment, (a) CLP, (b) CCL, (c) BLP and (d) NLP ................ 239
Figure 6.4 Illustration of shallow injection depth in the selected tunnel ............................... 241
xvi
List of Tables
Table 2.1 Water inflow rating of 10 m long tunnel section (Bieniawski, 1990) ..................... 19
Table 2.2 Methods for tunnel maintenance for water ingress and the correspondent
mechanisms .............................................................................................................................. 27
Table 3.1 Assumed parameters used for a two-dimensional simulation ................................. 94
Table 3.2 Parameters used in the worked example for three-dimensional simulation ............ 99
Table 3.3 Parameters used in the calculation ......................................................................... 101
Table 3.4 Variation range of hydraulic parameters for β value ............................................. 115
Table 3.5 Framework for water inflow prediction models .................................................... 117
Table 3.6 Parameter ranges used for validation of the water inflow factor chart .................. 134
Table 3.7 Procedure for using water inflow factor chart ....................................................... 135
Table 4.1 Rock mass hydraulic conductivity and tunnel dimensions .................................... 144
Table 4.2 Hydraulic head of tunnel sections (in m above centre line) .................................. 145
Table 4.3 Summary of site inspections .................................................................................. 146
Table 4.4 Summary of using the finite length water collection method for different flow
shapes ..................................................................................................................................... 161
Table 4.5 Water flow measurement and the correspondent crack length .............................. 162
Table 4.6 Concrete compressive strength in CCL ................................................................. 164
Table 4.7 Concrete compressive strength in BLP .................................................................. 166
Table 4.8 Crack length and crack density of each tunnel segment in the mapped sections .. 172
Table 4.9 Number of wet cracks from year 0 to year 3 and rate of linear interpolation of the
increase in number of wet cracks. Total numbers for each tunnel are shown in bold. .......... 180
Table 4.10 List of the collected samples ................................................................................ 185
Table 4.11 Sample properties based on EDS and XRD tests ................................................ 187
Table 4.12 Properties of water sample and local groundwater .............................................. 188
xvii
Table 5.1 Means and Coefficient of variation of different parameters used in water inflow
reliability calculation ............................................................................................................. 198
Table 5.2 Mean functions and the coefficient of variation functions for different tunnel
sections ................................................................................................................................... 199
Table 5.3 Acceptable component failure number ( ) for different component system in all
tunnel sections, the total number of each tunnel sections in bracket ..................................... 208
Table 5.4 Service life of each section of different tunnels subject to water inflow through
cracks (years) ......................................................................................................................... 214
Table 5.5 Means and Coefficient of variation of different parameters used in water seeping
reliability calculation ............................................................................................................. 216
Table 5.6 Mean functions and the coefficient of variation functions for different tunnel ..... 217
Table 5.7 Service life of each section of different tunnels subject to water seeping through
lining (years) .......................................................................................................................... 220
Table 5.8 Service life of each section of different tunnels due to water ingress (years) ....... 227
Table 5.9 Description of five risk categories for water ingress ............................................. 228
Table 5.10 Time-dependent risk rating of tunnel sections of CLP ........................................ 228
Table 5.11 Time-dependent risk rating of tunnel sections of CCL........................................ 229
Table 5.12 Time-dependent risk rating of tunnel sections of BLP ........................................ 229
Table 5.13 Time-dependent risk rating of tunnel sections of NLP ........................................ 229
Table 6.1 Number of failed treatments in different tunnels and the total number of treated
cracks is in brackets ............................................................................................................... 238
xviii
1 Introduction
1.1 Background
Tunnels are an important part of the engineering infrastructure in most cities. Tunnels are
often used as an option for transportation of both passengers and resources. They are
commonly used to transport resources such as water in hydropower projects. Tunnels for
passengers include road tunnels for road vehicles and rail tunnels for trains. Amongst all the
tunnels for passenger transportation, railway metro tunnels play a particularly important role
for daily commuters especially in urban areas. The tunnelling industry in Australia is
developing rapidly especially in major coastal cities. Recent tunnelling projects include the
Cross River rail project in Brisbane, Westconnex project in Sydney and Westgate tunnel
project in Melbourne. In the city of Melbourne, which is the second largest city in Australia,
the metro system forms a railway loop in the central business district of the city. The average
daily patronage carried by the city railway system, including both tunnel and railway above
ground, is 415,000 as reported by Metro Trains Melbourne.
Tunnels have a significant advantage in a modern city compared with surface transport routes.
Space on the ground surface is getting more crowded in a modern city due to the growing
population. Tunnels provide the option of travelling underground which saves space on the
ground surface significantly. More than space saving in urban areas, the development of
tunnels also enables travel across natural barriers including rivers and mountains and saves
time for transporting materials and passengers. Furthermore, extreme weather conditions can
affect the traffic on ground surface. Tunnels can protect the transportation system from those
extreme weather conditions since tunnels are buried underground.
1
The underground infrastructures are often considered to be a significant engineering
achievement due to their complexity during design and construction. The depth of tunnelling
is an important aspect of the difficulty that can be experienced in tunnelling. In order to
withstand the stress from the load of ground above a tunnel, ground support is required to be
installed during tunnel construction. Some of the ways this is achieved include the use of
structural linings. The lining of a tunnel is typically made from steel-reinforced concrete. In
many modern tunnels, the tunnel segments are precast and then installed on site. The concrete
lining can also be cast in situ by installing forms which separated by segment joints on-site
within the tunnel.
The length of a tunnel also is an aspect of difficulty in tunnel construction due to the need to
transport materials and personnel to the excavation faces. The longest railway tunnel
worldwide is the Gotthard base tunnel which is located in the central Swiss Alps (Lüth et al.,
2008). This tunnel has a total length of 151.8 km. This tunnel is also currently the world’s
deepest traffic tunnel.
Despite the achievements of tunnelling projects all over the world, many problems and
hazards remain for tunnel construction. The Hengqin tunnel in China partially collapsed due
to the failure of tunnel support on 19 July 2012 during the construction stage (Liu et al.,
2012). The roof of a ventilation tunnel in Lane Cove tunnel in Sydney collapsed on 2
November 2005 during the construction stage (Brown, 2006). Approximately 150 concrete
ceiling panels in the Sasago tunnel in Yamanashi prefecture, Japan collapsed and caused
multiple casualties on 2 December 2012 (Nakahara, 2014).
While railway tunnels carry a large amount of city traffic, the tunnels themselves can
deteriorate with time due to environmental conditions. Cracking of concrete due to various
2
mechanisms can often occurs in a deteriorating tunnel lining. The cracks and joints on the
lining would become the weakest location of the lining under environmental conditions.
There are multiple deterioration mechanisms including, reinforcement corrosion, concrete
degradation, soil swelling and water ingress, which could cause the failure of a tunnel.
Among all the hazardous elements, water ingress was identified to be one of the most
common problems from a survey of tunnels in many countries (Cesano et al., 2003, Shin et
al., 2012). It was reported that water ingress in a tunnel can also act as a trigger for the other
failure mechanisms due to the aggressive chemical content of groundwater. The water ingress
in a tunnel would worsen the tunnel service condition over time. Therefore it is necessary to
evaluate the safety and serviceability condition of urban metro tunnels based on water ingress.
Remediation works in a tunnel are necessary especially when the safety or serviceability
conditions of the tunnel became unacceptable.
1.2 Significance
The maintenance cost for a tunnel is considered be a major cost, based on a project review
conducted by Infrastructure Australia (Infrastructure Australia, 2008). The maintenance of
tunnels often costs millions of dollars due to the complexity of concrete linings (ITA, 1991).
The maintenance cost of a train tunnel in Switzerland was reported to be as much as 11
million Australian dollars in 1980s (ITA 1991). The high cost of maintenance is required to
mitigate or eliminate the problems which could affect the safety or the serviceability of a
tunnel.
Water ingress into tunnels is a widely recognised problem worldwide. The report published
by the International Tunnelling Association (ITA, 1991) investigated 48 tunnels in 12
countries which were found to have water ingress problems. It was found that water ingress
initiated safety problems in tunnels including concrete degradation, reinforcement corrosion

3
and soil swelling. Aggressive groundwater ingress through concrete was reported to degrade
concrete under extreme conditions (Gérard et al., 2002). Degradation of concrete linings can
be critical since concrete linings are often acting as a protection for the traffic inside the
tunnel. Water ingress can also initiate the corrosion of steel reinforcement of tunnel (Song et
al., 2009). The support of a tunnel would be weakened once the steel reinforcement elements
become corroded. The structural deterioration in a tunnel could reduce the strength of the
lining and then affect the overall tunnel support. Those safety problems could cause structural
failure or collapse of a tunnel in extreme conditions.
Unlike the safety problems which could cause tunnel collapse, service problems in a tunnel
will not directly result in catastrophic failure of a tunnel as a structure. The service problem
could cause difficulty for the operation of a tunnel. The financial loss due to a service
problem could be significant especially in a populated urban area. Excessive water ingress
into a tunnel is a hazardous element for tunnel operation. Water inflow into the tunnel would
either be dripping from crown or flowing along tunnel lining walls. Any forms of water
ingress would cause potentially adverse effects on the electrical system which are installed in
the tunnel. Furthermore, flowing water can cause mineral deposition on tunnel walls, in the
tunnel invert, in drainage system or on walkways of the tunnel. Mineral deposition on
walkways could adversely affect the working environment for working crews in the tunnel
and use of the walkways in emergencies. In general, water ingress is an important hazardous
element in tunnels. The condition of water ingress into a tunnel reaching a serviceability limit
is expected to occur at an earlier stage than would affect the structural safety of the tunnel.
As discussed earlier, water ingress is an important factor for tunnel serviceability. More
research needs to be done about this phenomenon. Most water ingress into a tunnel is from
cracks or joints of the lining. The features of cracks such as crack length, crack width and
4
crack orientation could be difficult to interpret due to the complexity of concrete linings.
Regular site investigations would only be able to observe the crack features on the lining
surface. The crack features within the lining would also contribute greatly to both water
inflow and structural stability. The crack features within a tunnel lining can be investigated
by the modern technologies including ground penetrating radar (GPR) and laser scanner.
However, those technologies are generally expensive to use and would not identify the entire
crack features inside the concrete lining when the lining is thick. A method is provided in this
study for analysing the crack features inside the tunnel lining using water inflow rate as the
indicating parameter.
Maintenance is an important approach to extend the service life of a tunnel. Polymer injection
is often considered to be an important method of maintenance of a tunnel to control water
ingress. This maintenance method is often expensive and time consuming since the locations
where maintenance is required are often spread through the whole tunnel. Timing and
locations of such maintenance is critical to achieve better maintenance efficiency and reduce
cost. The efficiency of tunnel maintenance can be low without a clear understanding of the
water ingress deterioration process. The deterioration process of a tunnel can be quantified by
reliability theory. The reliability of a tunnel structure has been studied based on the criteria
associated with stress and loading (Yue and Ang, 2015). The realiability assessement of a
tunnel based on water ingress has not been systematically studied yet.
1.3 Aim and objectives
This thesis aims to provide a time-dependent reliability prediction for an selected tunnel
based on water ingress. The key objective of this research is to develop a new methodology
to present the remaining safe life of a tunnel with concrete lining subject to water ingress.
5
The methodology for calculating water ingress into a tunnel was developed and examined.
The method developed is then applied to a specific lined tunnel to demonstrate its application.
The specific objectives and the correspondent methodologies of the research are listed below:
1) Develop a model for hydraulic conductivity of concrete by the connectivity of cracks
inside concrete. This objective is achieved by using analytical method combine with
Monte Carlo simulation.
2) Develop a method to calculate local water inflow rate in a concrete-lined tunnel with a
limited amount of site measurements. This objective is achieved by using analytical
method combine with site investigation.
3) Develop a method to calculate the crack width in a concrete-lined tunnel with
assumed homogenous (constant) hydraulic conductivity. This objective is achieved by
using analytical method.
4) Develop a method to calculate water inflow rate and hydraulic pressure distribution in
a tunnel lining with inhomogeneous (variable) hydraulic conductivity. This objective
is achieved by using analytical method.
5) Develop a method to predict the service life of the selected tunnel system due to water
ingress. This objective is achieved by using analytical method combine with site
investigation.
6) Evaluate the effectiveness of the remediation work of an selected tunnel system. This
objective is achieved by site investigation and data analysis.
1.4 Scope of research
The key scope of the study is a concrete lining in an existing tunnel where water is flowing
into the tunnel. Site inspection information on the condition of a tunnel lining in the past
6
three years was made available by the tunnel operator and was supported by eight site visits
by the author of this study. The concrete properties, local inflow rates, water properties and
crack distribution information are the only measured site data in this study. All other
information was obtained from literature.
This study only investigates the concrete lining of the tunnel. The properties of rock mass and
water table are based on the information in literature and are not discussed in detail.
The failure of a tunnel could be caused by many mechanisms including loading, water and
settlement. This study only investigates the failure mechanism of water ingress into the tunnel.
Other failure mechanisms are not considered in this study.
For the water ingress failure criterion, this study only considers the condition when water
ingress into the tunnel is saturated and steady state. The condition when water ingress is in a
transient state is not considered.
1.5 Outline of the thesis
This thesis is organised into seven chapters as described below:
Chapter one describes the significance and background for conducting the service life
prediction of the example metro tunnel system. The objectives and the scope of this research
are also presented.
Chapter two reviews the current literature. Four main aspects of knowledge are reviewed,
namely, construction of tunnels, concrete properties, engineering hydraulics and reliability
theory. Construction includes the design, construction and maintenance method for a tunnel.
Concrete is discussed as the construction material of tunnel linings in the following section.
Engineering hydraulics describes water movement and behaviour for an engineering
7
application. Reliability gives a basic theory and method for structure reliability calculation.
The research gaps are also identified after each section in this chapter.
Chapter three develops the theory of water ingress into a tunnel. The chapter is divided into
three sections. The first section defines a new parameter called connectivity coefficient. The
hydraulic conductivity of a cracked concrete is calculated based on the connectivity
coefficient. The second section develops a method to calculate the crack width in a tunnel.
The homogeneous lining assumption is used in this section. The probabilistic distribution of
crack width in the selected tunnel is given. A design chart which comprises multiple
parameters is given for crack width estimation in a generic tunnel. The third section develops
a method to consider the condition when water inflow through a tunnel with an
inhomogeneous lining. The potential water inflow and the hydraulic head difference caused
by the inhomogeneity is discussed in this section. A design chart which comprises multiple
hydraulic parameters is given in this section which could be used to interpret the level of
inhomogeneity of a tunnel lining.
Chapter four develops a method for site investigation. A method for water inflow
measurement in a tunnel is given in this chapter. The method is based on a finite length
measurement method. Water inflow rate measured in this study is collected by this method.
The water inflow rates measured in this section are used in the following sections. The
compressive strength of concrete of the tunnel lining is tested and used to calculate the
hydraulic conductivity of concrete. The crack feature on the lining, including the mapping of
surface crack, changing of number of wet crack and crack data ranking is presented in this
chapter. The chemical properties of the collected samples are presented to have a general
understanding of the condition of the tunnel.
8
Chapter five develops a method for the tunnel reliability calculation based on the criterion of
water ingress. The criterion of water ingress is contributed by two failure modes, which are
water inflow and water seeping. The time-dependent failure probability of different sections
of the selected tunnel is calculated for water inflow. The time-dependent failure probability of
the tunnel is also calculated for water seeping by two categories which are deep tunnels and
shallow tunnels. The failure probability of the water ingress criterion is developed based on
the time-dependent failure probability of water seeping and water inflow.
Chapter six evaluates the remediation works which have been conducted in early 2013. The
tunnel conditions in different tunnel section after remediation works were compared to the
conditions before remediation.
Chapter seven summarises the main findings of this research. This chapter also presents
recommendation for further research.
9
2 Literature Review
2.1 Introduction
This study aims to deliver a service life prediction of a tunnel under the water table. Four
major topics contribute to this study are reviewed, namely, construction of tunnels,
engineering hydraulics for underground structures, concrete properties and the reliability
theory.
Tunnels are structures excavated underground, usually for transportation purposes.
Knowledge of the construction of tunnels is required in order to understand their engineering
performance. The construction methods and the design of various tunnel types are reviewed.
The behaviour of tunnel linings are studied with reference to relevant maintenance methods.
The review aims to cover all relevant concepts of tunnels relative to water inflow problems.
Water ingress into a tunnel is a common problem for a tunnel under the water table.
Engineering hydraulics is reviewed to study the water ingress behaviour for an engineering
application. The knowledge of hydraulics is reviewed in order to quantify the phenomenon of
water inflow of tunnel. Darcy’s law and other commonly used equations are outlined for the
water ingress process. Analytical and numerical studies for water inflow prediction into a
tunnel are reviewed in that section.
The lining of a tunnel is often constructed using concrete as the material. The properties of
concrete, including the composition of concrete and the permeability of cement and the
hydraulic conductivity if concrete, are reviewed in this section for the water ingress
prediction. The theory of reliability is studied as the assessment method for a tunnel. Various
methods for reliability calculation are reviewed and discussed in that section.
10
This section aims to provide a summary of the required background knowledge for service
life prediction of a tunnel with concrete lining. The research gap of the current research are
identified and discussed.
2.2 Construction of tunnels
2.2.1 Design of tunnels
The design of tunnels usually varies case by case, and no universally accepted standards have
been reported on specific design for tunnel. Handbooks, government documents and
guidelines are used for tunnel design in most countries.
A general, method for design of a tunnel was published by the International Tunnelling
Association (ITA, 1988). The design criteria for a tunnel and its structural lining were
outlined in that document. These criteria include deformation and strains (ε), stress (σ) and
utilization of plasticity and lining strength of a cross-section (ITA, 1988). That study stated
that a national code may not always be appropriate for the design of tunnel. The variation of
ground information including geology formation, geology fracturing, texture and degree of
weathering are critical for the design of a tunnel and its support system. The water conditions
affecting tunnel design include water level, acceptable amount of water inflow (Q) and water
chemistry also vary according to the location of a given tunnel.
The International Tunnelling Association (ITA, 1988) study elaborated on the required
disciplines in a tunnel design which include geology, geotechnical engineering, excavation
technology and supporting elements and law. A British design guideline for tunnels aimed to
outline design of tunnel lining (BTS, 2004). The design parameters for linings in that
guideline include the geotechnical parameters, design life and lining settlement. The design
criteria of a tunnel in that guideline include the durability of tunnel lining, fire resistance of
11
tunnel and water proofing (BTS, 2004). The durability of a tunnel lining depends on the
selected material, ground conditions and water properties. Fire resistance of a tunnel depends
on the tunnel shape, types of fire and lining material. The water proofing of a tunnel includes
the selection of membranes, gaskets and grouting for leak prevention. The chemistry of water
is a main consider factor for the water proofing design of tunnels (BTS, 2004). Monitoring
methods and quality management were also discussed in that guideline (BTS, 2004).
A British tunnel design handbook described comprehensive knowledge which is required for
tunnelling (Kuesel et al., 2012). That handbook comprises the required knowledge in the
planning stage, design stage, construction stage and maintenance stage of tunnels. The design
criteria in that handbook were divided into design criteria for temporary support and design
criteria for permanent linings. The design criteria for a temporary lining include, lining
thickness (d), groundwater acceptance (Q) and length of unsupported ground from shaft (L).
The design criteria for a permanent lining includes properties of shotcrete, properties of
concrete, size of lining segment and properties of steel (Kuesel et al., 2012).
The general principles of structural design of tunnels including support of the ground and the
additional loading is outlined in a German tunnel design guideline (Maidl et al., 2014). That
design guideline also provides details in terms of acceptable water inflow during the
construction stage.
The design of tunnel linings is determined by the environmental requirement, groundwater
property and geology conditions (Maidl et al., 2014). The criteria for tunnel design include
structural design, safety of the tunnel and water inflow of tunnel (Maidl et al., 2014). The
detailed description of criteria for structural design, safety of tunnel and water inflow tunnel
will be further elaborated in the following paragraphs.
12
Most tunnels require ground support to meet safety standards. The lining of a tunnel can be
installed either permanently or temporarily after the excavation. A permanent lining should
be installed after the temporary lining. This may be followed by installing a non-structural
secondary lining to protect the inner devices for tunnel operation (Queensland Government,
2007). The acceptable factor of safety applied in tunnel design varies case by case. The factor
of safety for structural design of a tunnel should be in a range from 1.33 to 2 in a typical
engineering case (Maidl et al., 2014). The factors for the lining design should include ground
support and ventilation system based on a government tunnelling code (Queensland
Government, 2007). The capacity of the tunnel lining should be sufficient for the soil or rock
stresses that are induced by tunnel excavation and loading (Queensland Government, 2007).
The lining should be designed as a part of the ventilation system to provide adequate air
supply to the working surface to deal with the production of dust and heat.
Tunnel lining is often designed to mitigate the water ingress in water-rich zones (Maidl et al.,
2014). The water inflow rate of a tunnel section usually has a maximum limit which can vary
case by case and depends on multiple factors including pump capacity and the purpose of the
tunnel. Different lining types affect the behaviour of a tunnel in different ways. Tunnels have
been categorized in three categories based on lining types for water inflow modelling
(Butscher, 2012), namely, unlined tunnel, lined and drained tunnel and lined and water
sealing tunnel. Unlined tunnels were constructed with the exposed rock face (Maidl et al.,
2013). The exposure of rock surfaces can be reinforced with bolts and steel mesh to secure
the potential loose rock on the free face (Grob, 1975). Lined and drained tunnels were
designed to drain the water table and reduce the hydraulic pressure on the tunnel (Test, 1956).
Drained tunnels usually have more than three layers of lining. The first layer of lining is
designed to be installed at the rock interface and then followed by a drainage layer with high
13
hydraulic conductivity (Maidl et al., 2014). The secondary lining is designed to resist water
ingress and ensure no water leakage into the tunnel. Water inflow in the lining is drained by
the drainage layer and then pumped out. The lined and water sealing tunnels are often
preferred when the water table needs to be maintained and protected. The construction of
water sealing tunnels also comprises more than three layers for a modern tunnel (Maidl et al.,
2014). The first layer of lining is designed to support ground followed by a water-sealing
membrane layer which will reduce the lining hydraulic conductivity. The third layer is often
defined as finishing layer which would protect the water sealing membrane layer from
environmental erosion.
2.2.2 Construction methods of tunnels
There are three types of construction methods for rock tunnelling which are widely used in
engineering practice namely, drill and blast, tunnel boring machine and roadheaders (Hung et
al., 2009).
Drill and blast
Drill and blast is a method for tunnel advancement using drilling equipment and explosives.
The term drill refers to the drilling process to make blasting holes, blast refers to the
explosion caused by explosive material (Maidl et al., 2013). The drill and blast method was
used for tunnel construction before 1900 in the Alps of Switzerland. This method was
continuously used in 20th century. Examples includes the Niagara falls power station tunnel
in Canada from 1980 to 1958, the Mont Blanc road tunnel across France and Italy from 1959
to 1964 and the construction of Arlberg tunnel in Austria from 1974 to 1978 (Maidl et al.,
2013). The design parameters for the drilling process includes machine technology
parameters, rock mechanical parameters, geological factors, human factors and other
14
influences (Maidl et al., 2013). The speed of drilling operation usually varies from less than 1
metre/min to up to 6 metres/min which is determined by the quality of the rock mass and the
drilling equipment. The design parameters of blasting include rock mechanics parameters
such as uniaxial compressive strength, thickness of rock burden, blasting hole diameter and
hole spacing (Maidl et al., 2013).
Tunnel support is often required after excavation especially when ground conditions are poor
(Bieniawski, 1990). The primary support methods include rock bolts, shotcrete (including
fibrecrete) and steel ribs. The primary support is often installed shortly after the excavation of
the tunnel. The purpose of the primary support is to ensure the tunnel is stable until the
installation of the secondary lining which is often concrete lining (Bieniawski, 1990).
Tunnel boring machine
A tunnel boring machine (TBM) is a machine for tunnel advancement by cutting through the
rock mass. The first tunnel boring machine was designed by Charles Wilson in 1851 and
patented in 1856. A modern TBM is a complex system which comprises many interdependent
components (Maidl et al., 2013). A tunnel boring machine excavates the full surface of a rock
face with a rotating head which has multiple cutter heads embedded in the rotating head. The
excavated rock is transported from the cutting face to the tunnel portal and surface depending
on the transportation method onsite. Examples of the use of tunnel boring machines include
the Lotschberg base tunnel from 1999 to 2007 and the Gotthard base tunnel from 1999 to
2011. Both of these tunnels were constructed through the Alps of Switzerland. The length of
the Lotschberg tunnel and Gotthard base tunnel are 34.5 km and 151.8 km, respectively. The
design parameters for the TBM method include rock mechanics parameters, groundwater
profile and tunnel length (Maidl et al., 2013). Problems have occurred under the conditions
15
when water inrush occurred into the tunnel carrying sand (Kuesel et al., 2012). The sand and
rock fragment are known to become jammed between the TBM cutters and then reduce the
tunnelling efficiency (Kuesel et al., 2012). It was reported to be slow and difficult to clean
the jammed cutter head to regain the cutting speed (Kuesel et al., 2012).
Concrete support usually is applied shortly after the exposure of the rock surface in a modern
TBM (Maidl et al., 2013). The concrete lining is usually pre-cast as concrete segments and
loaded on the TBM behind the cutting face. The pre-cast concrete lining segments can be
installed semi-automatically right after the excavation as the support for the tunnel (Maidl et
al., 2013).
Roadheaders
The roadheader is a machine for tunnel advancement which can create a more versatile tunnel
shape compared to the tunnel boring machine. The roadheader was initially designed in
mining industry for the removal of small coal seams. The roadheader was patented by Dr.
Ajtay in 1949 in Hungary. The main advantage of roadheader is that it has a high level of
adaptability, availability and low cost (Hung et al., 2009). This method is often applied when
the tunnel project which is short in length. It is far less efficient in long tunnels compared to
TBM tunnelling method. The factors that need to be considered when using a roadheader for
tunnelling include, rock strength, tunnel shape and water pressure.
The support required for a roadheader tunnel is similar to that of the drill-blast tunnel as
discussed earlier. The primary support methods including rock bolts, shotcrete and steel ribs
are often installed shortly after the excavation followed by the installation of the secondary
lining which is often concrete lining (Bieniawski, 1990).
16
Different tunnelling methods have been discussed in this section. It is notable that all the
aforementioned tunnelling methods have been used in the selected tunnel in this study.
2.2.3 Tunnel behaviour
Cracking
In a typical lined tunnel, the cracking of lining concrete is a notable and important behaviour.
The mechanisms of concrete cracking can be reviewed and summarised as, thermal-induced
cracking, chemical-induced cracking, shrinkage-induced cracking and stress-induced
cracking (Neville, 2011). Thermal-induced cracking was found to be caused by temperature
differences of outer and inner concrete. The temperature difference would inducing stress can
cause cracking (Harrison, 1981). Chemical-induced cracking was found caused by the
material expansion inside the concrete. The typical examples from chemical reaction induced
cracking include the corrosion induced crack and the alkali–silica reaction. The stress-
induced cracking of concrete occurs when the load on concrete exceed the concrete capacity
(Neville, 2011). Polymer injection is a common approach to sealing cracks in a concrete
lining. A polymer injection work using self-healing material in Tokyo metro tunnel was
reported (Murakami et al., 2015). The sealant used in that study is considered to be a durable
material since it would heal the crack in long term. A revisit one year after the injection
shows that the injection works well in terms of crack sealing (Murakami et al., 2015).
Settlement
The settlement of a tunnel can cause differential movement of tunnel lining segments (Shen
et al., 2014). The differential movement of segments can potentially induce lining cracking
and water leakage. A case study of the settlement problem of Shanghai metro tunnels was
reported (Shen et al., 2014). The settlement of No. 1 line of Shanghai metro was measured for
17
10 years through different sections of the tunnel. The maximum settlement was reported to be
150 mm in the critical section in the whole metro tunnel. That study found that the
groundwater ingress could be an indicator of tunnel settlement since the settlement would
cause the displacement of the lining joints (Shen et al., 2014).
Water inflow
This section aims to describe the water inflow behaviour of both lined and unlined tunnels.
Water inflow into a tunnel has been reported as a commonly existed issued over last century
(ITA, 1991). The International Tunnelling Association has reviewed the water inflow
conditions of tunnel and the correspondent treatments method of different types of tunnel
worldwide. That study found that water inflow can be a very costly problem when the water
inflow affects the tunnel serviceability.
The study in this section aims to identify the water inflow problem in tunnel worldwide. The
methods for water inflow calculation are further reviewed in Section 2.4.
An empirical study was developed to incorporate the water inflow condition into tunnel
design (Barton et al., 1974). In combination with the properties of the rock mass, that study
used the water inflow as a parameter for tunnel stability evaluation (Barton et al., 1974). An
empirical study was conducted for the tunnel design in terms of both water flow and rock
mass condition (Bieniawski, 1990). That study developed a rating system for water inflow
rate into 10 m length tunnel section (Table 2.1). The tunnel section was rated as damp when
the flow rate is less than 10 L/min. The tunnel section was rated as wet when the flow rate is
10-25 L/min. The tunnel section was rated as dripping when the flow rate is 25-125 L/min.
The tunnel section was rated as flowing when the flow rate is over 125 L/min.
18
Table 2.1 Water inflow rating of 10 m long tunnel section (Bieniawski, 1990)
Water inflow rate of 10 m

tunnel section (Litres / min) None <10 10-25 25-125 >125
General condition Dry Damp wet Dripping Flowing
A case study of the groundwater inflow problem of a hard rock tunnel in Sweden was
conducted (Cesano et al., 2000). A rating method based on the water inflow rate of each
location was developed. The measured water inflow data in that tunnel were categorised as
major leakage, moderate leakage and minor leakage. That study found that the minor leakage
is caused by the water in rock mass while the major leakage is affected by the overburden
material (Cesano et al., 2000). That study used the measured water inflow rate and other
hydraulic parameters to evaluate the water inflow condition of a tunnel. However, the method
for local water inflow measurement has not been clearly identified. A method for local water
inflow rate measurement on site is further discussed in later chapters.
The water drainage caused by a rock tunnel in Norway was studied numerically (Kitterod et
al., 2000). That study simulated the inhomogeneity of rock mass by considering the fracture
distribution in the rock mass. The water inflow rate in Romeriksporten tunnel in Norway was
simulated using a boundary which has dimension 200 m wide × 600 m long × 300 m tall.
That study showed the impact of the water inflow field when place the drainage well around a
tunnel. That study showed the least successful drainage well would decrease 28% water
inflow of the tunnel while the most successful drainage well would decrease 81% water
inflow of the tunnel (Kitterod et al., 2000). The water inflow rate is the output parameter in
that study used to calibrate the model.
19
The water inflow through fractured rock was studied based on the formation of the rock
fractures (Cesano et al., 2003). The key parameters for tunnel water inflow prediction include
the mean and standard deviation of fracture orientation, the mean and standard deviation of
fracture spacing and the mean and standard deviation of fracture aperture. A heterogeneity
index which incorporated the fracture formation and was defined to quantify the
heterogeneity of the fracture rock mass in that study (Cesano et al., 2003). The data of the
statistical distribution of rock fracture were measured from a tunnel in the north-western
Italian Alps (Cesano et al., 2003).
A statistical method was developed to evaluate the water inflow condition in tunnel
(Vanarelli, 2008). The local hydraulic conductivity of rock mass in that study varies from
1×10-6 cm/s to 1×10-2 cm/s. The local hydraulic conductivity of the rock mass was
determined by conducting a packer test in bore holes. The water inflow rate in the entire
tunnel was reported to be 2670 gallon per minute (10107 litres per minute). However, the
method for measuring the water inflow rate of the entire tunnel was not given.
Water inflow problems of a railway tunnel in south-west of China has been observed (Long
et al., 2011). Water leakage in that study was found to occur through the joints in tunnel
lining. White precipitates were found on the tunnel wall where water flows through which
made the crack difficult to see directly. Precipitate samples were collected and tested by
scanning electron microscopy (SEM), X-ray diffractometry (XRD) and energy dispersive X-
ray spectroscopy (EDX). The chemical content and pH of water from site were collected and
tested in lab in that study. That study found that water inflow through joints could damage the
lining by leaching the calcium content out of the concrete. The finding in that study was
made based on the interpretation of the chemical content of solid precipitates and water
samples (Long et al., 2011).

20
The studies discussed in this section show that water inflow rate is an important factor for
tunnel condition assessment. The water inflow rate for a long tunnel section can be obtained
by measuring the pumping rate of the tunnel. However, the pumping rate cannot accurately
represent the water inflow rate of a local crack or joints. Measuring water inflow rate of a
local crack and joint can be difficult since water typically flows as a thin film along the
tunnel wall. No method for measuring water inflow rate of a local crack or joints has been
reported in detail base on the review.
The record of the water inflow rate of an urban tunnel in Barcelona was presented (Font-
Capó et al., 2011). The measured water inflow rate during tunnelling was plotted against the
identified geological conditions. The geological conditions in that study included weathered
granite, unaltered granite, dykes, alluvium and fault zones. That study predicted the potential
water inflow rate for the further tunnelling based on the geological formations (Font-Capó et
al., 2011). Detailed studies of water inflow calculation methods are further discussed in
Section 2.4.3.
Water inflow problem have been identified in a rock tunnels in Yucca mountain (Chen and
Tonon, 2012). That study identified the rock fractures is the main cause of water inflow
problem. Water flow into the tunnel was found to form a thin film flow along the tunnel wall
in that study. The parameters of the rock fracture include fracture orientation, sized
distribution and the spatial arrangement of fractures (Chen and Tonon, 2012). The rock
fractures intensity in a 100 m × 100 m domain was simulated using Monte Carlo simulation.
That study identified the number of fractures around the tunnel surface according to the
simulation.
21
An analytical study was developed to determine efficiency of the long term protection of the
tunnel lining (Shin et al., 2012). That study quantified the water inflow rate though the
sealing joint between lining segments. The assumption in that study was the lining joints are
not sufficient for the protection of the long term water leakage due to lining deterioration.
They also conducted a parametric study to quantify the water inflow rate variation caused by
different parameters in that study (Shin et al., 2012).
A guideline for tunnel design described water ingress into a tunnel by quantitatively
categorising different water inflow conditions (Maidl et al., 2014). In that guideline, sweating
and percolating were used as the terms to describe water seeping through a lining and water
inflow through cracks, respectively. The maximum acceptable water inflow rate of a tunnel
section from a cost point of view was discussed in that guideline (Maidl et al., 2014). That
guideline showed that the maximum acceptable water ingress rate for a short-term up to 4
hours is 12 L/min per metre tunnel length. The maximum acceptable water ingress rate for a
long-term over to 4 hours is 1.2 L/min per metre tunnel length in the top heading and 2.4
L/min for the bench and invert.
Research gap
It can be seen from the aforementioned studies in this section that water inflow rate was
considered to be an important issue for tunnel service condition assessment. The tunnel
condition was defined as “unacceptable” when the water inflow rate of a tunnel section
exceeded a limit. The water inflow rate of a tunnel is either measured on site, or calculated.
The water inflow rate in an entire tunnel can be quantified by measuring the pumping rate.
However, the pumping rate does not tell the local water inflow rate from cracks or joints. The
local water inflow rate through cracks is an important parameter for tunnel lining
22
maintenance. None of the aforementioned studies have provided a method for measuring the
local water inflow rate measurement on site in a tunnel. Therefore, local water inflow rate
measurement has been identified as a research gap. A method for local water inflow rate
measurement on site needs to be developed.
This section only aims to elaborate the existing water inflow problem in different tunnels.
The methodology for water inflow calculation is further discussed in Section 2.4.
2.2.4 Tunnel maintenance methods
The serviceability of a tunnel needs to be maintained to achieve the service requirement when
a tunnel is deteriorating over time. It has been reported that a large part of tunnel maintenance
is to deal with water inflow according to the report presented by International Tunnel
Association (ITA, 1988, ITA, 1991). The tunnel maintenance to control water inflow into a
tunnel is reviewed here. Four methods of water inflow treatment have been discussed, namely
drainage method, injection method, lining replacement method and patching method (BTS,
2004, Maidl et al., 2014).
 Drainage
The principle of the drainage method is to create a highly permeable zone behind or within
the tunnel lining (O'Rourke, 1984). The highly permeable zone would cause local
controllable high water flow and reduce the hydraulic pressure on the other parts of the lining.
The drainage method is often applied in a tunnel where the hydraulic pressure is intense
(Hudson and Feng, 2015). It is necessary to estimate how much water is expected when
designing the drainage of a tunnel (Kuesel et al., 2012). The key parameters of water draining
from a tunnel was found to be the drain hole length, drain hole spacing and the hydraulic
pressure around the tunnel (Shin et al., 2009).

23
An analytical study found that increasing the drain hole length will reduce the surrounding
water pressure of the tunnel (Shin et al., 2009). In that study, the optimum length and
longitudinal spacing of the drain hole are calculated to be 0.4 times the tunnel diameter. Four
drain holes around the tunnel perimeter is recommended to reduce the water table to a
controllable level (Shin et al., 2009).
The main concern for draining a tunnel is lowering the water table (BTS, 2004). The
lowering of water table could change the local hydrology environment and cause the surface
structures subsidence. A subsidence case caused by groundwater lowering was reported in
Bangkok, Thailand (Phien-Wej et al., 2006). The groundwater has been lowered as much as
70 m, and the subsidence rate is 30 mm per year (Phien-Wej et al., 2006).
High hydraulic pressure could cause high water inflow rates. The Burnley tunnel is a road
tunnel under 60 m of water pressure across Yarra river in Melbourne, Australia. Excessive
water inflow in Burnley tunnel up to 5 litres per second has been reported (ABC, 2001). It
was identified that the excessive water pressure also caused deformation of the tunnel lining
(Mothersille and Littlejohn, 2012). In order to mitigate the high water inflow rate, the
drainage method was applied in the Burnley tunnel. Approximately 1.04 million litres per day
of drinking water is then pumped into the aquifer above the tunnel to retain the water table
(Baker, 2001).
 Injection
The principle of the injection method is to reduce the local hydraulic conductivity of tunnel
lining or rock mass (O'Rourke, 1984). The voids and cracks of the tunnel lining and the
surrounding rock mass are filled to mitigate water inflow. The hydraulic pressure on the
24
water sealing lining would be much higher compared to that on the water draining lining
(Maidl et al., 2014).
There are three general steps before the actual polymer injection (Kuesel et al., 2012). Firstly,
the injection holes are drilled diagonally from the sides of the crack to intersect the crack
within the lining. The intercept depth of injection holes and crack depends on many
parameters including the crack width, concrete thickness and temperature. A typical drilling
depth is that required for the injection holes to intersect the crack in the middle of the lining
structure (WEBAC, 2017a). Secondly, a mechanical packer is installed near the opening of
the hole to provide a tight connection for injection. Thirdly, the surface of the crack is sealed
with gel before conducting the injection. The injection pressure needs to exceed the estimated
hydrostatic pressure for water inflow control treatment (BTS, 2004). A maximum injection
pressure for a concrete lining is 100 psi (689.5 kPa) (Kuesel et al., 2012). The recommended
working injection pressure is 40 psi (275.8 kPa) (Kuesel et al., 2012).
There are four major factors would affect the performance of the polymer injection process,
namely, strength of the repair material, durability of the repair material, environmental effects
and the time frame for repair (Kuesel et al., 2012).
The strength of the repair material is important when the sealant is expected to take loading.
The compressive strength, tensile strength and shear strength of 14 types of concrete crack
sealants were tested in lab (Zanzotto, 1996). It shows that the mechanical properties vary by
sealant types. The selection of sealant is critical when dealing with a case with structural
cracks.
A study for Botlek and Sophia rail tunnel showed that the injection around the tunnel lining
would change the local environment where the tunnel is located (Bezuijen and Talmon, 2004).
25
The ground injection will affect the loading on the tunnel lining during a TBM process
(Bezuijen and Talmon, 2004). For the condition when the tunnel is constructed in sand, the
loading and pore pressure is independent from the injection strategy due to the high
permeability of the sand material (Bezuijen and Talmon, 2004).
 Lining replacement
A lining could be heavily cracked and make the injection method uneconomic in an
engineering case. The heavily cracked or damaged lining usually has a much higher hydraulic
conductivity compare to the intact lining. Replacement of the damaged lining can decrease
the hydraulic conductivity and decrease the water inflow rate.
The main factor for the lining replacement method is to identify the lining thickness which
needs to be replaced. A study for a masonry railway tunnel in Frankfurt, Germany found the
tunnel lining was heavily weathered and damaged (ITA, 1991). Water inflow had been
reported from a large area of the lining due to intensive cracking. The lining was cut by a
thickness from 10 to 45 cm depending on the weathering condition of the lining. A new
concrete lining was then cast to replace the weathered lining sections to mitigate the water
seepage and to avoid the potential problem of spalling (ITA, 1991).
 Patching
The patching method is to apply material onto the crack surface to block the water flow.
Cementitious material is commonly used as the patching material for cracks based on a
review study (Morgan, 1996). The hydraulic conductivity of the leaking location would be
decreased by the patching process.
The patching method was applied to a water seeping problem in a road tunnel between
Goilling and Werfen in Austria (ITA, 1991). Water was reported to be seeping through cracks
26
and joints of the lining. Leaks were repaired with plaster and paint coat patching, but the
results were not satisfactory. Water continued to seep after the patching treatment until
injection works three months later (ITA, 1991).
A summary table for the methods of tunnel maintenance for water ingress is shown as
follows.
Table 2.2 Methods for tunnel maintenance for water ingress and the correspondent
mechanisms
Methods Mechanisms
Drainage Reduce hydraulic pressure
Injection Reduce lining hydraulic conductivity
Lining replacement Reduce lining hydraulic conductivity
Patching Reduce lining hydraulic conductivity
Research gap
Water inflow treatment of tunnels is considered to be an important aspect of tunnel
remediation as discussed in the literature. Multiple water treatment methods have been
described and discussed. A research gap has been identified that the effectiveness for water
inflow treatment of a tunnel has not been studied and justified according to the reviewed
literature. A study is conducted to analyse the global water inflow condition in an selected
tunnel. The study will illustrate the treatment effectiveness based on site data provided by the
tunnel operator and supported by eight site visits by the author of this study. A strategy for
site inspection and maintenance of the selected tunnel is given in the later chapters.
2.3 Concrete properties
Concrete is a commonly used construction material which has been in use for centuries
(Neville and Brooks, 1987). The properties of concrete, including the concrete composition
the strength of concrete and the permeability of concrete are discussed as follows.
27
2.3.1 Concrete composition
Concrete is fundamentally a mixture of cement, water and aggregate (Neville and Brooks,
1987). The mechanical property of a hardened concrete is provided by the bonding effect of
the cement paste during the hydration process (Neville and Brooks, 1987).
Aggregate
It has been observed that aggregate occupied at least three-quarters of the volume of the
concrete (Neville, 2011). The aggregate is known to increase the concrete stability and
decrease the volume cost of concrete. The aggregates have been classified by size into two
general groups, which are fine aggregate and coarse aggregate (Neville, 2011) The size of
fine aggregate usually not larger than 4 mm while the size of coarse aggregate usually lager
than 5 mm. The shape of aggregate is also a factor for consideration. A classification method
defined the shape of aggregate into 5 categorises as in standard BS EN 933-4:2008. The five
categories defined the roughness of the aggregate which include: well-rounded, rounded,
subrounded, subangular and angular. The bond between aggregate and cement is an important
factor in concrete strength. The bond is partially due to the interlocking of the hydrated
cemented paste and the aggregate as discussed by Neville (2011). The nature of the hydration
product of cement is an important factor of the bonding which is discussed below.
The permeability of concrete primarily depends on the water cement ratio and is not
sensitive to the aggregate in concrete, as discussed by Neville (2011). The aggregate in
concrete could slightly increase the local permeability by a creating interfacial transition zone
around aggregate which be discussed in later sections.
Cement
28
Cement is a material which consists mainly of lime, silica, alumina and iron oxide (Neville,
2011). The aggregate in concrete generally does not have chemical reaction with cement
paste during the hydration process (Neville and Brooks, 1987). The contact zone of the
cement paste and the aggregate, which is also known as interfacial transition zone was found
to have a different micro structure compared to the cement paste (Ollivier et al., 1995). That
study compared the porosity between the interfacial transition zone and the cement paste
using scanning electron microscopy. The study found that the porosity of the interfacial
transition zone could be 20% to 40% higher compare to the hardened cement paste. Image
processing was used as a tool to quantify the porosity increase caused by the interfacial
transition zone (Diamond and Huang, 2001). The result showed that the interfacial transition
zone can increase the local porosity of the concrete to as much as 50% within 5 m from the
aggregate (Diamond and Huang, 2001). The thickness of interfacial transition zone was found
to vary in a range from 9 to 51 m according to a theoretical study (Zheng et al., 2005) The
parameters used in the calculation include water cement ratio, maximum cement diameter
and this distance from the aggregate surface.
Mixture of water and cement
Water cement ratio is a key parameter to quantify the properties of hardened cement paste
and concrete (Neville, 2011). Chemical reaction will take place when mixing cement powder
and water. That chemical reaction is also known as the hydration process of cement paste
(Neville, 2011). The hydration process of cement paste creates hydration products and binds
the hydrated cement together. The strength of concrete mainly comes from the binding of the
hydrated cement. The hydration products of cement have been studied experimentally using
cement paste with different water cement ratios (Powers et al., 1954). That study defined two
types of pores in concrete namely gel pores and capillary pores. The gel pores were defined
29
as the pores that formed in the cement hydration product. The capillary pores were defined as
the space that were not filled by gel and other material (Powers et al., 1954). The micro-
structure and composition of the cement paste was investigated using X-ray (Powers, 1958).
Electron-optic and electron-diffraction techniques were used to study the micro structure of
cement paste under different hydration stages (Powers, 1958). That study found that the main
hydration product of the hardened cement paste is the crystalized calcium hydroxide
(Ca(OH)2) and the calcium silicate hydrates. The calcium silicate hydrates were defined as
cement gel (3CaO·2SiO2·3H2O) which is also known as C-S-H (Powers, 1958).
Water cement ratio is a determining factor for the permeability of both hardened cement paste
and concrete (Neville, 2011). Higher water cement ratio would increase the porosity of the
concrete and increase the permeability significantly.
The permeability of the cement paste at different hydration stages was studied by conducting
a laboratory test (Powers et al., 1954). The porosity of cement paste at different hydration
stage was obtained using electron-optic and electron-diffraction techniques. That study found
that the fresh cement paste tends to have a higher capillary porosity compared with the
matured cement paste. The permeability of the fresh cement paste was found to be several
magnitudes higher than that of the matured cement paste (Powers et al., 1954).
An experimental study was developed to study the hydraulic conductivity variation of cement
paste during the hydration process (Hughes, 1985). Ordinary Portland cement paste samples
with 0.47 water cement ratio were mixed and cured for 1 to 12 weeks before the permeability
test (Hughes, 1985). The permeability test found that changing of permeability of cement
paste can be calculated by the capillary pore size changes using Poiseuille’s formula (Hughes,
1985). The mechanism of porosity decreasing with cement paste hydration was studied
30
experimentally (Mehta, 1986). That study used electronic microscope to investigate the pore
sized and composition changes through cement paste hydration. That study found that the
unhydrated cement combines with water and causes the volume of the gel (C-S-H) to expand
and occupy a part of the original pores, and then cause significant decrease of the capillary
porosity (Mehta, 1986).
The main hydration product of cement paste is C-S-H and crystalized calcium hydroxide as
aforementioned. The phenomenon of water seeping through concrete was evaluated in two
phases by a theoretical study (Bentz and Garboczi, 1991). In that study, phase one was
defined as water seeping through the capillary pores, and phase two was defined as water
seeping through the hydration product. That study quantified the hydraulic conductivity of
both phrases based on the pore structures (Bentz and Garboczi, 1991)
The capillary pore changes were quantified by according to hydration process of the cement
paste by a theoretical study (Hansen, 1986). The variation of the capillary pore fraction ( ),
hydrated cement fraction ( ) and unhydrated cement fraction ( ) through hydration
was derived based on the empirical relationship between cement porosity and water cement
ratio as follows (Hansen, 1986).
⁄ (2.1)
⁄
(2.2)
⁄
(2.3)
⁄
where, is the hydration degree. w/c is the water cement ratio.
31
The permeability of a general porous media based on pore structure was studied by a
theoretical approach (Katz and Thompson, 1986). That study defined the pores which form
the first connected pore pathway through the material as the critical pores (Katz and
Thompson, 1986). That study found that the diameter of the critical pores has a major effect
to the permeability of a porous material as shown as follows.
(2.4)
where, is the permeability of a porous material (m2), is the critical pore diameter (m)
and V is the critical volume fraction of phase capillary pore.
The critical volume fraction of the capillary pore, which is a parameter in the analytical
solution of Katz and Thompson (1986) (Eq. (2.4) was calculated numerically (Bentz and
Garboczi, 1992). That study found the critical volume fraction does not exceed 18% when the
water cement ratio varies from 0.35 to 0.55 even under calcium leaching conditions. The
results of that study were further validated by experimental data in literature (Bentz and
Garboczi, 1992).
The permeability of the pore structure of ordinary Portland cement paste was also studied by
experiments (Cui and Cahyadi, 2001). Cement paste samples of various hydration degree (7
days, 35 days and 210 days) and different water cement ratio (0.3 and 0.4) were tested under
constant hydraulic pressure (Cui and Cahyadi, 2001). The experimental result of hardened
cement paste permeability showed a good agreement with the prediction of Katz and
Thompson (1986) (Eq. (2.4)).
An analytical solution was developed for calculating the hydraulic conductivity of cement
paste (Zheng and Zhou, 2008). That study was based on the behaviour of the capillary pores
32
of the cement paste during the hydration stage. Previously published experimental study data
was used to verify the analytical model and it was found to show a good agreement (Zheng
and Zhou, 2008).
2.3.2 Concrete strength
Strength is considered to be an important property for concrete (Neville, 2011). In practice,
the water cement ratio and the degree of compaction are considered to be the two primary
factors for concrete strength (Gilkey, 1961). The strength of concrete is often inversely
proportional to the water cement ratio (Neville, 2011). The ratio of cement aggregate and the
aggregate physical property were also found to contribute to the concrete strength (Gilkey,
1961). A study was developed for the relationship between strength of hardened cement paste
and cement water ratio (Nielsen, 1993). That study related the strength of the hardened
cement paste with the porosity of the hardened cement paste. It was found that the ratio of
cement and water have a linear relationship with strength of the hardened cement paste when
the ratio of cement and water is under 2.6 (equivalent to water cement ratio more than 0.38).
The rate of increasing in the strength of hardened cement paste starts to slow down when the
ratio between cement and water exceeds 2.6 (equivalent to water cement ratio less than 0.38)
(Nielsen, 1993). The strength of concrete are related to the concrete porosity and then related
to the permeability of concrete which is further discussed as in the following sections.
2.3.3 Permeability of concrete
Mortar is a mixture of cement, water and sand while concrete is a mixture of cement, water
and aggregates. The aggregate in both mortar and concrete include coarse aggregate and fine
aggregate. The permeability of cement mortar was tested experimentally (Reinhardt and
Gaber, 1990). That study was conducted to develop an equivalent pore size of the cement
mortar based on the pore size distribution. The maximum aggregate size used in the mortar
33
was 4 mm based on the Germany standard PZ-35-F. The water cement ratio in that study
varied between 0.4 to 0.75 and two curing conditions were applied. The pore size distribution
was developed using mercury intrusion porosimetry in that study (Reinhardt and Gaber,
1990). The water permeability set-up used in that experimental is shown in Figure 2.1. That
study found that the pore size in a hardened cement mortar can be made uni-sized for
calculating the equivalent pore radius. That study also found that the equivalent pore radius is
not constant and depends on the curing condition during the cement hardening process
(Reinhardt and Gaber, 1990).
Figure 2.1 Water permeability testing device (Reinhardt and Gaber, 1990)
The permeability of concrete was investigated for the protection of radioactive waste (Jacobs
and Wittmann, 1992). In that study, 28 year old concrete was tested using a permeability test
set-up for the permeability and porosity. The permeability and porosity of 91 days old
concrete was also test for a comparison. The water cement ratio of the samples in that study
varied from 0.45 to 0.80. The aggregate content varied from 75% weight to 81% weight of
the specimen. The permeability of concrete specimen in that study varied from1×10-19 to
34
1×10-16 m2. The result in that study showed that the continuing hydration process after 91
days does not affect the properties of concrete (Jacobs and Wittmann, 1992).
The effect of aggregates on the hydraulic conductivity of concrete was studied by conducting
a review of the permeability testing of concrete and cement paste (Breysse and Gérard, 1997).
The experimental data of concrete hydraulic conductivity was plotted against the concrete
porosity in that study. That study developed an empirical relation between the concrete
porosity (p) and hydraulic conductivity K (m/s) based on the reviewed experimental data
(Reinhardt and Gaber, 1990, Jacobs and Wittmann, 1992) as follows (Breysse and Gérard,
1997).
(2.5)
( )
It is notable that the hydraulic conductivity and the permeability are two traversable concepts.
The permeability is the property of a material while the hydraulic conductivity also related
with the viscosity and density of water. The detail descriptions are discussed in Section 2.4.1.
The permeability of concrete was tested experimentally after the concrete sample was
compressed under axial stress (Picandet et al., 2001). The permeability was test using Darcy’s
law in that study. The concrete permeability reflected the conditions of the pores of concrete.
That study experimentally investigated three types of concrete mixes, namely an ordinary
concrete with 0.5 water cement ratio and two of high-performance concretes with 0.3 water
cement ratio. All those concrete specimens were tested under compressive axial stress before
conducting a permeability test. The stress level in that study varied between 60% and 90% of
the ultimate strength of the concrete. That study found that the axial loading induced
damaged would increase the concrete permeability even after unloading. The mechanical
damage of concrete was found to be the main cause for the permeability rise in that study.
35
The result shows that the permeability of damaged samples would increase approximately
one order magnitude at 90% of the ultimate strength compared to an undamaged sample
(Picandet et al., 2001).
An experimental study was carried out to test the permeability of different types of concrete
(Leemann et al., 2006). That study investigated the interfacial transition zone between cement
and aggregate in self-compacting concrete and conventionally vibrated concrete. That
experimental study found the local porosity of the interfacial transition zone would increase
local porosity from 10% to 25%. The result of that study shows that the permeability of
interfacial transition zone in self-compacting concrete is approximately 10% lower than the
conventionally vibrated concrete (Leemann et al., 2006).
An experimental study of concrete permeability was conducted to investigate the effects
caused by stress and temperature (Choinska et al., 2007). The permeability of concrete
specimens was tested under high temperature from 80 to 150 . That test found that
cement pores would have a widening effect when the temperature is greater than 105 . The
micro-cracks caused by heating were found to be a major concern above 105 . The pore
widening effect was found to be much more significant when the temperature is greater than
150 (Choinska et al., 2007).
The permeability of the fibre reinforced concrete was studied experimentally (Lepech and Li,
2009). Cracked concrete was tested under a falling head permeability test in that study. That
study found that the crack was blocked after 70 days of testing. It was found that hydraulic
conductivity of the concrete specimen decreased by approximately two order magnitudes
compared to the hydraulic conductivity at the beginning of the test. X-ray diffraction test was
36
conducted in that study and found that the material which blocks the cracks is calcium
carbonate (Lepech and Li, 2009).
The porosity of concrete was studied under compressive stress by experiment (Lian and
Zhuge, 2010). The uniaxial compressive strength test showed that the majority of failure
happened between the cement paste and aggregate, which was found to be caused by the
interfacial transition zone (Lian and Zhuge, 2010). An empirical study using experimental
data was developed to study the relation between the effective concrete porosity ( ) and the
compressive strength ( ) MPa (Lian et al., 2011). Experimental data from literature were
used to statistically analyse the relationship between the concrete compressive stress and the
concrete porosity. That study found a solution which showed a good correlation with the
experimental data, and the equation was developed as follows (Lian et al., 2011).
(2.6)
2.3.4 Factors affecting water ingress in concrete
The water ingress depth of a material was developed analytically based on Darcy’s law
(Valenta, 1969). Concrete was used as the permeable material in that study. That study aimed
to investigate the water ingress depth of concrete in order to understand the initiation of
reinforcement corrosion in concrete. The assumptions used in that study were constant
hydraulic pressure and homogeneous concrete hydraulic conductivity. The factors which
would affect water ingress depth including the hydraulic head, hydraulic conductivity of
concrete and time. That study described the water ingress depth (x) by hydraulic head,
hydraulic conductivity of concrete and time as follows (Valenta, 1969).
37
(2.7)
where, K is the hydraulic conductivity of the material where water seeping through (m/s), x is
the water seeping depth (m), v is the water filled pore volume fraction during the test (m3/m3),
H is the pressure head (m), t is time (s).
The permeability of concrete in section 2.3.3 and 2.3.4 has been reviewed and discussed. The
permeability of concrete considers the pore structures and the bulk permeability of concrete.
The review provided in this case only considers the concrete as a bulk material. Cracks in
concrete for permeability are not considered in this section. The properties of concrete and
cracked concrete are very different in terms of permeability. Therefore, sound concrete and
cracked concrete are considered as two different materials in terms of permeability. For the
permeability of sound concrete, no research gaps have been identified. The permeability of
cracked concrete is discussed in Section 2.4.2.
2.4 Engineering hydraulics

Water travel in a system is a phenomenon that has been studied for centuries. One of the
earliest recorded studies was based on the description of mass conservation by Mikhail
Lomonosov in the middle 18th century (Lomonosov, 1756), which confirmed that the mass
of water in one system must be conserved. Another important conservation law for water
flow is the Navier-Stokes momentum equation (Navier, 1823, Stokes, 1846) which was
developed by and named after Claude-Louis Navier and Sir George Stokes. This equation
describes the conservation of momentum of a fluid in motion.
2.4.1 Basics of hydraulics
The early explanation of water movement in nature was developed by Daniel Bernoulli and
published in his book Hydrodynamica in 1738 (Bernoulli, 1738). The Bernoulli equation was
38
derived based on the principle of energy conservation. He found that in a steady,
incompressible flow, the summation of kinetic energy, potential energy and internal energy
remains in the system as a constant. This relationship is known as the Bernoulli equation as
follows.
(2.8)
where, z is the elevation head based on datum level, is the pressure head of water, is the
velocity head. The elevation head plus pressure head was defined as total head.
Water ingress into a cylindrical tube was studied experimentally by Jean Léonard Marie
Poiseuille and Gotthilf Heinrich Ludwig Hagen in early 19th century. They independently
investigated the water inflow mechanism through hollow cylindrical tubes with diameters
from 0.015 mm to 0.6 mm. The cylindrical tubes were placed under constant hydraulic
pressure, and the water flow rate out of each tube was measured and recorded in the
experiment. Their experimental works were further compared and described as the Hagen-
Posseuille equation as follows (Schiller et al., 1933).
(2.9)
where, Qp is the water flow rate (m3/s) through one single tube, ac is the radius of the tube
(m), µ is the viscosity of water (Pa.s) and ∇ is the hydraulic gradient.
The quantitative method for water flow through a porous material was undefined until studied
by Henry Darcy in 1856 (Darcy, 1856). Darcy studied the water ingress process by
conducting experiments on water seeping through a sand column as shown in Figure 2.2. The
sample sand column was held in a steel pipe with an inside diameter 0.35 m. The total height
39
of the column was 3.5 m as reported in the figure. The water inlet is from the left side of the
sand column. U-tube manometers were mounted on the right side of the sand column in order
to represent the water head at each location of the sand column. Water flow rate from the
water outlet was measured after 10 to 25 minutes when water flow rate through the sand
column is stabilized. Water flow from the outlet was collected by a tank and then the total
volume of water was measured over time. The measured water flow rate varied from 2.13 to
29.4 L/min while the head loss through the sand column varies from 1.11 m to 13.93 m.
Based on the measured water inflow rate, he proposed one of the most fundamental equations
in hydraulics which is known as Darcy’s Law.
(2.10)
where, Qd is the water flow rate (m3/s), K is the equivalent permeability coefficient or
hydraulic conductivity of a material (m/s), A is the area perpendicular to the water flow
direction (m2), hydraulic head difference ( ) over distance ( ) is also known as the
hydraulic gradient ( ).
40
Figure 2.2 Experimental set-up of water inflow through sand column (Darcy, 1856)
2.4.2 Mechanisms of water movement in concrete
The mechanism of water movement through a material was defined as water moving through
the interconnected capillary pores of a material (Buckingham, 1907). The unsaturated
condition of a soil was studied based on Darcy law in an experimental study (Buckingham,
1907). Unsaturated soil with different moisture content was used to investigate the water
movement through the capillary pores of the soil. Experimental work was conducted to
further investigate the moisture movement in material under unsaturated conditions (Richards,
1931). The solution of water movement in capillary pores of an unsaturated model was
41
derived based on Darcy’s law in that study. A governing equation for moisture movement
through a material was developed as follows (Richards, 1931).
(2.11)
where, K is the hydraulic conductivity (m/s), h is matric head which is caused by capillary
suction (m), z is the elevation head (m), is the water content, t is time (s).
The parameter of hydraulic conductivity in Darcy’s law represents the ability of passing
liquid through a material. The value is affected by both material properties and liquid
properties. The intrinsic permeability, as a material property, of a porous media was
investigated based on Darcy’s law (Muskat and Wyckoff, 1937). The hydraulic conductivity
was derived as two parts, intrinsic permeability of a material and the properties of liquid. The
intrinsic permeability of a material was found to be purely a function of material
microstructure. The liquid properties include the density of the liquid and the viscosity of the
liquid. The relation between hydraulic conductivity and intrinsic permeability was shown as
follows (Muskat and Wyckoff, 1937).
(2.12)
where, k is the intrinsic permeability of a material (m2), is the density of liquid (kg/m3) and
is the viscosity of liquid (Pa.s).
The theory of Darcy’s law is used widely to quantify water flow through a material. The
permeability of a material can be either high (like soil) or low (like concrete). The
fundamental mechanism is the same which is water travels through the interconnected
channels within the material. The application of Darcy’s law has been found to always hold
for the problem of water flow through a material.

42
2.4.3 Hydraulic conductivity of cracked material
The water flow rate through cracks was studied analytically based on the Navier-Stokes
equation and mass conservation (Snow, 1965). The condition in that study was water flow in
between two parallel plates. The boundary condition in that study was the water flow has zero
velocity at the water-plate interface. Water inflow through the aperture between two plates
(representing a crack) is defined to have a constant hydraulic head over the crack. The two
plates are considered as an impermeable material. That study found that the aperture between
two plates has a cubic relation with the water inflow rate (Figure 2.3). That study was used in
the further studies for the problems include water inflow through fractured rock mass and
water inflow through fractured concrete. That approach is also known as the water flow cubic
law as follows (Snow, 1965).
(2.13)
where, ∇ is the hydraulic gradient, as defined in Section 2.4.1, and are
the crack aperture (m) and length perpendicular to the water flow direction (m), and are
water density (kg/m3) and viscosity (Pa.s), respectively.
43
Figure 2.3 Water flow through an impermeable material with one aperture
The cubic law of water flow was applied for water inflow calculations for fractured rock
mass (Louis, 1969). That study investigated the effect of water inflow on rock mass stability.
Water inflow rate between the joints of a rock mass was calculated analytically using the
cubic law in that study. That study assumed the joints of the rock mass to act as two parallel
plates which is the same as in Snow’s study (Louis, 1969).
It was verified experimentally that the cubic law is applicable for water flow through rock
masses containing cracks (Witherspoon et al., 1980). The experiment investigated the water
inflow rate through fractured rock with crack width varying from 4 µm to 250 µm. Different
rock types including granite, basalt and marble were used in the permeability test in that
study. Rectangular blocks cut from the rock sample with width = 0.121 m, length = 0.207 and
height = 0.155 m were used in the permeability test. The cracks in the rock sample were
created under tensional stress. The created width of the cracks was found to vary from 0.2 µm
to 1540 µm. The study found water inflow rate through rock fractures has a cubic relation
with the crack aperture and therefore the cubic flow law holds (Witherspoon et al., 1980).
44
The theory of water flow through fractured rock mass was developed based on the fracture
formation within a rock mass (Long et al., 1982). A two-dimensional model for a fractured
rock mass system was developed to compare with the condition when the rock mass is
considered as a porous material. That study found that when the fracture density is high, the
fracture aperture is relatively constant and the fracture orientation is distributed the fractured
rock mass behaves like a porous material (Long et al., 1982).
The hydraulic conductivity of concrete was tested under a tensile stress up to 3.5 MPa
(Gédrard et al., 1996). The dimensions of the specimens were prepared as a concrete slab
with length of 160 mm, width of 110 mm and thickness of 40 mm. The cracks in the concrete
were formed by tensile stress. The relationship between the hydraulic conductivity of cracked
concrete and tensile stress was studied. That study found that the tensile stress would increase
the hydraulic conductivity of concrete specimen by increasing the crack aperture. The result
showed that the concrete hydraulic conductivity increased for three order magnitude when the
strain of the specimen increased to 0.0013 mm (Gédrard et al., 1996).
A review study was developed to study the hydraulic conductivity of material with multiple
cracks (Reinhardt, 1997). That study used a three-dimensional approach to represent the
hydraulic conductivity of a material with given crack width and crack length. The hydraulic
conductivity of bulk cracked media was studied based on the cubic flow law in that study.
That study presented a bulk hydraulic conductivity of cracked concrete consider crack width
and crack density as Eq. (2.14) (Reinhardt, 1997). The illustration is referring to Figure 2.3.
(2.14)
45
where, K (m/s) is the bulk hydraulic conductivity of the cracked concrete, (m) is the
crack aperture (width), is the crack density (m/m2) which equal to accumulated crack length
perpendicular to the water flow direction over the concrete area, is the crack roughness
reduction factor, (kg/m3) and (Pa.s) are water density and viscosity, respectively.
A falling head hydraulic conductivity test was carried out using cylindrical concrete
specimens (Wang et al., 1997). That study used the Brazilian splitting method (compression
across the diameter) to create a through crack in the specimen. That study showed that the
size of a crack controls the increasing rate of the hydraulic conductivity. When the crack
width is less than 50 µm, the crack does not have great impact on hydraulic conductivity.
When crack width increased from 50 to 200 µm, the hydraulic conductivity increased rapidly.
When crack width was beyond 200 µm, the permeability increased less rapidly compared
with the 50 to 200 µm condition (Wang et al., 1997).
An experiment was conducted to study the water and chloride permeability of different types
of concrete with cracks (Aldea et al., 1999a). The study used three types of concrete
specimen with different water cement ratio from 0.45 to 0.25. The specimens were placed
under a tensile test up to 4 MPa to crack the specimens. The crack width was measured after
the tensile test in that study. The result of that study showed that the permeability increased
approximately one order magnitude when the crack opened less than 200 µm. The
permeability increased significantly when the crack opening was more than 200 µm (Aldea et
al., 1999a).
It was found by a concrete permeability test that the permeability of cracked and uncracked
concrete depends on different parameters (Aldea et al., 1999b). Four types of material were
studied in the test, namely, paste, mortar, normal strength concrete and high strength concrete.
46
Nine different cracking conditions were assigned to the specimens from sound specimen to a
specimen with a 350 µm wide crack. The experimental results showed that permeability of
the cracked specimen purely depends on the crack width, while the permeability of the
uncracked concrete depends on the mortar type and the curing condition (Aldea et al., 1999b).
The healing effect of the cracked concrete was quantified by testing water flow through
concrete with different crack widths (Edvardsen, 1999). The changes of crack width over
time caused by accumulation of precipitates in the cracks were interpreted by the cubic law in
that study. The result showed that the hydraulic conductivity of the cracked concrete started
to become stable after approximately 800 water flow hours (Edvardsen, 1999). That study
found that the roughness factor, which was used to predict the concrete water inflow rate,
varied from 0.02 to 0.17 (Edvardsen, 1999).
The permeability of the cracked concrete under different temperature and healing stages was
studied experimentally by water permeability testing (Reinhardt and Jooss, 2003). That study
tested the water inflow rate of concrete specimens under varies temperature and different
crack width. That test validated the application of the cubic law for the cracked concrete
under various temperatures from 20°C to 80°C. That test also showed that high temperature
will accelerate the crack healing process, and therefore cause a more rapid decrease in water
inflow rate. Meanwhile, the high temperature was found to increase the final hydraulic
conductivity of the specimens (Reinhardt and Jooss, 2003). The healing effect is less rapid at
room temperature compared to the condition under high temperature.
The process of water transport in cracked concrete was measured three-dimensionally
(Carmeliet et al., 2004). In that study, X-ray computer tomography was used to determine the
interior structure of a crack. That study found that the stochastic nature of cracks is the major
47
factor affecting water travel within a concrete. The flow network in a concrete was measured
and showed that the two dimensional study can greatly underestimate the water seeping of
cracked concrete (Carmeliet et al., 2004).
The effect of crack roughness is a function of water flow rate as discussed in Eq. (2.14). An
experimental study was developed to quantify the effect of the crack roughness and tortuosity
to water inflow rate through cracked concrete (Akhavan et al., 2012). An artificially cracked
concrete specimen was used to study the water inflow rate through the crack. The calculated
hydraulic conductivity was compared with the measured value from experiments. That study
found that a rough crack will reduce the water inflow by 4 to 6 times compared to a smooth
crack (Akhavan et al., 2012).
The hydraulic conductivity of rock mass was discussed in a review study (Zhang, 2013). The
hydraulic conductivity of a sound permeable rock mass was calculated based on the rock
porosity, which is often very low, and depth as discussed in that study. The hydraulic
conductivity of several typical rock types, including, sandstone, shale, clay and volcanic
rocks were given in that study (Zhang, 2013). That study found that the hydraulic
conductivity of rock mass is mainly contributed by the discontinuities (including joints). The
hydraulic conductivity of a fractured rock mass was determined based on the fracture
aperture and density as aforementioned in this section. The depth of rock mass would cause
the fracture closure due to increased stress and therefore decrease the hydraulic conductivity
of the rock mass as discussed in that review (Zhang, 2013).
The underground structure and water pressure would affect the hydraulic conductivity of rock
mass (Huang et al., 2016). A site based experiment on the hydraulic conductivity of rock
mass was conducted in a coal mine in east China. The initial hydraulic conductivity of the
48
rock mass was tested by a pumping test combined with the original borehole geology
information. That study identified hydraulic fracturing was caused by increasing water
injection pressure. The hydraulic conductivity of rock mass started to show a significant
increase when the injection pressure was 0.5 MPa for the most critical case (Huang et al.,
2016).
2.4.4 Homogeneity of cracked material
The permeability of multiple fractures in rock mass was studied analytically based on the
hydraulic conductivity of a single fracture (Oda, 1985). The hydraulic conductivity of a
fracture within the rock mass was considered by three hydraulic conductivity component
based on the fracture orientation. The permeability of the rock mass was expressed by the
tensor of fractures at different directions in that study. That study concluded that a rock mass
containing cracks can be treated as a homogeneous, anisotropic porous media when the rock
mass contain many cracks (Oda, 1985).
The effect of stress on permeability change in cracked material was studied experimentally
(Raven and Gale, 1985). That study tested the permeability of five granite cores with
different diameters under axial compressive stress up to 40 MPa. That study found that a
larger sample would result in increasing of the crack roughness and therefore reducing the
water flow rate. That study also found that stress can have a significant impact on the
hydraulic conductivity due to closure of cracks. That study found that the hydraulic
conductivity can be decreased by three orders of magnitude when the compressive stress
increased from 0 to 40 MPa (Raven and Gale, 1985).
It is notable from the aforementioned study that the permeability test is based on an implicit
assumption that the crack is fully interconnected through the material. In a typical
49
engineering case, a crack in a material may not be fully interconnected. A crack could
terminate within concrete due to various reasons. The bulk hydraulic conductivity of a
material when a crack is terminated within a material has not been studied.
Research gap
The common approach to evaluate the hydraulic conductivity of a cracked material is based
on an implicit assumption that the crack is fully interconnected through the material based on
reviewed literature. There is no method for determining the hydraulic conductivity of a
cracked material when the crack is partially interconnected through a material. This study
will provide a method for calculating the hydraulic conductivity of cracked concrete when the
crack is partially interconnected through concrete in the following chapters.
2.4.5 Water inflow into tunnel
Excessive water inflow rate would flood the tunnel or affect the serviceability of the tunnel as
established in Section 2.2. The quantification methods of water inflow rate for different types
of tunnels are reviewed in this section. A study of water inflow into a tunnel was conduct in
1960s as a subject of hydrogeology (Polubarinova-Kochina, 1963). Tunnels with various
hydraulics boundary conditions are discussed as follows.
Classical water inflow models of tunnel
An early approach to predict water inflow rate into tunnels was developed by using a mirror
image method (Polubarinova-Kochina, 1963). In that study, the tunnel was considered as a
sink, and the imaginary source was considered to be the symmetry of the sink against the
water table, water flow from the source to the sink (Figure 2.4).
50
Figure 2.4 Tunnel profile of classical method
Water flow into a tunnel was expressed as follows according to Darcy’s Law based on the
assumption of homogeneous and isotropic generic rock mass around the tunnel (Harr, 1962).
(2.15)
where, Q is the water inflow rate of the tunnel (m3/s), r is expressed by the Cartesian
coordinate in the plane (m), K is the hydraulic conductivity of the surrounding rock mass
(m/s), h is the hydraulic head (m).
The hydraulic head was expressed based on mass conservation of water as follows (Strack,
1989).
(2.16)
where,
√ (2.17)
where, xtc ytc are the centre coordinate of the tunnel.
51
The hydraulic head was studied equal to the superimposition of the total head of the tunnel
and the imaginary tunnel.
Two boundary conditions were defined in this two dimensional flow problem (Barenblatt,
1996). The boundary condition for a unlined transport tunnel was expressed as follows
(Polubarinova-Kochina, 1963).
(2.18)
(2.19)
That equation for water inflow into a tunnel is also known as Goodman’s equation as noted
by Lei (1999). This tunnel was constructed in the rock mass, no lining is present in this type
of tunnel.
(2.20)
√
( )
√
where, Q is the water flow rate per unit length of tunnel (m3/s), K is the hydraulic
conductivity of the rock mass (m/s), a is the tunnel radius (m), H is the distance between the
water table and the tunnel centreline when the water table is below the ground surface (m).
The datum elevation was defined to be the ground surface when the water head is above the
ground (ie. tunnel under river or tunnel under sea) (Moon, 2007). The H value was defined to
be the distance from tunnel centre to the ground surface under this condition (Moon, 2007).
The final solution was written as Eq. (2.21).
52
(2.21)
√
√
where, H1 is the water height that above the ground (it is a negative value, as H1 is on the
negative side of y axis), H is the distance from the tunnel centre line to the ground surface
(when the water table is above the ground surface).
A review study presented a simplified solution for water inflow of tunnel based on the
classical water inflow solution as Eq. (2.22) (Domenico and Schwartz, 1990). The parameters
in the simplified equation were the same as in Eq. (2.20).
(2.22)
Based on literature and site data review and comparisons, it was found that Goodman’s
equation tends to overestimate the total water inflow rate of tunnels (Heuer, 1995). An
empirical reduction factor was developed based on Goodman’s analytical solution and site
data which compromised different ground conditions and boundaries (Heuer, 1995).
The classical solution of tunnel water inflow was developed based on the assumptions when
the rock mass around the tunnel is (1) saturated, homogeneous, isotropic and semi-infinite
aquifer, (2) the flow is steady, (3) water is incompressible and (4) the water head at the
ground surface and tunnel perimeter are constant (Polubarinova-Kochina, 1963).
Three kinds of lining types of tunnels for transportation were discussed in a review study.
Namely, unlined tunnels, drained lined tunnels, lined water sealing tunnels (Butscher, 2012).
The solutions for predicting the water inflow rate for all the three kinds of tunnels are
discussed in the following sections.
53
Water inflow model of unlined tunnel
The classical solution was developed to solve the water inflow problem of the unlined tunnels
(Goodman et al., 1965). For unlined tunnels, the water inflow rate was found mainly
determined by the permeability coefficient of the rock mass and water table (Goodman et al.,
1965). Different solutions for water inflow into unlined tunnel are presented as follows.
A method was developed to study the water inflow of tunnel by considering the rock mass
hydraulic conductivity variation (Zhang and Franklin, 1993). In that study, the hydraulic
conductivity of the rock mass was presented as a function of the burial depth as follows.
(2.23)
where, A is a hydraulic conductivity gradient, which is a parameter that considered the stress
increasing rate of the rock mass and water pressure, K is the hydraulic conductivity (m/s)
which is a function of burial depth h (m), the unit weight of water is , the unit weight of
rock is , is a constant equal to .
Fourier series was used to solve the solution of governing equation and the boundary
conditions in that study. The solution of water inflow rate was given as follows (Zhang and
Franklin, 1993).
[ ( ) ] ( ) (2.24)
( )
where, , the other parameters were given as above.
54
It was discussed in this study that when the A value is equal to 0, the hydraulic gradient is 0
and the ground is homogeneous. The solution is the same as Goodman’s solution under this
specific condition (Zhang and Franklin, 1993).
An analytical study raised the point that the classical solution for water inflow is an
approximation solution and the buried depth could affect the final solution (Lei, 1999). An
analytical method was developed to calculate the water inflow rate as well as the relative
error caused by the burial depth of the tunnel (Lei, 1999).
That study was derived based on mass conservation and Darcy’s law. Two dimensional flow
was described by Laplace’s equation . The water total head at the origin point
was assumed equal to the water head above ground. The total head at the tunnel
circumference was calculated as the elevation head (Lei, 1999). The exact water inflow
solution was expressed as follows.
(2.25)
√( ) √( )
where, Q is the water flow rate (m3/s), K is the hydraulic conductivity of the surrounding rock
mass (m/s), d is the depth of water that above the ground (m), pa is the atmospheric difference
pressure head which is often 0, h0 is the total head (pressure head + elevation head) at the
circumferential boundary of the lining, hp is the pressure head at the spring line of the tunnel,
D and R are the depth that the tunnel below the ground surface (m) and the tunnel radius (m)
respectively.
55
This solution was found to be equivalent to Goodman’s solution when the tunnel depth D>>R
(tunnel radius) (Lei, 1999). The comparison in that study showed that the relative error is
about 5% when D/R=2, the error becomes more significant when D/R get smaller.
A exact analytical solution was developed for general cases of tunnel water inflow based on
Mobius-transformation and Fourier-series (El Tani, 2003). The assumptions in that study
were the same as in the classical solution as above. The Mobius-transformation of a circular
tunnel profile was studied (Verruijt and Booker, 2000). In that study, the tunnel and water
surface on z-plane was transferred onto ζ-plane, with two concentric circulars of radius 1 and
(Figure 2.5).
Figure 2.5 Conformal mapping of physical plane on circular ring
The equation was developed found as follows (El Tani, 1999).
(2.26)
( )
( )
56
where, Q is the water inflow rate (m3/s), r is the tunnel radius (m), h is the distance from the
centre line to the water table (m), K is the hydraulic conductivity of the rock mass (m/s).
It has been emphasised that Eq. (2.26) is only applicable when the water table is underground
(El Tani, 1999). When the water table is above the ground, the solution was given based on a
different boundary condition as follows.
(2.27)
√( )
where, H is the water depth above the ground (m), h is the distance from the ground surface
to the tunnel centre line (m), √ .
A semi-analytical solution was developed for analysing the tunnel water inflow rate (Hwang
and Lu, 2007). That study used a convolution–deconvolution method to calculate the water
inflow rate. This study provided a method which is able to incorporate the water inflow
variation over time during the tunnel excavation.
Water inflow models of lined drained tunnel
The previous section described water inflow calculation models for unlined tunnel. The
hydraulic boundary conditions for the unlined tunnel is less complicated compared to a lined
tunnel. The hydraulic head at the interface of rock mass and concrete for a lined tunnel is
often unknown.
Water inflow was found to be driven by the water head above the tunnel (Maidl et al., 2013).
The pressure head was found to decrease significantly when water goes into the drainage
layer. The assumption in evaluating the drained tunnel water inflow problem was the outer
part of the lining usually has a much higher permeability coefficient comparing with the
57
surrounding rock mass (Butscher, 2012). The inner lining is relatively intact and much less
permeable in a typical lined drained tunnel (Butscher, 2012).
Analytical methods for both approximation and exact solution for a drained tunnel was
developed (Kolymbas and Wagner, 2007). The assumptions in that solution were the same as
Goodman’s solution. The difference was Kolymbas and Wagner’s solution assumed an inner
circumferential pressure around the tunnel in the drainage layer.
The water head expression was written as Eq. (2.28) along the tunnel circumference.
(2.28)
A rigorous solution was given for a shallow tunnel (Kolymbas and Wagner, 2007). The
Mobius transformation has been used to transform the ground surface and the tunnel to two
concentric circular. The water inflow rate for both deep and shallow tunnels was solved
exactly as Eq. (2.29).
(2.29)
√( )
A study was developed to make a comparison of analytical solutions for a drained tunnel
under the water table (Park et al., 2008). That study compared the water inflow calculation
result by using the El Tani (2003) method and the Kolymbas and Wagner (2007) method.
That study found that the estimation method is only accurate when water table
. The study also found that the approximation method is better for
practical use when the water table is above ground surface.
58
This section described water inflow in a drained tunnel under water table. The hydraulic
conductivity of a tunnel lining is a critical parameter for water inflow prediction as discussed
in the reviewed literature. The linings were considered to be homogeneous and have a
constant hydraulic conductivity as discussed in literature. The effect of an inhomogeneous
tunnel lining to the hydraulic behaviour around a tunnel has not been considered in the
literature.
Water inflow models of lined water-sealing tunnel
The water sealing tunnels have no drainage layer and are often constructed where the
groundwater needs to be maintained and protected (Maidl et al., 2013). The water-sealing
tunnel lining is usually designed to be impermeable when the tunnel is constructed (Maidl et
al., 2013).
A study of water inflow of water-sealing tunnels was developed (Fernandez and Moon,
2010b). The assumptions are the same as in the classical Goodman’s solution. Water inflow
of lined tunnels was separated into two different stages (Fernández, 1994). Stage one is
water travelling through the rock mass to the interface of the lining and rock mass. Stage two
is water flow through the lining.
The first stage is water flow through rock mass which was calculated using the classical
Goodman’s solution as follows.
(2.30)
where, Qm is the water inflow rate from the rock mass to the interface (m3/s), a is the tunnel
inner radius (m), d is the thickness of the lining (m), houter is the water head at the outer
59
surface of spring line of the lining (m) (using tunnel spring line as datum level), Km is the
hydraulic conductivity of the surrounding rock mas (m/s), h is the total water head (m).
The hydraulic conductivity of the tunnel lining was obtained by site inspection in that study
(Fernandez and Moon, 2010b). The crack pattern was measured during site inspection and the
hydraulic conductivity was then calculated based the cubic flow law as described in section
2.33.
The second stage is water inflow rate through the lining which considered as a radial flow as
follows (Fernández and Alvarez, 1994).
(2.31)
where, QL is the water rate from the rock mass to the interface (m3/s), hinner is the water head
at the inner surface of the lining (m), the hydraulic conductivity of the lining is Kl (m/s).
The boundary conditions used in that study (Fernandez and Moon, 2010b) were: 1. The
hydraulic head at the lining rock interface is a constant. 2. The pressure head at the inner
surface of the tunnel is 0, which means there is no pressurized water in the tunnel.
Under a saturated condition, water flow rate according to the continuity equation
(Fernandez and Moon, 2010b). houter was described as Eq. (2.32) by combing Eq. (2.30) and
(2.31),
(2.32)
where,
60
(2.33)
The water inflow of tunnel with a lining material was derived as Eq. (2.34).
(2.34)
The parameters are the same as in Eq. (2.30) and Eq. (2.31).
The study also provided the water pore pressure distribution of the surrounding rock mass
beyond the lining (Fernandez and Moon, 2010b).
( ) (2.35)
[ ]
( ) ( )
The difference between lined tunnel and unlined tunnel was represented by a ratio as Eq.
(2.36) (Fernandez and Moon, 2010b).
( ) (2.36)
( )
( ) ( )
where, QL is the water flow rate when the tunnel with a lining (m3/s), Q0 is the water flow
rate when the tunnel without a lining (m3/s). The result shows that the lining could reduce the
water inflow rate significantly.
The closure effect caused by stress was studied numerically (Fernandez and Moon, 2010a).
That study found the inhomogeneous rock mass which was caused by stress distribution
61
contributed greatly to the water inflow rate. It was found that stress decreased the crack
aperture and therefore decreased the hydraulic conductivity of the rock mass.
This section described water inflow in a water-sealing tunnel under water table. The
hydraulic conductivity of a concrete lining was obtained by observing the crack distribution
and then calculated based on the cubic flow law. The method for crack width measurement
on site has not been clearly defined in literature. The site information including crack width
can be difficult to measure directly due to the mineral precipitates which commonly form in
and around a crack which has been discussed in Section 2.2.4.
This section described the water inflow models for a tunnel. The water inflow rate and
hydraulic pressure evaluation model for three types of tunnel namely, unlined tunnel, lined
drained tunnel and lined water-sealing tunnel were reviewed and described. The main
differences between these three tunnels types are found to be the hydraulic boundary
conditions.
Research gap
The property of a tunnel lining was found to be an important factor for tunnel assessment.
The assessment of a tunnel lining depends on the measurement of crack width on site, which
has not been clearly defined in literature. Measuring crack width directly on site can be
difficult due to the precipitates on the surface of concrete as discussed in Section 2.2.4.
Therefore, the research gap is that the method for crack width measurement has not been
clearly defined. This study will provide a method to evaluate the crack width information in
an operating tunnel based on the measured water inflow rate.
The tunnel linings in literature were considered as a homogeneous material. Meanwhile, the
tunnel lining could be inhomogeneous due to various reasons in practical engineering cases.
62
The research gap is the water inflow model for a tunnel with an inhomogeneous lining has
not been previously developed. This study will provide a method to evaluate the potential
inhomogeneous condition of a tunnel lining for the water inflow and hydraulic pressure
estimation to fill the research gap.
2.5 Reliability theory
Reliability is the generic probabilistic measure for assessing safety or serviceability. The
failure of a structure does not necessary refer to the catastrophic failure such as collapse. The
failure of a structure represents the failure relative to a selected criteria for the structure to be
functional (Melchers, 1999). The selection of criteria was found critical for reliability
calculation (Melchers, 1999).
2.5.1 Basics of reliability theory
An early modern approach to quantify the safety or serviceability of a structure was
developed in 1950s (Newmark, 1953). That study assessed a blast design for engineering
structures. The stress in a structure was considered to be the criteria for safety calculation in
that study. The permissible stress in a structure was defined as the resistance of the structure.
The applied stress caused by the blasting was defined as the load effect on the structure. The
ratio of the permissible stress and the applied stress was defined as the factor of safety. That
study defined that the structure is not safe when the factor of safety is less than 1. In other
words, the criterion would be failed when the applied load on the structure exceeded the
structure capacity (Newmark, 1953).
The assessment of a structure depends on the chosen criteria. The variation of a failure
criterion was considered for the safety of a structure (Freudenthal, 1962). In that study, the
mechanisms which could harm the structures were defined as load effect ( ). The resistance
63
of the structure to the failure mechanisms were defined as resistance . Both load effect and
resistance were defined as probabilistic parameters for an operating structure in that study.
The limit state function ( ) was defined to describe the assessment criterion (Freudenthal,
1962). The value is a basic random variable which contributes to the limit state function. A
typical function for reliability calculation would comprise multiple basic random variables as
follows.
The structure is defined as failed when the limit state function is less than 0.
(2.37)
The failure probability of a structure can be illustrated by multiple-dimensional integration as
follows (Melchers, 1999).
(2.38)
∫ ∫
The deterministic value for parameters was found not to be appropriate for practical
engineering cases (Cornell, 1967). The parameters were described as a random variables in
that study. The probabilistic approach was applied to assess the reliability of a structure
(Cornell, 1967). The resistance ( ) and load effect ( ) of a criterion was quantified by the
probability density functions and respectively (Cornell, 1967). The probability of
failure of a criterion was noted as , which is depends on the probability distribution of both
resistance ( ) and load effect ( ). The mathematical expression of the reliability of a
structure was described as follows (Cornell, 1967).
64
(2.39)
where, is limit state function.
The reliability of the limit state functions in that study was solved based on the input
variables when the variable follows a normal distribution (Cornell, 1969). That study used the
first order of Taylor series and the first two moments of the input variables, which are the
mean and the standard deviation to define the distribution of the resistance and the load effect
(Cornell, 1969). The mean values of resistance and load effect were denoted as and ,
and the variances were denoted and respectively. The probability of failure of a
structure was rewritten as follows (Cornell, 1969).
(2.40)
[ ]
where, was defined as ‘safety index’, is the standard normal distribution function (zero
mean and unit variance). The value of can be found from the standard normal table
(Larson and Farber, 2012).
A fundamental analysis for the second moment reliability calculation has been developed
(Hasofer and Lind, 1974). That study developed a transformation method to simplify the limit
state function. The fundamental purpose of that method is to transform all variables to their
standardized form N(0,1), which is a normal distribution with zero mean and unit variance.
That transformation is also known as Hasofer-Lind transformation as follows (Hasofer and
Lind, 1974).
65
(2.41)
where, Yi is a standardized normal distribution has and .
A generalized second moment reliability index was defined for the condition where no high
quality information is available (Ditlevsen, 1981). That study was developed based on the
method of Lagrange multipliers. The physical means of the reliability index was defined as
the minimum distance from the checking point to the failure surface. The application of this
method for a multiple failure mode system in a civil engineering case was presented in that
study (Ditlevsen, 1981).
The probabilistic information was often found to be incomplete in practical engineering
applications (Der Kiureghian and Liu, 1986). The fundamental requirement for using that
method was that the distribution model for the variables should be constant with the available
information. That study discussed several scenarios for cases with incomplete information
including the condition when the safety margin was known, the condition when marginal and
partial joint distribution were known and the condition when less than marginal distribution
was known. That study developed a comprehensive framework for each scenario to
incorporate the incomplete information in a reliability study (Der Kiureghian and Liu, 1986).
A study for an exact solution for the reliability index was developed based on the first order
second moment theory (Melchers, 1999). The solution for a linear limit state function in
reliability approach was presented in that study. The solution for the reliability index for a
linear limit state function was described as follows (Melchers, 1999).
66
(2.42)
(∑ )
where the represent the coordinates of any point on the limit state surface.
The equation of the safety index for the linear limit state function is subject to the condition
when g(y)=0.
The function g(y)=0 at y* was developed as follows for the linear limit state function as
(2.43)
∑
The safety index of a linear limit state function was determined directly from the checking
point coordinate y* as follows.
(2.44)
∑
That study also provided a solution for non-linear limit state function in the reliability index
approach. The solution for the reliability index was described as follows (Melchers, 1999).
∑ (2.45)
where, is equal to . is the ith partial derivative that evaluate at design point. The limit
state function g(y)=0 was evaluated using Taylor expansion.
The function g(y)=0 at y* was developed as follows (Melchers, 1999).
67
(2.46)
∑
In that study the reliability index for a non-linear performance limit function was give as
∑ (2.47)
The reliability index can be solved accordingly.
Reliability theory has been applied in geotechnical engineering (Duncan, 2000). That study
analysed the stability of a retaining wall and a dam using reliability theory. The highest
conceivable value and lowest conceivable value of each variable was determined using three
standard deviations (99.7%) of average of a normally distributed parameter in that study.
That study confirmed that the theory of reliability can be used in geotechnical engineering
practice when the on-site parameters are considered as variables (Duncan, 2000).
2.5.2 Outcrossing method
An early study using the outcrossing method for solving reliability problem was developed
by Ditlevsen (1983). The reliability calculation was conducted for a system having a convex,
polyhedral safety set in the space of basic variables in that study. The structure in that study
was subjected to a Gaussian vector-load-effect process (Ditlevsen, 1983). The safety region
of a structure within a given time interval t was defined by the trajectory of the vector process.
The failure probability in this study was described as the first passage probability, which is
the probability for the first violation of the safety margin, is determined as follows (Ditlevsen,
1983).
68
(2.48)
∫
where, is the probability of failure at time 0, is the outcrossing rate.
The mean outcrossing rate in that study was calculated from generalized Rice formula (Rice,
1944). The two-dimensional vector process and an safety domain D were used
combining with the generalised Rice formula (Ditlevsen, 1983). The safety domain D was
described by . In the case of a stationary Gaussian scalar process, the
mean outcrossing rate as follows (Ditlevsen, 1983).
(2.49)
∫ [ { ∑ ̇ }| ]
where, is the probability density function of , is the piece-wise continuous
differentiable boundary of D, is the unit vector normal to at and ̇ is the
time derivative of .
The upcrossing rate for the non-stationary Gaussian process was developed analytically (Li
and Melchers, 1993). The upcrossing problem was formulated using the original Rice
formula rather than the generalized one. The multiple-dimensional integral was simplified to
one dimensional integral. The approach in that method would be able to extend the
formulation to the time-dependent domain. The upcrossing rate in that study was described as
follows (Li and Melchers, 1993).
̇| ̇| ̇| ̇| (2.50)
{ ( ) }
̇| ̇| ̇|
The time dependent reliability theory was developed to evaluated the time when the first
occurrence of an excursion of the random vector out of the safe domain D (Melchers, 1999).
69
The probability of outcrossing in a range [0, t] was approximated by Poisson distribution in
that study.
The service of a structure caused by deflection was studied in a time dependent approach (Li,
2005). The criterion was considered as deflection for a concrete structure in that study. The
deflection of the structure was considered as a stochastic process in that study. The
probability of upcrossing was described based on the theory of stochastic process as follows
(Li, 2005). The up-crossing rate in that study was calculated using the same theory as in Eq.
(2.49). The notation represented the failure criterion which is deflection.
∫ (2.51)
̇| ̇ ̇| ̇ ̇| ̇ ̇| (2.52)
{ ( ) }
̇| ̇| ̇|
̇
̇| [ ̇| ] ̇
(2.53)
̇| ̇ (2.54)
(2.55)
̇
(2.56)
̇ |
̇ (2.57)
̇ ̇
(2.58)
̇
70
where, and are standard normal density and cumulative distribution function of the
function in bracket respectively, and were defined as the mean and standard deviation of
(stochastic process of deflection), is the auto correlation coefficient for the auto-
covariance function at time and , is the auto-covariance function, ̇ is
the cross-variance function.
A reliability study for a corrosion affected concrete structure was conducted (Li and Melchers,
2005). In that study, the failure probability of a concrete structure was determined using the
upcrossing method. The function of the load effect and the resistance was fitted by linear
regression. The threshold of serviceability was a deterministic value in that study and the
failure probability using the upcrossing method was described as follows (Li and Melchers,
2005).
∫
̇|
{ (
̇|
)
̇| ̇|
}
(2.59)
̇| ̇| ̇|
where, is the failure probability of a structure, the other parameters are the same as in
Eq. (2.52).
A closed-form solution was developed for the outcrossing rate calculation (Li et al., 2016).
The non-stationary lognormal process was discussed in that study based on the rice formula.
The upcrossing rate of the lognormal processes showed good agreement with Monte Carlo
simulation and safety index method. The analytical solution of the upcrossing rate in that
study is shown as follows (Li et al., 2016).
71
̇| ̇ ̇|
(2.60)
( ) {√ ( ( ))
̇|
̇ ̇| ̇ ̇|
( ) ( )}
̇| ̇|
(2.61)
√
(2.62)
√
where, is the standard deviation function of the stochastic process, is the mean function of the
stochastic process, is the outcrossing rate, the other parameters are shown as Eq. (2.52).
2.5.3 Monte Carlo simulation
The Monte Carlo method was initially developed in 1940s in Los Alamos National
Laboratory for the Manhattan project. The fundamental concepts of Monte Carlo simulation
were discussed in an early study (Buslenko, 1966). That study described Monte Carlo
simulation as a method for solving function by giving a large number of trials. The desired
results can be quantified by recording of the trials and analyse the correspondent output
information (Buslenko, 1966).
The general approach for using Monte Carlo simulation was to sample each random variable
, and give it a random value ̂ (Melchers, 1999). The limit state function of the system
̂ can be calculated as a result. When the limit state function of the system is less
than 0, the function was marked as ‘failed’. The methodology repeats this process a large
number of times. The mathematical expression was written as follows (Melchers, 1999).
72
̂ (2.63)
where, ̂ is the number of trails n for which ̂ . The number N of trials
required, which is depends desired accuracy of .
Four general principles for conducting a Monte Carlo simulation were given (Melchers,
1999). Firstly a systematic method for sampling the basic variable was found required in that
study. Secondly, a reliable simulation technique for the basic variable was found to be
required. Thirdly, the effect of the number of variables and the complexity of the limit state
function was found to be important and needs to be evaluated. Finally, a reasonable number
of trials was found to be important to determine the probability of failure of a limit state
function (Melchers, 1999).
The efficiency of multivariate control variability in Monte Carlo simulation was discussed
(Rubinstein and Kroese, 2011). The method for processing variables includes controlling the
variance of the variable. The Markov chain Monte Carlo simulation was recognised as an
algorithm which is dependent with the trail steps in that study (Rubinstein and Kroese, 2011).
A study for tunnel risk assessment was developed using Monte Carlo simulation (Vanorio
and Mera, 2012). A Monte Carlo software package @RISK was used in that study. An
example urban metro tunnel with was analysed in that study. The length of the selected tunnel
is 18 Km, operating 17 hours a day. The frequency between trains is approximately 120
seconds. The selected tunnel has been operated for 30 years with a total train kilometre over
200 million. Four types of risk were identified in an urban metro tunnel in that study, namely,
derailment, collision, impact with obstacle and fire. The probability of failure for each
criterion was obtained based on data provided by the local statistical department. The failure
73
probability of the metro tunnel was then evaluated by using Monte Carlo simulation software
package (Vanorio and Mera, 2012).
The probability of failure of a tunnel support system was studied using Monte Carlo
simulation (Bukaçi; et al., 2016). A tunnel in hydropower plant in Albania was studied to
investigate the application of the Monte Carlo simulation in that study. The variables in that
study was the rock mass condition which quantified by the uniaxial compression strength and
the rock mass rating (Bukaçi; et al., 2016). The stress on the tunnel was defined as the failure
criterion in that study. The tunnel support system which comprises concrete and steel
reinforcement was design to support the tunnel. The failure probability of the tunnel was
calculated using Monte Carlo simulation based on the variables include the stress in the
tunnel and the supporting capacity (Bukaçi; et al., 2016).
A probabilistic risk analysis of tunnel during construction stage was developed (Yu et al.,
2017). Multiple parameters which could contribute to the construction failure have been
identified in that study. That study established a framework for tunnel safety using a complex
system to simulate the working site during the construction stage of a tunnel. A twin tunnel
for hydropower station was used as the project case in that study. The criteria in that study
include blasting, ventilation, shotcrete-bolt and security inspection. The failure probability of
the system was quantified using Monte Carlo simulation in that study (Yu et al., 2017).
To analyse the application of reliability theory on tunnel, water ingress into a tunnel is an
important concerning for tunnel industries as discussed in Section 2.2. Water ingress into a
tunnel has not been studied as a stochastic process for tunnel reliability calculation. This
study will provide an application of the reliability theory on the tunnel service life prediction
using water ingress as a stochastic process.
74
2.5.4 System reliability
The reliability of a system often comprises more than one failure criterion (Madansky, 1965).
An early study was developed to calculate the reliability of a system (Madansky, 1965). The
failure of any component in a series system would result the failure of a criterion in that study.
That study found that the connection of the components in a system would affect the failure
probability of the system. The connection of components in a given system was simplified
into two categories, which are series connection of components and parallel connection of
components (Madansky, 1965).
A series connection of components was described as a chain in a study for structure design
(Freudenthal, 1962). That study described a system which comprises purely series component
connection as the weakest link system (Figure 2.6). The failure probability of the weakest
link system was found largely depends on the component which has the highest failure
probability. The system would fail if any component in the system fails (Freudenthal, 1962).
Figure 2.6 Series system.

The mathematical expression of a series system was described as follows (Freudenthal, 1962).
(2.64)
A parallel connection of components was described as a redundant system in a study for
structure design (Freudenthal, 1962). That study described the components which involved in
a parallel connection as redundancies (Figure 2.7). The failure probability of a redundant
system was found largely to depend on the component which has the lowest failure
75
probability. The parallel system would fail if all the components in the system are failed
(Freudenthal, 1962).
Figure 2.7 Parallel system.

The mathematical expression of a parallel system was described as follows (Freudenthal,
1962).
(2.65)
where, each failure mode was expressed by a limit state function . If the failure
probability of each failure mode is known, and the failure probability of each section was
calculated.
A study was developed to apply system reliability theory to a nuclear reactor system
(Rackwitz and Krzykacz, 1978). The concept of safety region was developed to quantify the
reliability of a system in that study. The safe region for a system was denoted as ̅ in that
study. The expression of the safe region of a system with series connection and parallel
connection was described Eq. (2.66) (Rackwitz and Krzykacz, 1978).
76
̅ ̅ ̅ (2.66)
̅ {
̅ ̅ ̅
where, ̅ is when limit state functions of the failure mode is larger than 0 ( ).
Combining with the safety region of a system, the probability of safety of a system was
written as follows (Rackwitz and Krzykacz, 1978).
(2.67)
∫ ∫
̅
where, is the probability of safety of a system, X is the vector of random variables, is
the survival function that determined by each component.
The failure probability of a parallel system which consists of multiple components was
discussed (Rausand and Arnljot, 2004). The status of components in the system in that study
only has two possible outcomes, failed or not failed. The number of component is defined to
be an integer in that study. The probability when more than x number of component failure in
that system was written as follows (Estes and Frangopol, 1998).
(2.68)
∑( )
where, is the probability of occurrence when more than x number of component fail, n is
the total number of component in this system, is the probability of failure of one
component. The value x and n are defined to be integers.
A probability density evolution method (PDEM) was developed to determine the global
reliability of a complex engineering structure (Chen and Li, 2005). The method developed in
that study was capable to determine the dynamic response of the random variables of a
77
structure. That study found that the method would able to deal with both linear and nonlinear
stochastic processes (Chen and Li, 2005).
The framework of reliability assessment of tunnels was developed by considering multiple
failure mechanisms of a tunnel (Yuan et al., 2012). The failure mechanisms including
concrete spalling, water seeping, corrosion and concrete degradation were discussed in that
study as a combined system. That study classified the safety grade of tunnels in five risk
grades by qualitatively studying all the aforementioned failure mechanisms as a combined
failure system (Yuan et al., 2012).
A study was developed to assess the reliability of a tunnel during earthquakes (Yue and Ang,
2015). A numerical finite element model was developed to determine the dynamic load on
the tunnel during earthquakes in that study. The aforementioned probability density evolution
method (PDEM) was employed to calculate the reliability of the tunnel subjected to an
earthquake (Yue and Ang, 2015). That study concluded that the probability density evolution
method (PDEM) could be used as an effective tool to deal with the earthquake response of a
tunnel.
Research gap
The study of reliability of tunnel is mainly focuses on the criteria which associated with stress
load on a tunnel as discussed in previous section. Water ingress is a significant problem as
discussed in section 2.2.3. Reliability studies based on the criteria of water ingress have not
been presented in the literature.
78
2.6 Summary
This Chapter reviewed and described the required literature for this study. The research gaps
have been identified through the review. Six major research gaps were identified as discussed
early in this chapter. Each research gap has been identified in text and listed below.
1. The hydraulic conductivity of a concrete with a partially interconnected cracks has not
been studied in detail as identified in Section 2.4.4.
2. Method for local water inflow rate measurement from a crack on site in a tunnel has
not been studied as identified in Section 2.2.3.
3. Method for crack width evaluation in a tunnel lining has not been studied as identified
in Section 2.4.5.
4. The potential inhomogeneity (variation) of hydraulic conductivity of a tunnel lining
has not been studied in detail as identified in Section 2.4.5.
5. Water ingress into a tunnel has not been studied as a criterion for tunnel reliability
calculation as identified in Section 2.5.4.
6. The efficiency of water inflow treatment methods has not been reported and evaluated
in detail as identified in Section 2.2.4.
The correspondent six actions for the six research gaps are listed as follows.
1. This study develops a method to calculate the hydraulic conductivity of a concrete
with partially interconnected cracks as shown in Chapter 3.
2. This study develops a method for water inflow measurement onsite in a tunnel as
shown in Chapter 4
3. This study develops a method to estimate the crack width in an operating tunnel based
on the measured water inflow rate and the other hydraulic parameters as shown in
Chapter 3.
79
4. This study develops a method to quantify the response of the water inflow and the
hydraulic pressure caused by the potential lining inhomogeneity as shown in Chapter
3.
5. This study provides a study for tunnel reliability prediction using water ingress as the
assessment criterion as shown in Chapter 5.
6. This study provides a study on the efficiency of water inflow treatment of an selected
tunnel as shown in Chapter 6.
80
3 Water Ingress into Concrete Lining
3.1 Introduction
The prediction of water ingress through concrete lining is essential for tunnel durability
assessment and service life prediction (ITA, 1991). Based on the crack formation, concrete is
defined into three categories in this chapter, namely, sound concrete, concrete with fully
interconnected crack and concrete with partially interconnected crack. The crack width is
used to quantify the hydraulic conductivity of concrete with fully interconnected cracks. A
new parameter called crack connectivity coefficient is developed to quantify the crack
interconnection within a concrete in this chapter. The crack connectivity coefficient is then
used to calculate the hydraulic conductivity of concrete with partially interconnected cracks.
The term water ingress is used here to refer to general process of water movement through
concrete. A saturated and steady-state condition is assumed in this study. The two different
terms are used to differentiate the water ingress rate, namely, water seeping and water inflow.
Water movement through concrete is categorised as water seeping when water ingress rate is
low and does not form rapid water flow. Water movement through concrete is categorised as
water inflow when the water ingress rate is high and forms rapid water flow. Water would
take a certain time to seep through a concrete with low hydraulic conductivity. The water
seeping time through the concrete lining in the selected tunnel is calculated based on the data
measured from site. The crack connectivity coefficient is considered as an input parameter for
calculating the time for the water to seep through concrete. The effect of surrounding soil and
rock can also affect the behaviour of concrete lining, however it is not in the scope of this
study and therefore not further discussed. The study in Section 3.2 in this chapter aims to
develop a method to calculate the hydraulic conductivity of a concrete with a partially
81
interconnected crack (research gap 1). The method developed in this study is then further
used in the reliability calculation in Chapter 5.
The cracking condition of a concrete lining is a critical factor for water inflow assessment as
discussed earlier. The crack width on a lining would be difficult to obtain due to reasons
include, difficult accessibility and mineral deposits on the crack surface (research gap 3).
This Chapter will provide a new method to evaluate the crack width on the tunnel lining
based on the measured water inflow rate through cracks as in Section 3.3.
The potential inhomogeneity of a tunnel lining has not been considered in literature (research
gap 4). The existing analytical models of water flow through a tunnel lining under steady-
state, saturated conditions are extended to incorporate a linear variation of hydraulic
conductivity with distance from the tunnel wall as in Section 3.4. The impact of
inhomogeneity of a lining on water inflow rate and water pressure distribution is evaluated.
The relation between lining inhomogeneity and other hydraulic parameters is established.
Section 3.2 develops a model of hydraulic conductivity based on crack features. Section 3.3
provides a method for effective crack width calculation based on the existing relation of
water inflow rate and hydraulic conductivity under the assumption of homogeneous hydraulic
conductivity. Section 4.3 develops a new water inflow and hydraulic head model based on the
relation of water inflow rate and hydraulic conductivity under the assumption of
inhomogeneous hydraulic conductivity.
This chapter aims to address the three research gaps (research gaps 1, 3 and 4) as in Section
3.2, Section 3.3 and Section 3.4 respectively. Each section provides a correspondent solution
for the identified research gaps through literature review. A list for terminology and symbols
used in this study is listed as follows.
82
Terminology Synonym Symbol Explanation
Crack width Crack Distance between the surfaces bounding a crack
aperture
Crack density Crack Length of cracks present in a unit area
abundance
Hydraulic conductivity Permeabilit Capability for water to move through a material
y
coefficient
Flow area Area perpendicular to the water inflow direction
Spring line The mid height of a tunnel
Depth from the water table (or phreatic pressure
Hydraulic
Total head H head) to springline, when the datum level is spring
head
lining.
Tunnel radius
Lining thickness
Flow rate Discharge Quantity of water flow through a unit area in a
unit time
Inflow Seepage Quantity of water flow into a length of a tunnel

Inflow rate Seepage Quantity of water flow into a length of a tunnel in
rate a unit time
Inner surface Intrados* Surface of the lining exposed inside the tunnel
Outer surface Extrados* Surface of the lining at the interface of the lining
and surrounding rock
Inhomogeneous Ratio of the hydraulic conductivity at outer lining
hydraulic conductivity and inner lining
coefficient
Water inflow factor A dimensionless factor
Water inflow factor A graphical approximation of an analytical

chart solution representing a dimensionless water
inflow parameter and
Homogeneous lining The hydraulic conductivity through lining is
hydraulic conductivity constant
Inhomogeneous lining The hydraulic conductivity through lining is non-
hydraulic conductivity constant
Constrictive lining Hydraulic conductivity of lining decrease from
hydraulic conductivity rock-tunnel interface to tunnel inner surface
Expansive lining Hydraulic conductivity of lining increase from
hydraulic conductivity rock-tunnel interface to tunnel inner surface
83
3.2 Hydraulic conductivity of different categories of concrete
This section presents a theory to calculate the hydraulic conductivity of a concrete with
different cracking condition (research gap 1).
The velocity of water seeping in a porous media was described by the hydraulic conductivity
and the hydraulic gradient as discussed in Chapter 2. The total time for water seeping through
a material was described using the transformation of Darcy’s law as discussed previously
(Valenta, 1969) as follows. Water seeping in this case refers to the water movement through
concrete when the water ingress rate is low.
(3.1)
where, K is the hydraulic conductivity (m/s), x is the water seeping depth (m), v is the water
filled pore volume fraction during the test (m3/m3), H is the pressure head (m), t is time (s).
That study assumed that the concrete section where water has seeped through will become
saturated (Valenta, 1969). All the pores and voids are filled by water after the water seeps
through. Under this assumption, the water filled pore volume fraction (v) in Eq. (3.1) is 1.
The process of water seeping through a material is a function of saturated material hydraulic
conductivity, hydraulic head and time.
The hydraulic conductivity of a saturated cracked concrete is discussed based on the crack
interconnection and the hydraulic conductivity of uncracked concrete. Water tends to migrate
through a material via the section with higher hydraulic conductivity. The section of concrete
with higher hydraulic conductivity, which is the cracked area, is more important in terms of
water ingress. The hydraulic conductivity of three different categories of concrete (sound
84
concrete, concrete with fully interconnected cracks and concrete with partially interconnected
cracks) are discussed as below.
3.2.1 Different categories of concrete
Sound concrete
The hydraulic conductivity of sound concrete is mainly determined by the intrinsic properties
of concrete, including the aggregate size, water cement ratio and hydration degree as
discussed in the literature review section. All these parameters are related with the porosity of
concrete as discussed in Section 2.3.1. The relation between porosity and permeability of
concrete was developed empirically by experiment (Breysse and Gérard, 1997). The porosity
is a function of compressive strength (Lian et al., 2011). The porosity of concrete could
potentially be calculated by measuring the compressive strength of concrete. The
permeability of concrete could be quantified if the porosity is available. The conversion
between concrete compressive strength and concrete hydraulic conductivity is based on the
relation between the strength ( ) -porosity ( ) of concrete as Eq. (3.3) (Breysse and Gérard,
1997) and permeability ( )-porosity ( ) of concrete as Eq. (3.2) (Lian et al., 2011).
(3.2)
( )
(3.3)
The hydraulic conductivity of the sound concrete is assumed to be 2.57×10-12 m/s for the
general calculation in this chapter.
Concrete with fully interconnected crack
For a material with low hydraulic conductivity, cracks in the direction perpendicular to water
ingress can be simply ignored and the material is taken as intact material (Fetter, 2000).
85
Cracks along the water ingress direction are more important for hydraulic conductivity
calculations. The water flowing through a crack along water flow direction was studied
analytically as cubic flow law (Reinhardt, 1997). The cubic flow law was applied to quantify
the water flow rate through cracked concrete when cracks in concrete are fully interconnected.
The total flow area ( ) of cracks is the crack width times crack density of concrete. Based on
Darcy’s law and the cubic flow law, the equivalent hydraulic conductivity of a single crack is
a transformation of Eq. (2.14) as follows.
(3.4)
where, is the equivalent hydraulic conductivity of a crack, (m) is the crack aperture
(width), is the crack roughness reduction factor, (kg/m3) and (Pa.s) are the water
density and viscosity, respectively.
Concrete with partially connected crack
Water tends to move faster through cracked zones compared to intact zones due to the high
hydraulic conductivity of the cracked zones. Sound concrete is the main barrier for water to
travel through a partially cracked concrete block. A smaller concrete fraction will result an
easier water seeping environment. The seepage path is defined to be the water migrating path
through a cracked concrete which takes the shortest time. The seep path has the highest local
hydraulic conductivity among the whole block of the cracked concrete. The schematic
illustration of the seepage path in a cracked concrete in a tunnel lining is shown as in Figure
3.1.
Where water exits a crack into the tunnel the flow will vary according to the position on the
tunnel and the orientation of the crack relative to the tunnel. Water would partially drip from
86
the crack and partially flow on the side of the tunnel when the crack is on the crown of the
tunnel. Water would flow along the tunnel wall and forms a fan shape flow area when the
crack is on the side of the tunnel. The following illustration only shows the circumstance
when the crack is on the side of the tunnel. The behaviour of water and the relevant
techniques for water inflow measurement will be further discussed in Chapter 4.
Figure 3.1 Schematic illustration of a seepage path of concrete
3.2.2 Development of hydraulic conductivity
To determine the hydraulic conductivity of the cracked concrete, the cracked concrete is
investigated under a saturated condition.
The hydraulic conductivity of concrete is very low compared to the cracks in concrete. The
process of water seeping through a seepage path is calculated as one dimensional problem.
The length of seepage path is written as Eq. (3.5).
87
∑ ∑ (3.5)
where, ∑ is the total crack depth in x-y plane, ∑ is the distance of remaining concrete
thickness between cracks, is the total length of seepage path.
The connectivity coefficient is introduced to quantify the connectivity of crack in concrete. In
this study, crack connectivity coefficient is defined to be the ratio between total effective
crack length in x-y plane and the correspondent length of seepage path. The connectivity
coefficient ( ) is shown as Eq. (3.6).
∑ (3.6)
The connectivity coefficient is simplified in one dimensional seepage. The other
parameters are defined as in Eq. (3.5).
The seepage path is fully connected when by the definition in this study. However,
cracks may have some blockage effect due to the roughness of the crack or location of
aggregates (Akhavan et al., 2012). The aggregate size of concrete typically ranged from
9.5mm to 37.5mm (Neville, 2011). It is difficult to interpret the connectivity condition when
the total concrete thickness in seepage path is smaller than an average aggregate size.
Therefore, the seepage path is considered as effective fully interconnected beyond this
accuracy.
Based on Darcy’s law, water seeping through seepage path is described as Eq. (3.7).
(3.7)
88
where, is the water flow rate through seepage path, the equivalent hydraulic conductivity
of seepage path is , the hydraulic head difference over seepage path is , the area that
perpendicular to water flow direction is , which is equal to width of seepage path ( ) times
total seepage path length in z direction (unit value). The seepage path is assumed to have the
same width as the cracks in the concrete.
The total flow rate through the seepage path can be represented by water flow rate through
the concrete part of the seepage path. Under a saturated condition, the total flow is the same
as water flow rate through the cracked part of seepage path due to mass conservation. The
water flow rate through each cross-section of the seepage path is equal. Therefore, Eq. (3.7)
is rewritten as Eq. (3.8).
(3.8)
where, the water flow rate through seepage path is , the flow rate travel through the crack
part of seepage path is , the flow rate travel through the concrete part of seepage path is
Combining Eq. (3.5) and Eq. (3.6), the total crack length (∑ ) and the total concrete length
(∑ ) within the seepage path is rewritten as follows.
∑ (3.9)
∑ (3.10)
The hydraulic head loss remains at the same rate through the same material. The rate of loss
of hydraulic gradient head (hydraulic gradient) through any cross section of concrete is a
constant. The hydraulic gradient over the crack part of the seepage path is . The hydraulic
89
gradient over the concrete part of the seepage path is . The total hydraulic head difference
( ) over seepage path is equal to ∑ ∑ .∑ is the summation of hydraulic head loss
over cracks. ∑ is the summation of hydraulic head loss over concrete. The hydraulic
∑
gradient over sound concrete is ∑
by definition. The hydraulic gradient over crack is
∑
∑
by definition.
To quantify the flow rate through the seepage path, and are further determined based
on Darcy’s law as follows.
∑ (3.11)
∑
∑ (3.12)
∑
where, the equivalent hydraulic conductivity of the crack part of seepage path is , the
hydraulic conductivity of the concrete part of seepage path is which has been determined
in Section 3.2.1.
Based on the aforementioned equations, equation 3.5 is further developed as follows.
∑ ∑ (3.13)
The relation between water ingress rate through the crack part of seepage path and the
concrete part of seepage path is written as Eq. (3.14).
∑ (3.14)
∑
90
Based on Eq. (3.14), the hydraulic head over the cracked zone (∑ ) of seepage path is
described as follows.
(3.15)
∑
The equivalent hydraulic conductivity of the seepage path ( ) is described based on Eq.
(3.13) and Eq. (3.15) as follows.
(3.16)
where, the equivalent hydraulic conductivity of seepage path is , the equivalent hydraulic
conductivity of the crack part of seepage path is , the hydraulic conductivity of the
concrete part of seepage path is , connectivity coefficient is .
The equivalent hydraulic conductivity of seepage path Eq. (3.16) is further written as follows.
(3.17)
The equivalent hydraulic conductivity of a crack ( ) is usually much higher than the
hydraulic conductivity of sound concrete ( ). The ratio of and is infinitely close to
0. The equation is further simplified as follows.
(3.18)
where, the equivalent hydraulic conductivity of saturated seepage path is , the hydraulic
conductivity of the concrete part of the seepage path is , connectivity coefficient is .
91
The equivalent hydraulic conductivity of a saturated seepage path has been provided above.
A higher connectivity coefficient means higher hydraulic conductivity and therefore water
will take less time to seep through that media.
3.2.3 Crack connectivity coefficient
The parameters used in Eq. (3.18) are considered to be deterministic values. The hydraulic
conductivity of the seepage path is a function of the hydraulic conductivity of sound concrete
and the crack connectivity coefficient as calculated in Eq. (3.17). The formation of crack
connectivity (α) of the concrete is a key factor in dealing with the hydraulic conductivity of a
seepage path. The statistical distribution of crack features is required to determine the
characteristic value of the connectivity coefficient (α) using Monte Carlo simulation. The
characteristic value of crack connectivity used in this study is when the accumulated
probability is equal to three standard deviations (99.7%). The criteria of three standard
deviations has been applied in the reliability analysis in multiple engineering cases (Duncan,
2000) as discussed in Chapter 2. The characteristic value of crack connectivity obtained in
the Monte Carlo simulation can be used in Eq. (3.18).
Method of two-dimensional simulation
The crack connectivity coefficient ( ) is determined by the spatial parameters of cracks. The
crack connectivity coefficient ( ) of the seepage path is then determined using Monte Carlo
simulation.
One approach to study the crack connectivity coefficient ( ) is to consider this problem in a
two-dimensional way. The number of Monte Carlo realizations is denoted as MC. The
concrete thickness (d) is 0.56 m. The cracking in concrete is defined by four geometry
92
parameters, number of crack (n), crack centre coordinate ( ), crack length in x-y plane ( )
and crack orientation to x-z plane ( ).
The parameters used in the worked example are assumed as follows. The position of crack
centre is uniformly distributed in a space of ([0, 0.56], [0, 0.56]). The crack length is assumed
to have a normal distribution with a mean of 0.17 m and standard deviation of 0.11 m. Crack
orientation is normally distributed with a mean of 90° and standard deviation of 15°. The
parameters which define the crack are shown as in Figure 3.2.
Figure 3.2 Two-dimensional crack and the relevant parameters

The assumed values used in the worked example of a two-dimensional simulation are shown
as in Table 3.1.
93
Table 3.1 Assumed parameters used for a two-dimensional simulation
Parameters Symbol Distribution type
Concrete
Constant (0.56)
thickness
Crack number P (n, 3)
Crack centre
[x,y] Uniform [(0,0.56), (0,0.56)]
coordinate
Crack length x-y
Normal (0.17, 0.11)
plane
Crack orientation
Normal (90°, 15°)
to x-z plane
The steps for the Monte Carlo simulation for simulating the crack connectivity are shown as
follows.
1. Define a boundary of concrete for the Monte Carlo simulation.
2. Randomly generate the number of cracks which follows Poisson distribution.
3. Randomly generate the centre coordinate of each crack within the defined boundary
which follows uniform distribution.
4. Randomly generate the crack length and crack orientation of each crack follow
normal distribution.
5. All cracks can be plotted based on steps 1 to 4.
6. Calculate the seepage path with the minimum concrete distance in between.
7. Calculate if there is any intersection of cracks in the concrete and conduct
correspondent action which will be further discussed.
8. The connectivity coefficient is then calculated directly once the seepage path is
known. Record the connectivity coefficient for this realization.
94
9. Repeat Step 1 to 8 and run a large number of realizations until the mean value of the
crack connectivity converged within in 0.5%.
The method for Step 6 is further discussed as follows.
In order to calculate the seepage path, the distance between each crack is calculated and
presented in a matrix based on the geometric information of the cracks. Since there are two
boundaries of concrete body, the size of the matrix is equal to (n+2)2. The value in the matrix
represents the distance between each crack and boundaries as
| |
| |
where, D is the distance between two cracks or boundaries. UB and LB are the upper
boundary and the lower boundary respectively.
All the possibilities of crack connection are performed based on the shortest path theory by
Dijkstra (1959) method once the crack distance matrix was calculated. The seepage path is
identified by finding the crack interconnection with the minimum concrete ( ) in-between.
The crack IDs that contribute to the critical seepage path are identified.
The method for Step 7 is further discussed as follows.
The length of the crack area is re-calculated after the crack ID that contributed to the seepage
path has been identified. It is noted that if one crack contributes to the seepage path it does
not mean the whole length of that crack contributes to the seepage path. The cracked length
in seepage path is calculated by finding the crossing point of cracks when the cracks joint
together as shown in Figure 3.3. The cracked length in the seepage path is denoted to be ( ).
95
Figure 3.3 Illustration of the seepage path when cracks are crossing
Therefore, the connectivity of one realization of the simulation is calculated as . The
probability distribution of the connectivity coefficient ( ) is obtained by conduct multiple
numbers of realizations. The flow chart for determining the connectivity coefficient of the
seepage path in a cracked concrete is shown as Figure 3.4.
96
97
Figure 3.4 Flow chart of Monte Carlo simulation for the connectivity coefficient of
critical seepage path
The worked example in a two-dimensional condition is simulated based on the parameters
presented in Table 3.1.
The result of the worked example shows that the cracks in the concrete lining are most likely
to be half-way through ( ) when using the assumed input parameters in Table 3.1.
There is a slim possibility that the cracks goes all the way through the concrete. The mean
crack connectivity of the two-dimensional Monte Carlo simulation is calculated to be 0.52.
The characteristic value of crack connectivity coefficient is approximately 0.93. The
parameters used in this simulation are assumed values from Table 3.1.
Method of three-dimensional simulation
A crack is expressed as a plane with finite dimension in a three-dimensional space. The crack
in concrete is defined by seven geometry parameters in three-dimensional study, namely,
number of crack (n), crack centre coordinate ( ), crack length in x-y plane ( ), crack
orientation relative to x-y plane ( ), crack length in x-z plane , crack orientation relative
to x-z plane ( ) and the rotation angle relative to x-y plane ( ).
The assumed parameters for the three-dimensional simulation are shown as follows. The
position of the crack centres is uniformly distributed in a space of ([0, 0.56], [0, 0.56],
[0,0.56]). The crack length in x-y plane ( ) is normally distributed with a mean of 0.17 m and
standard deviation of 0.11 m. Crack orientation relative to x-y plane ( ) is normally
distributed with a mean of 90° and standard deviation of 15°. The crack length in x-z plane ( )
is normally distributed with a mean of 0.17 m and standard deviation of 0.11 m. The crack
orientation relative to x-z plane ( ) is normally distributed with a mean of 90° and standard
98
deviation of 15°. The rotation angle relative to x-y plane ( ) is normally distributed with a
mean of 0° and standard deviation of 15°.
The summary of the assumed value used in the work example for three-dimensional
simulation is shown as Table 3.2.
Table 3.2 Parameters used in the worked example for three-dimensional simulation
Parameters Symbol Distribution type
Concrete
Constant (0.56)
thickness
Crack number Poisson (n, 3)
Crack central Uniform [(0,0.56),
[x,y,z]
coordinate (0,0.56),(0,0.56)]
Crack length x-y
Normal (0.17, 0.11)
plane
Crack orientation
Normal (90°, 15°)
to x-y plane
Crack Length x-z
Normal (0.17, 0.11)
plane
Crack orientation
Normal (90°, 15°)
to x-z plane
Crack orientation
Normal (0°, 15°)
to x-y plane
The Illustration of a three-dimensional crack in concrete lining and the parameter which used
to define the crack formation is shown as Figure 3.5.
99
Figure 3.5 Three-dimensional crack and the relevant parameters
The modelling process is the same as Figure 3.4. The crack connectivity under three-
dimensional conditions is noticeably smaller compared to the two-dimensional conditions
using similar input parameters. The connectivity coefficient ( ) in the three-dimensional
worked example are more likely to be 30% through the concrete ( ) when using the
input parameters in Table 3.2. The mean crack connectivity of the three-dimensional Monte
Carlo simulation is calculated to be 0.35. The characteristic value of the crack connectivity
coefficient is approximately 0.71.
3.2.4 Worked example and discussion
The hydraulic conductivity of sound concrete is a value determined based on the concrete
intrinsic property as discussed above. The hydraulic gradient is based on the water table
information (assumed to be 15 m) and the lining thickness. The parameters assumed in this
study are shown in Table 3.3.
100
Table 3.3 Parameters used in the calculation
Parameter Mean Comments

Determined value based on
Concrete hydraulic
2.57×10-12 m/s concrete property as discussed in
conductivity
Section 3.2.1
Lining thickness 0.56 m From the design drawing
Based on water table height and
Hydraulic gradient 26.8
lining thickness
The hydraulic conductivity of the seepage path increases while the connectivity coefficient
increases as shown in Figure 3.6 based on Eq. (3.18). The hydraulic conductivity increased
significantly when the connectivity coefficient ( ) getting close to 1.
8.0E-10 4.0E-05
7.0E-10 3.5E-05 When connectivity
When connectivity
Seepage path hydraulic
coefficient is 1
6.0E-10 coefficient is 0.99 3.0E-05
conductivity (m/s)
conductivity (m/s)
5.0E-10 2.5E-05
4.0E-10 2.0E-05
3.0E-10 1.5E-05
2.0E-10 1.0E-05
1.0E-10 5.0E-06
0.0E+00 0.0E+00
0 0.2 0.4 0.6 0.8 1 0.999990 0.999993 0.999996 0.999999
Connectivity coefficient ( ) Connectivity coefficient ( )
(a) (b)
Figure 3.6 Seepage path hydraulic conductivity increase when connectivity coefficient
increases from 0 to 0.99 (a) and 0.99999 to 1 (b), the other parameters are fixed
determined value
The water flow rate through the seepage path is calculated according to the seepage path area
and the hydraulic gradient. In this specific case, the concrete thickness is 0.56 m, and the
average aggregate size is larger than 6 mm based on the literature in Section 3.2.3. The
seepage path is defined as an effective fully interconnected crack when the connectivity
coefficient is larger than 0.99. The water flow rate can be calculated based on the hydraulic
101
conductivity and hydraulic gradient which are all given as above. The relation between water
flow rate and the connectivity coefficient is shown as Figure 3.7 based on the relation
between crack connectivity and the hydraulic conductivity.
4.0E-08
Water flow rate through Seepage
3.5E-08
Water flow rate
3.0E-08 when connectivity
coefficient is 0.99
path (L/min)
2.5E-08
2.0E-08
1.5E-08
1.0E-08
5.0E-09
0.0E+00
0 0.2 0.4 0.6 0.8 1
Connectivity coefficient ( )
Figure 3.7 Increase of water flow rate because of connectivity coefficient increase from 0
to 0.99, the other parameters are fixed determined value
Figure 3.7 can be used to identify the effective fully interconnected seepage path. The result
shows that the connectivity coefficient would exceed 0.99 and the system is defined as an
effective fully interconnected crack when the water flow rate through the seepage path
exceeds 3.68×10-8 L/min for one seepage path.
The water flow rate ( ) of several locations in the selected tunnel is shown in Table 4.5,
which is further discussed in Chapter 4. The lowest and the highest measured water flow rates
are 9.34×10-4 (L/min) and 1 (L/min) respectively. Since even the lowest measured value is
greater than the threshold value 3.68×10-8 L/min, the sample measurement locations are all
considered to be fully effectively interconnected. The correspondent lowest and highest
102
hydraulic conductivity of the seepage path where water flow rates were measured are
calculated to be 6.7×10-7 (m/s) and 2.39×10-5 (m/s) respectively. Based on the relationship
between connectivity coefficient and hydraulic conductivity as in Eq. (3.18), the collected
data is plotted by the correspondent hydraulic conductivity as shown in Figure 3.8. It shows
that the crack should be effectively fully interconnected to form this amount of water flow
even for the location with the least collectable amount of water.
4.0E-05
3.5E-05
Highest calculated
3.0E-05 hydraulic conductivty

conductivity (m/s)
2.5E-05
2.0E-05
1.5E-05
Lowest calculated
1.0E-05 hydraulic conductivty
5.0E-06
0.0E+00
0.999990 0.999993 0.999996 0.999999
Connectivity coefficient ( )
Figure 3.8 Lowest and highest calculated connectivity coefficient based on the site
measured data
Water can form wet areas on the tunnel wall or drip from the tunnel crown only if the water
flow rate is relatively high. Seepage will not form water flow when the water flow rate is
relatively slow. Instead, water flow would be observed as dampness on concrete surface.
Minerals which are dissolved in the water can form precipitates on the surface of the concrete.
These surface mineral deposits occur once the water seeps through the concrete as discussed
in Chapter 2, and are known as concrete efflorescence. The chemical property of the
103
precipitates is further discussed in Chapter 4. The precipitates on concrete are used as an
indication of water seepage through concrete in this study. Water dripping and water seeping
deposits in the selected tunnel are shown in Figure 3.9 a and b, respectively.
(a) (b)
Figure 3.9 Water ingress through concrete causes (a) water dripping and (b)
efflorescence
Water flow or drips can only be observed at places where the cracks are effective fully
interconnected through the concrete ( ). A partially interconnected seepage path
( ) will not be able to form water flow or drips when the hydraulic gradient is low.
The seepage path zone when is more likely to form concrete efflorescence, as
shown above.
Water seeping in a partially cracked concrete
In this study the tunnel lining is assumed to be in a saturated condition after water has seeped
through the concrete lining. The water ingress rate is very low when the crack is partially
interconnected. Water seeping through the partially interconnected seepage path usually takes
much longer time compared to water flow through a fully interconnected crack due to the
difference of hydraulic conductivity as shown in Figure 3.6.
104
The time taken for water to seep through a concrete lining is dependent on its thickness. The
thickness of the selected tunnel lining is discussed in Chapter 4. Based on Eq. (3.1), the time
for water to seep through the tunnel lining under different connectivity coefficient value for
two representative lining thicknesses is shown in Figure 3.10.
140
Water seeping through time
120
100 Series1
Series3
(year)
80
60
40
20
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Connectivity coefficient (α)
Figure 3.10 Water seeping through time under different crack connectivity coefficient
It will take up to 130 years for water to seep through the lining of the selected tunnel when
the lining thickness is 0.56 m under the given condition in Table 3.3. It will take up to 80
years for water to seep through the 0.44 m thick lining when the lining is intact. It can also be
seen that the time for water to seep through from one end of the lining to the other is very
sensitive to the connectivity coefficient. The time for water to seep through the lining and the
connectivity coefficient has a linear relation under a one-dimensional flow assumption, as
used in this study. Water will flow through the lining almost instantly when the crack is
effectively fully interconnected ( ).
105
For a more generic case, when tunnel lining varies from 0.4 m to 0.6 m. The time for water to
seep through the lining is shown as Figure 3.11.
120
Water seep through time (year)
100 Series1
Series2
80 Series3
60
40
20
0
0.4 0.45 0.5 0.55 0.6
Lining thickness (m)
Figure 3.11 Time for water to seep through lining relative to thickness
A thicker lining will increase the time for water to seep through the lining according to Figure
3.11. Increase in crack connectivity coefficient ( ) will reduce the time for water to seep
through the lining significantly. It has been calculated that water would take approximately
45 years to seep through a 0.4 m thick lining with 0.3 connectivity coefficient. In contrast, the
time for seep through time would be 100 years for a 0.6 m thick lining with 0.3 connectivity
coefficient.
This method can be used to predict the time for water to seep through a material. This method
cannot predict the water seeping depth inside the concrete in the case of partially cracked
concrete. The hydraulic conductivity of a seepage path is a bulk hydraulic conductivity which
comprises sound concrete and cracks. The crack pattern inside the concrete is not regular and
106
the water seeping depth over time is not continuous and therefore not predictable by this
method.
The study in this section developed a method to calculate the hydraulic conductivity of
concrete with partially interconnected cracks (research gap 1). The method developed in this
study is used in further chapters.
A connectivity coefficient ( ) is introduced to quantify the interconnection of cracks in
concrete. The relation between the hydraulic conductivity of a seepage path and the
connectivity coefficient ( ) is established. The hydraulic conductivity of a seepage path
increases as crack connectivity coefficient ( ) increases.
It is found that the water ingress rate is usually too low to form any water drips when the
crack is not fully effectively interconnected ( ). Water only forms efflorescence on
the surface of concrete when the water flow rate is too low to form water drips. The crack is
very likely to be effective fully interconnected ( ) through concrete when water flow
rate exceed a limit and forms water drips or flows as a film on the concrete surface.
The method developed in this chapter is applied to predict the water seeping time through a
concrete lining under the water table. Time for water to seep through through a concrete
lining is largely affected by the crack connectivity coefficient ( ). Water will take less time
to seep through the seepage path when the crack connectivity coefficient increases. The
process of water seeping through a lining can takes over 130 years in the selected tunnel
when concrete is intact ( ). In contrast, it will takes less than 20 years when the crack
connectivity coefficient ( ) increases to 0.9.
107
3.3 Crack width evaluation of a homogeneous tunnel lining
The study in this section will develop a new method to evaluate crack width in a tunnel lining
(research gap 3) as discussed in the conclusion section of Chapter 2.
The hydraulic conductivity of the tunnel lining was found to be determined by the crack
width and crack distribution. Crack width is often difficult to obtain since the crack width of
concrete is usually narrow and covered by precipitates for a wet crack. An analytical model is
presented to calculate the effective crack width in a lined tunnel based on the water inflow
rate and the other hydraulic parameters. This section is conducted under an assumption of
homogeneous crack distribution through the lining. This section presents a method to
evaluate the general cracking condition of a tunnel lining based on the transformation of the
previous study of water inflow into tunnel when the lining is homogeneous.
3.3.1 Profile of tunnel for effective crack width evaluation
A tunnel is defined in a cylindrical coordinate system for the calculation of the water inflow
rate of a tunnel lining. The origin of the coordinate system is defined as the centre point of
the tunnel. All points within the cylindrical coordinate system are defined by an angle , and
the distance (Figure 3.12).
108
×1m
Figure 3.12 Illustration of tunnel geometry parameters

Water inflow of a tunnel is affected by multiple parameters including, water table, lining
thickness, hydraulic conductivity of the surrounding rock mass and that of the tunnel lining.
Cracks in a concrete tunnel lining are studied as the main contributing factor for the hydraulic
conductivity of a tunnel lining. The water inflow rate has a cubic relation with the crack
width and a linear relation with the crack density has been discussed in Section 2.3.3. The
crack density has been defined as crack length divided by a concrete surface area. The
hydraulic conductivity of a concrete lining is calculated as a constant value under the
homogeneous assumption. Cracks through a concrete lining are considered to have a constant
crack width and crack density within the lining under the homogeneous condition. It is
notable that the lining thickness and the concrete surface area from the lining inner surface to
the lining outer surface are defined in cylindrical coordinates (Figure 3.12). Therefore the
reference area will change though the lining and a constant crack density does not mean a
constant crack length due to the varying concrete surface area through the lining.
109
3.3.2 Development of effective crack width model
The effective crack width of a crack from which water is flowing is difficult to obtain in a
tunnel site. This is primarily caused by the difficulty of crack accessibility and mineral
deposits formed on the surface of cracks as shown in Figure 3.13.
(a) (b)
Figure 3.13 Examples of a crack difficult to access for crack width measurement (a) and
crack difficult to measure width due to the deposit on crack surface (b)
The width of cracks in a concrete lining is critical for both water inflow prediction and
structural safety evaluation. The crack width in the lining contributes to the water inflow rate.
The crack width in this case is calculated based on the measured water inflow rate from
cracks. The flow of water into a tunnel, through the surrounding rock mass and the tunnel
lining, has important implications for tunnel design and tunnel condition assessment. The
hydraulic conductivity of the lining is a major controlling factor on the water inflow rate.
The hydraulic gradient is described as Eq. (3.19) for a water sealing tunnel with
homogeneous lining.
(3.19)
where, dh is the change in head and dr is the change in radius.

110
The water discharge per unit length of the tunnel for a homogenous lining (m3/s) is written
as Eq. (3.20).
(3.20)
where, is the hydraulic conductivity of the concrete lining (m/s).
The hydraulic conductivity of the concrete lining (K) is primarily determined by the effective
crack width and density. The equation is given based on Eq. (2.14) in Chapter 2 as follows.
(3.21)
For a permeable lining of a tunnel, the rate of water inflow was derived by Fernández and
Alvarez (1994) as in Eq. (2.30) in Chapter 2. This equation is rewritten as Eq. (3.22) using
the parameters in this study.
(3.22)
where, (m) is defined as water head at the outer surface at the spring line of the lining;
(m) is the water head at the inner surface at the spring line of the lining; a (m) is the
tunnel inner radius (from the centre to the inner surface) and d (m) is the lining thickness.
For a tunnel with no internal water pressure, the pressure head at the inner surface of the
tunnel lining is zero. The hydraulic head loss across the thickness of the lining is equal to the
pressure head at the outer surface of the lining (at the spring line). The datum level is defined
as the spring line of the tunnel (Figure 3.12).
111
The water head reduction caused by a lining, and the water head at the spring line at the outer
surface, was described as follows based on Eq. (2.32).
(3.23)
where, H is the vertical distance from the spring line of the tunnel to the water table (m) (or
equivalent phreatic pressure head); (m/s) is the equivalent hydraulic conductivity of the
lining; (m/s) is the hydraulic conductivity of the surrounding rock mass and is a
coefficient related to tunnel radius and lining thickness, which is defined as follows based on
Eq. (2.33).
(3.24)
Water inflow rate through homogeneous tunnel can be calculated based on the analytical
solution (Eq. (3.22)) when all the input parameters are available.
The required parameters include crack density and effective crack width as shown in Eq.
(2.14) in Chapter 2. The crack density, which is the accumulated crack length per tunnel area,
has been mapped at various locations during site inspections as shown in Section 4.5.1.
The relation between the hydraulic conductivity of lining (K) and the water flow rate of a
tunnel has been established previously. The analytical solution for homogeneous lining (Eq.
(3.22) to Eq. (3.24) is rewritten as Eq. (3.25) to calculate the hydraulic conductivity of the
lining.
112
(3.25)
where, K is the hydraulic conductivity (m/s) of the cracked tunnel lining, is the hydraulic
conductivity of the rock mass, all the other parameters have been defined previously.
The hydraulic conductivity of the lining (K) in Eq. (3.25) is on both side of the equation. The
hydraulic head at the inner side of a lining is 0 for a metro tunnel as all water flowing into the
tunnel is drained immediately. Equation (3.25) is further re-written as in Eq. (3.26).
(3.26)
The relation between crack width and concrete hydraulic conductivity has been validated as
Eq. (2.14) and discussed in Section 2.3.3. The relation between effective crack width and
other hydraulic parameters is obtained by combining Eq. (3.21) and Eq. (3.26). The
description of effective crack width is shown as Eq. (3.27).
(3.27)
√
where, (m) is the effective crack width, is the crack density (m/m2) which equal to
accumulated crack length perpendicular to the water flow direction over an area
perpendicular to the flow direction, is the crack roughness reduction factor, (kg/m3) and
(Pa.s) are the water density and viscosity, respectively.
113
The hydraulic parameters including water table and hydraulic conductivity of rock mass are
usually obtained by site test before tunnel construction. A relation between water flow rate
and the correspondent crack width is shown as Figure 3.14. The other hydraulic parameters
are based on the given information in Table 4.1 and Table 4.2.
3.00E-04
2.50E-04
Crack width (m)
2.00E-04
1.50E-04
1.00E-04
5.00E-05
0.00E+00
Water inflow rate (m3/s)
Figure 3.14 Relation between water inflow rate and crack width when the other
parameters are given
The hydraulic parameters underground are often highly variable when considering the joints
of a rock mas and the variation of groundwater table elevation. A parameter β is defined to
comprise all the hydraulic parameters based on Eq. (3.27) as follows.
(3.28)
These parameters vary in a range to cover a more generic case as shown in Table 3.4. It is
calculated that the β value in the range as provided in Table 3.4 varies from almost 0 to 1×10-
11
.
114
Table 3.4 Variation range of hydraulic parameters for β value
Min Max
3
Water flow rate Q (m /s) 1.00E-09 5.00E-04
Water table H (m) 5 30
Hydraulic conductivity of rock
1.00E-10 1.00E-05
mass Km (m/s)
Tunnel radius a (m) 2 6
Lining thickness d (m) 0.2 1.5
All variables used in the calculation are considered in a chart (Figure 3.15). The chart can be
applied to determine the crack width profile by monitoring the water inflow rate and crack
density over a section of a tunnel. The extrapolation of the crack width in Figure 3.15 extends
to 5×10-4 m (0.5 mm). Any crack width greater than 5×10-4 m (0.5 mm) is considered to be a
rare case. Those cracks which have a large width could potentially be measured on site
directly.
5.00E-04
4.00E-04
Crack width (m)
3.00E-04
2.00E-04
1.00E-04
0.00E+00
β value
Figure 3.15 Multiple-dimensional design chart for crack width evaluation
115
The analytical solution for crack width is simplified to three parameters, namely, crack width,
crack density and the β value which comprises all the other hydraulic parameters. The crack
width can be determined from Figure 3.15 when all the hydraulic parameters are known.
An analytical method is presented to calculate the crack width profile on a concrete tunnel
lining based on water inflow rate and the other hydraulic parameters. The water inflow rate is
measured on site and the other hydraulic parameters are known prior to the construction of a
tunnel. This study found that the width of cracks in the selected tunnel follows a lognormal
distribution with a mean value approximately 0.12 mm. The maximum width of cracks based
on the measured flow rate in the selected tunnel is calculated to be approximately 0.29 mm.
The probabilistic data of crack width is used in the reliability chapter (Chapter 5).
The hydraulic parameters used in the selected tunnel calculation are deterministic values. The
water table and the hydraulic conductivity of rock mass are often highly variable for many
tunnels. A multi-dimensional design chart which comprises all the variable parameters is
presented in this study. The multi-dimensional design chart developed in this study can be
used to determine concrete crack width when all the other hydraulic parameters are known.
3.4 Water inflow into a tunnel with inhomogeneous lining
The study in this section will develop a solution to incorporate the potential inhomogeneity of
hydraulic conductivity of a tunnel lining (research gap 4) as discussed in the conclusion
section of Chapter 2.
Previously published work has been mainly based on the assumption of constant hydraulic
conductivity of tunnel linings. The water inflow rate was measured as a known factor to back
calculate the lining hydraulic conductivity and then calculate the effective crack width in the
study of the previous section.

116
This section will develop a solution for the hydraulic behaviour of a lining when the
hydraulic conductivity of lining is inhomogeneous. The study in this section will incorporate
a linear variation of hydraulic conductivity with distance from the inner surface of a tunnel
wall. The model in this section can be used to predict the hydraulic pressure at the outer face
of the lining with significantly increased accuracy compared with the existing models based
on constant hydraulic conductivity. Design charts are also developed for engineering
applications.
Most approaches for solving tunnel water inflow problems assume that the lining has a
homogeneous hydraulic conductivity. The surrounding rock mass is also commonly assumed
to have a homogeneous hydraulic conductivity, although two studies with inhomogeneous
properties have been conducted (Table 3.5). Schematic illustration of the hydraulic
conductivity distribution of the three types of tunnel under homogeneous conditions is shown
in Figure 3.16.
Table 3.5 Framework for water inflow prediction models
Lining/Lining-like
Type of Tunnel Rock mass zone Reference
Homogenous Goodman et al., 1965
Lei, 1999
Unlined
El Tani, 2003
Inhomogeneous Zhang and Franklin, 1993
Homogenous Homogenous Kolymbas and Wagner, 2007
Park et al., 2008
Lined Drained Homogenous Inhomogeneous None
Inhomogeneous Homogenous None
Inhomogeneous Inhomogeneous None
Homogenous Homogeneous Fernandez and Moon, 2010a
(unlined tunnel with a lining-like
zone)
Lined water Homogenous Inhomogeneous Addressed in this study
sealing Inhomogeneous Homogenous Fernandez and Moon, 2010b
(unlined tunnel with a lining-like
zone)
Inhomogeneous Inhomogeneous None
117
Type of Tunnel Illustration
Unlined
Lined Drained
Lined water
sealing
Figure 3.16 Schematic illustration of hydraulic conductivity of three types of

homogeneous tunnel
Calculations of water inflow are very sensitive to the hydraulic conductivity of the lining and
inhomogeneity of the lining can potentially cause significant differences in the results. The
hydraulic conductivity of a lining is typically deduced from observations of joint and crack
118
features on the inner face of the lining. Such features can vary through the lining which
would affect the hydraulic conductivity along the water seepage path. However such
variation within the lining cannot be directly observed, and thus there is a need for methods
to predict and account for this.
This study presents an analytical solution for water inflow rate and hydraulic pressure
distribution in tunnel linings with inhomogeneous hydraulic conductivity, assuming saturated
conditions and steady-state flow. The solution uses the hydraulic conductivity of the surface
of the lining inside the tunnel as the reference value. Variation in hydraulic conductivity is
considered linear within the lining. In combination with the observations of cracking at the
inner surface and the water inflow rate, this model can be used to estimate the level of
inhomogeneity within the lining including the water pressure and as the crack condition at the
outer face of the lining.
3.4.1 Development of water inflow solution of a inhomogeneous lining
Previous research only considered the condition when the lining is homogeneous, i.e. K is a
constant. A homogeneous hydraulic conductivity is a reasonable assumption when a lining is
thin and inhomogeneity of the lining would not affect water inflow greatly. The
inhomogeneity of lining needs to be considered when the lining is relatively thick. For a
lining which is inhomogeneous in terms of hydraulic conductivity, the hydraulic conductivity
K(r) is a function of r. A study of crack spacing in reinforced concrete showed that
differences in strain in reinforcing elements, for example, would result in a linear variation of
crack spacing (Bazant and Oh, 1983). The hydraulic head variation caused by the lining
inhomogeneity is shown below (Figure 3.17).
119
Figure 3.17 Head loss under both homogeneous (right) and inhomogeneous (left)
conditions
The flow of water through a tunnel lining needs to be understood in terms of the area across
which the flow occurs. In simple terms, the water flows from the outer surface of the tunnel
to the inner surface of the tunnel progressively passing through an area defined by the
circumference of a circle. The flow area per metre length of tunnel is equal to the
circumference of the lining times unit length of tunnel. The flow area through the lining
decreases progressively with the value of r, as shown in Figure 3.12. Referring to Darcy’s
law, the water inflow rate for a unit arc length of a unit length of tunnel lining ( ) is written
as
(3.29)
120
where, is the flow area of unit arc length of a unit length tunnel (m2) and is the
hydraulic conductivity of the lining (m/s), which varies with distance from the tunnel centre.
An inhomogeneous hydraulic conductivity coefficient can be defined as
(3.30)
where, and is the hydraulic conductivity (m.s) of the inner surface of the lining
and outer surface of the lining, respectively. The value of can potentially be
determined directly by observation and measurement of crack density and crack width as Eq.
(2.14) in Chapter 2.
The hydraulic conductivity of the outer surface of the lining ( ) typically cannot be
determined directly by observation. The hydraulic conductivity of any position in the lining
can be represented as
(3.31)
The function that contains radius r can be rearranged as
(3.32)
where and .
Based on Eq. (3.29), the hydraulic head through the lining can be written as
(3.33)
Integration of Eq. (3.33) gives
121
(3.34)
[ | |]
where, is a constant.
To determine Eq. (3.34), the boundary conditions need to be considered. For a lined tunnel
with relatively low water seepage velocity, it is assumed that any frictional loss at the rock
mass-lining interface is very small and can be neglected. When the water head loss along the
rock tunnel interface is negligible, the decrease of elevation is equal to the increase of
pressure head along the interface. Therefore, the hydraulic head (total head) which is equal to
the summation of the pressure head and elevation head at the boundary of the rock and lining
is considered a constant. Since there is no water pressure at the inner surface of the lining, the
total head at the inner surface is equal to the elevation head when the datum level is the
spring line of the tunnel. Therefore the boundary conditions (Kolymbas and Wagner, 2007)
are as follows.
| (3.35)
| (3.36)
The boundary conditions Eq. (3.35) and Eq. (3.36) are used to determine water head within
the lining with respect to the cylindrical coordinate system.
Substituting the boundary condition Eq. (3.35) into Eq. (3.34), the value of C can be obtained
as
(3.37)
[ | |]
122
Substituting the boundary condition Eq. (3.36) into Eq. (3.34), the water head can be further
developed as
(3.38)
[ | |]
Equation (3.38) can be rewritten as
(3.39)
[ | | [ | |]]
Considering and , Eq. (3.39) can be further
developed as
(3.40)
[ | | | |]
The equation of water flow rate across a unit arc length of the lining can be expressed by
simple transformation of Eq. (3.40) as
(3.41)
| |
The total water inflow rate of a unit length of tunnel with a cylindrical lining can be
obtained by integrating , to give
(3.42)
| |
123
If the tunnel has a constant total head at the interface of rock and lining, previous works as
discussed in section (2.4.5) have shown that the water flow rate through the rock mass can be
written as
(3.43)
Where, D is the distance from tunnel spring line to the water table or the ground surface (m),
depending on which one is smaller; the hydraulic conductivity of the surrounding rock mass
is (m/s).
Since water is considered to be an incompressible fluid in this study, water flow rate through
a lining under saturated conditions is equal to the water flow rate through the rock mass,
based on mass conservation.
Combining Eq. (3.42) and Eq. (3.43), the following equation can be obtained
(3.44)
| |
√
Transforming Eq. (3.44), the water head at the outer surface ( ) of the lining can be
expressed as
(3.45)
| |
124
Substituting Eq. (3.45) into Eq. (3.42), the water inflow rate per unit length of tunnel can
be expressed as
(3.46)
√
| |
| |
[ ]
All the parameters have been defined above.
To calculate the hydraulic head in a lining, Eq. (3.37) and Eq. (3.41) are substituted into the
water head distribution function Eq. (3.34) and combined with the water head at the outer
surface of the lining (Eq. (3.45)). The water head through the lining can be calculated as
(3.47)
| |
| |
For any position within the lining with given and , the water head in the lining can be
calculated accordingly.
3.4.2 Hydraulic pressure and water inflow rate of lining
Parameters used in application
An selected tunnel (Figure 3.18) was selected for the purpose of illustrating the influence of
variable hydraulic conductivity through a lining. The selected tunnel is cylindrical and
approximately 35 m underground. The water table is approximately 15 m above the spring
line of the tunnel. The radius ( ) of the tunnel from the tunnel centre to the inner surface of
the lining is 3000 mm and the thickness of the cast in situ concrete lining ( ) is 560 mm. The
tunnel lining comprises steel beams as primary support embedded in the cast-in-situ lining
125
near its outer surface. Steel reinforcement bars were cast in the concrete close to the inner
surface of the lining. The primary lining also included application of shotcrete to the tunnel
wall prior to the installation of the secondary lining. The primary and secondary lining
elements are together considered to comprise the lining. The hydraulic conductivity can be
variable through the lining due to the different allocation of reinforcement as in Figure 3.18.
Figure 3.18 Illustration of the example concrete lining with steel reinforcement elements
(a) transverse section (b) longitudinal section
The tunnel was designed as a water-sealed tunnel to prevent water table draw down. Even
though the concrete lining was designed to minimise the water ingress, cracks in the concrete
lining have been observed. The mean crack width ( ) is assumed to be 0.3 mm, and the
crack density is assumed to be 0.5 m/m2. The hydraulic conductivity of the inner surface of
126
the lining is approximately 1×10-8 m/s based on the observed crack density and crack width.
The mean value of hydraulic conductivity of the surrounding rock mass is approximately
1×10-7 m/s based on previous geological surveys (Robinson and Kenna, 1992).
It has also been reported generally, that the type of reinforcement in a concrete lining can
result in variations of crack width and crack density in different parts of the concrete (Bazant
and Oh, 1983). While observations of crack density and crack width exposed on the inner
surface of the lining of the tunnel were used to estimate the hydraulic conductivity, the crack
density and crack width within the lining (and therefore potential variation in hydraulic
conductivity within the lining), is not known. Based on the distribution of support elements
within the concrete lining, inhomogeneous hydraulic conductivity is considered a likely
condition for the lining of the selected tunnel.
In this study we incorporate the potential variability of hydraulic conductivity through the
lining into the model, as described in Section 3.4.1. We consider the linear inhomogeneous
hydraulic conductivity coefficient across a range from 0.01 to 100. The concept of is
proposed as a practical way to frame the problem as the condition of the inner surface of a
tunnel lining can be observed by direct inspection or by use of devices such as televiewers.
Water head variation
The result of the analytical solution (Eq. (3.46)) shows the water head behaviour at the spring
line through the lining for values of 0.01, 0.1, 1, 10 and 100 (Figure 3.19). For the
homogeneous case ( ) it can be seen that the water head distribution is slightly non-
linear due to the effect of flow convergence described above (Figure 3.19). For cases of
inhomogeneous hydraulic conductivity coefficient ( ) the water head variation through the
lining is distinctly non-linear (Figure 3.19). When >1, the hydraulic gradient decreases
127
from the inner surface to the outer surface. This relationship becomes more pronounced as
increases. The relationship is due to the relatively minor influence of the flow area change
and relatively major influence of the increase in average hydraulic conductivity, represented
by >1. In contrast, the hydraulic gradient decreases through the lining when <1.
Figure 3.19 Relation of the water head at different lining position r-a with different
value when =10. The horizontal axis is the distance outward from the inner
surface of the lining, equal to r-a. Parameters derived from selected tunnel
Increasing represents an increase in hydraulic conductivity at the outer surface while the
hydraulic conductivity at the inner surface is fixed. Therefore, as increases, the average
hydraulic conductivity of the lining would also increase. The average or equivalent hydraulic
conductivity of the lining has not been defined or used in this study because it cannot be
128
independently verified. Rather, the hydraulic conductivity at the inner surface has been used
as the reference (as this can be estimated using direct observation) and hydraulic conductivity
at the outer surface is a variable factor defined according to . The variation of hydraulic
conductivity in an inhomogeneous lining will cause hydraulic head redistribution at the outer
surface. The water head at the outer surface (x axis=0.56, Figure 3.19) will decrease if the
average hydraulic conductivity of lining decreases based on Eq. (4). In general, lower average
hydraulic conductivity of the lining will allow greater hydraulic pressure at the outer surface,
if all other parameters are constant.
Figure 6 shows the water head at the outer surface for various values of . When increases
from 0.01 to 1, the water head at the outer surface decreases. When increases from 1 to
100, the water head at the outer surface of the lining drops from 6.7 m to less than 1 m. The
homogeneous case represents an inflection point on the curve representing the range of
inhomogeneous conditions (Figure 3.20).
129
Figure 3.20 The relationship of water head and inhomogeneous hydraulic conductivity
coefficient at outer surface of the lining. Parameters derived from selected tunnel
The homogeneous point means the hydraulic conductivity through the lining are constant.
Water inflow rate variation
In addition to influencing water head distribution, the value will also influence the water
inflow rate. When increases from 0.01 to 1, the water inflow rate increases (Figure 3.21).
The variation of water inflow rate is controlled by both hydraulic conductivity and hydraulic
head. The homogeneous condition again represents an inflection point in the interplay of
these parameters (Figure 3.21). For the homogeneous condition the calculated water inflow
rate for the selected tunnel is 2.46×10-6 m3/s. When increases from 0.01 to 100, the water
inflow increases, ranging from 9.7×10-7 m3/s to 4.3×10-6 m3/s, respectively.

130
By using Figure 3.21, the overall condition of the lining can be predicted based on the
observed water inflow rate. An example of water inflow rate data collected in the tunnel is
studied based on the Figure. The measured water inflow rate is 1.64×10-6 m3/s. Based on
Figure 3.21, the projected value is 0.16, indicating the hydraulic conductivity increases in
the direction of flow, from the outer surface to the inner surface of the tunnel lining.
5.00E-06
5 10-6
4 10-6
4.00E-06
3 10-6
3.00E-06
2 10-6
2.00E-06
Homogeneous
1 10-6
1.00E-06
0
0.00E+00
0.01 0.1 1 10 100
Inhomogeneous hydraulic conductivity coefficient (Cl)
Figure 3.21 The relationship of water inflow rate and inhomogeneous hydraulic
conductivity coefficient at the outer surface of the lining when varies from 0.01 to
100. Parameters derived from the selected tunnel
The homogeneous point means the hydraulic conductivity through the lining are constant.
It is noted that an increase of would increase the water inflow rate even beyond the
discussed range. The variation discussed in this section aims to illustrate a range of two
131
orders of magnitude each side of the homogeneous case. The observed variation within that
range indicates that beyond the range from 0.01 to 100 there would be less significant
water inflow variation.
Design chart and crack prediction
In an engineering application such as a tunnel, the parameters are expected to vary, for
example, along the length of the tunnel. Parameters such as depth to water table and
hydraulic conductivity of the rock mass and the tunnel lining would have a range of values.
To further demonstrate the application of the analytical equation, a water inflow factor has
been defined. This is a dimensionless parameter comprising four main variables, water inflow
rate per length of tunnel ( ), lining thickness ( ), total water head above the spring line of
the tunnel ( ) and the rock mass hydraulic conductivity ( ). One dependent variable ( ) is
also included in the water inflow factor. The dependent variable ( ) is a simplification of the
intersection of the main variables. The other term used in the chart is the relation between
rock mass hydraulic conductivity ( ) and hydraulic conductivity at the inner surface
( ). The relationship between water inflow factor and the inhomogeneity of a tunnel
lining is shown in a chart that can cover a range of hydraulic conductivity conditions (Figure
3.22).
132
Figure 3.22 The water inflow factor chart for parameters of the selected tunnel. An
example of collected data is shown
Since the dependent variable ( ) is a simplification of the intersection of basic variables, the
chart is not perfectly accurate; to quantify the inaccuracy a comparison of the calculated
water flow value and the chart-derived value is presented (Figure 3.23). The input parameters
used in the comparison were taken from an arbitrary range shown in Table 3.6. The result
shows a strong correlation between calculated values and the simplified chart (Figure 3.23).
133
Table 3.6 Parameter ranges used for validation of the water inflow factor chart
(m/s) (m/s) (m) (m) (m)
Low 1×10-7 1×10-9 2.8 0.5 5 0.01

Middle 1×10-6 1×10-7 3 0.56 15 1
High 1×10-5 1×10-5 3.2 0.6 20 100
The design chart is generated based on the analytical solution of water inflow as in Eq. (3.46).
The design chart is valid as long as the analytical solution holds. The dependent variable ( )
is generated and calibrated based on a selected range of hydraulic parameter as shown in
Table 3.6. A larger error value is possible for the hydraulic parameters beyond the calibrated
range.
Figure 3.23 Water flow rate is calculated compared with the water inflow rate
according to the water inflow factor chart
134
A measurement of water flow from the inner surface of the selected tunnel is plotted on
Figure 3.22 to illustrate the application of this chart. For this specific location, the value
has been obtained from Figure 3.22. This value represents the relation between inner and
outer lining hydraulic conductivity. The variation of hydraulic conductivity can be attributed
to crack density variation or crack width variation according to the cubic law (Figure 3.24).
The procedure for using such data (Tan et al., 2016) is shown in Table 3.7.
Table 3.7 Procedure for using water inflow factor chart
Steps General explanation Application in selected tunnel

1 Measure the water inflow rate ( ) per Measured local water inflow rate is
length of the tunnel. 1.64×10-6 m3/s (0.099 L/min).
2 Calculate the equivalent hydraulic Inner surface crack width ( )
conductivity of the inner surface ( ) is 0.3 mm and inner surface crack
based on observed crack width and crack density ( ) is 0.5 m/m2, the
density. calculated hydraulic conductivity of
inner surface is approximately
1×10-8 m/s.
3 Calculate water inflow factor based on the Water head is approximately 15 m,
water head ( ), lining thickness ( ), lining thickness is 0.56 m, and the
dependent variable ( ) and the hydraulic hydraulic conductivity of rock mass
conductivity of the rock mass ( ). is approximately 1×10-7 m/s based
on literature. Water inflow factor is
calculated to be 0.62.
4 Determine the ratio between hydraulic Based on step 2) and 3), the ratio
conductivity of the rock mass ( ) and the ( / ) is 1×101.
equivalent hydraulic conductivity of the
inner surface ( ).
5 Identify the intersecting point of y axis The intersecting point is shown in
value and / curve, and then find Figure 3.22, the value is found to
value. be 0.16.
6 Predict the crack features through concrete Inner surface crack width ( )
lining according to the calculated value. is 0.3 mm, and value is 0.16 in
The relation between hydraulic this example. For a constant crack
conductivity of cracks features and density is 0.16 mm. For a
hydraulic conductivity is quantified by constant crack width is
Reinhardt (1997). approximately 0.35 m-1 (Figure
3.24).
135
Figure 3.24 Relation between crack features and inhomogeneous hydraulic conductivity
coefficient ( ). Parameters for the selected tunnel and collected data are marked
Models of hydraulic conductivity of tunnels
The typical conditions of hydraulic conductivity variation in a tunnel lining can be divided
into three general types. The first type is homogeneous lining ( =1), where the hydraulic
conductivity through the lining is constant. This type of lining has been studied previously
(Fernandez and Moon, 2010a). The water flow rate through the homogenous lining ( ) is
also used as a reference value to contrast the flow rates when the hydraulic conductivity is
136
inhomogeneous in this study. The hydraulic head and water inflow rate under homogeneous
condition is marked to compare with the inhomogeneous condition in Figure 3.20 and Figure
3.21.
The second type is the constrictive lining ( >1). In this type of lining the hydraulic
conductivity decreases from the outer surface to the inner surface of the lining. The
decreasing trend can be continuous or non-continuous. In this study, we have discussed a
circumstance when the decreasing trend of hydraulic conductivity is continuous and linear.
We showed that the hydraulic head difference through the lining decreased from
approximately 7 m to less than 1 m when increased from 1 to 100 as in Figure 3.20. The
flow rate is almost doubled when increase from 1 to 100 as in Figure 3.21.
The third type is the expansive lining ( <1). In this type of lining, the hydraulic conductivity
increases from the outer surface to the inner surface of the lining. The expansive water flow
could induce unsaturated conditions, which would be more complex compared to the
homogeneous and constrictive conditions.
The hydraulic conductivity of the surrounding rock mass is usually higher than that of the
concrete lining, as in our case study tunnel. The condition when the rock mass is less
permeable than the concrete lining is not considered in this case. Three conditions with
respect to the hydraulic conductivity of the tunnel lining are illustrated as graphs in Figure
3.25.
137
(a) (b) (c)
Figure 3.25 Illustrations of types of lining and rock mass hydraulic conductivity
conditions. (a) Lining hydraulic conductivity homogeneous; (b) Lining hydraulic
conductivity constrictive; (c) Lining hydraulic conductivity expansive
The approach in this study analysed all the three lining hydraulic conductivity conditions. In
a practical engineering application, it is often difficult to identify which of the three cases a
lining belong to. This information can be obtained by using the design chart as shown in Fig.
8. When the identified >1, the cracking condition inside the lining is inferred to be worse
(more cracks or wider cracks) than it appears on the lining inner surface. When the identified
<1, the cracking condition inside the lining is inferred to be better than it appears on the
lining inner surface. The distribution of cracking along and within a lining could be
interpreted with a range of value readings along a tunnel. For example, a tunnel with
differential settlement could have different values along its length. The collected water
inflow data from the case study tunnel represents a value smaller than 1 (Figure 3.25(c)).
Hence, the hydraulic conductivity of the inner surface is predicted to be higher than the outer
surface of the lining, according to our model.
The case studied is a tunnel designed to be water-sealed with no drainage layer. However, in
modern tunnels, a drainage layer is typically incorporated to drain away the potential inflow
water and relieve the hydraulic pressure from the lining. In a drained tunnel, the primary and
secondary lining is separated by a drainage layer. In such cases, the secondary lining is
138
considered effectively impermeable as most of the water pressure has been relieved by the
drainage layer. The hydraulic conductivity of the drainage layer is designed to be higher than
that of both the lining and the rock mass. Impermeability of the lining is an idealised situation
even where a drainage layer is present. The flow models of tunnels with a drainage layer for
comparison with the simpler examples are shown in Figure 3.26.
(a) (b) (c)
Figure 3.26 Illustrations of types of hydraulic conductivity conditions of a tunnel with a

drainage layer. (a) homogeneous lining hydraulic conductivity; (b) constrictive lining
hydraulic conductivity; (c) expansive lining hydraulic conductivity
Water flow rate in an inhomogeneous lining
The water flow rate through an inhomogeneous lining can be affected by the inhomogeneity
of the lining as demonstrated above. The overall inflow rate (flux) through a lining depends
on both hydraulic gradient and hydraulic conductivity of the lining; under saturated
conditions, the flux is a combination of hydraulic gradient and hydraulic conductivity,
according to Darcy’s law. For an underground, lined tunnel under steady state flow, variation
of hydraulic conductivity will cause corresponding variation of the hydraulic gradient
through the lining. An increase in value will decrease the overall hydraulic conductivity of
the tunnel lining, and the inflow rate will also decrease. Variation of basic variables like
139
water table and rock mass hydraulic conductivity will also cause water flow rate variation.
This study developed a multiple parameter chart to quantify the effect of these different
parameters on water inflow rate (Figure 3.22). The parameter values (e.g., ) can also be
inferred if the water flow rate is measured.
Water pressure in an inhomogeneous lining
The water pressure distribution in a concrete lining can also be affected by the inhomogeneity
of the lining. This effect has not been considered in previous research. Based on our study, if
the value is greater than 1, the water pressure at the outer surface is lower than for the
homogeneous condition (e.g., Figure 3.19). However, the water pressure close to the inner
surface could be higher than for the homogenous condition. The hydraulic pressure in the
lining is an important parameter for structure stability analysis. Higher than expected water
pressure close to the inner surface could trigger concrete spalling and then lead to structural
problems (Jansson and Boström, 2010). For a selected location (Table 3.7), the value is
calculated to be 0.16, based on the collected water inflow rate and other hydraulic parameters.
According to that value and Figure 3.20, the hydraulic head at the outer face of the lining
is approximately 9.8 m in this method. In contrast, the hydraulic head at the outer face of the
lining would be approximately 7.5 m for a homogeneous lining, which would underestimate
the hydraulic pressure by approximately 30%.
Outer face crack prediction
The degree of cracking of the outer face of a lining can be different from that of the inner
face of the lining as discussed above. Variation of the degree of cracking will affect lining
hydraulic conductivity causing water inflow variation. The relation between water inflow
variation and the degree of cracking is quantified in this study. Based on our study, when the
140
value is greater than 1, cracks at the outer face of the lining are wider or more abundant
than at the inner face of the lining. Cracks at the outer lining are narrower or less abundant
than at the inner face of the lining when the value is less than 1. The degree of cracking at
the outer lining is predicted for one selected location based on the water inflow rate and other
hydraulic parameters (Table 3.7). The crack width and crack density of the inner face of the
lining at the selected location is 0.3 mm and 0.5 m/m2 respectively. Based on a homogeneous
lining, the crack formation of the outer face of the lining would be assumed to be the same as
the inner face. In the method presented in this study, the cracking at the outer face of the
lining is predicted to have narrower crack width (0.16 mm) or less abundant crack density
(0.35 m/m2) when other hydraulic parameters are fixed.
The analytical model of water inflow rate into a cylindrical tunnel under steady state,
saturated conditions has been extended to the case of tunnel lining with linearly
inhomogeneous hydraulic conductivity. The linear change of hydraulic conductivity of the
lining is represented by an inhomogeneous hydraulic conductivity coefficient, . This
coefficient is used to quantify the ratio of the hydraulic conductivity at the outer surface and
inner surface of the lining. Features such as cracks and joints which may allow water to flow
through a concrete lining can be observed on the inner surface of a tunnel lining. The
hydraulic conductivity of the inner surface of the lining can be calculated from the observed
crack density and crack width. The approach proposed in this study is to project the hydraulic
conductivity from the inner surface of the lining into the body of the lining with a continuous
linear distribution.
Using the parameters of a selected tunnel, the water head distribution through the lining was
calculated for a range of values. The tunnel lining is considered to be homogeneous when
the value equal to 1. Water inflow rate and the hydraulic head of the homogeneous
141
condition are used as the reference to make comparison to the inhomogeneous condition. As
the value increases, the hydraulic conductivity of the lining also increases.
Correspondingly, the water head near the outer surface of the lining decreases. In general, the
assumption of homogeneous hydraulic conductivity of a tunnel lining can result in water
pressure at the rock-lining interface being under-estimated by as much as 90% if the value
is less than 1. Conversely, the assumption of homogeneous hydraulic conductivity of a tunnel
lining can result in water pressure near the inner surface of the lining being under-estimated
by as much as 30% if the value is greater than 1. This finding has implications for stability
analyses as the water pressure in and behind the tunnel lining is an important contribution to
stability of a tunnel lining. A water inflow factor chart relating several different parameters
has been developed to enable these parameters to be assessed for a given case. A method to
project the cracking condition into the concrete lining using the water inflow factor chart has
been developed. The cracked concrete within the lining is described by the value. The
value can be determined once the other hydraulic parameters including, water table elevation
and water inflow rate are known. An example has been demonstrated to show the procedure
for using the water inflow factor chart and the degree of inhomogeneity of a tunnel lining,
due to cracking, has been predicted by measuring water inflow.
3.5 Summary
This chapter provided a study of water inflow through a tunnel lining. This chapter comprised
three parts for this topic. The three topics are present in three sections from Section 3.2 to 3.4
which correspond to three research gaps as discussed in the literature review chapter.
A method is developed in Section 3.2 for calculating the hydraulic conductivity of a concrete
with partially interconnected cracks (research gap 1). The method developed in this chapter is
used to calculate the hydraulic conductivity of partially cracked concrete in further chapters.
142
A method is developed in Section 3.3 for estimating the crack width in a tunnel. The
measured site data were used in the calculation (research gap 3). The crack width of the
concrete lining in the selected tunnel is evaluated using the new developed method.
A method is developed in Section 3.4 for calculating the water inflow rate and the hydraulic
pressure within a tunnel with an inhomogeneous lining. The variability of hydraulic
conductivity of a tunnel lining was quantified by a linear inhomogeneity coefficient. A design
chart for engineering application is presented in Section 3.4. This study developed a solution
to incorporate the potential inhomogeneity of hydraulic conductivity of a tunnel lining
(research gap 4) as summarised in Section 2.6.
143
4 Site Investigation
4.1 Introduction
This chapter provides details of the site investigation which has been conducted in the
selected tunnels. There are four parallel selected tunnels in this case, which will be referred to
as BLP, NLP, CCL and CLP. Each tunnel was divided into four sections by three
underground stations. The underground stations will be referred to as FGS, MCS and PS. The
length of each tunnel is approximately 2.5 km. BLP and NLP are the deeper tunnels
compared to CCL and CLP.
The average value of the hydraulic conductivity of the rock mass is approximately 1×10 -7 m/s
based on the literature (Peck, 1992). The tunnel radius is approximately 3 m with a 0.56 m
thick lining based on the original design drawings.
Table 4.1 Rock mass hydraulic conductivity and tunnel dimensions
Parameter (m/s) (m) (m)
Value 1×10-6 3 0.56
The water table for different sections of the selected tunnels has been interpreted from the
existing ground water survey in Victoria (Leonard, 2006) and the alignment of the four
tunnels. The water table varies along tunnel sections. The representative values of hydraulic
head for each section of the tunnels are shown in Table 4.2 based on the local water table and
the elevation of different tunnel section.
144
Table 4.2 Hydraulic head of tunnel sections (in m above centre line)
Portal to FGS FGS to MCS MCS to PS PS to Portal
CLP 3 5 9 4
CCL 3 5 6 4
BLP 7 14 18 8
NLP 7 14 15 8
The deepest section of the tunnel system is between stations MCS to PS. The BLP and CLP
are deeper in this section due to the placement of the sump and pumping system. The
elevation of the tunnels starts to rise as they approach the portals of each end which is
reflected in Table 4.2.
It was found from the site investigation that water inflow into the tunnel has occurred through
cracks in the lining. A method is developed to calculate the water inflow rate from cracks
with limited measured data (research gap 2) in this chapter. The procedure for water inflow
measurement is further discussed in Section 4.2. Water inflow rate for the selected tunnels is
then measured on site using this method.
The water inflow rate measurement was conducted in BLP and CLP between MCS and PS.
At the measurement locations water inflow was visible and measureable. The compressive
strength of lining concrete was tested in two tunnels in various locations. The concrete
strength measurement locations are at the deepest section of the tunnel between MCS and PS.
The crack pattern of the lining was mapped in a 50 m section between MCS to PS in all four
tunnels. The number of wet cracks was monitored by the tunnel operator in eight site
inspections through the whole tunnel over three years. The crack inspection data is
summarised in this chapter. Solid samples of mineral deposits and water samples were
145
collected in the selected tunnels during the site investigation. The chemical content of the
water and the precipitates around cracks can be used as an indication for chemical reactions
in concrete lining as discussed in Section 2.2.4. The chemical content of the collected water
and the precipitates was tested and interpreted. The crack pattern and the crack changing rate
recorded in inspections are investigated in this chapter.
4.2 Techniques for site investigation
The techniques for the site investigation conducted by the author are discussed in this section.
The investigations in the tunnels include four main aspects, namely, water inflow
measurement, lining compressive strength field test, crack feature investigation and chemical
analysis of samples. All these four items require site investigation. A total of eight site
inspections were conducted, including visits to all four of the parallel tunnels from 2014 to
2016. The compressive strength of the concrete lining was investigated along with recording
of crack features of the lining and sample collection in the first five inspections in 2014.
Water inflow measurement was conducted in the last three site inspections from 2015 to 2016.
Table 4.3 Summary of site inspections
Date Tunnel Activities

Inspect general tunnel condition, inspect inner structure of tunnel,
16/08/2014 BLP
Crack mapping, Collect samples
24/08/2014 CLP
7/09/2014 NLP
19/10/2014 CCL
Crack mapping, Collect samples, Test concrete strength
26/10/2014 BLP
Crack mapping, Collect samples, Test concrete strength
7/11/2015 CLP Water inflow measurement
16/11/2015 BLP Water inflow measurement
13/10/2016 BLP Water inflow measurement
146
The objective for water inflow measurement is to measure the local water inflow rate from
cracks and joints. These data are correspondent to the water inflow rate (Q) as discussed in
Section 3.3. These data are used to calculate the effective crack width ( ) in tunnel as
discussed in section 3.3.
The objective of compressive strength field testing is to measure the concrete compressive
strength of ( ) as discussed in Section 3.2. These data are used to calculate the hydraulic
conductivity of sound concrete ( ) in the tunnel as discussed in section 3.2.1.
The objective for crack feature measurement is to record the crack density and the crack
geometry information in the selected tunnels. These data are correspondent to the crack
density ( ) as discussed in Section 3.3. These data are used to calculate the water inflow rate
in a selected section of the tunnels.
The objective for water and mineral sample collection is to have a general understanding of
the lining properties in the selected tunnels. These data are correspondent to the ambient
environmental condition as discussed in Section 2.2.
The technique for conducting each item on site is discussed as follows.
4.2.1 Technique of water inflow measurement
In order to measure the water inflow rate along a tunnel wall, a collector (Figure 4.1), a
measuring cylinder and timer is used. The width of the collector is a known value Lc. The
edge of the collector is placed against the tunnel wall where water is flowing. The water flow
rate for the finite length (Lc) is obtained.
147
Figure 4.1 Device used for surface water collection
The flow area could be wider than the width of the collector. Using a collector will only
measure a portion of the water inflow of one location. It is not practical to measure the entire
flow for one location since the inspection time is very limited. Therefore, it is necessary to
make an estimation of the total flow rate of one flow location based on measurement of a
portion of the flow. The method for calculating the total water inflow rate of one location
based on the measured portion of the inflow rate is further discussed in Section 4.3.
4.2.2 Technique of concrete compressive strength test
The concrete compressive strength of the selected tunnel lining was tested by using Schmidt
hammer. The compressive stresses of three types of locations were tested, namely, plinth,
walk way and tunnel lining as shown in Figure 4.2.
148
Figure 4.2 Illustration of plinth, walk way and tunnel lining in a tunnel
Two investigations in 2014 for the compressive strength test of tunnel lining were conducted
in in CCL and BLP using Schmidt hammer. The Schmidt hammer was held with different
angles against the concrete and values were adjusted according to the instructions provided
with the apparatus. The outer and inner in this collection are referring to the inner side wall
and outer side wall relative to the tunnel curvature. An alternative description of inner and
outer is the wall on the left hand side is the outer wall when facing direction is the same as
the tunnel distance reference (chainage) increasing direction.
4.2.3 Technique of crack feature investigation
The surface features of crack were mapped in a 50 m section in a similar location for all the
four tunnels. The mapping location is between MCS to PS stations which is also the deepest
the section of the whole tunnel system.
149
The mapping was used to record the relative location, orientation and length of cracks on the
lining surface as shown below. The surface crack mapping shows the crack pattern on the x-z
plane as in Figure 3.1.
Eight inspections over three years were conducted by the operating company to monitor the
number of cracks in each section of the selected tunnels. The description includes recognition
if the crack is wet, meaning water can be observed emanating from the crack. The cracks are
recorded according to their location given as a chainage distance. The number of wet cracks
of one tunnel component is a key parameter for determining crack density. The number of
wet cracks observed in the lining of the four tunnels varied in each inspection. In some cases
new cracks were identified, whereas in other cases existing cracks changed from dry to wet or
vice versa. The change of number of wet cracks per 100 m of tunnel ( ) of each tunnel was
recorded based on the crack data recorded from 2013 to 2016. The increasing rate of number
of wet cracks is defined as the slope of the linear regression. It is not implied that the rate of
change is linear – only that the current amount of data does not support more complex
interpretations.
The crack increasing trend obtained in this section indicates a time dependent increasing of
water inflow rate over time which will be further used in Chapter 5 for reliability calculations.
4.2.4 Technique of chemical analysis of samples
The solid precipitates and water were collected from selected locations in the tunnels.
Approximately 30 samples were collected by the author during the site inspections. The main
purpose of sample collection is to identify the potential chemical reaction in the tunnel lining
based on the chemical content.
150
The microstructure of each solid sample was inspected using Scanning Electron Microscopy
(SEM). The image of the SEM for each sample is shown in Appendix 1. The composition is
confirmed by both X-ray diffraction (XRD) and Energy-dispersive X-ray spectroscopy (EDX
or EDS) as shown in Appendix 1. The calcium ion content and iron ion content of the water
samples were tested by Atomic Absorption Spectroscopy (AAS).
The chemical content of water (including pH and salinity) and solid precipitates can be used
to indicate the chemical reaction in a tunnel lining.
4.3 Water inflow rate of lining
The water flow along a tunnel wall was considered to be a shallow thin film flow (Lebeau
and Konrad, 2010). The water flow rate was described by the conveyance (K) and the
hydraulic gradient in the literature (Lebeau and Konrad, 2010). The conveyance (K) of the
thin film flow along a tunnel wall is affected by the roughness of the tunnel wall and the
geometry of flow. The hydraulic gradient is determined by the elevation of water inflow area
on tunnel lining.
This Section aims to provide a method to estimate the local water inflow rate of a crack or
joint by measuring the water flow rate over a finite section of the flow area. The
measurement of water inflow rate only considers the condition when water flows along the
tunnel wall and forms a fan shape flow area which is the most typical case observed in the
selected tunnel. A detailed description for measuring local water inflow rate on site will be
given in this section. The study in this section will develop a method for water inflow
measurement in a tunnel (research gap 2) as discussed in the conclusion section of Chapter 2.
Water inflow data measured from site as discussed in this section will be used as an input
parameter for further chapters. The general crack width of the lining of the selected tunnel
151
will be estimated by calculations based on the site measured water inflow rate and the theory
presented in Section 3.3.
4.3.1 Development of local water inflow rate calculation
Water flow shape refers to the geometry of the water flow pattern observed on a tunnel wall
as water emanates from a crack or joint. The water flow shape on a tunnel wall can vary or
even form a point flow due to the orientation and interconnection of cracks. The variation of
water flow shape affects the water inflow measurement strategy and choice of segment for
measurement. One specific water flow shape, which is, the fan shaped flow caused by
transverse cracks, will be studied. The detailed illustration of the fan shape of the water flow
will be elaborated in following paragraphs. Nine different locations of water inflow in the
selected tunnel were measured to illustrate the application of this method.
Transverse wall cracks or joints occur across the direction of the tunnel and are vertical on
the tunnel wall. Each point along a transverse wall crack has a different elevation and the
flow pattern was observed to form a triangle or fan shape on the wall (Figure 4.3). Water
flow rate within the flow area depends on both height and location along the wall.
152
Figure 4.3 A transverse wall crack with a fan shaped flow is observed in the centre of
the photograph (as arrow). A complex crack comprising diagonal and horizontal
segments is observed from top centre to the right hand side of the photograph. The
latter crack is currently dry but previously flowing evidenced by the white mineral
staining
As discussed above, this section only studies the most typical case in the selected tunnel
when water flow along the tunnel wall and forms a fan shape flow area. The case when water
dripping from crown can be measured by a measuring cylinder directly and is not further
discussed. The schematic illustration of water flow from a transverse crack is shown as in
Figure 4.4. The following figure is comparable to Figure 3.1.
153
Figure 4.4 Illustration of measuring water inflow rate of a fan shaped flow caused by
transverse cracks
The velocity of water flowing down along the tunnel wall varies by location within the flow
area. The velocity profile across the fan shaped flow is modelled to provide a framework for
relating sample measurements to the total water flow. Estimation of the total water flow rate
of a transverse wall crack is based on the location where the measurement is made.
An imaginary sink method is developed to quantify the water flow velocity profile across the
flow area of a transverse wall crack. Water flow along the tunnel wall of a transverse wall
crack is considered as a two dimensional water flow problem. A circle at the maximum y
axial value on the water inflow source, which is the top of the crack, is defined as the
“representative point” with a radius of r (Figure 4.5). The water head of a representative point
(h) is expressed in a Cartesian coordinate as Eq. (4.1) (Polubarinova-Kochina, 1963).
(4.1)
where, is the centre coordinate of the representative point, C is an unknown constant.
154
Figure 4.5 Illustration of the representative point in a Cartesian coordinate
The hydraulic field in this system is expressed using the imaginary method (Goodman et al.,
1965). The hydraulic head of the representative point ( ) and the hydraulic head of the
imaginary representative point ( ) is expressed as Eq. (4.2) and Eq. (4.3) respectively.
(4.2)
(4.3)
where, (0, ) is the centre of the representative point, (0, ) is the centre of the imaginary
representative point, K is the unknown tunnel wall conveyance as discussed above, and
are unknown constants.
According to the superposition principle, the hydraulic head in the system of Figure 4.5
caused by two representative points is further modified, as given in Eq. (4.4).
155
(4.4)
[ ]
where, C3 is an unknown constant.
The water source is a vertical line that has a length . In this case, is referring to the
length of flowing crack. The water head distribution is calculated as follows.
(4.5)
∫ ∫ ( [ ] )
The elevation datum level in this case is the base of the linear source. No pressure head is
considered for the case when water flows along the tunnel wall. The unknown constant is
derived by combining with the boundary condition (Eq. (4.6)). The function of total head ( )
is expressed by x, y, Q and K.
| (4.6)
The measured water flow rate is primarily determined by the flow velocity along y direction
when the collector is placed under the water flow.
The hydraulic gradient along the y direction is calculated to determine the water flow velocity
along the y direction. The hydraulic gradient is determined by the hydraulic head difference
over distance. The water head is caused by the different elevation of the water source for a
transverse crack. The hydraulic gradient along the y direction is calculated by Eq. (4.7). As
can be seen, the hydraulic gradient along the y direction is a function of x and y. This means
the hydraulic gradient along the y direction varies within the Cartesian coordinates as shown
in Figure 4.5.
156
(4.7)
The flow velocity is known to be proportional to the hydraulic gradient . The velocity
along the y direction is expressed in Eq. (4.8). The flow velocity along the y direction at any
x,y point is calculated as follows.
(4.8)
The water flow rate is expressed as follows
(4.9)
where, is the thickness of water layer on the tunnel wall, K is the unknown conveyance of
the tunnel wall surface. In this study, and K are considered to have no variation through the
flow area.
The total flow rate of a fan shaped flow area is expressed as the integration as follows.
(4.10)
∫ | ∫ |
The measured water inflow rate is calculated based on the collector size as follows.
(4.11)
∫
where, is the centre line coordinate of the collector. To simplify the measuring process,
the centre line of the collector needs to be aligned with the crack ( ).
It is not possible to solve Eq. (4.10) and Eq. (4.11) individually since the factor including the
conveyance is unknown. Instead, the equation is expressed by the flow rate ratio ( ) of the
measured water flow rate ( ) and the total water flow rate ( ) as Eq. (4.12). The ratio
157
( ) between Eq. (4.10) and Eq. (4.11) would eliminate the variables which do not join the
integration.
(4.12)
The value is a relation between the measured water flow rate and the total water flow rate.
It can only be obtained once a measurement of is obtained on site.
4.3.2 Ratio between measure flow rate and total flow rate
The fan shaped flow caused by vertical transverse wall cracks is described in a two
dimensional way under the Cartesian coordinates. The water flow velocity of each position of
the flow area is different. For a certain elevation, the water flow relation is described by a
ratio between the flow velocity across the flow area (vp) and the flow velocity at the centre
line of that given elevation (vs).
The flow velocity ratio across the bottom of the crack ( ) and the flow velocity ratio
across one third elevation of crack ( ⁄ ) are shown as Figure 4.6.
1 1
Relative water flow velocity
Relative water flow velocity
0.9 0.9
0.8 0.8
0.7 0.7
0.6 0.6
(vp/vs)
(vp/vs)
0.5 0.5
0.4 0.4
0.3 0.3
0.2 0.2
0.1 0.1
0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Relative distance away from the source Relative distance away from the source
(2x/Lf) (2x/Lf)
158
(a) (b)
Figure 4.6 Flow velocity variation through given elevation within the flow area, when (a)
, (b) ⁄ (b)
It can be seen that the source (when the x axis is 0) tends to have a greater flow velocity than
the other parts of the flow area. The contrast is much more significant at the y=0 position.
The flow velocity on the y axial direction (vs) is a function of y and varies with the y value.
The water flow velocity at the source point when is not equal to the flow velocity
when ⁄ . The value is the length of crack as discussed in Eq. (4.5). The bottom
source point ( ) tends to have a higher water flow velocity than the source point
with higher elevation ( ). The flow velocity comparison for source points along
the elevation direction is shown as Figure 4.7.
1
Relative water flow velocity of
0.9
0.8
0.7
source point
(vs/vsly=0)
0.6
0.5
0.4
0.3
0.2
0.1
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Relative source point position
(y/Lcl)
Figure 4.7 Flow velocity comparison on the water source along elevation direction
It shows that the bottom of a flowing crack (y=0) has the greatest flow velocity. The flow
velocity decreases with increasing of the relative elevation (y increasing).
159
The centre of the collector is aligned with the crack for practical applications. The collector
only has a finite length and collects a finite amount of water ( ). When the collector is
placed at bottom of the source ( ), the flow rate ratio ( ) between measured water flow
rate ( ) and total water flow rate ( ) is shown as follows.
1
0.9
0.8
Flow rate ratio
0.7
0.6
(Cr)
0.5
0.4
0.3
0.2
0.1
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Relative size of collector and flow area
(Lc/Lf)
Figure 4.8 Relation between flow rate ratio and the collector dimension ( )
A flow rate value is measured on site. The total flow rate is equal to ⁄ . For
example, when the flow area length Lf is 1 m and the collector size is 0.2 m, the value is
approximately 0.5 as shown in Figure 4.8. The total flow rate is estimated to be .
It may not be always feasible to place the collector at the position. A chart describes
multiple measuring elevations is plotted as shown in Figure 4.9.
160
1.00E+00
Flow rate ratio

(Cr) 1.00E-01
y=0
1.00E-02
y=0.3
y=0.6
1.00E-03
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Relative size of collector and flow area
(Lc/Lf)
Figure 4.9 Water flow rate ratio chart when and

This figure can be used as an index chart to identify appropriate value. The total water
flow rate can be estimated by using Eq. (4.12).
A summary for measuring water flow rate and estimating total water flow rate of different
flow shapes using this method are presented as Table 4.4. The total water flow rate on a
tunnel wall can be estimated as long as all the listed parameters are available.
Table 4.4 Summary of using the finite length water collection method for different flow
shapes
Collector Flow Water
Collector Crack Measured Total water
width area flow rate
Flow shape Elevation length rate flow rate
width ratio
y (m) (m3/s) (m3/s)
(m) (m) (m3/s)
Rectangular N/A Measured Measured Required N/A Calculated

Required
Index
Fan Required Measured Measured Required Calculated
Figure
161
4.3.3 Water inflow rate measured on site
Water inflow rate has been measured using the aforementioned method during tunnel site
inspections from 2015 to 2016. The water flow rate in CLP and BLP has been measured at
nine locations as shown in Table 4.5. The column of measured water flow rate records the
water flow rate of the measured finite segments in one flow area. The ratio of is the
relative size of the collector and the maximum width of the flow area. Total water flow rate is
the calculated using the measured water flow rate based on Eq. (4.12).
Table 4.5 Water flow measurement and the correspondent crack length
Crack
Time Volume Measured water flow Total water flow
Date Tunnel length
(s) (ml) rate (L/min) rate (L/min)
(cm)
07/11/15 CLP 60 31.5 3.15×10-2 0.44 75 4.38×10-2
07/11/15 CLP 60 35.5 3.55×10-2 0.67 145 4.08×10-2
07/11/15 CLP 600 8.5 8.50×10-4 0.91 70 9.34×10-4
07/11/15 CLP 30 85 1.70×10-1 0.40 150 2.43×10-1
16/11/15 BLP 60 58 5.80×10-2 0.21 140 1.12×10-1
16/11/15 BLP 60 99 9.90×10-2 0.17 200 2.20×10-1
16/11/15 BLP 30 74.5 1.49×10-1 0.19 160 3.04×10-1
13/10/16 BLP 330 2 3.64×10-4 0.16 50 8.26×10-4
13/10/15 BLP 5 79.5 9.54×10-1 0.90 90 1.00×100
The average wet crack length based on the data measurement is approximately 1.2 m. It was
found that the maximum measured water flow rate during the inspection is 1 L/min (1.59×10 -
5
m3/s) per location. The minimum water inflow rate could be determined using this method
was found to be 8.26×10-4 L/min (1.38×10-8 m3/s) based on limitations of the time taken to
collect a measurable amount of water.
There are 9 locations where water inflow rate was measured as in Table 4.5. The crack widths
are calculated based on Eq. (3.27) using the measured water inflow rate. It is notable that
Figure 4.10 is not presented for the validation purpose of Eq. (3.27). Instead, Figure 4.10 is
162
presented to show the general cracking condition in this selected tunnel. The water inflow
rate (Q) is then used to calculate the effective crack width ( ) as proposed in equation
(3.27) in section 3.3.
3.00E-04
Effective crack width (m)
2.50E-04
2.00E-04
1.50E-04
1.00E-04 Calculated effective

crack width
5.00E-05
0.00E+00
Figure 4.10 Calculated effective crack width under a given water inflow rate when the
other parameters are given
The effective crack width of the selected tunnel is calculated based on the collected water
inflow rate (Table 4.5) and Eq. (3.27) as shown above. The minimum calculated effective
crack width is calculated to be 1.8×10-5 m (0.018 mm). The maximum calculated effective
crack width is 2.9×10-4 m (0.29 mm).
4.4 Compressive strength of lining concrete
This section provides the compressive strength of concrete in the tunnel which can be used to
interpret the hydraulic conductivity (K) of the sound concrete as shown in Eq. (3.2) and Eq.
(3.3). The tested compressive strength refers the compressive strength of the sound concrete.
163
4.4.1 Investigation in CCL
The first investigation of compressive strength was conducted in CCL in 19/10/2014. Four
different chainage locations were tested in an approximately 15 m interval from chainage
5550 to 5590. The Plinth, walk way and tunnel lining is defined as in Figure 4.2.
Table 4.6 Concrete compressive strength in CCL
Schmidt CCL 19-10-2014

Hammer
Position 5590
Operator JS
Inner Outer
Data Strength(MPa) Data Strength(MPa)
Plinth 30 28.5 Plinth 32 32
38 37 29 22
Walkway Walkway 24 20
49 57
Lining 45 51 Lining 51 60.5
48 56 49 58
Position 5575
Operator JS
Inner Outer
Plinth 30 28.5 Plinth 30 28.5
41 42 20 10
32 27
Lining 52 63 Lining 43 47
50 59 38 38
51 61
Position 5565
Operator YT
Inner Outer
Plinth 35 37 Plinth 45 54
32 27 25 16
30 24
164
45 50 48 57
Position 5550
Operator XZ
Inner Outer
38 37 29 22
38 37
55 69 50 59
The summary for the data collection in CCL is as follows.
80
70
Concrete Strength of CCL
60
50
(MPa)
40
30
20
10
0
Plinth (N=16) Walk way (N=8) Wall (N=17)
Figure 4.11 Statistical comparison of concrete compressive strength of plinth, walkway

and tunnel lining in CCL
The median hydraulic conductivity of sound concrete is calculated based on the compressive
strength of tunnel lining as discussed in section 3.2.1. The calculated mean value of hydraulic
conductivity of concrete is approximately 2.57×10-12 m/s which is the same as the value used
in the worked example in section 3.2.4.
165
4.4.2 Investigation in BLP
The second investigation of compressive test was conducted in BLP in 26/10/2014. The
sampling strategy is the same as in CCL. Four different chainage locations were tested in an
approximately 15 m interval from chainage 5550 to 5594. The testing method are the same as
the first test as mentioned above.
Table 4.7 Concrete compressive strength in BLP
Schmidt BLP 26-10-2014

Hammer
Position 5594
Operator YT
Inner Outer
25 16 30 24
Walkway 18 10 Walkway
10 10
49 59 53 66
Position 5580
Operator YT
Inner Outer
38 37 44 47
31 25
50 59 48 57
48 55
Position 5565
Operator YT
Inner Outer
22 12 46 51
166
45 50
52 64 49 58
Position 5550
Operator YT
Inner Outer
Plinth 37 40 Plinth 30 28.5
33 28 50 59
37 35
48 56 32 27
54 67
The summary for the data collection in BLP is as follows.
80
Concrete Strength of BLP
70
60
50
(MPa)
40
30
20
10
0
Plinth (N=16) Walk way Wall (N=18)
(N=8)
Figure 4.12 Statistical comparison of concrete compressive strength of plinth, walkway

and tunnel lining in BLP
4.4.3 Summary of compressive strength of lining concrete
The overall condition of the tunnel concrete is summarised based on the two site test as
follows.
167
80
Concrete Strength (MPa)

70
60
50
40
30
20
10
0
Plinth (N=32) Walk way (N=16) Lining (N=35)
Figure 4.13 Statistical comparison of concrete strength of plinth walk way and tunnel
lining (overall condition)
The concrete of plinth and walkway generally have a low compressive strength. The median
compressive strength of concrete of plinth and walkway is approximately 25 to 30 MPa. The
concrete of lining is found to have a much greater compressive strength compared with plinth
and walkway. The median compressive strength of the tunnel lining is approximately 58 MPa.
The median hydraulic conductivity of the sound concrete lining in the tunnel is calculated to
be 2.57×10-12 m/s based on Eq. (3.2) and Eq. (3.3). Only the compressive strength of the
lining was used in the hydraulic conductivity calculation.
4.5 Crack features on the lining
This section provides the general cracking information of the lining of the selected tunnel.
Five site inspections were conducted in all four tunnels in 2014. The cracking pattern of the
tunnel lining will be presented in Section 4.5.1. The information in Section 4.5.1 can be used
to determine the crack connectivity coefficient which has been discussed in Section 3.2 and
to evaluate the crack density which has been discussed in Section 3.3. The monitoring of
crack changes over time will be presented in Section 4.5.2. The information in Section 4.5.2
168
will be used to calculate the water inflow rate changes over time and to calculate the
reliability of the tunnel which will be further used in Chapter 5.
4.5.1 Crack mapping
The pattern of cracks is a factor that affects the bulk hydraulic conductivity of the lining. The
crack pattern observed on the tunnel lining inner surface was mapped in a section between
MCS to PS from chainage 5550 to 5595 in all the four tunnels as in Figure 4.14.
Figure 4.14 Crack mapping on site

The length of each mapping zone is approximately 45 m. The width between dash lines in the
mapping is 3 m. The surface crack mapping shows the crack feature on the x-z plane as
shown in Figure 3.1.
169
Figure 4.15 Crack mapping from chainage 5550 to 5595 in BLP
Figure 4.16 Crack mapping from chainage 5550 to 5595 in CLP
170
Figure 4.17 Crack mapping from chainage 5550 to 5595 in NLP
Figure 4.18 Crack mapping from chainage 5550 to 5595 in CCL

Crack density is the accumulated crack length per area of concrete. The crack length of each
lining segment was estimated based on the crack maps provided above. The crack density of
each lining 3 m lining segment is calculated in the table below. It is found that the crack
density for BLP, CLP, NLP and CCL is 0.22, 0.18, 0.21 and 0.17 respectively. The overall
crack density of the tunnel system is 0.21 m/m2.
171
Table 4.8 Crack length and crack density of each tunnel segment in the mapped sections
Crack Segment Crack Crack Segment Crack

Tunnel Tunnel
Length(m) Area(m²) density(m/m²) Length(m) Area(m²) density(m/m²)
BLP 8.41 37.7 0.22 CLP 10.77 37.7 0.29

5.64 37.7 0.15 6.79 37.7 0.18
8.9 37.7 0.24 2.83 37.7 0.08
3.41 37.7 0.09 Average 15.3 37.7 0.41
Average crack
crack length 5.8 37.7 0.15 length and 8.92 37.7 0.24
and density density for
for BLP is 17.24 37.7 0.46 CLP is 4.3 37.7 0.11
8.31m and 6.86m and
0.22 m/m2 12 37.7 0.32 7.73 37.7 0.21
0.18 m/m2
respectively 5.7 37.7 0.15 respectively 12.64 37.7 0.34
7.34 37.7 0.19 5.91 37.7 0.16
5.57 37.7 0.15 8.18 37.7 0.22
3.1 37.7 0.08 9.41 37.7 0.25
5.22 37.7 0.14 5.89 37.7 0.16
7.9 37.7 0.21 14.96 37.7 0.4
19.39 37.7 0.51 7.04 37.7 0.19
9 37.7 0.24 4.26 37.7 0.11
NLP 9.81 37.7 0.26 CCL 10.8 37.7 0.29
12.09 37.7 0.32 4.8 37.7 0.13
6.75 37.7 0.18 1.7 37.7 0.05
Average
crack
Average 4.62 37.7 0.12 15.6 37.7 0.41
length and
crack length
13.9 37.7 0.37 density for 6.3 37.7 0.17
and density
BLP is
for NLP is 12.57 37.7 0.33 2 37.7 0.05
6.32m and
8.03m and
4.5 37.7 0.12 0.17 m/m2 1.8 37.7 0.05
0.21 m/m2
respectively
respectively 3.1 37.7 0.08 0 37.7 0
5 37.7 0.13 3.7 37.7 0.1
9.26 37.7 0.25 5.8 37.7 0.15
6.87 37.7 0.18 7.8 37.7 0.21
8.82 37.7 0.23 3.9 37.7 0.1
10.85 37.7 0.29 7.63 37.7 0.2
3.64 37.7 0.1 10.7 37.7 0.28
8.66 37.7 0.23 10.9 37.7 0.29
Overall crack density = 0.21
172
The depth of cracks is another important factor controlling the connectivity coefficient of the
seepage path as discussed in Section 3.2. A three-dimensional ground penetrating radar (GPR)
test was conducted by the tunnel operator to assess the concrete lining. A representative
section of the ground penetration radar result is shown as Figure 4.19.
Figure 4.19 Representative section of GPR data

This ground penetration radar inspection was conducted by the tunnel operator and therefore
not further elaborated. This information can be used to interpret the depth of crack, which
cannot be observed directly on the lining surface. The steel reinforcement elements are
identified in the ground penetration radar scanning. The formation of a crack, which can be
seen at the lining surface, is illustrated through the lining. The crack depth as shown in Figure
3.1 is interpreted by the scale accordingly.
4.5.2 Changing of number of wet crack
The total number of wet cracks per 100 m tunnel length (n) recorded in the site inspection
data is shown with a linear regression fit in Figure 4.20. Although the data points show
irregular variations, the linear fit will be used in the analysis. Other data fitting models were
trialled, but none showed better regression coefficients than the linear model. It is anticipated
that with further accumulation of site inspection data the trend and choice of fitting method
may be revised in the future.
173
The number of wet cracks in each of the CLP, CCL and NLP tunnels shows an increase over
time with an acceptable linear fit (R2 value). The number of wet cracks in the BLP appears to
be changing little compared to the other three tunnels.
35 35
30 CLP 30 CCL
yy==3.5084x+5.3092
3.5084x - 7057.1
R² = 0.889
R2 = 0.889 y =y 1.929x+5.277
= 1.929x - 3877.8
Wet crack number

Wet crack number
25 25
R² = 0.822
R2 = 0.822
(100*m-1)
(100*m-1)
20 20
15 15
10 10
5 5
0 0
2013
0 2014
1 2015
2 2016
3 2017
4 2013
0 2014
1 2015
2 2016
3 2017
4
Time (year) Time (year)
(a) (b)
35 35
30 BLP 30 NLP
yy==0.5082x+9.6066
0.5082x - 1013.4
Wet crack number
Wet crack number
25 R2 = R² = 0.4506
0.4506 25
(100*m-1)
(100*m-1)
20 20 y = 2.2955x+4.9415
y =22.2955x - 4615.9
R = 0.9198
15 15 R² = 0.9198
10 10
5 5
0 0
2013
0 2014
1 2015
2 2016
3 2017
4 2013
0 2014
1 2015
2 2016
3 2017
4
(c) (d)
Figure 4.20 Average number of wet cracks per 100 m length of tunnel, (a) CLP, (b)
CCL, (c) BLP and (d) NLP.
To investigate the crack register data further, each tunnel has been separated into four
sections based on the location of the portals and stations. Each of the four tunnels passes
174
through three underground stations (FGS, MCS and PS). The section between MCS to PS has
the longest distance which is approximately double the other three sections in all the four
tunnels. It was found that the rate of change in the number of wet cracks in each section of
tunnel differs from the overall recorded data for each tunnel. The number of increase of wet
cracks in the section from portal to FGS is shown as in Figure 4.21.
35 35
30 CLP 30 CCL
yy==3.6451x+3.7863
3.6451x - 7333.8
R2 = R²
0.8895
= 0.8895
Wet crack number

Wet crack number
25 25
y =y 2.7585x-0.3395
= 2.7585x - 5553.2
(100*m-1)
(100*m-1)
20 20 R2 = 0.8716
R² = 0.8716
15 15
10 10
5 5
0 0
2013 2014 2015 2016 2017 2013
0 2014
1 2015
2 2016
3 2017
4
0 1 2 3 4
(a) (b)
35 35
30 BLP 30 NLP
Wet crack number
Wet crack number
25 25
yy ==1.1305x
1.1305x+3.0965
- 2272.6
(100*m-1)
R2 =R²0.7873
(100*m-1)
20 = 0.7873 20
15 15
10 10 yy==5.4528x+5.4864
5.4528x - 10971
R2 = 0.84
R² = 0.84
5 5
0 0
2013
0 2014
1 2015
2 2016
3 2017
4 2013 2014 2015 2016 2017
0 1 2 3 4
Time (year)
Time (year)
175
(c) (d)
Figure 4.21 Number of wet cracks per 100 m length of tunnels from portal to FGS, (a)
CLP, (b) CCL, (c) BLP and (d) NLP
The increasing rate of the number of wet cracks in the NLP in this section (portal to FGS) is
more than double compared to the overall condition of the NLP. The number of wet cracks in
BLP in this section increases relatively slowly compared with other tunnels in the same
section. But it greater compared with the overall condition of the BLP and has a better linear
fit coefficient.
The number of increase of wet cracks in the section from FGS to MCS is shown as in Figure
4.22.
35 35
30 CLP 30 CCL
3.7036x - 7453.4
y = 3.7036x+1.9468
Wet crack number
25
Wet crack number
25 R2 =R² = 0.8758
0.8758 y == 2.2637x
2.2637x+1.6281
- 4555.2
R2 = R²0.766
= 0.766
(100*m-1)
(100*m-1)
20 20
15 15
10
10
5
5
0
0 2013 2014 2015 2016 2017
0 1 2 3 4
2013
0 2014
1 2015
2 2016
3 2017
4
Time (year)
Time (year)
(a) (b)
176
35 35
30 BLP 30 NLP
yy==-1.1899x + 2409.6
Wet crack number

Wet crack number
25 -1.1899x+14.3313 25
R2 = R² = 0.3553
0.3553
y = y0.7999x+3.9987
(100*m-1)
= 0.7999x - 1606.2
(100*m-1)
20 20 R2 = 0.8849
R² = 0.8849
15 15
10 10
5 5
0 0
2013
0 2014
1 2015
2 2016
3 2017
4 2013
0 2014
1 2015
2 2016
3 2017
4
(c) (d)
Figure 4.22 Number of wet cracks per 100 m length of tunnels from FGS to MCS, (a)
The increasing rate of the number of wet cracks in NLP in this section (FGS to MCS) is
relatively slow. It can be seen that, the number of wet cracks appears to decrease over the
three years inspection for BLP in this section.
The number of wet cracks over time in the section from MCS to PS is shown as in Figure
4.23.
35 35
30 CLP 30 CCL
Wet crack number
Wet crack number
25 25
(100*m-1)
(100*m-1)
20 20
15 15
y =y0.0854x+25.0702
= 0.0854x - 146.84
y =y 5.6712x+14.1256
= 5.6712x - 11402
R2 = 0.0009
R² = 0.0009
10 R² = 0.8223
R2 = 0.8223 10
5 5
0 0
2013
0 2014
1 2015
2 2016
3 2017
4 2013
0 2014
1 2015
2 2016
3 2017
4
(a) (b)
177
35 35
30 BLP 30 NLP y y= =4.7628x+5.7164

4.7628x - 9581.8
R² = 0.8335
R2 = 0.8335
Wet crack number

Wet crack number
25 25
(100*m-1)
(100*m-1)
20 20
15 15
yy == 2.389x+18.757
2.389x - 4790.3
R2 =R² = 0.7184
0.7184 10
10
5 5
0 0
2013 2014 2015 2016 2017 2013
0 2014
1 2015
2 2016
3 2017
4
0 1 2 3 4
Time (year)
Time (year)
(c) (d)
Figure 4.23 Number of wet cracks per 100 m length of tunnels from MCS to PS, (a)
The CCL in this section (MCS to PS) does not increase in the number of wet cracks. A rapid
increase in wet cracks is observed in CLP, BLP and NLP in this section.
The number of wet crack over time in the section from PS to portal is shown as in Figure 4.24
35 35
30 CLP 30 CCL
Wet crack number
Wet crack number
25 25
y = 2.389x+3.7066 y y= =2.1521x+0.8773
2.1521x - 4331.3
Ry2 =
= 0.8837
2.3082x - 4642.7
(100*m-1)
(100*m-1)
20 20 R2 = 0.9029
R² = 0.9029
R² = 0.8837
15 15
10 10
5 5
0
0
2013
0 2014
1 2015
2 2016
3 2017
4
2013
0 2014
1 2015
2 2016
3 2017
4
Time (year)
Time (year)
(a) (b)
178
35 35
30 BLP 30 NLP
Wet crack number
Wet crack number

25 25
y = 0.112x+6.776
(100*m-1)
(100*m-1)
20 y = 0.112x - 218.68 20
R2 = 0.0104
R² = 0.0104 y = 0.624x+5.112
15 15 R2 =y 0.8412
= 0.624x - 1251
R² = 0.8412
10 10
5 5
0 0
2013
0 2014
1 2015
2 2016
3 2017
4 2013
0 2014
1 2015
2 2016
3 2017
4
(c) (d)
Figure 4.24 Number of wet cracks per 100 m length of tunnels from PS to portal, (a)
The number of wet cracks of BLP and NLP shows a very low rate of change in the section
between PSs to portal.
The summary of the increasing rate of number of wet cracks and the wet crack number is
shown in Table 4.9.
179
Table 4.9 Number of wet cracks from year 0 to year 3 and rate of linear interpolation of
the increase in number of wet cracks. Total numbers for each tunnel are shown in bold.
Increasing rate
of number of
Wet crack Number
wet cracks
(100* m-1/year)
Sep Dec
May-1 Jul-1 Dec-1 Aug-2 Dec-2 Aug-3
-0 -3
BLP 24
256 245 237 238 260 259 302 0.51
3
Portal to FGS 20 19 16 31 31 31 34 38 1.13
FGS to MCS 70 71 70 51 48 52 46 64 -1.19
MCS to PS
88 88 81 111 104 109 109 123 2.39
PS to Portal 65 78 78 44 55 68 70 77 0.11
CCL 13
159 165 137 255 248 251 277 1.93
8
Portal to FGS 1 4 4 12 39 37 38 42 2.76
FGS to MCS 11 15 15 9 39 38 37 40 2.26
MCS to PS 11
124 124 78 113 115 117 118 0.09
6
PS to Portal 10 16 22 38 64 58 59 77 2.15
CLP 12
167 170 191 337 337 336 360 3.51
9
Portal to FGS 17 26 26 28 58 58 60 62 3.65
FGS to MCS
6 17 17 40 54 54 56 58 3.7
MCS to PS 66 78 78 73 133 133 128 138 5.67
PS to Portal 40 46 49 50 92 92 92 102 2.31
NLP 14
158 158 237 287 287 303 340 2.3
8
Portal to FGS 35 36 36 73 110 110 109 113 5.45
FGS to MCS 38 39 39 38 48 48 53 60 0.8
MCS to PS 28 30 30 73 73 73 79 96 4.76
PS to Portal 47 53 53 53 56 56 62 71 0.62
180
4.5.3 Cracking data ranking
The increasing rate of wet cracks has been categorised into “high rate of increase”, “moderate
rate of increase”, “low rate of increase” and “no sign of increase” based on the slope of the
linear fit to the data. The increasing rate of number of wet cracks is categorised as high rate
of wet crack increase when the value is greater than 3 per 100 m tunnel. It is categorised as
moderate rate of wet crack increase when the value ranges from 2 to 3 per 100 m tunnel. It is
categorised as low rate of wet crack increase when the value ranges from 0.5 to 2 per 100 m
tunnel. It is categorised as no sign of wet crack increase when the value less than 0.5 per 100
m tunnel. There are 16 tunnel sections for the whole tunnel system as discussed above. The
increasing rate of number of wet cracks of all the tunnel sections is compared as in Figure
4.25.
6 Portal to FGS
increasing rate of number of wet crack
FGS to MCS
5 High rate of wet
MCS to PS crack increase
4
PS to Portal
(100*m-1/year)
3
Moderate rate of
wet crack increase
2
Low rate of wet
1 crack increase
0
No sign of wet
crack increase
-1
-2
0 1
BLP 2
CCL 3
CLP 4
NLP 5 6
Figure 4.25 Increasing rate of number of wet cracks by tunnels and sections
It is found that the high rate of wet crack increase is concentrated in CLP and NLP. The MCS
to PS part of CLP has the most rapid increase rate of number of wet cracks at 5.67 cracks per
181
0%
100%
0%
20%
60%
80%
40%
20%
40%
80%
100%
60%
0%
20%
40%
60%
80%
100%
3000 3000 3000
3070 3070 3070
3140 3140 3140
3210 3210 3210
3280 3280 3280
3350 3350 3350
3420 3420 3420
3490 3490 3490
3560 3560 3560
3630 3630 3630
3700 3700 3700
3770 3770 3770
3840 3840 3840
3910 3910 3910
3980 3980 3980
4050 4050 4050
4120
FGS FGS
4120 4120
FGS
4190 4190 4190
4260 4260 4260
4330 4330 4330
4400 4400 4400
4470 4470 4470
4540 4540 4540
4610 4610 4610
4680 MCS 4680
MCS 4680 MCS
4750 4750 4750
4820 4820 4820
4890 4890 4890
4960 4960 4960
5030 5030 5030
5100 5100 5100
5170 5170 5170
5240 5240 5240
5310 5310
5310
5380
(c)
(a)
5380
182
5380
(b)
5450 5450
5450
5520
5520 5520
5590
5590 5590
5660
5660 5660
5730
5730 5730
5800
5800 5800
5870
PS PS
5870 5870
PS
5940
5940 5940
6010
6010 6010
6080
6080 6080
6150
6150 6150
6220
6220 6220
6290
6290 6290
6360
6360 6360
6430
6430 6430
Number of wet crack increasing ranking (BLP)

6500
Number of wet crack increasing ranking (CCL)

6500 6500
6570
Number of wet crack increasing ranking (CLP)
6570 6570
6640
6640 6640
6710
6710 6710
6780
6780 6780
6850
6850 6850 6920
6920 6920 6990
6990 6990
Station
5.45 cracks per 100 m tunnel length per year of the section from portal to FGS of NLP.
The tunnel sections are ranked based on the wet crack increasing rate as in Figure 4.26.
Station
Station
High rate
Low rate
Moderate rate
Moderate rate
Moderate rate
100 m tunnel length per year. It is followed by the increase rate of number of wet cracks of
No sign of increasing
No sign of increasing
100%
80%
Station
MCS
FGS
60%
PS
40% High rate
20% Low rate
0%
3000
3070
3140
3210
3280
3350
3420
3490
3560
3630
3700
3770
3840
3910
3980
4050
4120
4190
4260
4330
4400
4470
4540
4610
4680
4750
4820
4890
4960
5030
5100
5170
5240
5310
5380
5450
5520
5590
5660
5730
5800
5870
5940
6010
6080
6150
6220
6290
6360
6430
6500
6570
6640
6710
6780
6850
6920
6990
Number of wet crack increasing ranking (NLP)
(d)
Figure 4.26 Tunnel section ranking based on increasing rate of number of wet cracks, (a)
The ranking of tunnel sections only provides a qualitative illustration of the increasing rate of
number of wet cracks observed in inspections. The total wet crack number of a section
depends on both initial crack number and crack increasing rate. Therefore, a high ranking
level does not necessarily mean a high total wet crack number. It means a high rate of
increase in the number of wet cracks.
4.6 Chemical analysis of samples
4.6.1 Sample list
This section aims to provide a general elaboration of general condition of the selected tunnel
by collecting soil and water samples on site. Sample collection on site was conducted in the
selected tunnels. The site data collection is shown in Figure 4.27.
183
Figure 4.27 Site data collection
The sample type indicates the onsite morphology of the sample. Inner wall and outer wall is
described in Section 4.4. Both solid samples and water samples were summarised in Table
4.10.
184
Table 4.10 List of the collected samples
Chainage Descripti
Sample Code Tunnel Sample type Brief Comments
(m) on
Inner wall, close to plinth, soft red
MU 1 CLP 5610 Sandy Solid red
fraction
Inner wall, close to plinth, soft black
MU 2 CLP 5610 Soft Solid black
fraction
MU 3 CLP 5505 Sandy Solid red Outer wall close to centre line
Outer wall close to centre line, flow
MU 4 CLP 5505 Water Clean
along the wall
MU 5 CLP 5640 Powder Solid White Inner wall, crystal salt
Outer wall, the skin of this Material is

MU 6 CLP 5640 Powder Solid white
very hard, but inside is powder
MU 7 CLP 5550 Hard Solid Black Inner wall
MU 8 CLP 5550 Hard Solid White Inner wall
Stalactite Stalactite of crown around 10cm in
MU 9 CLP 6130 Solid
of crown length
Stalactite Stalactite of crown around 10cm in
MU 10 CLP 6130 Solid
of Crown length, adjacent to MU 9
Clay on At the intersection of rectangular and
MU 11 CLP Portal Solid
the floor circle tunnel
Precipitation around the wet zone or
MU 12 NLP 5580 Sandy Solid red
the outer wall
Precipitation around the wet zone or
MU 13 NLP 5580 Hard Solid black
the outer wall
Yellow- Material Collected from ditch looks
MU 14 NLP 5580 Sandy Solid
Brown like clay
Brown-
MU 15 NLP 5589 Hard Solid Material collected from ditch
Black
Black-
MU 16 NLP 5323 Hard Solid From outer wall
white
MU 17 NLP 5323 Hard Solid Brown From inner wall
MU 18 NLP 5143 Powder Solid White At the plinth of inner wall
MU 19 NLP 5143 Soft Solid Black Very wet material
Rectangular salt slab, the skin is
MU 20 NLP 5520 Hard Solid White
Black, and inside of the slab is white
MU 21 NLP 6129 Water Clean Flowing, Outer wall
Concrete
MU 22 NLP 5368 Concrete Spalled concrete from plinth
of Plinth
MU 23 CCL 5652 Water Water Outer wall, flowing
MU 24 CCL 5653 Soft Solid Black Outer wall, flowing
MU 25 CCL 5685 Soft Solid Black Inner wall
MU 26 BLP 6135 Soft Solid Black Inner wall, 26-Oct-14
MU 27 BLP 6135 Water clean Dripping from crown, 26-Oct-14
From accumulated water of ditch, 16-
MU 28 BLP 5550 Water Clean
Aug-14
MU 29 BLP 5630 Water Clean Dripping from crown, 16-Aug-14
185
4.6.2 Properties of solid samples
It was found that dry cracks tend to be clean on the crack surface while wet cracks usually
have deposits forming on the crack surface. The samples taken from the tunnel include the
deposit around wet cracks as summarised above. The composition of the white deposit
around a wet crack was found mainly calcium carbonate. The properties of the collected solid
samples based on the EDS and XRD test is as follows.
186
Table 4.11 Sample properties based on EDS and XRD tests
EDS (wt%)
Sample
Tunnel XRD Na Mg Al Si P S Cl K Ca Mn Fe Zn Zr Ba
Code
MU 1 CLP Non-crystalized 2 2.2 4.2 1 1 1.4 1.3 32 26
SiO2 CaCO3, Nacl
MU 2 CLP and Non- 1.3 25 7.3 2.9 3.1 6.7
Crystallization
MU 3 CLP Non-crystalized 4.1 1.7 15 39
MU 5 CLP NaCl 6 5.4 16 5.9
MU 6 CLP CaCO3 1.4 43
NaCl, Na2SO4,
MU 7 CLP northupite and Non- 8.7 24 1 7 0.7 4.7
Crystallization
MU 8 CLP NaCl 22 8.1 21 6.9
MU 9 CLP CaCO3, NaCl 1.4 1 43 2.3
MU 10 CLP CaCO3, NaCl 0.9 43 1.4
SiO2, Labradorite,
MU 11 CLP Muscovite and Non- 1.3 2.6 6 22 1 1.1 1 2.5 18 1
crystallization
MU 12 NLP Non-crystalized 1.6 3 2.2 1 1.1 1.4 45 17
Mg5(CO3)4(OH)2
MU 13 NLP 5H2O and non- 36 4.5 4.5 1.7 3.7
crystallization
CaCO3 and non-
MU 14 NLP 1.9 2.1 3.3 22 18
crystallization
MU NaCl and non-
NLP 2.2 5.6 2.3 1 1 1.9 0.4 0.6 53
15(1) crystallization
MU
NLP 1.1 2.6 31.1 1.5 2.4 1.6 4.9 1 12.3 2
15(2)
Na3H(CO3)2 2H2O,
MU 16 NLP NaCl and non- 14 1 5.1 1 2.7 1 3.2 34
crystallization
NaCl, CaCO3 and
MU 17 NLP 0.5 0.2 0 0.5 0 0.1 46 0.3 3.1
non-crystallization
MU 18 NLP NaCl 3.1 16 6.6
MgCO3
2H2O,SiO2,CaCO3,
MU 19 NLP 1 5.2 2 27 0.6 12 4.2
NaCl and non-
crystallization
MU 20 NLP CaCO3 1.2 0.9 44
CaCO3, Ca2SiO3Cl2
MU 24 CCL and non-
crystallization
NaCl,
MU 25 CCL Na3Ng(CO3)2Cl and
non-crystallization
NaCl,
MU 26 BLP Na3Ng(CO3)2Cl and
non-crystallization
187
4.6.3 Properties of water samples
Water samples were collected from various locations on site. The pH test of the water
samples show a higher alkaline profile compared to the local ground water. The calcium and
iron content of the water sample is also higher than the groundwater data as shown in Table
4.12.
Table 4.12 Properties of water sample and local groundwater

Sample Conductivity Ca2+ Fe2+
Tunnel Chainage (m) Brief Comments pH
Code (ms/cm) (mg/L) (mg/L)
outer wall close to
MU 4 CLP 5505 centre line, flow 8.32 20.2 62.115 0.501
along the wall
MU 21 NLP 6129 Outer wall, flowing 8.76 11.51 55.964 1.446
MU 23 CCL 5652 Outer wall, flowing 8.59 7.35 137.222 10.768
Dripping from
MU 27 6135 8.37 8.59 177.372 1.295
crown
MU 28 BLP 5550 From ground Floor 9.33 29.6 6.183 1.776
Dripping from
MU 29 5630 8.45 8.69 46.146 0.101
crown
Local ground water (Leonard, 2006) Average 7.45 1.56 40 0.5
Maximum 8.65 4.33 92 2.1
Minimum 4.52 0.15 2 0
The high electrical conductivity of water indicates the high salinity of the water samples. The
calcium content in the water sample is greater than the background value. The increasing of
calcium ions in water sample could imply the loss of calcium content in concrete. The
comparison of water sample content and background value is shown in Figure 4.28.
188
9 14000
8.8
Groundwater maximum 12000
8.6
8.4 10000
TDS (mg/L)
8.2
8000
pH
8
7.8 Groundwater average 6000
7.6 4000
7.4
2000
7.2 Groundwater maximum
7 0
5500 5600 5700 5800 5900 6000 6100 6200 5500 5600 5700 5800 5900 6000 6100 6200
Chainage Chainage
(a) (b)
200 12
180
160 10
140
8
Ca2+ (mg/L)
Fe2+ (mg/L)
120
100 6
Groundwater maximum
80
4
60
40 Groundwater average 2 Groundwater maximum
20 Groundwater average
0 0
5500 5600 5700 5800 5900 6000 6100 6200 5500 5600 5700 5800 5900 6000 6100 6200
Chainage Chainage
(c) (d)
Figure 4.28 Comparison of water samples chemistry from tunnel and local ground
water property (a) pH comparison, (b) Salinity comparison, (c) Calcium ion content
comparison and (d) Iron ion content comparison
The chemical analysis of the measured water showed a different chemical profile compared
to the groundwater. The elevation of water salinity and pH value of water sample indicates
the potential for active chemical reactions occurring in the tunnel lining. The higher value of
pH in water samples relative to average pH value of groundwater could be caused by the
alkaline composition of the concrete.
189
The calcium ion content in the water samples is higher than the average of groundwater. Two
water samples showed a higher calcium ion content above the range of groundwater. The
solid precipitates collected from site were found to be calcium carbonate as shown in
Appendix A. It can be interpreted that the calcium content in concrete is experiencing
chemical reaction based on the information of both water samples and solid samples. The loss
of calcium from the concrete indicates a calcium leaching effect occurring in the lining.
The iron ion content in the water samples was found to be within the range of the
groundwater. One sample showed higher iron ion content compared to that of the
groundwater which could indicate the existence of corrosion product for that location.
This section provides a general elaboration of the conditions of the selected tunnel. The
potential behaviour of the selected tunnel can be interpreted from the mineral samples and the
water samples. The data provided in this section may provide baseline data for future tunnel
maintenance and evaluation.
4.7 Summary
The summary for the site investigation was outlined in this section. A technique for water
flow rate estimation along tunnel walls has been proposed (Research gap 2). The water
inflow rates of various locations were measured in the selected tunnels. The compressive
strength in the selected tunnels was tested at multiple locations using a Schmidt hammer. The
crack pattern and crack density has been mapped on site. The changes in number of wet crack
observed in site inspections has been summarised in this Chapter. The chemical content of
the collected water and the solid precipitates in the tunnel have been analysed to understand
the chemical reaction in the lining.
190
The local water inflow rate of one crack or joint source is estimated by measuring the water
inflow rate of a limited segment within the flow area. The water inflow measurement method
was applied to fan shaped flows caused by transverse wall cracks. The water inflow rate
estimation method is presented here with the following results.
(a) An imaginary sink method is developed to calculate the water flow velocity within the
flow area of the transverse crack. The ratio between the measured water flow rate and
the total water flow rate ( ) was calculated based on the collector size and the
maximum width of the water inflow area. An index chart for water flow rate ratio ( )
was developed to incorporate more site conditions.
(b) A summary table of the required parameters is provided for site application. The
water flow estimation method in this study could potentially serve as an effective tool
for site inspection where water inflow rate data is beneficial.
(c) A table of the on-site measured water inflow data was given. It shows that this
collection and estimation method can be successfully applied to cracks with flow rates
between 8.26×10-4 L/min and 1 L/min.
The compressive strength test in the selected tunnel shows that the plinth and walkway have a
much lower compressive strength compared to the actual tunnel lining. The median
compressive strength of concrete of the tunnel wall is approximately 58 MPa. The hydraulic
conductivity of sound concrete was estimated based on the compressive strength of lining.
The crack patterns of the four selected tunnels were mapped in a 50 m section between MCS
to PS. The length, location and orientation of every crack within the mapping section were
recorded. The number of wet cracks was recorded in all the four tunnels by the tunnel
operator in eight inspections in three years. The increasing of the number of wet cracks was
191
summarised. Each tunnel section has been ranked based on the increasing rate of the number
of wet cracks.
The chemical analysis of the water samples showed a different chemical content compared to
that of the groundwater. The salinity, pH and calcium content of water entering the tunnel
were found higher than the local groundwater. The solid samples were found mainly calcium
carbonate deposits. The calcium leaching in the selected tunnel was interpreted to be a
potential detrimental chemical reaction in the concrete lining.
192
5 Tunnel Service Life Prediction
5.1 Introduction
Water ingress into a tunnel is a widely recognised problem for tunnels under the water table.
The reliability of a tunnel is an important aspect for tunnel assessment and maintenance. The
reliability of a tunnel depends on the defined failure mechanism. Studies of tunnel risk
assessment have been developed based on criteria including, collision, impact with obstacles
and fire (Vanorio and Mera, 2012). Water ingress into a tunnel has been recognised as a
problem based on a review study that was developed by International Tunnelling Association
(ITA, 1991). Water ingress into a tunnel could, for example, drip onto electrical cables in the
tunnel when the water inflow rate is high. The reliability of a tunnel based water ingress has
not been studied as a stochastic process for tunnel reliability calculations according to the
literature review in Chapter 2. The life span of a tunnel with respect to water ingress was
defined by two failure modes which are water inflow through cracks and water seeping
through concrete.
The first failure mode ( ) for tunnel service life calculation is water inflowing through
cracks and joints of the tunnel lining. If a concrete lining cracked very soon after casting due
to concrete shrinkage, for example, water would flow through the lining almost instantly if
the cracks were fully interconnected. In a typical engineering case, only selected priority
cracks can get immediate treatment. Even fully interconnected crack with low flow rates
could be left untreated. The flow rate through untreated crack could increase over time due to
the deterioration of lining. The tunnel lining affected by failure mode ( ) is defined to be
failed once the water inflow rate per metre length of tunnel exceeds a limit. The mechanism
of water inflow through cracks was studied in Chapter 3. Water inflow through fully
193
interconnected cracks and the joints of the lining was determined by the cubic flow law. The
model to quantify the water inflow rate into a lining was discussed as in Eq. (3.22).
The second failure mode ( ) for tunnel service life calculation is water seeping through the
tunnel lining. That is, water seeping into a tunnel through partially interconnected cracks and
sound concrete. The hydraulic conductivity of concrete with partially interconnected cracks is
much lower compared to a fully interconnected crack as recognised in Chapter 3. It would
take a long time for water to seep through concrete with partially interconnected cracks due
to the low hydraulic conductivity. This failure mode would not form visible water on the
lining surface due to the low flow rates. This failure mode ( ) is defined to be failed once
water seeps through the whole thickness of the concrete lining. The mechanism of water
seeping through concrete was studied in Section 3.2. Water seeping through sound concrete
and concrete with partially interconnected cracks was determined based the hydraulic
conductivity of the concrete which was quantified by Darcy’s law under saturated condition.
The model to quantify the hydraulic conductivity of partially interconnected cracks was
developed as in Eq. (3.18).
The limit state function of both failure modes are determined by both environmental effects
and acceptance as follows.
(5.1)
where, is the limit state function; is the acceptance; is the environmental
effect; t is time. The environment effect refers to the water inflow rate and the water seeping
depth for the first and second failure modes respectively.
The probability of failure of these two failure modes can be calculated accordingly. The time
when the failure probability of the tunnel exceeds a limit is defined as the service life of the
194
tunnel. The service life of the tunnel can be determined based on the time when the calculated
failure probability exceeds the maximum acceptable failure probability.
This chapter will provide a reliability prediction of an selected tunnel using both water inflow
through cracks and water seeping through concrete as the failure modes. The failure
probability of the selected tunnel will be calculated based on both these failure modes.
The assessment of water inflow through cracks are discussed in Section 5.2 in this chapter.
The assessment of water seeping through concrete are discussed in Section 5.3 in this chapter.
The system failure probability of the tunnel due to water ingress, which includes both of the
failure modes, are discussed in Section 5.4 in this chapter.
The study in this chapter provides an application of reliability assessment method for water
ingress problems of a tunnel (research gap 5).
5.2 Reliability of tunnel lining subjected to water inflow
As discussed in previous chapters, an excessive amount of water inflow, for example,
dripping on the electrical cables, can cause potential hazard in the selected tunnel. Water
inflow in the selected tunnel is highly localized as discussed in Chapter 3. The selected tunnel
is divided into multiple tunnel components to increase the accuracy of the reliability
prediction. A tunnel component is defined as 1 m tunnel length. The component failure
probability means the failure probability of 1 m tunnel length.
A crack was found to be fully interconnected if it is able to form any visible water inflow
and/or water drips based on the study in Chapter 3. The partially interconnected crack will
only form efflorescence during a seepage process. Under the failure mode of water inflow
( ), a component of concrete lining is defined as failure when the water flow rate in that
195
component exceeds a limit. The component failure probability due to water inflow, , is
determined as follows.
(5.2)
where, is the water inflow rate per component (environmental effect). is the water
inflow capacity of tunnel component (acceptance). The tunnel component is calculated as
failure when the water inflow rate of a component ( ) is greater than the acceptance of the
water inflow rate of a component ( ).
The failure of one tunnel component does not mean the violation of the water inflow failure
system mode for the tunnel. The failure of the tunnel ( ) subject to water inflow is
defined when the expected number of component failure ( ) exceeds the threshold
component failure number ( ) as follows.
(5.3)
The expected number of component failure ( ) is determined by the product of total number
of components ( ) and the failure probability of each component ( ), which is written as
. The threshold component failure number ( ) in this case is defined as a certain
percentage of the failure ( ) of the total component number ( ), which is written as
. The total component number of different tunnel sections is different which will
be further discussed in Section 5.2.3.
The service life of the tunnel subjected to water inflow ( ) is reached when the failure
probability of the tunnel ( ) at time exceed the maximum acceptable risk level ( ) as
follows.
196
(5.4)
The threshold component failure probability ( ) and maximum acceptable risk level ( )
are further discussed in Section 5.2.3.
The upcrossing method which has been discussed in Section 2.5.2 is used for calculating the
stochastic model of water inflow (Li, 2005).
5.2.1 Model of environmental effect
An analytical solution for predicting the water inflow rate into a fully cracked concrete lining
was presented in the previous section (Eq. (4.17)). The phenomenon of water inflow is
considered as a stochastic process. The water inflow rate is the function of basic random
variables and time as follows.
(5.5)
where, is the hydraulic conductivity of the tunnel lining which can be determined using
the cubic flow law based on crack width (w) and crack number, Km is the hydraulic
conductivity of the rock mass, h is the hydraulic head. The probabilistic information of those
basic random variables is available from the study in previous Chapter as summarised in
Table 5.1. The basic random variables will be further used for the one metre component
failure probability calculation for in different tunnel sections.
197
Table 5.1 Means and Coefficient of variation of different parameters used in water
inflow reliability calculation
Parameters Mean COV Source

Calculated based on
Crack width ( ) 0.11×10-3 m 0.16
onsite measurement
Hydraulic head ( ) Different by sections 0.2 Literature
Hydraulic conductivity
1×10-6 m/s 0.2 Literature
of rock mass (Km)
Crack number (n(t)) Different by sections - Onsite measurement
Lining thickness (d) 0.56 (m) - Literature
Crack length ( ) 1.2 m - Onsite measurement
The water inflow process of the tunnel is expressed by a stochastic model. A model for water
inflow increase can be described in the form as follows.
(5.6)
where, is the stochastic increasing function for water inflow rate, is the original water
inflow rate.
The statistics of mean function and covariance function of water inflow process is obtained
using the technique of Monte Carlo simulation. Each tunnel has been divided by four sections
as discussed above. With the value given in Table 5.1, the stochastic process of the water
inflow functions of tunnel component in each section is developed as the mean functions
( ) and the coefficient of variation functions as shown in Table 5.2.
198
sections
Mean functions ( ) Coefficient of variation functions
BLP
Portal to FGS
FGS to MCS * *
MCS to PS
PS to Portal * *
CCL
Portal to FGS
FGS to MCS
MCS to PS * *
PS to Portal
CLP
Portal to FGS
FGS to MCS
MCS to PS
PS to Portal
NLP
Portal to FGS
FGS to MCS
MCS to PS
PS to Portal
*sections not calculated due to not showing wet crack increase from year 0 to year 3
The co-variance function for for all the tunnel sections is shown as follows.
199
( ) ( ) (5.7)
where, is the auto-correlation coefficient for Q(t) between two points in time and , is
the coefficient of variance, is the mean function of Q(t).
The co-variance function will be affected by the value. Three correlation values ,
and will examined for the reliability calculation.
5.2.2 Function of acceptance
This section aims to identify the term in Eq. (5.2). The term of is the acceptance of
component failure of the first failure mode ( ). The component of the first failure mode is
defined to be 1 metre length of tunnel as discussed in previous sections.
For one metre tunnel length, it was defined as “damp” when the water inflow rate is less than
1 litre per minute (Bieniawski, 1990) as has been discussed in Table 2.1. A deterministic
value of the acceptance for this component failure of the failure mode 1 is used. When the
water inflow rate of 1 metre tunnel length more than 1 litre per minute, the tunnel is define as
“wet” (Bieniawski, 1990). A component is deemed to have failed if the total water inflow rate
over a 1 metre length tunnel exceeds a limit. Water inflow rate measurement has been
conducted on site and discussed in Chapter 4. One location where the water inflow rate is
greater than 1 litre per minute has been identified amongst the total nine measurements. An
illustration of a location where the water inflow rate exceeds 1 litre per minute in the tunnel is
shown as Figure 5.1.
200
Figure 5.1 Illustration of a location where the water inflow rate greater than 1 L/min
The limit applied in this study is 1 litre per metre per minute for the component failure
probability calculation ( ).
5.2.3 Reliability of tunnel due to water inflow
Failure probability of tunnel component
The component failure probability ( ) refers to the failure probability of 1 m length of
tunnel. The different crack increasing rate of different section of tunnel will differ according
to the component failure probability. The acceptance of water inflow rate for one component
( ) is 1 litre per minute as discussed in previous section. The stochastic process of water
inflow of a tunnel component of different tunnel sections is given as Table 5.2. The
component failure probability is calculated using the upcrossing method based on the
stochastic process of water inflow. The equation of the upcrossing method has been discussed
in chapter 2 and shown as follows.
∫
̇|
{ (
̇|
)
̇| ̇|
}
(5.8)
̇| ̇| ̇|
201
where, and are standard normal density and cumulative distribution function of the
function in bracket respectively, and were defined as the mean and standard deviation of
(stochastic process of water inflow). The description of the other parameters in the
upcrossing method is shown in Section 2.5.
The component failure probability ( ) of the section from portal to FGS is shown
graphically in Figure 5.2.
1 1
Portal to FGS CCL Portal to FGS
0.9 0.9
Probability of component failure

CLP
0.8 0.8
0.7 = 0.1
Series1 0.7 = 0.1
Series1
= 0.5
Series2 = 0.5
Series2
0.6 = 0.9
Series3 0.6 = 0.9
Series3
0.5 0.5
0.4 0.4
0.3 0.3
0.2 0.2
0.1 0.1
0 0
(a) (b)
1 1
Portal to FGS Portal to FGS
0.9 BLP 0.9 NLP
0.8 0.8
0.7 = 0.1
Series1 0.7 = 0.1
Series1
= 0.5
Series2 = 0.5
Series2
0.6 = 0.9
Series3 0.6 = 0.9
Series3
0.5 0.5
0.4 0.4
0.3 0.3
0.2 0.2
0.1 0.1
0 0
(c) (d)
202
Figure 5.2 Failure probability of one metre tunnel component from portal to FGS
section, (a) CLP, (b) CCL and (c) BLP and (d) NLP
The water table for this section is relatively shallow compared to the other sections since it is
close to the portal. The component failure probability of FGS to Portal section of CLP
increases significantly after 30 years. The CLP is an upper tunnel which is under lower water
pressure in this case. The primary reason of the component failure probability increase in
CLP in this section is caused by the high increasing rate of the number of wet cracks as
shown in Figure 4.26 and the measured high initial number of wet cracks as shown in Table
4.9.
The NLP also shows a high predicted component failure probability after 20 years. The
increasing of component failure probability of NLP in this section is caused by the high wet
crack increasing rate and the high water pressure.
The component failure probability in CCL and BLP in this section do not show a significant
increase in 50 years. The CCL shows a moderate wet crack increasing rate. However, the
initial number of wet cracks in CCL is less than that of the CLP. No significant increasing of
component failure probability can be observed in CCL in this section until 60 service years.
The component failure probability ( ) of the section from FGS to MCS is calculated as in
Figure 5.3.
203
1 1
CLP FGS to MCS FGS to MCS
0.9 0.9

CCL
0.8 0.8
0.7 = 0.1
Series1 = 0.1
Series1
0.7
= 0.5
Series2 = 0.5
Series2
0.6 = 0.9
Series3 0.6 = 0.9
Series3
0.5 0.5
0.4
0.4
0.3
0.3
0.2
0.2
0.1
0.1
0
0
Time (year)
Time (year)
(a) (b)
1 1
BLP FGS to MCS NLP FGS to MCS
0.9 0.9
0.8 0.8
0.7 0.7 = 0.1
Series1
= 0.5
Series2
0.6 0.6 = 0.9
Series3
0.5 No sign of wet crack increasing 0.5
Not calculated
0.4 0.4
0.3 0.3
0.2 0.2
0.1 0.1
0 0
Time (year)
Time (year)
(c) (d)
Figure 5.3 Failure probability of one metre tunnel component from FGS to MCS section,
(a) CLP, (b) CCL and (c) BLP and (d) NLP
The water table starts to increase in the section from FGS to MCS compared to the previous
section. No sign of wet crack number increasing is identified in the section between FGS to
MCS in BLP as shown in Figure 4.22(c). The component failure probability of CLP starts to
rise after 60 years even with a high wet crack increasing rate in this section. The primary
reason for the slow rise of the component failure probability is due to the low initial wet
crack numbers in this section of CLP.
204
The component failure probability of CCL and NLP shows a rising trend after 30 years.
These two sections show a moderate wet crack increasing rate and a low wet crack increasing
rate respectively. The rapid rise is caused by the moderate initial wet cracks numbers in these
two sections and the high water pressure as given in Table 4.2.
The component failure probability ( ) of the section from MCS to PS is calculated as in
Figure 5.4.
1 1
CLP MCS to PS CCL MCS to PS
0.9 0.9

0.8 0.8
0.7 = 0.1
Series1 0.7
= 0.5
Series2
0.6 = 0.9
Series3 0.6
No sign of wet crack increasing
0.5 0.5
Not calculated
0.4 0.4
0.3 0.3
0.2 0.2
0.1 0.1
0 0
(a) (b)
1 1
MCS to PS NLP
0.9 BLP 0.9
0.8 0.8
MCS to PS
0.7 = 0.1
Series1 0.7
= 0.5
Series2
0.6 = 0.9
Series3 0.6
0.5 0.5
0.4 0.4
= 0.1
Series1
0.3 0.3
= 0.5
Series2
0.2 0.2 = 0.9
Series3
0.1 0.1
0 0
(c) (d)
Figure 5.4 Failure probability of one metre tunnel component from MCS to PS section,
205
The section from MCS to PS has the deepest water table as given in Table 4.2. The failure
probability of this section shows a more rapid increasing rate relative to the other sections. It
is caused by a high initial wet crack number, high wet crack increasing rate and the high
hydraulic head. The three aforementioned factors are the worst combination in terms of
service life of a tunnel structure. The two deeper tunnels, BLP and NLP, are shown to have
the most critical conditions in terms of component failure probability. The failure probability
of CCL is not calculated since this tunnel is not showing a sign of increasing of wet crack
number from the three years of wet crack inspection records as presented above.
The component failure probability ( ) of the section from PS to portal is calculated as in
Figure 5.5.
1 1
CLP PS to Portal PS to Portal
0.9 0.9
CCL
0.8 0.8
0.7 = 0.1
Series1 = 0.1
Series1
0.7
= 0.5
Series2 = 0.5
Series2
0.6 = 0.9
Series3 0.6 = 0.9
Series3
0.5
0.5
0.4
0.4
0.3
0.3
0.2
0.2
0.1
0.1
0
0
Time (year)
Time (year)
(a) (b)
206
1 1
BLP PS to Portal NLP
0.9 0.9

0.8 0.8
= 0.1
Series1
PS to Portal
0.7 0.7 = 0.5
Series2
= 0.9
Series3
0.6 0.6
No sign of wet crack increasing
0.5 0.5
Not calculated
0.4 0.4
0.3 0.3
0.2 0.2
0.1 0.1
0
0
Time (year)
Time (year)
(c) (d)
Figure 5.5 Failure probability of one metre tunnel component from PS to portal section,
The water table decreases in the section from PS to portal. The failure probability of BLP is
not calculated since this tunnel does not show an increase of wet crack number in the section
from PS to portal according to the data provided in Figure 4.24. NLP does not show sign of
failure probability increasing in 100 years due to the low increasing rate of number of wet
cracks. The component failure probability of CLP and CCL starts to increase after 50 years.
The increasing rate of the number of wet cracks in this case is only correlated by linear
regression. Long term monitoring is required for wet crack number over tunnel length. The
probability of failure is not only controlled by the increasing rate of wet crack. The tunnel
depth is also a critical factor in the tunnel reliability assessment in this case.
Failure probability of component system of tunnel lining due to water inflow
The component failure in the tunnels means the water inflow rate of 1 metre length tunnel
exceeds a limit. The middle range value of time dependent correlation factor 0.5 is used for
further calculation. The component system is a parallel system which consists of multiple
tunnel components. A certain number of failures of tunnel components due to water inflow
207
are acceptable. The failure probability of a tunnel section due to water inflow depends on the
maximum acceptable component failure number ( ) and the expected number of
failed components ( ) as shown in Eq. (5.3). Sensitivity analysis of the maximum
acceptable number of component failure (when = 30%, 40% and 50%) is selected in all
the tunnel sections. The correspondent expected number of failed components when =
30%, 40% and 50% is as shown in Table 5.3. The number is rounded to integer in this study.
Table 5.3 Acceptable component failure number ( ) for different component system
in all tunnel sections, the total number of each tunnel sections in bracket

( ) ( ) ( ) ( )
30% 141 144 289 165
40% 189 192 386 220
50% 236 240 482 275
The failure probability of the tunnel component ( ) has been calculated above. The
expected number of failed components can be determined accordingly.
The failure probability of the component system ( ) is calculated from the failure
probability of the tunnel component ( ) and the acceptable number of component failures
as follows (Rausand and Arnljot, 2004). This equation is used when the and values are
integers. The value could be a decimal under a different value. The values in this
study are rounded to be integers in this study.
(5.9)
∑ ( )
where, is the acceptable number of component, is the failure probability of tunnel
component as calculated above, is the failure probability of a tunnel section.
208
The failure probability of the component system ( ) between portal to FGS is calculated
based on Eq. (5.9) as in Figure 5.6.
1 1
Probability of failure of water inflow

0.9 CLP 0.9 CCL
0.8 0.8
0.7 0.7
Series1 Series1
0.6 0.6
failure mode
failure mode
Series2 Series2
0.5 Series3 0.5 Series3
0.4 0.4
0.3 0.3
0.2 0.2
0.1 0.1
0 0
(a) (b)
1 1
0.9 BLP 0.9 NLP

0.8 0.8
0.7 0.7
Series1 Series1
0.6
failure mode
0.6
failure mode
Series2 Series2
0.4 0.4
0.3 0.3
0.2 0.2
0.1 0.1
0 0
(c) (d)
Figure 5.6 Failure probability of component system from portal to FGS sections, (a)
The predicted failure probability of the water inflow failure mode ( ) start to rise after
approximately 40 to 50 years of the component system from portal to FGS in CLP and NLP.
The calculation showed that the failure probability of CCL starts to increase after 50 years.
209
BLP does not show a significant increasing rate of failure probability in 100 years based on
the existing data.
The failure probability of the component system ( ) between FGS to MCS is calculated as
in Figure 5.7.
1 1
0.9 CLP FGS to MCS 0.9 CCL FGS to MCS

0.8 0.8
0.7 0.7
Series1 Series1
0.6 0.6
failure mode
failure mode
Series2 Series2
0.4 0.4
0.3 0.3
0.2 0.2
0.1 0.1
0 0
(a) (b)
1 1
BLP FGS to MCS
0.9 0.9 NLP FGS to MCS
0.8 0.8
0.7 0.7
Series1
failuremode
0.6 0.6
failure mode
Series2
0.5 No sign of wet crack increasing
0.5 Series3
Not calculated
0.4 0.4
0.3 0.3
0.2 0.2
0.1 0.1
0
0
Time (year)
Time (year)
(c) (d)
Figure 5.7 Failure probability of component system from FGS to MCS sections, (a) CLP,
(b) CCL and (c) BLP and (d) NLP
210
The predicted failure probability of the water inflow failure mode ( ) does not show signs
of system failure in 100 years in CLP in the section from FGS to MCS. The failure
probability of this component system in CCL and NLP are predicted to show a significant
increase after 40 years. BLP is not calculated in this section due to no sign wet crack
increasing.
The failure probability of the component system ( ) between MCS to PS is calculated as in
Figure 5.8.
1 1
CCL MCS to PS
0.9 CLP MCS to PS Probability of failure of water inflow 0.9
0.8 0.8
0.7 0.7
Series1
failuremode
0.6 0.6
failure mode
Series2 No sign of wet crack increasing

0.5 Series3 0.5 Not calculated
0.4 0.4
0.3 0.3
0.2 0.2
0.1 0.1
0 0
(a) (b)
1 1
0.9 MCS to PS 0.9
BLP NLP
0.8 0.8
0.7 0.7
MCS to PS
0.6 0.6
failure mode
failure mode
0.5 0.5
0.4 0.4 Series1

0.3 0.3 Series2
Series1 Series3
0.2 0.2
Series2
0.1 Series3 0.1
0 0
211
(c) (d)
Figure 5.8 Failure probability of component system from MCS to PS sections, (a) CLP,
The CCL is not calculated based on the low increasing rate of number of wet cracks as
recorded in 3 years data base. The failure probability of CLP and BLP shows a significant
increase after 20 years when the number of wet cracks is extrapolated linearly. The
component system between MCS to PS in NLP is the most critical section in the whole tunnel.
It is predicted that the probability of failure in MCS to PS section in NLP will reach 10% in 9
years.
The failure probability of component system ( ) between PS to portal is calculated as in
Figure 5.9.
1 1
0.9 CLP PS to Portal 0.9 CCL PS to Portal
0.8 0.8
0.7 0.7
Series1 Series1
0.6 0.6
failure mode
failure mode
Series2 Series2
0.4 0.4
0.3 0.3
0.2 0.2
0.1 0.1
0 0
(a) (b)
212
1 1
BLP PS to Portal
0.9 0.9 NLP

0.8 0.8
0.7 0.7
PS to Portal
failuremode
0.6 0.6
failure mode
0.5 No sign of wet crack increasing 0.5
0.4
Not calculated 0.4
0.3 0.3 Series1
0.2 0.2
0.1 0.1
0 0
(c) (d)
Figure 5.9 Failure probability of component system from PS to portal sections, (a) CLP,
The failure probability of BLP is not calculated since there is no sign of increasing of number
of wet cracks in the component system from PS to Portal. The failure probability of NLP
does not have signs of increase in 100 years in the component system between PS to portal.
The failure probability of CLP and CCL shows an increase after 60 and 90 years respectively.
Service life of tunnel sections
The acceptable failure probability ( ) is required in order to determine the service life of
tunnel sections ( ) as discussed in Eq. (5.4). The water inflow failure mode in this case has a
high detectability and relatively low consequences. Therefore, a high acceptable failure
probability ( ) is acceptable according to Australian standard (AS/NZS 4360:2004). A high
maximum acceptable failure probability is determined to be 16% ( = 1) in this study. This
maximum acceptable failure probability is a relative high value due to the high detectability
and low consequence of the failure mode.
213
The service life of tunnel is based on the failure of different sections ( ) as discussed above
and the acceptable failure probability ( ). A summary table of the service life of tunnel
subject to water inflow ( ) is given as follows based on Figure 5.6 to Figure 5.9.
The service life ( ) of each tunnel section is shown as in Table 5.4 in years.
Table 5.4 Service life of each section of different tunnels subject to water inflow
through cracks (years)
Portal to FGS FGSto MCS MCS to PS PS to Portal
CLP 53 99 29 77
CCL 75 54 * >100
BLP >100 * 18 *
NLP 52 51 12 >100
The failure condition selected is when there is 16%

probability of 30% of the lining components
experiencing water inflow exceeding 1 L / minute / m
Note of tunnel
*sections not calculated due to not showing wet crack
increase based on the site data
5.3 Reliability of tunnel lining due to water seeping
The water seeping process is calculated when the lining is under a saturated condition. The
process of water seeping through a tunnel lining is slower relative to water inflow through
fully cracked concrete. The comparison of water inflow and water seeping in the selected
tunnel was shown in Chapter 3.
Water travel through sound concrete is primarily through the capillary pores of the concrete
as discussed in the literature review chapter. The partially connected crack will increase the
hydraulic conductivity of the concrete and accelerate the water seeping process according to
214
Chapter 3. The concrete lining is defined as failed once water seeping through the thickness
of concrete. The failure probability due to water seeping, , is determined as follows
(5.10)
where, is the water seeping depth into concrete lining (environmental effect) and is
the maximum acceptable seeping depth (acceptance). The water seeping process of the lining
is calculated under a saturated condition. The probability of failure of tunnel lining due to
water seeping is calculated when .
The service life of the tunnel subjected to water seeping ( ) is reached when the failure
probability of the tunnel ( ) at time exceed the maximum acceptable risk level ( ) as
follows.
(5.11)
The maximum acceptable risk level ( ) for the failure mode of water seeping through tunnel
lining are further discussed in Section 5.3.3.
The upcrossing method which has been discussed in Section 2.5.3 is used for calculating the
stochastic model of water inflow (Li, 2005).
5.3.1 Model of environmental effect
A solution for predicting the water ingress depth into partially cracked concrete has been
discussed as a combination of Eq. (3.1) and Eq. (3.18) in Chapter 3. The water seeping is the
function of basic random variables and time as follows.
(5.12)
215
where, is the water seeping depth over time (m), t is time (s), Kc is the hydraulic
conductivity of sound concrete, is the crack concrete connectivity, h is the hydraulic head
over concrete.
The average value of hydraulic head across different tunnel sections is used in the reliability
calculation for water seeping failure mode. The reliability of two deep tunnels (BLP and NLP)
are calculated using a same hydraulic head while the two shallow tunnels (CLP and CCL) are
calculated using a same hydraulic head. The probabilistic information of those basic random
variables is available from the study in previous Chapters as summarised in Table 5.5.
Table 5.5 Means and Coefficient of variation of different parameters used in water
seeping reliability calculation
Parameters Mean COV Source

Sound concrete hydraulic Calculated based on onsite
2.57×10-12 m/s 1.1
conductivity measurement
Lining thickness d 0.56 m - Literature
15 m (BLP, NLP) Literature
Hydraulic head 0.2
5 m (CLP, CCL)
Calculated based on onsite
Connectivity coefficient 0.5 -
measurement
The water seeping process of the tunnel is expressed by a stochastic model. A model for
water seeping can be described in the form as follows.
(5.13)
where, is the increasing stochastic function for water seeping depth, is the original
water seeping depth.
The statistics of is obtained using the technique of Monte Carlo simulation. With the
value given in Table 5.5, the stochastic process of the water seeping functions of different
216
tunnel is developed as the mean functions ( ) and the coefficient of variation functions
as shown in Table 5.6.
Coefficient of variation functions

Mean functions ( )
BLP and NLP
CLP and CCL
5.3.2 Function for acceptance
This section aims to identify the term in Eq. (5.10). The term of is the acceptance of the
second failure mode ( ). The tunnel is considered as failed when water migrates through
the concrete lining for this failure mode. The acceptance for water seeping is the maximum
acceptable seeping depth ( ). The acceptance value in this case is a deterministic value
which is the thickness of the concrete lining (0.56 m).
As discussed above, water seeping through tunnel lining would not be able to form any
visible flowing or dripping water on the lining surface due to the low water inflow rate. An
indicator for water seeping through the lining is the white salt deposition on the lining surface.
A typical illustration of such efflorescence on the lining surface in the selected tunnel is
shown as Figure 5.10.
217
Figure 5.10 Illustration of a location where the white deposit was seen on the lining
surface
5.3.3 Reliability of water seeping failure mode
The failure probability due to water seeping is shown as Figure 5.11 based on the mean
function and coefficient of variation functions in Table 5.5. The failure probability difference
of the two deep tunnels (BLP and NLP) and the two shallow tunnels (CLP and CCL) is
differed by the hydraulic head. The failure probability of BLP and NLP tunnel is predicted to
rise slowly after 20 years. The prediction shows that the failure probability of the two deep
tunnels (BLP and NLP) subject to water seeping is less than 20% in 100 years. The failure
probability of the two shallow tunnels (CLP and CCL) subjected to water seeping has no sign
of increasing in 100 years.
218
1 1
Probability of failure of water seeping
Probability of failure of water seeping

failure mode (BLP and NLP) 0.9 0.9
failure mode (CLP and CCL)

0.8 0.8
0.7 0.7
0.6 0.6
0.5 Series2
= 0.5 0.5 Series2
= 0.5
= 0.9
Series3 = 0.9
Series3
0.4 0.4
0.3 0.3
0.2 0.2
0.1 0.1
0 0
0 20 40 60 80 100 0 20 40 60 80 100
(a) (b)
Figure 5.11 Failure probability of (a) BLP and NLP and (b) CLP and CCL due to water
seeping
Service life of tunnels
The acceptable failure probability ( ) is required in order to determine the service life of
tunnel section as discussed in Eq. (5.11). Similar to the water inflow failure mode, the water
seeping failure mode in this case has a high detectability and relatively low consequences.
Therefore, a high acceptable failure probability ( ) is acceptable according to Australian
standard (AS/NZS 4360:2004). In this case, a higher maximum acceptable failure probability
is determined to be 16%. This value is the failure probability when the safety index = 1.
The service life ( ) of different tunnel subject to water seeping failure mode is shown in
Table 5.7 based on the maximum acceptable failure probability .
219
Table 5.7 Service life of each section of different tunnels subject to water seeping
through lining (years)
Service life
(years)
BLP 93
NLP 93
CCL >100
CLP >100
The water seeping failure mode is much slower for all the four tunnels compared to the
failure mode of water inflow through inter-connected cracks and joints. Based on the
calculated failure probability due to water seeping, the service life of all the four tunnels
would exceed 100 years. Therefore, the water seeping failure mode is not critical for the
service life prediction of tunnel in this case.
5.4 System failure probability of tunnel lining
5.4.1 System of tunnel
The assessment of water ingress into tunnel comprises two failure modes, which are water
inflow rate failure mode and water seeping failure mode. Water ingress assessment is
idealized as a series system which comprises the two aforementioned failure modes. The
failure of either failure mode will cause failure of the tunnel based on water ingress. The
mathematical expression of the system failure probability of tunnel as a series system was
described as a weakest link system as follows (Rausand and Arnljot, 2004).
(5.14)
(5.15)
220
where is the failure probability of the tunnel based on water ingress; is the failure
mode of water inflow through the cracks and joints of the lining, is the failure mode of
water seeping through the lining.
The service life of the tunnel system ( ) is determined when the system failure probability
( ) exceed the maximum acceptable failure probability ( ) as follows.
(5.16)
The failure probability of the two failure modes has been discussed as above, which are water
inflow through crack ( ) and water seeping through concrete ( ). The failure probability
and service life of tunnel lining subject to each failure mode has been calculated as in section
5.2.3 and section 5.3.3 respectively.
The probability of system failure of a tunnel lining is determined by these two failure modes.
The failure mode one ( ) divided the tunnel into four sections. The failure mode two ( )
defined the tunnel as shallow tunnels and deep tunnels by the hydraulic head over the tunnel.
The failure mode one, which is water inflow through cracks and joints, is considered to be the
controlling failure mechanism based on the aforementioned calculation. The water seeping
failure mode is only considered for those sections where the water inflow failure mode is not
calculated. The system failure probability of tunnel lining due to water ingress is presented by
different sections of the tunnels.
5.4.2 Reliability of tunnel lining
The mathematical expression of the series system is shown as Eq. (5.14). The violation of any
failure mode will cause the failure of the tunnel. The system failure probability of the tunnel
due to water ingress from Portal to FGS is shown as Figure 5.12.
221
1 1
Probability of systemn failure of tunnel

Probability of system failure of tunnel
0.9 CLP 0.9 CCL Portal to FGS
Portal to FGS
0.8 0.8
0.7 0.7
0.6 0.6
0.5 0.5
0.4 0.4
0.3 0.3
0.2 0.2
0.1 0.1
0 0
(a) (b)
1 1

0.9 BLP 0.9 NLP

0.8 0.8
0.7 0.7
0.6 0.6
0.5 0.5
0.4 0.4
0.3 0.3
0.2 0.2
0.1 0.1
0 0
(c) (d)
Figure 5.12 Probability of system failure of tunnel due to water ingress from Portal to
FGS, (a) CLP, (b) CCL, (c) BLP and (d) NLP
In the section from Portal to FGS, NLP is the most critical tunnel. The failure probability of
NLP in this section starts to rise after 30 years. BLP of this section shows less probability of
failure. The failure probability of BLP of the section from Portal to FGS starts to rise after 70
years.
222
The probability of system failure of tunnels due to water ingress from FGS to MCS is shown
as Figure 5.13.
1 1

0.9 CLP FGS to MCS 0.9 FGS to MCS

CCL
0.8 0.8
0.7 0.7
0.6 0.6
0.5 0.5
0.4 0.4
0.3 0.3
0.2 0.2
0.1
0.1
0
0
Time (year)
Time (year)
(a) (b)
1 1
0.9 BLP FGS to MCS 0.9 NLP FGS to MCS

0.8 0.8
0.7 0.7
0.6 0.6
0.5 0.5
0.4 0.4
0.3 0.3
0.2 0.2
0.1 0.1
0 0
(c) (d)
Figure 5.13 Probability of system failure of tunnel due to water ingress from FGS to
MCS, (a) CLP, (b) CCL, (c) BLP and (d) NLP
In the section from FGS to MCS, NLP is the most critical tunnel. The failure probability of
NLP in this section starts to rise after 40 years. CCL also shows a relative high failure
223
probability rising after 45 service years. CLP in this section shows a less probability of failure.
The failure probability of BLP of the section from FGS to MCS starts to rise after 80 years.
The probability of system failure of tunnels due to water ingress from MCS to PS is shown as
Figure 5.14.
1 1

0.9 CLP MCS to PS 0.9 CCL MCS to PS
0.8 0.8
0.7 0.7
0.6 0.6
0.5 0.5
0.4 0.4
0.3 0.3
0.2 0.2
0.1 0.1
0 0
(a) (b)
1 1
MCS to PS
0.9 BLP 0.9 NLP

0.8 0.8
0.7 0.7
MCS to PS
0.6 0.6
0.5 0.5
0.4 0.4
0.3 0.3
0.2 0.2
0.1 0.1
0 0
(c) (d)
Figure 5.14 Probability of system failure of tunnel due to water ingress from MCS to PS,
(a) CLP, (b) CCL, (c) BLP and (d) NLP
224
The section between MCS to PS is calculated to be the most critical section in the whole
tunnel for CLP, BLP and NLP. NLP is the most critical tunnel of the section from MCS to PS.
The failure probability of NLP of this section is showing a rising trend at the moment. BLP
also shows a failure probability rise after 10 service years. CLP of this section shows a rise of
probability of failure after 20 years. The failure probability of CCL of the section from MCS
to PS does not show sign of rising in 100 years. The low calculated rising trend of failure
probability in CCL in this section is primarily due to no sign of number of wet cracks
increasing be observed in the site inspection records. A continuous inspection of the number
of wet cracks is recommended for better understanding of the deterioration process of the
lining.
The probability of system failure of tunnels due to water ingress from PS to portal is shown
as Figure 5.15.
1 1
0.9 CLP PS to Portal 0.9 CCL

0.8 0.8 PS to Portal
0.7 0.7
0.6 0.6
0.5 0.5
0.4 0.4
0.3 0.3
0.2
0.2
0.1
0.1
0
0
Time (year)
Time (year)
(a) (b)
225
1 1

0.9 BLP 0.9 NLP
0.8 PS to Portal 0.8
0.7 0.7
PS to Portal
0.6 0.6
0.5 0.5
0.4 0.4
0.3 0.3
0.2 0.2
0.1 0.1
0
0
(c) (d)
Figure 5.15 Probability of system failure of tunnel due to water ingress from PS to
portal, (a) CLP, (b) CCL, (c) BLP and (d) NLP
The probability of system failure of tunnels due to water ingress into CLP, CCL and NLP is
identical to the failure probability due to the failure mode 1 ( ). Water inflow through cracks
and joints is the controlling mechanism to serviceability in the four tunnels as a series system.
The critical section for the tunnel system is the section between MCS to PS and shown in
Figure 5.14.
5.4.3 Service life of the tunnel lining
As discussed in both Section 5.3 and Section 5.4, the water inflow ingress into a tunnel has a
high detectability and relatively low consequences. The maximum acceptable system failure
probability is selected to be 10% in this study which is a lower failure probability
compared to that of the failure modes.
The service life ( ) based on the system failure probability of the tunnel (Eq. (5.16) is shown
as in Table 5.8.
226
Table 5.8 Service life of each section of different tunnels due to water ingress (years)
CLP 48 92 25 71
CCL 72 52 >100 95
BLP 88 90 14 90
NLP 41 47 9 90
5.4.4 Risk rating
The risk rating of the tunnel section is determined by the time-dependent reliability prediction.
The acceptance of a risk is considered as an important factor for the risk management
(AS/NZS, 2004). The rating of risk has been categorised by the reliability index (β) from β=1
to β=4. The correspondent probability from β=1 to β=4 is 1.59×10-1, 2.28×10-2, 1.35×10-3 and
3.16×10-5 respectively. Five risk categories from risk 5 to risk 1 were defined using the four
boundaries created by the four reliability indexes. The description and the correspondent
actions of the five risk categories for water ingress are shown as in Table 5.9.
227
Table 5.9 Description of five risk categories for water ingress
Risk Probability
Description Action
categories of failure
The tunnel section is relatively dry.
Occasional wet area caused by Monitoring the wet and dry
1 <3.16×10-5 cracks can be seen on the tunnel condition of the identified
lining. Most of the water inflow rate cracks
is low and not measurable.
The tunnel section is damp.
Occasional flowing area caused by Monitoring the wet and dry
3.16×10-5
2 cracks can be seen on the tunnel condition of the identified
~1.35×10-3
lining. The water inflow rate along cracks
the lining is enough to be measured.
The tunnel section is damp. Flowing
Monitoring the wet and dry
area caused by cracks is commonly
-3 condition of the identified
1.35×10 present on the tunnel lining. The
3 cracks. Measure the water
~2.28×10-2 flow rate of most of location is
inflow rate of the critical
measurable but not enough to form
cracks.
intensive flow.
The tunnel section is wet. Flowing Monitoring the wet and dry
area caused by cracks is commonly condition of the identified
present on the tunnel lining. The cracks. Measure the water
-2
2.28×10 flow rate of most of location is inflow rate of critical
4
~1.59×10-1 measurable. A small number of cracks. Conduct small scale
cracks can form intensive flow remediation work for the
locally and affect the serviceability cracks with intensive water
of a short tunnel section. inflow.
Monitoring the wet and dry
The tunnel section is wet. Flowing
condition of the identified
area caused by cracks is commonly
cracks. Measure the water
present on the tunnel lining. A large
inflow rate of critical
5 >1.59×10-1 number of cracks can form intensive
cracks. Conduct large scale
flow locally and affect the
remediation work for the
serviceability of a long tunnel
cracks with intensive water
section.
inflow.
The time-dependent risk rating for each tunnel is shown as Table 5.10 to Table 5.13.
Table 5.10 Time-dependent risk rating of tunnel sections of CLP

Risk
categories Portal to FGS FGS to MCS MCS to PS PS to Portal
1 2017~2045 2017~2074 N/A 2017~2060
2 2045~2048 2074~2080 2017~2029 2060~2072
3 2048~2056 2080~2099 2029~2033 2072~2081
4 2056~2066 2099~2112 2033~2042 2081~2090
5 2066~ 2112~ 2042~ 2090~
228
Table 5.11 Time-dependent risk rating of tunnel sections of CCL
Risk
categories Portal to FGS FGS to MCS MCS to PS PS to Portal
1 2017~2064 2017~2046 * 2017~2081
2 2064~2074 2046~2055 * 2081~2090
3 2074~2080 2055~2062 * 2090~2103
4 2080~2088 2062~2067 * 2103~>2113
5 2088~ 2067~ * >2113
*not calculated due to no sign of wet crack increase
Table 5.12 Time-dependent risk rating of tunnel sections of BLP
Risk
categories
1 2017~2086 * N/A *
2 2086~2090 * N/A *
3 2090~2108 * 2017~2024 *
4 2108~>2113 * 2024~2031 *
5 >2113 * 2031~ *
*not calculated due to no sign of wet crack increase
Table 5.13 Time-dependent risk rating of tunnel sections of NLP
Risk
categories
1 2017~2035 2017~2028 N/A 2017~>2113
2 2035~2040 2028~2048 N/A >2113
3 2040~2047 2048~2055 N/A >2113
4 2047~2065 2055~2064 2017~2025 >2113
5 2065~ 2064~ 2025~ >2113
Water ingress is the failure mechanism for the selected tunnel in this study. Water ingress is
considered to have a high detectability and low consequence compared to structural failure.
Therefore, this tunnel can accept a higher risk as maintenance intervention procedures can be
applied. In this case, risk category five is defined as the threshold risk. Injection works is
necessary in the 5 category.
The risk rating is determined from the time-dependent reliability prediction which is
calculated using the existing database. The future condition of the tunnel is extrapolated from
229
the existing data. A continuous monitoring of wet crack information is necessary to further fit
the extrapolation.
5.5 Summary
This chapter provided a study of reliability prediction using water ingress as the assessment
criterion (research gap 5). Two failure modes which are water inflow through cracked
concrete or joints, and water seeping through sound concrete are calculated for the reliability
analysis of water ingress criterion. Site data and models presented in previous chapter are
used in this Chapter.
Water inflow into a tunnel is defined as the first failure mode of the water ingress criterion
( ). There are multiple components of the tunnels which contribute to the first failure mode.
One component is defined as one metre length of tunnel. Sensitivity analysis is conducted by
varying the correlation factor of two time point. Excessive components failure will cause the
failure of the water inflow mode. The failure probability of the failure mode 1 ( ) depends
on the acceptable percentage of the component failure. Four sections of each tunnel are
calculated. The MCS to PS section is most critical section for CLP, BLP and NLP. The
section from FGS to MCS is the most critical section for CCL tunnel. The BLP and NLP is
predicted to be failed in 13 years under the condition of 16% probability of 30% of the lining
experiencing water inflow exceeding 1 L/min/m of tunnel.
The section between FGS to MCS and PS to portal does not show signs of wet crack increase
in BLP based on the three years of inspection reports. The section between MCS to PS in
CCL also shows no sign of increasing number of wet cracks. The three sections are not
included in the reliability calculation of the water inflow failure mode ( ).

230
Water seeping into a tunnel lining is defined to be the second failure mode of the water
ingress criterion ( ). This failure of water seeping occurs slower compared to water inflow
through lining. The failure probability of the section failure mode ( ) of BLP and NLP is
predicted to start to rise after 80 years. The failure probability of the water seeping failure
mode is calculated to be less than 20% after 100 years for BLP and NLP. The failure
probability of the water seeping failure mode is predicted to reach 0.002 after 100 service
years for CLP and CCL.
The system failure probability of the tunnel subject to water ingress is calculated in this
chapter. The study found that water inflow through fully interconnected cracks and joints is
the main failure mechanism of the water ingress criterion.
231
6 Water Inflow Treatment of Tunnel Lining
6.1 Introduction
Water inflow is a detrimental factor to be addressed in tunnel design and assessment as
discussed in Section 2.2.4. Water inflow treatments could mitigate the potential hazards
through the service stage of the tunnel. Water inflow through cracks has been identified to be
the major factor that affects the tunnel service life according to the reliability calculation in
Chapter 5. The treatment of cracks is necessary for extending the service life of tunnels. The
injection method was used for water inflow treatment of cracks in the selected tunnel. The
identified cracks in the selected tunnel were treated in early 2013 by the tunnel operator. Four
water inflow treatment methods were discussed as in Section 2.2.4, namely, drainage,
injection, patching and lining replacement. In this specific case, water inflow into tunnels
occurred through cracks and lining joints. The injection method was used for water inflow
treatment in the selected tunnel. The tunnel condition was then inspected after the injection
treatment from September 2013 to December 2016 by the tunnel operator. The effect of the
treatment of this selected tunnel is evaluated based on the inspection data available.
This chapter evaluates the effectiveness of water inflow treatment in the selected tunnel
(research gap 6). The number of wet cracks is compared with the number of cracks treated in
each of the tunnel sections. The number of failed treatments was identified and counted in
different tunnel sections after the injection works.
6.2 Injection material
As discussed in Section 6.1 that injection method was used in the selected tunnel. The
properties of different types of sealant are discussed in this section. The chemical and
232
mechanical properties of sealant used in injection treatment can vary. The properties and the
applications of three types of sealant are discussed as follows.
6.2.1 Epoxy injection resins
The epoxy injection resins were used for structural filling of cracking in concrete in an
experimental study (Karayannis et al., 1998). The adhesive strength of the epoxy resin is high
and provides structural bonding of cracks. The epoxy resin usually has low viscosity and is
able to fill cracks with widths smaller than 0.1 mm. The viscosity of the epoxy resins varies
from 95 mPa.s (0.95 poise) to 700 mPa.s (7 poise) when the temperature is 23 degrees
Celsius (WEBAC, 2017b). The workable life of the epoxy resins varies from 20 min to 100
min when the temperature is 20 degrees Celsius (WEBAC, 2017b).
6.2.2 Polyurethane injection resins
Polyurethane resins were used for the concrete polymer injection due to their high elasticity
and good adhesion (WEBAC, 2017c). Polyurethane can be quite flexible allowing for
movement when cured. The water pressure in the crack will determine whether a preliminary
injection is required. The preliminary injection is usually conducted with foam resin to
temporarily stop the water flow locally under high water pressure. A permanent higher
density polyurethane injection should be then injected into the crack (WEBAC, 2017c).
6.2.3 Silicate injection resins
Silicate injection resins are known to be a non-flammable material (WEBAC, 2017d). It is a
rigid material which is suitable for mechanical cutting during a tunnel construction stage.
Multiple applications including rock stabilization and tunnel face stabilisation have been
reported (WEBAC, 2017d).
233
6.3 Crack treatment in tunnel lining
It has been observed that water inflow occurs through cracks and joints of the lining of the
selected tunnel. The dripping and flowing of water inflow have caused concerns to the tunnel
operator about the serviceability of the tunnel.
A large-scale tunnel investigation has been conducted by the tunnel operator. The cracks on
the tunnel lining have been ranked in five categories based on both visual inspection and
ground penetrating radar scanning. The cracks have been separated as ‘category five crack’
and ‘non-category five crack’ by the tunnel operator. The category five cracks were generally
identified to have more water inflow problems than the non-category five cracks. The
category five cracks represent the most critical cracks which need treatment.
Polymer injection works have been done in early 2013 in order to control water inflow and
extend the tunnel service life. The injection holes were drilled 45 degrees into the concrete to
intersect the crack at a depth of approximately 100 mm inside the lining at 200 mm spacing
on alternate sides of the crack. The diameter of the drill hole is approximately 10 mm . Low
viscosity foam polyurethane (TamPur 130) was used as the first stage injection for polymer
injection to provide an initial control water flow in the identified cracks. The first stage
injection for one crack was finished when the sealant could be seen from emanating from the
crack at the lining surface. The second stage of injection was conducted using a polyurethane
resin (TamPur 150) to push into the low density foam of the first stage injection and complete
the seal. The second stage of injection takes approximately 2-3 minutes per packer under low
pressure. The whole injection is completed after 12 hours curing time for the resin. The
injection pressure of the selected tunnel treatment varies by location. The injection pressure
was controlled to be under the pressure which may lead to concrete cracking. The typical
234
pressure for polymer injection varies from 40 psi (275.8 kPa) to 100 psi (689.5 kPa) as
discussed in Section 2.2.4.
There were 254 cracks treated in all four tunnels. It has been recorded that there were 85
treated cracks in BLP, 78 in CCL 42 in CLP and 49 in NLP. Each tunnel was further divided
into four sections by the station locations as illustrated in Sections 5.2.1.
Three years of visual inspection monitoring has been conducted by the tunnel operator from
September 2013 after the cracks treatment. The data of the number of wet cracks has been
presented in Section 5.2. The new wet cracks emerging of each section was evaluated through
eight inspections in 3 years as shown in in Section 5.2. There are a total of 16 sections in the
tunnel system since each tunnel has four sections which divided by the three stations. The
number of wet cracks shows an increasing trend based on the site inspections in the selected
tunnel. A linear regression was used to fit the number of wet cracks as shown in Section 5.2.
The slope of the linear regression line was defined as the increasing rate of the number of wet
cracks. The increasing rate of the number of wet cracks was summarised by different tunnels
and sections in Section 5.2. A high increasing rate of number of wet cracks would indicate a
rapid worsening process of the tunnel lining. The increasing rate of number of wet cracks is
not directly proportional to the reliability of the tunnel, since factors including hydraulic
pressure and rock mass property are also contributing factors as discussed in Chapter 5.
The polymer injection treatment aims to repair the cracks with rapid water inflow and to
extend the service life of tunnel. The number of crack treatments of different tunnel sections
varies depending on the original lining conditions. The number of treated cracks in each
tunnel section was recorded. The tunnel sections have different length, both the number of
treated cracks and increasing rate of number of wet cracks are normalised and expressed
235
using the unit of per 100 m length of tunnel. The normalised number of treated cracks in
different tunnel sections was plotted against the normalised increasing rate of number of wet
cracks as shown in Figure 6.1.
7
Increasing rate of number of wet crack
6
BLP
5
CCL
4 CLP
(100*m-1/year)
3 NLP
-1
-2
0 2 4 6 8
Number of treated crack (100*m-1)
Figure 6.1 Relation of increasing rate of number of wet cracks and number of treated
crack, BLP, CCL, CLP and NLP refer to different tunnels
It can be seen from Figure 6.1 that the rate of wet crack increase in relative to the number of
treated cracks is highly variable. However, the number of treated cracks shows a general
negative relation with the increasing rate of number of wet cracks. That would suggest that
the more crack treatment conducted in a section, the less new wet cracks develop in that
section.
The section from MCS to PS was selected as the focus section for the study in the initial
stages of the project. The general depth of MCS to PS section is 35m for BLP and NLP
which is the deepest section of the whole tunnels. It is likely that the hydraulic pressure in
236
MCS to PS section is greater than the other sections. The crack treatment and the increasing
rate of number of wet cracks in this section of the tunnels are compared in Figure 6.2.
6
Increasing rate of number of wet crack
MCS to PS
5
BLP
(100*m-1/year)
4
CCL
CLP
3
NLP
0
0 2 4 6 8
Number of treated crack (100*m-1)
Figure 6.2 Relation of increasing rate of number of wet cracks and number of treated
crack in section between MCS and PS, BLP, CCL, CLP and NLP refer to different
tunnels
The increasing rate of number of wet cracks has a strong correlation with the number of
treated cracks in the MCS to PS section (Figure 6.2). This observation suggests that the
treatment of cracks by injection with a foam polyurethane and resin products has been
beneficial in reducing the water inflow into the tunnels.
6.4 Failed treatments
It has been found from the records that water inflow of some cracks does not stop even after
the injection treatment. Those locations were defined as failed treatments. The failed
237
treatments in the tunnel have been recorded by the tunnel operator from September 2013 to
December 2014. In total, 41 category five crack treatments were reported as failed based on
the operator inspection in September 2013. The summary from the tunnel operator does not
show the number of failed treatments from December 2014 to December 2016. The location
of each treated category five crack has been compared with the record of wet cracks from
December 2014 to December 2016 in this study. The number of failed treatments from
December 2014 to December 2016 has been identified manually as part of this study. The
number of failed treatments in the selected tunnel is shown in Table 6.1.
Table 6.1 Number of failed treatments in different tunnels and the total number of
treated cracks is in brackets
Total
BLP (85) CLP (42) CCL (78) NLP (49)
(254)
Sep-2013 24 13 2 2 41
May-2014 26 17 4 3 50
Jul-2014 21 18 4 5 48
Dec-2014 24 11 7 11 53
Aug-2015 19 11 9 9 48
Dec-2015 20 11 11 9 51
Aug-2016 18 10 11 9 48
Dec-2016 28 12 11 11 62
The time-dependent changing of the number of failed treatment in Figure 6.3 four tunnels is
plotted in based on the data in Table 6.1.
238
30 30
Number of failed treatment

25 25
20 20
15 15
10 10
CCL
CLP
5 5
0 0
Apr-12 Aug-13 Dec-14 May-16 Sep-17 Apr-12 Aug-13 Dec-14 May-16 Sep-17
(a) (b)
30 30

25 25
20 20
15 15
10 10 NLP
BLP
5 5
0 0
Apr-12 Aug-13 Dec-14 May-16 Sep-17 Apr-12 Aug-13 Dec-14 May-16 Sep-17
(c) (d)
Figure 6.3 Number of failed treatment, (a) CLP, (b) CCL, (c) BLP and (d) NLP
The number of failed treatments in all the four tunnels varies based on Table 6.1. The number
of failed treatments in CLP varies between 10 and 18. The data in CLP do not show a clear
trend of time-dependent variation. The number of failed treatment in BLP varies between 18
and 28. The number of failed treatments in BLP shows a decreasing trend until the last
inspection. The number of failed treatments in CCL shows an increasing trend from the first
five inspections. The number of failed treatments stabilised at 11 according to the last three
inspections in CCL. The number of failed treatment in NLP shows an increasing trend overall.
239
A sudden increase in number of failed treatments was observed in the fourth inspection in
NLP.
It is notable that the depth of water table in this selected tunnel would vary seasonally. The
water table in this case would be lower in dry season (February to April) compared to that in
wet season (August to October). It is possible that the seasonal variation of water table drying
cracks which are flowing in the wet season. Thus, some failed treatments would not be
observed in the inspections in dry seasons.
The efficiency of crack treatment depends on both sealant properties and environmental
effects. The properties of the sealant used in the selected tunnel are beyond the scope of this
research. The environmental effects in the selected tunnel mainly refer to the changing of
hydraulic pressure on the lining which caused by the injection of cracks.
The injection of the cracks in the example is approximately 100 mm into the lining as
discussed in Section 6.3. The injection depth is shallower compared to a typical injection
depth which is halfway into the lining as discussed in Section 2.2.4. The injection into a
shallow depth may not be able to fill the entire cracks under hydraulic pressure. The injected
material in the selected tunnel would be under greater hydraulic gradient accordingly as
shown in Figure 6.4.
240
Figure 6.4 Illustration of shallow injection depth in the selected tunnel
The outer lining surface is the interface of rock mass and tunnel lining. The inner lining
surface is the lining surface we can see when walking in the tunnel. The definition of the
inner and outer lining surface was defined in Chapter 3.
The injection works conducted in the selected tunnel would decrease the overall hydraulic
conductivity of the lining by reducing the effective crack numbers. The decrease of hydraulic
conductivity would increase the hydraulic pressure on the lining based on the analytical
relation as discussed in Section 4.4. Therefore, the lining of the selected tunnel could be
under extra hydraulic pressure on the lining after injection. It is possible that the sealant could
fail under the extra hydraulic pressure in a certain circumstance combining with the
possibility that the sealant does not fulfil the entire crack as shown in Figure 6.4.
The injection depth of the crack is shallow (approximately 20% of the lining thickness)
relative to the lining thickness in the selected tunnel. The partially sealed crack would create
a constrictive inhomogeneous hydraulic conductivity of the lining. The inhomogeneous
241
hydraulic conductivity would affect the water inflow rate and hydraulic pressure on the lining.
This situation is relevant to the analysis of the inhomogeneous hydraulic conductivity as was
discussed in Chapter 3. The shallow injection depth would cause the potential for
inhomogeneity of the lining. A deeper injection depth is recommended for further injection
works (approximately 50% of the lining thickness) as discussed in Chapter 2 in order to push
the high water head away from the lining inner surface.
It is difficult to establish a statistical significant interpretation from the data of the number of
failed treatments. A long-term investigation is required in order to understand the changing
mechanism of the number of failed treatments as many factors are involved in this process.
Water could find alternative pathways through the un-filled crack network even if the original
flow channel is blocked by injection works. This could be one reason for the variation of the
number of failed treatments. Another possible reason for the variation would include the
property changes of the sealant over time. A study to evaluate the polymer injection
effectiveness could be developed when data on the rate of sealant breakdown and the rate of
water finding new pathways is obtained.
6.5 Summary
The study in this chapter evaluated the effectiveness of water inflow treatment in the selected
tunnel (research gap 6). The wet crack numbers have been compared with the water inflow
treatment numbers in different tunnel sections in this chapter. The result shows that the
number of crack treatments generally has a negative correlation with the increasing rate of
wet cracks. For the tunnel section between MCS to PS, the increasing rate of number of wet
cracks has a strong correlation with the number of treated cracks. The data of the selected
tunnel show that the crack treatment program in the selected tunnel would mitigate the water
242
inflow rate of the tunnel. The overall crack injection program in the selected tunnel is
effective and the objective is therefore achieved.
The number of failed treatments in different tunnel sections has been summarised in Section
6.4. The number of failed treatments measured in this study is not sufficient to establish a
statistically significant interpretation. The reasons which could lead to the failure of
treatments have been discussed in Section 6.4. The failure mechanism of the treatments could
be interpreted when a long-term inspection of the wet crack number is developed.
243
7 Conclusions and Recommendations
This chapter aims to present the main findings and conclusions based on the water ingress
study of the selected tunnel investigated in this thesis. A method is developed in this study for
evaluating the service life of a tunnel based on the criterion of water ingress. The developed
theory was used to calculate the time dependent failure probability of a tunnel based on the
criteria of water ingress.
Six research gaps identified in Chapter 2 have been filled through the study.
The deliverables of this thesis include, model of hydraulic conductivity of concrete with
partially interconnected cracks, method of measuring water inflow rate on a tunnel lining,
method of crack width calculation based on site measured data, model of water inflow rate
through inhomogeneous lining and time-dependent reliability prediction of the selected
tunnels.
The following conclusions of this thesis are drawn as follows based on the solution of each
research gap.
1. A model is developed to calculate the hydraulic conductivity of cracked concrete. The
concrete is divided into three categories by cracking condition, namely sound concrete,
concrete with partially interconnected cracks and concrete with fully interconnected
cracks. A connectivity coefficient is developed to quantify the crack connectivity in
concrete. The connectivity coefficient range 0 to 1 is used to describe the cracking
profile from sound concrete to a fully through going crack. That study found that the
hydraulic conductivity of a cracked concrete increases steadily when the crack
connectivity increases from 0 to 0.99. The hydraulic conductivity of concrete with a
fully interconnected crack will be several orders magnitude higher than the partially
244
cracked condition. The calculation shows that the cracks in the lining of the selected
tunnel have to be fully interconnected through the lining in order to form the least
amount of the measured water flow rate. The partially interconnected crack can only
form efflorescence on the lining surface.
2. Water inflow rate measurement on site in a tunnel is difficult when the water is
flowing along the tunnel wall. The water flow can form a fan shape or a triangle shape
on the wall when water flowing onto the tunnel lining surface. A method was
developed for total water inflow rate measurement by only measuring a section of the
flow area. The water inflow rate of different sections of the flow area is expressed in
an analytical format. A general application for this method is given in that chapter. It
shows that the most rapid water flow occurs along the crack, which is usually the
centre line of the fan shape of the flow area.
3. Crack width within concrete is often difficult to measure directly due to the narrow
opening and the deposits of mineral salts around cracks. A method was developed to
calculate the crack width based on the water inflow rate through the crack when the
other hydraulic parameters are known. This study is based on the assumption that the
tunnelling lining is homogeneous. The water inflow rate measured from site is used in
the calculation. The result shows that the cracks in the selected tunnel have a mean
crack width of approximately 0.12 mm from 9 measured locations. A design chart
which comprises all the hydraulic parameters is developed for the general application
of crack width calculation.
4. A model is developed to study the water inflow rate and the hydraulic pressure
through an inhomogeneous tunnel lining. Traditional methods for calculating water
inflow through a lining are largely based on the assumption that water flow through a
homogeneous lining. The condition of tunnel with an inhomogeneous lining is

245
evaluated in this study. The inhomogeneity of lining is quantified by an
inhomogeneity coefficient which describes a linear hydraulic conductivity variation
through the lining. The water inflow variation and the hydraulic head variation caused
by the inhomogeneity of the lining hydraulic conductivity are quantified.
Comparisons of water inflow rate and hydraulic pressure are made to illustrate the
difference between a homogeneous lining condition and an inhomogeneous lining
condition. A design chart is developed for general applications of the model. The
design chart could be used to interpret the lining condition at the rock tunnel interface
which cannot be seen using tradition method.
5. The reliability of the selected tunnel system is predicted with a time-dependent
approach. Water ingress is defined as the failure criterion in this thesis. The water
ingress criterion comprise two failure modes, namely water seeping and water inflow.
The tunnel is defined as failed once water seeping front seep through the lining or
water inflow rate of a tunnel section exceed an acceptable limit. The site measured
data from year 0 to year 3 is used for reliability calculation. Each tunnel is divided by
four sections by stations. It is found that the water inflow through crack is the
controlling failure mode for most tunnel sections in the selected tunnel. The critical
section is identified and the remaining life of the selected tunnel is given under a
certain condition.
6. The effectiveness of the crack treatment in the selected tunnel is discussed. The crack
treatment numbers seems to have a negative relation with the crack increasing in the
MCS to PS section. The number of failed treatments has been manually identified
from the inspection data which provided by the tunnel operator. This research
objective is achieved based on the short term data. However the data of the failed
treatments is not sufficient to establish a statistical significant interpretation of the

246
failure mechanism. The failure mechanism of the treatments could be interpreted
based on a long-term inspection of the wet crack number and sealant breakdown rate.
It is notable that the rock mass is considered as a homogeneous material in this study. The
inhomogeneity and stratification of rock mass can often affect the design of tunnels
significantly in a typical engineering case. The hydraulic conductivity variation of rock mass
could be a potential direction of further research.
The water ingress was found to cause loss of calcium ion of the concrete lining during the site
inspection. Analyse the effect of calcium leaching of a tunnel lining could also be a direction
of further research.
247
8 Reference
ABC. 2001. Mysterious crack floods Melbourne's Citylink tunnel [Online]. Available:
http://www.abc.net.au/7.30/stories/s250493.htm 12/07/2017].
AKHAVAN, A., SHAFAATIAN, S.-M.-H. & RAJABIPOUR, F. 2012. Quantifying the
effects of crack width, tortuosity, and roughness on water permeability of cracked
mortars. Cement and Concrete Research, 42, 313-320.
ALDEA, C.-M., SHAH, S. P. & KARR, A. 1999a. Effect of cracking on water and chloride
permeability of concrete. Journal of materials in civil engineering, 11, 181-187.
ALDEA, C., M, S.P.SHAH & A.KARR 1999b. Permeability of cracked concrete. Materials
and Structures, 32, 370-376.
AS/NZS 2004. Standards Australia and Standards New Zealand 4360: 2004 risk management.
BAKER, R. 2001. Burnley tunnel still using one million litres. THE AGE.
BARENBLATT, G. I. 1996. Scaling, self-similarity, and intermediate asymptotics:
dimensional analysis and intermediate asymptotics, Cambridge University Press.
BARTON, N., LIEN, R. & LUNDE, J. 1974. Engineering classification of rock masses for
the design of tunnel support. Rock mechanics, 6, 189-236.
BENTZ, D. P. & GARBOCZI, E. J. 1991. Percolation of phases in a three-dimensional
cement paste microstructural model. Cement and Concrete Research, 21, 325-344.
BENTZ, D. P. & GARBOCZI, E. J. 1992. Modelling the leaching of calcium hydroxide from
cement paste: effects on pore space percolation and diffusivity. Materials and
Structures, 25, 523-533.
BEZUIJEN, A. & TALMON, A. Grout pressures around a tunnel lining, influence of grout
consolidation and loading on lining. World Tunnel Congress 2004, 2004 Singapore.
Singapore Tunnelling Society, 22-27.
BIENIAWSKI, Z. T. 1990. Tunnel design by rock mass classifications. DTIC Document.
BREYSSE, D. & GÉRARD, B. 1997. Modelling of permeability in cement-based materials:
Part 1-Uncracked medium. Cement and concrete research, 27, 761-775.
BROWN, E. 2006. Forensic engineering for underground construction. Rock Mechanics In
Underground Construction: (With CD-ROM). World Scientific.
BTS, I. 2004. Tunnel Lining Design Guide, London: Thomas Telford, British Tunneling
Society and the Institution of Civil Engineers.
BUCKINGHAM, E. 1907. Studies on the movement of soil moisture. US Department of
Agricultural, 38.
BUKAÇI;, E., KORINI;, T., PERIKU;, E., ALLKJA;, S. & SHEPERI, P. 2016. Probability
of failure determination for tunnels in rock by using Monte Carlo simulation.
Engineering Structures&Materials, 3.
BUSLENKO, N. P. 1966. The Monte Carlo method: the method of statistical trials, Oxford,
Pergamon.
BUTSCHER, C. 2012. Steady-state groundwater inflow into a circular tunnel. Tunnelling and
Underground Space Technology, 32, 158-167.
CARMELIET, J., DELERUE, J. F., VANDERSTEEN, K. & ROELS, S. 2004. Three‐
dimensional liquid transport in concrete cracks. International journal for numerical
and analytical methods in geomechanics, 28, 671-687.
CESANO, D., BAGTZOGLOU, A. C. & OLOFSSON, B. 2003. Quantifying fractured rock
hydraulic heterogeneity and groundwater inflow prediction in underground
excavations: the heterogeneity index. Tunnelling and Underground Space Technology,
18, 19-34.
248
CESANO, D., OLOFSSON, B. & BAGTZOGLOU, A. 2000. Parameters regulating
groundwater inflows into hard rock tunnels—a statistical study of the Bolmen tunnel
in southern Sweden. Tunnelling and Underground Space Technology, 15, 153-165.
CHEN, J.-B. & LI, J. 2005. Dynamic response and reliability analysis of non-linear stochastic
structures. Probabilistic Engineering Mechanics, 20, 33-44.
CHEN, R. & TONON, F. Fracture Cluster Modeling for Groundwater Inflow Prediction into
Rock Tunnels Using Geostatistics. GeoCongress 2012, 2012 Oakland, California.
2372-2381.
CHOINSKA, M., KHELIDJ, A., CHATZIGEORGIOU, G. & PIJAUDIER-CABOT, G. 2007.
Effects and interactions of temperature and stress-level related damage on
permeability of concrete. Cement and Concrete Research, 37, 79-88.
CORNELL, C. A. 1967. Bounds on the reliability of structural systems. Journal of the
Structural Division, 93, 171-200.
CORNELL, C. A. 1969. A probability-based structural code. Journal of the American
Concrete Institute, 66, 974-985.
CUI, L. & CAHYADI, J. H. 2001. Permeability and pore structure of OPC paste. Cement and
Concrete Research, 31, 277-282.
DER KIUREGHIAN, A. & LIU, P.-L. 1986. Structural reliability under incomplete
probability information. Journal of Engineering Mechanics, 112, 85-104.
DIAMOND, S. & HUANG, J. 2001. The ITZ in concrete–a different view based on image
analysis and SEM observations. Cement and Concrete Composites, 23, 179-188.
DIJKSTRA, E. W. 1959. A note on two problems in connexion with graphs. Numerische
mathematik, 1, 269-271.
DITLEVSEN, O. 1981. Uncertainty modeling with applications to multidimensional civil
engineering systems, New York, McGraw-Hill International Book Co.
DITLEVSEN, O. 1983. Gaussian outcrossings from safe convex polyhedrons. Journal of
Engineering Mechanics, 109, 127-148.
DOMENICO, P. A. & SCHWARTZ, F. W. 1990. Physical and chemical hydrogeology, New
York, Wiley.
DUNCAN, J. M. 2000. Factors of safety and reliability in geotechnical engineering. Journal
of geotechnical and geoenvironmental engineering, 126, 307-316.
EDVARDSEN, C. 1999. Water permeability and Autogenous Healing of Cracks in Concrete.
ACI Materials Journal, 448-454.
EL TANI, M. Water inflow into tunnels. Proceedings of the World Tunnel Congress ITA-
AITES, 1999 Oslo, Norway. 61-70.
EL TANI, M. 2003. Circular tunnel in a semi-infinite aquifer. Tunnelling and Underground
Space Technology, 18, 49-55.
ESTES, A. C. & FRANGOPOL, D. M. 1998. RELSYS: A computer program for structural
system reliability. Structural engineering and mechanics, 6, 901-919.
FERNÁNDEZ, G. 1994. Behavior of pressure tunnels and guidelines for liner design.
Journal of Geotechnical Engineering, 120, 1768-1791.
FERNÁNDEZ, G. & ALVAREZ, T. A. 1994. Seepage-induced effective stresses and water
pressures around pressure tunnels. Journal of geotechnical engineering, 120, 108-128.
FERNANDEZ, G. & MOON, J. 2010a. Excavation-induced hydraulic conductivity reduction
around a tunnel–Part 2: Verification of proposed method using numerical modeling.
Tunnelling and Underground Space Technology, 25, 567-574.
FERNANDEZ, G. & MOON, J. 2010b. Excavation-induced hydraulic conductivity reduction
around a tunnel – Part 1: Guideline for estimate of ground water inflow rate.
249
FETTER, C. W. 2000. Applied hydrogeology, Merrill, Columbus, OH, Prentice hall.
FONT-CAPÓ, J., VÁZQUEZ-SUÑÉ, E., CARRERA, J., MARTÍ, D., CARBONELL, R. &
PÉREZ-ESTAUN, A. 2011. Groundwater inflow prediction in urban tunneling with a
tunnel boring machine (TBM). Engineering Geology, 121, 46-54.
FREUDENTHAL, A. 1962. Safety, reliability and structural design. Transactions of the
American Society of Civil Engineers, 127, 304-318.
GÉDRARD, B., BREYSSE, D., AMMOUCHE, A., HOUDUSSE, O. & DIDRY, O. 1996.
Cracking and permeability of concrete under tension. Materials and Structures, 29,
141-151.
GÉRARD, B., LE BELLEGO, C. & BERNARD, O. 2002. Simplified modelling of calcium
leaching of concrete in various environments. Materials and Structures, 35, 632-640.
GILKEY, H. J. Water-cement ratio versus strength-another look. Journal Proceedings, 1961.
1287-1312.
GOODMAN, R. E., MOYE, D. G., VAN SCHALKWYK, A. & JAVANDEL, I. 1965.
Ground water inflows during tunnel driving, Berkeley, College of Engineering,
University of California.
GROB, H. 1975. Swelling and heave in Swiss tunnels. Bulletin of Engineering Geology and
the Environment, 14, 55-60.
HANSEN, T. C. 1986. Physical structure of hardened cement paste. A classical approach.
Materials and Structures, 19, 423-436.
HARR, M. E. 1962. Groundwater and seepage, New York, NY, McGraw Hill.
HARRISON, T. 1981. Early-age thermal crack control in concrete, London, Construction
Industry Research & Information Assoc.
HASOFER, A. M. & LIND, N. C. 1974. Exact and invariant second-moment code format.
Journal of the Engineering Mechanics division, 100, 111-121.
HEUER, R. E. Estimating rock tunnel water inflow. Proceedings of the Rapid Excavation
and Tunneling Conference, 1995. SOCIETY FOR MINING, METALLOGY &
EXPLORATION, INC, 41-60.
HUANG, Z., JIANG, Z., ZHU, S., WU, X., YANG, L. & GUAN, Y. 2016. Influence of
structure and water pressure on the hydraulic conductivity of the rock mass around
underground excavations. Engineering Geology, 202, 74-84.
HUDSON, J. A. & FENG, X.-T. 2015. Rock engineering risk, Boca Raton, FL, CRC Press.
HUGHES, D. 1985. Pore structure and permeability of hardened cement paste. Magazine of
HUNG, C. J., MONSEES, J., MUNFAH, N. & WISNIEWSKI, J. 2009. Technical manual for
design and construction of road tunnels–civil elements. US Department of
Transportation, Federal Highway Administration, National Highway Institute, New
York.
HWANG, J.-H. & LU, C.-C. 2007. A semi-analytical method for analyzing the tunnel water
inflow. Tunnelling and Underground Space Technology, 22, 39-46.
ITA 1988. Guidelines for the design of tunnels. Tunnelling and underground space
technology, 3, 237-249.
ITA 1991. Report on the damaging effects of water on tunnels during their working life.
JACOBS, F. & WITTMANN, F. 1992. Long term behavior of concrete in nuclear waste
repositories. Nuclear engineering and design, 138, 157-164.
KARAYANNIS, C., CHALIORIS, C. & SIDERIS, K. 1998. Effectiveness of RC beam-
column connection repair using epoxy resin injections. Journal of Earthquake
Engineering, 2, 217-240.
250
KATZ, A. & THOMPSON, A. 1986. Quantitative prediction of permeability in porous rock.
Physical review B, 34, 8179.
KITTEROD, N. O., COLLEUILLE.H, WONG.W.K & PEDERSEN.T.S 2000. Simulation of
groundwater drainage intoa tunnel in fractured rock and numerical analysis of leakage
remediation, Romeriksporten tunnel, Norway. Hydrogeology Journal, 8, 480-493.
KOLYMBAS, D. & WAGNER, P. 2007. Groundwater ingress to tunnels–the exact analytical
solution. Tunnelling and Underground Space Technology, 22, 23-27.
KUESEL, T. R., KING, E. H. & BICKEL, J. O. 2012. Tunnel engineering handbook,
Melbourne, FL, Springer Science & Business Media.
LARSON, R. & FARBER, E. 2012. Elementary statistics, New York, Pearson Prentice Hall.
LEBEAU, M. & KONRAD, J. M. 2010. A new capillary and thin film flow model for
predicting the hydraulic conductivity of unsaturated porous media. Water Resources
Research, 46.
LEEMANN, A., MÜNCH, B., GASSER, P. & HOLZER, L. 2006. Influence of compaction
on the interfacial transition zone and the permeability of concrete. Cement and
LEI, S. 1999. An Analytical Solution for Steady Flow into a Tunnel. Groundwater, 37, 23-26.
LEONARD, J. 2006. Hydrogeology of the Melbourne Area. Australia Geomechanics Society,
41, 63-74.
LEPECH, M. D. & LI, V. C. 2009. Water permeability of engineered cementitious
composites. Cement and Concrete Composites, 31, 744-753.
LI, C. Q. 2005. Time dependent reliability analysis of the serviceability of corrosion affected
concrete structures. International Journal of Materials & Structural Reliability, 3,
105-116.
LI, C. Q., FIROUZI, A. & YANG, W. 2016. Closed-Form Solution to First Passage
Probability for Nonstationary Lognormal Processes. Journal of Engineering
Mechanics, 142, 6-23.
LI, C. Q. & MELCHERS, R. E. 1993. Outcrossings from convex polyhedrons for
nonstationary Gaussian processes. Journal of engineering mechanics, 119, 2354-2361.
LI, C. Q. & MELCHERS, R. E. 2005. Time-dependent risk assessment of structural
deterioration caused by reinforcement corrosion. ACI Structural Journal, 102, 754-
762.
LIAN, C. & ZHUGE, Y. 2010. Optimum mix design of enhanced permeable concrete–an
experimental investigation. Construction and Building Materials, 24, 2664-2671.
LIAN, C., ZHUGE, Y. & BEECHAM, S. 2011. The relationship between porosity and
strength for porous concrete. Construction and Building Materials, 25, 4294-4298.
LIU, T., CHEN, Y.-B. & LIU, H. 2012. Case study of ultra-deep foundation pit by island and
cofferdam construction in soft soils in coastal areas. Chinese Journal of Geotechnical
Engineering, S1.
LONG, G.-C., XIE, Y.-J., DENG, D.-H. & LI, X.-K. 2011. Deterioration of concrete in
railway tunnel suffering from sulfate attack. Journal of Central South University of
Technology, 18, 881.
LONG, J., REMER, J., WILSON, C. & WITHERSPOON, P. 1982. Porous media equivalents
for networks of discontinuous fractures. Water Resources Research, 18, 645-658.
LOUIS, C. 1969. A study of groundwater flow in jointed rock and its influence on the
stability of rock masses, Imperial College of Science and Technology.
LÜTH, S., GIESE, R., OTTO, P., KRÜGER, K., MIELITZ, S., BOHLEN, T. &
DICKMANN, T. 2008. Seismic investigations of the Piora Basin using S-wave
251
conversions at the tunnel face of the Piora adit (Gotthard Base Tunnel). International
Journal of Rock Mechanics and Mining Sciences, 45, 86-93.
MADANSKY, A. 1965. Approximate confidence limits for the reliability of series and
parallel systems. Technometrics, 7, 495-503.
MAIDL, B., THEWES, M. & MAIDL, U. 2013. Handbook of Tunnel Engineering I :
Structures and Methods, Somerset, NJ, USA, John Wiley & Sons.
MAIDL, B., THEWES, M. & MAIDL, U. 2014. Handbook of Tunnel Engineering II: Basics
and Additional Services for Design and Construction, John Wiley & Sons.
MEHTA, P. K. 1986. Concrete. Structure, properties and materials, New York, Prentice-Hall.
MELCHERS, R. E. 1999. Structural reliability : analysis and prediction, Chichester, West
Sussex, England, John Wiley & Sons, Ltd.
MOON, J.-S. 2007. Evaluation and assessment of inflow rates in tunnels excavated in jointed
rock mass, University of Illinois at Urbana-Champaign.
MORGAN, D. 1996. Compatibility of concrete repair materials and systems. Construction
and building materials, 10, 57-67.
MOTHERSILLE, D. & LITTLEJOHN, S. 2012. Grouting of anchors to resist hydrostatic
uplift at Burnley Tunnel, Melbourne, Australia. Grouting and Deep Mixing 2012.
MURAKAMI, T., AHN, T., HASHIMOTO, T., OGURA, N. & KISHI, T. A study on the
new water leak repair method for subway tunnels using crack self-healing repair
materials. Fifth International Conference of Self-healing Materials, 2015. 1-4.
MUSKAT, M. & WYCKOFF, R. D. 1937. Flow of homogeneous fluids through porous
media, New York, McGraw-Hill.
NAKAHARA, R. 2014. Partnership marketing for prevention. Journal of Global Scholars of
Marketing Science, 24, 160-171.
NAVIER, C. 1823. Mémoire sur les lois du mouvement des fluides. Mémoires de l’Académie
Royale des Sciences de l’Institut de France, 6, 389-440.
NEVILLE, A. M. 2011. Properties of concrete, Harlow, Pearson.
NEVILLE, A. M. & BROOKS, J. J. 1987. Concrete technology, Harlow, Longman Scientific
and Technical.
NEWMARK, N. M. 1953. An engineering approach to blast resistant design. DTIC
Document.
NIELSEN, L. F. 1993. Strength development in hardened cement paste: examination of some
empirical equations. Materials and Structures, 26, 255-260.
O'ROURKE, T. D. Guidelines for tunnel lining design. 1984. ASCE Technical Committee on
tunnel lining design of the Underground Technology Research Council.
ODA, M. 1985. Permeability tensor for discontinuous rock masses. Geotechnique, 35, 483-
495.
OLLIVIER, J., MASO, J. & BOURDETTE, B. 1995. Interfacial transition zone in concrete.
Advanced Cement Based Materials, 2, 30-38.
PARK, K.-H., OWATSIRIWONG, A. & LEE, J.-G. 2008. Analytical solution for steady-
state groundwater inflow into a drained circular tunnel in a semi-infinite aquifer: a
revisit. Tunnelling and Underground Space Technology, 23, 206-209.
PECK, W. A. 1992. Engineering geology of Melbourne : proceedings of the Seminar on
Engineering Geology of Melbourne, Melbourne, Victoria, Australia, 16 September
1992, Rotterdam, A.A. Balkema.
PHIEN-WEJ, N., GIAO, P. & NUTALAYA, P. 2006. Land subsidence in Bangkok, Thailand.
Engineering geology, 82, 187-201.
252
PICANDET, V., KHELIDJ, A. & BASTIAN, G. 2001. Effect of axial compressive damage
on gas permeability of ordinary and high-performance concrete. Cement and Concrete
Research, 31, 1525-1532.
POLUBARINOVA-KOCHINA, P. Y. 1963. Theory of Ground Water Movement, New
Jersey, Princeton University Press.
POWERS, T. C. 1958. Structure and physical properties of hardened Portland cement paste.
Journal of the American Ceramic Society, 41, 1-6.
POWERS, T. C., COPELAND, L., HAYES, J. & MANN, H. 1954. Permeablity of Portland
Cement Paste. ACI Journal proceedings. American Concrete Institute.
RACKWITZ, R. & KRZYKACZ, B. 1978. Structural reliability of reactor systems.
Probabilistic analysis of nuclear reactor safety.
RAUSAND, M. & ARNLJOT, H. 2004. System reliability theory: models, statistical
methods, and applications, John Wiley & Sons.
RAVEN, K. & GALE, J. Water flow in a natural rock fracture as a function of stress and
sample size. International Journal of Rock Mechanics and Mining Sciences &
Geomechanics Abstracts, 1985. Elsevier, 251-261.
REINHARDT, H.-W. & JOOSS, M. 2003. Permeability and self-healing of cracked concrete
as a function of temperature and crack width. Cement and Concrete Research, 33,
981-985.
REINHARDT, H. W. 1997. Penetration and permeability of concrete : barriers to organic and
contaminating liquids : state-of-the-art report prepared by members of the RILEM
Technical Committee 146-TCF, London ; New York, E & FN Spon.
REINHARDT, H. W. & GABER, K. 1990. From pore size distribution to an equivalent pore
size of cement mortar. Materials and Structures, 23, 3.
RICE, S. O. 1944. Mathematical analysis of random noise. Bell Labs Technical Journal, 23,
282-332.
RICHARDS, L. A. 1931. Capillary conduction of liquids through porous mediums. Physics,
1, 318-333.
RUBINSTEIN, R. Y. & KROESE, D. P. 2011. Simulation and the Monte Carlo method,
Chichester, John Wiley & Sons.
SCHILLER, L., HAGEN, G., POISEUILLE, J. L. M. & HAGENBACH-BISCHOFF, E.
1933. Drei Klassiker der Strömungslehre: Hagen, Poiseuille, Hagenbach,
Akademische Verlagsgesellschaft mbh.
SHEN, S.-L., WU, H.-N., CUI, Y.-J. & YIN, Z.-Y. 2014. Long-term settlement behaviour of
metro tunnels in the soft deposits of Shanghai. Tunnelling and Underground Space
Technology, 40, 309-323.
SHIN, H.-S., YOUN, D.-J., CHAE, S.-E. & SHIN, J.-H. 2009. Effective control of pore
water pressures on tunnel linings using pin-hole drain method. Tunnelling and
Underground Space Technology, 24, 555-561.
SHIN, J.-H., KIM, S.-H. & SHIN, Y.-S. 2012. Long-term mechanical and hydraulic
interaction and leakage evaluation of segmented tunnels. Soils and Foundations, 52,
38-48.
SNOW, D. T. 1965. A parallel plate model of fractured permeable media. University of
California, Berkeley.
SONG, H.-W., PACK, S.-W. & ANN, K. Y. 2009. Probabilistic assessment to predict the
time to corrosion of steel in reinforced concrete tunnel box exposed to sea water.
Construction and Building Materials, 23, 3270-3278.
STOKES, S. G. G. 1846. On the friction of fluids in motion and the equilibrium and motion
of elastic solids, JW Parker, Pitt Press.
253
STRACK, O. D. 1989. Groundwater mechanics, Harlow, Prentice Hall.
TAN, Y., SMITH, J. V., LI, C. Q. & DAUTH, J. Water infiltration estimation in tunnels: with
reference to the Melbourne underground. Transforming the Future of Infrastructure
through Smarter Information: Proceedings of the International Conference on Smart
Infrastructure and Construction, 2016 Cambridge, UK. 597-602.
TEST, M. 1956. Pressure tunnel with drained lining. Gidrotekh. Stroitel.
VALENTA, O. 1969. Kinetics of water penetration into concrete as an important factor of its
deterioration and of reinforcement corrosion. RILEM International Symposium:
Durability of Concrete-1969, 1, A177-A189.
VANARELLI, M. J. Statistical Determinations of Steady-State, Groundwater Inflow in Rock
Tunnels for the Radial Flow Condition. World Environmental and Water Resources
Congress 2008, 2008. 1-10.
VANORIO, G. & MERA, J. 2012. Methodology for risk analysis in railway tunnels using
Monte Carlo simulation. WIT Transactions on The Built Environment, 127, 673-683.
VERRUIJT, A. & BOOKER, J. R. Complex variable analysis of Mindlin’s tunnel problem.
Proceedings of Development of Theoretical Geomechanics, 2000 Balkema, Sydney.
3-22.
WANG, K., JANSEN, D. C. & SHAH, S. P. 1997. Permeability study of cracked concrete.
Cement and Concrete Research, 27, 381-393.
WEBAC. 2017a. Crack Repair [Online]. Available: https://www.webac-
grouts.com/uploads/tx_produkte/BRO__Risssanierung__en.pdf 04/07/2017].
WEBAC. 2017b. Epoxy Injection Resins [Online]. Available: https://www.webac-
grouts.com/solutions/classic-product-line/epoxy-injection-resins/ [Accessed
04/07/2017 2017].
WEBAC. 2017c. PU Injection Resins [Online]. Available: https://www.webac-
grouts.com/solutions/classic-product-line/pu-injection-resins/ 04/07/2017].
WEBAC. 2017d. WEBAC® SILform [Online]. Available: https://www.webac-
grouts.com/solutions/consolidation-line/silicate-resins/webac-silform/ 04/07/2017].
WITHERSPOON, P., WANG, J., IWAI, K. & GALE, J. 1980. Validity of cubic law for fluid
flow in a deformable rock fracture. Water Resources Research, 16, 1016-1024.
YU, J., ZHONG, D., REN, B., TONG, D. & HONG, K. 2017. Probabilistic Risk Analysis of
Diversion Tunnel Construction Simulation. Computer‐Aided Civil and Infrastructure
YUAN, Y., BAI, Y. & LIU, J. 2012. Assessment service state of tunnel structure. Tunnelling
and Underground Space Technology, 27, 72-85.
YUE, Q. & ANG, A. H. Dynamic response and reliability of tunnel under earthquakes.
International Conference on Applications of Statistics and Probability, 2015
Vancouver 1-8.
ZANZOTTO, L. 1996. Laboratory testing of crack sealing materials for flexible pavements,
Ottawa, ON, Transportation Association of Canada, 171 pp.
ZHANG, L. 2013. Aspects of rock permeability. Frontiers of Structural and Civil
ZHANG, L. & FRANKLIN, J. 1993. Prediction of water flow into rock tunnels: an analytical
solution assuming an hydraulic conductivity gradient. International Journal of Rock
Mechanics and Mining Sciences & Geomechanics, 30, 37-46.
ZHENG, J.-J. & ZHOU, X.-Z. 2008. Analytical method for prediction of water permeability
of cement paste. ACI materials journal, 105, 200-206.
ZHENG, J., LI, C. & ZHOU, X. 2005. Thickness of interfacial transition zone and cement
content profiles around aggregates. Magazine of Concrete Research, 57, 397-406.
254
255
Appendix A
X-Ray Diffraction test result
Sample MU 1
Sample MU 2
Sample MU 3
256
Sample MU 5
Sample MU 6
Sample MU 7
257
Sample MU 8
Sample MU 9
258
Sample MU 10
Sample MU 11
Sample MU 12
259
Sample MU 13
Sample MU 14
260
Sample MU 15
Sample MU 16
Sample MU 17
261
Sample MU 18
Sample MU 19
262
Sample MU 20
Sample MU 24
Sample MU 25
263
Sample MU 26
264
Scanning Electron Microscopy (SEM) and Energy-dispersive X-ray spectroscopy (EDX)
test result
Sample MU 1
SEM
EDS
265
Sample MU 2
SEM
EDS
266
Sample MU 3
SEM
267
EDS
Sample MU 5
SEM
268
EDS
Sample MU 6
SEM
269
EDS
Sample MU 7
270
SEM
EDS
271
Sample MU 8
SEM
EDS
272
Sample MU 9
SEM
EDS
273
Sample MU 10
SEM
EDS
274
Sample MU 11
SEM
275
EDS
Sample MU 12
SEM
276
EDS
Sample MU 13
SEM
277
EDS
Sample MU 14
278
SEM
EDS
Sample MU 15
279
SEM
EDS
280
Sample MU 16
SEM
EDS
281
Sample MU 17
SEM
EDS
282
Sample MU 18
SEM
EDS
283
Sample MU 19
SEM
284
EDS
Sample MU 20
SEM
285
EDS
286
View publication stats

Service Life Prediction of Metro Tunnel Lining Based On Water Ingress

Uploaded by

Copyright:

Available Formats

Service Life Prediction of Metro Tunnel Lining Based On Water Ingress

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Service Life Prediction of Metro Tunnel Lining Based On Water Ingress

Uploaded by

Copyright:

Available Formats

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

Service Life Prediction of Metro Tunnel Lining Based on Water Ingress

Thesis · December 2017

Service life prediction of a metro tunnel View project

The user has requested enhancement of the downloaded file.

A thesis submitted in fulfilment of the requirements for the degree of

procedures and guidelines have been followed.

19th December 2017

ingress is considered to be the major failure criterion.

lining was established analytically.

rock mass properties are known.

of water ingress significantly when the inhomogeneity degree is high.

without their support.

privilege to be their student.

company through every moment of my life.

and Underground Space Technology, 71, 115-125.

Bergen, Norway. Norwegian Tunnelling Society, 1-5.

tunnels: with reference to the Melbourne underground. Transforming the Future of

Infrastructure through Smarter Information: Proceedings of the International

Conference on Smart Infrastructure and Construction. Cambridge, UK, 597-602.

of fines on concrete tunnel lining. In: TATIANA ROTONDA, M. C., FRANCESCO

CRC Press/Balkema, 191-193.

List of Publication ...................................................................................................................... v

List of Figures ........................................................................................................................... xi

List of Tables .........................................................................................................................xvii

1.1 Background ................................................................................................................. 1

1.2 Significance ................................................................................................................. 3

1.3 Aim and objectives ...................................................................................................... 5

1.4 Scope of research ........................................................................................................ 6

1.5 Outline of the thesis..................................................................................................... 7

2 Literature Review ............................................................................................................. 10

2.1 Introduction ............................................................................................................... 10

2.2 Construction of tunnels ............................................................................................. 11

2.2.1 Design of tunnels ............................................................................................... 11

2.2.2 Construction methods of tunnels ....................................................................... 14

2.2.3 Tunnel behaviour ............................................................................................... 17

2.2.4 Tunnel maintenance methods ............................................................................ 23

2.3 Concrete properties ................................................................................................... 27

2.3.1 Concrete composition ........................................................................................ 28

2.3.3 Permeability of concrete .................................................................................... 33

2.3.4 Factors affecting water ingress in concrete ........................................................ 37

2.4 Engineering hydraulics .............................................................................................. 38

2.4.1 Basics of hydraulics ........................................................................................... 38

2.4.2 Mechanisms of water movement in concrete..................................................... 41

2.4.3 Hydraulic conductivity of cracked material ....................................................... 43

2.4.4 Homogeneity of cracked material ...................................................................... 49

2.4.5 Water inflow into tunnel .................................................................................... 50

2.5 Reliability theory ....................................................................................................... 63

2.5.1 Basics of reliability theory ................................................................................. 63

2.5.2 Outcrossing method ........................................................................................... 68

2.5.3 Monte Carlo simulation ..................................................................................... 72

2.5.4 System reliability ............................................................................................... 75

2.6 Summary ................................................................................................................... 79

3 Water Ingress Into Concrete Lining ................................................................................. 81

3.1 Introduction ............................................................................................................... 81

3.2 Hydraulic conductivity of different categories of concrete ....................................... 84

3.2.1 Different categories of concrete ......................................................................... 85

3.2.2 Development of hydraulic conductivity............................................................. 87

3.2.3 Crack connectivity coefficient ........................................................................... 92