3-Sivalingam Koneshwaran Thesis

BLAST RESPONSE AND SENSITIVITY
ANALYSIS OF SEGMENTAL TUNNEL
By
Sivalingam Koneshwaran
BSc Eng (Hons,1)
A THESIS SUBMITTED TO
FACULTY OF SCIENCE AND ENGINEERING FACULTY
QUEENSLAND UNIVERSITY OF TECHNOLOGY
IN PARTIAL
FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
November 2014
Dedicated to my parents, family and friends
Page i
Page ii
Acknowledgements
I wish to express my deepest gratitude to my principal supervisor, Professor David

Thambiratnam, for his excellent guidance, caring, patience, and rightful leniency
throughout my doctoral studies. I also want to thank my associate supervisor Dr.
Chaminda Gallage for taking the time to review my work and for providing me with
valuable technical observations.
My sincere thanks also goes to Mr. Ashley Wright and other staff members at the
High Performance Computer and Research Support Group for providing technical
assistance with numerical simulations and data manipulation. I wish to extend my
thanks to Dr. Luke Mosse, of Livermore Software Technology Corporation who
provided immeasurable assistance on the working of the Finite element software LS-
DYNA. I also wish to thanks to Associate Professor Anirban De, Manhattan College,
USA and Dr. Arulmoli, Principal of Earth Mechanics, Inc., Canada for providing
extensive experimental data set for validating the modelling techniques.
I gratefully acknowledge the financial support granted by Australian Postgraduate

Award and Science and Engineering Faculty, Queensland University of Technology
(QUT) to conduct the research. I am very grateful to my colleagues at QUT for
sharing their knowledge, valuable guidance and encouragement in carrying out the
research.
Finally, I am extremely grateful to my parents and wife who have been a source of
encouragement and inspiration throughout my career.
Sivalingam Koneshwaran
Science and Engineering faculty,
Queensland University of Technology,
Brisbane, Australia.
November 2014
Page iii
Page iv
TABLE OF CONTENTS
ACKNOWLEDGEMENTS ....................................................................................III
TABLE OF CONTENTS ......................................................................................... V
LIST OF FIGURES ................................................................................................. IX
LIST OF TABLES ................................................................................................ XIII
ABSTRACT .............................................................................................................XV
KEY WORDS ...................................................................................................... XVII
PUBLICATIONS .................................................................................................. XIX
STATEMENT OF ORIGINAL AUTHORSHIP ............................................... XXI
1 CHAPTER 1: INTRODUCTION ......................................................................... 1
1.1 Background ......................................................................................................... 1
1.2 Research problem ............................................................................................... 5
1.3 Aims and objectives ........................................................................................... 7
1.4 Significance of the research ................................................................................ 8
1.5 Thesis outline...................................................................................................... 8
2 CHAPTER 2: LITERATURE REVIEW........................................................... 11
2.1 Introduction ...................................................................................................... 11
2.2 Explosion and related phenomenon .................................................................. 11
2.2.1 Explosives and TNT equivalent .............................................................. 11
2.2.2 Blast scaling laws .................................................................................... 12
2.3 Shock wave propagation ................................................................................... 13
2.4 Crater formation ............................................................................................... 14
2.5 Soil-Structure Interaction (SSI) ........................................................................ 15
2.6 Blast response of buried tunnels ....................................................................... 16
2.6.1 Experimental approach............................................................................ 16
2.6.2 Numerical approach ................................................................................ 17
2.7 Research gap ..................................................................................................... 19
3 CHAPTER 3: NUMERICAL MODELLING ................................................... 21
3.1 Introduction ...................................................................................................... 21
3.2 Arbitrary Lagrangian Eulerian.......................................................................... 22
3.3 Smooth Particle Hydrodynamics ...................................................................... 23
3.4 Simulation of blast loading ............................................................................... 25
3.4.1 Element selection .................................................................................... 25
3.4.2 Boundary condition ................................................................................. 25
3.4.3 Contact interface ..................................................................................... 26
3.4.4 Reinforced concrete model ..................................................................... 26
Page v
3.4.5 Stress initialization .................................................................................. 26
3.4.6 Data manipulation ................................................................................... 27
3.5 Material models ................................................................................................ 28
3.5.1 Air model ................................................................................................ 28
3.5.2 Explosive model ...................................................................................... 29
3.5.3 Soil model ............................................................................................... 30
3.5.4 Concrete model ....................................................................................... 33
3.5.5 Reinforcement model .............................................................................. 35
3.6 Chapter summary.............................................................................................. 36
4 CHAPTER 4: VALIDATION OF NUMERICAL MODELS .......................... 37
4.1 Introduction ...................................................................................................... 37
4.2 Validation of ALE simulations using centrifuge test results ............................ 38
4.2.1 Description of centrifuge test .................................................................. 38
4.2.2 Development of numerical model ........................................................... 39
4.2.3 Material models and parameters ............................................................. 41
4.2.4 Description of three stage simulation...................................................... 42
4.2.5 Shockwave propagation .......................................................................... 44
4.2.6 Tunnel response ...................................................................................... 48
4.2.7 Summary ................................................................................................. 55
4.3 Validation of SPH simulations using centrifuge test results ............................ 55
4.3.2 Shockwave propagation .......................................................................... 57
4.3.3 Tunnel response ...................................................................................... 58
4.3.4 Comparison of strain output .................................................................... 59
4.3.5 Comparison of computational efficiency ................................................ 62
4.3.6 Summary ................................................................................................. 62
4.4 Validation of ALE simulations using conweb test results ................................ 63
4.4.1 Description of CONWEB test ................................................................. 63
4.4.3 Material models and parameters ............................................................. 67
4.4.4 Free-field simulation ............................................................................... 68
4.4.5 Test slab response ................................................................................... 69
4.4.6 Summary ................................................................................................. 74
4.5 Validation of SPH simulations using conweb test results ................................ 74
4.5.2 Free-field simulation ............................................................................... 75
4.5.3 Test slab response ................................................................................... 76
4.5.4 Summary ................................................................................................. 80
4.6 Chapter summary.............................................................................................. 80
5 CHAPTER 5: EFFECT OF SEGMENTS ON THE BLAST RESPONSE OF
BURIED TUNNELS ................................................................................................ 83
5.1 Introduction ...................................................................................................... 83
5.2 Blast response of segmented bored tunnels with rectagular segments ............. 83
5.2.1 Description of segmented tunnel............................................................. 83
5.2.2 Description of numerical model .............................................................. 85
Page vi
5.2.3 RS Tunnel response ................................................................................ 86
5.3 Blast response of segmented bored tunnels with different types of segments . 91
5.3.1 HS Tunnel response ................................................................................ 91
5.3.2 IS Tunnel response .................................................................................. 93
5.4 Numerical results and discussion ..................................................................... 95
5.5 Chapter summary.............................................................................................. 96
6 CHAPTER 6: CRUCIAL PARAMETERS AFFECTING THE BLAST
RESPONSE OF BURIED TUNNELS ................................................................... 99
6.1 Introduction ...................................................................................................... 99
6.2 Blast performance of shallow bored tunnels in different soils ....................... 101
6.2.1 Peak pressures in various soil types ...................................................... 101
6.2.2 Description of segmented tunnel........................................................... 102
6.2.3 Description of numerical model ............................................................ 103
6.2.4 Blast response of segmented tunnel ...................................................... 104
6.3 Effect of radial joint types .............................................................................. 108
6.4 Effect of number of segments......................................................................... 114
6.5 Discussion of numerical results ...................................................................... 121
6.6 Chapter summary............................................................................................ 122
7 CHAPTER 7: BLAST IMPACT AND MITIGATION MEASURES ........... 125
7.1 Introduction .................................................................................................... 125
7.2 The effect of tunnel depth and stand-off distance in blast impact .................. 126
7.3 Mitigation of blast Impact using different approaches ................................... 133
7.3.1 Wrapping the tunnel with flexible and compressible barrier
consisting of a layer of polyurethane foam. .......................................... 133
7.3.2 Wrapping the tunnel with a secondary in-situ reinforced concrete
lining ..................................................................................................... 136
7.3.3 Replacing Reinforced Concrete (RC) segments by Steel Fibre
Reinforced Concrete (SFRC) segments. ............................................... 138
7.3.4 Introducing energy absorbing flexible honeycomb elements between
radial joints. ........................................................................................... 140
7.3.5 Introducing a reinforced concrete buster slab above the tunnel crown. 143
7.4 Discussion of Numerical results ..................................................................... 147
7.4.1 Wrapping the tunnel with a flexible and compressible barrier
consisting of a layer of polyurethane foam. .......................................... 148
7.4.2 Wrapping the tunnel with a secondary in-situ reinforced concrete
lining ..................................................................................................... 148
7.4.3 Replacing Reinforced Concrete (RC) segments by Steel Fibre
Reinforced Concrete (SFRC) segments. ............................................... 149
7.4.4 Introducing energy absorbing, flexible honeycomb elements between
radial joints. ........................................................................................... 149
7.4.5 Introducing a reinforced concrete buster slab above the tunnel crown. 150
7.5 Chapter summary............................................................................................ 150
8 CHAPTER 8: CONCLUSION AND RECOMMENDATIONS .................... 153
Page vii
8.1 Review of research work ................................................................................ 153
8.2 Main findings of this research and recomendations ....................................... 154
8.3 Future research ............................................................................................... 157
REFERENCES ....................................................................................................... 159
Page viii
LIST OF FIGURES
Figure 1.1: Comparison of types of terrorist attacks on tunnel (GTD, 2013) ........................... 1
Figure 1.2: Schematic view of London underground transit tunnels (Skyscrapercity.com,
2013) .................................................................................................................... 2
Figure 1.3: Schematic view of Precast Segmented Rings ......................................................... 4
Figure 1.4: Interaction of shockwave with the tunnel ............................................................... 6
Figure 3.1: Lagrangian-Eulerian coupling .............................................................................. 23
Figure 3.2: Coupled SPH-FE Method..................................................................................... 24
Figure 4.1: Flowchart used for validation of numerical techniques and material models ...... 37
Figure 4.2: Setup of experimental model (All dimensions are in prototype scale) (De &
Zimmie, 2011) .................................................................................................... 38
Figure 4.3: A quarter symmetrical prototype model (Lagrangian structure) .......................... 40
Figure 4.4: Kinetic energy vs. time during the initialization .................................................. 43
Figure 4.5: Displacement vs. time for three gauges ................................................................ 43
Figure 4.6: Insertion of ALE background mesh into Lagrangian structure ............................ 45
Figure 4.7: Kinetic energy vs. time plot of ALE background mesh ....................................... 45
Figure 4.8: Propagation of shock wave in dry sand ................................................................ 46
Figure 4.9: Free field numerical model for wave propagation study ...................................... 47
Figure 4.10: Comparison of peak pressure distributions along Grid 1 and Grid 2 ................. 47
Figure 4.11: Kinetic energy vs. time plot for tunnel ............................................................... 48
Figure 4.12: Pressure wave propagation through the tunnel ................................................... 50
Figure 4.13: Arrangement of measuring gauges on the tunnel exterior surface ..................... 51
Figure 4.14: Axial and circumferential strain histories........................................................... 51
Figure 4.15: Deformation of the tunnel during the peak response.......................................... 53
Figure 4.16: Comparison of numerical and experimental axial strains .................................. 54
Figure 4.17: Comparison of peak axial and circumferential strains ....................................... 54
Figure 4.18: Coupled SPH-FE method. .................................................................................. 56
Figure 4.19: Explosive-soil interaction ................................................................................... 57
Figure 4.20: Shock wave propagation in soil.......................................................................... 57
Figure 4.21: Interface pressure contours................................................................................. 59
Figure 4.22: Interface pressure vs. time .................................................................................. 59
Figure 4.23: Comparison of axial and circumferential strains. ............................................... 60
Figure 4.24: Comparison of axial strain from Gauge 2 between simulation and experiment. 61
Figure 4.25: Comparison of peak axial and circumferential strains. ...................................... 61
Figure 4.26: Experimental setup of CONWEB test (elevation) (Hayes, 1989) ...................... 64
Figure 4.27: A half symmetry numerical model for ALE simulation ..................................... 65
Page ix
Figure 4.28: Test structure components .................................................................................. 66
Figure 4.29: Comparison of peak pressures ............................................................................ 69
Figure 4.30: Progress of test slab deformation ....................................................................... 70
Figure 4.31: Broken slab reinforcement ................................................................................. 70
Figure 4.32: Propagation of cracks on test slab surfaces ........................................................ 71
Figure 4.33: Instrumentation plan of test slab (Hayes, 1989). ................................................ 72
Figure 4.34: Comparison of displacements ............................................................................ 74
Figure 4.35: A half symmetrical numerical model for SPH simulation ................................. 75
Figure 4.36: Comparison of peak pressures ............................................................................ 76
Figure 4.37: Interface pressure contours ................................................................................. 77
Figure 4.38: Comparison of peak interface pressures ............................................................. 77
Figure 4.39: Propagation of cracks on interior surface ........................................................... 78
Figure 4.40: Comparison of displacements ............................................................................ 80
Figure 5.1: Rectangular Segments (RS).................................................................................. 84
Figure 5.2: Configuration of the coupled numerical model. ................................................... 85
Figure 5.3: Gravity initialization............................................................................................. 86
Figure 5.4: Tunnel response in longitudinal direction ............................................................ 87
Figure 5.5: Tunnel crown displacement.................................................................................. 88
Figure 5.6: Drifting response ................................................................................................... 89
Figure 5.7: Crack patterns obtained from numerical simulations ........................................... 90
Figure 5.8: Different types of segments. ................................................................................. 91
Figure 5.9: Immediate HS ring response. ............................................................................... 92
Figure 5.10: Hexagonal segmented tunnel response. .............................................................. 93
Figure 5.11: Immediate IS ring response. ............................................................................... 94
Figure 5.12: Interlocking segmented tunnel crown response ................................................. 95
Figure 5.13: Interface of radial joints ..................................................................................... 96
Figure 6.1: Soil skeleton ....................................................................................................... 100
Figure 6.2: Comparison of free-field peak pressure ............................................................. 102
Figure 6.3: General arrangement of segments ...................................................................... 103
Figure 6.4: A half symmetrical numerical model ................................................................. 104
Figure 6.5: Diametric distortion in dry soil .......................................................................... 105
Figure 6.6: Diametric distortion in partially saturated soil ................................................... 106
Figure 6.7: Diametric distortion in saturated soil ................................................................. 106
Figure 6.8: Radial bolt failure adjacent to key segment ....................................................... 108
Figure 6.9: Initial crack formation after impact of shockwave in load case 1N ................... 111
Figure 6.10: Radial drift response of an immediate ring ...................................................... 112
Figure 6.11: Isometric view of segmental lining .................................................................. 115
Page x
Figure 6.12: Radial drift response of an immediate ring ...................................................... 116
Figure 6.13: View of cracks (>0.3mm) in Type 1 tunnel in load case 4N ............................ 117
Figure 6.16: View of bolt failure in Type 3 tunnel (Load case 4N) ..................................... 120
Figure 7.1: Variation of tunnel depth and ground distance................................................... 126
Figure 7.2: Failure modes of the tunnel under different tunnel depths (load case 3) ........... 128
Figure 7.3: Failure modes of the tunnel under different ground distances (load case 3) ...... 129
Figure 7.4: Critical explosive weight vs. stand-off distance for different damage state ....... 132
Figure 7.5: Comparison of crack response of tunnels (Load case1 and depth=1D) ............. 134
Figure 7.6: Time history of total kinetic and strain energies ................................................ 135
Figure 7.7: Critical explosive weight vs. tunnel depth for polyurethane foam ..................... 136
Figure 7.8: Comparison of shockwave interaction with the barrier materials ...................... 137
Figure 7.9: Critical explosive weight vs. tunnel depth for Reinforced Concrete secondary
lining ................................................................................................................ 138
Figure 7.10: The crack response of SFRC tunnel (Load case1 and depth=1D) .................... 139
Figure 7.11: Critical explosive weight vs. tunnel depth for SFRC lining ............................. 140
Figure 7.12: Arrangement of flexible parts in a ring ............................................................ 141
Figure 7.13: Critical explosive weight vs. tunnel depth for flexible honeycomb elements .. 142
Figure 7.14: Tunnel deformation after 300 ms of the explosion for different load cases ..... 143
Figure 7.15: Numerical models ............................................................................................ 144
Figure 7.16: Shockwave propagation through the buster slab-Depth= 1D & Load case 1 ... 145
Figure 7.17: Critical explosive weight vs. tunnel depth for buster slab introduction ........... 146
Page xi
Page xii
LIST OF TABLES
Table 1.1: Weight of explosive (TNT equivalent) filling a vehicle (Conterterrorism 2014
Calendar, 2014)....................................................................................................... 3
Table 2.1: TNT equivalent of common explosives materials (Jayasooriya et al., 2011) ........ 12
Table 3.1: Material properties for air (Yang et al. 2010) ........................................................ 29
Table 3.2: Material parameters for TNT explosive................................................................. 29
Table 4.1: Scaling law (Kramer, 1996)..................................................................................... 39
Table 4.2: Conversion to prototype model (De, 2012) .............................................................. 39
Table 4.3: Material parameters for copper. ............................................................................... 41
Table 4.4: Material parameters for soil ..................................................................................... 42
Table 4.5: Comparison of computational efficiency............................................................... 62
Table 4.6: Material properties of backfill materials ................................................................ 67
Table 4.7: Material properties of concrete .............................................................................. 68
Table 4.8: Material properties of steel reinforcement ............................................................. 68
Table 5.1: Load cases.............................................................................................................. 86
Table 5.2: Crack response under different load cases ............................................................. 89
Table 6.1: Soil properties for numerical simulation (Jayasinghe et al., 2014)...................... 102
Table 6.2: Numbers of radial bolt failure.............................................................................. 107
Table 6.3: Radial joint types and corresponding force transmission capacity ...................... 109
Table 6.4: New Load cases ................................................................................................... 109
Table 6.6: Maximum crack width in mm.............................................................................. 113
Table 6.7: Maximum crack width in mm.............................................................................. 119
Table 6.9: Numbers of circumferential bolt failure .............................................................. 121
Table 7.1: Maximum crack width in mm / nos. of bolt failures ........................................... 129
Table 7.2: Maximum crack width in mm / nos. of bolt failures ........................................... 130
Table 7.3: Material properties of SFRC................................................................................ 138
Page xiii
Page xiv
Abstract
Increasing threat of terrorism highlights the importance of enhancing the resilience of

infrastructure to all types of hazards. A significant amount of studies have been
carried out to scientifically develop appropriate protective measures for above
ground structures such as buildings and bridges to mitigate the adverse effects of
terrorist attacks. However, only a few studies have been carried out on underground
transportation tunnels which are equally or even more important in the infrastructure.
Bored tunnels driven by tunnel boring machines (TBM) are now common in many
cities providing a quick and cost effective alternative to surface rails and roads.
Although a few studies have been carried out on underground tunnels subjected to
surface blasts, they do not cover bored tunnels which consist of prefabricated
reinforced concrete segments placed together by bolted joints.
In this research field, full-scale tests are often impractical as it is extremely risky and
expensive. Although few studies have been conducted using scale-down centrifuge
tests to investigate the blast response of pipe structures, limitation in the centrifuge
test makes it virtually impossible for investigating the blast response of bored tunnel
with multiple bolted joints which are essential to resist the blast load by allowing the
joints to rotate, slide and dissipate energy to achieve equilibrium before the segments
are damaged. The possible alternative, therefore, is to use numerical techniques
which can provide valuable data in a timely and cost effective manner to enable the
development of mitigation measures.
This research develops and applies a fully coupled technique to study the response of
bored tunnels to credible surface blasts using explicit dynamic nonlinear finite
element software LS-DYNA. Appropriate material models were utilized with strain
rate effects in the analyses. Damage states of tunnels were identified based on the
displacements, drifting responses between segments and cracks in the segments. The
influence of soil type, joint type and number of segments forming the tunnel ring has
been investigated. Results show that the tunnel buried in saturated soil displayed
severe damage compared to that buried in partially saturated or dry soils. For weaker
blast loads, joint types influenced the tunnel response by allowing the segments to
Page xv
rotate about the joints, whereas for higher blast loads, the segments were damaged
before activating the flexibility of the joints. Although an increase in the number of
segments increases the number of joints, beyond a certain limit, increase in the
number of segments did not improve the blast performance of the tunnel. The tunnel
is more vulnerable to surface explosions which occur directly above the centre of the
tunnel than those that occur at any equivalent scale distances by moving the
explosive away from the tunnel centre.
The finding from this research offer significant new information on the blast
performance of segmented bored tunnels and will guide their future implementation
in civil engineering applications.
Page xvi
Key words
Blast analysis, Surface explosion, Segments, Bored tunnel, Joints, Soil, Finite
element method, Fully coupled analysis, Numerical simulation, Material models,
cracks, Soil-Structure Interaction.
Page xvii
Page xviii
Publications
International Journal Papers
 Sivalingam, K., Thambiratnam, D. P., and Gallage, C (2014), “Numerical

Performance of buried tunnels subjected to surface blast incorporating fluid
structure interaction”. ASCE Journal of Performance of Constructed
Facilities. Published
 Sivalingam, K., Thambiratnam, D. P., and Gallage, C (2014), “Response of
Segmented Bored Transit Tunnels to Surface Blast”. Engineering Structures,
Elsevier. Submitted in July 2014
 Sivalingam, K., Thambiratnam, D. P., and Gallage, C (2014), “Blast Response and
Failure Analysis of Segmented Bored Tunnel using Coupled SPH-FE
Method”. Computers and Structures, Elsevier. Under review
 Sivalingam, K., Thambiratnam, D. P., and Gallage, C (2014), “Effects of
Segments on the Blast Response of Buried Tunnels”. ASCE Journal of
Structural Engineering. Under review
International Conference Papers
 Sivalingam, K., Thambiratnam, D. P., and Gallage, C (2013), “Response of a

Buried Tunnel to Surface Blast using Different Numerical Techniques”.
Proceedings of the Fourteenth International Conference on Civil, Structural
and Environmental Engineering Computing: Civil-Comp Press, Stirlingshire,
UK. Published
Page xix
Page xx
Statement of original authorship
The work contained in this thesis has not been previously submitted to meet
requirements for an award at this or any other higher education institution. To the
best of my knowledge and belief, the thesis contains no material previously
published or written by another person except where due reference is made.
QUT Verified Signature
Koneshwaran Sivalingam
November 2014
Page xxi
Page xxii
Chapter 1: Introduction
1Chapter 1: Introduction
1.1 BACKGROUND
Underground transit tunnel system provides a quick and cost effective alternative to
laying surface rails and roads. This system plays an important role in addressing
transportation needs in many cities. Terrorist attacks, such as the Belarus bombing in
2011, the Moscow metro bombing in 2004 and the London subway bombing in 2005,
highlight that underground transit tunnels are vulnerable to potential terrorist attacks.
According to University of Maryland Global Terrorism Database in Figure 1.1 the
terrorist attacks on tunnels continue to increase and explosive is the main source for
those attacks over the past four decades. The Department of Homeland Security of the
United States also issued a statement that terrorists may attack New York subway
system using explosive.
6
Explosive
5
Firearms
4 Incendiary
0
1972 1977 1982 1987 1992 1997 2002 2007 2012
Figure 1.1: Comparison of types of terrorist attacks on tunnel (GTD, 2013)
Terrorist attacks on transportation tunnels must therefore be closely considered for two
reasons; i) the importance and centrality of such infrastructure, and ii) the level of
public use. The failure of such underground tunnels would not only cause delays and
transport network interruptions, but also result in severe loss of lives with considerable
financial implications.
Page 1
The central part of a rapid transit network in cities is usually within tunnels. These
tunnels are mostly bored tunnels constructed using tunnel boring machines (TBM) with
the support of permanent linings. The principle of bored tunnel construction has been
known for a long time. Tunnels constructed in the beginning of the nineteenth century
are still in use in many cities. They have a direct relationship with the identity of the city
as they illustrate its history, culture, and its economic, political and social states. Some
tunnels are designed to resist natural disasters such as earthquakes and extreme events
such as fires, but most are not designed to resist blast loading. This demands great
attention for protecting such structures from terrorist attacks.
Figure 1.2: Schematic view of London underground transit tunnels (Skyscrapercity.com,

2013)
Tunnel structures may experience either internal or external blasts. Internal blasts are
less likely to occur as it is difficult to place an explosive device within a tunnel as well
as having a moving vehicle filled with explosives is not feasible in subway systems.
Advanced security and surveillance systems installed in the subway can also easily
detect such interior explosive devices. However, external blasts are unfortunately more
likely and have the potential to cause most damage. For example, Figure 1.2 shows a
schematic view of a part of London underground transit network. The congestion shown
in this picture is indicative of the situation in many cities worldwide. The picture
illustrates that the tunnels are shallow and such shallow tunnels are vulnerable to
external explosion from a moving vehicle having potential sources of ignition.
Page 2
Furthermore, the crowded nature of underground developments increase risk to both

underground and above ground structures in the vicinity.
A Vehicle Borne Improvised Explosive Device (VBIED) is commonly used as a

weapon of terrorism to damage constructed facilities. As shown in Table 1.1, VBIED
can be carried in a wide range of vehicles such as cars, vans, containers and trailers to
transport a large amount of explosive without attracting suspicion. As the VBIED is
relatively close to the ground surface, the explosion from the VBIED generates air-blast
pressure as well as ground shock.
Table 1.1: Weight of explosive (TNT equivalent) filling a vehicle (Conterterrorism 2014
Calendar, 2014)
Vehicle Explosives capacity

Type of vehicles *
description TNT Equivalent (kg)
Sedan 227
SUV/van 454
Small delivery
1814
truck
Container/water
4536
truck
Semi-trailer 27216
* Section 2.2.2 describes TNT (Trinitrotoluene).
Ground shocks travel through both soil and rock. In soil, the tunnel structure resists
most of the ground shock, whereas in rock, the medium itself carries a large amount of
the ground shock. As the shock wave intercepts the geostatic soil structure interface
between the tunnel structure and surrounding soil, the tunnel experiences an effect of
inertia lasting for a few milliseconds. The surrounding medium significantly constrains
the tunnel motion. However, the tunnel response to the shock wave is similar to
Page 3
instantaneous vibration as damping from the surrounding medium is ineffective. The

shock wave reaches the tunnel in a very short time before the damping activates itself
(Shin et al., 2011). The tunnel can therefore exhibit excessive level of stress affecting
the structural integrity. This leads to a localized failure and potential collapse of the
tunnel with serious consequences.
Bored tunnels with segmented tunnel lining are more popular in underground railway
systems. The tunnel lining is installed directly behind the TBM. The space between the
bored soil/rock face and the lining is filled with annular concrete grout. The lining
consists of prefabricated reinforced concrete segments placed together to complete one
lining ring to span the circumference of the tunnel as shown in Figure 1.3. The segments
are connected to adjacent segments at contact joints by bolts which create an
interlocking system providing the tunnel with resistance against external pressure from
the surrounding ground. The stiffness of the joints determines the load transmission
throughout the ring. The joints in both radial and circumferential directions consist of
grooves to accommodate the elastomeric gasket that provides water-resistance. The
bolts help stabilise the ring during the grouting as well as ensure the water-resistance by
compressing the gasket in the years of service after construction.
Figure 1.3: Schematic view of Precast Segmented Rings
In order to ensure the safety and stability of bored tunnels, it is necessary to consider the
variability in the prediction of load carrying capacity during their service life. The four
key factors which affect the tunnel response under blast loading are: (i) surrounding
ground type (ii) explosive mass, (iii) standoff distance and (iv) lining stiffness. Though
Page 4
the lining stiffness is the engineer’s choice, it may be difficult to evaluate how the
tunnel response is influenced by its stiffness. The response of the segmented tunnel
lining to the blast loading is more complex than other loadings such as geo-static and
earthquake loadings. Performance of the segmented tunnels under static and earthquake
loads has been the subject of several studies. However, there is inadequate information
on response of bored tunnels to blast loading. Munfah (2009) described that thin precast
segmented tunnel linings are more vulnerable to blast loads than thick cast-in-place
concrete tunnels. It is therefore of interest to investigate the vulnerability of segmented
tunnels to credible blast loading.
In this research field, there are no records on full-scale prototype field experiments
investigating the tunnel response to surface blast. Full-scale experiments are extremely
risky and unattainable in civilian research as they involve the use of a large quantity of
explosives. On the other hand, significant studies have been conducted using scaled-
down centrifuge modelling techniques to treat the blast response of buried cylindrical
structures. However, limitations in the centrifuge test makes it virtually impossible for
investigating the blast response of bored tunnel with multiple bolted joints which are
essential to resist the blast load by allowing the joints to rotate, slide and dissipate
energy to achieve equilibrium before the segments are damaged. The possible
alternative, therefore, is to use numerical techniques which can provide valuable data in
a timely and cost effective manner to enable the development of mitigation measures.
1.2 RESEARCH PROBLEM
The primary load resisted by the tunnel is through the combination of circumferential
(hoop) stress and bending stress induced by external pressure from the surrounding
ground acting on the circumference of the tunnel. The external pressure is mostly a non-
uniformly distributed pressure, and not a concentrated load. The VBIED-induced
shockwave applying pressure on the tunnel surface is however more complex. Figure
1.4 illustrates how VBIED induced shockwave propagates through the soil surrounding
a tunnel structure. When the expanding shockwave front strikes the tunnel surface as
shown in Figure 1.4(b), it exerts a dynamic concentrated load. As the shockwave
progresses, it applies a distributed load. Meanwhile, the shockwave is reflected from the
tunnel surface and the reflected wave diminishes the intensity of oncoming incident
Page 5
wave. If the tunnel is closer to the explosive, the reflected wave becomes more
significant, and peak reflected pressure appears to be higher than the oncoming incident
wave pressure. At this point, it generates a negative pressure on the tunnel surface.
(a) Propagation of shockwawe (b) Shockwave striking tunnel crown
(c) Reflection of shockwave
Figure 1.4: Interaction of shockwave with the tunnel
As the surrounding pressure changes through a concentrated load, distributed positive

and negative pressures, the tunnel lining experiences diametric distortions in all
directions. In such situations, the tunnel lining applies its flexibility to deform and
interact with the surrounding medium to resist the shockwave. The lining segments are
Page 6
generally designed to resist the axial bearing loads and buckling from the TBM thrust
loads as well as bending stresses from the surrounding soil pressure. Under geostatic
loading, the segmental linings permit a degree of rotation at the joints without
significant loss of load carrying capacity. However, the potential damage due to ground
shock resulting from the blast has not received proper attention in the current practice of
tunnel designing, particularly transit tunnels with very thin segmented linings and joints,
which are at risk of terrorist attacks. Lack of knowledge on analysing the tunnel
performance to blast load impedes the progress of the study in this area. This
emphasises the need for research to evaluate the blast performance and vulnerability of
the segmented bored tunnel.
1.3 AIMS AND OBJECTIVES
The main aim of this research is to study the response of bored tunnel (subway) to
credible surface blast as well as to study the effects of controlling parameters. This
research will examine the vulnerability of bored tunnel with precast segmented lining
under different blast scenarios. In order to achieve this, the specific objectives of the
present study are as follows:
 To develop fully coupled numerical models incorporating different material

models to study the effects of bored tunnel with segments subjected to surface
blast loading.
 To provide validation of the proposed modelling techniques using available
experimental results.
 To conduct parametric studies on the influence of soil type, segment geometry,
joint type and number of segments in a ring on the blast response.
 To provide mitigation measures to protect the tunnel from credible surface
blasts.
 To develop design information that can be used as a reference to check the blast
capacity and vulnerability of a tunnel.
Page 7
1.4 SIGNIFICANCE OF THE RESEARCH
Terrorist attacks are not limited to government and military structures. Terrorists have
expanded their attacks with an intention to cause heavy human casualties and economic
losses to common and crowded public facilities such as transportation structures,
shopping malls and office buildings. A significant number of appropriate protective
measures have been put in place to protect above ground structures against blast
loading. However, there is no design guidance available to mitigate the adverse blast
effects on bored tunnels with segmented joints. This highlights a need to develop
appropriate guidance and recommendation for design of newly built bored tunnels to
provide safety to both passengers and structure. It is also expected that this research will
develop design information for strengthening typical existing bored tunnels.
This research will develop and apply fully coupled numerical techniques incorporating
Fluid Structure Interaction (FSI) and using Arbitrary Lagrangian-Eulerian (ALE)
techniques. It provides new information on the influence of important parameters such
as soil types, geometrical shape of segments, joint types and numbers of segments in a
ring. The study is further extended to evaluate mitigation measures in order to reduce
the blast impact on the tunnel response. In general, this research is intentionally carried
out on a common single tube railway tunnel system subjected to credible surface blast
from Vehicle Borne Improvised Explosive Device (VBIED).
1.5 THESIS OUTLINE
The present thesis on “Blast Response and Sensitivity Analysis of Segmental Tunnel”
consists of eight chapters, including introduction (Chapter 1) and conclusion (Chapter
8). Each chapter covers one particular theme of study in the research. The outline of this
thesis is organized as follows.
 In Chapter 2, a comprehensive literature review of blast loads and resulting

wave propagation through soil, soil-structure interaction and tunnel response
under blast loads are presented.
 Chapter 3 describes the basic principles of numerical techniques, including
different types of spatial discretisation solvers, Arbitrary Lagrangian-Eulerian
Page 8
(ALE) techniques and Smooth Particle Hydrodynamics (SPH) techniques as

well as suitable material models adopted in this study.
 Chapter 4 discusses in detail the existing experimental records used to validate
the numerical techniques used in this study. It then describes the validation of
the fully coupled numerical models.
 The developed numerical techniques are applied in Chapter 4 to investigate the
blast response of segmented bored tunnel by varying the geometry of segments
in Chapter 5.
 Chapter 6 describes crucial parameters affecting the blast response of segmented
tunnel in soft soils. The effect of soil properties on the tunnel response under
surface explosion and the influence of tunnel flexibility by varying geometric
properties of the tunnel lining are discussed in this chapter.
 Chapter 7 presents the results of a parametric study to evaluate the performance
of a segmented tunnel by introducing different types of protective barriers,
segment materials and energy absorbing pads between segments to mitigate the
blast effects.
 Finally, Chapter 8 summarises and discusses the main findings of this research
and presents practical implications, and recommendation for future work.
Page 9
Page 10
Chapter 2: Literature Review
2Chapter 2: Literature Review
2.1 INTRODUCTION
This chapter explores the current knowledge on explosion, shockwave propagation in

soil, soil-structure interaction (SSI) and the response of buried tunnels to ground shock.
2.2 EXPLOSION AND RELATED PHENOMENON

An explosive is a chemical compound or a complex mixture that explodes due to either
a rapid chemical reaction or electric sparks or flame. As a result, the surrounding
environment experiences an increase in volume of the explosive gas and release of
energy in an extreme manner. During the explosion in both the air and the ground, a
shock wave composed of a high pressure wave front expands violently from the centre
of the detonation with pressure intensity decaying with distance and time.
2.2.1 Explosives and TNT equivalent
The characteristics of blast loading are determined by many factors such as charge
weight, stand-off distance and type of explosive. There are many types of explosive, but
not all kinds of explosives release the same amount of blast energy. As a result, the TNT
(Trinitrotoluene) equivalent is introduced as an “Explosive Bench Mark” for
quantifying the energy released in explosions. The “a ton of TNT” is a unit of energy
equal to 4.184GJ, which is nearly the amount of energy generated in the explosion of
the one ton of TNT. Each explosive material has a unique TNT equivalent value which
depends on several factors such as detonation rate, effectiveness, and amount of energy
released. This definition is essential, for instance, when placing limits in blast load
investigations. Table 2.1 illustrates the TNT equivalent of common explosives
materials.
Page 11
Table 2.1: TNT equivalent of common explosives materials (Jayasooriya et al., 2011)
Explosive TNT Equivalent
ANFO 0.82
Composition A-3 1.09
Composition B 1.1-1.2
Composition C-4 1.37
Cyclotol (70/30) 1.14
HBX-1 1.17
HBX-3 1.14
HMX 1.3
H-6 1.38
Minol II 1.2
Nitro-glycerine 1.5
Octol(70/30) 1.06
PBX-9010 1.29
PETN 1.27
Pentolite 1.38-1.5
Picratol 0.9
RDX 1.2
Tetryt1 1.07
TNETB 1.36
TNT (Trinitrotoluene) 1
TRITONAL 1.07
2.2.2 Blast scaling laws
The Hopkinson-Cranz scaling law, or cube-root scaling, is the most common form of
scaling to describe blast wave characteristics (Cooper & Kurowski, 1996). This states
that similar explosive waves are generated at identical scaled distances when two
explosive charges of similar geometry and type are detonated in the same atmosphere.
In the Hopkinson-Cranz law, the scaled distance Z is described as a function of the
charge weight of explosive in terms of TNT equivalent (W) and the distance R from the
centre of the charge (stand-off distance) , given by Equation 2.1.
Page 12
R Equation 2.1
Z 1
3
W
The scaling factor Z allows a compact and efficient representation of shock wave data
for a wide range of air-blast situations. De (2012) used the scaling law to investigate the
response of buried tunnel subjected to surface explosion over dry sand. For buried
explosion, Yankelevsky et al. (2011) found that a relatively loose sand with a
considerable high volumetric strain potential well describes the scaling law, but
saturated clays with a relatively low volumetric strain potential showed a large
discrepancy in the results. Lee (2006) also describes the limitation of this approach for
buried explosion in saturated soils as both water and soil transmit the shock wave at
significantly greater rates with greater shock impedance and less compressibility than in
air.
2.3 SHOCK WAVE PROPAGATION
A ground shock occurs due to the sudden release of energy from an explosion at or near
the ground surface. The ground shock is produced in two primary ways, each of which
sets the ground in motion. Firstly, air-induced ground shock occurs if the energy is
transmitted through the air and it runs over the ground surface. Secondly, direct-induced
ground shock results from the energy transmission process due to the direct coupling of
blast energy around formation of crater.
The air-induced ground shock results from an explosion occurring in free-air high above
ground surface where the blast waves propagate radially with a spherical wave front
from the charge centre. The shock waves primarily compress the ground and transmit a
stress pulse into the ground. The intensity of the motion is typically downward with a
peak at the ground surface (Joint Departments of the Army, the Navy and the Air Force,
1990). Procházka et al. (2008) found that the contact explosions on the soil surface
transfer up to 30% of the total blast energy into the ground.
Direct-induced ground shock resulting from the surface explosion creates a direct-
induced motion as well as cratering-induced motion. The ground-induced ground shock
is more important in underground engineering as the direct-induced shock has more
Page 13
effect in terms of intensity and duration of the shock wave than the air-induced ground
shock. The intensity and duration of the ground shock primarily depends on soil
properties such as density and degree of water saturation (Zhongqi et al., 2004).
However, the effect of atmospheric moisture in the air-blast pressure is considered to be
small (Glasstone & Dolan, 1977). The blast orientation largely influences the degree of
blast energy imparted to the ground during the direct-induced motion. The Alekseenko
test described that the proportion of blast energy imparted to the air and the ground was
around 53% and 47% respectively when an explosive was buried in such a way that its
upper surface was at the same level as the ground surface (Yang et al., 2010). On the
other hand, explosion from deep buried charges produce greater effects on the buried
structures in magnitude and are much longer in duration than the corresponding blast in
the air (Department of the Army, 1986).
The ground shock generally produces primary waves (P-waves), shear waves (S-waves)
and surface waves. P-waves involve compression and dilation in the same direction as
the waves progresses through the ground. They are rich in blast induced ground shock
(Kramer, 1996). S-waves propagate through the ground by causing particles to move
perpendicular to the direction of the waves. P-waves travel faster than the other wave
types. Seismic velocities of S and P waves assist in the estimation and prediction of the
properties of the subsurface soil materials. Kramer (1996) provides additional insights
on the evaluation of soil properties based on the seismic velocities.
2.4 CRATER FORMATION
The mechanism of crater formation depends on the explosive mass and the height or
depth of the detonation centre with respect to the ground surface. If the explosion is well
above the surface, it creates a conical-shaped crater in the ground. For buried
explosions, gravitational effects govern the crater mechanism. As the depth increases,
larger amounts of overburden subsoil need to be disintegrated and ejected outwards.
This result in increase of crater size to a certain depth, from which the crater size
decreases, or in other word, no crater formation is observed on the ground (Bull &
Woodford, 1998). This is where a cavity is formed.
Page 14
The theory of the crater formation has been discussed in the literature (Ambrosini et al.,
2004; Ambrosini & Luccioni, 2005; Kinney & Graham, 1985; Luccioni et al., 2009).
Ambrosini and Luccioni (2005) conducted a series of experiments to derive empirical
relationships between the depth and the diameter of the crater for surface explosions. It
showed that an insignificant variation of 5% accounts for the differences between the
soil properties. This suggests that the crater formation study may not be adequate to
investigate and differentiate the blast effects on different types of soils.
2.5 SOIL-STRUCTURE INTERACTION (SSI)
The interaction between a structure and the surrounding soil has been of technical
interest for many years. This aspect, however, is not a simple problem due to the
complex effects on interaction between the soil and the structure. This indeterminacy
arises from the process in which the response of the soil affects the structural motion as
well as the structural motion affects the soil response. In other words, structural
deformation influences the distribution and magnitude of the surrounding earth pressure
at the soil-structure interface.
In the past, designers ignored the SSI effects in designing buried structures to dynamic
loading with a conservative simplification that the SSI has a beneficial effect on the
structural response. As a result, simplified free-field deformation methods have
conventionally been considered (Wang, 1993). However, the assumption is not always
true and the SSI causes detrimental effects on the structural response. With the
introduction of advance numerical techniques, drawbacks associated with the free-field
deformation methods are eliminated by considering the SSI effects (Xu & Liu, 2008).
The nature of the SSI effect on buried structures varies with the confining stress field. If
the stress due to the surrounding load has a significant inertia effect, then the structural
response would be transient. This is called dynamic SSI. The transient phase continues
for a few milliseconds during an explosion-induced ground shock, but the duration is
relatively long for earthquake excitation. Apart from the loading, factors such as (i)
structural material, size and stiffness; (ii) surrounding soil, and (iii) construction method
(example, bored tunnel, mined tunnel and cut-cover-tunnel), affect the SSI.
Page 15
Krauthammer and Chen (1989) indicated that the introduction of interface model
between the structure and the surrounding soil had a significant effect on the response of
buried structures. Sandford (2000) further described that the buried structures cannot
resist the external loads, including surrounding soil, without utilizing the strength of the
soil in an extreme situation. Overall, the literature survey reveals that the SSI effect is
an essential aspect in the tunnel-soil interaction under blast loads.
2.6 BLAST RESPONSE OF BURIED TUNNELS
In the past decades, a series of experiments and numerical simulations have been
conducted to investigate the consequences of explosion on super structures, such as
buildings and bridges. Nevertheless, blast response of buried transportation structures is
still the subject of investigation. In this field, there are no records of information on
field prototype experiments in the literature as full-scale field tests are limited by
constraints of cost, risk and time. A few recent and current studies however used scaled-
down centrifuge modelling techniques to investigate the blast response of tunnels.
2.6.1 Experimental approach
When reliable real-time field data are difficult to obtain, centrifuge modelling is
beneficial for scale modelling of large-scale nonlinear problems in geotechnical
engineering. A number of researchers (Davies, 1994; Davies & Williams, 1992; De,
2012; De et al., 2013; Kutter et al., 1988; Whittaker, 1987) have successfully
implemented the centrifuge modelling to simulate the blast response of buried
structures. De (2012) and De et al. (2013) conducted a series of centrifuge tests to study
the surface blast effect on cylindrical structure buried in dry sand. These experiments
provide quantitative information on the effects of surface explosion on the buried
tunnels. Due to varying gravitational field in the test bucket, the centrifuge models are
limited to smaller models which may not be adequate to predict the blast response of
bolted joints in the segmental lining of bored tunnels. On a small scale, Hayes (1989)
conducted a Conventional Weapon Effects Backfill (CONWEB) test series to
investigate the blast response of buried reinforced concrete structures with differing
backfills. This study used a small model of a reinforced concrete slab bolted to a
reaction structure to study its blast response.
Page 16
2.6.2 Numerical approach
Since the advent of super computers, explicit dynamic nonlinear finite element
approaches provide an alternative technique to solve a broad range of blast problems.
These techniques allow the quality and quantity of blast simulations to be increased in
an affordable and time-efficient manner. Yang et al. (2010) conducted a numerical
simulation of the blast response of a metro tunnel in Shanghai using an advanced
general purpose multi-physics simulation software LS-DYNA (LSTC, 2007). They used
Arbitrary Lagrangian-Eulerian (ALE) method where Eulerian meshes were used for air,
soil and explosive while Lagrangian mesh was employed for the tunnel. Merged nodes
at the interface of the two parts facilitated the stress transfer between the soil and the
tunnel. Feldgun et al. (2008) investigated the soil-structure separation of a buried tunnel
subjected to internal explosion. The comparison of analysis with and without soil-
structural separation shows a significant variation in the tunnel response. Olarewaju
(2012 (a), 2012 (b), 2013) describes a recent series of numerical studies that had been
carried out on buried pipes to investigate the pipe response due to accidental explosions
using the commercial Finite Element program ABAQUS. In this study, the interaction
between the soil and pipe was simulated by perfectly bonded ’no slip’ condition which
may be appropriate for small pipe structures as the pipe deforms with soil, in which the
separation and sliding effects of the interface are negligible. However, for larger tunnel
structures, it is vital to incorporate the contact model that allows separation, recontact
and sliding.
Olarewaju (2012 (a), 2012 (b), 2013) simplified the soil and pipe materials to be linear,
homogeneous and isotropic. This assumption may be valid for far field blast and
subsequent structural response, but the behaviour of soil and resistance of pipe material
under blast loading are highly nonlinear. In addition, the material models should include
strain rate effects.
Morris et al. (2004) studied the blast response of buried structures using distinct element
method (DEM). This study simulated a rock excavated lined tunnel and compared the
results with experimental data. Although the predicted peak velocities and
dispalcements are reasonable, this study was unable to provide complete damage. Gui
Page 17
and Chien (2006) investigated a bored tunnel passing beneath Taipei Shongsan airport
under buried explosion. Their model employed the Finite Difference software FLAC 2D
with a simplification of three dimensional (3D) blast wave into a two dimensional (2D)
blast wave. It was a conservative approach, though it considerably minimized the
computation cost. The model also simulated the soil-tunnel interface with static and
dynamic frictions in order to permit sliding. Liu (2009; 2012) investigated the dynamic
response of subway tunnels subjected to internal blast using the computer code
ABAQUS, where the cast iron tunnel was modelled as a shell structure, assuming a
triangular blast pressure on the internal surface of the tunnel. Although the focus of the
study was on the SSI and stiffness of the lining, the simulation did not consider the
important aspects of air-structure interaction, a 3D blast effect and direct heat effect
from the internal blast. The heat effect may not be very important to the buried tunnels
subjected to surface blast as the surrounding soil medium will act as a good heat
insulation material. When an explosive detonates in contact with the ground, the blast
waves propagate hemispherically in the ground and affect the performance of structures
in all directions. This emphasizes the need for 3D analysis in order to simulate a more
realistic problem (Higgins, 2011). The 3D analysis requires considerable computational
time and large memory. Application of an effective modelling technique greatly reduces
the computation time and memory.
Using the centrifuge test results, De (2012) validated his numerical model developed
from AUTODYN. Eulerian meshes were used to model the air and explosive while the
soil and the copper-tunnel were modelled with Lagrangian meshes. Penalty based
interaction ’Lagrange-Lagrange’ method simulated the interface between the soil and
tunnel. He utilised a default sand model based on Laine and Sandvik (2001). The
material model simulates the strain rate behaviour of sand (Higgins et al., 2012). Soil
moisture content has a significant influence on the response of buried structures under
blast loads. To date, no one has studied the effects of pore water pressure in the soil on
the blast response of underground tunnels. De (2012) also highlighted the need for soil
models to incorporate pore water effects.
Page 18
The studies performed to date have ignored the important aspect of the tunnel joints.
Munfah (2009) also indicated that the precast segmental lining is more vulnerable to
blast than thick cast-in-place tunnels.
2.7 RESEARCH GAP
In recent years a number of researchers have performed analyses using various finite
element hydrocodes with implementation of various constitutive material models to
simulate a variety of blast related problems. Although blast performance of above
ground structures like buildings and bridges has gained increasing interest, less attention
has been paid to the blast performance of buried structures with a common belief that
buried structures are much safer than aboveground structures. In fact, it is necessary to
consider the critical factors, such as intensity of shockwave, flexibility of structure and
the surrounding ground that influence the vulnerability of buried structures to blast
loading.
An underground transit tunnel is a critical piece of infrastructure as a large number of

people use it. Some of them are designed to resist natural disasters such as earthquakes
and fires, but most tunnels are not designed to resist credible explosions. As stated
previously, there are no records in the open literature regarding the blast performance of
segmental tunnel lining (bored tunnel by TBM) with multiple bolted joints. These joints
are semi-rigid joints assisting the segments to transfer loads to neighbouring segments
by allowing rotation without reducing the load carrying capacity under geostatic
loading. The influence of the joints on the tunnel response to surface blast is hence less
understood. Since there is no comprehensive guidance to evaluate the performance of
segmental tunnel subjected to surface blast, a detail investigation is required. This
research will address the knowledge gaps identified above.
The literature review revealed that numerical modelling technique can be utilized to
provide valuable data in a timely and cost effective manner to predict the blast
performance of segmental tunnel. The next chapter describes the finite element
modelling technique used in the present research.
Page 19
Page 20
Chapter 3: Numerical Modelling
3Chapter 3: Numerical Modelling
3.1 INTRODUCTION
Hydrocodes are the computational tools that can be used to solve a wide range of non-
linear problems in solid, fluid flow and explosions. There are several numerical
techniques available in hydrocodes, but selection of an appropriate numerical technique
is wholly dependent on the type of problem and computational cost. The numerical
techniques described in this section evaluate the appropriate techniques and to expand
the state of the art in surface blast response of underground tunnels. The general
purpose finite element hydrocode LS-DYNA (LSTC, 2007) is employed for this
numerical simulation. This chapter describes the theory behind the numerical technique
and various constitutive material models used in this study.
The numerical simulations divide a system into finite elements, a process called
discretisation. The discretisation takes place with respect to time (temporal) and space
(spatial). In numerical simulations, temporal discretisation uses the explicit method
which calculates the state of a system at a later time as a function of time step from the
current state of the system. In order to capture any activity within an element, the time
step should satisfy the Courant-Friedrich-Levy (CFL) condition for which the time step
(∆t) is less than the duration for sound to cross the smallest element. Usually, it is
advised to use a Safety Factor (SF) of 0.67 for blast loads in LS-DYNA, which could be
different for other computer software. This condition can be generally expressed as
below:
l
0  t  N Equation 3.1
c
where N is the safety factor, l is the least element size and c is the speed of sound
through the element.
Different types of spatial discretisation solvers are available to model various materials
and their conditions, utilising the suitable features of each solver. A Lagrangian solver
is used in the finite element method, in which the elements move with the material
Page 21
during deformation. There is no flow of material from one element to another element.
This solver is more suitable for solid objects where the deformation is addressed by the
distortion of the mesh. For large deformation problems, this solver undergoes severe
element distortion which may result in very small time steps (∆t) and grid tangling. An
Eulerian solver is used in the finite difference method and/or finite volume method, in
which the mesh remains undeformed while the material flows freely from one to
another element. This solver is most suitable for fluid materials such as a gas or a liquid.
The Eulerian solver can also be used to model solid objects where larger deformation is
inevitable, but this solver has difficulty dealing with the material boundary conditions
involving surface slippage in contact (Wang et al., 2005).
3.2 ARBITRARY LAGRANGIAN EULERIAN
The Arbitrary Lagrangian Eulerian (ALE) technique is introduced by combining the

best features of pure Lagrangian and Eulerian solvers, while reducing their respective
weaknesses. In ALE, the nodes in the computational mesh may be moved with the
continuum in a Lagrangian manner, or held fixed in an Eulerian manner, or moved in
some arbitrarily specified way to give a continuous rezoning capability. With this
feature of freedom to move the computational mesh, the ALE efficiently handles the
problem with greater distortions of the mesh than the purely Lagrangian method. It also
offers more resolution than the purely Eulerian method. The ALE technique allows
modelling of fluid-structure interactions (FSI) with a fluid structure coupling algorithm
which satisfies the governing equations describing the conservation of mass, momentum
and energy (Hallquist, 2006).
FSI is a coupled multi-physics simulation process for solving highly non-linear

problems in a single model. Examples of this type of simulation include surface
explosions, where the air is Eulerian and the soil and the tunnel are Lagrangian, and
buried explosions, where the soil is Eulerian and the target tunnel is Lagrangian. In all
cases, the Lagrangian mesh moves through the Eulerian mesh. Figure 3.1 shows the
overlapping of the Lagrangian boundary with the Eulerian mesh. The coupling
algorithm first searches for the intersections of the Eulerian mesh with Lagrangian
mesh. When the Lagrangian surface is detected inside the Eulerian mesh, the algorithm
applies pressure boundary conditions on the Largrangian mesh from the Eulerian mesh,
Page 22
causing displacement on the structure. In return, the Lagrangian mesh boundaries act as
velocity boundary (geometric constraints) in the Eulerian mesh (McMaster, 1984).
Figure 3.1: Lagrangian-Eulerian coupling
The simulation uses the ALE Multi-Material formulation to model a mixture of two or
more different materials in each mesh (Alia & Souli, 2006). The explosive is prescribed
into the surrounding (air or soil) using INITIAL_VOLUME_FRACTION option in LS-
DYNA as a sphere or a cylinder or a cube. This allows the explosive substances to
expand into the meshes initially occupied by the air or soil. The coupling between the
Lagrangian meshes and Eulerian meshes is accomplished using CONSTRAINED_
LAGRANGE_IN_SOLID, where the penalty factor controls the penetration of the
explosive-air volume fraction into the Lagragian mesh.
3.3 SMOOTH PARTICLE HYDRODYNAMICS
Smooth Particle Hydrodynamics (SPH) is a mesh-free computational Lagrangian

hydrodynamic particle method. This method originated about three decades ago in
astrophysical problems (Gingold & Monaghan, 1977; Lucy, 1977), where it dealt with
the simulation of interacting fluid masses in vacuum without boundaries. It was then
modified as a deterministic mesh-free particle method and implemented to continuum
solid and fluid mechanics (Monaghan, 1992, 1994). SPH treats the solid mechanics
problems in which large deformations and fragmentation occur. SPH is mathematically
based on interpolation theory by utilizing kernel approximation of a function which is
adequately smooth even for higher order derivatives and provides stable and accurate
results. When SPH was initially implemented to fields other than astrophysics, it
Page 23
suffered from the enforcing of finite boundary conditions. Modelling of finite element
(FE) meshes as the boundaries for SPH particles offers a possible solution (Thiyahuddin
et al., 2012).
SPH provides extensive ability as a numerical tool for modelling problems having large
distortions and deformations. Unlike conventional Lagrangian meshes, SPH is free from
mesh tangling and hourglassing effects. As such, SPH particles are used where large
deformation or severe material failure occurs in near field domain and FE meshes are
used where intermediate or small deformation is expected in far field domains.
a) Tied interface b) Hybrid element coupling c) ‘Nodes-to-surface’ contact
Figure 3.2: Coupled SPH-FE Method
Although SPH is more expensive in terms of computation, the coupled SPH-FE

approach reduces high computational demand. There are methods that allow the
coupling interaction between SPH particles and FE meshes (Wang et al., 2005). A
constraint interface ties SPH particles to the corresponding surfaces of FEM meshes as
shown in Figure 3.2(a). This method requires due consideration in FEM mesh creation
at the interface as it could cause unphysical penetration and system instability. Hybrid
element coupling is a new feature in LS-DYNA. As shown in Figure 3.2(b), these
elements constrain SPH particles and act as transit layers between SPH particles and FE
meshes. The main advantage of this method is that no tied/contact interface is required.
However, this method suffers from intruding SPH particles into FE mesh if hybrid
elements are close to the explosive. The third method is a simple method where the
interaction is achieved by the penalty based ‘Nodes-to-surface’ contact where solid
elements as the master and SPH particles as the slave nodes as shown in Figure 3.2(c).
Page 24
The ‘Node-to-surface’ coupling method is employed in the present study as SPH

particles are close enough to achieve steady interface.
3.4 SIMULATION OF BLAST LOADING
Previous studies have shown that both ALE and SHP techniques are well suited to simulate
the blast response of buried tunnels. The following sections describe the most significant
key words used in the modelling of the problem.
3.4.1 Element selection
LS-DYNA provides several element formulation options for solids. Solid elements
(Lagrangian elements) can be modelled using hexahedral-eight node fully integrated
elements (ELFORM=2) or constant stress solid elements (ELFORM=1), which is the
default formulation. Although ELFORM=2 is free from hourglass stabilization, it is
computationally expensive for hi-fidelity models as well as too stiff, especially for poor
aspect ratios. For most models, ELFORM=1 provides reliable results when suitable
hourglass control is used. Although there are ALE element formulations (ELFORM=5,
6 and 12), the recommended ELFORM=11 is used for the multi-material ALE
(Jayasinghe et al., 2013).
3.4.2 Boundary condition
As with the explosive charge centre, node displacements at the symmetry boundaries in
the model can be restrained in their normal directions using symmetry boundary
condition. The top surfaces of air and soil-air interface are considered as offering no
constraints while the base of the model can be fixed in all directions to represent the bed
rock. In SPH simulation, a special symmetry boundary BOUNDARY_SPH_
SYMMETRY_PLANE is used for the SPH particles at the symmetric boundaries. As
reflection of shock waves at the far-field can affect the accuracy of the numerical
simulation, non-reflecting boundary conditions (BOUNDARY_NON_REFLECTING)
are placed to allow the shock wave to flow out the soil skeleton at the far-field infinite
domain without reflection.
Page 25
3.4.3 Contact interface
The SSI contact interface between tunnel and soil is a vital part of simulation involving
penetration, sliding and separation. The selection of contact type depends on the
interaction phenomenon between the SSI interfaces. LS-DYNA performs this operation
by checking potential penetration of the slave set through a master set. Penetration
algorithms conduct this check at each time step. AUTOMATIC contact option detects
the penetration from both sides of contact. Therefore, orientations of slave set and
master set are not important, making them more robust than non-automatic option. In
this present study, the Soil-Structural Interaction (SSI) at the contact interface between
the soil and the tunnel is described by CONTACT_AUTOMATIC_SURFACE_TO_
SURFACE type contact. The most popular contact option CONTACT_AUTOMATIC_
SINGLE_SURFACE is defined for self-contact between tunnel segments.
3.4.4 Reinforced concrete model
Precast tunnel segments are modelled using eight node hexahedral solid elements with
Lagrangian processor while steel reinforcing bars are defined using two node Hughes-
Liu beam elements with cross sectional integration. These reinforcing elements are
simple, computationally efficient and compatible with most material types. The beam
elements within the concrete segments can be modelled using the CONSTRAINED_
LAGRANGE_IN_SOLID (CTYPE=2) coupling. This eliminates the burden of aligning
the beam nodes to the solid element nodes. Minimum of three hexahedral elements are
specified across its thickness to facilitate the non-linear stress distribution through the
thickness.
3.4.5 Stress initialization
Stress initialisation is essential in geotechnical modelling to induce a steady state

preload prior to the application of the transient load simulating the blast. In order to
bring the model to a state of static equilibrium, a small amount of damping is
incorporated and hence kinetic energy approaches zero. This can be achieved using time
dependent mass damping option DAMPING_GLOBAL to impose near-critical damping
until the preload is established in the model. The solution will be computed dynamically
Page 26
until it converges to equilibrium. For optimal convergence, lowest mode of vibration

frequency can be obtained from the damping constant DS which is defined by:
2
DS  Equation 3.2
T
where T is the longest period of vibration for the model being computed from an
undamped transient analysis.
The gravity load can be applied to the model using LOAD_BODY_Z option. The
simulation used two curves: (i) gravity load curve includes a ramp load function and (ii)
damping curve defines damping constant only during the stress initialization phase. For
stability reasons, the stress initialization is normally conducted with a default time step
SF of 0.90.
3.4.6 Data manipulation
LS-DYNA provides two formats of analysis data output such as binary files and ASCII.
The binary data files contain the model’s finite element response information (d3plot,
d3thdt, d3dump, interface force file (intfor)). The binary file recording interval can be
varied, for example, from 250sec, 500sec to 1000sec, depending on how accurate
peak measurements are required to capture the tunnel response during the explosion.
LS-PrePost processes output from LS-DYNA to visualize the three-dimensional
response. LS-PrePost reads results data files such as d3plot, intfor and all ASCII time
history data files. The powerful animation and colour contour images on meshes
produced by LS-PrePost include acceleration, velocity, displacement and element
pressure time-history data. Furthermore, LS-PrePost computes a variety of strains,
bending moments and reaction forces along the constrained boundaries. The binary
d3dump files are automatically written at the end of normal termination can be used to
restart the model by removing redundant elements in the model.
In addition, the Winfrith concrete model (LSTC, 2007) used in this study to simulate the
blast response of reinforced concrete lining provides an attractive feature of monitoring
crack propagation in the database. This database includes the information on crack
Page 27
location, directions and widths. This crack database can be visualized on the deformed
mesh output.
3.5 MATERIAL MODELS
The following section describes the material models considered for modelling air, high
explosive, soil and reinforced concrete tunnel segments made of concrete and steel bars
in the numerical simulation. They are:
3.5.1 Air model
The air is modelled as an ideal gas (LSTC, 2007) utilising MAT_NULL material model
with a linear polynomial Equation of State (EOS). The null material model can
effectively model fluids and hydrodynamic substances. The pressure is expressed by:
 
P  C0  C1  C2  2  C3 3  C4  C5   C6  2 E Equation 3.3
where E is the internal energy per unit volume, C0, C1, C2, C3, C4, C5, and C6 are
 
constants and    1 , where is the ratio of current density to initial density.
0 0
The linear polynomial equation represents an ideal gas with the gamma law EOS, in
which C0=C1=C2=C3= C6= 0 and C4 = C5=   1 , where  is the ratio of specific heat at
constant pressure per specified heat at constant volume. The pressure is then described
by:

P  (  1) E Equation 3.4
0 0
where: E0  0cv  T Equation 3.5
 is an adiabatic constant for air behaving as an ideal gas (estimated value for  =1.4), ρ
is the density, cv is the specific heat at constant volume and E0 is the initial internal
energy per unit volume. Table 3.1 shows the material parameters used for the air.
Page 28
Table 3.1: Material properties for air (Yang et al. 2010)
Parameter Value
ρ 1.29g/cm3
C0, C1, C2, C3 and C6 0
C4 and C5 0.4
E0 0.25kJ/m3
3.5.2 Explosive model
The Jone-Wilkin-Lee’s EOS (LSTC, 2007) is used to describe the explosive as it is the
most popular one and the easiest to calibrate. This EOS defines the pressure as below:
    R1V     R2V E
P  A1  e  B1  e  Equation 3.6
 R1 
V  R 2V  V
In the above equation V is the relative volume or the expansion of the explosive, E is the
initial energy per volume, vD is the detonation velocity, PCJ is the pressure at the critical
point, A, B, R1, R2 and ω are empirically derived constants for the explosive. This
equation allows using a wide range of different high explosive parameters such as for
TNT (Wang et al., 2005) and C-4 (Wang et al., 2001) explosives. Table 4.2 shows the
material parameters for TNT and C-4 explosive charges.
Table 3.2: Material parameters for TNT explosive
Parameters TNT C-4

ρ (g/cm3) 1.630 1.601
vD (m/s) 6930 8193
PCJ (GPa) 21 28
A (GPa) 373.77 609.97
B (GPa) 3.747 12.95
R1 4.15 4.5
R2 0.90 1.4
ω 0.35 0.25
V 1 1
3
E0 (kJ/m ) 6.0e+06 9.0e+06
Page 29
3.5.3 Soil model
Soils with different moisture contents give different responses to high strain-rate
dynamic loading, such as direct impact and blast loading. This research uses an
appropriate soil model that incorporates the various soil compositions, in particular,
moisture content. By evaluating several material models in LS-DYNA,
MAT_FHWA_SOIL model is identified as a suitable soil model that includes strain
softening, kinematic hardening, strain rate effects, element deletion, excess pore water
effects and stability with no soil confinement (Lewis, 2004; Saleh & Edwards, 2011).
Furthermore, the model simulates an isotropic material behaviour with failure. The
material model was developed by the Federal Highway Administration (FHWA) in
2004. The FHWA soil model is based on a modified Mohr-Coulomb yield criterion
(Abbo & Sloan, 1995). The modified yield surface is a smooth hyperbolic surface which
enables an accurate, robust, and a significant computational saving in numerical
simulation.
There are a total of 25 material parameters in the FHWA soil model. Among those
parameters, the material model requires main parameters such as, mass density, specific
gravity, bulk modulus, shear modulus and moisture content. These soil parameters are
generally determined through laboratory tests. Parameters required for defining strain
softening, kinematic hardening, strain rate effects and pore water effects can be
evaluated through laboratory tests and/or equations in the material manual (Lewis,
2004). By considering the pore-water effect, Sivalingam et al. (2014), Jayasinghe et al.
(2013) and Ortman and Catherine (2008) and Lee (2006) successfully employed the
model for studying the blast effect in fully saturated soil. If the soil is unsaturated or
dry, the same model can be utilised by eliminating pore-water effects (Lee, 2006). Saleh
and Saleh and Edwards (2011) also used this model for investigating the interaction of
landmine explosion with protective structures by ignoring the pore-water effects. Values
recommended by Lee (2006) were used in this study to include the strain softening,
kinematic hardening and strain rate effects in the soil.
Shear and bulk moduli are two of the main parameters. Shear wave velocity (VS) is a
valuable indicator of the dynamic soil properties because of its relationship with the
shear modulus (G) as given by Equation 3.7 (Kramer, 1996):
Page 30
G  VS2 Equation 3.7
where ρ is the soil density.
A wide variety of field and laboratory tests are available to evaluate the magnitude of
primary wave (VP) and shear wave (VS) velocities. The velocities of VP and VS depend
on the compressibility of the soil body through Equation 3.8 (Kramer, 1996):
VP 2  2v
 Equation 3.8
VS 1  2v
where v is the Poisson’s ratio of soil.
This ratio is typically assumed to be constant in the studies of both dry and saturated
soils of homogeneous and isotropic media. However, the Poisson’s ratio varies with the
degree of saturation. For a typical Poisson’s ratio of 0.3 for dry sandy soil, the ratio
VP/VS is 1.87. However, for low shear velocity materials such as saturated clay, the
Poisson’s ratio can be close enough to 0.5. When it approaches to 0.5, the ratio VP/VS is
infinite.
The bulk modulus (K) is a degree of resistance against volumetric compression. The
compressibility of the material is the reciprocal of the bulk modulus that relates the
shear modulus (G) with the Poisson’s ratio ( v ), given by Equation 3.9:
2G (1  v)
K Equation 3.9
3(1  2v)
The soil skeleton has pores filled with air and/or water. The water filled in the pores
creates pore water pressure. The MAT_FHWA_SOIL soil model handles pore water
pressure build-up using the relationship between pore-water pressure (u) and volumetric
compression strain (  v ) as shown in Equation 3.10 (Lewis, 2004).
K sk Equation 3.10
u v
1  K sk D2 ncur
Page 31
where Ksk = skeleton modulus, ncur = current porosity due to air voids and D2 = a
function of Skempton’s pore-water pressure.
Equation 3.11 defines the parameter D2 from Skempton pore-water pressure parameter
B, as shown below:
1 1 B
B , D2 
1 n
K sk BK sk [n(1  S )] Equation 3.11
K
where n = soil porosity and S = degree of saturation.
The parameter D2 is constant for partially saturated soil. For fully saturated case, the
constant D2 has no effect on pore-water pressure (i.e D2 = 0) (Lee, 2006). As described
by Lee (2006), Ksk is the volumetric strain factor which varies between 5% and 20% of
material bulk modulus.
The effects of excess pore-water pressure are defined in the material model using
Equation 3.12 (Lewis, 2004).
Ki
K Equation 3.12
1  Ki D1ncur
where Ki = nonporous bulk modulus and D1 = stiffness of the soil by adjusting the bulk
modulus before the air voids are collapsed. For fully saturated case, Lee (2006)
estimated D1 to be 4.63GPa-1.
Inclusion of water in soil changes the magnitude of shear and bulk moduli with degree
of saturation. Water has very low compressibility (higher bulk modulus≈ 2GPa) and
negligible resistance to shear or distortional loads (negligible shear modulus).
Naesgaard et al. (2007) reported that the shear wave velocity (VS) is essentially
independent of the Skempton parameter B. Therefore, the saturated shear modulus (Gsat)
can be assumed to be the same as the dry shear modulus (G). If laboratory or field data
are available, Equation 3.13 can be used to evaluate Gsat.
Page 32
Gsat  satVS2sat Equation 3.13
The bulk modulus of water has more influence over the soil-water mixture than the bulk
modulus of solid soil particle. Therefore, the bulk modulus of the soil-water composite
mixture should be within those of water and soil bulk moduli. Lee (2006) described
additional insights of the evaluation of the saturated bulk modulus (Ksat). Alternatively,
laboratory and field tests can help to evaluate the primary wave velocity (VP) of
saturated soil. The magnitude of the saturated bulk modulus (Ksat) can be evaluated from
Equation 3.14.
4
K sat   satVP2 sat  Gsat Equation 3.14
3
where VP-sat is the primary wave velocity in saturated soil.
3.5.4 Concrete model
Dynamic performance of concrete structures subjected to blast effects is a complex

nonlinear and rate-dependent process, in which the apparent concrete strength can
increase significantly. LS-DYNA material library has several advanced constitutive
material models developed to simulate the concrete material behaviour. However, in
many circumstances, the necessary material parameters for the concrete cannot be found
in the literature. This study utilized a common material model MAT_WINFRITH_
CONCRETE (LSTC, 2007) which has the automatic generation capability of concrete
law parameters. This material model originated in the nuclear industry to simulate the
response of reinforced concrete structures subjected to blast and impact loading.
This model is based on the shear failure surface proposed by Ottosen (1977) and
includes the strain rate effects and strain softening in tension by incorporating crack
opening width or fracture energy. The shear failure surface is expressed by a quadratic
equation given as:
J2 J I
F ( I1 , J 2, cos 3 ) a ' 2
  '2 b 1'  Equation 3.15
( fc ) fc fc
Page 33
where I1 is the first invariant of stress component representing volumetric responses and
J2 is the second invariant of the deviatoric stress component which accounts for
deviatoric response. The constants a and b control the meridional shape of the shear
failure surface. These parameters are internally generated in the Winfrith concrete
model based on input parameters such as unconfined tensile and compressive strengths.

 k cos  cos (k2 cos 3 )  ,
-1
 1   for cos 3  0
  3 
   cos-1 (-k cos 3 )  Equation 3.16
 1k cos  - 2
 , for cos 3  0
 

3 3 
where k1 and k2 are functions of the ratio of the unconfined tensile to compressive
strength ( ft/ fc’ ). They define the shape of the shear failure surface in the octahedral
plane.
3 3 J3
cos 3  Equation 3.17
2 J 21.5
where J2 and J3 are the deviatoric stress components and the angle  is referred to as the
Lode Angle.
This is a simple input concrete model which requires the unconfined compressive
strength and tensile strength. For known concrete weight and unconfined concrete
compressive strength, Young’s modulus (E) is calculated from an equation
recommended by ACI Committee 318 as below:
E  33w1.5 fc' Equation 3.18
where w is the concrete weight in lb/ft3 and fc’ is the compressive strength of normal
strength concrete in psi.
Page 34
3.5.5 Reinforcement model
In the present study, material model MAT_PLASTIC_KINEMATIC is employed to

model the steel reinforcement. This model is more efficient in terms of the
computational demand. The steel is an isotropic elastic-plastic material with the same
initial yield stress for both uniaxial tension and compression. It uses Von-Mises yield
criterion to define the yield function of the steel, as given by:
F (J2 )  0
3J 2   y  0
Equation 3.19
3
y  sij s ji
2
where J2 is the second invariant of the deviatoric stresses sij.
The strain rate effect is incorporated by using the Cowper Symonds strain rate
relationship given by Equation 3.20.
 1


 

 y  1      0   EP eff 
P
P
Equation 3.20
 C  
 
where,  y is the dynamic flow stress at a uniaxial plastic strain rate  ,  0 is the
associated flow stress, EP is the plastic modulus and eff is the effective plastic strain
P
of the material. This equation computes the strain rate dependent factors to scale the
yield stress. Parameter  is included to differentiate the type of plastic hardening in the
material model. For elastic perfectly-plastic material with kinematic hardening (  = 0)

P
and hence the surplus stress of the plastic hardening part  E P eff is omitted. C and P are
strain rate Cowper-Symonds coefficients which were evaluated based on best fitting of
the Dynamic Increased Factor (DIF) of steel reinforcement as described by Malvar and
Crawford (1998). This model also includes failure upon reaching a predefined value of
failure strain.
Page 35
3.6 CHAPTER SUMMARY
This chapter focused on the theory behind the numerical techniques and constitutive
material models for simulation of buried tunnel structure subjected to surface explosion.
Two different numerical approaches, Arbitrary Lagrangian-Eulerian (ALE) and Smooth
Particle Hydrodynamics (SPH), for fluid-structure interaction in blast problems were
described. The general purpose finite element hydrocode LS-DYNA will be used to
perform those techniques, incorporating the fully coupled numerical simulation and
different material models utilising strain rate effects.
In the present study, atmospheric air surrounding explosive is modelled using null
material model with a linear polynomial equation of state while the explosion process is
modeled using the Jones-Wilkens-Lee (JWL) equation of state with high burn material
model. FHWA material model is used for the soil to incorporate the strain softening,
kinematic hardening, strain rate effects, element deletion, and excess pore water effects.
The Winfrith concrete model is chosen to simulate the concrete, in which the
reinforcement is separately modelled as an isotropic elastic-plastic material by
‘Constrained Lagrange In Solid’ coupling. Gravity preload is incorporated using time
dependent mass damping. Self-contact surfaces are used to define joint interfaces of
segments using the CONTACT_AUTOMATIC_SINGLE_SURFACE type contact. The
contact interface between the tunnel exterior surface and soil is simulated using the
CONTACT_AUTOMATIC_ SURFACE_TO_SURFACE type contact.
Page 36
Chapter 4: Validation of Numerical Models
4Chapter 4: Validation of Numerical Models
4.1 INTRODUCTION
Model verification and validation are essential for the development of numerical models
that can be used to make engineering predictions with quantified confidence. This
chapter presents the validation of the two numerical approaches reported in the previous
chapter:
1. Coupled FSI in ALE (ALE method).

2. Coupled SPH-FE (SPH method).
For the validation purpose, two experiments from the literature are considered. These
are: (1) a series of tests conducted by De (2012) and De et al. (2013) that study the
surface blast effect on a buried copper pipe in dry sand and (2) Conventional Weapon
Effects Backfill (CONWEB) test series carried out by Hayes (1989) to investigate the
underground blast response of a buried Reinforced Concrete (RC) structure in various
backfill soil conditions. Figure 4.1 shows the flowchart of the validation process of
numerical techniques and relevant material models.
Experimental records
Validation of surface blast Validation of blast response of

response of tubular buried buried RC structural models.
structural models. (Centrifuge test) (CONWEB test)
ALE method SPH method ALE method SPH method
Figure 4.1: Flowchart used for validation of numerical techniques and material models
Page 37
4.2 VALIDATION OF ALE SIMULATIONS USING CENTRIFUGE TEST

RESULTS
The study first simulated and validated the experiment conducted by De (2012) who
used a 70g centrifuge testing machine, where g is the gravitational acceleration. De
(2012) investigated the blast response of a tubular tunnel structure using the centrifuge
models.
4.2.1 Description of centrifuge test
The scaled-down model composed of a tubular copper tunnel buried in dry Nevada sand
(relative density (Dr) = 60%) at a depth of 3.6m equivalent to prototype scale as shown in
Figure 4.2. A spherical shape explosive was symmetrically placed above the midspan,
directly over the centreline of the copper tunnel, such that the ground surface was tangent to
the spherical surface of the explosive.
Figure 4.2: Setup of experimental model (All dimensions are in prototype scale) (De &
Zimmie, 2011)
The copper tunnel was instrumented with 19 strain gauges to monitor axial, circumferential
and shear strains at different locations on the tunnel. For comparison purposes, three gauge
readings reported by De (2012) and De & Zimmie (2011) were considered. Figure 4.3
describes these gauge locations in the prototype scale.
Centrifuge scaling factors show how a field model and its dynamic events are correlated in
the centrifuge test, in which the scaled model is sufficiently raised to N times the
gravitational acceleration. Scaling factors for different parameters are shown in Table 4.1.
Page 38
Based on the centrifuge scaling laws, Table 4.2 lists the corresponding prototype model
dimensions for the numerical simulation.
Table 4.1: Scaling law (Kramer, 1996)
Model dimension
Type of event Parameter
Prototype dimension
Stress 1
Strain 1
Length 1/N
All events Mass 1/N3
Density 1
Force 1/N2
Gravity N
Time 1/N
Dynamic events Velocity 1
Acceleration N
Table 4.2: Conversion to prototype model (De, 2012)
Parameters Model dimension Prototype dimension
Copper tunnel diameter 76mm 5.32m

Copper tunnel thickness 2.5mm 175mm
Explosive weight of TNT 2.6g 888kg
4.2.2 Development of numerical model
Symmetric modelling capabilities play an important role in this numerical simulation to

reduce the computational cost by considering a quarter symmetry-geometrical prototype
models with a size of 20m x 16m x 15m as shown in Figure 4.3. This prototype model was
used to represent the Lagrangian structure which is composed of only two major parts:
copper tunnel and soil.
Page 39
Figure 4.3: A quarter symmetrical prototype model (Lagrangian structure)
Eight-node solid elements were used with different spatial discretisation solvers.
Lagrangian meshes were used to model the soil and copper tunnel while ALE meshes
(background mesh) were used separately to model the surrounding air and explosive. A
mesh consistency condition was achieved through a series of cases with different meshes to
capture the analytical solution in the limit of a mesh refinement process. The Lagrangian
soil structure used smaller element size of 12.5cm x 12.5cm x 12.5cm in the region adjacent
to the explosive and larger element size of 40cm x 40cm x 40cm for far field region.
The copper tunnel was modelled with due consideration for a greater mesh refinement of
curved surfaces. Peaks and valleys on the curved surface of copper tunnel interlock with the
interface of soil elements and the element interlocking can cause initial penetration and
crossed edges. Therefore, nodes on the soil interface were modelled to be coincident with
those of the tunnel interface to minimize the element interlocking. The tunnel used three
elements across its thickness to facilitate a nonlinear stress distribution through the
thickness. The tunnel was further refined with a gradual increase in element size in both
axial and circumferential directions away from the incident blast. The size of the smallest
element in the tunnel was 10.25cm x 12.5cm x 5.83cm.
The spherical shape explosive was defined into the background mesh by using INITIAL_
VOLUME_FRACTION_GEOMETRY, by specifying its radius and detonation point. The
contact interface simulation between the soil and the tunnel surfaces is crucial to an accurate
analysis of tunnel response. The contact interface was modelled using CONTACT_
Page 40
AUTOMATIC_SURFACE_TO_SURFACE type contact. The translational displacements

of symmetry boundaries XZ and YZ planes are constrained in their normal directions. The
non-reflecting boundary condition was applied to the infinite domain of both the Eulerian
and the Lagrangian meshes while the base was fixed in all directions to represent the bed
rock.
4.2.3 Material models and parameters
The material models and relevant material properties for the air and explosive described in
Chapter 3.5 were employed in the simulation. The copper tunnel was modelled using
MAT_PLASTICIY_KINEMATIC material model which incorporated both non-linear
material behaviour and high strain rate effects due to the ground shock. Material parameters
for copper (Matuska, 1984; Peroni et al., 2009) are described in Table 4.3. The main
parameters include mass density (ρ), Young’s modulus (Es), Poisson’s ratio (v), yield stress
(σy), tangent modulus (Etan), hardening parameter (β) and strain rate parameters (C) & (P)
for Cowper Symonds strain rate model.
Table 4.3: Material parameters for copper.
ρ Es v σ Etan β C P
g/cm3 GPa MPa MPa s-1
8.93 117 0.35 400 100 0 1.346e+06 5.286
For the Civil and Mechanical Systems Program of the National Science Foundation (NSF),
Nevada sand (at a relative density (Dr) of 60%) was used for centrifuge tests by De (2012).
Arulmoli et al. (1992) conducted an extensive laboratory test for the Nevada sand with
different Dr values including: 40% and 60% in the VELACS (Verification of Liquefaction
Analyses by Centrifuge Studies) Program. From Cyclic Triaxial Test data for Nevada sand
at Dr = 60% (Arulmoli et al., 1992), the main soil parameters such as mass density (ρ),
specific gravity (Gs) and moisture content were reported.
Based on initial void ratio, porosity of the sand was derived as 0.4. De (2012) presented data
for density (ρ) versus sound speed (c) that was used for back-calculation of shear modulus
(G). For Nevada sand v = 0.33 (De, 2012), the ratio VP/VS is 1.985. Using this ratio of
Page 41
VP/VS, the shear wave velocity (VS) can be calculated using Equation 4.1 for known initial
sound speed (c) at initial mass density of 1.6g/cm3.
3(VP2  c 2 )
VS  Equation 4.1
4
Equation 3.7 was used to compute the shear modulus (G), which is consistent with the shear
modulus evaluated from resonant column test for dry Nevada sand at Dr = 60% (Arulmoli et
al., 1992). The Bulk modulus (K) was derived using Equation 3.9. Table 4.4 shows the
material parameters for the Nevada dry sand.
Table 4.4: Material parameters for soil
Parameters Dry Nevada sand
Density (g/cm3) 1.60
Specific gravity 2.67
Shear modulus (MPa) 56.0
Bulk modulus (MPa) 146.0
Cohesion (MPa) 6.20e-03
Friction angle 35o
Moisture content (%) 0
PWD1 (MPa-1) 0
PWD2 (MPa-1) 0
4.2.4 Description of three stage simulation
In order to minimise the computational cost, the numerical modelling technique was
divided into a time-ordered sequence of interrelated phases which described the entire
simulation. LS-DYNA’s restart feature enables the entire simulation to be broken into
three stages such as stress initialization, ALE/Lagrangian coupling and deletion of ALE
background mesh.
Page 42
The model was first preloaded quasi-statically in the form of a ramped load curve with
DAMPING_GLOBAL option which removed the unwanted oscillations in the model.
Figure 4.4 shows the Kinetic Energy (KE) vs. time plot during the stress initialization,
where the KE converged to zero at 150ms.
In addition, three gauge points such as on tunnel crown, springline and invert of the
tunnel cross section at midspan verified the initialization process by using vertical
displacement vs. time plots as shown in Figure 4.5.
30
Soil
25
Copper Tunnel
20
KE (kJ)
15
10
0
0 250 500 750 1000 1250 1500
Time (ms)
Figure 4.4: Kinetic energy vs. time during the initialization
*
*
*
*
*
Figure 4.5: Displacement vs. time for three gauges
Figure 4.6 describes the background (ALE) mesh insertion into the preloaded
Lagrangian model. The option CONSTRAINED_LAGRANGE_IN_SOLID enabled the
Page 43
coupling of the Eulerian meshes across the Lagrangian meshes during load transfer.
Degree of accuracy in ALE/Lagrangian coupling depends on the appropriate refinement
of Eulerian mesh inside the Lagrangian mesh. Smaller Eulerian mesh size increases
computational cost. Therefore, for adequate solution, at least 10 elements in the
Eulerian mesh flow passage were modelled to be nearly the same size as the Lagrangian
mesh.
The ALE/Lagrangian coupling phase is more expensive than the other two phases as it
deals with FSI which is complex to solve analytically. However, the duration for the
blast load transfer from ALE domain to Lagrangian parts is considerably small as
evident from Figure 4.7. It can be observed that the KE of ALE background mesh is
sufficiently reduced to zero in about 180ms. This highlights that ALE background mesh
is only necessary for duration of the load transfer and is ineffective beyond that time
(even if the analysis continues without removing it).
LS-DYNA’s restart features DELETE_PART and DELETE_FSI were used to remove

the redundant ALE background mesh and ALE/Lagrangian coupling respectively.
Deletion of redundant elements from the model reduced the computational time
considerably.
4.2.5 Shockwave propagation
Immediately after the detonation (at t0 = 1500 ms), spherical shock waves propagated in
the air from the charge epicentre which was 50.5cm above the ground. The shock waves
travelled into the ground in the form of hemispherical waves while forming a crater in
the ground, as shown in Figure 4.8. The area of the wave front expanded with the wave
propagation. A set of fringe range on the right side of Figure 4.8 illustrate that the
variation of shock wave pressure in Mbar between positive and negatives values. The
wave front reached the tunnel surface after 7ms of explosion.
Without the tunnel structure, a free field numerical model, as shown in Figure 4.9(a),
was considered to monitor the shock wave propagation in soil. Figure 4.9(b) illustrates
the arrangement of measuring gauges at 3m spacing along two grid directions. Grid 1 is
vertically below the explosive while Grid 2 is inclined by 450 to the ground surface.
Page 44
Figure 4.6: Insertion of ALE background mesh into Lagrangian structure

200
Kinetic energy of ALE
background mesh
150
KE (MJ)
100
50
0
1500 1550 1600 1650
Time (ms)
Figure 4.7: Kinetic energy vs. time plot of ALE background mesh
Page 45
Gauges in Grid 1 are located in the central zone (Yang et al., 2010) to capture the
intensity of shock wave which is mainly due to the compressive waves in the soil.
Gauges in Grid 2 are located at the interface between the central and surface zones
(Yang et al., 2010). Figure 4.10 shows the peak pressure variation with respect to the
equivalent scaled distance of R/W1/3 of the explosive. The propagation and attenuation
of the shock waves in soil are clearly demonstrated by the two plots in Grid 1 and Grid
2. By comparing readings in Grid 1 and Grid 2, it is evident that gauges in Grid 1
experienced slightly higher peak pressures than the corresponding gauges in Grid 2.
This could be due to the fact that the corresponding gauges in Grid 2 are shallower and
located at the interface between the central and surface zones.
Figure 4.8: Propagation of shock wave in dry sand
Empirical equations of the power law in the technical manual TM5-855-1 (Department
of the Army, 1986) can be used to estimate the peak pressure range. In order to derive
equations for the type of soil considered in the simulation, the more appropriate soil
properties of acoustic impedance (pc) and attenuation coefficient (n) are selected as
51.41kPa/ms−1 and 2.625±0.125 (Department of the Army, 1986) respectively, for a
known seismic speed. The following two empirical equations are derived by substituting
the upper and lower limits of n in the TM5-855-1 empirical equations for free field
pressure. These two equations are plotted on a logarithmic scale as two straight lines
shown in Figure 4.10. The slopes of a straight line represent the attenuation coefficients.
Page 46
a) Free filed numerical model b) Measuring gauge points

Figure 4.9: Free field numerical model for wave propagation study
2.75
 R 
P0  f  22   3  ; n  2.75 Equation 4.2
 W
2.50
 R 
P0  f  27   3  ; n  2.50 Equation 4.3
 W
where P0 is the peak pressure in MPa, f is the coupling factor which has a recommended
constant value of f = 0.14 for explosion in air, R is the distance to the explosive in m,
and W is the explosive weight in kg.
Figure 4.10: Comparison of peak pressure distributions along Grid 1 and Grid 2
Page 47
Gauge readings along Grid 1 agreed reasonably well with the estimated peak pressure
values from Equation 4.3. Shallower depth readings along Grid 2 slightly deviated from
both straight lines, but they fell into the straight lines as the depth increased beyond the
scale distance of 1.60m/kg1/3. This discrepancy may be the result of the limitation of
power law for a certain distance range (Yankelevsky et al., 2011). According to
Yankelevsky et al. (2011) some types of soils may not be represented by a linear
relationship on a logarithmic scale, but need to be described by either bi-linear or tri-
linear relationships.
4.2.6 Tunnel response
The tunnel started to respond to the blast when the shock wave hit the tunnel and the
response continued until the shock wave completely attenuated in the soil. Figure 4.11
shows the time history of kinetic energy (KE) of the tunnel response after the explosion.
The tunnel response commenced at t1 = 1507ms by imparting inertia load to the tunnel
from the surrounding soil and the response reached its peak within 4.5ms. The tunnel
dissipated 90% of its KE within 23ms of initial response period.
The process of the pressure wave propagation through the tunnel is presented in Figure
4.12 at different time intervals of the tunnel response. While the pressure wave travelled
along both the longitudinal and the circumferential directions of the tunnel, positive and
negative phases of pressure contours changed with time.
1000
Kinetic energy of tunnel
750 Commencement of tunnel response

Critical period of tunnel response
KE (KJ)
500
250
t1=1507 t2=1530
0
1500 1550 1600 1650
Time (ms)
Figure 4.11: Kinetic energy vs. time plot for tunnel
Page 48
(a) 1507 ms
(b)1510 ms
(c) 1513 ms
Figure 4.12: Pressure wave propagation through the tunnel
Page 49
(d)1520 ms
(e) 1525 ms
(f) 1530 ms
Figure 4.12: Pressure wave propagation through the tunnel
Two gauges Gauge 1 and Gauge 2 were introduced along the surface of tunnel crown to
record the axial and circumferential strains in the numerical analysis. Gauge 1 was
located directly below the explosive center on the tunnel surface while Gauge 2 was
placed 10.6m away from Gauge 1 as shown in Figure 4.13. Gauge 3 was placed on the
springline of the midspan to monitor the circumferential strain in the numerical
simulation.
Page 50
(a) A half-symmetrical prototype tunnel. (b) A quarter-symmetrical numerical model.
Figure 4.13: Arrangement of measuring gauges on the tunnel exterior surface
Studies were performed initially without considering the mesh alignment on the curved
contact interface between the soil and tunnel. Interlocking due to peaks and valleys on
the interface governed the tunnel response rather than the frictional forces. Interlocking
of contact surfaces constrained the tunnel response in circumferential direction as in a
perfectly bonded interface. Results reported are not presented here, but the important
finding has been used in this numerical simulation to avoid the interlocking of elements.
Figure 4.14 shows the time histories of the axial and circumferential strains at the above
mentioned Gauge positions. Both plots show that the tunnel response decreases with
increase in distance from the explosive, in both axial and circumferential directions, as
the intensity of shock wave decays along the tunnel. The plots further indicate that all
deformations at gauge points travelled through a peak response cycle over a period of
23ms (1507ms - 1530ms). In this period, the tunnel lost about 90% of peak KE.
Therefore, it would be appropriate to say that, this might be the critical period of tunnel
response.
900 1000
450 0
-1000
0
Mircrostrain
-2000
Mircostrain
-450
-3000
-900
-4000 Gauge 1 circumferential strain
-1350 Gauge 1 axial strain Gauge 2 circumferential strain
-5000
Gauge 2 axial strain Gauge 3 circumferential strain
-1800 -6000
1500 1550 1600 1650 1500 1550 1600 1650
Time (ms) Time (ms)
(a) Axial strains (b) Circumferential strains
Figure 4.14: Axial and circumferential strain histories
Page 51
Furthermore, it is observed in Figure 4.14(a) that most of the axial deformation is

recoverable although the response continued with a noticeable fluctuation in axial strain.
It is also evident that the circumferential deformation, obtained from gauges 2 & 3,
recovered as shown in Figure 4.14(b). However, from the circumferential residual strain
at Gauge 1, it is clear that the tunnel suffered permanent deformation due to the surface
explosion. This state of strains is reasonable because the tunnel has an infinite surface
area in the longitudinal direction which constrained its movement due to the grip in the
contact surface between the tunnel and the soil. In the transverse direction, excessive
compressive stress from the blast load changed its deformation mode to a horizontal
ovalisation. The deformation mode depends on degree of flexibility of the tunnel
structure. Due to this, some researchers (Gui & Chien, 2006; Shin et al., 2011) have
treated the tunnel response using two dimensional plane-strain analysis by omitting the
axial deformation.
The higher value of the circumferential strain at Gauge 1 compared to that at Gauge 3
highlight that the upper part of tunnel above the springline is vulnerable to the
explosion. This suggests possible blast mitigation of the tunnel by providing a
protective cover on the top half of the tunnel and if tension is evident providing
appropriate tensile reinforcement (where necessary) for non-metallic tunnels. Figure
4.15(a) demonstrates how the axial strain varies along the surface of the tunnel crown
when Gauge 1 responded to the peak axial strain (deformation) at 1513ms. During this
time, as illustrated in Figure 4.12(c), the shock wave front was slightly less than 5.0m
away from the midspan of the tunnel. These two figures show that Gauge 1 exhibited
the peak axial tensile strain after the shock wave travelled through the target.
Propagating shock wave compressed the portion of the tunnel ahead of the shock wave.
Furthermore, it displayed a decrease of the axial compressive strain followed by a peak
axial compressive strain at 5.0m from the midspan. The compressive zone extended
over 5.0m, beyond which there were no significant effect in axial strain. Similarly,
Figure 4.15(b) describes the distribution of the circumferential strain along the arc
distance away from the top of the tunnel mid-span when the same target was subjected
to peak deformation at 1510ms. In this time, propagating shock waves were within an
angle of 450 from the crown, which can be seen from Figure 4.12(b). The
circumferential tensile strain rapidly decreased to a small magnitude at a position of
Page 52
32.50. Due to the globalized vertical in-plane response, the circumferential strain was in
tension ahead of the shock wave and it extended even below the springline. Although
most surfaces were in tension, the region close to the invert level was slightly in
compression.
(a) Axial strain distribution along the (b) Circumferential strain distribution
crown at t=1513ms along the arc angle at t=1510ms
Figure 4.15: Deformation of the tunnel during the peak response
Comparisons of the tunnel response with available experimental results of a centrifuge

test (De, 2012; De & Zimmie, 2011) were carried out. Figure 4.16 compares the
numerical strain history at Gauge 2 with that from the centrifuge test (De & Zimmie,
2011). The proximity of the two curves, with closely matching peaks, indicates a
reasonably good correlation between the numerical and the experimental results and
provides adequate confidence in the present modelling techniques. Furthermore, in
Figure 4.17, magnitudes of peak axial and circumferential strains at Gauge 1 and Gauge
3 respectively, are plotted against the equivalent scaled distance by varying the
explosive weight and the soil cover. These strains obtained from the LS-DYNA analysis
are compared with those from the centrifuge test (De, 2012) (measured for a specific
value of the scaled distance). There is a small discrepancy in the axial strains in Figure
4.17(a), as was also observed in the numerical simulations reported in (De, 2012). This
could be due to the uncertainties in the end condition of the pipe and the limitations in
its movement in the experiment. A real buried tunnel in a soil medium has no movement
restrictions, but the experimental model had restrictions from the four sides of the test-
bucket containing the soil. These constrained boundaries may restrict the combined
motion of soil and the pipe.
Page 53
750
Axial strain at Gauge 2 (Centrifuge)
500 Axial strain at Gauge 2 (Numerical)
Microstrain
250
-250
-500
1500 1550 1600 1650
Time elapsed (ms)
Figure 4.16: Comparison of numerical and experimental axial strains
The circumferential strains at Gauge 3 obtained from the numerical simulations were
compared with the experimental results at the two points CS1 & CS2 as illustrated in
Figure 4.17(b). These two points are symmetrically located on either side of the
springline at the midspan in the experimental setup. The comparisons show that the
numerical best-fit line lies between the experimental values at CS1 & CS2. Under
symmetric condition of blast loading and the symmetric locations of gauges CS1 &
CS2, the peak strains measured in the experiment should be the same at both locations.
However, this was not the case and a variation of 280 microstrain was noted. This lack
of symmetry and variation in results at CS1 & CS2 could be due to possible movement
of the explosive from its initial position inside the test-bucket before the blast occurred
or due to a rotation of the copper pipe about its axis during the placement and sand
filling into the test-bucket. In centrifuge tests, controlling measures are very difficult to
implement, particularly, in blast loading.
2000
2000
Axial strain at top of midspan (Centrifuge)
Circumferential strain at Gauge 3 (Numerical)
Axial strain at Guage 1 (Numerical)
Circumferential strain @ CS1 & CS2 (Centrifuge)
Peak mircostrain
Peak mircostrain
1500 1500
AS1 CS2
1000 CS1
1000
500 500
0.325 0.375 0.425 0.475 0.525 0.575 0.625 0.64 0.74 0.84 0.94 1.04 1.14
Scaled distance (m/kg1/3) Scaled distance (m/kg1/3)
(a) Peak axial strains (b) Peak circumferential strains
Figure 4.17: Comparison of peak axial and circumferential strains
Page 54
These observations of different circumferential strains at two initially symmetric

locations demonstrate the possible movement of the tunnel or explosive during testing
and resulting in the actual value of the peak axial strain (at the top of the tunnel at
midspan) to be more than the value measured in the experiment. This justifies the higher
value of the numerical axial strain compared to the measured value.
4.2.7 Summary
Coupled FSI in ALE technique and material models to analyse the tunnel response to
surface blast were validated using previous experimental data of centrifuge test. The
main finding of this study can be summarised as follows:
• The axial and circumferential deformations decrease with increasing distance

from the explosive, as expected.
• The circumferential strain histories indicate that the upper part of tunnel above
the spring-line is more vulnerable to the explosion than the lower part.
• The validated numerical results provide confidence in the modelling techniques

used in this study and allow developing rational procedures for predicting the
blast response of buried cylindrical structures in dry sand.
4.3 VALIDATION OF SPH SIMULATIONS USING CENTRIFUGE TEST

RESULTS
The experiment described in Section 4.2.1 was further simulated using the coupled
SPH-FE approach using the same material models as described in the earlier section.
As shown in Figure 4.18(b), a portion of the soil experiencing large deformations and
the explosive were modelled with SPH particles while the rest of the geometry was
modelled with Lagrangian meshes. No attempt was made to model the interior volume
of the tunnel. The surrounding outside space of the explosive was assumed to be a
vacuum which ignored the later interaction process between the explosion-produced gas
and surrounding atmosphere.
Page 55
(a) A quarter symmetrical model. (b) Insertion of explosive.
Figure 4.18: Coupled SPH-FE method.
First, a number of models were developed to determine the optimal size of a box filled
with SPH soil particles. For a quarter symmetric model, the optimum size of the box
was determined as 350cm x 350cm x 276cm. The SPH particles were 10cm in diameter
with equal inter-particle distance of 10cm. Beyond the SPH region, the soil and the
tunnel were replicated with the same mesh resolution as described in the ALE
simulation.
The coupling interaction between the SPH and Lagrange FEM is formed by the penalty
based contact CONTACT_AUTOMATIC_NODES_TO_SURFACE. Though the
boundary conditions were identical to the ALE model, a special symmetry boundary
BOUNDARY_SPH_SYMMETRY_PLANE was applied to those SPH particles at the
symmetry planes. The simulation was considered in two stages, stress initialization and
blast analysis. The model shown in Figure 4.18(a) was used for stress initialization with
a time-dependent mass damping. Upon initializing the model, as illustrated in Figure
4.18(b), the explosive SPH particles were added into the preloaded model for the blast
analysis.
Page 56
4.3.2 Shockwave propagation
Figure 4.19: Explosive-soil interaction
As illustrated in Figure 4.19, during the explosion process, the quick interaction of the
explosive with the neighbouring soil SPH particles implies distortion, where those
explosive SPH particles are dispersed by the flow of the expanding explosion. The blast
produced shock waves transmitted hemispherically through the soil and propagate at a
higher rate than the crater formation as shown in Figure 4.20. The shock waves reached
the tunnel crown after 7 ms of the detonation. When SPH particles interacted with solid
elements, the SPH-FE coupling enabled the stress transfer at the interface without
penetration of SPH particles.
Figure 4.20: Shock wave propagation in soil
Page 57
4.3.3 Tunnel response
The tunnel response commenced when the shock wave struck the tunnel surface. Figure
4.21 illustrates the interface pressure variation along the tunnel, as the shock waves
progress through the soil-structure interface. The interface between the structure and the
surrounding soil was simulated using thin-layer elements. Four elements were
considered along the crown of the tunnel interface and Figure 4.22 illustrates
corresponding interface pressure distributions. The peak interface pressures are high in
the immediate area of the explosive. The element EL#32 experienced the highest peak
pressure which is more than 600 times the geo-static stress due to the overburden soil.
The peak pressure line in Figure 4.22 illustrates that the peak pressure drop is very high
in the immediate region whereas the drop is insignificant beyond the region of element
EL# 895 which is 4.0m from the mid-span.
(a) t = 1507ms
(b) t = 1510ms
Figure 4.21: Interface pressure contours
Page 58
(c) t = 1514ms
(d) t = 1536ms
50
EL# 32
EL# 3736
Interface pressure, MPa
40
EL# 895
EL# 1097
30 Peak pressure
20
10
0
1500 1510 1520 1530 1540 1550 1560
Elapsed time, ms
Figure 4.22: Interface pressure vs. time
4.3.4 Comparison of strain output
Figure 4.23 shows a comparison of axial and circumferential strain histories at two
locations (Gauge 1 and Gauge 3) obtained from ALE and SPH methods. Figure 4.23(a)
Page 59
shows that the peak axial strain in the SPH simulation is about 7 per cent more than that
in the ALE simulation. General trend of circumferential strain histories are similar as
shown in Figure 4.23(b). However, the peak values from the SPH simulations are
slightly more than those from the ALE simulations in both the positive and negative
phases. After the first peak, there were inconsistent peaks and valleys in the positive
phase. This could have resulted from the existing air background mesh in the ALE
simulation. Both curves display some fluctuations in strains after 1575ms. Repeated
reflection of shock waves caused this fluctuation which continued until the shock waves
completely attenuated in the soil.
300 1200
0 900
600
-300
Mircrostrain
Mircostrain 300
-600 0
-900 -300
-600
-1200
-900 ALE method
ALE method
-1500 -1200
SPH method SPH method
-1800 -1500
1500 1550 1600 1650 1500 1550 1600 1650
Time elapsed, ms Time elapsed, ms
(a) Axial strain at Gauge 1 (b) Circumferential strain at Gauge 3
Figure 4.23: Comparison of axial and circumferential strains.
Figure 4.24 shows the comparison of the numerical strain histories at Gauge 2 with the
results from the centrifuge test. The ALE predictions are closer to the experimental
results compared to the SPH predictions which were somewhat conservative across
most parts of the duration. This could be due to the assumption in the SPH simulation
that the surrounding medium of the explosive SPH particles was considered to be a
vacuum. This assumption ignored the importance of the interaction of the SPH
explosive particles with the air and hence the energy imparted from the explosive into
the soil was significantly larger in the SPH simulations than that in the ALE
simulations.
Page 60
750
Centrifuge test
500 ALE method
SPH method
Microstrain
250
-250
-500
1500 1550 1600 1650
Time elapsed, ms
Figure 4.24: Comparison of axial strain from Gauge 2 between simulation and
experiment.
Figure 4.25 compares the magnitudes of peak axial strains at Gauge 1 and the peak
circumferential strain at Gauge 3 with respect to the equivalent scaled distance of
R/W1/3 of the explosive. Results for these strains obtained from both ALE and SPH
simulations are compared with those from the test data and overall the results from both
simulations agree reasonably well with each other and with the test result. Figure
4.25(a) shows that the SPH prediction is slightly (about 7%) higher than the ALE
prediction which is closer to the test data. There is however a small discrepancy in the
peak axial strain between the test data and the ALE simulation due to the confinement
of soil in the test-bucket.
2000 2000
Centrifuge test Centrifuge test
1750 SPH method 1750 SPH method
ALE method
Peak mircostrain
1500 ALE method

1500
Peak mircostrain
CS2
1250 AS1 1250
1000 CS1
1000
750 750
500 500
0.325 0.375 0.425 0.475 0.525 0.575 0.625 0.64 0.74 0.84 0.94 1.04 1.14
Scaled distance, m/kg1/3 Scaled distance, m/kg1/3
(a) Comparison of peak axial strains (b) Comparison of peak circumferential strains
Figure 4.25: Comparison of peak axial and circumferential strains.
The circumferential strains at Gauge 3 obtained from the numerical simulations are
compared with the test data, as illustrated in Figure 4.25(b). The comparisons show that
Page 61
both numerical best-fit lines are very near and they fall within the range of test data at
CS1 and CS2. The test data CS1 and CS2 should be the same for this case with
symmetry, but the magnitude of CS2 is 25% more than that of CS1. This could also be
another factor that affected the experimental peak axial strain at gauge AS2.
4.3.5 Comparison of computational efficiency
Computer simulations were conducted using nine parallel processors on a standard high
performance workstation. Each simulation involved two stages of stress initialization
and blast analysis. Table 4.5 shows the comparison of the number of elements and the
computational time of both ALE and SPH simulations. The ALE simulation is much
faster for the stress initialization than the SPH as the ALE simulation used the
Lagrangian structural mesh alone. For agreed mesh resolution, the ALE simulation took
slightly more CPU time than the SPH to simulate the blast problem for a period of
180ms.
Table 4.5: Comparison of computational efficiency
ALE simulation SPH simulation
Initialization Blast-analysis Initialization Blast-analysis
Nos. of Lagrangian
220255 220255 213472 213472
elements
Nos. of Eulerian
- 307530 - -
elements
Nos. of SPH
- - 34300 34438
particles
Simulation duration
1500 180 1500 180
(ms)
Timestep (  s ) 1.06e+01 5.89e+00 1.06e+01 4.72e+00
Total CPU time
(hr:min:sec) 10:51:51 164:10:11 96:31:49 142:54:41
4.3.6 Summary
In this section, coupled SPH-FE techniques have been developed and validated for
treating the blast response of underground tunnel subjected to surface blast. The coupled
Page 62
FSI in ALE technique was compared with the coupled SPH-FE technique using the
same centrifuge test data. The comparison shows that, in terms of accuracy and
computational efficiency, the coupled FSI in ALE outweighed the coupled SPH-FE
technique for dealing with above ground explosion problems. The coupled FSI in ALE
provides a comprehensive solution in a more efficient manner by handling the
numerical model through two different phases which are stress initialization and blast
analysis.
4.4 VALIDATION OF ALE SIMULATIONS USING CONWEB TEST

RESULTS
This section describes simulations that replicate the CONWEB test (Hayes, 1989) using
ALE method in order to validate the response of reinforced concrete slab bolted to the
reaction structure.
4.4.1 Description of CONWEB test
A test described in the CONWEB test series (Hayes, 1989) was simulated using the
coupled FSI approach in ALE formulation. In this test series, Hayes (1989) considered
two low seismic velocity backfill soil materials: a low shear strength reconstituted clay
and a high shear strength compacted concrete sand. This simulation utilizes a test with
the reconstituted clay which is nearly saturated.
As illustrated in Figure 4.26, the specimen was prepared with a reusable reaction
structure which was first placed in an excavated test pit (6.1m x 6.1m x 2.7m). A test
slab was bolted to the reaction structure by means of bolts with 22.86 cm spacing along
the base and the roof of the reaction structure. The reconstituted clay was backfilled into
the pit with a great attention to ensure the consistency of soil properties throughout the
backfilling. During the backfilling, a series of interface pressure gauges were fixed on
the right side of the charge to monitor the free-field motion of the soil as shown in
Figure 4.26. The test slab was also instrumented with interface stress gauges on the
exterior face to monitor the interface pressure as well as accelerometers on the interior
face to measure horizontal motion. A 7kg C-4 cased cylindrical explosive was placed
152.4cm away from the center of the test slab at a depth of 152.4cm.
Page 63
Figure 4.26: Experimental setup of CONWEB test (elevation) (Hayes, 1989)
The test slab was 4.57m in length, 1.65m high and 10.9cm thick, reinforced in both
vertical and horizontal directions with adequate shear links. The reinforcement and
concrete cover details can be found in Hayes (1989). The density and average
unconfined compressive strength of the test slab concrete were 2.24g/cm3 and 42.0MPa
respectively. The reaction structure was a 28cm thick heavily reinforced concrete box
open on one side to mount the test slab. The exposed surface of the reaction structure
was cast with a 16mm thick steel plate which provided a hard smooth bearing interface
and performed as a protective layer against the blast impact.
The numerical model composed of mainly four components consisting of a buried

explosive, soil, air and the test structure. Symmetric modelling capabilities played
important roles in this simulation by reducing the computational demand in terms of the
problem size. Therefore, a quarter of each of the soil and explosive and a half the
structure were modelled as shown in Figure 4.27.
Page 64
Figure 4.27: A half symmetry numerical model for ALE simulation
The reaction structure and the test slab were modelled using the eight-node hexagonal
solid elements with Lagrangian meshes. The soil and air within the interior volume of
the structure were modelled using the eight-node hexagonal solid elements with
Eulerian meshes. The cylindrical explosive was defined into the soil mesh using
INITIAL_VOLUME_FRACTION_GEOMETRY by specifying its radius, height and
detonation point. A series of mesh sensitivity studies provided an appropriate mesh
refinement to capture the detonation process and subsequent response of the structure.
The soil was refined with a gradual increase in mesh size in both X and Y directions
from the explosive center. The size of the smallest element in the central part of the
explosive was 2.25cm x 2.25cm x 3.0cm. The interface between the backfill soil and the
in-situ clay were modelled with merged nodes. Since material properties were not
available for the in-situ clay, it was assumed to be the same as the backfill material.
In the simulation, a fully coupled FSI approach combining both Lagrangian and
Eulerian solvers was adopted to allow for the incorporation of the essential processes
using CONSTRAINED_LAGRANGE_IN_SOLID. Nodes in the symmetry boundaries
of XZ and YZ planes were constrained in their normal directions. The bottom of the
mesh was modelled as fixed in all directions. Along the infinite boundary of the entire
computational domain, non-reflecting boundaries were set as a flow out boundary to
avoid shock wave reflection.
Page 65
(a) Reaction structure (b) Steel plate/bolts
(c) Test slab (d) Reinforcement
Figure 4.28: Test structure components
Figure 4.28 illustrates half the test structure components which were modelled with
appropriate boundary conditions at the symmetry planes. The reinforcement details
described in the CONWEB test (Hayes, 1989) was simulated in the model. The
reinforcement steel and bolts were modelled using Hughes-Liu beam elements (LSTC,
2007) with cross sectional integration. The test slab was modelled with a very fine mesh
to achieve adequate accuracy in the deflection. A minimum of two hexahedral elements
were specified for the reinforcement cover. There were 72960 solid elements in the test
slab attached to the steel plate with 20 bolts. The simulation was free from pretension in
bolts. In the test slab, reinforcing beam elements and bolts were modelled as discrete
elements immersed in the slab meshes using the CONSTRAINED_LAGRANGE_IN _
SOLID coupling. In the steel plate, the same nodes were shared for a perfect bond
Page 66
between the steel plate and bolts. The CONTACT_AUTOMATIC_SINGLE_

SURFACE contact was also used to define the interface between the test slab and the
steel plate.
As the steel plate was cast integrally together with the reaction structure, the perfectly
bonded ‘no slip’ condition was simulated between the steel plate and reaction structure.
The material properties for the steel plate were specified by Bessette (2004) and the
steel was modelled with MAT_PLASTIC_KINEMATIC material model. The reaction
structure was a heavily reinforced thick concrete structure (Hayes, 1989). Since there
were no reinforcement details available, for simplicity, the reaction structure was
modelled as a smeared concrete as also adopted by others (Yang et al., 2010).
4.4.3 Material models and parameters
The material models and relevant material properties for the air and explosive were
described in Chapter 3.5. The soil properties such as density, specific gravity and water
content are shown in Table 4.6. The shear and bulk moduli were evaluated based on the
density, seismic velocity and Poisson’s ratio (Bessette, 2004). The cohesion and friction
angle were selected based on the modified material properties described by Baylor
(1992). Magnitudes of PWD1 were estimated based on best fit material analysis using
the free-field blast simulation reported in CONWEB test series.
Table 4.6: Material properties of backfill materials
Reconstituted clay Compacted concrete

Parameters
sand
Density (g/cm3) 1.96 1.865

Specific gravity 2.71 2.70
Shear modulus (MPa) 20.0 70.2
Bulk modulus (MPa) 193.3 117.0
Cohesion (MPa) 2.275e-02 3.723e-02
Friction angle 22o 40o
Moisture content (%) 23.3 5.0
PWD1 (MPa-1) 4.810e-03 2.520e-02
PWD2 (MPa-1) 4.974e-02 1.152e-02
Page 67
Table 4.7 shows the material parameters of concrete adopted in the simulation of
CONWEB tests. The fracture energy for the concrete containing limestone aggregate,
used in the experiment, was considered as 70N/m (Darwin et al., 2001). The material
properties of the steel are described in Table 4.8.
Table 4.7: Material properties of concrete
Parameter Test slab Test slab

(Test with clay) (Test with sand)
Density (g/cm3) 2.24 2.24

Unconfined compressive strength (MPa) 42.0 40.4
Poisson’s ratio 0.19 0.19
Maximum aggregate size (mm) 9.5 9.5
Table 4.8: Material properties of steel reinforcement
Density Es σ Etan C
v β P
(gcm-3) (GPa) (MPa) (GPa) (S-1)
Principal steel 7.85 200 0.3 465 2.0 0 1080 5.48

Temperature steel 7.85 200 0.3 563 2.0 0 9650 5.50
Shear steel 7.85 200 0.3 505 2.0 0 2150 5.49
4.4.4 Free-field simulation
Figure 4.29 compares the peak pressure attenuations obtained from the free-field
simulation with those from the CONWEB free-field test (Hayes, 1989) for saturated
reconstituted clay. For a magnitude of PWD1 (= 6.30GPa-1), numerical results are only
marginally lower than those from the experiment in the immediate vicinity of the charge
(152.4cm away from the charge center). As distance increases form the charge, the
numerical values become slightly higher than the experimental values. This could be
due to the confinement effect of the charge. The casing of the charge was not considered
in the simulation which used a bare charge. The proximity of the two curves, however,
demonstrates a reasonable agreement between the present numerical results and the
measured free-field peak pressures. It can be seen that the shock pressure attenuation for
the reconstituted clay was not aligned with a linear power law on a logarithmic scale. As
Page 68
highlighted by Yankelevsky et al. (2011), behaviour of some types of soil may be

represented by either bi-linear or tri-linear trend lines. As the numerically obtained free-
field response reasonably agreed with that from the experiment, it provided a means to
evaluate the best fit PWD1 constant for the saturated reconstituted clay as 6.30GPa-1.
100
ALE method
CONWEB (Clay)
Peak pressure, MPa
10
1
0.1 1 10
Scaled distance, m/kg1/3
Figure 4.29: Comparison of peak pressures
4.4.5 Test slab response
The response of the test slab under buried blast was investigated. Figure 4.30 shows the
progress of the slab deformation as the shock wave travels through the structure over
20ms of duration. After igniting the explosive at 300ms, blast induced shock waves
travelled through the soil and compressed the test slab surface. The shock wave front
impacted the exterior surface of the slab at 303.1ms.
(a) t = 300ms (b) t = 303.1ms
Figure 4.30: Progress of test slab deformation
Page 69
(c) t = 303.5ms (d) t = 304ms
(e) t = 310ms (f) t = 320ms
Figure 4.30: Progress of test slab deformation

The slab displayed some kind of breaching failure at the center and the failure extended
away from the center, as the shock wave progressed through the slab. Before the
reinforcement bars near the center of the slab failed, those at the top and bottom support
edges broke due to both tensile failure and shear failure mechanisms. Figure 4.31
displays the broken reinforcement at the end of the simulation. The observed damages
in the reinforcement are similar to the experiment as described by Hayes (1989). Along
the exterior edges of the support, there was a series of separation between the slab and
steel plate due to both localized rotation and deformation of bolts.
Figure 4.31: Broken slab reinforcement
Page 70
Crack opening capabilities in the concrete material model illustrate the failure or
damage in the slab. Cracks first appeared in both interior and exterior surfaces after 4ms
of the explosion as shown in Figures 4.32(a) and 4.32(d). These figures show that most
of the damage of the test slab resulted within a very short period of time from the
excessive impact of blast pressure. During the early stages of the simulation, cracks
were vertical and parallel to the principal steel. Later on a number of diagonal cracks
initiated on the interior slab surface. At later stages, large number of diagonal cracks
emanated from the bolt points which act as a stress initiating points from the
deformation mechanism. The interior and exterior views of the slab damages, cracks
patterns and deformed shapes at the end of the simulation are similar to those observed
in the CONWEB test (Hayes, 1989).
(a) t = 304ms (Exterior surface) (b) t = 310ms (Exterior surface)
(c) t = 320ms (Exterior surface) (d) t = 304ms (Interior surface)
(e) t = 310ms (Interior surface) (f) t = 320ms (Interior surface)
Figure 4.32: Propagation of cracks on test slab surfaces
Page 71
In order to compare the displacement history of the slab with the results from the
CONWEB test, six gauges mounted on the slab surface are considered. Figure 4.33
describes the interior view of the slab and gauges. Gauge AHS-2 is close to the center of
the slab and gauges AHS-1, AHS-0 and AHS-3 are vertically aligned with ASH-2 on
the center line of the slab. Gauges AHS-1, AHS-5 and AHS-6 are on a horizontal line
which is about 19cm below the slab center.
Figure 4.33: Instrumentation plan of test slab (Hayes, 1989).
Figure 4.34 compares the present results for displacement histories at the six gauges
with those from the CONWEB test (Hayes, 1989). It shows that the numerical
simulations of the displacements compared well with the corresponding experimental
displacements and are within the deviation limits with few exceptions. At first, a small
lag in the time of arrival of shock wave was observed in the simulation. This could be
due to the small deviation in soil material properties evaluated based on several
assumptions and empirical formulae. Secondly, comparisons clearly show that
displacement-time responses are slightly steeper during the first half of the simulation.
The displacement-time response is dependent on the Young’s modulus of the concrete
used in the CONWEB test (Hayes, 1989) where the concrete was made with a limestone
aggregate. Since there are no records of stress-strain relationship available for the
concrete containing limestone aggregate, Young’s modulus used in the present
simulation was evaluated based on Equation 3.17. As highlighted by Oluokun et al.
(1991), this equation predicts less accurate values of the Young’s modulus.
Along the transverse direction at gauges AHS-1 and AHS-5 numerical predictions are
reasonably close to the experimental values, but at AHS-6 they seem to diverge with
Page 72
time. The reason for this discrepancy is unclear and it was also observed in the
numerical simulation reported by Bessette (2004). Numerical predictions are also
reasonably close to the experimental values along the vertical center line of the slab,
except at AHS-2 where there is over-prediction of the results. This was also observed by
Bessette (2004). By comparing experimental displacements from surrounding gauges,
this variation at AHS-2 could be due to possible disturbance of the gauge caused by
excessive cracking near the center of the slab. Overall, the numerical displacements are
reasonably close to the experimental values at many locations. The numerical results of
reinforcement bar failure, crack patterns in the slab and slab deflections agree
reasonably well with those from the experiment and provide adequate confidence in the
present modelling techniques.
30 30
25 25
20 20
Displacement, cm
Displacement, cm
15 15
10 10
5 AHS-0 (Experiment) 5 AHS-1 (Experiment)
AHS-0 (Numerical) AHS-1 (Numerical)

0 0
300 305 310 315 320 300 305 310 315 320
Time elapsed,ms Time elapsed, ms
(a) AHS-0 (b) AHS- 1
30 30
AHS-2 (Numerical) AHS-3 (Experiment)
25 AHS-2 (Experiment) 25 AHS-3 (Numerical)
20 20
Displacement, cm
Displacement, cm
15 15
10 10
5 5
0 0
300 305 310 315 320 300 305 310 315 320
Time elapsed, ms Time elaspsed, ms
(c) AHS-2 (d) AHS-3

Figure 4.34: Comparison of displacements
Page 73
30
30
AHS-6 (Experiment)
25 AHS-6 (Numerical)
25
20 20
Displacement, cm
Displacement, cm
15 15
10 10
5 AHS-5 (Experiment) 5
AHS-5 (Numerical)
0 0
300 305 310 315 320 300 305 310 315 320
(e) AHS-5 (f) AHS-6
4.4.6 Summary
In this section, the coupled FSI in ALE technique and material models to analyse the
underground structural response to belowground explosion was validated using
CONWEB test. The result shows that the simulation agreed reasonably well with the
experiment in the aspects of reinforcement bar failure, crack patterns in the slab and
slab deflections.
4.5 VALIDATION OF SPH SIMULATIONS USING CONWEB TEST

RESULTS
This section simulated CONWEB test (Hayes, 1989) using SPH method in order to
validate the response of the same reinforced concrete structure buried in partially
saturated sand.
Similar to the model in Section 4.4.2, a quarter symmetric model was utilised as shown
in Figure 4.35. The explosive and near field soil media were modelled using SPH
particles while the far field soil and the test structure were modelled using the eight-
node hexagonal solid Lagrangian elements.
Page 74
Figure 4.35: A half symmetrical numerical model for SPH simulation
All SPH particles within a box of 46.5cm x 46.5cm x 127.5cm were modelled with an
equal inter-particle distance of 1.5cm. A series of mesh sensitivity studies provided an
appropriate mesh refinement to capture the detonation process and subsequent response
of the structure. The soil was refined with a gradual increase in mesh size in both X and
Y directions beyond the SPH region. The size of the smallest element adjoining the SPH
region was 3.7cm cube. Backfill soil elements were merged with the in-situ clay soil
elements at the interface. Since material properties were not available for the in-situ
clay, it was considered to be the same as the backfill reconstituted clay. This will avoid
any problems with the wave refraction/reflection at the interface as both backfill
materials have the same seismic velocity.
The same modelling procedure as described in Section 4.4.2 was used to model the test
structure. The interface between the test structure and the backfill soil was modelled
using CONTACT_AUTOMATIC_SURFACE_TO_SURFACE contact. Section 4.4.3
described the soil properties for the backfill compacted sand used in this simulation.
4.5.2 Free-field simulation
Free-field analysis was performed to evaluate a best fit material parameter of PWD1. In
this numerical simulation, the charge was modelled using SPH particles without the
casing as SPH is incapable of modelling shell elements. Figure 4.36 shows the
comparison of peak pressure from the present numerical study with those from the
CONWEB free-field test (Hayes, 1989) for partially saturated sand. It can be seen that
Page 75
the best fit line of the sand has a perfect match with the measured free-field peak
pressure and it also complies with a linear power law on a logarithmic scale. Hence the
free-field simulation agreed reasonably well with the test results and provides a means
to evaluate the best fit PWD1 parameter for the partially saturated sand as 25.0GPa-1.
SPH method
CONWEB (Sand)
10
Peak pressure, MPa
1
0.1 1 10
Scaled distance, m/kg1/3
Figure 4.36: Comparison of peak pressures
4.5.3 Test slab response
The structure began to respond when the shock waves reached the soil structure
interface. Figure 4.37 illustrates the variation of interface pressure contours at the
interface of the soil structure.
(a) t = 500ms (b) t = 504ms
Figure4.37: Interface pressure contours
Page 76
(c) t = 504.5ms (d) t =5 05ms
(e) t = 505.5ms (f) t = 506ms
The peak pressure which results from the interaction of the structure and the
surrounding soil is the dominant parameter that influences the response of the test slab.
This peak interface pressure was therefore considered in the present study. Figure 4.38
compares the peak interface pressures with those obtained from the CONWEB test. The
peak pressures evaluated by using the Winfrith concrete model agree better with those
from the experiment at many locations.
7
SPH method
Peak interface pressure, MPa
6
CONWEB (Sand)
5
0
IF1 IF2 IF3 IF6 IF8
Interface stress gauges
(a) Exterior surface of test slab (b) Interface peak pressure
Figure 4.38: Comparison of peak interface pressures
Page 77
The response the test slab under blast load was investigated. The simulation showed a
kind of breaching failure at the center of the test slab and the failure extended away
from the center. Shear failure was observed at the end support along the floor and roof
edges of the reaction structure. A separation between the test slab and steel plate was
also observed due to both localized rotation along the support edges and straining of
bolts. Figure 4.39 illustrates the formation of cracks in the test slab. During early stages
of the simulation, cracks were vertical and parallel to the principal steel. At later stages
of the simulation, number of diagonal cracks emanated from the bolt locations.
Comparisons of the interior view of the test slab damage (Hayes, 1989) and numerical
simulations illustrate that the damage patterns at the interior face of the slab are similar
in terms of qualitative assessment.
(a) t = 504ms (b) t = 508ms
(c) t = 520ms
Figure 4.39: Propagation of cracks on interior surface
Finally, the displacement histories of six gauges are compared as shown in Figure 4.40.
The slight lags in the time-of-arrival are similar to those observed in the earlier
simulation. Though the deflections obtained from the simulation were less than those
observed in the experiment, the predictions of final displacements were closer to the
experimental results at many locations. Overall, the experimental results such as the
Page 78
interface pressure, crack propagation and deflection obtained from for this configuration
agrees well with the chosen concrete model.
6 6
5 5
Displacement, cm
Displacement, cm
4
3 3
2 2
AHS-0 (Experiment)
1 1 AHS-1 (Experiment)
AHS-0 (Numerical) AHS-1 (Numerical)
0 0
500 505 510 515 520 500 505 510 515 520
(a) AHS-0 (b) AHS-1

6 6
AHS-3 (Experiment)
5 5 AHS-3 (Numerical)
4 4
Displacement, cm
Displacement, cm
3 3
2 2
1 AHS-2 (Experiment) 1
AHS-2 (Numerical)
0 0
500 505 510 515 520 500 505 510 515 520
(c) AHS-2 (d) AHS-3
Page 79
6 6
AHS-5 (Experiment) AHS-6 (Experiment)
Displacement, cm 5 AHS-5 (Numerical) 5 AHS-6 (Numerical)
4 4
Displacement, cm
3 3
2 2
1 1
0 0
500 505 510 515 520 500 505 510 515 520
(e) AHS-5 (f) AHS-6

4.5.4 Summary
In this section, the coupled SPH-FE technique and material models to analyse the
underground structural response to belowground explosion were validated using the
CONWEB test. The results show that the simulation agreed reasonably well with the
experiment in the aspects of interface peak pressure and crack patterns in the slab,
although the simulation slightly under-predicted the displacement response at few
locations.
4.6 CHAPTER SUMMARY
This chapter demonstrates the applicability of the state of the art of numerical
techniques to accurately predict the blast-induced buried structure response. The
validation of numerical techniques for the prediction of the tunnel response to
aboveground (surface) and belowground explosions has been presented in this chapter.
Two different numerical techniques (i) coupled FSI in ALE and (ii) coupled SPH-FE
techniques have been developed and applied for treating the blast response of buried
structures using the software LS-DYNA. The modelling techniques were compared for
the surface-blast induced tunnel response in dry soil and validated using results from the
centrifuge test (De, 2012; De et al., 2013). For aboveground surface explosion and
subsequent response of buried tunnels, the coupled FSI in ALE outweighed the coupled
SPH-FE technique in terms of accuracy and computational efficiency. Validation of
Page 80
reinforced concrete model and contact between two structural elements connected by
bolts was performed using CONWEB tests (Hayes, 1989) in both numerical techniques.
The results confirmed the validity of the ALE modelling techniques, reliability in
employing material models and their applicability to extend the study to treat the blast
response of segmental buried tunnels. The remaining chapters of the thesis employed
the ALE modelling techniques to investigate the influence of important parameters on
the blast response of segmental tunnels.
Page 81
Page 82
Chapter 5: Effect of segments on the blast response of buried tunnels
5Chapter 5: Effect of segments on the blast

response of buried tunnels
5.1 INTRODUCTION
Numerical techniques validated in the previous chapter provide confidence in

investigating the response of segmented bored tunnels subjected to surface blast. The
previous chapter revealed that ALE modelling technique in LS-DYNA is a suitable
platform for the modelling of the surface blast response of buried structures. A common
single tube railway tunnel system with six identical rectangular segments (RS) spanning
the circumference of the tunnel is simulated by using fully coupled ALE modelling
techniques as described in Chapter 4. Two types of segments, hexagonal segments (HS)
and interlocking segments (IS), are considered to study their effects on the dynamic
response of bored tunnels subjected to surface blast. The effect of explosive weight on the
blast response is also considered. The chapter first describes the simulations and the set-up,
followed by a comparison of results.
5.2 BLAST RESPONSE OF SEGMENTED BORED TUNNELS WITH

RECTAGULAR SEGMENTS
Bored tunnels run through densely populated city centres where the tunnels come to the
ground surface via tunnel portals and/or underground stations. In this area, the tunnels
appear to be shallow. The required minimum depth of the tunnel is determined by the
demand of TBM that requires a certain amount of soil (overburden) above the tunnel in
order to drive the TBM and the vertical stability (under effect of buoyancy) of the tunnel
after construction. As a “rule of thumb”, a minimum depth of cover is considered as 1D
where D is the outer diameter of the tunnel. The tunnel linings consist of precast reinforced
concrete segments connected by bolts at their interface.
5.2.1 Description of segmented tunnel
As far as urban rapid transit railway tunnels are concerned single tube (single track) tunnels
with inner diameter between 5.40m and 6.40m are common. Figure 5.1 describes the details
of a single tube transit tunnel. The tunnel has an inner diameter of 5.80m and an outer
Page 83
diameter of 6.35m. A lining ring (width of 1.4m) consists of six similar precast reinforced
concrete segments spanning the circumference of the tunnel. The reinforcement provides
resistance against varying earth pressure due to both static and dynamic situations. The
segments are reinforced concrete (grade 60-Appendix B) with a concrete cover of 40mm.
The segments are rotated from ring to ring by 30o in order not to line up along the
longitudinal direction. Every single segment is attached to its neighbouring segments by
bolts in both radial and circumferential directions, in which joints are considered as flat
surfaces, as described in Figure 5.1. The bolts are M24 grade 8.8 curved bolts.
(a) Tunnel cross section (b) Reinforcement arrangement
(c) Radial joint (d) Circumferential joint
Figure 5.1: Rectangular Segments (RS)
Page 84
5.2.2 Description of numerical model
The problem was modelled using the ALE modelling technique and material models
discussed in Chapter 3. As the lining has no-uniform properties about its periphery because
of such joints, a half symmetric model was considered. As shown in Figure 5.2, the model
involved a half of the air domain, soil domain, explosive and tunnel structure with
appropriate boundary conditions. The surrounding soil represented the dry Nevada sand
discussed in Chapter 4. The CONTACT_AUTOMATIC_SINGLE_SURFACE contact was
used to define the interface between the segments. The curved bolts were simulated as a
chain of four numbers of straight beam elements making the curvature. Reinforcing beam
elements and bolts were modelled as discrete elements immersed in the segments meshes
using the CONSTRAINED_LAGRANGE_IN_SOLID coupling.
Cylindrical disk type explosives were placed on the ground surface directly above the
tunnel crown. In order to examine the effect of explosive weight, five load cases were
considered by varying the explosive weight from 250 to 1250kg of TNT, by equal amount
of 250kg as shown in Table 5.1, for the minimum tunnel depth of 6.35m.
Figure 5.2: Configuration of the coupled numerical model.
Page 85
Table 5.1: Load cases
Case 1 2 3 4 5
TNT explosive, (kg) 250 500 750 1000 1250
5.2.3 RS Tunnel response
Prior to the blast, gravity loading was applied to preload the model. As shown in the
displacement vs. time plot in Figure 5.3, the tunnel was brought to the geostatic equilibrium
state where the maximum displacement was within the allowable deflection of 1% internal
diameter. The blast simulation was conducted for a duration of 50ms in order to evaluate
the initial damage.
0
Crown displacement
-0.5
Displacement, cm
-1
-1.5
-2
0 100 200 300 400 500 600 700
Time, ms
(a) Displacement-initialization (b) Initialized tunnel under gravity

load
Figure 5.3: Gravity initialization.
Figure 5.4(a) illustrates the blast load transfer mechanism in the tunnel. When the shock
waves impacted the tunnel, the segments along the tunnel crown were compressed in the
longitudinal direction as the shock waves progressed. However, as a result of soil
confinement, energy dissipation of joint interfaces and shock wave attenuation, this effect
was localized within a limited number of segments from the explosive, as shown in Figure
5.4(b). At the invert level, blast induced tension force in the longitudinal direction was
redistributed to adjoining segments through the bolted circumferential joints. Although out
of plane movements in the tunnel are naturally restrained along the longitudinal tunnel axis,
the blast induced tension triggered the yielding of bolts. It was found that gaps which
Page 86
opened between segments were insignificant (maximum of 1.2mm for load case 5) due to
the fact that the bolts were strong in tension.
As the shock waves travelled through the tunnel, ovaling of the tunnel profile (in-plane
response) was observed in the lining ring. Within the lining ring, segments responded in an
arch like behaviour that directly transferred the load to adjoining segments through the
radial joints. In the longitudinal direction, the vertical component of the blast load was
redistributed to its adjacent segments through the shear force transmission capacity at the
circumferential joints. In order to assess the in-plane response of the ring immediately
below the explosive, displacement history of closest point on the tunnel crown was
considered as shown in Figure 5.5. As expected, the vertical deformation increased with
explosive weight. For load case 1, the tunnel crown experienced more than six times its
geostatic displacement. For load cases 2 to 5, the peak crown displacements were several
times of the geostatic displacement. This can be a hazard for the operational envelop which
accommodates train and other services.
(a) Blast load transfer mechanisms
(b) Numerical simulation
Figure 5.4: Tunnel response in longitudinal direction
Page 87
-5
Crown displacement, cm
-10
-15
Case 1
-20 Case 2
-25 Case 3
Case 4
-30
Case 5
-35
700 710 720 730 740 750
Time elapsed, ms
Figure 5.5: Tunnel crown displacement
Figure 5.6 demonstrates that the segments moved relative to each other (in the form of
drifting response). This mainly occurred perpendicular to the tunnel axis between the rings
along the longitudinal direction caused by the vertical component of the blast load
redistributed to the adjoining rings through the bolted joints. These were weak in shear and
in the vicinity of the explosion, many bolts failed in shear in the common planes and not
through global deformation of the rings. As there were no bolt failures at the circumferential
joints for load cases 1 to 3, the drifting responses were significantly recovered. However,
load cases 4 and 5 generated a permanent drifting response as shown in Figure 5.6(b)
whereas the in-plane drifting response has insignificant effects on the tunnel response as
shown in Figure 5.6(c). The drifting response resulted in offset between waterproofing
gaskets as shown in Figure 5.6(a). Further drifting may accelerate the soil and water inflow
which may make the situation even worse by affecting the integrity of the above ground
structures.
Page 88
(a) Drift between adjacent lining rings
3.5 0.5
Case 1
3
0.4 Case 2
Drift response, cm
Drift response, cm
2.5 Case 3
0.3 Case 4
2 Case 1
Case 2 Case 5
1.5 0.2
Case 3
1
Case 4 0.1
0.5 Case 5
0 0
700 710 720 730 740 750 700 710 720 730 740 750
(b) Drifting response between (c) In-plane drifting response between

lining rings segments
Figure 5.6: Drifting response
The lining ring that was closest to the explosive ruptured first with cracks and a significant
number of progressive cracks developed during the phase of lining vibration following the
blast. Cracks were mainly generated by bending stresses. The number of cracks and the
maximum crack widths increased with explosive as presented in Table 5.2. The crack width
measurements were verified with a single element simulation in Appendix A. In all load
cases, the maximum crack widths exceeded the design crack limiting value of 0.3mm
(Engineering Group, 2010).
Table 5.2: Crack response under different load cases
Maximum crack width in mm Nos. of cracks (crack width>0.3mm)

Case1 Case2 Case3 Case4 Case5
1.92 4.11 5.43 5.79 7.33
396 907 1211 1472 1500
Page 89
Figure 5.7 shows the cracks induced after 27 ms of detonation in load case 5. In the
proximity to the tunnel crown as shown in Figure 5.7(a), the flexural cracks were first
triggered by residual circumferential stresses from the combined action of hoop stresses and
circumferential bending stresses on the interior surface of the segments. As displayed in
Figure 5.7(b), further flexural cracks developed on the top surface of the segments due to
the similar action, but the bending stresses were on the exterior surface. The main cracks
that appeared in the longitudinal direction have the expectable orientation due to the
bending stresses produced in the circumferential direction according to the ring-arch
mechanism.
(a) view of cracks in the tunnel interior
(b) view of cracks in the tunnel exterior
Figure 5.7: Crack patterns obtained from numerical simulations
Page 90
5.3 BLAST RESPONSE OF SEGMENTED BORED TUNNELS WITH

DIFFERENT TYPES OF SEGMENTS
The simulations indicated in the previous section that the drifting response resulted from
the continuous vertical joints between the rings in a tunnel with RS system. In order to
prevent the development of drift, two types of segment were considered. The first one is
the precast concrete hexagonal segmental (HS) system which has the advantage of non-
continuous ring joint. It is offset by a half segment width between the neighbouring
segments along the longitudinal direction as shown in Figure 5.8(a). As there could be
construction difficulties associated with the handling of the hexagonal segments by the
conventional TBM which uses horizontal segment erection method, an interlocking
segment (IS) system was also considered where a part of the segment is stepped and
interlocked with neighbouring segments as shown in Figure 5.8(b).
(a) Hexagonal segments (HS) (b) Interlocking segments (IS)
Figure 5.8: Different types of segments.
5.3.1 HS Tunnel response
Static responses of HS were similar to RS system under geostatic conditions, but there
were significant changes in the blast responses due to the behaviour of joints. In the HS
system, drifting response along the tunnel axis was essentially improved and converged
to a very minimal residual displacement as shown in Figure 5.9(a). However, for weaker
blast load, there were no significant change in the peak crown displacements (as shown
in Figure 5.9(b)) compared to the RS system. For higher blast load, the crown
displacements were slightly reduced because of the adopted joint system.
Page 91
1 0
Case 1
0.75 -5
Drift response, cm Case 2
Case 3 -10
0.5
Case 4 -15
0.25 Case 1
Case 5 -20 Case 2
0 Case 3
-25
Case 4
-0.25 -30
Case 5
-0.5 -35
700 710 720 730 740 750 700 710 720 730 740 750
(a) Drifting response between lining rings (b) Tunnel crown displacement
Figure 5.9: Immediate HS ring response.
Figure 5.10(a) shows that for load case 3, the segments along the tunnel crown were
subjected to extreme drift movements perpendicular to the tunnel axis. As the shock
waves progressed, the crown-segments were disengaged and slipped along the adjacent
segments upon failure of bolts in shear. While the drifting response developed and
continued along the weakest plane as described by the zigzag lines, segments drifted
relative to each other along the springline. Though this was a transient phenomenon and
quickly returned to rest with a negligible residual displacement, the immediate crown-
segment exhibited a large residual drifting of 11.0cm in the transverse direction as
shown in Figure 5.10(b). Considering Figures 5.9(b) & 5.10(b), it is evident that the
localized drift in the crown segment dominated the displacement response by more than
30%. Besides the localized drifting response, the global response of the overall HS
system is stiffer than RS system. This resulted from the “V” shaped (two-surface) radial
joints in HS system affecting the rotational ability (degree of flexibility) of the tunnel
even though the “V” shaped joint consisting of two-surfaces significantly increased the
sliding contact area compared to RS system which has a single surface radial joints (the
radial joints are parallel to the tunnel axis).
Page 92
(a) Drifting response (Load cases 2 and 3)

12
10
Drift response, cm
8
Case 1
6
Case 2
4 Case 3
Case 4
2
Case 5
0
700 710 720 730 740 750
Time elapsed, ms
(b) In-plane drifting response between segments
Figure 5.10: Hexagonal segmented tunnel response.
5.3.2 IS Tunnel response
Finally, the IS system response is compared with RS system. The tunnel suffered
permanent drifts between tunnel rings in all load cases, as shown in Figure 5.11(a), and
the drifts were comparatively more than RS system. As shown in Figure 5.11(b),
displacements were slightly more than RS system with unsmooth lines due to a series of
sudden drift movements. Although the radial joints are parallel to the tunnel axis,
alignment of radial joints (multi-surface) in the interlocking segment affected the
flexibility of the radial joints.
Page 93
-4 0
Case 1
-5
Drift response, cm -3 Case 2
Case 3 -10
-2
Case 4 -15
Case 1
-1 Case 5
-20 Case 2
0 -25 Case 3
Case 4
-30
1 Case 5
-35
700 710 720 730 740 750 700 710 720 730 740 750
(a) Drifting response between lining rings (b) Tunnel crown displacement
Figure 5.11: Immediate IS ring response.
Figure 5.12 illustrates, for load case 5, that the tunnel rings suffered permanent drifts
not only in the longitudinal direction, but also in the transverse direction (in-plane
drifts) of the tunnel. As shown in Figure 5.12(a), nosing of the interlocking segments
were subjected to a high concentrated stress due to the extreme hoop stress from the
crushing of adjoining segments. This resulted in contact-element distortion and it
appeared with cracks around the nosing of contact elements. It can be clearly seen that
the downward drift at the crown, as displayed in Figure 5.12(b), increased the crown-
displacement response. Overall, the interlocking joint system affected the tunnel
response in both longitudinal and transverse directions of the tunnel.
(a) Drifting response (Load case 5)
Figure 5.12: Interlocking segmented tunnel crown response
Page 94
5
Case 1
4 Case 2
Drift response, cm
Case 3
3 Case 4
Case 5
2
0
700 710 720 730 740 750
Time elapsed, ms
(b) In-plane drifting response between segments

elapsed
Figure 5.12: Interlocking segmented tunnel crown response
5.4 NUMERICAL RESULTS AND DISCUSSION
The study provides confidence in adopting the established techniques to evaluate the
response of segmented bored tunnels subjected to surface explosion using LS-DYNA.
The blast performance of the segmented bored tunnels buried in dry sand was
demonstrated for various blast load cases. The analysis showed that segments along the
tunnel crown were compressed in the longitudinal direction while the segments in invert
level were exposed to tension which generated a gap between the rings. The gap
opening response was relatively insignificant compared to the compression induced
effect at tunnel crown. However, the drifting response between the lining rings along
the longitudinal axis is the main concern affecting the water tightness in the commonly
used ring type modern tunnels. The hexagonal segmented tunnel system significantly
reduced the longitudinal drifts, but it suffered in-plane drifts resulting from weak
continuous joint alignments parallel to the tunnel axis. As shown in Figure 5.13(b), the
‘V’ shaped radial joints in the hexagonal segments affected the flexibility of the tunnel
as well. The interlocking segmented tunnel system showed slightly higher crown-
deflections than the commonly used tunnel system. It also suffered from drifts in all
directions. Although the radial joints are parallel to the tunnel axis in the interlocking
segmented tunnel, alignment of multi-surface contact in the interlocking segments, as
shown in Figure 5.13(c), affected the flexibility of the radial joints.
Page 95
(a) RS-joint (b) HS-joint (c) IS-joint
Figure 5.13: Interface of radial joints
Comparison of different types of segment geometries demonstrated that the commonly

used ring system with single-surface radial joints as presented in Figure 5.13(a)
exhibited better performance with regards to the overall drift responses. In the
transverse direction, this system is more flexible than other two systems in terms of in-
plane drifting response. This suggests that providing a secondary lining around the
primary lining will assist to redistribute the blast load to the neighbouring rings. The
drifts in the circumferential joints needs to be further investigated to improve the joint
performance under different soil conditions.
5.5 CHAPTER SUMMARY
In this chapter, a common single tube railway tunnel system with six identical
rectangular segments (RS) were used to investigate the response of the tunnel subjected
to surface blast. Fully coupled ALE numerical modelling technique incorporating Fluid
Structure Interaction (FSI) as described in Chapter 4 was developed and applied. For a
given tunnel depth of 6.35m, the influence of the explosive weight on the tunnel
response was studied by varying the explosive weight. It was found that the number of
cracks and the maximum crack widths in the segments increased with the explosive
weight, as expected. Even for a small amount of explosive of 250kg of TNT, the
maximum crack widths exceeded the design limiting value of 0.3mm. The tunnel
experienced differential movements (in the form of drifting response) between segments
in both longitudinal and circumferential directions. The drifting response in longitudinal
direction, i.e between rings, affects the water tightness between the rings. However, in
the circumferential direction, the drifting between segments (i.e in-plane drifting
response) was insignificant.
Page 96
Two types of segments, the hexagonal segments (HS) and interlocking segments (IS),
were considered to study the drifting effects between the rings. HS tunnel system
significantly reduced the drifts between the rings, but it suffered in-plane drifts resulting
from weak continuous joint alignments in the longitudinal direction. IS tunnel system
also suffered from the drifting between the rings as well as between segments in the
plane of the ring. The drifting response revealed that the tunnel rings in the RC tunnel
system were able to withstand higher blast loading compared to other two tunnel
systems though they displayed noticeable drifting between the rings.
The coupled modelling techniques which were developed and applied to treat the blast
response of buried tunnels provided useful information. The results on drift responses
and vertical displacements offer new and valuable information in the area of blast
response of segmented tunnels and will provide design information in future modelling
and analysis in this area.
Page 97
Page 98
Chapter 6: Crucial parameters affecting the blast response of buried tunnels
6Chapter 6: Crucial parameters affecting the

blast response of buried tunnels
6.1 INTRODUCTION
This chapter focuses on the investigation of several factors affecting the performance of
shallow bored tunnels in soft soils. Blast performance of bored tunnel structures is
influenced by the flexibility of the tunnel structure in relation to the properties of the
surrounding soil medium. The tunnel lining needs to be adequately flexible to be able to
deform and interact with the surrounding medium to resist the blast loading. After the
blast, the tunnel must be able to return to operation without suffering unacceptable
damages in the lining. The flexibility in the lining is determined by its ring bending
stiffness which depends on the elastic modulus of the material as well as on the
geometric properties of the lining, such as type of joints and number of segments used
to form a ring.
The segmental linings are primarily made of reinforced concrete (RC) which behaves in
a nonlinear manner. As the lining is subjected to blast loading, it cracks, behaves in a
less stiff manner, and benefits by redistribution of moment along the circumference of
the ring. The lining has joints which act like partial hinges. The stiffness of joints
depends greatly on the stiffness of the individual bolts and joint orientations. Muir
Wood (1975) suggested that the influence of joints may reduce the stiffness of the lining
to one-fourth of the full lining value for an eight-segment lining. The reduction in the
lining stiffness by increasing the number of segments increases the flexibility. However,
the number of segments per ring depends on many parameters including the aspect ratio
of the segment, internal diameter of the tunnel and the length of the segments (in the
longitudinal direction). The number of segments per ring reduces in keeping with
smaller tunnel diameters. For example, a common single tube railway tunnel system in
Singapore (Wen et al., 2004) has five segments plus a key segment.
There are different types of radial joints including flat, convex and curved. The flat
joints are more effective in terms of axial load transferring mechanism while both
Page 99
convex and curved joints facilitate rotation at the joint, allowing the segment to deform
and transfer moments at joints. Selection of an appropriate joint is essential to provide
structural flexibility without the segments being drifted or displaced with one another in
the radial direction.
The surrounding soil (soil skeleton) is a multiphase medium that includes solid soil
particles, and pores filled with water and air as illustrated in Figure 6.1. The soil is dry if
all pores are filled with air. If all pores are filled with water, the soil is saturated.
Otherwise, the soil is partially saturated in which the water can be either continuous or
discontinues in the pores depending on its degree of water saturation. Wang et al. (2004)
described that the soil involves two basic deformation mechanisms under loading: (i)
low pressure generates elastic deformation of bonds on the contact surfaces of particles;
and high pressure generates both a failure in bond and displacements of the particles;
and (ii) deformation occurs in all the soil phases based on their respective volume
compression. When a soil experiences loading, both mechanisms act simultaneously.
Figure 6.1: Soil skeleton
Under natural conditions, dry soil contains air and a small amount of water, whose
compressibility significantly exceeds that of the skeleton. Under static and dynamic
loading, the first mechanism initially governs the deformation. With increasing
pressure, the bond between the particles deforms and displaces and the second
mechanism becomes more important as the soil is compacted. In a saturated soil, the
pores are filled with water and a little air. Under rapid impulsive loading, both the water
and air provide higher load resistance than the particle bonds. The deformation and
resistance are predominately determined by the second mechanism, specifically by
Page 100
water and air deformation and the solid phase is only significant at relatively high
pressure.
Under static loading, the water and air are squeezed out of the pores and the
compressibility depends mainly on the solid skeleton. On the other hand, under blast
loading, the soil skeleton deforms with the trapped air and water within the pores as the
duration is not long for the air and water to flow through the skeleton. In this study, the
soil model is considered as a single phase material at the macro view though the soil has
a three phase material at the micro view.
This chapter first evaluates the blast performance of shallow bored tunnels in different
soils. A common single tube railway tunnel system is considered and the tunnel ring
includes five segments with a key segment along the tunnel circumference. A number of
blast cases as described in Chapter 5 are considered by varying the mass of the
explosive for a given tunnel depth of 6.35m. This chapter further investigates the effect
of structural flexibility by varying the radial joints types and the number of segments
forming the tunnel ring.
6.2 BLAST PERFORMANCE OF SHALLOW BORED TUNNELS IN

DIFFERENT SOILS
6.2.1 Peak pressures in various soil types
The present study investigates the blast response of shallow bored tunnel in different
soil types. Three types of soils considered by Jayasinghe et al. (2014) to investigate the
blast effect on buried piles were used in this study. Table 6.1 presents the three soil
types, such as saturated soil, partially saturated soil and dry soil. As described by
Jayasinghe et al. (2014), a similar free-field study was conducted. The free-field peak
pressure attenuation responses for those three soils were consistent with the plot
presented by Jayasinghe et al. (2014). Figure 6.2 shows comparison of the peak pressure
attenuation for the three types of soils as a function of scaled distances. It shows that
both the soil type and the degree of water saturation played a large role in determining
the peak pressures. The comparison shows that higher peak pressures occurred in the
saturated clay soil.
Page 101
Table 6.1: Soil properties for numerical simulation (Jayasinghe et al., 2014)
Soil properties Saturated soil Partially saturated soil Dry soil
Composition Clay Sand & Clay Sand
Density 2065 kg/m3 1960 kg/m3 1450 kg/m3
Degree of saturation 100% 85% (Va> 4%) 0%
Seismic velocity 1575 m/s 500 m/s 175 m/s
100
Saturated soil
Peak Pressure (MPa)
10 Partially
saturated soil
Dry soil
1
0.1
0.01
0.1 1 10
Scaled distance (m/kg1/3)
Figure 6.2: Comparison of free-field peak pressure
6.2.2 Description of segmented tunnel
As described in Figure 6.3, a common single tube railway tunnel system (Wen et al.,
2004) was considered with a 150mm thickness annulus concrete grout of concrete grade
15 (Appendix B) around the tunnel. The inner diameter and the thickness of the tunnel
lining were 5.8m and 275mm respectively. The segment was 1.4m length in
longitudinal direction. As illustrated, the segments were rotated from ring to ring by
22.5 degree angle to the tunnel centreline (CL). Both radial and circumferential joints
were flat and reinforcement details are described in Figure 5.1.
Page 102
(a) Key segment: left-hand CL (b) Key segment: right-hand CL
Figure 6.3: General arrangement of segments
6.2.3 Description of numerical model
Figure 6.4 represents the three dimensional numerical model to study the effects of
surface blast loading on tunnel response under influence of the soil properties. By
considering the symmetries, half of the geometry was modelled with a cylindrical
explosive on the ground surface. The annulus gap grout achieved a full round
embedment with the surrounding soil. Therefore, the interface between the grout and
soil was modelled using merged nodes. However, the interface between the segments
and grout was modelled using the penalty-based contact surface-surface type. The soil
properties were changed to investigate their effects on the different aspect of the tunnel
response.
Page 103
Figure 6.4: A half symmetrical numerical model
6.2.4 Blast response of segmented tunnel
Prior to the blast, the model was brought to equilibrium under gravity loading. As the
simulations are quite expensive in terms of computational time and memory space, the
process of capturing the blast responses was recorded over 300ms of duration. Blast
induced tunnel deformations were three dimensional. However, as the tunnel is
naturally restrained along the direction of tunnel axis, in-plane (ZX plane) deformation
of the ring immediately below the explosive is critically more important than out of
plane deformation in the blast analysis. Before the blast, tunnel exhibited a gravity-
induced elastic diametric distortion. After the blast, the diametric distortion of the tunnel
is significant as it changed its shape to ovalisation before the tunnel failed due to either
the joint distortion or excessive cracks in the segmented tunnel. Blast induced tunnel in-
plane deformation can be expressed by Equation 6.1.
D(t )  D0  (t ) Equation 6.1
where: D(t) is the diameter of deformed tunnel at time t, D0 is the initial diameter of the
tunnel and (t ) is diametric distortion at time t.
Figures 6.5 to 6.7 show the time histories of the diametric distortion which resulted
from the displacement of the tunnel crown with respect to the tunnel invert in different
Page 104
soils. The figures illustrate that the tunnel experienced unrecoverable deformation in all
the cases. In general, after first peak, tunnels subjected to a series of noticeable
fluctuations in distortion before steadying to residual distortions. It can be seen that the
amplitude of the fluctuation depends on the soil type. For instance, the tunnel buried in
the partially saturated soil displayed a quicker residual distortion than that in the other
two soils. The tunnel buried in the dry soil suffered higher peak diametric distortions
than the tunnel in the partially saturated soil. As expected, the tunnel buried in saturated
soil produced considerably larger peak distortions. Beyond load case 3 in the saturated
soil, the key segment in the immediate plane disintegrated from the neighbouring
segments due to the drifting response resulting from the bolt failure and this may lead to
progressive collapse of the tunnel structure. Therefore, it is recommended that the
tunnels should be buried above the ground water table. However, this may not be
practicable because of minimum soil cover requirement and variation of the ground
moisture level with seasonal changes.
0
Case 1
Case 2
Diametric distortion, cm
-4 Case 3
Case 4
-8 Case 5
-12
-16
500 550 600 650 700 750 800
Time elapsed, ms
Figure 6.5: Diametric distortion in dry soil
Page 105
0
Case 1
Case 2
-4 Case 3
Case 4
-8 Case 5
-12
-16
500 550 600 650 700 750 800
Time elapsed, ms
Figure 6.6: Diametric distortion in partially saturated soil
0
Case 1
Case 2
-4
Case 3
-8 Failure due to drifting
-12
-16
500 550 600 650 700 750 800
Time elapsed, ms
Figure 6.7: Diametric distortion in saturated soil
Table 6.2 compares the number of radial bolt failures during the first peak distortion and
at the end of the simulation of the tunnel in different soils. There were no bolt failures in
the tunnel buried in the dry soil for all load cases. Though the tunnel in the partially
saturated soil displayed bolt failure for the highest load cases 4 and 5, in-plane tunnel
profile were not affected by any form of drifting responses. For the saturated soil, there
were bolt failures in all load cases. A comparison of peak pressure attenuations study, as
shown in Figure 6.2, illustrates that the peak pressure of the saturated and partially
saturated soils were 40 and 5 times the peak pressure of the dry soil at the depth of
6.35m for load case 1 (scaled distance ≈ 1.0m/kg1/3). As a result, a large number of
radial bolts failed in the saturated soil by the high intensity compressive blast wave
impacting the tunnel crown. The number of bolt failure at the end of the simulation
shows a rapid increase from load case 2 to 3 for the saturated soil.
Page 106
Table 6.2: Numbers of radial bolt failure

Numbers of radial bolt failure during
the first peak distortion end of the simulation(@ 800ms)
Case 1 Case 2 Case 3 Case 4 Case 5
Dry 0 0 0 0 0
0 0 0 0 0
Par. saturated 0 0 0 3 3
0 0 0 3 3
Saturated 1 4 7 9 11
4 11 32 41 51
Figure 6.8 illustrates the contours of plastic strain developed along the bolts connecting
the key segment with neighboring segments during their respective first peak
distortions. In dry soil, effective plastic strains in the bolts are very small for load case 5
as shown in Figures 6.8(a) and 6.8(b). As part of the modeling technique, a failure
strain erosion criterion was used such that the beam elements within the concrete were
eroded once the plastic strains within those elements exceed 0.12. In the partially
saturated soil, for load case 4, the first bolt failed after 15ms of the explosion as
displayed in Figure 6.8(d). At the end of the simulation, there were three bolt failures in
both load cases 4 and 5. In Figures 6.8(e) and 6.8(f), the scenario was more critical in
the saturated soil as all radial bolts failed immediately after the first peak distortion.
After the bolt failures, as shown in Figure 6.7, the key segment drifted upwards due to
the presence of blast induced negative pressure at the tunnel crown.
(a) t = 578ms (Dry -Case 5) (b) t = 579ms (Dry -Case 5)
Figure 6.8: Radial bolt failure adjacent to key segment
Page 107
(c) t = 515ms (Partially saturated -Case 4) (d) t = 554ms (Partially saturated -Case 4)
(e) t = 507ms (Saturated-Case 3) (f) t = 508ms (Saturated-Case 3)
Figure 6.8: Radial bolt failure adjacent to key segment
6.3 EFFECT OF RADIAL JOINT TYPES
This section examines how the tunnel’s flexibility influences its response to the surface
blast by varying the radial joint/contact types. The radial joints can be either designed as
flat, convex-concave or convex-convex. The capacity of a joint to resist blast loading
depends on the load transmission behaviour and its ultimate capacity. Table 6.3
compares the ability of joints to rotate and transfer the loads based on the geometry of
the joints. Due to the inherent complex behaviour of segmental joints under blast
loading, it is essential to study the combined effect resulting from rotation, axial/hoop
force transmission and shear force transmission at the joints.
In this study, the circumferential joints were all considered as flat. The tunnels were
buried in saturated clay soil. Other than the geometrical aspect of radial joints, the
tunnels were identical in terms of structural aspects such as number of segments,
dimensions, materials and reinforcement arrangements. Since the tunnel in saturated soil
was severely deformed for the explosive of 750kg of TNT, the load cases were refined
as presented in Table 6.4 with five new load cases to study the effect of explosive mass
on the tunnel response.
Page 108
Table 6.3: Radial joint types and corresponding force transmission capacity
Types of radial joints Shear force Axial force Rotational

transmission transmission capacity
Medium High Low

(a) Flat joint
High Medium Medium

(b) Convex-concave joint
Low Low High

(c) Convex-convex joint
Table 6.4: New Load cases
Case 1N 2N 3N 4N 5N
TNT explosive, (kg) 250 375 500 625 750
The shockwave was transmitted to the tunnel lining in the form of compressive hoop
stress and normal stress. The hoop stress acts along the circumference of the lining
while the normal stress acts perpendicular the lining surface. Load case 1N was
considered as an example to study the crack formation immediately after impact of the
shockwave. There were no cracks in the flat joints as shown in Figure 6.9(d). As the
hoop stress at the tunnel crown was greatly increased by the shockwave, in addition to
the bolt resistance, the flat joints in the vicinity of the tunnel crown greatly resisted the
blast induced normal stress by the increased frictional resistance capacity between the
flat joints which are effective in transferring the hoop stress. However, a large number
of elements of the convex-concave joints were cracked before the bolt failure due to a
high intensity of shear force transmitted through the edge elements making the
Page 109
periphery as displayed in Figure 6.9(f). It should be noted that the edge elements
marking the periphery of the convex-concave joints were not reinforced because of
minimum required concrete covers. Due to rapid increase in hoop stress at the tunnel
crown, the convex-convex joints exhibited high stress concentrations at the radial
contact causing splitting forces which resulted cracks in the segments as shown in
Figure 6.9(g). In these joints, the normal stress was primarily resisted by radial bolts
rather than the frictional forces at the interface. Until the bolts reached ultimate strength,
the segments were slightly rotated about the joints. After the bolts failure, the segments
were moved up and down with respect to the neighboring segments. As a result, the
number of cracks appeared to be increased on the edges.
(a) Shockwave at 504ms (b) Shockwave at 505ms
(c) Flat joint at 504ms (d) Flat joint at 505ms
(e) Convex-concave joint at 504ms (f) Convex-concave joint at 505ms
Figure 6.9: Initial crack formation after impact of shockwave in load case 1N
Page 110
(g) Convex-convex joint at 504ms (h) Convex-convex joint at 505ms

Figure 6.9: Initial crack formation after impact of shockwave in load case 1N
Figure 6.10 describes the in-plane radial drifting response in different types of radial
joints. In the first three load cases, the variations of the drifting responses were
insignificant although the number of radial bolt failures varied significantly as
summarized in Table 6.5. In load case 4N, both flat and convex-convex joints displayed
larger drifting at the end of the simulation than in convex-concave joint. As the
explosive is further increased to case 5N, the tunnel with flat joints suffered large
drifting as shown in Figure 6.10(e). Although the tunnel with convex-concave joints
displayed a larger number of bolt failures at the end of the simulation in load cases 3N
to 5N, the high shear transmission capability resulting from the interlocking system in
the convex-concave joint largely protected the tunnel profile at failure by minimizing
the drifting responses.
0.6 1.2
Flat Flat
0.5 Convex-concave Convex-concave
Convex-convex 0.9 Convex-convex
Radial drift, cm
Radial drift, cm
0.4
0.3 0.6
0.2
0.3
0.1
0 0
500 550 600 650 700 750 800 500 550 600 650 700 750 800
(a) Case 1N (b) Case 2N
Figure 6.10: Radial drift response of an immediate ring
Page 111
1.8 2.5
Flat Flat
1.5 Convex-concave Convex-concave
2
Convex-convex Convex-convex
Radial drift, cm
Radial drift, cm
1.2
1.5
0.9
1
0.6
0.3 0.5
0 0
500 550 600 650 700 750 800 500 550 600 650 700 750 800
(c) Case 3N (d) Case 4N
15
Flat
12 Convex-concave
Convex-convex
Radial drift, cm
0
500 550 600 650 700 750 800
Time elapsed, ms
(e) Case 5N

Case 1N Case 2N Case 3N Case 4N Case 5N
1 3 4 7 7
Flat joint
4 8 11 18 32
Convex-concave 0 0 0 4 6
joint 2 6 17 30 39
Convex-convex 1 3 4 4 4
joint 2 5 11 26 32
Table 6.6 compares the maximum crack width in both the first peak distortion and at the
end of the simulation (at 800ms). It appears that the maximum crack width in the tunnel
lining exceeded its limiting design value of 0.3mm at the end of the simulation in all
load cases. As expected, the maximum crack width increases with the explosive mass.
The cracks were opened and closed during the phase of vibration. Under small
explosive loads (load case 1N), the convex-convex joint performance is better in terms
Page 112
of the crack formation as the maximum crack width slightly exceeded the limiting value
due to its greater flexibility allowing the segment to deform and dissipate the bending
moment. During the first peak deformation, the joint proximity to the tunnel crown
exhibited a maximum crack width of 0.42mm which is more than the crack width in the
crown segment at the end of the simulation. The number of bolt failures was reduced by
almost half in the flat jointed tunnel as summarized in Table 6.5.
In load case 2N, the convex-convex joint transferred a significant amount of moment to
the mid-span of the segment. As a result, the tunnel was subjected to an increase in the
crack width due to the increase in sagging moment. The convex-concave joint displayed
large crack width during the first peak distortion due to the lack of reinforcement in the
periphery of the joints. Table 6.5 illustrates that there were no bolt failures in the
convex-concave joint during the first peak distortion, but the numbers bolt failure was
considerably increased at the end of the simulation because the radial bolts resisted
significant amount of share after failure of the edge elements making the periphery.
Table 6.6: Maximum crack width in mm
Maximum crack width in mm during

0.12 0.51 0.65 0.74 0.92
Flat joint
0.62 0.63 0.97 1.61 2.20
Convex-concave 0.74 0.70 0.92 0.81 1.15
joint 0.93 1.07 1.14 1.67 2.25
Convex-convex 0.42 0.48 0.65 0.69 0.85
joint 0.37 0.89 1.90 2.12 2.57
As the explosive is increased to load case 5N, intensity of the blast induced shock wave
is more concentrated on the tunnel crown. The rigid nature of the flat joints around the
tunnel crown dissipated the blast energy by allowing the segments to slide against each
other. This dissipation of local effects is further facilitated by failure of radial bolts.
Consequently, the flat joint exhibited more radial drifting and hence induced an
eccentricity at the joint causing additional moment in the segments. Similar to the flat
joint, the convex-convex jointed tunnel suffered from high drifting response with the
same number of bolt failures at the end of the simulation. Although the convex-concave
Page 113
joint displayed more bolt failures, the joints sustained the higher load of explosive at
failure. The maximum crack widths were nearly the same in all types of joints. This
shows that choice of joint types has insignificant influence on the blast response of
segmental tunnel for higher blast loads as the segments were damaged before activating
the flexibility of the joints. However, for weaker/smaller blast loads (  250kg of TNT),
the joint types influenced the tunnel response by allowing the segments to rotate about
the joints.
Performance of the convex-concave joint can be improved by providing adequate

reinforcement to facilitate the load transfer from one segment to another without
cracking. However, the particular shape of the edges creating the periphery of the
segment makes it difficult to place the reinforcement as a minimum required cover is
necessary for protection against corrosion in the segments. On the other hand, the flat
joints are very efficient in transferring the hoop stress between segments with less end
reinforcement. Furthermore, the flat joints effectively provide grooves to accommodate
the watertight gasket within the joints.
6.4 EFFECT OF NUMBER OF SEGMENTS
With an increase in the number of segments forming a ring, the number of joints
increases accordingly. This section examines how the number of segments affects the
flexibility of the tunnel. The stiffness of the joint is always less than that in the body of
the segment. Therefore the flexibility of the tunnel lining increases with an increase in
the number of segments. Equation 6.2 computes the equivalent stiffness of the lining
from Muir Wood (1975). This equation demonstrates that the lining will always be less
stiff than an unjointed lining if the number of joints is more than 4.
 n
2 Equation 6.2
Ie  I j  I 4
where: Ie is moment of inertia of jointed lining

Ij is the moment of inertia of the joint
n is the number of joints (if >4)
I is the moment of inertia of the unjointed lining
Page 114
In order to examine the effect of number of segments in a ring on the blast response of
the tunnel, three types of segmented tunnels are considered by varying the number of
segments as shown in Figure 6.11. These are labelled as Type 1, 2 and 3 and consist of
five, six and seven segments respectively, including the key segment along the tunnel
circumference. The key segments are of the same size in all three tunnel types. Types 1,
2 and 3 tunnels are similar in terms of structural aspects such as dimensions, materials,
reinforcement arrangements and joint details. The joints were considered as flat in all
three types.
Figure 6.11: Isometric view of segmental lining
Figure 6.12 compares the in-plane radial drifting response between segments closest to
the tunnel crown of the immediate ring. In load case 1N, there were no significant
changes in the drifting response of those three tunnels. However, as the load increased
from case 2N to 4N, Type 3 tunnel showed very minimal drifting compared to other two
tunnels. Beyond load case 4N, all the tunnels suffered excessive drifting and wide
cracks that lead to high strain in the reinforced steel and also many bolt failures at the
joints. Excessive drifting response can lead to a progressive collapse of the tunnel
structure.
Page 115
0.25 1.2
Type 1 Type 1
Type 2 Type 2
0.2
Type 3 0.9 Type 3
Radial drift, cm
Radial drift, cm
0.15
0.6
0.1
0.3
0.05
0 0
500 550 600 650 700 750 800 500 550 600 650 700 750 800
(a) Case 1N (b) Case 2N

1.8 4
Type 1 Type 1
1.5 Type 2 Type 2
Type 3 3 Type 3
Radial drift, cm
Radial drift, cm
1.2
0.9 2
0.6
1
0.3
0 0
500 550 600 650 700 750 800 500 550 600 650 700 750 800
(c) Case 3N (d) Case 4N

15
Type 1
12 Type 2
Type 3
Radial drift, cm
0
500 550 600 650 700 750 800
Time elapsed, ms
(e) Case 5N
Figure 6.13 shows the crack development after 300ms of the explosion in Type 1 tunnel
for load case 4N. The lining ring that was closest to the explosive ruptured first with
flexural cracks, followed by inclined flexure-shear cracks above the springline of the
tunnel. The flexural cracks formed when the tensile stresses due to bending exceeded
the flexural strength of the concrete under arch mechanism of the tunnel behaviour. The
cracks developed and extended to adjoining rings during the phase of shock wave
propagation. Within the ring, the segment on the right-hand side of the key segment
suffered excessively and a deep crack progressed through the entire segment material at
Page 116
the tunnel crown. This segment is therefore at risk of failing. However, as the key
segment has a small arc angle, it showed minor flexural cracks which opened and closed
during vibration and degree of severity. As expected, number of cracks, crack widths
and depths (crack depth to width ratio) increased with explosive mass. Observed
flexural cracks at the tunnel crown were much more than the shear cracks and the
flexural crack widths varied in range from 0 to 3.41mm.
Full-depth flexural cracks rarely occur in the segments as one part of the cross section
(in the longitudinal direction) should be in compression at all times to provide the
internal bending moment to resist the external moment which results from both the soil
and the blast loads. Most of the cracks formed during the phase of blast impact were
closed on the compression side of the segment under the action of the hoop stress.
However, a crack which opened fully at the tunnel crown did not close during the tunnel
fluctuations and the bending moment was significantly reduced to a small value. This
could be possible as the doubly reinforced section could provide the essential internal
resisting moment, without the need of compression in the concrete.
Figure 6.13: View of cracks (>0.3mm) in Type 1 tunnel in load case 4N
The stiffness of the segment is relatively higher than that in the radial joint. The
stiffness of joints using bolts in tension and shear depends greatly on the stiffness of the
individual bolts and joint orientations. Blast induced global response of the tunnel is
governed by the joint stiffness provided that the segments are adequately reinforced.
Page 117
Reduction of the tunnel stiffness due to excessively wide and deep cracks in the
segments altered the load transfer mechanism, by which the segments were prompted to
rotate about the weakest cracked plane rather than about those radial joints. This could
be the reason that the in-plane drifting response is significantly smaller in Type 1 tunnel
than that in Type 2 tunnel, in which the key segment drifted relative to the right-hand
segment as shown in Figure 6.14. As the number of segments in a ring increased from
Type 1 to Type 2, the numbers of elements cracked (crack width > 0.3mm) were
reduced by nearly half and cracks were shallower and thinner.
Page 118
Table 6.7: Maximum crack width in mm
Maximum crack width in mm during

0.13 0.80 1.16 1.51 0.95
Type 1
1.46 1.95 2.76 3.41 3.77
0.12 0.51 0.65 0.74 0.92
Type 2
0.62 0.63 0.97 1.61 2.20
0.38 0.54 0.57 0.82 0.95
Type 3
0.49 0.84 1.42 1.46 2.78

Type 1 0 0 2 2 3
(10 radial bolts/ring) 1 2 4 14 21
Type 2 1 3 4 7 7
Type 3 0 1 1 1 3
The numbers of elements cracked in Type 3, as displayed in Figure 6.15, were more
than twice those for Type 2. As presented in Table 6.7, the maximum crack widths were
not much different compared to those in Type 2 tunnel. On the other hand, the numbers
of radial-bolt failures were largely reduced in all load cases as shown in Table 6.8. As a
result of less numbers of bolt failures at the radial joints, in-plane drifting responses
were insignificant. While Type 3 tunnel offered better performance in terms of reduced
radial bolt failure because of its increased flexibility, Type 2 tunnel suffered a large
number of radial bolt failures among other tunnels in all load cases.
Besides the radial bolt failures, the tunnels experienced a significant number of bolt
failures at the circumferential joints. There were no circumferential bolt failures for load
cases 1N to 3N. However, in load case 4N, the circumferential bolts in the ring at the
tunnel crown failed because of the high stress concentration from the blast induced
shear transmission force induced between the rings at the crown. The number of
Page 119
circumferential bolt failures varied between the tunnels and this was influenced by the
localized relative movement between the segments in the common plane between rings.
Figure 6.16 shows the contours of stress developed along the bolt before and after the
bolt failures in Type 3 tunnel for load case 4N. As presented in Table 6.9, for load case
4N, Type 1 tunnel exhibited three circumferential bolt failures closest to the tunnel
crown whereas Type 2 tunnel was free from the circumferential bolt failures. For load
case 5N, Type 1, Type 2 and Type 3 tunnels displayed 10, 3 and 7 circumferential bolt
failures respectively. Due to the influence of high flexibility in Type 3 tunnel, the tunnel
ring closest to the explosive attracted more blast energy. This resulted in higher relative
deformation in the closest ring with respect to the adjacent rings. The relative movement
between the rings increased the number of circumferential bolt failures. As Type 1
tunnel is considerably rigid, the closest ring distributed the blast load to its adjacent
rings through the common plane, in which a large number of the circumferential bolts
failed in shear. However, in terms of the circumferential bolt failures, Type 2 tunnel
displayed an intermediate response compared to Type 1 and Type3 tunnels.
(a) First circumferential bolt failure (b) First radial bolt failure
Figure 6.16: View of bolt failure in Type 3 tunnel (Load case 4N)
Page 120
Table 6.9: Numbers of circumferential bolt failure

Type 1 0 0 0 3 10
(17 circumferential bolts/ring)
Type 2 0 0 0 0 3
Type 3 0 0 0 1 7
6.5 DISCUSSION OF NUMERICAL RESULTS
Comprehensive studies were carried out by varying several parameters affecting the
blast response on a segmental tunnel buried in various homogenous soils. First, a tunnel
having five segments plus a key segment in a ring was buried in different soil types,
such as saturated soil, partially saturated soil and dry soil, to examine the effect of soil
properties on the tunnel response. The study showed that the tunnel buried in the dry
soil has higher peak diametric distortion than the tunnel in partially saturated soil.
Though the intensity of shockwave is smaller in dry soil than the partially saturated soil,
the deformation of the dry soil might be high due to poor bond between the soil grains
and the soil skeleton which consists of a large amount of air causing high
compressibility. Compared to other types of soils, the tunnel buried in the partially
saturated soil displayed a quick residual distortion. The tunnel buried in the saturated
soil was found to have the highest peak distortion and the tunnel suffered a large
number of bolt failures.
The effects of joint types as well as the number of segments forming a ring were also
investigated when the tunnel is buried in the saturated soil. The study showed that, for
weaker/smaller blast loads (  250kg of TNT), the choice of joint types has a great
impact on the blast response of the tunnel in terms of bolt failure and crack formation.
However, for higher blast loads, the segments were damaged before activating the joint
induced flexibility in the tunnel. There were no significant changes in the bolt failures
and crack formation for higher blast loads, but the convex-concave joint sustained the
tunnel profile by minimizing the drifting response at the joints. Flat joints were effective
Page 121
in transferring the hoop stress and it encouraged friction induced shear resistance at the
joints. The nosing elements making the periphery in convex-concave joint resisted shear
force by mechanical interlocking of those elements and many elements failed after
shockwave impact. While the flat joints are effective in transferring hoop stress at the
joints, convex-convex joint resulted in less number of bolt failures and significant
improvement in the crack response for weaker blast loading. Besides the joint responses
to the blast loading, the benefits associated with constructability in terms of
accommodating watertight gasket and less joint reinforcement outweigh the flat joint
from other joints such as convex-concave joint and convex-convex joint.
A tunnel with seven segments (six segments plus a key segment) in a ring has resulted
in very minimal drifting response compared to other types of tunnel up to 625kg of TNT
explosive. There was a sudden reduction in the maximum crack widths as the number of
segments increased from five segments (four segments plus a key segment) in a ring to
six segments (five segments plus a key segment) in a ring. For example, the maximum
crack width was reduced to 0.62mm from 1.46mm for 250kg of TNT explosive.
However, as the number of segments further increased to seven segments in a ring,
change in the maximum crack widths was insignificant. Although the tunnel with seven
segments in a ring displayed less radial bolt failures, the numbers of circumferential bolt
failures were significantly higher than the tunnel with six segments in a ring.
6.6 CHAPTER SUMMARY
In this chapter, the effect of crucial parameters on the tunnel response under surface
blast provides another perspective to the study of segmental tunnels. The influence of
the surrounding soil properties and geometrical characteristics of the segmental tunnel
lining were investigated. First, a commonly used single tube tunnel having six segments
(five segments plus a key segment) was used to examine the effect of soil properties on
the tunnel response. The tunnel buried in saturated soil, partially saturated soil and dry
soil were considered. Secondly, the influence of segment geometry was studied with
respect to the joint types and the number of segments forming the ring. The main
findings of this chapter are as follows:
Page 122
1. The blast response of buried tunnel in saturated soil is more severe in terms of
crack formation and bolt failures than the tunnel buried in either partially saturated
soil or dry soil when subjected to the same surface explosion.
2. In all soil conditions, the diametric distortions increase with the explosive mass and
the distortions are unrecoverable in all load cases (from 250 to 1250kg of TNT).
As the explosive mass increased to more than 750kg of TNT, the tunnel buried in
the saturated soil displayed segment-disintegration from neighbouring segments.
3. For weaker blast loads (  250kg of TNT), joint types influenced the tunnel response
by allowing the segments to rotate about the joints, whereas for higher blast loads,
the segments were damaged before triggering the flexibility of the joints.
4. While the flat joints are effective in transferring hoop stress at the joints, convex-
convex joint resulted in less number of bolt failures and significant improvement in
the crack response for weaker blast loading.
5. As the number of segments increased in a ring, the maximum crack width

decreased. However, as the number of segments further increased to seven
segments (six segments plus a key segment) in a ring, it did not make much
difference in the magnitude of the maximum crack width.
6. Although the tunnel with seven segments in a ring displayed less radial bolt
failures, the numbers of circumferential bolt failures were significantly more than
the tunnel with six segments in a ring.
Page 123
Page 124
Chapter 7: Blast impact and mitigation measures
7Chapter 7: Blast impact and mitigation

measures
7.1 INTRODUCTION
The dynamic behaviour of the tunnel under the influence of a series of surface blast
loads were previously investigated by varying a number of parameters, such as soil
types, joint types and number of segments in a ring. Blast induced ground shock had a
high level of impact on the performance of buried tunnel structures. This chapter
considers damage evaluation factors such as cracks in the segments, bolt failure, drifting
or sliding of segments against each other. The tunnel failure may be driven by a single
factor or by a combination of several factors. These factors are discussed in this chapter
along with mitigation measures that could be implemented to minimize the blast impact.
This chapter begins by analysing the blast response of the buried tunnel in saturated soil
with clay composition (Table 6.1) under a number of parametric conditions; soil cover
(tunnel depth), distance of explosive from the tunnel centreline (ground distance) and
explosive weight. Critical factors identified from the sensitivity study are considered to
mitigate the blast effect by the following measures:
1. wrapping the tunnel with a flexible and compressible barrier consisting of a

layer of polyurethane foam;
2. wrapping the tunnel with a secondary in-situ reinforced concrete lining;
3. use of Steel Fiber Reinforced Concrete (SFRC) in precast segments;
4. energy absorbing flexible honeycomb elements;
5. introducing a buster slab between the tunnel crown and the ground surface.
In the first two measures, the space between the bored soil face and the lining is filled
using the protective barrier materials which are in contact with the tunnel surface. As a
third measure, Steel Fiber Reinforced Concrete (SFRC) is used instead of conventional
reinforced concrete segments to produce the tunnel segments. Deformable honeycomb-
type cell elements are introduced at the radial joints as a fourth measure in order to
Page 125
absorb a considerable amount of blast energy. Finally, as an external measure, a buster

slab is placed between the tunnel crown and the explosive in order to combat the
shockwave before reaching the target.
7.2 THE EFFECT OF TUNNEL DEPTH AND STAND-OFF DISTANCE IN

BLAST IMPACT
There are numerous empirical relationships relating stand-off distance to blast effects
from various explosive weights for free-field explosions (Department of the Army, 1986
and Yankelevsky et al., 2011). However, the relationship between stand-off distance
and the segmented tunnel response due to a surface blast are not reported. There is no
established guidance for predicting either the tunnel response or the characteristics of
the blast loading. In this section, the effects of tunnel depth, ground distance and
explosive weight on the tunnel response are studied. Figure 7.1 (a) shows that the
tunnel depth is varied from 6.35m (= 1D) to 12.70m (= 2D), and the ground distance is
varied from the tunnel centreline to 12.70m (= 2D) as shown in Figure 7.1(b). As
described in Table 5.1, the cylindrical disk type explosive weight is varied from 250 to
1250kg of TNT.
(a) Variation of tunnel depth (b) Variation of ground distance
Figure 7.1: Variation of tunnel depth and ground distance
Cracks in segments, bolt failure and drifting response are considered in this study as
Page 126
critical factors to evaluate the blast performance of the tunnel structure. The damage
state of tunnels can be divided into the following four groups:
1. no damage (the tunnel is considered to behave elastically with some minor

cracks in segments (maximum crack width < 0.3mm) and no bolt failure);
2. slight damage (a small number of cracks exceeded the crack limiting value of
0.3mm and a few incidents of bolt failure at joints, but the drifting response is
insignificant);
3. moderate damage (a large number of cracks exceeded the crack limiting value of
0.3mm, a large number of failed bolts triggered significant drifting or sliding of
segments at joints, however, the tunnel remains functional by keeping the in-
plane tunnel profile due to hoop compression);
4. severe damage or collapse (formation of fully depth cracks, large number of bolt
failures resulting in large drifting between segments).
The damage state rises as intensity of shockwave impacting the tunnel increases. Low-
energy blast impact causes no and slight damages in the tunnel lining while high-energy
blast impact causes moderate and severe damages.
In the present study, more than 30 models were considered to identify critical situations
under the influence of tunnel depth, ground distance and explosive weight, but only a
few important results are shown in this section. Figure 7.2 illustrates the failure modes
of the tunnel under a load of 750kg of TNT (Load case 3 refers to Table 5.1) at different
tunnel depths. It can be seen that the tunnel depth showed significant influence on the
failure modes of the tunnel when the surface explosive was directly above the tunnel
crown. For the tunnel depth of 6.35m (1D), the tunnel was severely damaged with wide
and deep cracks, segments were crushed and drifted away from the adjacent segments
and a large number of bolts failed as shown in Figure 7.2(a). There was not much
difference in the tunnel buried at depth 2D in terms of number of bolt failures.
However, when referring to damage due to the crack, a significant change in the crack
response retained its in-plane tunnel profile without large drifting between segments as
displayed in Figure 7.2(b). The damage level corresponds to the state of “moderate
damage”. As the depth further increases to 2D, the tunnel responded with very small
Page 127
cracks, in which the maximum crack width was 0.24mm, less than the crack limiting
value of 0.3mm. From this point of view, it is appropriate to say that the tunnel is safe
with “no damage”.
(a) Depth = 1D (b) Depth = 1.5D (c) Depth = 2D

Max. crack width = 2.20mm Max. crack width = 0.57mm Max. crack width = 0.24mm
Nos. of cracks (>0.3mm) = 81674 Nos. of cracks (>0.3mm) = 60 Nos. of cracks (>0.3mm) = 0
Nos. of bolt failure = 35 Nos. of bolt failure = 10 Nos. of bolt failure = 0
Severe damage Moderate damage No damage
Figure 7.2: Failure modes of the tunnel under different tunnel depths (load case 3)
Table 7.1 compares the damage scenarios for the tunnel based on the drifting response,
maximum crack width and corresponding numbers of bolt failures under a series of
explosives from case 1 (250kg of TNT) to case 5 (1250kg of TNT). For the first two
damage states, the drifting response is insignificant, although few radial bolts failed
resulting in “slight damage” state. The combined effect of the cracking and bolt failure
influenced the drifting response in “severe damage” state where the crown segments
were severely drifted with respect to the adjacent segments in both longitudinal and
transverse directions. This resulted in a significant number of circumferential bolt
failures beyond load case 3 for the tunnel depth of 6.35m (1D). There were no
circumferential bolt failure in both depth 1.5D and depth 2D in all load cases. The
segments were slightly drifted and stabilized in “moderate damage” state while
maintaining the in-plane tunnel profile.
Page 128
Table 7.1: Maximum crack width in mm / nos. of bolt failures
Depth Case 1 Case 2 Case 3 Case 4 Case 5

0.62 0.97 2.20 3.00 4.40
Depth 1D
4 11 35 45 63
0.20 0.37 0.57 0.60 0.90
Depth 1.5D
0 2 10 17 32
0.15 0.21 0.24 0.36 0.54
Depth 2D
0 0 0 2 6
: No damage : Slight damage
: Moderate damage : Severe damage
Figure 7.3 compares the failure modes of the tunnel under load case 3 at different
ground distances for a specified tunnel depth of 6.35m (1D). The cracks on the
segments were observed from different ground distances. The ground distance from the
explosive has a significant influence on the damage response, which is similar to the
tunnel response due to the variation of the tunnel depth. As compared in Table 7.2, there
were no bolt failures at a distance of 12.70m (2D) for load cases 1 to 4. In load case 5
for a distance 2D, the tunnel displayed considerable drifting between segments.
(a) Distance = 0 (b) Distance = 1D (c) Distance = 2D

Max. crack width = 2.20mm Max. crack width = 0.40mm Max. crack width = 0.10mm
Nos. of cracks (>0.3mm) = 81674 Nos. of cracks (>0.3mm) = 34 Nos. of cracks (>0.3mm) = 0
Nos. of bolt failure = 35 Nos. of bolt failure = 4 Nos. of bolt failure = 0
Severe damage Moderate damage No damage
Figure 7.3: Failure modes of the tunnel under different ground distances (load case 3)
Page 129
Table 7.2: Maximum crack width in mm / nos. of bolt failures
Distance Case 1 Case 2 Case 3 Case 4 Case 5

0.62 0.97 2.20 3.00 4.40
Distance 0
4 11 35 45 63
0.10 0.31 0.40 0.46 0.65
Distance 1D
0 0 4 11 21
0.01 0.07 0.10 0.35 0.60
Distance 2D
0 0 0 0 2
: No damage : Slight damage
: Moderate damage : Severe damage
The effects of explosive weight versus stand-off distance of the tunnel were used to plot
different damage zones for two situations described in Figure 7.1. Figure 7.4(a) depicts
the variation of damage states due to the influence of the tunnel depth, in which the
stand-off distance is equal to the tunnel depth. Three critical lines are drawn using best-
fitting lines to divide the boundary between different damage zones. Some intermediate
coordinates were selected and relevant simulations were carried out to fine-tune the
critical lines. The following equations mathematically express the relationship between
the stand-off distances to the explosive weight:
Line1 : R  2.94ln( w)  6.78 


Line 2 : R  3.94ln( w)  15.02  6.35m  R  12.70m Equation 7.1
Line3: R  5.40ln( w)  28.42 

where R is the stand-off distance in m, w is the explosive weight in kg.
Line 1 illustrates the threshold border between no damage zone to slight damage zone,
points on the left side of the line produces no damage in the tunnel lining. Line 2 depicts
the threshold border between slight damage zone to moderate damage zone. The
coordinates between Line 1 and 2 denotes that the tunnel exhibited minor cracks with
few bolt failures. Under slight damage, the tunnel can return to operation after some
surface repairs. The third line divides the moderate and the severe damage zones. The
region between Line 2 and 3 produces moderate damage. Table 7.1 shows that 250kg of
TNT explosive can cause moderate damage on shallow buried tunnels. In addition to
Page 130
deep and wider cracks, the tunnel segments suffered a large number of bolt failures
causing drifting or sliding displacement between segments. It may be considered that
those affected segments are at risk of failing. Furthermore, infiltration of water and soil
through the joints can alter the surrounding ground condition and speed up the damage
state. This may require complete replacement of affected segments. Replacing segments
are often associated with time consuming labour and in turn correspondingly cost
intensive repair work.
The right side of Line 3 illustrates that the combination of explosive (w) and stand-off
distance (R) resulted in very severe damage to the structural integrity of segmented
lining. The most severe damage state associated with deep and wider cracks and several
bolt failures. The cracks extended from the bottom of interior surface to top of the
segments due to high bending stresses developed from the explosion. Several bolt
failures caused drifting between segments at the radial joints. When the drifting
exceeded a certain degree, the segments lost contact hoop force transmission capacity at
those radial joints. As a result, the segmented lining was subjected to progressive
collapse as displayed in Figure 7.2(a).
(a) Explosive weight vs tunnel depth for different damage state
Figure 7.4: Critical explosive weight vs. stand-off distance for different damage state
Page 131
(b) Explosive weight vs ground distance for different damage state
Figure 7.4: Critical explosive weight vs. stand-off distance for different damage state
A ground distance of 1.32D in Figure 7.4 (b) is equivalent to the stand-off distance of
1.5D (tunnel depth of 1.5D in Figure 7.4 (a)). Two critical explosive weights for stand-
off distance of 1.5D corresponding to Line 1 and Line 2 are projected as shown in
Figure 7.4(a). The projected values for Line 1 & 2 are 250 and 500kg of TNT
respectively. Similar arrows shown in Figure 14 (b) denote that those values for Line 1
& 2 are significantly greater than 350 and 650kg of TNT respectively. For the same
scaled distances, the study shows that the tunnel is more vulnerable to the surface
explosion from the explosive placed on the centerline of the tunnel than any other
locations on the surface. When the shockwave impacted the tunnel, the sideways
component of the shockwave influenced the tunnel motion in a direction perpendicular
to the ground surface. Due to shallow soil cover above the tunnel crown, the tunnel is
more flexible in crown-invert direction than the lateral direction as the infinite soil
medium constrains the tunnel in the lateral direction. In addition, the previous study in
section 4.2.5 demonstrates that shallower gauge points experienced slightly small peak
pressure than the corresponding gauge points directly below the explosive.
Furthermore, Figure 7.4 (b) assists in developing a protection and safety zone in order to
reduce detrimental effects of any possible surface explosions. Incorporation of proper
and adequate precaution/protection measures by restricting any activities within the
zones will protect the tunnel from credible blasts. For example, the study shows that the
introduction of a safety zone 1.0D distance from the tunnel centerline completely
Page 132
protects the tunnel from an explosion caused by car bombs and the tunnel is even safe
with slight damages if the explosion is caused by SUV/van bomb (Table 1.1).
7.3 MITIGATION OF BLAST IMPACT USING DIFFERENT

APPROACHES
The previous study demonstrated that the surface explosive placed directly above the
tunnel crown creates detrimental effects to the tunnel. In this section, different
mitigation measures as described in Section 7.1 are considered to minimise the tunnel
damage induced by the surface explosion.
7.3.1 Wrapping the tunnel with flexible and compressible barrier consisting of a
layer of polyurethane foam.
The efficacy of a protective barrier material in mitigating the blast effects on the buried
tunnel is studied in this section where a compressible porous barrier consisting of a layer of
polyurethane foam is installed around the tunnel. The barrier material is placed into the
space between the bored soil face and the lining. In the conventional tunnel, the space is
filled by annulus concrete grout.
The polyurethane layer was approximated as isotropic, nearly incompressible, and

hyperplastic rubber material. The constitutive behaviour was described with the formulation
of Blatz and Ko (1962). This material is considered to be rate-independent and implemented
in LS-DYNA as MAT_BLATZ-KON_RUBBER which was used for treating the blast
response of sandwich structures (Bahei-El-Din & Dvorak, 2007). Material properties of
polyurethane foam (Appendix B) used in the study were provided by De et al. (2013).
Figure 7.5 compares the crack responses of the tunnel with concrete grout and the tunnel
with polyurethane foam after 300ms of explosion. In this study, the tunnel depth and
explosive weight were 6.35m (1D) and 250kg of TNT respectively. The tunnel, with
concrete grout displayed moderate damage where the maximum crack width and number of
bolt failure were 0.62mm and 4 respectively, whereas the tunnel with polyurethane
exhibited no damage with allowable minor cracks of 0.1mm.
Page 133
(a) Crack response of conventional tunnel with concrete

grout
(b) Crack response of tunnel with polyurethane
Figure 7.5: Comparison of crack response of tunnels (Load case1 and depth=1D)
The total energy imparted by the shockwave to the tunnel structure is converted into kinetic
energy and strain energy. The strain energy includes elastic strain energy and plastic strain
energy due to permanent deformation. The elastic strain is kept mainly in the inner elements
in the segments. Energy dissipation occurs in the crushable polyurethane foam.
Figure 7.6(a) compares the time histories of the total kinetic energy of the tunnel structure
(total tunnel length of 25.2m) considering the above-defined situation. The reported energy
magnitudes in figures refer to the total volume of segments, reinforcement, bolts and
polyurethane foam. The exterior concrete grout exhibited immediate permanent crushing
with cracks due to blast induced ground shockwave. Therefore, the conventional tunnel
with concrete grout displayed a sharper increase in the total kinetic energy (at a peak value
of 2300kJ) than the tunnel with polyurethane. Replacement of polyurethane (instead of
Page 134
concrete grout) reduced the maximum total kinetic energy by 300kJ. The total strain energy
computed in both cases is compared in Figure 7.6(b). Both cases show an initial peak before
reaching the residual total strain energy. During the peak, the tunnel with polyurethane
dissipated significantly more strain energy by the polyurethane foam during localized
plastic deformation. After the peak, the tunnel with concrete grout dissipated a large amount
of strain energy in the tunnel by cracking of the segments as it is evident from the
comparison of cracks in Figure 7.5. This was reduced by about 50% in the tunnel with
polyurethane, which is entirely due to the polyurethane layer around the tunnel. Therefore,
the behaviour of the tunnel with polyurethane shows better energy absorption over the
conventional tunnel.
2500 2500
Tunnel with concrete grout Tunnel with concrete grout
Total kinetic energy, kJ
Total strain energy, kJ
2000 Tunnel with polyurethane 2000 Tunnel with polyurethane
1500 1500
1000 1000
500 500
0 0
500 550 600 650 700 750 800 500 550 600 650 700 750 800
(a) Kinetic energy vs. time (b) Strain energy vs. time
Figure 7.6: Time history of total kinetic and strain energies
Figure 7.7 illustrates how the critical lines defining the damage zones move with the
introduction of protective barrier of polyurethane foam around the tunnel. The arrow
indicates the line of improvement. Figure 7.7 clearly depicts the improvement in the tunnel
protection in first three damage zones compared to Figure 7.4 (a). Lower part of Line 1
moved to the right-hand-side about the coordinate of (750, 2D) as Line1_Poly. A similar
trend was observed in Line 2 where the slight damage zone moved to a higher order of
explosive range. This resulted in a significant reduction in the moderate damage zone
compared to the conventional tunnel with concrete grout. However, there was no difference
in the severe damage zone.
In the severe damage zone, though the tunnels showed significant improvement in terms of
crack response at a micro level, the tunnel failure is triggered by a significant increase in
Page 135
number of bolt failures in macro level. The immediate ring attracted more blast shock
pressure through the barrier material as the polyurethane is more flexible than the concrete
grouting material in the conventional tunnel. As a result, a large number of circumferential
bolts failed in the immediate ring joint.
Figure 7.7: Critical explosive weight vs. tunnel depth for polyurethane foam
7.3.2 Wrapping the tunnel with a secondary in-situ reinforced concrete lining
A similar study as described in the previous section was considered by replacing the
concrete grout with in-situ reinforced concrete, which acts as a secondary lining to the
segmented tunnel. For the secondary lining, the same grade of concrete as used for the
precast segments was considered. The secondary lining was 150mm thick and double
reinforced with 10mm diameter rebars at 175mm spacing in both circumferential and
transverse directions with 10mm diameter shear links. A similar modelling technique to
that used for the primary segmented lining was used to model the secondary lining.
Both flexible and rigid protective barrier materials provide an adequate level of
protection of the tunnel. Figure 7.8 compares the shockwave interaction with those
barrier materials at different times after the explosion. Propagation of the shockwave
through the soil is visible in two separate situations. At t = 504ms (Figures 7.8(a) &
(b)), the shockwave impacted on the barrier materials. When the shockwave impacted
on the barrier materials, the compression shockwave expanded on the surface of the
flexible polyurethane barrier as shown in Figure 7.8(a). The compression shockwave
reflecting from the polyurethane barrier resulted in a temporary stage of wave
Page 136
stagnation around the tunnel crown as depicted in Figure 7.8(b) whereas, in the rigid RC
barrier, the reflected shockwave was quickly moved out of the surface as displayed in
Figure 7.8(d).
(a) Polyurethane barrier @ t = 504ms (b) RC barrier @ t = 504ms
(c) Polyurethane barrier @ t = 505ms (d) RC barrier @ t = 505ms
(e) Polyurethane barrier @ t = 506ms (f) RC barrier @ t = 506ms
Figure 7.8: Comparison of shockwave interaction with the barrier materials
While the flexible barrier material provides protection by allowing the material to
deform and absorb the blast energy, the rigid RC barrier provides passive blast
protection by allowing material to crack and reflect the shockwave. Figure 7.9 illustrates
that the RC barrier protected tunnel against low-energy blast impact. The critical Line 1
equally moved to the right by expanding the no damage zone while shrinking the slight
damage zone. Changes in the moderate damage zone are insignificant. Mitigation
measure by replacing the concrete grout by the secondary RC layer has not altered the
severe damage zone, although the numbers of circumferential bolt failures were
significantly reduced by the grater load distribution capacity of the secondary layer.
Page 137
Figure 7.9: Critical explosive weight vs. tunnel depth for Reinforced Concrete
secondary lining
7.3.3 Replacing Reinforced Concrete (RC) segments by Steel Fibre Reinforced

Concrete (SFRC) segments.
This section examines how the tunnel responds to the surface blast when its segments are
generated from Steel Fibre Reinforced Concrete (SFRC) without rebars. A similar model
described in the earlier section was considered. SFRC was modelled using the same
material model of MAT_WINFRITH_CONCRETE which was validated and used for
treating the blast response of SFRC slab (Ågårdh, 1997). SFRC material parameters for the
segmental lining are provided by Beňo and Hilar (2013) as shown in Table 7.3.
Table 7.3: Material properties of SFRC
Parameter Value
Density (g/cm3) 2.29

Unconfined compressive strength (MPa) 57.2
Tensile strength (MPa) 4.2
Poisson’s ratio 0.2
Specific fracture energy (N/m) 7500
After shockwave impact on the tunnel, a number of cracks were visible in the concrete
grout whereas no cracks appeared in the segments for all load cases because of higher
Page 138
fracture energy and tensile strength. Figure 7.10 shows the crack formation at the end of the
simulation for Load case1 (refers to Table 5.1) and depth=1D. The tunnel segments were
deformed while keeping the numbers of bolt failures small than the conventional RC
segmented tunnels. For instance, the tunnel buried at a depth of 6.35m showed only one
radial bolt failure for load case 1 whereas the tunnel with RC segments displayed four bolt
failures for the same situation.
Figure 7.10: The crack response of SFRC tunnel (Load case1 and depth=1D)
As the intensity of the shockwave impacting the tunnel increased, SFRC segments
displayed rigid response compared to the RC segmented tunnel. The dynamic response of
the tunnel lining is governed by the flexural stiffness defined as EI where E is the Young’s
modulus and I is the second moment of area. Though E is nearly the same for both
segments, excessive crack width opening in RC segments largely reduced its I value and
resulted in increased flexibility in RC tunnel response. Figure 7.11 shows the influence of
the critical lines by replacing the conventional RC with SFRC segments. The movement of
Line 1 to the right displayed a great expansion in the area of no damage zone. Similarly, the
movement of Line 2 shifted the slight damage zone to the right. As a result, the moderate
damage zone significantly shortened. For instance, (750, 1.5D) coordinate in the moderate
damage zone is moved to slight damage zone due to the replacement of SFRC segments.
However, there was no change in severe damage zone.
Page 139
Figure 7.11: Critical explosive weight vs. tunnel depth for SFRC lining
7.3.4 Introducing energy absorbing flexible honeycomb elements between radial

joints.
In the bored tunnels, there are no flexible (compressible) elements of type used in
conventional tunnelling. However, in this section, an adjustable lining system has been
developed and applied based on honeycomb flexible (compressible) elements principle in
order to minimise the blast effects on the segments by allowing them to expand along the
circumference and rotate about the joints. The flexible (compressible) elements are
expressly used to take up the blast load in order to reach equilibrium before the segments
are damaged. The parts consisting of the flexible elements can be integrated into the
segment on one side or on both sides and they form a complete segment ring as shown in
Figure 7.12.
In this case, every single flexible part was incorporated into the segment on right side
(circling the ring in clockwise direction) and the flexible part was cast together with
appropriate anchor keys into the segment. The other side of the flexible part was connected
to the segment face using bolts during the installation. The radial bolts were M24 grade 8.8
straight bolts. The ring joints were similar to the conventional tunnel. The flexible parts
were made of a series of cells that were assembled using welded vertical and horizontal 4
mm thick mild steel plates as shown in Figure 7.12(a).
Page 140
(a) Cut section of a flexible part (b) A ring with flexible parts
Figure 7.12: Arrangement of flexible parts in a ring
The mild steel plates were modelled as shell elements using MAT_PLASTICIY_
KINEMATIC material model. The fitted faces between the segments and the flexible parts
were simulated using CONTACT_TIED_NODES_SURFACE_OFFSET type contact
whereas the free surfaces (at the radial joint) between the segments and the flexible parts
were modelled using CONTACT_AUTOMATIC_SURFACE_TO_SURFACE type
contact. As one end of the bolt was bolted to the flexible part, the connection was
considered as fully fixed.
Figure 7.13 shows the blast influence diagram by varying both the explosive mass and
tunnel depth with the introduction of flexible honeycomb elements between the radial
surfaces of the segments. It can be clearly seen that the no damage zone has significantly
increased. In this zone, deformation of flexible parts dominated the global deformation. The
flexible parts were crushed and deformed in the circumferential direction in order to allow
the neighbouring segments to expand and rotate at their radial interface. After reaching a
state of equilibrium, the segments started to crack due to combined bending and hoop
stresses. Crushing of flexible elements has significantly delayed the commencement of
crack propagation in the segments compared to the conventional segmental lining.
As the intensity of shockwave impacting the tunnel increased, both the flexible parts and
segments responded simultaneously to the blast load, resulting in no improvement in the
Page 141
crack response. In this tunnel, the flexible parts were not only deformed due to the
circumferential hoop stress, but also due to the shear force transmission at the radial joints.
The shear deformation of the flexible part in the radial direction has significantly reduced
the number of radial bolt failures. For example, the coordinate (250, 1D) in the conventional
tunnel denoted 0.64mm maximum crack width and four number of bolt failures. With
introduction of flexible parts, the same coordinate moved to slight damage zone as it
displayed 0.6 mm maximum crack width and no bolt failures. Figure 7.14(a) illustrates the
deformed tunnel profile and the closest flexible part showed significant plastic damage. As
shown in Figure 7.13, the slight damage zone displayed a slight improvement as the
difference between Line 2 and Line 2_Honey are insignificant.
Figure 7.13: Critical explosive weight vs. tunnel depth for flexible honeycomb elements
In the severe damage zone, immediately after the explosion the segments were damaged
before the flexible parts deformed and absorbed significant amount of blast energy.
After that, the segments together with the flexible parts resisted the blast load. Crushing
of the flexible parts in the circumferential direction increased the segments’ bending
ability. As a result, segments displayed a large number of deep and wider cracks. For
example, a tunnel with 1D depth under 750kg of TNT surface explosive displayed a
large number of full depth cracks as shown in Figure 7.14(b). The similar situation in
the conventional tunnel displayed a large drifting between the crown segments as shown
in Figure 7.2(a). However, in this case, the drifting response is significantly reduced due
to the introduction of the flexible part which deformed in the radial direction, but saved
the number of radial bolt failures. As the blast response is severe in terms of the overall
Page 142
crack response compared to the conventional tunnel, the coordinate (750,1D) is still in
the severe damage zone. As displayed in Figure 7.13, increased soil confinement of the
tunnel buried at 1.5D depth slightly reduced the upper limit of the severe damage zone.
(a) Deformed shape of the tunnel buried in 1D depth for load case 1
(b) Deformed shape of the tunnel buried in 1D depth for load case 3
Figure 7.14: Tunnel deformation after 300 ms of the explosion for different load cases
7.3.5 Introducing a reinforced concrete buster slab above the tunnel crown.
This is an external blast mitigation measure combating the shockwave before reaching the
tunnel structures. This method can be applied to both a newly designed tunnel and to blast
retrofitting of existing tunnel structures. In this method, a reinforced concrete slab (buster
slab) was buried between the tunnel crown and the ground surface and at the exact center of
the tunnel as shown in Figure 7.15(a). The buster slab was 8.5 m width, 14.0m long and
Page 143
90cm thick and reinforcement details are shown in Figure 7.15(b). The same concrete
strength grade as used for producing the segments was used to make the buster slab.
In this modelling, the interaction between the soil and the buster slab was modelled using
CONTACT_AUTOMATIC_ SURFACE_TO_ SURFACE type contact which allowed the
sliding and separation at the contact interface.
(a) Incorporated buster slab model
(b) Reinforced concrete buster slab model
Figure 7.15: Numerical models
Figure 7.16 illustrates the shockwave interaction with the buster slab for a situation, in
which the tunnel depth and the explosive weight were 6.35m (1D) and 250kg of TNT
respectively. Following the impact of a shockwave on the buster slab, a significant amount
of the shockwave refraction occurred beneath the slab as shown in Figure 7.16(b).
Page 144
(a) Shockwave after 2ms of explosion
(b) Shockwave after 3ms of explosion
(c) Shockwave after 4ms of explosion
(d) Shockwave after 5ms of explosion
Figure 7.16: Shockwave propagation through the buster slab-Depth= 1D & Load case 1
Page 145
As displayed in Figure 7.16(c), a reflection of incident shockwave at the slab surface caused
considerable negative pressure in the soil above the top of the buster slab. As a result, a
slight gap remained between the soil and the slab. Besides the small gap opening, the slab
moved with the soil due to the soil confinement. The reflected wave interacted with
oncoming incident wave disturbing the spherical pattern of the reflected wave near to the
crater which is shown in Figure 7.16(d). Overall, the buster slab significantly reduced the
intensity of the shockwave striking the tunnel through a series of reflections and rarefaction
of the shockwave.
The blast influence diagram is shown in Figure 7.17. This mitigation approach shows a
significant improvement in all the damage zones. Expansion of no damage zone illustrates
that the shallow buried tunnel with the soil cover of 6.35m can resist up to 250kg of TNT
without critical damages. The same tunnel can even resist up to 750kg of TNT with
moderate damages. The introduction of the buster slab has dramatically reduced the
negative pressure around the tunnel crown whereas, in the conventional tunnel without the
buster slab (for 750kg of TNT), segments moved relative to each other and left extreme
offsets between segments due to the present of negative pressure. Even in the severe
damage zone, maximum crack widths and number of bolt failures were significantly
reduced compared to the conventional tunnel without the buster slab.
Figure 7.17: Critical explosive weight vs. tunnel depth for buster slab introduction
Page 146
7.4 DISCUSSION OF NUMERICAL RESULTS
The analysis was further extended to predict the blast response of the segmented tunnel to
different scale distances by varying the tunnel depth as well as the ground distance of
explosive from the tunnel centreline. The tunnel was considered as buried in saturated soil.
Four damage categories for the segmented tunnel lining subjected to surface blast were
investigated, such as no damage, slight damage, moderate damage and severe damage. The
numerical results mathematically transformed to the critical lines to represent the critical
explosive weight versus the stand-off distance relationships corresponding to the four
damage states. It is clear that the tunnel depth as well as the ground distance of the
explosive from the tunnel centreline have significant influence on the damage levels of
tunnel response. The obtained analytical curves allow a quick and simple assessment of the
vulnerability of buried tunnels subjected to surface blast. Comparison of critical lines for an
identical scale distances illustrates that the tunnel lining is vulnerable to surface explosions
which occurred directly above the centre of the tunnel than those that occurred at any
equivalent scale distances by moving the explosive away from the tunnel centre on the
ground. For the tunnel considered in this study, the safe explosive weight to resist the
explosion that occurred directly above the centre of the tunnel are recommended: if a tunnel
is buried at a depth of 6.35m, the safe explosive weight to avoid severe damage should be a
maximum of 625kg of TNT, whereas for a depth of 9.52m, the safe explosive weight is
1125kg of TNT which is almost double. Providing a safety zone (a distance of 6.35m on
either side of the tunnel centreline) on the ground will protect the shallow buried tunnel
from severe damage from up to 1125kg of TNT explosive.
Different blast mitigation measures were included in this study to decrease the vulnerability
by adopting prevention and mitigation measures to reduce the structural damage and to
increase the coping capacity. Five mitigation approaches are discussed in this study as
follows:
Page 147
7.4.1 Wrapping the tunnel with a flexible and compressible barrier consisting of
a layer of polyurethane foam.
A compressible porous barrier consisting of a layer of polyurethane foam was used to fill
the circular annular gap between the bored soil face and the lining. Energy absorption
capability of the polyurethane layer significantly reduced the tunnel damage in terms of
cracks and bolt failure. The first two critical lines were moved to the right side of a higher
order of explosive range. For the tunnel considered in this study, the following
improvements are suggested: if a tunnel depth is 6.35m, it can resist up to 500kg of TNT
with slight damage, whereas the conventional tunnel can resist only up to 230kg of TNT.
The tunnels did not show any improvement for high-energy blast impact. As a result, the
severe damage zone remained unchanged. In this zone, the immediate ring in the tunnel
lining has attracted more blast load as the polyurethane layer is less stiff to distribute the
blast load to its neighbouring rings.
7.4.2 Wrapping the tunnel with a secondary in-situ reinforced concrete lining
The annular gap was filled with a secondary reinforced concrete lining to protect the
primary segmented lining. This barrier provides a passive blast protection by allowing the
secondary lining to crack and reflecting the shockwave. It also helped to redistribute the
blast load to its neighbouring rings. This approach significantly improved the blast
performances against low-energy blast impact. For example, if a tunnel depth is 6.35m, it
can resist up to 240kg of TNT with no damage, whereas the conventional tunnel can resist
only up to 85kg of TNT. However, there was no significant difference in high-energy blast
impact since the severe damage zone remained unchanged. It should be noted that the
secondary lining approach is more expensive and time consuming than the spraying the
concrete grout. Since the annular gap (the thickness is 150mm in this study) is
comparatively smaller than the lining thickness, the capping may not be sufficient to resist
the blast induced bending stresses. Increasing the annular gap may not be feasible with the
conventional TBM.
Page 148
7.4.3 Replacing Reinforced Concrete (RC) segments by Steel Fibre Reinforced

Concrete (SFRC) segments.
The tunnel lining was constructed using Steel Fibre Reinforced Concrete (SFRC) segments
without reinforced bars. After the explosion, the segments were free from major cracks due
to high specific facture energy than in the conventional reinforced concrete (RC). It was
observed that, for low-energy blast impact, SFRC lining displayed a small number of radial
bolt failures compared to the conventional RC lining. As a result, the first two critical lines
were significantly moved to right side, resulting significant reduction in the tunnel damage
than that of previous mitigation measures such as wrapping the tunnel with polyurethane
foam and secondary RC layers. However, for high-energy blast impact, SFRC lining is not
very effective as it has reduced flexural strength compared to the conventional RC. This
resulted in a large number of both radial and circumferential bolt failures. It is suggested
that increasing the number of segments in a ring may help to increase the flexibility of
SFRC segments.
7.4.4 Introducing energy absorbing, flexible honeycomb elements between radial

joints.
Honeycomb flexible (compressible) cell elements were introduced between the radial joints
in order to allow the segments to expand along the circumference and rotate about the
joints. Under low-energy blast impact, crushing of flexible elements has significantly
delayed the commencement of crack formation in the segments as well as the reduced
number of bolt failures compared to the conventional segmented lining. As a result, a tunnel
buried at 9.52m showed there was a significant rise in its no damage resistance from 250kg
to 375kg of TNT. For high-energy blast impact, both the flexible elements and the
segments deformed simultaneously and displayed no improvement in the crack response.
However, deformation of flexible elements in the radial direction considerably minimised
the number of radial bolt failures. By considering the additional costs associated with
introducing the flexible elements, it seems that it is not worth in terms of overall
improvement in the blast performance of the tunnel considered in this study.
Page 149
7.4.5 Introducing a reinforced concrete buster slab above the tunnel crown.
Introduction of a buster slab expressively reduced the vulnerability of the tunnel response to
the surface explosion. The buster slab interrupted and combated the shockwave on its way
to the target. After reflection of the shockwave, a significant amount of rarefaction of the
shockwave transmitted through the buster slab and interacted with the tunnel structure.
As the intensity of the shockwave has significantly reduced, the tunnel displayed huge
improvement in overall blast performance. Movement of all the critical lines to the right
enlarged the area of no damage zone as well as diminished the other damage zones. Overall,
this method is very effective in mitigating blast effects on the tunnel as it can be applied to a
newly designed tunnel and to blast retrofitting of exiting tunnel as well.
7.5 CHAPTER SUMMARY
In this chapter, a parametric study was considered by varying the tunnel depth as well as the
ground distance of explosive from the tunnel centerline. The tunnel was considered as
buried in the saturated soil. The study showed that the tunnel lining is vulnerable to surface
explosions which occurred directly above the centre of the tunnel than those that occurred at
any equivalent scale distances by moving the explosive away from the tunnel centre on the
ground. The finding from the parametric study was further extended to mitigate the blast
effects by introducing different mitigation strategies. The main findings of this study are as
follows:
1. When a tunnel is subjected to a surface explosion, the tunnel depth as well as the ground
distance of the explosive from the tunnel centreline have a significant influence on
tunnel response.
2. If a tunnel is buried at a depth of 6.35m, the safe explosive weight to avoid severe
damage should be less than 625kg of TNT, whereas for a depth of 9.52m, the safe
explosive weight is 1125kg of TNT which is almost double.
3. Providing a safety zone (a distance of 6.35m on either side of the tunnel centreline) on
the ground will protect the shallow buried tunnel from severe damage from up to 1125kg
of TNT explosive.
Page 150
4. For low-energy blast impacts, a strategy of wrapping the tunnel with a flexible and
compressible barrier consisting of a layer of polyurethane foam significantly reduced the
tunnel damage. For example, if the depth of a wrapped tunnel is 6.35m, it can resist up to
500kg of TNT with slight damage, whereas the conventional tunnel can resist only up to
230kg of TNT. The lack of load redistribution capability of the polyurethane did not
make any difference in high-energy blast impacts. As a result, the severe damage zone
remained unchanged.
5. An approach of wrapping the tunnel with a secondary reinforced concrete lining offered
a significant improvement for low-energy blast impact due to its passive blast resistance
as well greater redistribution ability. For example, if the tunnel depth is 6.35m, the
wrapped tunnel can resist up to 240kg of TNT with no damage, whereas the
conventional tunnel can resist only up to 85kg of TNT. For high-energy blast impacts,
there was no change in the severe damage zone.
6. Replacing the conventional reinforced concrete segments by steel fibre reinforced

concrete (SFRC) segments has a significant impact on the blast response. As a result of
improved crack response and reduction in the number of bolt failures, upper boundaries
of both no damage and slight damage zones displayed a significant expansion towards
higher blast level. For example, if the tunnel depth is 9.52m, it can resist up to 875kg of
TNT with slight damage, whereas the conventional tunnel can resist only up to 500kg of
TNT. However, for high-energy blast impact, SFRC lining displayed stiff behaviour as
SFRC has reduced flexural strength compared to the conventional RC. As a result, the
severe damage zone remained unchanged.
7. As a result of the introduction of honeycomb flexible (compressible) cell elements

between the radial joints, crack formations in the segments were delayed and the
numbers of radial bolt failures were reduced for low-energy blast impacts compared to
the conventional segmented lining. For example, the tunnel buried at 9.52m showed
there was a significant rise in its no damage resistance from 250kg to 375kg of TNT. For
high-energy blast impacts, there was no difference in terms of crack response. Hence the
severe damage zone has an insignificant effect.
Page 151
8. Introduction of a buster slab expressively improved the blast performance of the tunnel
in terms of crack formation, number of bolt failure and drifting between segments.
Unlike other approaches mentioned above, this approach significantly influenced the
tunnel response from low to high energy blast impacts as the buster slab considerably
reduced the intensity of shockwave transmission. For example, if the tunnel is buried at a
depth of 6.35m, the safe explosive weight to avoid severe damage is at 750kg of TNT,
whereas the conventional tunnel can resist only up to 625kg of TNT.
Page 152
Chapter 8: Conclusion and Recommendations
Chapter 8: Conclusion and

8
Recommendations
8.1 REVIEW OF RESEARCH WORK
The principal objectives of this thesis were to investigate the performance of segmented
bored tunnels subjected to surface blast loads, and to study the behaviour of joints in the
segmented tunnel. This research developed and applied fully coupled numerical
techniques incorporating Fluid Structure Interaction (FSI) to accurately predict the
dynamic response of segmented tunnels to blast loads using an advanced general
purpose multi-physics simulation software LS-DYNA. Throughout the thesis, the aim
has been accomplished.
Two different numerical techniques (i) coupled FSI in ALE and (ii) coupled SPH-FE
techniques were considered for treating the response of buried tunnels subjected to
surface blast loads. In the coupled FSI in ALE, eight-node solid elements were used with
different spatial discretisation solvers. Lagrangian meshes were used to model the soil and
the tunnel while Eulerian meshes were used separately to model the surrounding air and
explosive. In the coupled SPH-FE, the portion of soil experiencing large deformations
and the explosive were modelled with SPH particles while the rest of the geometry was
modelled with Lagrangian meshes. The surrounding outside space of the explosive was
assumed to be a vacuum. The reinforced concrete segments were modelled using eight-
node solid Lagrangian elements. The reinforcement steel and bolts were modelled using
Hughes-Liu beam elements with cross sectional integration. The reinforcing beam
elements and bolts were modelled as discrete elements immersed in the segment meshes
using Constrained_Lagrangian_In_Solid coupling.
The modelling techniques were validated using previous experimental records. Based
on the comparison of the two techniques in terms of computational efficiency and
accuracy, it was found that the coupled FSI in ALE technique outweighed the coupled
SPH-FE technique to treat surface blast loads. To achieve the basic objectives, a
parametric study was performed to observe the behaviour of the segmented tunnels of
Page 153
various segment geometries, different joint types and number of segments forming the
tunnel ring. Three types of soil: saturated, partially saturated and dry soils were
considered in order to investigate the effect of soil properties on the blast response of a
buried tunnel.
A second phase of the research consisted of investigating the effect of stand-off

distances by varying the tunnel depth as well as the ground distance of the explosive
from the tunnel centreline. Four damage categories of the tunnel lining subjected to a
surface blast were investigated, such as no damage, slight damage, moderate damage
and severe damage. In order to improve damage performance and to decrease the
vulnerability of the tunnel, five different blast mitigation measures were illustrated.
8.2 MAIN FINDINGS OF THIS RESEARCH AND RECOMENDATIONS
The main findings of this research are presented below:
This research developed and applied two different numerical modelling techniques (i)
coupled FSI in ALE and (ii) coupled SPH-FE for treating the blast response of buried
tunnels subjected to surface blast. Comparison of the two techniques using results from
a centrifuge test (De, 2012) shows that the coupled FSI in ALE technique outweighed the
coupled SPH-FE technique in terms of accuracy and computational efficiency. This
provides confidence in the coupled FSI in ALE modelling technique for predicting the
response of segmented bored tunnels subjected to surface blast.
The validated numerical technique was then extended to study the effects of the geometric
shapes of the segments on the blast response of the tunnel. Hexagonal segments (HS)
significantly reduced the drifting between the tunnel rings, but they suffered severe in-
plane drifting within the ring. Interlocking segments (IS) also suffered from the drifting
between rings as well as in-plane drifting. Though the tunnel with rectangular segments
(RS) displayed considerable drifting between the rings, in-plane drifting was
insignificant because of increased flexibility due to the orientation of radial joints.
The blast response of buried tunnel in saturated soil is more severe in terms of crack
formation and bolt failures than the tunnel buried in either partially saturated soil or dry
Page 154
soil when subjected to the same surface explosion. In all soil conditions, the diametric
distortions increase with the explosive mass and the distortions are unrecoverable in all
load cases (from 250 to 1250kg of TNT) treated in this study. As the explosive mass
increased to more than 750kg of TNT, the tunnel buried in the saturated soil displayed
segment-disintegration from neighbouring segments. For weaker blast loads (250kg of
TNT), joint types influenced the tunnel response by allowing the segments to rotate
about the joints, whereas for higher blast loads, the segments were damaged before
triggering the flexibility of the joints. While the flat joints were effective in transferring
hoop stress at the joints, convex-convex joints resulted in less number of bolt failures
and significant improvement in the crack response for weaker blast loading.
As the number of segments increased in a ring, the maximum crack width decreased.
However, as the number of segments further increased to seven segments (six segments
plus a key segment) in a ring, it did not make much difference in the magnitude of the
maximum crack width. Although the tunnel with seven segments in a ring displayed less
number of radial bolt failures, the numbers of circumferential bolt failures were
significantly higher than the tunnel with six segments (five segments plus a key
segment) in a ring.
When a tunnel is subjected to a surface explosion, the tunnel depth as well as the ground
distance of the explosive from the tunnel centreline have a significant influence on
tunnel response. If a tunnel is buried at a depth of 6.35m, the safe explosive weight to
avoid severe damage should be less than 625kg of TNT, whereas for a depth of 9.52m,
the safe explosive weight is 1125kg of TNT which is almost double. Providing a safety
zone (a distance of 6.35m on either side of the tunnel centreline) on the ground will
protect the shallow buried tunnel from severe damage from up to 1125kg of TNT
explosive.
This research also provides recommendations for blast mitigation of the segmental
tunnels subjected to surface explosion. For low-energy blast impact, a strategy of
wrapping the tunnel with a flexible and compressible barrier consisting of a layer of
polyurethane foam significantly reduced the tunnel damage. For example, if the depth of
a wrapped tunnel is 6.35m, it can resist up to 500kg of TNT with slight damage,
Page 155
whereas the conventional tunnel can resist only up to 230kg of TNT. The lack of load
redistribution capability of the polyurethane did not make any difference in the severe
damage zone which remained unchanged in the damage influence diagram.
An approach of wrapping the tunnel with a secondary reinforced concrete lining offered
a significant improvement for low-energy blast impacts due to its passive blast
resistance as well greater load redistribution ability. For example, if a tunnel depth is
6.35m, a wrapped tunnel can resist up to 240kg of TNT with no damage, whereas the
conventional tunnel can resist only up to 85kg of TNT. For high-energy blast impacts,
there was no change in the severe damage zone.
Replacing the conventional reinforced concrete segments by steel fibre reinforced

concrete (SFRC) segments has a significant impact on the blast response. As a result of
improved crack response and reduction in the number of bolt failures, upper boundaries
of both no damage and slight damage zones showed a significant expansion towards
higher blast level. For example, if the tunnel depth is 9.52m, this type of tunnel can
resist up to 875kg of TNT with slight damage, whereas the conventional tunnel can
resist only up to 500kg of TNT. However, for high-energy blast impacts, SFRC lining
displayed stiff behaviour as SFRC has reduced flexural strength compared to the
conventional RC. As a result, the severe damage zone remained unchanged.
As a result of the introduction of Honeycomb flexible (compressible) cell elements

between the radial joints, crack formations in the segments were delayed and the
numbers of radial bolt failures were reduced for low-energy blast impacts compared to
the conventional segmented lining. For example, the tunnel buried at 9.52m showed
there was a significant rise in its no damage resistance from 250kg to 375kg of TNT.
For high-energy blast impacts, there was no difference in terms of crack response.
Hence the severe damage zone has an insignificant effect.
Introduction of a buster slab expressively improved the blast performance of the tunnel
in terms of crack formation, number of bolt failure and drifting between segments.
Unlike other approaches as mentioned above, this approach significantly influenced the
tunnel response from low to high energy blast impacts as the buster slab considerably
Page 156
reduced the intensity of shockwave transmission. For example, if the tunnel is buried at
a depth of 6.35m, the safe explosive weight to avoid severe damage is at 750kg of TNT,
whereas the conventional tunnel can resist only up to 625kg of TNT.
Based on the present study of the response of the segmented tunnel subjected to surface
blast loads, it can be concluded that the present modelling techniques and research
findings can serve as a benchmark reference in future development in this research area
and in the verification and validation of numerical models. This study also made an
effort to provide a confidence level for mitigating the blast effects on the tunnel
response.
8.3 FUTURE RESEARCH
From the experience gained in this numerical research study the following
recommendations are made for future studies in this area:
 Numerical simulation of a tunnel buried in a layered soil profile is recommended

for future studies in order to simulate real field conditions. In the present study,
the soil is considered as a homogeneous single layered soil.
 The present research did not consider the bolt holes in the segments as it is
numerically more expensive. The bolts were modelled as discrete elements
coupled to the solid elements of the segments throughout its length. A detailed
numerical study is required in order to evaluate and verify the influence of bolt
holes in the segments under blast loads.
 Due to the limitation in the non-reflecting boundaries, surcharge loads from

moving vehicle and super imposed dead loads from aboveground structures are
not considered in the present modelling. Therefore, further research is required
to include those loads to simulate the real the conditions.
 In the present thesis, small diameter single tube railway tunnels were considered
to investigate the blast effects. However, there are many large diameter
segmented tunnels which need to be investigated.
Page 157
 Some of the blast mitigation measures considered in this study are expensive.
Since soils are capable of absorbing blast energy, extensive research should be
conducted in future studies for mitigating the blast effects on the tunnels by
introducing appropriate backfilling materials to cover the top ground surface of
the tunnels.
Page 158
References
References
Abbo, A. J., & Sloan, S. W. (1995). A smooth hyperbolic approximation to the Mohr-
Coulomb yield criterion. Computers & Structures, 54(3), 427-441. doi:
http://dx.doi.org/10.1016/0045-7949(94)00339-5
Ågårdh, L. (1997). Fe-Modeling of Fibre Reinforced Concrete Slabs Subjected to Blast
Load. J. Phys. IV France, 07(C3), C3-723-C723-728.
Alia, A., & Souli, M. (2006). High explosive simulation using multi-material formulations.
Applied Thermal Engineering, 26(10), 1032-1042. doi:
http://dx.doi.org/10.1016/j.applthermaleng.2005.10.018
Ambrosini, D., Luccioni, B., & Danesi, R. (2004). Influence of the soil properties on craters
produced by explosions on the soil surface. Computational Mechanics, XXIII, 571-
590.
Ambrosini, R. D., & Luccioni, B. M. (2005). Craters Produced by Explosions on the Soil
Surface. Journal of Applied Mechanics, 73(6), 890-900. doi: 10.1115/1.2173283
Arulmoli, K., VELACS Project, Earth Technology Corporation, National Science
Foundation (U.S.) (1992). VELACS Verification of Liquefaction Analyses by
Centrifuge Studies Laboratory Testing Program: Soil Data Report. Earth
Technology Corporation.
Bahei-El-Din, Y. A., & Dvorak, G. J. (2007). Behavior of Sandwich Plates Reinforced with
Polyurethane/Polyurea Interlayers under Blast Load. Journal of Sandwich
Structures and Materials, 9, 261-281.
Baylor, J. T. (1992). Parameters Affecting Loads on Buried Structures Subjected to
Localized Blast Effects: Army Engineer Waterways Experiment Station Vicksburg
Ms Structures Lab.
Beňo, J., & Hilar, M. (2013). Steel Fibre Reinforced Concrete for Tunnel Lining-Verification
by Extensive Laboratory Testing and Numerical Modelling. Acta Polytechnica
Journal of Advanced Engineering, 53(4), 329-337.
Bessette, G. C. (2004). Modeling Coupled Blast/Structure Interaction with Zapotec,
Benchmark Calculations for the Conventional Weapon Effects Backfill (CONWEB)
Tests: Sandia National Laboratories Report SAND20044096.
Blatz, P. J., & Ko, W. L. (1962). Application of Finite Elastic Theory to the Deformation of
Rubbery Materials. Transactions of the Society of Rheology., VI, 223-251.
Bull, J. W., & Woodford, C. H. (1998). Camouflets and their effect on runway support.
Computers & Structures, 69(6), 695-706. doi: http://dx.doi.org/10.1016/S0045-
7949(98)00135-7
Conterterrorism 2014 Calendar (2014), Bomb Threat Stand-off Distances. Retrieved March
2014, from http://www.nctc.gov/site/technical/bomb_threat.html
Cooper, P. W., & Kurowski, S. R. (1996). Introduction to the Technology of Explosives:
VCH.
Darwin, D., Barham, S., Kozul, R., & Luan, S. (2001). Fracture energy of high-strength
concrete. ACI Materials Journal, 98(5).
Davies, M. C. R. (1994). Dynamic Soil Structure Interaction Resulting from Blast Loading.
Paper presented at the Centrifuge 94. Rotterdam:Balkema.
Davies, M. C. R., & Williams, A. J. (1992). Centrifuge Modelling the Protection of Buried
Structures Subjected to Blast Loading. Structures Under Shock and Impact II, 663-
674.
De, A. (2012). Numerical simulation of surface explosions over dry, cohesionless soil.
Computers and Geotechnics, 43(0), 72-79. doi:
http://dx.doi.org/10.1016/j.compgeo.2012.02.007
Page 159
References
De, A., Morgante, A. N., & Zimmie, T. F. (2013). Mitigation of Blast Effects on
Underground Structure Using Compressible Porous Foam Barriers. Paper presented
at the Poromechanics V. http://ascelibrary.org/doi/abs/10.1061/9780784412992.116
De, A., & Zimmie, T. F. (2011). Modeling of Surface Blast Effects on Underground
Structures. Paper presented at the Geo-Frontiers Congress.
Department of the Army (1986). Fundamental of Protection Design for Conventional
Weapons, TM5-855-1, US Army Engineers Waterways Experimental Stations,
Vicksburg.
Engineering Group (2010). Civil Design Criteria for Road and Rail Transit Systems, Land
Transport Authority, Singapore.
Feldgun, V. R., Kochetkov, A. V., Karinski, Y. S., & Yankelevsky, D. Z. (2008). Internal
blast loading in a buried lined tunnel. International Journal of Impact Engineering,
35, 172 - 183.
Gingold, R. A., & Monaghan, J. J. (1977). Smoothed particle hydrodynamics-theory and
application to non-spherical stars. Monthly Notices of the Royal Astronomical
Society, 181, 375-389.
Glasstone, S., & Dolan, P. J. (1977). The Effects of Nuclear Weapons (3 ed.): United States
Department of Defense and the Energy Research and Development Administration.
Gui, M., & Chien, M. (2006). Blast-resistant Analysis for a Tunnel Passing Beneath Taipei
Shongsan Airport–a Parametric Study. Geotechnical and Geological Engineering,
24(2), 227-248.
Hallquist, J. O. (2006). LS-DYNA THEORY MANUAL, Livermore Software Technology
Corporation(LSTC). California, USA.
Hayes, P. G. (1989). Backfill Effects on Response of Buried Reinforced Concrete Slabs.
Vicksburg, Mississippi: Waterways Experiment Station, Corps of Engineers.
Higgins, W., Chakraborty, T., & Basu, D. (2012). A high strain-rate constitutive model for
sand and its application in finite-element analysis of tunnels subjected to blast.
International Journal for Numerical and Analytical Methods in Geomechanics
(2012).
Higgins, W. T. (2011). Development of a high strain-rate constitutive model for sands and
its application in finite element analysis of tunnels subjected to blast. (M.S. thesis),
University of Connecticut, Storrs, Connecticut, USA.
Jayasinghe, L. B., Thambiratnam, D. P., Perera, N., & Jayasooriya, J. H. A. R. (2013).
Computer simulation of underground blast response of pile in saturated soil.
Computers & Structures, 120(0), 86 - 95.
Jayasinghe, L. B., Thambiratnam, D. P., Perera, N., & Jayasooriya, J. H. A. R. (2014). Effect
of soil properties on the response of pile to underground explosion. IABSE,
Structural Engineering International Journal, (in press).
Jayasooriya, Ruwan, David P. Thambiratnam, Nimal J. Perera and Vladis Kosse. 2011. Blast
and residual capacity analysis of reinforced concrete framed buildings. Engineering
Structures 33 (12): 3483-3495.
Joint Departments of the Army, the Navy and the Air Force (1990). Structures to Resist the
Effect of Accidental Explosions, TM5-1300/NA VFAC P-397/AFR 88-22.
Kinney, G.F. and Graham, K.J. (1985). Explosive shocks in air (2nd edition). Springer-
Verlag, 282pp.
Kramer, S. L. (1996). Geotechnical Earthquake Engineering. USA: Prentice-Hall.
Krauthammer, T., & Chen, Y. (1989). Soil-structure interface effects on dynamic interaction
analysis of reinforced concrete lifelines. Soil Dynamics and Earthquake
Engineering, 8(1), 32-42.
Kutter, B. L., O'Leary, L. M., & Thompson, P. Y. (1988). Gravity-Scaled Tests on Blast-
Induced Soil-Structure Interaction. Journal of Geotechnical Engineering,, 114, 431-
447.
Laine, L., & Sandvik, A. (2001). Derivation of Mechanical properties for sand. Proceedings
of the 4th Asia-Pacific Conference on Shock and Impact Loads on Structures, CI-
Premier PTE LTD, Singapore, 361-368.
Page 160
References
Lee, W. Y. (2006). Numerical modeling of blast-induced liquefaction. (doctoral dissertation),

Department of Civil and Environmental Engineering, Brigham Young University,
USA.
Lewis, B. A. (2004). Manual for LS-DYNA Soil Material Model 147. Federal Highway
Administration, McLEAN, VA.
Liu, H. (2009). Dynamic Analysis of Subway Structures Under Blast Loading. Geotechnical
and Geological Engineering, 27(6), 699-711.
Liu, H. (2012). Soil-Structure Interaction and Failure of Cast-Iron Subway Tunnels
Subjected to Medium Internal Blast Loading. Journal of Performance of Constructed
Facilities, Vol. 26, 691-701.
Livermore Software Technology Corporation, LSTC (2007). LS-DYNA Keyword User's
Manual v971. California, USA.
Luccioni, B., Ambrosini, D., Nurick, G., & Snyman, I. (2009). Craters produced by
underground explosions. Computers & Structures, 87(21-22), 1366-1373.
Lucy, L. (1977). A numerical approach to the testing of the fission hypothesis. The
Astronomical Journal, 82, 1013-1024.
Malvar, L. J., & Crawford, J. E. (August 1998). Dynamic Increase Factors for Concrete.
Paper presented at the Twenty-Eight DDESB Seminar, Orlando, FL,.
Matuska, D. A. (1984). HULL Users' Manual. Air Force Armament Laboratory, Florida,
USA: Orlando Technology, Incorporated, Florida, USA.
McMaster, W., H. (1984). Computer codes for fluid-structure interactions. American
Society of Mechanical Engineers.
Monaghan, J. J. (1992). Smoothed particle hydrodynamics. Annual review of astronomy and
astrophysics, 30, 543-574.
Monaghan, J. J. (1994). Simulating Free Surface Flows with SPH. Journal of Computational
Physics, 110(2), 399-406. doi: http://dx.doi.org/10.1006/jcph.1994.1034
Morris, J.P., M.B. Rubin, S.C. Blair, L.A. Glenn and F.E. Heuze. (2004). Simulations of
underground structures subjected to dynamic loading using the distinct element
method. Engineering Computations 21 (2/3/4): 384-408.
Muir Wood, A. M. (1975). The circular tunnel in elastic ground. Geotechnique, 25(1), 115 –
127.
Munfah, N. (2009). Safety and Security of Tunnels and Underground Transportation
Facilities. Parsons Brinckerhoff, National Tunnelling Practice Leader, New York,
U.S.A.
Naesgaard, E., Byrne, P., M, & Wijewickreme, D. (2007). Is P-wave velocity an indicator of
saturation in sand with viscous pore fluid? International Journal of Geomechanics,
7(6), 437-443.
Olarewaju, A. J. (2012a). Study on the Impact of Varying Degrees of Underground
Accidental Explosions on Underground Pipes by Simulation. Earth Science
Research, Published by Canadian Centre of Science and Education, Vol.1, No. 2,
189-199. doi: 10.5539/esr.v1n2p189
Olarewaju, A. J. (2012b). Effect of Loose Sand and Dense Sand on the Response of
Underground Empty Pipes due to Accidental Explosions. Electronic Journal of
Geotechnical Engineering (EJGE), Vol. 17 Bundle G, 879-891.
Olarewaju, A. J. (2013). Prediction and Assessment of Loads from Various Accidental
Explosions for Simulating the Response of Underground Structures using Finite
Element Method. Electronic Journal of Geotechnical Engineering (EJGE), Vol. 18
Bundle B, 375-396.
Oluokun, Francis A, Edwin G. Burdette, & Deatherage, J. H. (1991). Young's modulus,
Poisson's ratio and compressive strength relationships at early ages. ACI Materials
Journal, 88, 3-9.
Ortman, & Catherine, M. (2008). The Effect of Diameter on Dynamic Seabed Penetration
(ADA486586 ed.): NAVAL ACADEMY ANNAPOLIS MD.
Ottosen, N. S. (1977). A Failure Criterion for Concrete. Journal of Engineering Mechanics
Division, 103(4), 527-535.
Page 161
References
Peroni, M., Peroni, L., & Dallocchio, A. (2009). Thermo-mechanical model identification of
a strengthened copper with an inverse method DYMAT 2009-9th International
Conference on the Mechanical and Physical Behaviour of Materials under Dynamic
Loading (Vol. 2, pp. 1367-1373). France: EDP Sciences.
Procházka, P., P, Kravtsov, A., N, & Peskova, S. (2008). Blast impact on structures of
underground parking: Wessex Institute of Technology, UK.
Saleh, M., & Edwards, L. (2011). Application of a soil model in the numerical analysis of
landmine interaction with protective structures Paper presented at the 26th
International symposium on blastics MIAMI.
Sandford, T. C. (2000). Soil-Structre Interaction of buried strcutres A2K04: Committeee on
Subsurface Soil-Structure Interaction, Universtiy of Maine, U.S.A.
Schwer, L. (2011). The Winfrith Concrete Model: Beauty or Beast? Insights into the Winfrith
Concrete Model. Paper presented at the 8th European LS-DYNA Users Conference,
Strasbourg, France.
Shin, J. H., Moon, H. G., & Chae , S. E. (2011). Effect of blast-induced vibration on existing
tunnels in soft rocks. Tunnelling and Underground Space Technology, 26(1), 51-61.
Sivalingam, K., Thambiratnam, D. P., & Gallage, C. (2014). Performance of Buried Tunnels
Subjected to Surface Blast Incorporating Fluid Structure Interaction. ASCE Journal
of Performance of Constructed Facilities. doi: doi:10.1061/(ASCE)CF.1943-
5509.0000585
Skyscrapercity.com (2013). A cross-section of Moorgate Retrieved March 2013, from
http://www.skyscrapercity.com/showthread.php?t=236257&page=168
Thiyahuddin, I., Gu, Y., & Thambiratnam, D. P. (2012). A coupled SPH/FEM analysis of
portable water ﬁlled barriers. Paper presented at the 4th International Conference on
Computational Methods (ICCM2012), Gold Coast, Qld.
Wang, J.N. (1993). Seismic Design of Tunnels ( A Simple State-of-the-Art Design Approach).
Parsons Brinckerhoff Quade & Douglas, Inc., U.S.A.
Wang, J. (2001). Simulation of Landmine Explosion Using LS-Dyna3d Software: Benchmark
Work of Simulation of Explosion in Soil and Air: DSTO Aeronautical and Maritime
Research Laboratory.
Wang, Z., Hao, H., & Lu, Y. (2004). A three-phase soil model for simulating stress wave
propagation due to blast loading. International Journal for Numerical and Analytical
Methods in Geomechanics, 28(1), 33-56.
Wang, Z., Lu, Y., Hao, H., & Chong, K. (2005). A full coupled numerical analysis approach
for buried structures subjected to subsurface blast. Computers & Structures, 83(4-5),
339-356.
Wen, D., Poh, J., & Ng, Y. W. (2004). Design considerations for bored tunnels at close
proximity. Tunnelling and Underground Space Technology, 19(4–5), 468-469. doi:
http://dx.doi.org/10.1016/j.tust.2004.02.072
Whittaker, J. P. (1987). Centrifugal and numerical modeling of buried structures. Volume 3.
A centrifuge study of the behavior of buried conduits under airblast loads. Final
report: Colorado Univ., Boulder (USA). Dept. of Civil, Environmental, and
Architectural Engineering.
Xu, J.-x., & Liu, X.-l. (2008). Analysis of structural response under blast loads using the
coupled SPH-FEM approach. Journal of Zhejiang University - Science A, 9(9),
1184-1192.
Yang, Y., Xie, X., & Wang, R. (2010). Numerical simulation of dynamic response of
operating metro tunnel induced by ground explosion. Journal of Rock Mechanics
and Geotechnical Engineering, 2(4), 373-384.
Yankelevsky, D. Z., Karinski, Y. S., & Feldgun, V. R. (2011). Re-examination of the shock
wave's peak pressure attenuation in soils. International Journal of Impact
Engineering, 38(11), 864 - 881.
Zhongqi, W., Yong, L., & Hong, H. (2004). Numerical Investigation of Effects of Water
Saturation on Blast Wave Propagation in Soil Mass. Journal of Engineering
Mechanics, 130(5), 551-561.
Page 162
Appendices
APPENDIX A
Controlled single concrete element simulation
This study uses a single element of a concrete, as shown in Figure A.1(a), to verify
the output of crack width measurement and its unit in LS-DYNA which did not
describe the unit of the crack width output in its technical manual. A consistent unit
system is used in this study: mass (g), length (mm), time (ms), force (N) and stress
(MPa). As described by Schwer (2011), a similar study is conducted with his
material properties. The two parts of the element are allowed to move apart due to
the prescribed displacement, as illustrated in Figure A.1(b). When the displacement
exceeds the crack width (w) at failure of 0.127mm, the failure occurs on the element.
0.2
0.15
0.1
Displacement, mm
0.05
0
0 10 20 30 40
-0.05
-0.1
-0.15
-0.2
Time, ms
(a) A single solid element (b) Prescribed axial displacement (Schwer, 2011)
Figure A.1: A single element uniaxial test
Figure A.2 shows the simulation results. As the displacement increased in the tension
portion, a single horizontal crack appeared perpendicular to the displacement in the
element at 9.05ms and it satisfied the crack width displacement criterion, i.e. w =
0.127mm. As the displacement was reversed, closing of the horizontal crack
occurred immediately after 30ms in the compression portion and it allowed to resist
the compression. Consequently, two orthogonal cracks formed at 30.1ms when the
element reached its unconfined compressive strength.
The maximum crack width of the first crack is consistent with unit system as both the
prescribed displacement and the maximum crack width were nearly the same. The
Page 163
Appendices
orthogonal tensile crack width was very small in micrometres. This finding
demonstrates that the crack width output unit in LS-Dyna is consistent with the
adopted unit system. For example, if the input unit for the length is cm, the output
crack width would be the same in cm.
(a) Single tensile crack (b) Double orthogonal cracks
(d) The maximum crack width= 0.127mm (c) The maximum crack width= 0.352m
Figure A.2: Crack formations and corresponding maximum crack widths
Page 164
Appendices
APPENDIX B
Input parameters of concrete material model
B. 1 Concrete Grade 60
B. 2 Concrete Grade 15
Page 165
Appendices
B. 3 Polyurethane foam
Unit system: Mass (g), Length (cm), Time (s) and Stress (Mbar)
Page 166

3-Sivalingam Koneshwaran Thesis

Uploaded by

Copyright:

Available Formats

3-Sivalingam Koneshwaran Thesis

Uploaded by

Document Information

Original Description:

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

3-Sivalingam Koneshwaran Thesis

Uploaded by

Copyright:

Available Formats

BLAST RESPONSE AND SENSITIVITY

ANALYSIS OF SEGMENTAL TUNNEL

FACULTY OF SCIENCE AND ENGINEERING FACULTY

QUEENSLAND UNIVERSITY OF TECHNOLOGY

FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF

I wish to express my deepest gratitude to my principal supervisor, Professor David

I gratefully acknowledge the financial support granted by Australian Postgraduate

Increasing threat of terrorism highlights the importance of enhancing the resilience of

International Journal Papers

 Sivalingam, K., Thambiratnam, D. P., and Gallage, C (2014), “Numerical

International Conference Papers

 Sivalingam, K., Thambiratnam, D. P., and Gallage, C (2013), “Response of a

Figure 1.1: Comparison of types of terrorist attacks on tunnel (GTD, 2013)

Figure 1.2: Schematic view of London underground transit tunnels (Skyscrapercity.com,

Furthermore, the crowded nature of underground developments increase risk to both

A Vehicle Borne Improvised Explosive Device (VBIED) is commonly used as a

Vehicle Explosives capacity

* Section 2.2.2 describes TNT (Trinitrotoluene).

instantaneous vibration as damping from the surrounding medium is ineffective. The

Figure 1.3: Schematic view of Precast Segmented Rings

1.2 RESEARCH PROBLEM

(a) Propagation of shockwawe (b) Shockwave striking tunnel crown

(c) Reflection of shockwave

Figure 1.4: Interaction of shockwave with the tunnel

As the surrounding pressure changes through a concentrated load, distributed positive

1.3 AIMS AND OBJECTIVES

 To develop fully coupled numerical models incorporating different material

1.4 SIGNIFICANCE OF THE RESEARCH

1.5 THESIS OUTLINE

 In Chapter 2, a comprehensive literature review of blast loads and resulting

(ALE) techniques and Smooth Particle Hydrodynamics (SPH) techniques as

2Chapter 2: Literature Review

This chapter explores the current knowledge on explosion, shockwave propagation in

2.2 EXPLOSION AND RELATED PHENOMENON

2.2.1 Explosives and TNT equivalent

Explosive TNT Equivalent

2.2.2 Blast scaling laws

2.3 SHOCK WAVE PROPAGATION

2.4 CRATER FORMATION

2.5 SOIL-STRUCTURE INTERACTION (SSI)

2.6 BLAST RESPONSE OF BURIED TUNNELS

2.6.1 Experimental approach

2.6.2 Numerical approach

2.7 RESEARCH GAP

An underground transit tunnel is a critical piece of infrastructure as a large number of

3Chapter 3: Numerical Modelling

3.2 ARBITRARY LAGRANGIAN EULERIAN

The Arbitrary Lagrangian Eulerian (ALE) technique is introduced by combining the

FSI is a coupled multi-physics simulation process for solving highly non-linear

Figure 3.1: Lagrangian-Eulerian coupling

3.3 SMOOTH PARTICLE HYDRODYNAMICS

Smooth Particle Hydrodynamics (SPH) is a mesh-free computational Lagrangian

a) Tied interface b) Hybrid element coupling c) ‘Nodes-to-surface’ contact

Figure 3.2: Coupled SPH-FE Method

Although SPH is more expensive in terms of computation, the coupled SPH-FE

The ‘Node-to-surface’ coupling method is employed in the present study as SPH

3.4 SIMULATION OF BLAST LOADING

3.4.1 Element selection

3.4.2 Boundary condition