Asce 7690 9780784480038

Selected Papers from the Proceedings of the
Fourth Geo-China International Conference
Geotechnical Special
Publication No. 260
Emerging Technologies in
Tunnel Engineering, Modeling,
Design, Construction, Repair,
and Rehabilitation
Edited by
James C. Ni, Ph.D., P.E.
Junsheng Yang, Ph.D., P.E.
Shong-loong Chen, Ph.D., P.E.
Tong Qiu, Ph.D., P.E.
GEOTECHNICAL
SPECIAL
PUBLICATION
NO.
260
GEO
O-CHINA
H A 2016
EMERGIN
M
NG TECH
HNOLOGIIES IN TUNNEL ENGINE
EERING,
MODELIING, DESIGN
E
, CONSTRU
UCTION, REPAIR
R, AND
REHABBILITATI
TION
SE
ELECTED PAPERS FROM TH
HE PROCE
EEDINGS OF THE F
FOURTH
GE
EO-CHINA
A INTERN
NATIONA
AL CONFE
ERENCE
July 2527,
2
20116
Shandong, Chinna
SPON
NSORED BY
Y
Shando
ong Univerrsity
Shando
ong Deparrtment of T
Transportattion
University of Oklahhoma
Chineese Nation
nal Sciencee Foundatioon
Geo--Institute of
o the American Socieety of Civiil Engineerrs
ED
DITED BY
James C.. Ni, Ph.D.., P.E.

Junsheng
J
Yang,
Y
Ph.D
D., P.E.
Sh
hong-loong
g Chen, Phh.D., P.E.
Tong Qiu,
Q Ph.D., P.E.
Published
P
by
b the Amerrican Societyy of Civil En
ngineers
Published by American Society of Civil Engineers

1801 Alexander Bell Drive
Reston, Virginia, 20191-4382
www.asce.org/publications | ascelibrary.org
Any statements expressed in these materials are those of the individual authors and do not
necessarily represent the views of ASCE, which takes no responsibility for any statement
made herein. No reference made in this publication to any specific method, product, process,
or service constitutes or implies an endorsement, recommendation, or warranty thereof by
ASCE. The materials are for general information only and do not represent a standard of
ASCE, nor are they intended as a reference in purchase specifications, contracts, regulations,
statutes, or any other legal document. ASCE makes no representation or warranty of any
kind, whether express or implied, concerning the accuracy, completeness, suitability, or
utility of any information, apparatus, product, or process discussed in this publication, and
assumes no liability therefor. The information contained in these materials should not be used
without first securing competent advice with respect to its suitability for any general or
specific application. Anyone utilizing such information assumes all liability arising from such
use, including but not limited to infringement of any patent or patents.
ASCE and American Society of Civil EngineersRegistered in U.S. Patent and Trademark
Office.
Photocopies and permissions. Permission to photocopy or reproduce material from ASCE
publications can be requested by sending an e-mail to permissions@asce.org or by locating a
title in ASCE's Civil Engineering Database (http://cedb.asce.org) or ASCE Library
(http://ascelibrary.org) and using the Permissions link.
Errata: Errata, if any, can be found at http://dx.doi.org/10.1061/9780784480038
Copyright 2016 by the American Society of Civil Engineers.
All Rights Reserved.
ISBN 978-0-7844-8003-8 (PDF)
Manufactured in the United States of America.
Geo-China 2016 GSP 260
iii
Preface
This Geotechnical Special Publication contains 34 peer-reviewed technical papers
presented at the 4th GeoChina International Conference: Sustainable Civil
Infrastructures: Innovative Technologies for Severe Weathers and Climate Changes,
which took place in Shandong, China, from July 25 to 27, 2016. This proceeding
examines topics such as:
- Tunnel Management and Inventory, Monitoring and Settlement Control
- Emerging Technologies, Lining Design & Precast Segment Advances
- Innovation in Tunneling Design, Construction, Repair, Rehabilitation
- Fire & Life Safety, Vulnerability & Security
- Tunneling in Soft Ground, Ground Conditioning and Modification
- Advanced prediction technology of tunnel construction geology
- Deep excavations and urban tunneling
This publication should be valuable to geotechnical engineering professors and
students, as well as geotechnical engineers and professionals.
Acknowledgements
The following individuals have assisted in reviewing the papers:
Wen-Chieh Cheng, Ph.D. (Ove Arup & Partners HK Ltd.), Cheng-Tao Ho, Ph.D.
(Supertek Counsltant Co.), Shen-Chung Lee, Ph.D. (National Taipei University of
Technology), Xin Kang, Ph.D. (Missouri University of Science and Technology),
Omid Ghasemi Fare, Ph.D. (University of Louisville), and Ali Naeimipour (Ph.D.
Candidate, Penn State)
ASCE
iv
Contents
A Jet-Grouted Wall in Mitigating Tunnelling Effects on Adjacent
Structures .................................................................................................................... 1
Jinyang Fu, Junsheng Yang, and Shuangting Zhu
The Kriging Overlapping Method of Numerical Simulation for
Complex Models........................................................................................................ 10
Qin Yan, Bangshu Xu, Shuchen Li, and Yan Zhao
Simple Method for the Safety Evaluation of Underground Pipelines Induced
by Urban Shallow Tunneling ................................................................................... 19
Yong Wang, Lingwei Kong, and Yanli Wang
Mechanical Behavior Analysis for Parallel Small Spacing Tunnels in a Loess
Region ........................................................................................................................ 27
Xiangxing Kong
Analysis of Tunnel Monitoring Results Based on the Modulus Maxima
Method of Wavelet Transform ................................................................................ 34
Xiangxing Kong
Study of the Diseases of Shield Tunnels and Its Reasons ...................................... 40
Baotong Shi and Xiangxing Kong
Preparation and Failure Modes of Simulated Transversely Isotropic Rock ...... 46
X. M. Zhang, X. F. Ou, D. Yang, J. S. Yang, and J. Y. Fu
Sensitivity Research on a Construction Space for a Multi-Partition
Excavation Tunnel .................................................................................................... 56
W. Wang, J. S. Bian, and W. S. Pan
Numerical Research on the Waterproof Failure Mechanism of the Sealing
Gasket of an Underwater Shield Tunnel ................................................................ 64
J. C. Wang, L. W. Sun, X. J. Lv, Z. Wang, S. B. Hu, and C. F. Chao
Analysis of the Stability of Mud Inrush Induced by Fillings Sliding Failure
in a Karst Cave Based on the Simplified Bishop Method and
Its Application ........................................................................................................... 73
P. Lin, S. C. Li, Z. H. Xu, L. P. Li, X. Huang, and S. J. He
ASCE
Study on the Reasonable Cover Thickness of a Subsea Tunnel with the

Numerical Calculation Criterion Method .............................................................. 81
B. S. Xu, S. C. Li, R. C. Liu, and C. L. Zhao
A Study on the Strategy of Seepage Leakage Treatment of Highway
Tunnels under a Long-Term Groundwater Table ................................................ 92
Y. L. Zhang and Q. Y. Tian
A Case Study for Escape Route Optimization after Water Inrush in a
Backward Excavated Karst Tunnel ...................................................................... 100
S. C. Li, J. Wu, Z. H. Xu, L. P. Li, X. Huang, Y. G. Xue, and Z. C. Wang
Studies on the Characteristics of Traffic Accidents in Expressway
Tunnels..................................................................................................................... 109
Dingwen Zhong, Ling Pan, and Qingyan Tian
Earth Pressure in a Head Chamber and Grouting Issues for
EPB Tunneling in Soft Ground: A Case Study in Tianjin Area ........................ 117
Jianjun Zhou and Dongyuan Wang
The Effects of Relative Tunnel Depth and Volume Loss on
Vertical Settlements above Tunnels in Dense Sands ........................................... 125
Andrea Franza, Bo Zhou, and Alec M. Marshall
Performance of a Long Pipe Roof for a Large Span Shallow Tunnel
below an Existing Highway .................................................................................... 133
Yu-Feng Shi, Da-Xin Geng, Chang-Jie Xu, Jun-Sheng Yang, and Jin-Yang Fu
Analysis of the Impacts of a Tunnel on a Normal Fault Rupture through
Uniform Soil Cover ................................................................................................. 141
Mehdi Sarayloo and Reza Mahinroosta
Plasticity Solutions of an Undrained Cavity Contraction for the
Prediction of Soil Behaviour around Tunnels ...................................................... 150
Pin-Qiang Mo and Hai-Sui Yu
Numerical Analysis of a Temperature Field within a Vertical Frozen Soil
Wall .......................................................................................................................... 158
Jun Hu, Yong Liu, and Kai Yao
Improvement of Off-Shore Immersed Tunnel Foundations in
Soft Ground by Composite Foundation Methods ................................................ 166
Hongtao He, Yougao Lin, and Jianyu Li
ASCE
The Excavation Damage Zone of a Subway Tunnel Considering Stress

Release ..................................................................................................................... 174
Feng Huang, Hehua Zhu, Shuping Jiang, and Bo Liang
Variation of Slurry Pipe Jacking Force in Coarse and Fine Soils ..................... 182
James C. Ni, Louis Ge, and Wen-Chieh Cheng
Research on the Comprehensive Geological Prediction Method on a
Metro Tunnel Karst Disaster and the Numerical Simulation of
Corresponding Treatment Measures .................................................................... 189
Shucai Li, Haotian Wu, Ke Wu, Wen Zhang, Maoxin Su, and Yajun Wang
Risk Assessment of a Tunnel Collapse in a Mountain Tunnel Based on the
Attribute Synthetic Evaluation System ................................................................ 198
Benshuo Wang, Shucai Li, Qianqing Zhang, Liping Li, Qian Zhang, and Fei Xu
Gaussian Process Model of an Advanced Surrounding Rock Classification
Based on Tunnel Seismic Predictions ................................................................... 210
Peng He, Li-Ping Li, Qian-Qing Zhang, Fei Xu, Jie Hu, and Jian Zhang
Energy Characteristics of Micro-Seismic Events Associated with
Immediate Rockbursts in Deep Tunnels............................................................... 218
Yang Yu, Bing-Rui Chen, Chang-Jie Xu, Xin-Hong Diao, Li-Hong Tong, and
Yu-Feng Shi
Numerical Study on Tunnel Instability in a Fault Zone:
A Case Study of the Wuzhuling Tunnel in China ................................................ 226
Yingchao Wang, Hongwen Jing, Haijian Su, and Jiangyue Xie
The Effect of Foundation Pit Excavations on Adjacent Open-Cut Tunnels
for High-Speed Railways ........................................................................................ 234
Zhong Zhou, Xiang-Can Wang, Hao Yang, and Qi-Fang Zhang
Analysis on the Response of Ground Surface Settlement for a
Metro Station during Construction Using the Pile-Beam-Arch (PBA)
Method Subjected to Pore Pressure ...................................................................... 243
Fei Xu, Li-Ping Li, Qian-Qing Zhang, Zhe-Chao Wang, and Peng He
Tunneling beneath Existing Buildings Supported on
Shallow Foundations .............................................................................................. 251
C. F. Leung, J. Yu, Y. K. Chow, Y. S. Ng, H. T. Tan, and C. J. Hua
A Winkler-Based Method for the Assessment of Tunnelling-Induced
Deformations on Piled Structures ......................................................................... 259
Andrea Franza, Twana K. Haji, and Alec M. Marshall
ASCE
vi
Effectiveness of Buttress and Cross-Wall in Deep Excavations ......................... 267

Shong-Loong Chen, Cheng-Tao Ho, Chia-Yu Yeh, and Meen-Wah Gui
Effects of Cut-and-Cover Tunnel Construction on Adjacent Piles of an
Overpass in Bangkok Soft Clay ............................................................................. 276
Zaw Zaw Aye, Thayanan Boonyarak, Nutthapon Thasnanipan, and
Sereyroath Chea
ASCE
vii
A Jet-Grouted Wall in Mitigating Tunnelling Effects on Adjacent Structures

Jinyang Fu1; Junsheng Yang2; and Shuangting Zhu3
1
Postdoctoral Researcher, School of Civil Engineering, Central South Univ., Changsha 410075, China.
E-mail: jyfu2010@163.com
2
Professor, School of Civil Engineering, Central South Univ., Changsha 410075, China. E-mail:
jsyang@edu.csu.cn
3
Power China Huadong Engineering Corporation Limited, HangZhou 310014. E-mail:
zhushuangting07@163.com
Abstract: Technical measures, such as underground partition wall, could sever as

recourse to mitigate the ground movements induced by excavation. In this paper field
monitoring and numerical analysis were used to examine the performance of
underground jet-grouted partition wall in mitigating the effects of shield tunnel
construction on existing pier of Xin-Zhong-Road viaduct in the project of Changsha
Subway Line 1 in China. The performance of the jet-grouted partition wall was
calibrated by the reasonable agreement found between the observed field measurements
and the numerical results. Underground jet-grouted-column partition wall was testified
to serve as an effective measure in ground movement control given the need of
tunnelling nearby piled structures.
1
INTRODUCTION
Tunnelling in urban areas may cause damage to buildings founded adjacent. Underground
partition wall installed between the buildings and the tunnel was expected to cut off the
displacement induced by tunnelling so that the nearby buildings can be protected. Some
researches have been carried out to study the mechanism of underground partition wall
in reducing the damage impact of tunnelling to adjacent structures.
Harris et al. (1994) presented that the compensation grouting could effectively
reduce the ground settlement and nearby building distortion. Bilotta and Taylor (2005)
carried out centrifuge tests to study how the diaphragm wall reduces the tunnelling
induced movements at the two side of the wall. Kirsch and Piazzi (2009) also identified
the influential parameters of bored pile wall to protect sensitive buildings from
settlement resulting from tunnelling using numerical simulations. They concluded fully
mobilized friction between soil and wall leads to no significant reduction of settlement
on the far side of the wall. Bilotta (2008) performed a series of centrifuge tests and
ASCE
numerical analyses to investigate the effects of a diaphragm wall. They found length and
thickness of the wall, and the roughness of the soil-wall interface, the length of the wall
plays the main role in the effectiveness of reducing ground movements. Wu et al. (2012)
studied the pile wall in controlling the tunnelling effects on the nearby wood piles of a
church through a numerical analysis. Zou and Xu (2013) carried out three-dimensional
numerical simulation to investigate the mitigation effects of the separation pile and
diaphragm wall to tunnel induced ground movements, considering the effects of soil at
small strain stiffness.
In this paper, the performance of underground jet-grouted partition wall in
mitigating the effects of shield tunnel construction on existing pier of Xin-Zhong road
viaduct in the project of Changsha Subway Line 1 was examined based on field
measurements and numerical analysis. The performance of the jet-grouted partition wall
in reducing the tunnelling influence on the adjacent pier is illustrated by a comparison
between the field measurement and numerical results. The effectiveness of using
jet-grouted underground partition wall in reducing tunnelling induced displacement was
testified and discussed.
2
PROJECT OVERVIEW
2.1 Construction techniques

The tunnelling project referred to is located closely on the east side of L-off-ramp of
Xinzhong-road viaduct in the project of Changsha Subway Line 1, as shown in Figure 1.
The construction consists of two parallel tunnels. The two tunnels are circular and their
axes are 21.5 m below the ground surface level while there is a central line separation of
17.3 m between them. The studied tunnel section is about 155 m long from the distance
mark DK24+245 to DK24+400. Composite EPB shield tunnel boring machine of 6.25 m
diameter was used to construct the tunnels. Each construction cycle involved an
excavation in advance of the shield 1.5 m to install the segment lining before shoving
the shield forward. The precast concrete segments were each 0.3 m thick. Each segment
ring was formed from 6 precast concrete segments and has an internal diameter of 5.4 m.
Six piers of the L-off-ramp of Xinzhong Road viaduct are situated on the east side
of the east bound tunnel, and their marks vary from L01# to L06#. Each pier was
founded by four piles except L01# was supported by a single pile. The distances from
the piles to the tunnel central line vary between 6.7 to 9.9 m, which was no more than
1.5 times of the tunnel diameter. The pier L04# is located at DK24+330 founded by four
friction-cum-end bearing piles. Each pile has a diameter of 1.2 m enlarging to 1.8 m at
the end and is separated by a spacing of 3 m to the other piles. The distance of the
nearest pile to the tunnel profile was 6.9 m.
In order to minimize the tunnelling effect on the pier, three rows of jet-grouted piles
were installed to form a at least 4 m wide underground partition wall to protect the pier
from tunnelling induced deformation. The jet-grouted columns have a diameter of 0.8 m
while their axis spacing is 0.5 m. That means there is always about 0.3 m jet-grouted
ASCE
column body interlocked with the adjacent one. As a result, the jet-grouted columns can
act as an underground wall to partition the soil movements off. Since the piles of pier
L04# are founding above the tunnel invert level, this paper focuses on the analysis of
using jet-grouted column wall to mitigate the influence of tunnelling on the piled-pier
L04#.
Tunnelling direction
West bo un
L06#
mileage
DK24+400
mileage
DK24+340
East boun d tunnel
mileage
DK24+330
North
d tu nnel
L05#
L02#
L03#
L04#
L01#
scale 1 0 m
centreline of L-off-ramp of Xinzhong Road viaduct

ground surface
pier L06#
1.2/1.6m
L20m
pier L05#
1.2/1.6m
L20m
silty clay
34.00
34.00
52.41
plain fill
pier L04#
1.2/1.8m
L14m
pier L03#
1.2/1.7m
L21m
pebble
36.00
tunnel
29.00
pier L02#
1.2/1.7m
L19m
pier L01#
1.2/1.8m
L19m
31.00
29.00
clayey sandy grevel with pebble

conglomerate
Figure 1 plane view and geologic profile of the Changsha Subway Line 1 pass
through the L-off-ramp of Xinzhong-Road viaduct
2.2 Geological condition
The shield tunnel drove in the clayey sandy gravel with pebble ground which is overlain
by silty clay followed by plain fill up to the ground surface. The typical geological
condition at the transverse section of pier L04# was also shown in Figure 1. The clayey
sandy gravel with pebble consists of silty clay, silica sand, gravel and pebble, among
which the silica sand is the main content mixed with mud. Site investigation shows that
the soil contains 50.7% to 62.7% grain coarser than 20 mm while it contains 20% sand
grain finer than 2 mm, and the fines are less than 10%. This manifests that the tunnel is
ASCE
excavated in a gap-graded soil. The ground water table was measured approximately 7~8
m below the surface level in this area.
Spoil taking from the screw conveyor, after flushed, shows that the gravel and sand
were not well graded. In-situ pump test results indicate that the coefficient of permeability
of the clayey sandy gravel ground is about 1.73610-4 cm/s while the above silt clay
results in a relatively low coefficient of permeability about 5.79010-8 cm/s. Table 1
shows the geotechnical properties of each soil layers used for analysis in this study.
3
NUMERICAL ANALYSIS
A 3D numerical model was built using the commercial code ANSYS to investigate the
field performance of jet-grouted partition wall in protecting the pile-supported pier #L04
in the studied project. The finite element mesh is shown in Figure 3 (a). The model had
100 m in transverse direction and 60 m in the longitudinal direction. The bottom of the
model was 46 m down below the surface. All the subjects were modelled with 202400
quadratic volume elements with the dimension described in section 2. The three soil layers
were modelled to follow the Druck-Prager failure criterion in ANSYS while the
pile-supported pier and jet-grouted partition wall were modelled as elastic. The tunnelling
process was simulated based on the method presented by Cheng et al. (2007). The
mechanical parameters for all materials in the analysis are listed in Table 1.
Table 1 Mechanical parameters of the materials used for Druck-Prager failure
criterion in ANSYS
material
Unit weight
(kN/m3)
thickness
(m)
Elastic modulus
(MPa)
Poissons
ratio
Cohesion c
(kPa)
Friction
angle ()
Plain fill
19
8.9
10
0.40
40
17
Silty clay
20
6.6
30
0.38
65
20
Clayey sandy gravel

with pebble
22
11.1
50
0.35
10
36
25
15.0
2e3
0.2
25
23.5
30e3
0.2
Jet-grouted partition
wall
Piled pier
The relative positions of the tunnel, the pile-supported pier and the partition wall are
shown in Figure 2 (b). The pier was simplified at the ground surface level supported by
four piles and it was loaded at the up surface with a distributed load of 1080 kN which was
computed from the actual weight multiply a safety factor. The interfaces between the pile
and the soil and between the partition wall and the soil were modelled with
surface-to-surface contact. The behavior of the interfaces follows the coulomb friction
criterion with the friction coefficient of 0.5 and 0.6 for the pile-soil interface and wall-soil
interface, respectively. The jet-grouted columns were assumed to be interlock with each
other tightly and uniformly. Hence they were simplified as a wall between the tunnel and
ASCE
the piled
p
pier. Th
he mechaniccal propertiess of the jet-ggrouted partiition wall weere evaluated
based
d on the prin
nciple presen
nted in Pavlo
ovic et al. (2010).
Since the prresent paper most concerrn of the worrking mechaanism of the underground
partition wall, th
he tunnel waas simulated
d by the sim
mple displaceement controolled method
und volume loss
l of 1.1%. A similar an
analysis withhout the partition wall waas
with a given grou
carried out to maake a compaarison of the pile responnse to tunnellling. Only thhe short term
m
ground movemen
nts were con
nsidered in this
t study, hhence the num
merical anallyses were inn
undrained condittion.
gure 2 Num
merical modeel for simula
ating tunne l near piled
d pier with jet-grouted
Fig
collumn wall: (a)
( numericcal model, (b
b) piled pierr, partition w
wall and shield tunnel
4
FIELD
F
MO
ONITORING
G
In orrder to evalu
uate the effeectiveness off the jet-groouted partitioon wall in m
mitigating thhe
ground movemen
nts. Extensiv
ve measurem
ments were ccarried out too monitor thee response oof
the ground
g
during
g the constru
uction of the shield tunneel (see Fig. 3). Settlemennt observationn
pointts were instaalled along th
he transversee section (S--A) where thhe pier L04## protected byy
the jet-grouted partition
p
walll, and anoth
her comparaative transveerse sectionn (S-B) 10 m
y to the south
h where therre were no pier and partiition wall weere monitoreed as well.
away
ASCE
Fence
Tunnel centre line
Partititon wall
C2
C1
DK24+330
A4
Pile L04#
A3
Section SA
A2
A1
B2
B1
DK24+340
A0
A6
A7
A8
Temporary panel house

B5
B4
B3
Section SB
Figure 3 Scheme of field measurements

Meanwhile, horizontal displacement were measured with the two inclinometer tubes
installed at the two sides of the jet-grouted partition wall (see Fig. 3). The inclinometer
tubes were placed as deep as the partition wall which is enough for a fixed ending.
Furthermore, the groundwater drawdown was recorded using a water level meter at the
same inclinometer tubes where special filtering hole was placed at the deep end of the
tubes before the installation (see Fig. 3).
5
GROUND RESPONSE
Figure 4 compares the ground horizontal displacement measured and that computed
from the numerical analysis. Reasonably good agreement was found between the
measured and calculated horizontal displacements at tube C2, while the measured
horizontal displacements at tube C1 are significantly less than computed. This is because
during the field monitoring the displacement sensor cannot be lowered more than 16 m
down in tube C1. However, deformation trend shows that more horizontal displacement
would be experienced close to the surface level. A fact is that the jet-grouted column wall
was not extended to the ground surface but below the surface at a certain level. Hence the
ground deformation close to the surface level therefore cannot be effectively partitioned
off.
ASCE
Horizontal displacement at C2 tube (mm)

0
2
4
6
8
10 12 14
0.0
0.0
-2.5
-2.5
-5.0
-5.0
Vertical depth below sruface (m)
Vertical depth below surface (m)
Horizontal displacement at C1 tube (mm)

0
2
4
6
8
10 12 14
-7.5
-10.0
-12.5
-15.0
-17.5
-20.0
-22.5
-25.0
-27.5
Numerical results
Field measurements
-7.5
-10.0
-12.5
-15.0
-17.5
-20.0
-22.5
-25.0
-27.5
Numerical results
Field measurements
Figure 4 Horizontal displacement at the two side of partition wall

Figure 5 shows both the measured and computed surface settlements. The
effectiveness of the jet-grouted partition wall was illustrated by a significant discontinuity
in settlements at the two sides of the position. On the side of the wall close to the tunnel,
the ground has experienced much more settlement than on the natural ground surface
while on the side where structures located on the ground has experienced significant less
differential settlement than on the natural ground surface.
The numerical results have revealed the discontinuity in the settlement curve, but it
shows a shallow and wide curve if comparing with measured data. It should be pointed out
that the calculated settlement trough is relatively shallow and wide compared with the
measured results. Such differences should be ascribable to the relative simple soil
constitutive model used in this study (Addenbrooke et al., 1997; Hejazi et al., 2008),
hence it is suggested that further detailed investigation of this problem should be
evaluated with advanced soil constitutive model.
Since the coefficient of permeability of the clayey sandy gravel with pebble ground is
relatively high. The shield tunnel driving would lead to variation of groundwater and
consequently induce further ground deformation. It is of great interest to monitor the
influence of the jet-grouted partition wall to the groundwater level. Figure 6 shows the
variation of groundwater level at C1 and C2 tubes with the advance of the shield tunnel
construction. The results show that when the tunnel face approaches the monitoring
section, as results of the shield extrusion, the groundwater level raised quickly about 0.25
m and 0.7 m in tube C1 and tube C2, respectively. After the tunnel face passed there were
significant groundwater drawdown about 0.5 m and 1 m for tube C1 and tube C2,
respectively, and then back to the original groundwater level. It should be emphasized that
the groundwater variation in tube C2 is always larger than in tube C1, which highlights the
ASCE
effect of the jet-grouted partition wall. That means the jet-grouted column wall can also
partition some groundwater variation off.
-50
-40
-30
Distance from the tunnel centreline (m)

-20
-10
0
10
20
30
40
50
0.0
-3.0
Settlement (mm)
position
-6.0
-9.0
Measured settlement at SA
Numerical results (Vl=1.1%)
Measured settlement at SB
-12.0
-15.0
Figure 5 Surface settlement profiles

-50
Distance of tunnel face to the monitoring section

-40 -30 -20 -10
0
10
20
30
40
50
Groundwater depth (m)
5.0
5.5
Shield passing the

monitoring section
6.0
C1 tube
C2 tube
6.5
7.0
7.5
8.0
8.5
Figure 6 Groundwater variation at tube C1 and C2 during tunnel passing

6
SUMMARY AND CONCLUSIONS
A relatively small number of case histories are available on the topic of using
underground partition wall to mitigate the possible ground movements. In this paper,
field measurements and numerical analysis were carried out to investigate the
performance of jet-grouted partition wall in mitigating the effects of shield tunnel
construction on adjacent pile-supported pier.
The underground partition wall can work effectively to improve the mechanical
performance of tunnelling-pile-pier interaction. The presence of the partition wall can
relieve the piles behind the wall suffering from differential displacement. In front of the
wall at the tunnel side, the ground would experience an increase in displacements.
ASCE
Besides to redistribute the ground displacements of the structures to be less

differential, another notable benefits are that the underground partition wall can also
cut-off groundwater variation off during the tunnel construction stage. More
sophisticated numerical model is suggested to achieve a detailed analysis of the working
mechanism of underground partition wall.
REFERENCES
Addenbrooke, T.I., Potts, D.M. and Puzrin, A.M. (1997). The influence of pre-failure
soil stiffness on the numerical analysis of tunnel construction. Gotechnique, 47, 3,
693-712.
Bilotta, E. (2008). Use of diaphragm walls to mitigate ground movements induced by
tunnelling. Gotechnique, 58, 2, 143-155.
Bilotta, E. and Taylor, R. (2005). Centrifuge modelling of tunnelling close to a
diaphragm wall. International Journal of Physical Modelling in Geotechnics, 5, 1,
27-41.
Harris, D., Mair, R., Love, J., Taylor, R. and Henderson, T. (1994). Observations of
ground and structure movements for compensation grouting during tunnel
construction at Waterloo station. Gotechnique, 44, 4, 691-713.
Hejazi, Y., Dias, D. and Kastner, R. (2008). Impact of constitutive models on the
numerical analysis of underground constructions. Acta Geotechnica, 3, 4, 251-258.
Kirsch, A. and Piazzi, L. (2009). Numerical investigation of the effectiveness of a bored
pile wall for the minimisation of settlement resulting from tunnel driving.
Geomechanik und Tunnelbau, 2, 6, 753-765 10.1002/geot.200900063.
Pavlovic, M.N., Cotsovos, D.M., Dedic, M.M. and Savidu, A. (2010). Reinforced
jet-grouted piles. Part 1: analysis and design. Proceedings of the Institution of Civil
Engineers - Structures and Buildings, 163, SB5, 229-308.
Wu, C.J., Zhang, Z.X., Ding, W.Q. and Zhang, D.Y. (2012). Influences of construction
of side-crossing shield tunnel on adjacent ancient architectures and reinforcement
effect of protection measures. Chinese Journal of Geotechnical Engineering, 34, 1,
158-165. (In Chinese)
Zou, W.H. and Xu, M. (2013). 3D numerical analysis of mitigation effect of separation
pile and diaphragm wall considering small strain stiffness of soils. Chinese Journal
of Geotechnical Engineering, 1, s1, 203-209. (In Chinese)
ASCE
10
The Kriging Overlapping Method of Numerical Simulation for Complex

Models
Qin Yan1; Bangshu Xu2; Shuchen Li; and Yan Zhao
1
2
Doctor, School of Civil & Hydraulic Engineering, Shandong Univ. E-mail: 282182074@qq.com
Associate Professor, School of Civil & Hydraulic Engineering, Shandong Univ. E-mail:
xubangshu@sdu.edu.cn
Abstract: With the rapid development of urban construction and the traffic congestion
problems arise, underground engineering has developing rapidly throughout the
country. The form of underground engineering is also more and more diverse.
Numerical simulation method has become increasingly widely used in geotechnical
engineering. However, when there are multiple excavation areas or channels and even
intersections in a model, it will cause great inconvenience for modeling and
calculation. To solve the modeling and calculation problem of complex models,
Kriging interpolation method is improved to be suitable for geotechnical engineering.
Kriging overlapping method of numerical simulation for complex models means that
model is divided into a global model and a local model. Using Kriging interpolation
method, the displacement of local model is lead to the global model as initial
displacement field. And then further excavation for global model is carried out to get
the final displacement field. Compared the results of numerical simulation and
monitoring data, the results show that Kriging can consider spatial structure
characteristics of the displacement field in underground engineering, and it has high
reliability. As expected, the results of numerical simulation and measurement come to
an agreement, which confirming the feasibility of Kriging overlapping method.
INTRODUCTION
In recent years, with the rapid development of urban construction, which
accompanying by the emergence of traffic congestion and land shortage (Lin Q. Y.,
2011), more and more metro engineering arise (Dong J. X., 2002; Liu J. H., 1993;
Qian Q. H., 1998; Wang N. T., 2002; Zhou S. H., 2005). Deformation caused by
subway construction not only endanger security and stability of project itself, but also
affect the safety of surrounding buildings and municipal facilities. It can cause serious
economic losses and social impacts. Therefore, the study on safety and stability of
subway construction is essential.
Underground engineering have characteristics of non-homogeneous, noncontinuous, non-linear, complex loading and unloading conditions, and boundary
ASCE
conditions, and so on, which make underground engineering problems often cannot
simply be solved by analytic methods. In contrast, numerical analysis method can
simulate complex mechanical behavior and structural properties of rock, and it can
also analyze complex boundary value problems and the construction process easily
(Kumar, 2000; Li J. Y., 2012; Zhang L. M., 2007; Zhou J. W., 2009; Zhou Y., 2012).
Subway construction process may consider tunnel excavation, transfer station
excavation, as well as interaction of open cut pit and bored tunnel. With the
development of subway construction and appearance of diverse design form of
subway construction, numerical calculation method encounters problems. It is difficult
to build complex model and compute effectively.
Zhu Weishen put forward the concept of construction process mechanics. Base on
mechanics analysis, he thinks that underground construction and support process is
actually a repeated loading apply to surrounding rock in space and time. This concept
provides a powerful rationality basis (Zhu W. S., 2003). Xu Bangshu come up with
displacement overlapping method of numerical analysis for deformation influence on
the adjacent buildings in complicated subway station construction. The inverse
distance weighted method and elementary interpolation method are compared. Study
results indicate that inverse distance weighted method has higher precision (Xu B. S.,
2014).Currently, It is rarely reported that acquiring final displacement by
superimposing construction information using Kriging interpolation.
SUPERPOSITION PRINCIPLE AND IDEAS
In order to solve numerical simulation problem of complex subway stations, and
simulate the deformation response of surrounding buildings to station excavation more
accurately, new numerical simulation implement method is studied. Kriging
overlapping method of numerical simulation is proposed by taking a tower-style
double hole tunneling engineering as the background and using Kriging interpolation
method, which is widely used for data visualization (Jin G. D., 2003). The model is
considered as a combination of global model and local model in the same coordinate
system. Calculating and obtaining the displacement of local model firstly. And then
lead this displacement field into global model as initial displacement field by Kriging
interpolation method. Finally completing construction and support process simulation
of global model on this basis, and obtaining a new displacement field, which
approximated the final result of overall model.
KRIGING INTERPOLATION METHOD
Improving traditional Kriging interpolation method to make it be suitable for
geotechnical engineering. Let P1, , Pn as a series of sample points in model, D(x1)
D(xn) as corresponding displacement. P1, , Pn are only determined by
coordinate values. According to ordinary Kriging principle, D*(P0) at location P0 is a
weighted summation result of n known sample points, and it can be estimated by a
linear combination:
ASCE
11
12
D* ( P0 ) = i D ( Pi )
i =1
wherei are the unknown weights assigned to the data samples, and are determined
by unbiased condition and minimum estimation variance.
Based on unbiased requirements
E D* ( P0 ) -D ( P0 ) = 0
Since mathematical expectation of D(P) exists and is equal to a constant, then
n
n
E i D ( Pi ) D ( P0 ) = i m m = 0
i =1
i =1
Yield
n
=1
i =1
Let estimation variance minimal in unbiased conditions, that is
2
Var D * ( P0 ) D ( P0 ) = E ( D * ( P0 ) D ( P0 ) ) E D * ( P0 ) D ( P0 )
2
= E ( D * ( P0 ) D ( P0 ) )
= min
Conditional extreme value can be obtain by Lagrange multiplier method
n
2
*
2 = 0, j = 1, , n
E
D
P
D
P
(
)
(
)
(
)
i
0
0

i =1
where, is Lagrange multiplier.

Further derivation obtain n+1 order linear equations, namely Kriging equations:
n
C ( Pi Pj ) i = C ( P0 Pj )
i =1
n
=1
i
i =1
( j = 1, , n )
Variogram is defined as:

2
1
( Pi Pj ) = E D* ( Pi ) D ( Pj )
Structural analysis of regionalized variables

After obtaining the experimental variogram value, the right variogram theoretical
model should be determined according to the experimental variogram, and then curve
ASCE
13
fitting can be performed for selecteed theoreticaal model. T

This processs is calledd
strructural analy
ysis. Variog
gram parameters used too describe thheoretical moodel includee
cod
domain, abuttment, sagittaa and nuggeet constant. M
Model param
meter diagraam shown inn
Figu
ure 1.
FIIG. 1. Modeel parameterr diagram

The
T diagram illustrate thaat curve is in
n a horizontaal position w
when distancce exceeds a
certtain thresho
old. This sp
pecific disttance calledd codomainn when currve reachess
horiizontal posittion for the first time. Two
T
points aare correlativve in spatiall when theirr
disttance is less than codom
main value, otherwise
o
be uncorrelateed. The varioogram valuee
at codomain position is callled abutmentt. Abutment minus nuggget is equal too sagitta.
Theeoretical mod
del of variog
gram
Theoretical
T
models
m
of variogram mainly
m
incllude sphericcal model, exponentiall
mod
del and Gau
ussian modell. One of thee most comm
monly used is the spherrical model.
Sph
herical modeel equation iss:
3 h 1 h 3

( h ) = C0 + C
2
a
2 a
C + C
0
h=0
ha
ha
PRO
OJECT CA
ASE
Pro
oject profile
The
T pillar typ
pe station, lo
ocated near the
t road intersection, is uundergroundd excavationn
of double
d
hole.. Station lin
ne is from north
n
to sout
uth. Buildinggs upon subbway stationn
maiinly include residential buildings, hospitals,
h
gaas stations aand commeercial shops.
Stattion is locateed in core areea of business and with developed trransport. Tottal length off
maiin station is 145.55m, wh
hile total wid
dth is 33.9m
m. The main station incluude platform
m
ASCE
14
tunn
nel for left and right lin
ne, north an
nd south upllink, cross ttunnel, northh and southh
statiion hall and
d end wind tunnel.
t
Therre are two bbored duct aat both endss of left andd
righ
ht platform, where south
hern wind tu
unnel constru
ruct at the saame time wiith southernn
statiion hall.
n height of leeft and right platform aree all 9.2m, w
while width aare all 9.5m.
Excavation
Two
o platform are
a linked by
y three crosss tunnel, whhich has a heeight of 6.5m
m, 7.5m andd
7.5m
m and a wid
dth of 6.5 m, 8.5m and 8.5m,
8
respecttively. In thee middle of ccross tunnell
are two uplink tunnel, which connect north
n
and soouth station hall. Uplinkk tunnels inn
the south and north
n
have a height of 4.3m
4
and a w
width of 6.55m, and theiir length aree
m and 54m, respectively.
r
40m
Station is located
l
in th
he lower mo
onadnock errosion. soil ffrom top to bottom aree
artifficial filled of
o Quaternary Holocenee and coarsee granite of llate Yanshann. The mainn
statiion is located in slighttly weathereed granite w
with less undderground w
water, whilee
statiion in the no
orth located in
i moderately
y weatheredd granite.
Modeling and parameter
p
determinati
d
ion
In
nternal struccture model of
o the tower--style doublee hole station is shown iin Figure 2.
FIG. 2.
2 Internal structure
s
moodel of statiion
The
T subway station include platforrm tunnel ffor left and right line, three crosss
tunn
nels, two up
plinks, tow station hallls and one w
wind tunnell. Southern station halll
con
nstruct at thee same timee with end wind tunnell. Model is big and coomplex. Thee
who
ole model is divided into
o one global model and oone local moodel, where local modell
con
nsiders only transverse channels excavation
e
aand global model connsiders onlyy
long
gitudinal channels excavation. Com
mputing locaal model andd lead its ddisplacementt
field
d to global model
m
and excavate
e
nex
xt step. Buillding two models solvess difficultiess
in three-dimens
t
sional modeeling and meshing.
m
Groound surfacee settlementt caused byy
excavation is sttudied. Dim
mensions of two
t
models are all 100m
m84m2000m, and thee
surfface has a certain
c
slopee. Numericaal models arre built by finite elemeent softwaree
FLA
AC3D and co
omputed by finite
f
difference softwaree.
The
T bottom of
o the model is fixed. The
T upper booundary of ccalculation m
model is thee
grou
und, which is free. And
A
unidirecctional consstraints are applied on the laterall
bou
undaries of model.
m
The constitutive model of ssoil is the cllassical Mohhr-Coulombb
ASCE
15
ideaal elastic-plaastic constitu

utive model.. The mechaanical param
meters of soiil are shownn
in Table
T
1, whicch provided by engineerring geologicc investigatiion and simiilar projects.
Calculation mod
del considerrs weight streess only, witthout considering tectoniic stress.
Table 1 Physico-mech
hanical paraameters of th
he soil
Friction
nal
Deformattion
Densiity Cohesioon
Poisson
angle
modulu
us
(kN/m
m3)
(kPa))
ratio
(MPa))
()
Miscellaneeous fill
1.5
20
19
14
0.38
Highly
H
weatheered granite
80
20
300
31
0.3
Mod
derately weatthered granitee
800
20
500
42
0.25
slightly weatheered granite
1500
20
1000
45
0.25
Soil
Ana
alysis of calculation ressults
After
A
calculaating transveerse channelss excavationn of local moodel, lead thhe results off
disp
placement field
f
to glo
obal model by Krigingg interpolattion methodd as initiall
disp
placement field. Excavatte longitudin
nal channels on this basiis to obtain ffinal results.
A number
n
of keey points in
n the surfacee are taken aand compareed with site monitoringg
dataa.
Results of numeerical simula
ation
Results
R
of lo
ocal model and global model are shown in F
Figure 3 annd Figure 4,,
resp
pectively. Fig
gure 4 show
ws the final reesults.
FIG. 3. Results of
o transverse channels eexcavation (local modeel)
Fig. 4. Final
F
resultts by Krigin
ng overlappiing method (global mod
del)
ASCE
16
Results comparrison of Krig

ging overlapp
ping methodd with monitooring
n order to veerify the feassibility of Krriging overlaapping methhod, a numbeer of surfacee
In
key
y points at left
l
side of models aree taken as m
monitoring points. arraangement off
mon
nitoring poin
nts is shown in Figure 5.
Fig. 5. Arrangem
ment of moniitoring poin
nts
Adding
A
calcu
ulation resultts of two mo
odels individdually and coomparing wiith results off
Krig
ging overlaapping meth
hod, as shown in Figuure 6(a). Positive valuues indicatee
settllement, negaative values indicate upllift. Results comparison of Kriging overlappingg
metthod with mo
onitoring aree shown in Figure
F
6(b).
(a)
(b)
FIG. 6.
6 Final resu
ults
(a) Results
R
of num
merical simullation; (b) Co
omparison of nnumerical ressults and monnitoring data
The
T results are
a basically the same ob
btained by ccalculating inndividually aand Krigingg
overlapping meethod accord
ding to figurre 6(a). Andd through sim
mple modell tests usingg
Krig
ging interpo
olation meth
hod, it was found Krigging can sim
mulate spatiial structuree
charracteristics of
o displacem
ment field an
nd predict unnknown dispplacement acccurately. Inn
the calculation results
r
of glo
obal model, all monitoriing points m
move up exceept point J3.
c
is deep and has strongg stiffness, the middlee
Beccause the trransverse channels
ASCE
excavation of the model cause slight bulge on both sides. After longitudinal channels
construct, station hall generates maximum displacement, which with shallowest depth.
Therefore, surface settlement caused by shallow tunnel excavation might be more
obvious. This conclusion match the former study obtained (Liang R., 2006; Liu J. F. et
al, 2009).
Comparison shows that Kriging overlapping method results are in good agreement
with experimental results according to figure 6(b). Although there are some
differences in the two curves on the value, the trend of deformation is the same and the
differences are within allowable range. The results acquired by Kriging are ideal. It
confirmed the feasibility of Kriging overlapping method.
CONCLUSION
In order to solve numerical simulation problem of complex subway stations, new
numerical simulation implement method is studied. Kriging overlapping method of
numerical simulation is proposed by taking a tower-style double hole tunneling
engineering as the background. The whole model is divided into a global model and a
local model. Kriging interpolation method is introduced to achieve input and output of
displacement field at different coordinates system according to nodal coordinates. The
displacement results of local model is the initial displacement field of global model.
Final results can be obtained after global model excavation, and compared with
measured results. The main conclusions are as follows:
(1) The more shallow tunnels embedding, the greater surface subside. Since the
longitudinal channels are deep enough, and their diameter are small, excavation have
little effects on surface deformation. The method dividing the whole model into two
models not only can reflect the displacement response of tunnel excavation, but also
bring a significant reduction in time and effort of modeling and computation. The
method have important significance and reference value.
(2) Kriging, which considers spatial structure characteristics, can simulate
displacement field in underground engineering reasonably. Hence, Kriging
interpolation can also be used to predict the unknown displacement of model, even
though there are no grid nodes at unknown point position. Kriging overlapping method
add displacement field of local model and global together, and the result is similar to
the experimental results. Kriging overlapping method is feasible.
ACKNOWLEDGMENTS
Financial supports for this paper provided by National Science Foundation of China
(No.50909056), Provincial Science Foundation of Shandong Province
(No.ZR2014EEM014, ZR2014EEM029), and Science and Technology Project Plan in
2015. Ministry of Housing and Urban-Rural Development of China (No. 2015-K5-004)
are gratefully acknowledged.
ASCE
17
REFERENCES
Dong J. X., and Wang X. X. (2002). "Construction of 21.6 m span section of
Guangzhou metro" J. Railway Construction Technology, Vol. (3): 29-32.
Kumar, P. (2000). "Infinite elements for numerical analysis of underground
excavations" J. Tunnelling and Underground Space Technology, Vol. 15 (1): 117124.
Li J. Y., and Zhang Z. X. (2012). "3D numerical analysis of deep station excavation
constructed in round gravel strata" J. Chinese Journal of Underground Space and
Engineering, Vol. 8 (1): 71-76,110.
Lin Q. Y. (2011). "The problem analysis of urban road traffic" J. China New
Technologies and Products, Vol. (22): 188.
Liu J. H., Hou X. Y. (1993). Excavation engineering manual.
Qian Q. H. (1998). The development of urban underground space" J. Chinese Journal
of Geotechnical Engineering, Vol. 20 (1): 112-113.
Wang N. T., Chen R. Y., and Xie J. (2002). "Study on construction technique of multitunneling at shallow depth of urban subways" J. Rock and Soil Mechanics, Vol. 23
(2): 208-212.
Xu B. S., Ding W.T., and Liu L. J. (2014). "Displacement superposition method of
numerical analysis for determining influence of complex metro station construction
on deformation of adjacent buildings" J. Rock and Soil Mechanics, (S2): 619-625.
Zhang L. M., Zheng Y. R., and Wang Z. Q. (2007). "Application of strength reduction
finite element method to road tunnels" J. Rock and Soil Mechanics, Vol. 28 (1): 97101.
Zhou J. W., and Xu W. Y. (2009). "Application of rock strain softening model to
numerical analysis of deep tunnel" J. Chinese Journal of Rock Mechanics and
Engineering, Vol. 28 (6): 1116-1127.
Zhou S. H., and Zhang X. F. (2005). "Study on excavation method under shallow
overburden of soft-flow ground" J. Chinese Journal of Rock Mechanics and
Engineering, 24 (3): 526-531.
Zhou Y., and Wang T. (2012). "Strain softening model of rock based on FLAC3D" J.
Journal of Yangtze River Scientific Research Institute, Vol. 29 (5): 51-56,61.
Zhu W. S., and Li S. C. (2003). "Some developments of principles for construction
process mechanics and some case history studies" J. Chinese Journal of Rock
Mechanics and Engineering, Vol. 22 (10): 1586-1591.
ASCE
18
19
Simple Method for the Safety Evaluation of Underground Pipelines Induced by

Urban Shallow Tunneling
Yong Wang1; Lingwei Kong1; and Yanli Wang2
1
State Key Laboratory of Geomechanics and Geotechnical Engineering, Institute of Rock and Soil
Mechanics, Chinese Academy of Sciences, Wuhan, Hubei 430071, China. E-mail:
wang831yong@163.com
2
Key Laboratory of Geotechnical Mechanics and Engineering of the Ministry of Water Resources,
Yangtze River Scientific Research Institute, Wuhan, Hubei 430010, China. E-mail: wyldhh@126.com
Abstract: Excavation of urban-shallow buried tunnel would cause adverse effects on

above or neighboring undergound pipelines, even leads to major accidents. According
to the joint connection and deformation rigidity, municipal pipelines are divided into
the flexible pipeline and the rigid pipeline. Based on the pipeline failure modes, safety
criteria of various underground pipelines are discussed. Through the commonly
accepted formula of soil movement induced by tunnelling, this paper proposes a
simple method on safety evaluation of the flexible pipeline, and gives a case of
engineering application, which takes into account spatial position relations between
the pipeline and tunnelling, and deformation curve shape of the flexible pipeline. The
calculation formula has fewer parameters to be used easily, especially suitable for
quantitative safety prediction of underground pipelines in a condition that there are no
enough formation datas and detailed pipeline informations before tunnelling. And it
also provides a theoretical support for reasonably determining the limit alarming value
of ground settlement in tunnel construction.
INTRODUCTION
Underground pipeline is the lifeline of the city, which is an important support for its
survival and development. It is almost inevitable for city tunnel engineering to traverse
kinds of existed underground pipelines, especially for the urban pedestrian underpass
crossing municipal roads. Compared with the metro tunnel, the depth of urban
pedestrian underpass is mainly determined by the buried depth of above existed
underground pipelines. Moreover, in terms of tunnel design and choice of the
excavation method, effects induced by tunnelling on the safety of above underground
pipeline need to be paid more attention. It is necessary to properly handle the
contradiction between tunnelling and the protection of underground pipeline. If not, it
would lead to the pipeline strength failure or structural damage, causing the sudden
accident so as to affect the normal daily life, and even other major disasters. The
accidents of pipeline leakage, pipe bursting, nature gas explosion, cable breaking and
ASCE
others caused by tunnelling in urban city are often reported in newspapers. According
to an incomplete statistics, accidents of pipelines and cables occur more than ten
thousand cases annually in China, causing direct economic losses of about 100 million
dollars (Wang et al. 2006).
The types of underground pipelines mainly include the water supply pipe, drainage
pipe, gas pipe, thermodynamic pipe, cable of telecommunication and television, and so
on. Various type of pipeline has different sensitivity and tolerance to the stratum
deformation due to the differences of its working pressure, material, connection mode,
usage status, construction quality, and service life (Moser 2003). Domestic and foreign
researchers have done a lot of works on the safety evaluation of underground pipeline,
in which main methods include the theoretical calculation, model test, and numerical
simulation. Based on the theory of elastic foundation beam, Attewell et al. (1986)
firstly proposed analysis method of the effects of tunnelling on underground pipelines.
Using an elastic continuum model, Klar et al. (2005) analyzed the deformation of the
adjacent underground pipeline induced by tunnelling, and established a strict elastic
solution of pipeline deformation. Further on, Vorster et al. (2005) evaluated the
bending deformation of pipelines by the elastic continuous solution, and verified it
through the centrifugal simulation test. Xiang et al. (2010) based on the Peck empirical
formula and the theory of elastic continuous solution of interactions between the pipe
and soil, established an internal connection between the ground settlement and
deformation of pipeline. Liu & Hou (1997) suggested the deformation curvature radius
of pipeline as criteria of damage evaluation with the hypothesis of the same curvature
of joint and pipe section. By means of mathematical statistic analysis, Tian et al. (2012)
discussed the deformation controlling standard that the construction of urban
underground engineering affects the pipeline. In terms of limiting the section
deformation rate to ensure serviceability of pipes, Huang et al. (2013) studied the
control criterion of section deformation for the flexible pipelines. In a word, most
studies used the analytical theory of elastic foundation beam or the numerical
simulation method to quantitatively evaluate influences of tunnel excavation on above
pipelines (Wei et al. 2009). However, which method is more suitable or closed to the
truth has not reached a definitive conclusion yet.
By means of the mathematical model of settlement curve suggested by Attewell &
Yeates (1986), based on the fact that there is no unified method for safety evaluation
of underground pipelines at present, this paper proposes a simple method on safety
prediction for the flexible pipelines affected by tunnelling, and gives a reference for
the practical engineering.
THE PIPELINE FAILURE MODE
The form of pipeline function failure varies. Tunnelling inevitably gives
disturbances to surrounding soil of the pipeline, which leads to the pipe cause
additional stress and deformation. Once the stress or deformation reaches or exceeds
tolerance of the pipe, leakage or structural damage may be generated. During the
construction of underground engineering, the main reason of damage or function
failure of pipelines is differential settlement of the surrounding soil induced by
excavation. Generally, there are two main failure modes of underground pipelines
ASCE
20
21
under the action of soil movement caused by construction of the underground

facilities. The first is that pipes lose working ability due to the crack caused by
additional tensile stress, and even break. The second is that pipe sections are intact, but
the joint position is damaged because the joint rotation angle is too large or pulled out,
resulting in the leakage. The failure mode of pipeline may be one of them or all of
them.The resistence of underground pipelines for surrounding soil movement is
mainly determined by their relative stiffness between soil and pipe and admissible
pipe joints displacement. If the existed underground pipelines are vertical above the
tunnel, they mainly affected by the increase of pipe section bending stress and the joint
rotation angle due to subsidence displacement of surrounding soil induced by
tunnelling. If they are parallel above the tunnel, their infulences are mainly the action
that additional axial stress caused by surrounding soil movement. Figure 1 is the
influence feature of tunnelling on underground pipelines.
(a) Pipeline vertical above the tunnel
(b) Pipeline parallel above the tunnel
FIG. 1. Properties of pipeline additional movement induced by tunnelling

CRITERIA OF PIPELINE SAFETY EVALUATION
Underground pipelines can be generally divided into two types: the jointed ductile
pipeline, also known as the flexible pipeline, and the rigid pipeline. Generally, the
flexible pipe refers to the non-integral section-coupling pipe, with obvious joints
between pipe sections and a small amount of allowed rotation. Namely, after the
mechanical deformation, the pipe can still work normally under a condition that the
joint rotation angle and additional stress do not exceed their allowed values, such as
concrete sewage pipe with spigot and socket, cast iron pipe, flexible joint pipe and etc.
However, the rigid pipe is often integral and continuous. Its joints is not obvious,
usually the welding. The pipe does not allow large distortion; otherwise it would lead
to fracture failure for excessive bending stress, such as welded pipe, gas line and etc.
To judge whether the tunnel construction affects the safety of above underground
pipelines, first of all is to determine the safety control stanards of pipelines, then to
adopt an appropriate method to predict their additional deformations induced by
construction. They are safe if the amount of deformation is less than the control
standard; otherwise it is not. In practice, the joint rotation angle, the joint displacement,
and the additional stress are generally selected as the safety control index of pipelines.
ASCE
22
As deformation of the flexible pipeline concentrates on joints between the sections

accompanying with soil movement, the safety control indexes are the joint rotation
angle and the joint displacement. But for the rigid pipelines, the safety control index is
mainly the additional stress, because soil movement may cause the bending internal
force exceeding limits. Moreover, for the electric power and telecommunication
cables, they are the flexible materials with properties of strong deformation
adaptability and easy maintenance. Therefore, there is no particularly requirement for
safety controls for them in practice.
This paper focuses on study of the municipal flexible pipeline, and discusses the
methods of pipeline safety evaluation caused by tunnelling, distingusihing two typical
relative position between the pipe and tunnel the pipeline vertical above tunnel and
the pipeline parallel above tunnel. In order to conveniently link up with different
industrial standards of pipeline and apply easily, hypothesis that the cross section of
pipe sections only generates bending stress without bending deformation,
accompanying with the soil movement, is adopted in this paper. Considering the
mutual conversion between the joint allowed rotation angle and gaping dispacement of
the pipe, the joint rotation angle can be used as the safety control index to analyze
the tolerable differential settlement of pipes and the slope of settlement curve ratio of
surrounding soil. If [ ] , the pipeline is safe;otherwise, the pipeline is unsafe.
DETERMINATION OF THE JOINT ROTATION ANGLE
Soil Movement Laws
According to Attewell et al. (1986), the vertical and horizontal displacement of
ground surface induced by tunnelling can be expressed as follows:
W = Wmax exp(
V=
G (s) =
y2
x
) G( )
2i 2
i
y
W
zt z
(1)
(2)
t 2
exp
dt
2
(3)
where the direction of x is the axial of tunnel, and at the position of working face
x = 0 ; y is the horizontal direction of tunnel cross section, and at the center y = 0 ; z is
vertical direction of tunnel cross section; Wmax is the maximum settlement of ground
surface; i is the surface settlement trough width coefficient; zt is the depth of tunnel
axis; G(s) is Gauss normal distribution function.
According to Mair et al. (1993), the law of soil movement induced by tunnelling
also conforms to the surface settlement trough curve, which can be discribed by Gauss
normal distribution function. The volume of surface settlement trough formed by
tunnel excavation is equal to the ground loss under the undrained condition. Therefore,
the relationship between the maximum soil settlement of depth of z and the maximum
surface settlement ( z = 0 ) can be obtained:
Wz ,max = Wmax
i
iz
(4)
With increase of the depth, the settlement trough width coefficient of soil iz will
reduce. From the study of Jiang et al. (2004), the settlement trough width coefficient
of soil in different depth is given by
ASCE
23
z n
(5)
)
zt
where exponent n from -0.45 to -0.2, general value is -0.3.
Mostly, joints of the flexible pipe are provided with joint fillers to adapt a certain
degree of joint open under conditions of small deformation of stratum. In this paper,
the pipeline is assumed not to separate from surrounding soil, thus the deformation
curve of pipeline can be approximately replaced by soil settlement trough curve in the
depth which pipelines lie in.
iz = i (1
Joint Rotation Angle of Pipeline Vertical above the Tunnel

The Gauss normal curve can be used to discribe soil movements of tunnel cross
section induced by tunneling, which means that the maximum deformation of soil is
above the axis of tunnel. In that case, the maximum deformation of jointed ductile
pipeline occurs on vertical plane through the tunnel axis. Therefore, if a pipeline is
vertical above tunnel and the length of pipe sections is same, the most unfavorable
condtion is that the pipeline joint is located above axis of tunnel. And the relationship
between soil movement and pipeline deformation induced by tunnelling can be shown
in Figure 2.
Wp
Wpmax
D
h
FIG. 2. Geometric relationship between soil movement and pipe deformation

induced by tunnelling
In Figure 2, l is assumed to the length of pipe section with the diameter of D , and
the depth of pipe is z p . Pipe sections after deformation induced by tunnelling are
symmetrical along the vertical plane of tunnel axis, being located in a circular curve
with radius of R . If the maximum deformation of pipe sections is Wp, max , ignoring the
rise of arch curve at place of open joints, Eq. (6) and Eq. (7) can be given
Wp = Wp,max h = Wp.max e
h = l sin( / 2) =
l
2D
( / 2 + l cos )2
2i p2
(6)
(7)
where i p is the soil settlement trough width coefficient in depth of pipe. Depending on
the relationship between the pipe joint rotation angle and the joint displacement,
approximatively, / 2 + l cos / 2 l , the following formula can be derived
ASCE
24
=2
Wp,max D
l
(1 e
l2
2i p 2
(8)
Substituting Eq. (8) into Eq. (4) and Eq. (5), and considering D and
Wp, max = Wz, max , the pipe joint rotation angle becomes
=
2Wmax
l2
[1
exp(
)]
2i 2 (1 z p /z t )0.6
l (1 z p / zt )0.3
(9)
where z p is the depth of pipe. Obviously, if given in advance the allowed pipeline
joint rotation angle , the allowed maximum ground settlement Wmax can be easily
calculated by Eq. (9), thus providing a basis for determining the limit alarming value
of ground settlement in practice.
Joint Rotation Angle of Pipeline Parallel above the Tunnel
If the horizontal distance between the pipe and tunnel is assumed to y p , and the
tunnel is excavated paralleling along pipelines, the soil movements of cross section
induced by tunnelling also can be discribed by Eq. (1). At the positons where it is in
front of the excavation face x = i and behind x = i , the soil deformation curvature is
the largest, and so does the pipe. Assuming the same length of pipe sections as well,
and considering the most dangerous condition that the working face of tunnel is
located behind the pipe joint with the distance x = i , the relationship between soil
movement and pipeline deformation induced by tunnelling can be shown in Figure 3.
x=0
x=i
Wmax
S1
Si
S2
FIG . 3. Geometric relationship between soil movement and pipe deformation

induced by tunnelling
By this time, ground settlements at the locations of the dangerous pipe joint and
adjacent joints are Si , S1 and S2 , respectively. Considering the fact that allowed joint
rotation angle of the flexible pipeline is very small in actual engineering, its
calculation can be given as the following equation:
= sin -1 (
S1 -Si
S -S
S -2S + S2
)-sin -1 ( i 2 ) 1 i
l
l
l
(10)
Through Eq. (10), the pipeline joint displacement can also be obtained D .
Further considering the effect of horizontal displacement of soil on pipeline joints
deformation, the soil movement induced by tunnelling is the vectorial sum of the
vertical displacement W and lateral horizontal displacement vector and V , namely:
ASCE
25
S = 1 + [y p / ( zt zp )]2 Wp
(11)
Solving this equation for x = i p 1 , i p and i p + 1 , and y = y p , S1 , Si and S2 can be

obtained receptively. Similarly, with the relation of Wp, max = Wz, max , substituting the
results into Eq. (4) and Eq. (5) leads to an expression of pipeline joint displacement
discribed by the ground sectional groove coefficient.
ENGINEERING APPLICATION
An urban pedestrian underpass was excavated to pass through a cast iron water
supply pipe with the socket and spigot joint, those buried depth is 8m and radius is 3m
And the pipe is vertical above the tunnel, with radius of 0.6m, depth of 3m and
section length of 5m. Internal friction angle of the soil around the pipeline is 30o, and
ground loss ratio of excavation is 5%. According to the criteria of socket and spigot
connected cast iron pipe (Attewell et al. 1986), an allowed joint rotation angle is
taken 1.5o as the safety evaluation criterion of pipeline deformation induced by
tunnelling. Through the above method in paragraph 4.2, the pipe joint deformation is
calculated as follows:
3
V = r 2 5% =1.41m where V is the groud loss; i =
zt
2 tg (45 / 2)
=5.54m
where i is the ground settlement trough width coefficient; thus the maximum ground
settlemtment Wmax =
V
2 i
=0.10m.
Due to z = 3m , zt = 8m , D = 0.6m , l = 5m , it can be derived iz = 4.8m ; subtituting to

Eq. (4), Wz ,max is gived by Wz ,max = 0.115m .
Considering Wz, max = Wp, max , =0.0115m; through Eq. (9), joint rotation angle of
the pipe is caculated =1.1o, which is less than 1.5o. Therefore, it can be initially
conclued that the pipe is safe under conditions of the case.
If the pipeline is parallel above the tunnel with horizontal distance of 2m, and
ceteris paribus, the joint rotation angle can be caculated by above method in
paragraph 4.3. According to Eq. (11), the result can be given by S = 1 + [2 / (8 3)]2 W .
Substituting x = 4.8 , 0.2 and 9.8 , and x = 4.8 , 9.8 and 0.2 into Eq. (1), Eq.(4)
and Eq.(5), and combining with Gauss normal distribution table, through Eq. (10), the
joint rotation angle of pipeline can be derived =0.28o, which is less than 1.5o.
Similarly, it can be initially determined that the pipe is safe under this condition.
CONCLUSIONS
Based on the commonly accepted formula of soil movement induced by tunnelling,
this paper proposes a simple method on damage assessment of the buried flexible
pipeline above urban shallow transit tunnel engineering, which takes into account
spatial position relations between the pipeline and tunnelling, and deformation curve
shape of the flexible pipeline, with properties of fewer parameters and easy to be
ASCE
mastered by engineers. Especially, it is suitable for quantitative safety prediction of

the shallow-buried pipelines in a condtion that there are no enough formation datas
and detailed informations of pipelines before tunnelling. At present, although some
corresponding norm of tunnel proposes alarm values of ground surface subsidence,
most of them are the empirical data and lack theories to support (Wang 2004). The
method of this paper provides a useful reference to scientifically determine the surface
monitoring alarm of urban shallow-buried tunnel engineering.
Need of special note is that the study is based on hypothesis that the pipe and round
soil has synchronous movement induced by tunnelling, ignoring the impact that
stiffness of a single pipe section resists round soil movement. But it is safe for
engineering evaluation. Further study should be carried out to consider interactions
between the pipeline and soil.
REFERENCES
Attewell P. B, Yeates J, Selby A. R. (1986). Soil movements induced by tunneling
and their effects on pipelines and structures. Blackie and Son Ltd, London.
Huang M. H., Zhao Z. H., Wang Y. X., Ma X. C. (2013). Controlling criteria of
sectional deformation of underground flexible pipelines. China Trenchless
Technology, Vol. (6): 1-5.
Jiang X. L., Zhao Z. M., Li Y. (2004). Analysis and calculation of surface and
subsurface settlement trough profiles due to tunneling. Rock and Soil Mechanics,
Vol. 25(10): 1542-1544.
Klar, A., Marshall, A. M., Soga, K., and Mair, R. J. (2008). Tunneling effects on
jointed pipelines. Canadian Geotechnical Journal, Vol. 45(1):131-138.
Liu J. H., Hou X. Y. (1997). Excavation engineering handbook. China Building
Industry Press, Beijing.
Mair R. J, Taylor R. N., Bracegirdle A. (1993). Subsurface settlement profiles above
tunnel in clays. Geotechnique, Vol. 43(2): 315-320.
Moser A. P. (2003). Design of underground pipeline. China Machine Press, Beijing.
Tian W. G., Feng Y. L. (2012). Study on deformation control criterion for
underground pipelines in the underg rond construction process. Special Structures,
Vol. 29 (6): 85-90.
Vorster, T. E. B., Klar, A., Soga, K., and Mair, R. J. (2005). Estimating the effects of
tunneling on existing pipelines. Journal of Geotechnical and Geoenvironmental
Engineering, Vol. 131(11): 1399-1410.
Wang M. S. (2004). Theory and technology of mining method for shallow tunnels.
Anhui Education Publishing House, Hefei.
Wang T., Liu W. N., Li X. G., et al. (2006). Current situation and prospect of studies
on tunneling induced effect of adjacent pipelines. ChinaRailway Science, Vol.
27(6): 117-123.
Wei G., Wei X. J., Qiu X. G, Yao N. (2009). 3D numerical simulation of effect of
underground urban street-passage tunnel construction on adjacent pipeline.
Chinese Journal of Rock Mechancs and Engineering, Vol. 28(Sup1.): 2853-2859.
Xiang W. G., Xu Y. S., Jiang H. H., Tan W. Z. (2010). Pipeline safey evalution of
tunnelling. Railway Engineering, Vol. (6): 70-73.
ASCE
26
27
Mechanical Behavior Analysis for Parallel Small Spacing Tunnels in a Loess

Region
Xiangxing Kong1
1
The First Highway Survey and Design Institute of China Communications Construction Company
Ltd., Xian 710075, China. E-mail: xiangxingkong8129@163.com
Abstract: For satisfying the requirements of normal running and enlarged outer
section in stabling line, parallel neighborhood metro tunnel with two different inner
sections is adopted in Xian Urban rail transit line one between North Zaoyuan Road
and Hancheng Road. Based on the characteristics of different inner sections and small
clear spacing, 2D-FEM was applied to simulate and analyze four construction
schemes, of which shield method was adopted in left tunnel with small inner section
and double-heading construction method or cross diaphragm method for right tunnel
with large inner section. The results show that compared with the construction scheme
that the tunnel with small inner section is firstly excavated, the larger tunnel by
double-heading constructed is first accomplished can evidently reduce the disturbance
effect on surrounding rock, reach better state mechanical of tunnel lining, in order to
ensure safety construction.
INTRODUCTION
As the city subway, underground tunnels are taken to two equal sections[1], only
meeting special engineering needs it is designed with two unequal sections, for
example crossover, contact lines and stop (deposit) car line and so on. The tunnel with
parallel non large section has carried out preliminary research. Through the
establishment of numerical model[2], the influence of supporting mechanical behavior
and stability of surrounding rock of unequal span highway tunnel are analysed; also,
construction mechanics and surface settlement of double line tunnel with
antisymmetric sections are also studied[3], but they are still need to further study.
In order to meet the two lane and the right line stop line section expanding
engineering function, the left line of small section tunnel with diameter 6m is
constructed by shield method, large section tunnel right line height with 9.31m width
11.24m is constructed by the new Austrian tunneling method[4]. A hole for the shield
and another hole for the new construction scheme of NATM is seldom in the domestic
Loess Area.
ASCE
28
In view of the novelty and atypical of large section tunnel, shield method and
NATM construction in loess area, and the small clear distance tunnel complexity, a
small section of the left line shield tunnel and the right line tunnel by double side
heading method or CRD method. The double line 4 scheme has dynamic numerical
simulation and scheme comparison and selection of construction mechanics state of
surrounding rock, focuses on the initial support, the two tunnel lining and roof lining
and segment structure sinking, surface displacement and middle soil stress were
compared and analyzed, in order to provide theoretical basis and technical support for
the best construction scheme and safe and smooth construction project. It also can
provide guidance and reference for the subway construction in Loess area.
ENGINEERING BACKGROUND
Xi'an city rail traffic engineering is the first subway tunnel construction in loess
areas, the 1st line Zao yuan Bei Lu Station - Hancheng Road station tunnel is located
in the city traffic main road Zaoyuan West below, the tunnel area is one of the hub city
of Xi'an traffic greatly, especially after the completion of the subway station and the
nearby intercity passenger transfer passenger station, its layout is shown in Figure 1.
Combined with the general linear programming, the thickness of tunnel soil covering
the range of 9~10 m, the vertical arrangement as shown in Figure 1.
Green belt
Tunneling
direction
Shield tunneling
7
Bus station
Zaoyuan road
8
Drain piple (PS)

Hencheng
9 road
7
Tunneling
Green belt
NATM tunneling
9
Green belt
direction
Residential area
Fig.1 Plan view of parallel neighborhood metro tunnels with two different inner
sections
NUMERICAL SIMULATION MODEL
As a non-large sections of shield method and NATM small net span loess tunnel, its
difficulty and complexity are large, in order to close to the actual conditions, selection
and optimization of reasonable the dynamic numerical analysis program is set up to
simulate construction scheme.
The calculation program uses two-dimensional plane strain model. It is assumed
that the soil with Drucker-Prager strength criterion and isotropic hardening
elastoplastic constitutive model of soil, the quadrilateral element mesh division, and
the shield segment is simulated by beam element, advance grouting pipe
reinforcement effect to improve the stability of the surrounding rock of the
quadrilateral element simulation, the initial support and steel the grid element
ASCE
29
simulation, two lining is simulated by beam element, considering the initial lining and
lining between the two waterproof board using contact element simulation. In order to
ensure the accuracy of the dense units around the tunnel are used in the construction
scheme of finite element model as shown in Figure 2.
Fig.2 Mesh of finite element model

According to the data of engineering geology, soil and calculation of related
parameters are shown in Table 1, and parameter calculation of retaining structure is
shown in Table 1.
CONSTRUCTION SCHEMES SIMULATION
The left line of small section tunnel excavation diameter is 6.0m, C50 concrete
segment thickness is 0.3m, the impermeability grade of S10, specific ring lining
section shown in figure 3. Large sections of the right line of NATM tunnel excavation
is high 9.31m, width 11.24m, C25 lining thickness 0.35m, HPB335 thick 0.25m steel
grille. In addition, in the loose loess shallow subway tunnel excavation, construction
of pre grouting pipe in the tunnel arch can improve the condition of surrounding rock
to ensure the stability of the tunnel face, and it is very important for increasing the
stability of tunnel. The specific scheme is diameter 42 3.5 (thickness 3.5mm)
grouting pipe, length 2.5m, angle 15 , circumferential spacing 0.3m, vertical spacing
of 1.0m; pressure injection of cement - water glass grout, grouting pressure control in
0.6~3.5MPa.
ASCE
Table 1 Parameters of support structures
Because the CRD method and the double side drift method is suitable for the large
section and small spacing shallow buried loess tunnel. The total of 4 schemes of
numerical simulation comparison with analysis to determine the interval tunnel
construction scheme optimization which is the small section tunnel large section of
shield method for double side heading method or CRD method in the construction
method of construction order double successively through and large section. The
following specific construction process dynamic simulation:
(1) Small section tunnel in left line firstly, the right line of large section tunnel by
double side heading method finally.
At first, completed the small section tunnel left line, simulation step of shield
construction into 2 steps: excavation; the application of segment lining. Then
the big cross section tunnel in right line, double side heading method construction
process simulation into 13 steps: the left and right up the steps: in the application of
heading arch part of advanced small pipe grouting; excavation; the vertical steel
grille, shotcrete; both sides heading down the steps followed in the excavation of
the frame; steel grille, shotcrete; central heading up the steps in the application
of arch part advanced small pipe grouting; the excavation; the vertical steel
grille, shotcrete; central drift step sequentially during excavation, removal step
temporary bracing; the vertical steel grille, shotcrete; central drift next step is (11)
excavation ;(12) sprayed concrete; (13) applied two lining.
Fig 4 Construction sequence of double-heading method
ASCE
30
Fig 5 Construction sequence of CRD method

(2) Small section tunnel in left line firstly, the right line of large section tunnel by
CRD method finally.
At first, completed the small section tunnel left line, simulation step of shield
construction into 2 steps: excavation; the application of segment lining. Then
the big cross section tunnel right line, simulation step construction with CRD method
into 11 steps: Step 1 right sequentially as arch advanced small pipe grouting;
excavation; the vertical steel grille, shotcrete; right down the steps in the
excavation; the vertical steel grille, shotcrete; left on the application of step
sequentially arch part of advanced small pipe grouting; the excavation; the
vertical steel grille, shotcrete; left down the steps in order to excavation; the
shotcrete lining ;(11) applied two times.
(3) Large section of double side heading method right tunnel first, shield small
section of left tunnel.
(4) CRD with large cross section tunnel right line first, shield small section of left
tunnel.
Scheme (3) and (4) respectively by the first double sidewall drift method and CRD
method through large section tunnel in right line, then the small section tunnel left line
was constructed by shield method, its construction step simulation method and scheme
(1) and (2) the same, is only has different construction step, not to repeat again.
RESULTS AND ANALYSIS
It is numerical simulation and analysis of the non-large sections of shield method
and NATM loess subway tunnel small spacing of 4 design and construction scheme,
especially surrounding rock mechanics state of initial support, two lining and lining
composed of tunnel lining construction mechanical state and vault subsidence, surface
displacement and in soil force.
(1) Mechanics analysis of the right line tunnel with large section
It can be seen from Table 2, scheme 3 produced in construction of initial support
axis force is maximum, scheme 4 is less than 3, but the difference is small, than
scheme 1 and 2. To the hole on the right side of haunch for example, scheme 1, 2, 4,
and 3, respectively, the 5.4%, 7.2% and 1.1%. This text through the NATM tunnel
with large cross section of shield tunnel after excavation of small section of initial
ASCE
31
32
support axial force than the first through hole after digging a hole was too big, the
double sidewall drift method for large cross-section tunnel results in maximum.
Accordingly, scheme 3 is the maximize of the tunnel lining supporting capacity.
Table 2 Calculation results of preliminary lining axial stress(unit: kN.m)
Construction
Left
vault
ratio/%
Right
Scheme
hance
hance
Left
vault
Right
scheme1
1263.0
761.3
1239.0
-6.4
-5.5
-5.4
scheme2
1237.0
720.5
1216.0
-8.4
-8.5
-7.2
scheme3
1350.0
835.1
1310.0
scheme4
1330.6
826.5
1295.0
-1.4
-0.6
-1.1
Ratio=(result of selected scheme-result of scheme3)/ result of scheme3

For the tunnel lining (lining of the two negative bending the outer surface, the inner
surface of the compression tension), scheme 3 and 4 significantly lower than scheme 1
and 2, of which 3 were minimal solution. Scheme 1, 2, 4, and 3 compared in the right
arch waist are respectively in 13.2%, 74.3%, and 8.3%. This shows that the two lining
of negative moment first through the NATM tunnel with large cross section of shield
tunnel after excavation of small section of a first through hole after digging a big hole
was small, the double side drift method is used for big hole results in minimum.
Accordingly, scheme 3 greatly increase the safety reserve of secondary lining.
Table 3 Calculation results of second lining moment(unit: kN.m)
Construction
Left
Scheme
hance
scheme1
48.4
scheme2
vault
ratio/%
Right
hance
Left
vault
Right
33.1
34.3
9.3
13.2
55.2
87.3
52.8
24.6
78.0
369.
4
scheme3
44.3
18.6
30.3
scheme4
46.6
22.1
32.8
5.2
18.8
74.3
-
8.3
(2) Mechanics analysis of left line tunnel with small section

It can be seen from Table 4, segment lining negative moment of scheme 3 is the
smallest, scheme 4, 1, and 2, were less than the scheme. This shows that the segment
lining moment first through the NATM tunnel with large cross section of shield
tunnel after excavation of small section of a first through hole after the excavation of
large hole to be small, the double sidewall drift method for large cross-section tunnel
results minimum. Accordingly, scheme 3 is most conducive to the safety to ensure the
linings.
Table 4 Calculation results of segment moment(unit: kN.m)
ASCE
Construction
Left
Scheme
hance
scheme1
7.8
scheme2
vault
ratio/%
Right
hance
Left
vault
right
10.2
8.1
290.0
251.7
326.3
11.1
13.6
12.9
455.0
369.0
578.9
scheme3
2.0
2.9
1.9
scheme4
4.5
3.9
5.6
125.0
34.5
194.7
CONCLUSION
The left small section tunnel constructed by shield and the right section tunnel by
double side heading method or CRD method with small spacing parallel sequence is
researched. The total of 4 kinds of construction schemes on theory and technology
selection and optimization analysis with the numerical simulation results of the
scheme of tunnel lining, draws the following conclusion:
(1) Constructed firstly by NATM tunnel with large section and small section tunnel
boring shield method construction scheme has less disturbance to the surrounding
rock.
(2) Constructed finally the shield of small section tunnel, a CRD method is less
safety and reliable lining than the use of double side heading method first large section
tunnel. Because of the double side heading method slow construction progress and
high cost, to ensure the safety of the tunnel under the premise, the adjustment will be
optimized for the CRD method is also worth exploring and trying.
(3) Through the dynamic simulation of the construction process, the subway tunnel
construction scheme of small spacing parallel analysis and scheme comparison,
according to the complexity of the project, we should strengthen the monitoring and
protection of surface deformation, and the mentioned research will provide technical
support and practical guidance for the construction of subway tunnel of Loess Area in
the future.
REFERENCES
Jiang Xiaorui. Finite Element Simulation of Construction Method of Crossover
Tunnel with a Unequal-span Double-arch [J]. Railway Standard Design2009
10:103-105.
Kong Xiangxing, Xia Caichu, Qiu Yuliang, et al. Study on Construction Mechanical
Behavior of Parallel Small Spacing Metro Tunnels Excavated by Shield and Cross
Diaphragm (CRD) Method in Loess Region [J]. Rock and Soil Mechanics2011
322): 516-524.
Shanghai Tunnel Engineering & Rail Transit Design and Research Institute.
Construction Design of rail transit Line One in Xian[R]. Shanghai: Shanghai
Tunnel Engineering and Rail Transit Design and Research Institute2008
Wang Min, Liao Shaoming, Hou Xueyuan. Ground Deformation Prediction of Shield
Tunnelling in Closely Overlapped Subway Tunnel Construction [J]. Journal of
Tongji University, 2004,32(9):1141-1146.
ASCE
33
34
Analysis of Tunnel Monitoring Results Based on the Modulus Maxima Method of

Wavelet Transform
Xiangxing Kong1
1
The First Highway Survey and Design Institute of China Communications Construction Company
Ltd., Xian, China. E-mail: xiangxingkong8129@163.com
Abstract: In recent years, the theory and method of wavelet analysis is widely used in
signal processing, pattern recognition, data compression, image processing, and
quantum physics. Compared with modulus maxima, wavelet packet decomposition
and coefficient shrinkage de-noising method of wavelet transform, their advantages
and disadvantages are analysed and summarized, and their respective scopes are
obtained. The Noissin chosen as the original signal with noise is analysed and denoised by the modulus maxima method of wavelet transform, meanwhile the usage
conditions and key computing parameters are also obtained. Finally, the modulus
maxima method of wavelet transform are successfully adopted to de-noised the
monitoring results of shield tunnel, the data revised are reliably provided for tunnel
healthy diagnosis.
INTRODUCTION
In 1910, Haar proposed the theory of wavelet orthonormal basis, which is the
earliest wavelet theory[1]. In 1936, Littlewood and Paley established binary frequency
component grouping theory for Fu Liye series, which carries on the division to the
frequency. And it is the earliest theory source of multi-scale analysis[2]. In 1981,
according to the group theory the French physicist Morlet firstly proposed the concept
of wavelet analysis in geological analysis of geological data[3-5]. As signal analysis
expert, Mallat proposed the concept of a multi-resolution analysis, and gave a general
method for constructing orthogonal wavelet basis[6]. The most commonly used multi
resolution analysis technique has two kinds: one is the analysis of time limited
resolution, and the other is spline multi-resolution analysis.
Comparative Analysis of De-noising Method

Three are a lot of signal de-noising methods based on wavelet transform, which can
be summarized as the following three categories.
(1) modulus maxima de-noising method based on the wavelet transform mainly is
suitable for the signal mixed with white noise, and contains more singular points. The
theory of signal singularity detection and the wavelet transform modulus is proposed
based on maximum principle by Mallat. It effectively uses propagation characteristics
ASCE
of different in multi-scale space in de-noising signal and white noise. In the de-noising
and singularity information, this method effectively retain signal and no excess
concussion, and is a very good estimate.
However, when reconstruction using alternating projection method, in order to
ensure the accuracy of reconstructed signal and improve the signal-to-noise ratio, it
usually is iterative dozens of times, so the calculation speed is slow. In addition, it is
important to choose the wavelet coefficients of the wavelet decomposition scale, and
small scale is easily affected by noise, and large scale could make signal loss some
important local singularity.
(2) wavelet packet decomposition de-noising method based on coefficient of
shrinkage mainly is suitable for the signal mixed with white noise case. The method
based on Donoho and Johnstone put forward discrete wavelet transform. Wavelet
packet decomposition can provide a more precise method for signal, by the signal
frequency band multi hierarchy further decomposition of the high frequency part is
without subdivision in wavelet transform. According to the analysis of the
characteristic of the signal, it is adaptive selection of frequency band, the frequency
and signal phase matching. Compared with wavelet transform, wavelet packet
decomposition can obtain more rich of the time-frequency localization information,
and more suitable for the analysis of nonstationary signal processing.
However, de-noising effect depends on ratio of the signal and noise, and is
especially suitable for high the ratio of signal and noise . In addition, in some cases,
such as the existence of discontinuous point signal, de-noising will produce Vo Gibbs
phenomenon. Also, the selection of threshold plays an important effect on the denoising effect.
(3) wavelet transform de-noising method based on translation invariant mainly is
suitable for the signal mixed with white noise and contains a plurality of discontinuous
point of the situation. This method is a kind of de-noising method based on Donoho
threshold method to improve the de-noising threshold method, in order to solve the
possible pseudo Gibbs phenomenon occurred in the signal the discontinuous point.
The basic idea is cyclic shifts of the original signal with noise in translation within a
certain range. In order to obtain a new signal phase difference in the time domain and
the original signal, the soft threshold function or hard threshold function of wavelet
coefficients are shrunk, which can get the signal de-noising by the inverse wavelet
transform.
The signal is contrary to the cyclic shift, which get de-noised signal and the original
signal with the same phase. Changed the translation, this process is repeated, on
average the results obtained to estimate signal noisy signal after de-noising.
However the process of de-noising in signal is convenient, and the calculation speed
is very slow.
Modulus Maximum Method Based on Wavelet Transform
Noissin is selected as the original signal with noise, and the wavelet decomposition
with 4series, wavelet function chosen as db3. The original signal is shown in Figure 1,
after wavelet transform coefficient levels similar to figure 2, the detail coefficient is
shown in Figure 3, the wavelet transform modulus maxima de-noising after all levels
remained modulus maxima is shown in Figure4, using wavelet modulus maxima de-
ASCE
35
noising method of reconstruction signal as shown in figure 5. From the figure, the
noise of original signal is mainly concentrated in the 1,2 level, through the two level
after de-noising, and then through the decomposition of 3,4 level, signal to remove the
noise coefficient of the vast majority, to retain the integrity of the signal coefficients,
and finally through the reconstruction on the restoration of the signal after de-noising.
Wavelet transform modulus maximum de-noising method not only can effectively
suppress noise, but also can well preserve the high frequency details in the original
signal. The de-noising signal distortion is small, and the de-noising effect is very
good. At the same time, the method of de-noising signal with no excess concussion is
the original signal and a very good estimate, and is good surface quality. It is
important to choose the wavelet decomposition scale, which should be a suitable
decomposition scale wavelet transform.
It is not yet to form a unified standard of perfection in the application of theoretical
research and the methods of wavelet de-noising. There are various methods applicable
conditions and limitations in practice. Overall, due to signal the discontinuous point
can produce pseudo Gibbs phenomenon, wavelet packet decomposition coefficients
shrinkage often signal to some difficulties in the process of noise based on wavelet
transform. Translation invariant de-noising method can improve the above
shortcomings, but in practice slow computational speed and poor convenience.
Analysis and Application
Health monitoring data is an important scientific basis for shield tunnel health
diagnosis, and its reliability and accuracy directly affect the health status of the shield
tunnel evaluation and analysis of results[7,8]. In fact, due to a variety of external
objective conditions, such as engineering environment, climate conditions and
measuring instrument, often contain errors caused by tunnel monitoring data. It is
difficult to accurately determine the non-stationary signal, which is self-mutation
expression, the effects caused by the mutation phenomenon, or is the comprehensive
embodiment of the two combined. Because the effects of these mutations are in the
signal curve showed spike and mutation, and they are all in the high frequency part of
signal. To accurately analyze this kind of signal, the wavelet transform modulus must
be suitable for the characteristics of the maxima method of de-noising, which can
effectively remove the noise interference, and can be well preserved the original signal
high frequency detail.
Based on the above principle, the wavelet modulus maxima de-noising algorithm
can be summarized as follows:(1) The signals contain noise of wavelet transform into
two, general 4~5 scales, and calculate each wavelet coefficient modulus
maxima.(2)With the biggest scale as the start and the threshold selected, if the absolute
value is less than the corresponding value of extreme point, then remove the extreme
point, or to be retained, which can get scale new modulus maxima..
ASCE
36
37
N/mm
time/s
Fig. 1 Noissin chosen as the original signal with noise
a1
d1
time/s
time/s
a
d2
a2
time/s
time/s
b
d3
a3
time/s
time/s
c
d4
a4
time/s
time/s
d
Fig. 2 Approximate coefficient
ASCE
d
Fig. 3 Detail coefficients
38
M
A
X
M
A
X
time/s
time/s
(a)j=1
M
A
X
(b)j=2
M
A
X
time/s
time/s
(c)j=3
(d)j=4
Fig. 4 Modulus maximum of wavelet transform after signal de-noising
N/mm
time/s
Fig. 5 Compared with health monitoring results and de-noising and reconstruction of tunnel lining stress
ASCE
Conclusion
Compared with modulus maxima, wavelet packet decomposition and coefficient
shrinkage de-noising method of wavelet transform, the advantages and disadvantages
are analyzed and summarized, and the respective scopes are obtained.
(1)The Noissin chosen as the original signal with noise is analyzed and de-noised
by the modulus maxima method of wavelet transform, meanwhile the usage conditions
and key computing parameters are also obtained.
(2)The modulus maxima method of wavelet transform are successfully adopted to
de-noised the monitoring results of shield tunnel, the data revised are reliably provided
for tunnel healthy diagnosis.
References
Jiang Peng. Application Research of Wavelet Theory in Noise and Data
Compression[D]. Hangzhou, Zhejiang University, 2004
Gao Zhi, Yu schallsea. Principle and Application of Wavelet Analysis in Matlab
Toolbox[M]. Beijing, National Defence Industry
Press, 2004
Ge Zhexue, Sha Wei. Wavelet Analysis Theory and Matlab R2007 Realization[M].
Beijing, Publishing House of Electronics
Industry, 2007
Guo Jian. Method of Bridge Health Monitoring Based on Wavelet Analysis [J].
Engineering Mechanics, 2006,23 (12): 129-135
Lin Dachao, Shi Huiji, Bai Chunhua. The blasting Vibration Wavelet Transform
Based on Time-Frequency Characteristic Analysis
[J]. Chinese Journal of Rock Mechanics and Engineering, 2004,23 (1):, 101-106
Panou K DSofianos A I. A Fuzzy Multi-criteria Evaluation System for the
Assessment of Tunnels Surface RoadsTheoretical
Aspects-Part I. Tunnelling and Underground Space Technology200217(2)
209219.
Kong Xiangxing. Analysis of Lining Deterioration Effect on Underwater Shield
Tunnel Deformation [J]. Highway Engineering,
2012,1 (37): 26-3
ASCE
39
40
Study of the Diseases of Shield Tunnels and Its Reasons

Baotong Shi1 and Xiangxing Kong2
1
The First Highway Survey and Design Institute of China Communications

Construction Company Ltd., Xian, China. E-mail: 117021654@qq.com
2
The First Highway Survey and Design Institute of China Communications
Construction Company Ltd., Xian, China. E-mail: xiangxingkong8129@163.com
Abstract: With more and more shield tunnels are constructed in highways and metro,
the tunnel structure damages become a key problem that has important effect on the
safe operation of the tunnels. In this paper, the possible damages of the shield tunnels
during its service life are summarized based on the analyses of the influencing factors.
The main damages of the shield tunnel include cracking and spalling of concrete, steel
corrosion, deformation of tunnel linings collar joint or juncture-open and the leakage
at segment, lining joint. On this bases, the reasons for these damages are discussed
detailedly.
INTRODUCTION
The shield tunnel is a kind of underground structure widely used in metros in China.
The safety and durability of shield tunnels will decrease with its operation time and
the damages of the shield tunnels appear gradually during the service life. Although
many researches on the damages of mountain tunnels and railway tunnels have been
performed, only a few researches on the damage of the shield tunnel have been
reported in the literatures.
As the shield tunnel used in metros usually locates in soft soil areas such as
Shanghai and Hangzhou, where the groundwater is very rich, the damages of the
shield tunnel are mainly caused by the groundwater. In this paper, the main diseases of
the shield tunnel are firstly summarized, then the reasons for these diseases are
explained. Finally, the treatments of the diseases of the shield tunnel are introduced
briefly.
MAIN DISESES OF SHIELD TUNNELS
According to the field investigations of many shield tunnel used in metro, the main
diseases of shield tunnels are shown as following:
Cracking and Spalling of Concrete
Cracks in the concrete are the most common disease in railway tunnel and shield
tunnel. Generally, the concrete often works with cracks. The effects of cracks on
shield tunnel are firstly characterized by the safety and durability of the structure.
Furthermore, cracks will also influence the performance of the structure. For example,
the groundwater will seep through the cracks, which will result in the spalling of the
ASCE
41
lining concrete. Fig.1 shows the typical cracks and spalling of lining concrete in a
metro shield tunnel constructed in soft soil area, among which fig.1a indicates the
micro-cracks existing in the segment, and fig1b indicates circumferential cracks in one
of the segments that have become the flow channel of groundwater and slurry, while
fig.1c and fig.1d indicate the longitudinal cracks in the segment, which may be caused
by the erosion and deterioration of concrete. Although the cracks shown in fig.1 are
near the surface, it will develop inward and result in the erosion of the steel bars if
nothing is done to the existing cracks.
(a)
(b)
(c)
ASCE
42
(d)
Figure 1 Typical cracks in the segment of shield tunnel
Steel Corrosion
The steel corrosion relates closely to the protection thickness of concrete, the degree
of cracks development and the chloride content of groundwater. The corrosion of the
steels will result in the decrease of the lining bearing capacity. Furthermore, the
chloride in the ground water will also corrode the concrete, which leads to the
decrease of the concrete strength and the development of macro-cracks in the segment.
So, the steel corrosion have important effect on the stability of the lining structure,
which is not permitted to happen in practices. Regular inspections of the concrete of
the lining are usually employed to prevent the corrosion of steel.
Deformation of Linings
The deformation of lining includes the deformation of tunnel cross-section,
staggered dislocation of segment and the subsidence of side wall, among which the
deformation of tunnel cross-section contains the lateral and longitudinal deformation
and the former is the main component.
The staggered dislocation of the segment often occurs at the tunnel arch wires or the
construction joint, which includes convex staggered dislocation and concave staggered
dislocation. The reason for the staggered dislocation is that the tunnel arch wires or the
construction joint are the relatively weak places in the tunnel, the staggered dislocation
will occur when the tunnel is subjected to shear stress.
The subsidence of side wall is usually accompanied by the deformation of the
tunnel bottom, which will cause the uplift and sink of the tunnel bottom. However, the
deformation of tunnel cross-section and subsidence of side wall are very common in
highway tunnels, while staggered dislocation of the segment is usually found in shield
tunnel. Fig.2 shows the staggered dislocation in different positions of the segment in a
shield tunnel.
(a) haunch of the tunnel
ASCE
43
(b) arch bottom of the tunnel

Figure 2 staggered dislocation of the segment
Leakage of Groundwater
As the tunnel is usually constructed underground where the groundwater is rich, the
leakage of groundwater is the most common disease for all tunnels, which have
important effects on the stability of the tunnel, safe operation of the trains or cars and
the ground water environment of the tunnel.
For shield tunnel, the leakage of groundwater can be divided into water inflow,
water leakage, water drop and seepage of water according to the leakage of water rate,
and the location of groundwater leakage usually occurs at circumferential and
longitudinal joint, injection holes and conjunction joints between metro station hall
and the regional tunnel. Fig.3 shows the typical leakage of groundwater in different
parts of the shield tunnel.
(a) circumferential joint
(b) longitudinal joint
ASCE
44
(c) arch bottom
(d) injection hole
(e) screw hole
(f) conjunction
Figure 3 Leakage in different parts of a shield tunnel
ASCE
REASONS of SHIELD TUNNEL DISEASES and CORESPONDING

TREAMMENTS
As both the geological conditions around the shield tunnel and operation conditions
are complex, the reasons for shield tunnel diseases are also complicated. In this
section, the reasons for each kind of disease are discussed and the corresponding
treatments are suggested.
Firstly, the reasons for the cracking and spalling of concrete can be classified as
follows,
geological factors: the inhomogeneity of the strata along the longitudinal
direction of the tunnel;design factor: unreasonable lining thickness;environmental
factor: the cyclic loading resulted from the operation of the trainsconstruction
factorthe excavation sequence, the quality of the concrete, and so on.
Secondly, the main reason for the deformation of the linings is that the tunnel is
subjected to non-uniform loads during construction and operation.
Thirdly, the reasons for leakage of groundwater are classified into subjective and
objective reasons, respectively. The objective reason is that the groundwater is rich in
soft soil area where the shield tunnel is constructed, while the subjective reason is that
the waterproof treatments of the lining and the drainage facilities of the tunnel are
poor.
Based on the reasons for the main diseases of the shield tunnel, the following
treatments are proposed: for the cracking of the concrete, carbon fiber cloth is the most
common used material to repair the linings; for the leakage of the groundwater, it is
important to improve the drainage facilities in the tunnel, furthermore, the chosen
waterproof materials should be able to resist the shear stress and corrosion of the
water.
CONCLUSION
In this paper, the main diseases of the shield tunnel, including cracking and spalling
of concrete, steel corrosion, deformation of lining and leakage of ground water, have
been introduced detailedly, of which the leakage of groundwater is the most important
disease impacting the safe operation of the tunnel. Then the reasons for these diseases
and the corresponding treatments are discussed.
REFERENCES
Zhao Guoqi. On the remedial measures to crevice disease in concrete lining of railway tunnels[J].
Chinese Journal of Rock Mechanics and Engineering, 1996, 15(4): 385-389.
Liu Tingjin, Zhu Hehua, Xia Caichu, et al. Analysis of site investigation of cracking and leakage
on arcade tunnel lining of Yunnan province[J]. China Journal of Highway and Transport,
2004, 17(2): 64-67.
Luo Xin, Xia Caichu. Current situation and problems of classification of tunnel diseases[J].
Chinese Journal of Underground Space and Engineering, 2006, 2(5): 877-880.
Liu Haijing, Xia Caichu, Zhu Hehua, Luo Xin. Studies on tunnel damage[J]. Chinese Journal of
Underground Space and Engineering, 2007, 3(5): 947-953..
Wu Jiangbin, Zhang Dingli, Wang Mengshu. Current damage situation of railway operation
tunnels and their inspection and evaluation[J]. China Safety Science Journal, 2003, 13(6): 4952.
Zhu Changchun. Reparation technology of road tunnel lining crack[J]. Geotechnical World, 2004,
7(9): 67-68.
ASCE
45
46
Preparation and Failure Modes of Simulated Transversely Isotropic

Rock
X. M. Zhang1; X. F. Ou2; D. Yang3; J. S. Yang4; and J. Y. Fu5
1
Associate Professor, School of Civil Engineering, Central South Univ., Changsha 410075, Hunan, China.
E-mail: zhangxm@csu.edu.cn
2
Doctor, School of Civil Engineering, Central South Univ., Changsha 410075, Hunan, China. E-mail:
253644082@qq.com
3
Assistant Engineer, Tianjin Railway Survey Design Institute Group Co., Ltd., Tianjin 300000, China.
E-mail: 489574275@qq.com
4
Professor, School of Civil Engineering, Central South Univ., Changsha 410075, China. E-mail:
jsyang@edu.csu.cn
5
Postdoctoral Researcher, School of Civil Engineering, Central South Univ., Changsha 410075, China.
E-mail: jyfu2010@163.com
Abstract: This paper describes a method for modelling the simulated transversely
isotropic rocks. Series of physical and mechanical experiments have been used to obtain
two materials A and B, which are chose to simulate rocks. Then the reasonableness of
two materials are verified by comparing their properties to typical natural rocks. Based
on this, a set of experimental apparatus has been developed for preparing the simulated
transversely isotropic rocks. Uniaxial compressive tests are carried out to study the
failure mode of the simulated rocks preliminarily. The results can provide a reference for
researches in study of transversely isotropic rocks.
Keywords: Transversely isotropic rock; Simulated material; Failure mode; Uniaxial
compressive test.
INTRODUCTION
Bedded rock is a typical rock with complex physical and mechanical properties.
Usually it has a characteristic of orthotropic or transversely isotropic properties.
Composition, physical and mechanical properties of the rock which parallel to
discontinuities are basically the same, but their properties are completely different in the
perpendicular orientations [1]. Anisotropic, as one of the important properties of rocks
which have a great effect on its engineering performance. Especially in large-scale
geotechnical engineering, ignoring the anisotropic properties may bring about
incalculable accident.
ASCE
It is commonly to take samples in field for experiments of anisotropy rocks in

previous studies, but there are many restrictions, such as equipment and field
environment, which made the preparation very difficult. Therefore, artificially
simulating transversely isotropic rocks becomes an inevitable trend. Tien and Tsao [2]
get some achievements in this area, they put forward a preparation method, and study
the physical and mechanical properties of simulated samples made by the method, his
work have some reference value in this paper.
Based on the relevant research, this paper carried out the following research: 1)
Obtain two raw materials A and B which can simulate rocks. 2) Preliminary obtaining
the failure mode at different orientation angle of the samples by using uniaxial
compression experiments.
SELECTION OF LAYERED ROCK MATERIAL
Selection principle
For artificially prepared layered rock, the raw material needs to have similar physical
and mechanical properties with typical natural layered rocks. So the following
requirements should be satisfied during the selection of materials:1) Composition
requirement; 2characteristics of natural rocks can be satisfied; 3) Physical and
mechanical properties is stable and can be easily controlled; 4) Preparation process is
simple; 5) Material has wide variety of sources and low cost.
Selection of source materials
Artificial rocks are usually prepared with aggregate and cementing materials .There
are commonly used aggregates like sand, clay, kaolin, expansive soil, iron powder,
aluminum powder, diatomite and cementing materials like cement, gypsum, paraffin,
polymer, and synthetic resin etc. In previous studies, Tien and Kuo [3] selected Ottawa
sand, silica fume and kaolin as aggregate while they select cement as cementing material
to simulate sandstone - shale rock.
PREPARATION OF TRANSVERSELY ISOTROPIC ROCKS
Selection of materials mix ratio
In order to simulate the significant anisotropic properties of the rocks, we first need to
prepare two single materials with significantly different mechanical properties. This
paper takes the suggestions by Tien and Kuo [3] to determine the material mix ratio. By
comparing the plasticity, bleeding, color discrimination and strength of the materials,
finally two kinds of mixture with different strength were determined. They were named
material A and material B, respectively. Their mix ratios are shown in Table 1.
ASCE
47
48
Table 1. Mix of material A and B

Material A
Material B
Water : Soil
3:1
1:1
Water : Cement
1:3
3:5
Material A has a relatively higher water to soil ratio which forms relatively higher
strength and stiffness. Material B has a relatively lower water to soil ratio so that the
simulated rocks have relatively lower strength and stiffness. This experiment use 20%
amount of cement and 40% amount of kaolin and consequently to calculate the amount
of water.
Physical and mechanical properties of materials A and B
For artificially preparing transversely isotropic rocks, physical and mechanical
parameters of the simulated materials should be determined at first. Physical properties
of material A and B obtained by experiments are shown in Table 2.
Table 2. Physical parameters of material A and B
Porosity (%)
Dry density (KN/m3)
material A
20.3
19
material B
18.6
28
Material
clay
shale
sandstone
limestone
gneiss
marble
andesite
Table 3. General rock physical parameters

Density
Density
Porosity (%) Material
Porosity (%)
(KN/m3)
(KN/m3)
15~22
10.1~62.9
basalt
26~33
>18.7
21~28
1.5~44.8
granite
23~27
0.4~4.1
18~28
2.0~18.4
gabbro
28.5~30.5
0.4~1.9
15~28
0.7~10
diabase
29~32
0.5~5.1
26~32
0.4~7.5
diorite
28.1
0.4~4.0
26~29
0.1~2.1
syenite
2.62
0.9~2.9
25~28
>6.0
-
Table 3 shows the typical range of physical parameters of natural rock [4]. It is clear
that material A and B are both in this range. After curing in the mold, material A appears
in dark gray while material B appears in white. The two materials are in contrast
significantly, which can truly reflect the characteristics of layered rocks.
Uniaxial compression test and indirect tensile test were conducted for material A and
B. By pasting strain slices on the middle surface of the samples, two kinds of
stress-strain curve and load-displacement curve of the material A and B were measured,
as shown in Figure 1 and Figure 2. In Figure 1, the solid line and the dash line represent
the results of two samples, respectively. Mechanical parameters of material A and B can
be obtained by calculation as shown in Table 4.
ASCE
49
Table 4. Material A and Bs meechanical paarameters

Uniaxial
Mean
M
nsile Meaan
Mean
Ten
compressive
E
value
value strength valu
ue
streength
(MPa)
(MPa)
(M
MPa)
(M
MPa) (MP a)
(M
MPa)
Mateerial
61.3
1.88
13124
60.9
6
1.944
12789
A
60
0.5
2
12454
0.4
1.35
11124
40
Mateerial
41.6
4
1.344
10940
B
42
2.8
1.33
10756
v
00.21
00.25
00.29
00.23
Meaan
valu
ue
0.233
0.255
FIG.1. stress-strain
s
n curve
According
A
to
o Deere and
d Miller's [5
5] theory, nnatural rockss are divideed into threee
categ
gories depen
nd on the sttrength ratio
o T. T is thhe ratio of tthe elastic m
modulus and
uniax
xial compresssive strengtth, given as
T=
When
W
T>500
0, the rock is
i high stren
ngth ratio roock; When 200<T<5000, the rock iis
mediium strength
h ratio rock; When T<20
00, the rock iis low strenggth ratio rock. According
to the experimen
nt result, matterial A and Bs strengthh ratio are 210 and 263, respectivelyy.
i
thaat material A belongs too medium sstrength ratio
Deerre and Milleers theory indicates
rock while Materrial B belong
gs to medium
m strength raatio rock.
ASCE
50
(a) materrial A
(b) materrial A
(c) materrial B
(d) materrial B
FIG.2. Loa
ad-displacem
ment curve
Taable 5 showss the method
d of rating in
ntact rock material accorrding to the rrock uniaxiaal
comp
pressive streength propossed by Ramaamurthy andd Arora [6]. Test results in this papeer
show
ws that material A belon
ngs to mediu
um strength rock, and m
material B beelongs to thhe
geneeral strength rock. The tw
wo materials have distincctly differentt strength.
Rating
R
A
B
C
D
E
F
ASCE
Ta
able 5. Strength ratingg of intact roock
Stren
ngth
Uniaxiaal compressiive strength
h (MPa)
Descriiption
Extremeely high
>2550
stren
ngth
High sttrength
100~
~250
Medium strength
50~1100
General strength
s
25~
~50
Low strrength
5~225
Extremeely low
<55
stren
ngth
51
Prep
paration pro
ocess of layeered rock
(a)
( Front vieew
(b) Top view

w
FIG
G.3. Cast iro
on mold sch
hematic diaggram
FIG..4. Wooden schematic ttemplate diaagram

Th
his experim
ment combin
nes cast irron molds and woodeen templatee to preparre
transsversely isotrropic rocks. The geomettry of the caast iron moldd is 150 150 550mm
m
shaped in a cubo
oid. There iss opening at the upper annd the four ssides are fassten by screw
w
bolt so that it is easily
e
to asssemble. Figu
ure 3 shows tthe mold in a schematicc way. On thhe
w
temp
plate is a recctangular sloot with a sizee of 15010550 mm, aas
otherr hand, the wooden
show
wn in Figure 4.
Th
he process of
o modelling transversely
y isotropic roocks are as ffollows:
1
Mold prep
paration: Cleean the inneer of the moold and spreaad oil evenlly to ensure
the separation off test block and
a the mold
d.
2
Material preparation:
p
According to
t the mixinng ratio deteermined in thhe previous
section, the amo
ount of kaoliin and cemeent can be caalculated. T
Then the matterials were
nd stirred theem uniformlyy. The mixer machine w
was kept for
put into a concreete mixer an
running for 2~3 minutes wh
hile quantitattive water w
was adding sllowly. Sincee kaolin has
p
of water
w
swellin
ng, there wo
ould be a veery obvious increase in tthe mixture
the property
volum
me of its con
ntent ratio iss higher. It is suggested that the totaal amount off material B
shou
uld be reduceed about 15%
%.
3
A layer off handi-wrap
p should be spread insidde wooden ttemplate. Thhe template
placeed on vibratting table after
a
filling proper amoount materiaal A into the template.
ASCE
52
Material should be paved evenly to fully fill the template while the shaking table is
working.
4A single layer of material A stands for about 2 minutes in the template, then the
template should reverse to put the material into the cast iron mold and remove the
handi-wrap to complete the fitting of a single layer material. Then the same method
should be used to add material B. The two steps are repeat in turn until the iron mold is
filled.
5The filled mold should put on vibration table vibrating for 1~2 minutes to
remove holes or air bubbles caused by improper operation.
6A loading of 100KN was put on the specimen surface after the preparation is
completed for 2 hours. The mold then was removed after standing for 24 hours. The
samples should be cured indoor with wet cloth for about 28 days.
The entire process should be completed within 2 hours. Otherwise, the cement would
become hard to pave evenly. Moreover, the layer of the material may be easily broken
which would affect the results of experiments.
Samples are then prepared by drilling along different orientation angle by a drill bit
with a diameter of 50mm. The orientation angle is to be defined the angle between the
discontinuities and the horizontal direction. Then the samples were cut and polished to
make its surface smooth and flat. Parallelism of the top and bottom sides should be
controlled within 0.5mm, and the flatness of Surface should be controlled within
0.1mm. The flatness is directly related to the accuracy of measuring the Poisson's ratio;
hence it should be more stringent.
UNIAXIAL COMPRESSION TEST
Uniaxial compression test scheme
In order to use uniaxial compression test to study the effects of samples with different
orientation angles, five groups of samples with the different orientation angle (0, 30,
45, 60, and 90) were tested.
P
P
y

y

x

y

x
P
z
x
(a ) =
0
ASCE
(b ) =
9 0
(a )
0 a n d 9 0
53
FIG.5 The
T uniaxia
al compress ion test exp
periment
Failu
ure mode off rock materrial under uniaxial
u
com
mpression sttress state
Taable 6 show
ws the failurre modes of
o simulatedd transverselly isotropic rocks undeer
uniax
xial compreession with different orrientation anngles and tthe consequuential macro
crack
ks sketch.
Table 6. Failure mo
odes under u
uniaxial com
mpression
Speecimen failu
ure modes
Maacro cracks sketch

= 0
= 30
= 45
ASCE
54
Speecimen failu
ure modes
Maacro cracks sketch

= 60
= 90
Th
he failure mode
m
of the specimen
s
wh
hich is show
wing in tablee 6 is quite complicatedd.
The rock samplees result in tension faillure when oorientation aangle of 0 aand 30, and
crack
ks developed
d perpendicu
ular to the diiscontinuitiess. The sampples result in shear sliding
along
g the discon
ntinuities form
ming a main
n shear failuure crack whhen the orienntation anglees
are 45
4 and 60, and test resu
ults show th
hat the failurre angle and orientation angle are thhe
samee. It is clearrly observed
d the crack developed
d
aalong disconntinuities durring the testt.
How
wever, due to
o the influen
nce of the heeterogeneity of rock matterial, a smaall amount oof
shearr or tensile failure
f
cracks were also observed.
o
W
When orientattion angle iss 90 the rockk
samp
ples result in
n splitting faiilure verticallly along thee discontinuiities.
Th
he results off uniaxial co
ompression experiment
e
sshow that thhe failure moode is almosst
the same with the results of Tien
T and Kuo [7], so thee experimentts are acceptaable.
CON
NCLUSION
NS
nsversely isootropic rockk and theoryy on similaar
Based on thee characteristic of tran
mateerials, new simulated
s
maaterials whicch consist oof cement, kaaolin at diffferent mixingg
ratio have been modelled
m
in this paper. Then with thhe given meethod layered rocks werre
artifiicially prepaared. The folllowing concclusions can be drawn:
Th
hrough the experiments
e
on materialls A and B, their basic pphysical andd mechanicaal
properties have been
b
obtaineed. The resullts show thatt properties of the prepaared materialls
c
to natu
ural rocks. Moderate-int
M
tensity or loow intensityy rocks can be simulated
are close
with materials A or B.
Details
D
of thee preparatio
on process have
h
been ggiven in thiss paper, inccluding mold
desig
gn and tech
hnological process.
p
Thee standard experiment samples m
made by thiis
meth
hod have obv
viously layeered texture, and the orieentation of ttexture can bbe controlled
accurately. This method can
n provide convenient annd reliable m
materials for studying thhe
hanical perfo
ormance of transversely
t
mech
isotropic roccks.
Th
he failure modes
m
of tran
nsversely iso
otropic rockss have been preliminaryy summarized
based
d on the uniaxial
u
com
mpression results
r
on tthe simulatted sampless. When thhe
disco
ontinuities orientation
o
an
ngle is with
hin the rangee 0~30, thhe rock mateerial result inn
ASCE
tension failure perpendicular the layer discontinuity surface. When the orientation angle
is within the range 30~60, the rock material results in shear failure sliding along the
discontinuities. When orientation angle is within the range 60~90, the rock material
showed inflation failure by compression or shear failure by pull with the crack
orientation parallel to the discontinuities.
ACKNOWLEDGMENTS
This paper is supported by National Natural Science Foundation of China (No.
51378505).
REFERERCES
Deere, D.U., Miller R.P. (1966). Engineering classification and index properties of
intact rock. Air Force Laboratory Technical Report No. AFNL-TR-65-116,
Albuquerque, NM.
Goodman, R.E. (1989). Introduction to rock mechanics. New York: Wiley.
Goodman, R.E. (1993). Engineering geology-rock in engineering construction. New
York: John Wiley and Sons, Inc., P. 293-332.
Ramamurthy, T. and Arora, V.K. (1993). A classification for intact and jointed rocks.
Geotechnical Engineering of Hard Soils-Soft Rocks [C], 235-242.
Tien, Y.M., Tsao, P.F. (2000). Preparation and mechanical properties of artificial
transversely isotropic rock. Int. J. Rock Mech. Min. Sci. 37:1001-12.
Tien, Y.M., Kuo, M.C. (2001). A failure criterion for transversely isotropic rocks. Int.
J. Rock Mech. & Min. Sci., 38:399-412.
Tien, Y.M., Kuo, M.C. (2006). An experimental investigation of the failure mechanism
of simulated transversely isotropic rocks. Int. J. Rock Mech. &Min. Sci., 43:
1163-1181.
ASCE
55
56
Sensitivity Research on a Construction Space for a Multi-Partition Excavation

Tunnel
W. Wang1; J. S. Bian2; and W. S. Pan3
1
Associate Professor, School of Civil Engineering, Central South Univ., Hunan 410075, China. E-mail:
wangweicsu@csu.edu.cn
2
Study at School of Civil Engineering, Central South Univ., Hunan 410075, China. E-mail:
986444323@qq.com
3
Study at School of Civil Engineering, Central South Univ., Hunan 410075, China. E-mail:
583371698@qq.com
Abstract: In order to improve the mechanical and deformation performance of

surrounding rock and supporting structure in the construction of multi-partition
excavation tunnel, and improve construction safety. Based on the tunnel construction
mechanical properties and principle of orthogonal test design, multiple conditions
three-dimensional numerical simulation study on construction space is carried out by
finite element software for Cross Diaphragm with Step method, CRD method and CD
method applied to partial excavation tunnel commonly. Comparatively analyze
response characteristics in aspect of crown settlement, surface displacement and
supporting structure stress in different conditions, reveal influence rule about
deformation and stress state of multi-partition excavation tunnel in different
construction space. In the meantime, consider the maneuverability of the actual
construction, then put forward reference comments to select reasonable construction
space for multi-partition excavation tunnel, and the final purpose is to guide the
construction.
INTRODUCTION
With the development of large-span tunnel construction, partial excavation method
is widely used to reduce each excavation area, such as CD method, CRD method,
Cross Diaphragm with StepCDSmethod and Both Side Heading method etc.
Currently, the study of partial excavation tunnel more focused on the construction
method comparison (ZHOU 2013. WANG et al 2009)and less consider the impact of
construction space on the safety. In actual construction, the construction space
selection is based on engineering experience, and lack of theoretical support.
According to Guide for Construction of Railway Tunnel EngineeringTZ204-2008,
article 5.5.4 points out that distance between the both sides of CD method should keep
10-20m and article 5.6.4 points out that the distance between the left and right
excavation face of the CRD method should not be greater than 15m. In addition,
ASCE
57
article 5.5.4 points out that the longitudinal spacing between the left and right
excavation face of the CRD or CD method should not be less than 15m in Technical
Specifications for Construction of Highway Tunnel (JTG F60-2009). Obviously,
construction space for multi-partition excavation tunnel is different because of
different research objects.
In order to explore the sensitivity of construction space for urban underground road
tunnel adopting partial excavation method under the weak surrounding rock conditions,
taking an urban underground road tunnel as the research object in Hangzhou, China,
reveal influence rule about surrounding rock deformation and supporting structure
stress in different construction space, and obtain the sensitivity of construction space
to different factors through the research on construction space for the CD method,
CRD method and CDS method. Then, put forward reference comments to select
reasonable construction space for multi-partition excavation tunnel, and the final
purpose is to guide the construction.
NUMERICAL COMPUTATION MODEL
Calculating Parameters
Table 1 Calculation Parameters in Different Stratum
Note:1-- Earth Fill,2-- Silty Clay,3-- Silty Clay with Gravel,4-- Silty Clay mixed
Gravel,5-- Strong weathering Grey Siltstone,6-- Middle weathering Grey Siltstone,7-Support Structure,8-- Advanced Support Structure.
Density
Material
kg/m3
1
2
3
4
3
5
6
7
8
1750
1940
1990
1980
1990
1970
2320
2300
2200
Elastic
Modulus
GPa
0.1
0.15
0.16
0.2
0.16
0.2
0.2
33.5
0.44
Poisson
Ratio
0.3
0.31
0.31
0.29
0.31
0.25
0.23
0.2
0.29
Internal
Cohesion Thickness
Friction
KPa
m
Angle
12
10
3.75
16
35
3
17.3
50
2
17
45
20
17.3
50
2.75
35
30
1.5
36
200
17
/
/
0.3
/
/
2
Numerical calculation parameters refer to the geological condition of a partial

excavation tunnel in Hangzhou, China, shown in Table 1. Rock and soil model adopts
the Mohr-Coulomb constitutive model, and support structure adopts linear elastic
material, and pipe roof reinforcement area is treated by improving formation
parameters. Besides, elastic modulus of steel arch converts into concrete based on
section equivalent principle and steel arch spacing, the formula is as follows.
ASCE
58
E=E c +
Sg Lg Eg
Sc L
wheere, E--elasttic moduluss of supporrting structuure after coonversion, E

Ec-- elasticc
mod
dulus of spraayed concrete, Eg-- elasstic moduluss of steel arcch, Sc--sectional area off
spraayed concrette, Sg--sectional area off steel arch, Lg--flange llength of steeel arch, L--steeel arch spacin
ng.
Num
merical Com
mputation Model
M
Taking
T
the teest section S0+00~S0+4
S
45m as protootype in the engineeringg east line, a
3D numerical model
m
is buillt by finite ellement softw
ware ABAQU
US (CHEN 2010)andd
its horizontal
h
leength is 90m
m, the verticaal length is 50m, and thhe longitudinnal length iss
45m
m. Besides, the
t thicknesss of the supp
porting struccture is 0.3m
m. Solid elem
ment C3D8R
R
is used
u
to sim
mulate suppo
orting structture and surrrounding roock, and thhe model iss
diviided into 83
3745 grids. In addition,, the same m
model is ussed to simullate the CD
D
metthod, CRD method
m
and CDS
C
method
d, shown in F
Fig.1.
Fiig.1 Numeriical Computtation Modeel

Note:
N
Letterss are not representative oof constructiion sequencee
CO
ONDITION DESIGN
D
BA
ASED ON ORTHOGO
O
ONAL TEST
T
Bassic Principlee of the 0rth
hogonal Test
Orthogonal
O
test
t is a prim
mary efficien
nt method inn multiple faactors experim
ment, and itt
is able
a
to selecct some reprresentative points
p
whosee characterisstics are "unniform" andd
"ord
der" from co
omprehensiv
ve test. Basic process oof the orthoogonal experriment is ass
follows: ascertaain test facttors and levelschoosee proper orthhogonal tabblelist testt
n and test ressultsanaly
yze test resullts.
plan
ASCE
59
Working Condition of Numerical Simulation

Five factors influence surrounding rock deformation and supporting structure stress
in sensitivity research on construction space for multi-partition excavation tunnel,
including the distance between a and d, a and b, b and c ,d and e, e and f. And, each
factor contains four levels, namely 2m, 4m, 6m and 8m. So, L1645orthogonal table
is selectedthen, get the working conditions of numerical simulation by substitution of
the factors and levels, shown in Table 2.
Table 2 Working Condition of Numerical Simulation
Conditions
Factors
a-dm
a-bm
b-cm
d-em
e-fm
1
2
2
2
2
2
2
2
4
4
4
4
3
2
6
6
6
6
4
2
8
8
8
8
5
4
2
4
6
8
6
4
4
2
8
6
7
4
6
8
2
4
8
4
8
6
4
2
9
6
2
6
8
4
10
6
4
8
6
2
11
6
6
2
4
8
12
6
8
4
2
6
13
8
2
8
4
6
14
8
4
6
2
8
15
8
6
4
8
2
16
8
8
2
6
4
ANALYSIS OF CALCULATION RESULT

Calculation Results
In order to improve the accuracy of results and eliminate the influence caused by
the difference of excavated volume, take value of crown settlement, surface
subsidence and maximum principal stress of supporting structure at initial section
S00+00m as output result when forming construction space and continuing digging
10m, G0, D0, Z0, G10, D10, Z10, respectively. Then, the results are calculated by formula
2 and shown in Table 3.
G=G10-G0
D=D10-D0
Z=Z10-Z0
(2)
Table 3 Numerical Simulation Results
Working
Condition
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
ASCE
GCRD
DCRD
ZCRD
GCD
DCD
ZCD
GCDS
DCDS
ZCDS
3.85
3.32
2.6
2.04
1.69
1.72
2.88
2.92
1.97
3.08
2.06
3.15
2.33
2.38
2.25
2.29
3.62
3.21
2.69
2.12
1.76
1.82
2.85
2.92
2.06
2.97
2.1
3.05
2.32
2.37
2.34
2.34
-5.49
-1.62
-0.64
0.02
-1.18
-1.14
-1.83
-3.13
-0.96
-2.71
-1.63
-0.84
-1.24
-2.5
-1.13
-1.6
6.53
6.39
5.92
5.54
4.5
4.42
5.79
5.68
4.8
6.22
4.9
6.22
5.45
5.51
4.85
5.01
5.43
5.64
5.27
4.83
4.21
4.11
5.16
5.11
4.29
5.46
4.6
5.61
4.9
5.04
4.35
4.48
-6.55
-4.95
-4.28
-3.74
-3.31
-2.8
-4.39
-5.76
-3.34
-3.33
-2.73
-4.75
-3.44
-3.9
-3.81
-4.8
8.4
8.01
7.54
7.14
6.32
6.07
7.02
7.02
6.01
7.36
6.6
7.92
6.59
6.77
6.06
6.39
6.29
6.48
6.15
5.89
5.26
4.98
5.85
5.69
4.84
6.07
5.45
6.32
5.33
5.7
4.91
5.27
-4.75
-3.79
-2.39
-2.19
-1.25
-1.34
-0.23
-0.34
-1.61
-0.87
-1.45
-3.69
-0.62
-1.64
-2.09
-2
60
Result Analysis
In order to get sensitivity regularity of construction space for multi-partition
excavation tunnel, define function indicating influence degree of different influence
factors on crown settlement, surface subsidence and maximum principal stress of
supporting structure, calculation results are shown in Table 4.
G =
G10 G0
100%
G0
D =
D10 D0
100%
D0
Z =
Z10 Z 0
100%
Z0
(3)
Table 4 of Different Working Conditions (%)

Working Condition
GCRD
DCRD
ZCRD
GCD
DCD
ZCD
GCDS
DCDS
ZCDS
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
45.36
25.83
15.98
10.85
10.70
10.91
19.28
19.41
12.87
23.03
13.40
22.53
15.71
16.12
15.37
15.69
90.04
42.97
26.22
16.79
17.66
18.18
30.14
30.75
21.51
36.34
21.39
35.11
25.16
25.35
25.46
25.28
58.14
15.13
5.59
0.17
8.80
8.23
19.53
33.43
8.32
33.71
13.45
8.80
12.62
30.57
10.72
14.48
63.19
38.27
25.95
19.23
21.23
21.88
27.39
28.26
23.99
35.49
23.92
31.77
26.79
26.55
26.36
27.16
100.47
53.99
33.68
23.14
29.19
30.20
35.50
37.54
31.79
46.94
32.79
42.81
35.15
35.20
35.54
36.59
70.74
67.97
66.54
62.23
28.34
20.53
89.87
134.99
35.66
48.17
20.81
67.66
41.07
41.03
43.26
45.71
106.57
63.14
42.87
31.91
27.53
31.24
39.20
47.53
27.90
44.54
34.76
52.88
28.94
31.40
35.83
40.20
126.54
68.70
44.47
32.72
30.70
33.59
44.36
50.08
30.74
50.49
36.75
55.59
32.15
35.42
38.62
43.02
48.02
49.41
31.17
32.37
11.67
11.19
2.38
3.70
17.02
9.67
12.41
51.89
6.26
16.73
22.33
18.21
Range analysis and variance analysis are always used to analyze orthogonal test
results, and range analysis method is adopted in this paper because of its high
visualization. Determine the effect degree of different factors on the test index by the
value of range, and the greater the value, the more important the factor. First, calculate
the average value of under different factors in different levels, shown in Fig.2.
ASCE
61
CD method
CRD method
60
Cross Diaphragm with Step method
50
40
30
20
8
)
d
a
(
6
)
d
a
(
4
)
d
a
(
2
)
d
a
(
10
(a-b)2 (a-b)4 (a-b)6 (a-b)8 (b-c)2 (b-c)4 (b-c)6 (b-c)8 (d-e)2 (d-e)4 (d-e)6 (d-e)8 (e-f)2 (e-f)4 (e-f)6 (e-f)8
Influence factors and level
a) Average value of G
70
CRD method
CD method
60
50
40
30
20
(a-d)2 (a-d)4 (a-d)6 (a-d)8 (a-b)2 (a-b)4 (a-b)6 (a-b)8 (b-c)2 (b-c)4 (b-c)6 (b-c)8 (d-e)2 (d-e)4 (d-e)6 (d-e)8 (e-f)2 (e-f)4 (e-f)6 (e-f)8
Influence factors and levels
b) Average value of D
80
CRD method
CD method
70
60
50
40
30
20
10
(a-d)2 (a-d)4 (a-d)6 (a-d)8 (a-b)2 (a-b)4 (a-b)6 (a-b)8 (b-c)2 (b-c)4 (b-c)6 (b-c)8 (d-e)2 (d-e)4 (d-e)6 (d-e)8 (e-f)2 (e-f)4 (e-f)6 (e-f)8
Influence factors and levels
c) Average value of Z
Fig.2 Average Value of
According to the above Fig.2, the value of range is calculated by formula 4 and
shown in Table 5.
R=max{2,4,6,8}min{2,4,6,8}
4
ASCE
62
Table 5 Range Value

Influence
factors
Vault Settlement
CR
D
a-d
9.4
a-b
5.2
b-c
5.2
d-e
13.3
e-f
13.0
CD
12.
0
7.9
7.8
5
14.
4
15.
6
Surface
Subsidence
Maximum Principal
Stress
CD
S
CRD
CD
CDS
CRD
CD
CDS
27.0
19.8
19.7
30.8
3.7
25.7
33.0
9.6
12.8
14.7
14.0
9.7
33.7
9.5
17.0
12.7
15.4
20.1
12.7
30.1
21.2
25.8
24.6
23.3
31.6
22.4
26.9
12.1
27.2
25.3
25
32.5
25.2
36.2
6.8
Figure 2 and Table 5 show that:

(1) Influence law of construction space to vault and surface subsidence is basically
uniform for partial excavation tunnel. Influence degree is as follows: CDS method
CD methodCRD method. Taking vault settlement of CDS method as an example,
different factors range value is as follows: the distance between e and
f 27.2a and
d27.0d and e25.8b and c17.0a and b9.6, which indicates that
the vault settlement is more sensitive to the distance between e and f, a and d, d and e.
(2) Taking the maximum principal stress of supporting structure as analysis project,
different factors range value of CRD method is as follows: the distance between e and
f25.2d and e22.4b and c12.7a and b9.7a and d3.7), and
the CD method is close, between 25 and 35, and the CDS method is as follows: the
distance between a and d33.0b and c21.16d and e12.07a and b9.47
e and 6.83
f
,which shows that the CD method is sensitive to every influence factor
comparing with the CRD method and CDS method.
3 The smaller the construction space, the greater the vault and surface
subsidence. When the distance between a and d is 4m, and the distance between a and
b, b and c is 6m, relatively, the vault and surface subsidence are smaller. When the
range of the distance between a and d, a and b is 4m to 6m, relatively, the maximum
principal stress of supporting structure is smaller.
CONCLUSIONS
(1) Influence law of construction space to vault and surface subsidence is basically
uniform for multi-partition excavation tunnel, and the smaller the construction space,
the greater the vault and surface subsidence. The vault and surface subsidence is more
sensitive to the distance between a and d, d and e, e and f, so pay more attention on
these factors under construction. But for different construction methods, the influence
degree of construction space is as follows: CDS methodCD methodCRD method.
ASCE
(2) Taking the maximum principal stress of supporting structure as the analysis
project, the sensitive of construction space is different due to the different construction
method. The CD method is sensitive to every influence factor, and the CRD method is
more sensitive to the distance between e and f, d and e, and the CDS method is more
sensitive to the distance between a and d , b and c.
ACKNOWLEDGMENTS
The authors appreciate the support of China Railway Tunnel Group Science and
Technology Innovation Project.
REFERENCES
CCCC First Highway Engineering Co., Ltd (2009). Technical Specifications for
Construction of Highway Tunnel (JTG F60-2009). Communications Press, Beijing,
China.
CHEN, W.Z., WU, G.J., JIA, S.P. (2010). The application of ABAQUS in tunneling
and underground engineering. Water Power Press, Beijing, China.
China Railway First Group Co., Ltd (2009). Guide for Construction of Railway
Tunnel EngineeringTZ204-2008. China Railway Publishing House, Beijing,
China.
WANG, G.X., XIAO, L.G., WANG, Y. L. (2009). Discussion on Excavation Methods
of Shallow-covered Extra-large Cross-section Urban Tunnels. Tunnel Construction,
29 (6):658-663.
ZHOU, D.J. (2013). Research on Shallow large-span Railway Tunnel Excavation
Methods under Urban Complex Environmental Conditions. Chongqing Jiaotong
University. Chongqing, China.
ASCE
63
Numerical Research on the Waterproof Failure Mechanism of the Sealing Gasket of

an Underwater Shield Tunnel
J. C. Wang1; L. W. Sun1; X. J. Lv2; Z. Wang3; S. B. Hu4; and C. F. Chao5,6
1
Zhejiang Univ., Zhejiang, Hangzhou 310058.
Zhejiang College of construction, Zhejiang, Hangzhou 311231.
CCCC Fourth Harbor Engineering Institute Co. Ltd Guangdong Guangzhou 510288.
Huadong Engineering Corporation Limited, Zhejiang, Hangzhou 310014.
Zhejiang Scientific Research Institute of Transport, Zhejiang, Hangzhou 310006.
Sichuan Province Univ. Key Laboratory of Bridge Non-destruction Detecting and Engineering
Computing.
Abstract: The research on the waterproof failure mechanism was conducted by the
software ABAQUS based on the project of waterproof design. The method used in the
paper was compared with the predecessor method. The results showed that not only the
predecessor method was workable but also the method in the paper was reasonable. The
waterproof failure mechanism was also revealed. The research showed that the leakage
was easier occurred on the contact surface of sealing gaskets under the condition of joint
opening. However, the leakage was easier occurred on the contact surface of sealing gasket
and concrete under the condition of joint dislocation.
INTRODUCTION
With the booming of social economy and the rapid development of construction
technology, shield tunnel has gradually become the main choice during construction of
cross-river. Since the cross-river tunnel bears higher water pressure than the subway tunnel,
waterproof design is of great importance. Failure of joint waterproof will cause water
leakage, which will also cause uneven settlement, endangering the safety of the tunnel
structure and operations. According to the field observation, shield tunnel leakage mainly
occurs in the joints (Zhu, 1995). Currently, sealing gasket is widely used in shield tunnel
waterproof design. Gaskets can be divided into elastic rubber gaskets and water swelling
rubber gaskets (Fig. 1).
ASCE
64
FiG.1 The diagram of elastic rubber gasket and water swelling rubber gasket
Research on the failure mechanism of elastic waterproof gaskets mainly focused on

experimental research and numerical analysis. Paul (1978) conducted waterproof failure
test on elastic rubber gaskets. The test results showed that most leakages occurred between
the elastic gasket and the trenches of steel gaskets caused by separation. It was also
revealed that there was a certain relationship between the contact pressure and the external
water pressure between the elastic gaskets when leakage occurred on the contact surfaces
of the elastic gaskets. Faisal Shalabi (2001) conducted waterproof failure test on elastic
rubber gaskets worked with the trenches of steel gaskets as well as concrete gaskets, the
protruding side on the longitudinal joint with certain angles were considered. Lu (2008)
conducted the water seal test on the T joint and circumferential joint based on the project of
Shanghai Changjiang tunnel, and the influence of different openings and dislocations of
joint were discussed. Deng (2008) studied the design requirement of tunnel segments based
on the project of Wuhan Changjiang tunnel. Liu (2011) insisted that uneven longitudinal
settlement would lead to the opening of longitudinal joint and weaken the waterproof
ability of sealing gaskets. The leakage would occurred on the longitudinal joints than the
circumferential joints. Tuo (2013), Zhao(2013) conducted the water seal test based on
Nanjiang Weisan tunnel, the optimized section of sealing gaskets were put forwarded.
Xiang et al. (2008) analyzed the deformation characteristic of elastic gaskets under
different compression by ANSYS and optimized the sealing section of joints. He (2009)
studied the relationship between the contact stress and compression force of gaskets by the
method of numerical analysis. Lei (2010) considered the factors of gasket deformations,
stress concentration, the stress redistribution under joint opening and put forward the
indexes of contact stress and compression force for the design of gaskets.
Numerical research on the failure mechanism of gaskets mainly focused on the stress
analysis and optimization of the gaskets section based on the symmetric characteristic of
the compressed gaskets on both sides of the joint. However, these researches cannot
analyze the contact stresses and deformation behavior of staggered gaskets on the situation
of gaskets dislocation. Therefore, based on a river-cross tunnel project, numerical research
on the waterproof failure mechanism was conducted using ABAQUS when facing elastic
rubber gaskets under deformation of opening and dislocation. The simulation results were
expected to show its value for elastic gaskets design.
ASCE
65
66
PROJECT BACKGROUND
The research was based on a cross-river tunnel project. The outer diameter and inner
diameter of the tunnel is separately 11.3 m and 10.3 m. The ring width is 2.0 m. C50
reinforced concrete segments are used and the design impermeability grade is S12. Thirty
six M30 oblique bolts are used to connect the tunnel rings. Between segments, two M36
oblique bolts are installed circumferentially. The deepest part of the tunnel is located in the
permeable stratum with a maximum pressure of 0.45 MPa. Dual waterproof measures
using EPDM rubber gaskets on the outside and water-swellable rubber gaskets adjacent to
it on the inside are employed
Considering the internal stress relaxation and the effects of aging of elastic gaskets
during the design life of the tunnel, the waterproof design value should be obtained from
the maximum pressure value multiplying a safety factor. According to Shield Tunnel
waterproofing Technical Specification, the actual design should withstand the pressure of
2 to 3 times the maximum water pressure. In this project, the safety factor is 2, and
therefore waterproof pressure is taken to be 0.9 MPa under maximum displacement of
opening and dislocation. The maximum opening displacement and the maximum
dislocation displacement were taken as 8 mm and 15 mm based on construction experience
and Shield Tunnel Construction and Acceptance Specification.
According to gaskets parameters and engineering experience, the cross-sectional
structure of the elastic gasket is shown in Fig.2. The hardness is taken to be 62 5 . A
compressed displacement of 20mm is required for the gasket to meet the design height.
FIG.2 The cross-sectional structure of the elastic gasket
NUMERICAL MODEL
The numerical model was completed according to section form of elastic gasket and
surrounding concrete segments, shown in Fig 3.
Boundary Conditions
The boundary conditions are described as followings.
For concrete segments C1: constraint is imposed in the x direction on the left edge and
in y direction on the upper and lower boundaries;
ASCE
67
For concrete segments C2: constraint is imposed in y direction on the upper and lower
boundaries; displacement or force load will be imposed on the right boundary according to
the calculating targets.
Elastic gasket
Segment C1
Segment C2
FIG.3 schematic diagram of modelling
Material Parameters
Elastic constitutive model was adopted to simulate C50 concrete. The elastic modulus is
taken as 3.45 104 MPa and the Poisson's ratio as 0.167. Rubber material is typically
treated as incompressible hyperelastic isotropic materials. Super-elastic material
non-linear relationship between stress and strain is defined by strain energy function. This
calculation uses Mooney-Rivilin two-parameter model (Wang, 2004) which is widely used
in engineering. The strain energy function expression:
U = C10 ( I1 - 3 ) + C01 ( I2 - 3 )
Where: U for the strain potential energy; I1, I2 for the strain invariants; C10, C01 for the material
parameters, obtained from the experiments. C10 was taken as 0.532 and C01 as 0.131.
RESULTS ANALYSIS
Comparison between two methods
Conventional numerical methods usually analyze one side of the elastic gasket
according to the structural symmetry. The surrounding concrete is considered as rigid body.
Rigid displacement on the upper side is applied to simulate the pressure of elastic gasket as
shown in Fig. 4. Completed modeling is employed in this work. Loads are imposed on
concrete segments C2 to simulate the elastic gasket compression. Fig. 5 is a comparison of
previous method and the proposed method in simulation of contact stress. It can be seen
that proposed method and the previous method had a good agreement, which not only to
verify the feasibility of the previous simplified method but also proved the reasonableness
of the proposed method. Fig. 5 (a) is the contact stress distribution of gasket, showing that
contact stress is lager in both ends due to the stress concentration. Fig. 5 (b) shows the
contact stress between the surrounding concrete and the gasket. It can be seen from the
figure that the contact stress distributed along the length of the contact surface
ASCE
68
symmetrically and the value of the contact force is basically lager than the waterproof
design pressure values. Calculation method of the average contact stress was proposed by
Zhao et al. (2013).
The average contact stress = assembly force / gasket contact surface width.
Over a comparison of chart (a) and chart (b), it can be seen that contact stress between the gaskets is less
than the stress between the concrete and the gasket.
Rigid
body
FIG.4 The conventional numerical method

6
proposed method
previous method
design water pressure
4.5
4.0
contact stress (MPa)
contact stress (MPa)
5.0
proposed method
previous method
4
3
2
1
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0.0
-0.5
12
16
20
24
28
32
length along contactsurface (mm)
10
20
30
(a) between gaskets
(a) between gasket and segment
FIG.5 The relationship between the contact stress and contact length
Hardness Analysis
80
assembly force (kN/m)
70
60
o
70
o
67
o
62
50
40
30
20
10
0
10
12
14
compression (mm)
16
18
20
FIG.6 Compressive stress-strain relationship under different gasket hardness
ASCE
40
length along contactsurface (mm)
69
According to the relationship between hardness and Young's modulus given by Wang et
al. (2004):
E = (15.75 + 2.15HA) / (100 - HA)
where: E for Young's modulus; HA for hardness. By changing the Young's modulus,
compression characteristics can be simulated under different hardness condition. Fig. 6 is a
numerical analysis result of compressive stress-strain relationship under different gasket
hardness. With a gasket hardness of 62, the assembly force is less than 60 kN/m to meet
the installing requirement. With the increase in hardness, the assembly force is also
increasing.
Waterproof Failure Mechanism
Opening condition
average contact stress (MPa)
3.5
stress between gaskets

stress between gasket and concrete
3.0
2.5
2.0
1.5
1.0
0.5
0.0
10
12
opening (mm)
14
16
18
20
FIG.7 Relationship between average contact stress and the opening displacement
Leakage at the
surface between
gaskets
FIG.8 Waterproof failure mechanism under opening condition
ASCE
70
Based on the average contact stress, relationship between average contact stress and the
opening displacement can be obtained. As can be seen from Fig. 7, in the case of gasket
opening, the average contact stress between gaskets is always less than the stress between
gasket and the concrete. When displacement reaches 9mm, the average contact stress
between gaskets reaches design pressure 0.9 MPa firstly. Therefore, in the case of gasket
opening, according to the waterproof failure criterion, which means the contact stress is
less than the design pressure, the waterproof failure mechanism is shown in Fig. 8. It can be
seen from Fig. 8, leakage will firstly occur between the elastic gaskets.
Dislocation condition
The relationship between average contact stress and the dislocation displacement is
shown in Fig. 9. The waterproof failure mechanism is more complicated than the
mechanism under opening condition. As it can be seen from Fig. 9, dislocation
displacement has little effects on the average contact stress between gaskets. But the
contact pressure between the gasket and the concrete increases with the displacement. For
this project, it can be found that the average contact stress between the gaskets is always
greater than the design water pressure 0.9MPa. When the dislocation displacement reaches
16mm, the average contact stress between the gasket and the concrete is reduced to 0.9MPa.
Therefore, according to the waterproof failure criterion, the failure mechanism of the
waterproof case is shown in Fig. 10. Leakage occurs between the gasket and the concrete.
average contact stress (MPa)
2.4
stress between gasket and concrete

stress between gaskets
2.1
1.8
1.5
1.2
0.9
0.6
0.3
0.0
12
15
18
compression (mm)
FIG.9 Relationship between average contact stress and the dislocation displacement
ASCE
71
Leakage at the
surface between
gasket and concrete
FIG. 10 Waterproof failure mechanism under dislocation condition
CONCLUSIONS
According to a waterproof design of an actual project, numerical simulation was applied
to reveal the failure mechanism of elastic gasket. The following conclusions are obtained:
(1) By comparison with the previous simplified studies, the proposed method is closer
to the real situation. The simulation method is reasonable and feasible. Distribution of
contact stress between gaskets and stress between trenches and gaskets are obtained, while
the previous one is usually smaller.
(2) Based on the conception of average contact stress, waterproof failure mechanism is
analyzed. The results show that leakages occur more on the contact surface of the gaskets
under the opening condition due to lower average contact stress. However, it is believed
that leakages occur more on the contact surface of the gasket and the concrete under the
dislocation condition since its average contact stress decreases with the dislocation
displacement increasing.
Although the study on the waterproof failure mechanism reaches some qualitative
conclusions, but the judgment is based on the concept of the average contact stress
proposed by Zhao. Therefore, it still requires future engineering and testing to be
corroborated.
ACKNOWLEDGEMENTS
This work was supported by the Opening Project of Sichuan University Key Laboratory of
Bridge Non-destruction Detecting and Engineering Computing and the Foundation
FX2014160 by Education Department of Zhejiang.
REFERENCES
Deng, C.H. (2008). The waterproof design of tunnel segment. Railway construction
technology, S: 157-163.
He, T.H., Wang, B.L., Liu, J.G. (2009). Rational Determination of Elastic Gasket Contact
Stress and Closure Pressure for Shield Tunnel Segments. Urban Mass Transit, 12(12):
60-64.
ASCE
He, T.H., Zhou, G.N., Liu, X.H. (2009). An optimal design research of rubber sealing
gasket in Hangzhou Metro Line 1. Underground Engineering and Tunnel, S: 41-45.
Lei, Z.Y. (2010). An Optimal Design Approach for the Rubber Sealing Gasket in Shield
Tunneling. Chinese Journal of Underground Space and Engineering, 6(4): 361-364.
Liu, Y., Zhang, D.M., Huang, H.W. (2011). Analysis of Mechanism of Water Leakage
Induced by Longitudinal Uneven Settlement of Shield Tunnel. Journal of Railway
Engineering Society, 5: 66-70.
Lu, M., Lei, Z.Y., Zhang, Y. (2008). Waterproofing Test of Lining joint and CrossPassage
of Shanghai Yangtze River Tunnel. Underground Engineering and Tunnel, 4: 12-16.
Paul, S.L. (1978). Sealability tests of gaskets between precast concrete tunnel lining
segments. Unpublished report, Univ. of Illinois at Urbana-Champaign.
Paul, S.L., Curtis, R.H., and Heuer (1978). Load testing precast concrete tunnel liner
segments and gasket sealability tests. Unpublished report, Univ. of Illinois at
Urbana-Champaign.
Tuo, Y.F., Shu, H., Guo, X.H. (2013). Design and experimental study on waterproof
gasket of large-diameter shield tunnel under ultra high water pressure. Chinese
Journal of Geotechnical Engineering, 35(S1): 227-231.
Code for construction and acceptance of shield tunneling method. Beijing, 2008
Shalabi, F. I. (2001). Behavior of gasketed segmental concrete tunnel lining. Ph. D.
Thesis, University of Illinois at Urbana-Champaign,.
Wang, W., Deng T., Zhao S.G. (2004). Determination for Material Constants of Rubber
Mooney-Rivlin Model. Special Purpose Rubber Products, 25(4): 8-10.
Xiang, K., Sh,i X.W. (2008). Design and Optimization of Elastic Gasket Section of Shield
Tunnel Lining. Chinese Journal of Underground Space and Engineering, 4(2):
361-364.
Zhao, M., Ding, W.Q., Peng, Y.C. (2013). Experimental Study on the Reliability of
Shield Tunnel Segment Joints to Remain Watertight Under High Water Pressure.
Modern Tunneling Technology, 50(3): 87-93.
Zhu, Z.X. (1995). Present situation and Prospect of urban tunnel waterproof technology.
Underground Engineering and Tunnel, 4: 18-24.
ASCE
72
Analysis of the Stability of Mud Inrush Induced by Fillings Sliding Failure in a Karst
Cave Based on the Simplified Bishop Method and Its Application
P. Lin1,*; S. C. Li2; Z. H. Xu3; L. P. Li4; X. Huang5; and S. J. He6
1
Ph.D. Student, Research Center of Geotechnical and Structural Engineering, Shandong Univ., Jinan,
Shandong 250061, China (corresponding author). E-mail: sddxytlp@163.com
2
Professor, Research Center of Geotechnical and Structural Engineering, Shandong Univ., Jinan,
Shandong 250061, China. E-mail: lishucai@sdu.edu.cn
3
Lecturer, Research Center of Geotechnical and Structural Engineering, Shandong Univ., Jinan,
Shandong 250061, China. E-mail: zhenhao_xu@sdu.edu.cn
4
Associate Professor, Research Center of Geotechnical and Structural Enginee ring, Shandong Univ.,
Jinan, Shandong 250061, China. E-mail: yuliyangfan@163.com
5
Shandong 250061, China. E-mail: hx19891018@yeah.net
6
Postgraduate Student, Research Center of Geotechnical and Structural Engineering, Shandong Univ.,
Jinan, Shandong 250061, China. E-mail: hsj686@126.com
Abstract: A simplified computational model used for the stability analysis of the
fillings is proposed based on the simplified Bishop method. In this model, sliding soil,
circular slip surface, exterior load, pore water pressure and boundary constraint is
defined by simplifying the complex system which consist of tunnel, filling materials
and karst cave. The global stability coefficient of fillings is got by solving
the equilibrium equation of force for individual soil slice and moment for failure mass
and by assuming that the direction of inter-slice forces is horizontal. By this way, the
risk of mud inrush due to the filled-typed karst caves can be identified. Using this
analytical method, this paper analyzes a case and conducts research on the influence of
hydraulic effect, shear strength parameters, mud content in cave and boundary
constraint on the global stability of fillings.
INTRODUCTION
This paper simplifies the complex system which consist of tunnel, filling materials
and karst cave into a computational model based on simplified Bishop method. By
solving this model, an innovative method is used to analyze the stability of mud inrush
induced by filled-typed karst cave. Using this method, conduct research on the
influence of hydraulic effect, shear strength parameters, mud content in cave and
boundary constraint on the stability of mud inrush.
ASCE
73
74
An Innovative Method for Stability Analysis of Mud Inrush

Computational Model
By the action of groundwater corrosion or seepage pressure, water infiltrate along
the interface of rock and soil. Then, shear strength and frictional resistance of the
interface is gradually reduced. So the contact interface is more easy to become critical
failure surface. The failure surface may appear inside soil, but the determination of the
critical slip surface will not be discussed in this paper.
These encountered karst caves but concealed caves are simplified to two types of
computational models: model 1, stability analysis without boundary restraint; model 2,
stability analysis with boundary restraint, as shown in Fig.1.
Boundary Restraint
Concealed Caves
Tunnel
Simplified Load
O
a
Cracked
Surface
Sliding
Soil
Sliding Surface
Model 2
Simplified Load
O
c
a
Sliding Soil
Sliding Surface
Model 1
FIG.1 Two different simplified computational models

There are six basic assumptions for above models:
a. The stability analysis is consider as plane-strain problem.
b. On sliding surface, shear failure obeys Mohr-Coulomb criterion and shear
strength is decided by cohesion and friction.
c. Fillings is assumed as an ideal rigid-plastic material, and there is no deformation
in fillings during loading process, and the cave itself is stable.
d. There are no forces on the contact interface because of the separation of fillings
and the upper boundary while fillings are sliding along the lower boundary.
e. The sliding surface is an arc, and forces on the surface satisfy the static force and
moment equilibrium, The direction of resultant of inter-slice forces is horizontal.
f. Only for model 2, there would generate a cracked surface within fillings result
from boundary restraint. Other assumptions: the cracked surface is also an arc; Shear
failure obeys Mohr-Coulomb criterion on the cracked surface too; The direction of
forces between soil and boundary restraint is horizontal.
Method and Principle
Firstly, for model 1, as shown in Fig.1, the arc ab(a boundary of cave) is selected as
the sliding surface. Through the top point a, draw a horizontal line ac. Fillings below
the line ac is regarded as sliding soil(soil abc), and above is simplified into exterior
load applied to the sliding soil. The sliding soil mass is divided equally into vertical
slices.
The forces acting on the whole sliding soil and the individual slice is shown in
Fig.2 (For simplifying the calculation, assume that the exterior load is uniform load
above each slice).
ASCE
75
R=15m
qi
h=9.6m
Free Face of Soil
Saturation line
qi
Wi
Fi+1
Fi
27
Ti
Ni
Ti
B=14m
(a)
Ni
(b)
FIG.2 Diagram indicating the forces acting on whole sliding soil and the
individual slice: (a) on whole sliding soil; (b) on the individual slice
Where Ni indicates normal force and Ti indicates tangential force, is the angle
between failure surface and horizontal plane, qi is the exterior load above a slice, Wi is
the weight of a slice,
indicates force between ith slice and i+1th slice.
For each single soil slice, static force equilibrium equation in the vertical direction
is established and for whole sliding mass, moment equilibrium equation of the circle
center O is established.
Secondly, for model 2, as also shown in Fig.1, the arc df is the cracked surface. Soil
abc is regarded as sliding mass.
Above the cracked surface, static force and moment equilibrium equation can be
established by forces analysis as is shown in Fig.3 (Where Fr0 indicates resultant on
constrained boundary). Here is the point: glyphs with a subscript ri are used in study
subject above the cracked surface, such as Nri, Tri, qri, Wri,
, Fri, implications are the
same as before.
qi
r=9.6m
qr
Fr0
Free Face
Fr0
Ti
(a)
Ni
Tri
Br=8m
hr=4.3m
Nri
(b)
FIG.3 Diagram indicating the forces acting on whole sliding soil and soil above
cracked surface: (a) on whole sliding soil; (b) on soil above cracked surface
Above the cracked surface, the static force equilibrium equation in the horizontal
and vertical direction and moment equilibrium equation of the circle center O are
established and Fr0 can be obtained.
ASCE
And above the sliding surface, static force equilibrium equation in the vertical
direction and moment equilibrium equation with Fr0 are established.
Safety Factor
The safety factor f reflects the stability of mud inrush induced by integral soil
=
sliding in karst cave. The f is defined by the following relation: = / ,
/ . Where c is cohesion and is angle of internal friction.The safety factor f is
got with iterative computation by putting ce and
e into front static force and
moment equilibrium equation.
Seepage Effect
Under seepage condition of karst water, Ni and Ti is defined by the following
expression: =
+(
)
. Where the indicates effective cohesion
and
indicates effective angle of internal friction, ui indicates pore water pressure on
sliding surface.
Examples Analysis and Discussion of Results
With or withou boundary restraint, the stabilities of mud inrush induced by failure
mass will be evaluated under 5 common conditions in real project using the above
methods.
Parameters under different conditions as shown in table 1.
Table 1 Parameters for Different Conditions
Condition no.
1
2
3/5
4
3
Natural weight(kN/m )
19.0
19.0
19.0
19.0
20.1
20.1
Saturated weightkN/m3
22.4
17.5
17.5
17.5
Friction angle ()
Cohesion c(kPa)
14.4
11.4
11.4
11.4
Pore water pressure
Not exist
exist
exist
Not exist
Buoyancy
Not exist
exist
Not exist Not exist
Condition 1:Ntural soil without underground water;
Condition 2:For soil with good water permeability, saturation line inside soil is
shown in Fig.2. The value of pore water pressure alone the sliding surface is 5m for
simplifying the calculation. Saturated soil is controlled by buoyancy.
Condition 3: Because of sudden drawdown of water level of the condition 2, there
are no more effects of buoyancy on sliding soil. While pore water pressure will not
quickly disappear.
Condition 4: For soil with poor water permeability, by the action of groundwater
corrosion, non-pressure water infiltrate along the boundary of karst cave. Then, only
the shear strength is reduced.
Condition 5: Also for soil with poor water permeability, by the action of
groundwater corrosion and seepage pressure, the shear strength is reduced and there
appear pore water pressure on the sliding surface.
The situation in condition 3 and condition 5 is similar. Soil is not affected by
buoyancy, but controlled by pore water pressure on sliding surface. This is the worst
situation.
ASCE
76
77
Hyd
draulic Effeccts on Stabiliity of Mud In
nrush
Safety
S
factorrs with or wiithout bound
dary restraintt are obtaineed, as shownn in Fig.4.
Safetyy Factor
3
2..39
2.5
withou
ut boundary rrestraint
with b
boundary resttraint
2.3
37
1.70
0
1.5
1.70
2 .00
1.20
0.94
0.71
0.92
0.71
0.5
0
1
Conditioon
FIG.4 Histogram
m about safetty factors u nder differeent conditioons
witth or withou
ut boundaryy restraint
As
A can be seeen from Fiig.4, the safe
fety factor o f condition 1 is larger than that off
otheers. It showss that underg
ground waterr play an impportant role in integral sliding.
The
T safety faactor of cond
dition 2 is laarger than thhat of condittion 3. It shoows that, forr
soil with good water
w
permeeability, sudd
den drawdow
wn of water will increase the risk off
mud
d inrush beccause of variiation of buoyancy and constant poore water preessure. Thiss
con
nclusion coin
ncides with engineering
e
actuality.
a
The
T safety factor
f
of con
ndition 5 is significantlyy larger thann that of conndition 2. Itt
show
ws that, at th
he same porre water pressure, fillinggs with poorr water perm
meability aree
morre readily to slide than porous filling
gs.
Influence of Sheear Strength on Stabilityy of Mud Inruush
Safety Factor
1.6
=18
=20
=22
1.4
1.2
1
0.8
0
10
20
30
40
Cohesion
C
(kP
Pa)
FIG
G.5 Relation curves beetween safetty factor and
d cohesion
The
T safety faactor of cond
dition 1 is laarger than thhat of conditiion 4 as show
wn in Fig.4.
It sh
hows that sh
hear strength
h has a greaat effect on iintegral slidding. Under ccondition 1,
reseearch about the influen
nce of shearr strength oon the stabiility is condducted withh
diffferent valuess of C and .. And relatio
on curves beetween safetyy factor and cohesion off
soils with differrent inner friiction angless is plotted aas shown in F
Fig.5.
Known
K
from
m the analysis of relation
n curves, thee safety factoor increase liinearly withh
the increase of cohesion
c
an
nd also increase with thee increase off inner frictiion angle. Itt
ws that the larger
l
the sh
hear strength
h, the strongeer the abilityy on resistingg the risk off
show
mud
d inrush.
ASCE
78
Influence of Fillings Content on Stability of Mud Inrush

Once the sliding surface and failure mass is defined, the magnitude of exterior load
is only controlled by fillings content. As a variable for research on influence, the
values of exterior load are set to Pi2Pi3Pi4Pi5Pi6Pi, where the value of Pi
is the initial exterior load.
Under condition 1, relation curves between safety factor and maximum shear force
at the cave boundary with the increase of fillings content(exterior load) are plotted as
shown in Fig.6.
2000
Safety factor with boundary restraint

safety factor without boundary restraint
maximum shear force
1.5
1600
1200
1.3
800
1.1
400
0.9
0
0
x Pi
Maximum Shear Force

(kN)
Safety Factor
1.7
Exterior Load (N/m)

FIG.6 Relation curves between safety factor and maximum shear force
with the increase of fillings content (exterior load)
By the analysis of relation curves, without boundary restraint, the safety factor
decrease gently with the increase of fillings content. But with the continuous increase
of fillings content, the change of safety factor is smaller and smaller. It shows that
fillings content has a certain extent effect on integral sliding, but is not the main
influence. And with boundary restraint, the influence on integral sliding is lesser.
And yet, with the increase of fillings content, the maximum shear force
significantly increase while the safety factor change slightly. It shows that fillings
content is not the main influence on integral sliding, but has a great effect on structural
security of cave boundary.
Influence of Boundary Restraint on Stability of Mud Inrush
As can be seen from Fig.4, safety factors with boundary restraint is significantly
higher than that without boundary restraint. It shows that boundary restraint plays an
important role in risk reduction of mud inrush.
Under condition 4, as a variable for research on influence, the length of boundary
restraint are set to 0.4m, 1.3m, 2.7m, 4.3m, 6.0m, 7.8m. And by calculation, relation
curves between safety factor and fillings content is plotted as shown in Fig.7.
ASCE
79
1600
safety factor
maximum shear force
3.2
1200
2.4
800
1.6
400
Maximum Shear
Force (kN)
Safety Factor
0.8
0
Length of Boundary(m)
FIG.7 Comparison between safety factor and maximum shear force
with the increase of length of boundary
Known from the analysis of relation curves, with the increase of the length, the
safety factor and the maximum shear force both significantly increase together. It
shows that the longer the cave boundary, the higher the risk of structural failure while
the less likely mud mass slide away. This coincides with engineering actuality: fillings
in a closed cave can not slide away before structural failure of cave boundary.
CONCLUSION
The influence of underground karst water on the stability of mud inrush is
summarized in three aspects: The shear strength of the interface is reduced by water;
The frictional resistance of the interface is reduced because of pore water pressure;
Porous soil below saturation line is affected by buoyancy. Porous soil is more likely to
encounter slip failure, while there happpen a sudden drawdown of water. But at the
same pore water pressure, soil with poor water permeability is more readily to slide
than porous fillings.
The shear strength has a great effect on integral sliding: the larger the shear
strength, the stronger the ability on resisting the risk of mud inrush.
Fillings content is not the main influence on integral sliding, but has a great effect
on structural security of cave boundary. In other words, while cave is filled with high
mud content, the mud mass may not slide because of the cave boundary, but the cave
boundary will first be at risk of structural failure. So, fillings content is also the main
influence on the stability of mud inrush.
Boundary restraint plays an important role in risk reduction of mud inrush: the
longer the cave boundary, the higher the risk of structural failure by the increase of the
maximum shear force, while the less likely mud mass slide away.
ACKNOWLEDGMENTS
The authors appreciate the support of the National Basic Research Program of China
(973 Program, No.: 2013CB036000), the National Natural Science Foundation of
China (Grant No.: 51479106), the promotive research fund for excellent young and
middle-aged scientists of Shandong Province (Grant No.: 2014GN028).
ASCE
REFERENCES
Li, S., Zhou, Z., Li, L., et al. (2013). Risk assessment of water inrush in karst tunnels
based on attribute synthetic evaluation system. Tunnelling and Underground Space
Technology. Vol. 38: 50-58.
Shi, S.S. (2014). Study on seepage failure mechanisim and risk control of water inrush
induced by filled disaster structure in deep-long tunnel and engineering
applications. Jinan: Shandong University.
Li, S.C., Lin, P., Xu Z.H., et al. (2015). The minimum safety thickness of water and
mud inrush induced by filled-type, karst water bearing structures based on the
concept of slice method. Rock and Soil Mechanics. Vol. 07:1989-1994+2002.
Xu, Z.H., Li, S.C., Li, L.P., et al. (2011). Cause, disaster prevention and controlling
of a typical kind of water inrush and lining fracturing in karst tunnels. Chinese
Journal of Rock Mechanics and Engineering. Vol. 30(7): 13961404.
Wang, X.Y., Tan, Z.S., Wang, M.S., et al. (2008). Theoretical and experimental
study of external water pressure on tunnel lining in controlled drainage under high
water level. Tunnelling and Underground Space Technology. Vol. 23(9): 552
560.
Zhu, D.Y., Lee, C.F., Huang, M.S., et al. (2005). Modifications to three well-known
methods of slope stability analysis. Chinese Journal of Rock Mechanics and
Engineering. Vol. 24(2): 183194.
Fang, Y.S. (2008). The lowest solution of slice method for slope stability analysis.
Chinese Journal of Geotechnical Engineering. Vol. 30(3): 331335.
Tong, Z.Y., Chen, C.X., Xu, J., et al. (2009). A slice-stability method for stability
analysis of slopes. Rock and Soil Mechanics. Vol. 30(5): 13931398.
Liu, Z.Y., Chen, S.W. (2002). Evolution model of progressive failure of strainsoftening soil slopes. Journal of Zhengzhou University (Engineering Science). Vol.
23(2): 3740.
Bishop, A.W. (1955). The use of slip circle for the stability analysis of slopes.
Geotechnique. Vol. 5(1): 717.
Morgenstern, N. R., Price, V.E. (1965). The analysis of the stability of general slip
surfaces. Geotechnique. Vol. 15(1): 7993.
Spencer, E. (1965). A method of analysis of the stability of embankments assuming
parallel inter-slice forces. Geotechnique. Vol. 17(1): 1126.
Janbu, N. (1973). Slope stability computations. Embankment Dam Engineering.
New York: Wiley: 4786.
Chen, Z.Y. (2003). Soil slop stability analysis-theory, methods and programs.
Beijing: China Water & Power Press.
Liu, G.B., Wang, W.D. (2009). Foundation pit engineering manual. Beijing:
China Building Industry Press.
Su, A.J. (2005). The principle and method of landslide stability evaluation.
Wuhan: China University of Geosciences Press.
ASCE
80
81
Study on the Reasonable Cover Thickness of a Subsea Tunnel with the Numerical Calculation
Criterion Method
B. S. Xu; S. C. Li; R. C. Liu; and C. L. Zhao
(School of Civil and Hydraulic Engineering, Shandong Univ., Jinan 250061, China.)
Abstract: The minimum rock cover thickness is a significant parameter during longitudinal
sections design of subsea tunnel by means of drilling and blasting. The thickness will impact the
cost and safety when subsea tunnel is constructed. On the basis of numerous elasto-plasticity
numerical tests and theoretical analysis, the minimum rock cover thickness criterion is provided,
including minimum displacement method and factor of safety (FOS). For a certain section of
subsea tunnel, with the growth of rock cover thickness, vault displacement will decrease first and
then increase. There always exists a minimal value. At the same time, the same section is
calculated and analyzed using a strength reduction method, and some clear regularities of subsea
tunnels FOS are summarized. That is, for any section of a subsea tunnel, with the increasing of
rock cover thickness, FOS increases first and then decreases, always with a maximum value.
There is a close correlation between FOS and vault displacements variation tendencies.
Accordingly, it can be thought that the technology feasible and economic reasonable minimum
rock cover thickness is a rational and effective thickness, considering the maximum FOS and
minimum vault displacement at the same time. Under this numerical method, the determined
minimum rock cover thickness can give great reference values to the planning of tunnels
vertical routes.
Keywords: Minimum vault displacement; Safety reserve coefficient; Qingdao Jiaozhou Bay
Subsea Tunnel; Minimum rock cover.
INTRODUCTION
In China, there are many subsea tunnels being planned, constructed and used. For example,
sea channel connecting Hong Kong, Macao and Zhuhaisubsea tunnel of Dalian Baysubsea
railway tunnel from Dalian to Yantai in Bohai Baysubsea railway tunnel of Qiongzhou Strait
subsea tunnel of Taiwan Strait, and so on. Xiang'an subsea tunnel and Qingdao Jiaozhou Bay
subsea tunnel have been constructed and are in use. It is a serious problem of how to determine
the subsea tunnels minimum rock cover thickness. At the same time, it designs tunnels safety
during construction and operation, while considering tunnels economical efficiency.
The minimum rock cover thickness is a significant parameter during longitudinal sections
design of subsea tunnel by means of drilling and blasting method. On the one hand, if rock cover
thickness is too thin, then there will be serious instability problems with the danger of seawaters
ASCE
inburst on the working face of subsea tunnel. Even if it is safe, it would increase the cost of
auxiliary construction, which will also increase the cost of tunnel support, seepage-proofing and
dewatering. On the other hand, thicker rock cover thickness means increasing the depth of tunnel,
increasing the length of tunnel, increasing the head pressure impact on lining structure, and thus
increasing the cost of subsea tunnelSun Jun 2006;Lu Ming 2005;Li shucai 2007;Wang mengshu
2205;Nilsen 1993; etc. So, it is obvious that there is always an optimum value of subsea
tunnels rock cover thickness.
The experience of constructed subsea tunnels in the world can be used for reference. Japanese
Seikan highway tunnel has an overall length of 53.85km, with 23.0 km in the subsea portion. The
water depth is 140m, and the minimum rock cover thickness is 100m. Japanese Kanmon subsea
railway tunnel is the earliest channel tunnel in the world. The tunnel has two holes, with roads in
both directions. The length is 1.14km, with height 5.75m and maximum slope 2.0~2.5 %. The
depth of seawater is 14 m, and the average rock cover thickness is 11m. The thinnest part is 9.5m,
located on the area near seabed rockfill and clay. Drilling and blasting methods for subsea tunnel
is used the most in Norway, compared with other countries. In the past 30 years, Norway has
built more than 40 tunnels, with a total length of more than 100km. Norway highway tunnel
design specifications define that subsea tunnels minimum rock cover thickness must be greater
than 50m. If the thickness is less than 50m, then detailed geological investigations and special
analysis must be conducted. Also, the design must be reported to the national highway
administration, and get their permission. However in fact, most of the Norway subsea highway
tunnels rock cover thickness is less than 50m. The shallowest part is only 23mNilsen 1993;Lu
Ming 2205; etc.
Shandong University has researched about minimum rock cover thickness through
engineering analogy, numerical calculation and model experiment on the bases of the constructed
Xiamen Xiangan subsea tunnel and Qingdao Jiaozhou Bay subsea tunnel. The system of subsea
tunnels minimum rock cover thickness has been built through engineering analogy, numerical
calculation and theoretical closed form analysis. In the mechanical analysis of engineering
problems, closed form solutions can be used only in the simple situation, while numerical
method can adapt to complicated boundary conditions and consider the mediums constitutive
relations. So, numerical method is widely used in engineering practice. At present, the numerical
methods usually used in the stability analysis of geotechnical engineering include finite element
method, finite difference method, discrete element method, boundary element method. In this
paper, the finite difference method is used to research the problem.
A subsea tunnels minimum rock cover thickness is a problem of optimization. There are
many criterions to evaluate a tunnels stability. For example, one is strength criterion. It estimates
surrounding rocks stability through comparing rock stress or stress combination to the ability of
the rock's resist compression, tensile, resisting shear and other strength factors during different
loads. Another is deformation or deformation rate criterion. The related standards at home and
abroad mainly focus on surrounding rocks deformation or deformation rate. It is thought that as
long as the deformation or deformation rate exceeds a certain value, the rock is thought to be
broken downShao Guojian,2003Li shuchen,2007. This paper based on Qingdao Jiaozhou
ASCE
82
83
Bay subsea tunnel, provides a vault displacement criterion method and security coefficient
criterion method of strength reduction theory, which has been applied in engineering practice
successfully.
Description of the project
Qingdao Jiaozhou bay subsea tunnel is a strategic road to connect main urban area and
assistant urban area, The south is Xuejiadao, north is Tuandao, and the tunnel passes through sea
area of Jiaozhou bays mouth. The tunnel is a city highway tunnel with six lanes in two
directions. The allowable speed of motor vehicles is 80km/h.Submarine tunnel under Jiaozhou
Bay(including connecting tunnel) has a length of about 7800m. Vehicular tunnel of Jiaozhou bay
tunnel is divided into left lines and right lines. The spacing of tunnel center lines is 55m. And it
is designed with a pedestrian cross every 200m to 250m, while there is a traveling crane cross
every 750m. A service channel is designed in the center of left line and right line, paralleled. In
order to cut down the length of the tunnel, longitudinal slope generally has a large shore slope
ratio, and the largest slope ratio is within 3.9%. The layout drawing of tunnels cross section and
vertical section are shown in Fig.1 and Fig.2. Normal cross section for inner contour of tunnel is
shown in Fig.3.
Fig.1 Distribution of tunnels cross sections
Fig.2 Distribution of tunnels vertical sections
ASCE
84
Fig.3 Normal cross section for inner contour of tunnel

From Fig.2, it is shown that in the deepest point of subsea tunnel, the seawaters depth is
about 40m, and the designed rock cover thickness is approximate 30m. The lithology of the
section in the area is mainly composed of rhyolite and andesite volcanic breccia.
Completely-weathered and highly-weathered zones of weathering don't grow, and
weakly-weathered regoliths thickness is often less than 10m. The geological survey shows that
the class of surrounding rock is level . According to the report of geological survey, the classes
of rock mass are divided by Basic Quality, which is an engineering classification of rock mass.
The physical and mechanical properties of surrounding rock are showed in Table 1.
Table.1 Physical and mechanical properties of surrounding rock
class of surrounding
rock
unit
weight
(kg.m-3)
23.5
modulus of
elastic
E(GPa)
1.36
Poissons
ratio
0.33
cohesion
C(MPa)
0.28
internal friction
angle
()
28.72
tensile
strength
t (MPa)
0.14
CRITERION METHOD
Using the designed section's deepest point of the Qingdao subsea tunnel, the surrounding
rocks stability and elastic-plastic numerical analysis were done for rock cover thicknesses. Vault
displacement criterion method and security coefficient criterion methods were provided and
applied in the numerical simulation.
Vault displacement criterion method
The paper uses finite difference software to do elastic-plastic numerical analysis research.
Large deformation, seawater pressure and water pressure of rock fracture water are considered.
The calculation model is showed in Fig.3. The width from tunnel to boundary is 3-5 times of
tunnels diameter. The horizontal direction is thought to be X-axial, with left and right boundary
to be 60m respectively. The vertical direction is thought to be Y-axial, with bottom boundary to
be 60m and up boundary to be seabed. The direction along the tunnel is Z-axial, with the
thickness to be 1. Elastic-plastic constitutive model and non-associated flow rule are used as well
as Mohr-coulomb yield criterion to be usedZhang Liming,2007. That is:
1
1
F = I1 sin + cos
sin sin J 2 c cos = 0
3
3
ASCE
(1)
Where, I1 J 1 are stress tensors first invariant, deviatoric stress tensors second invariant
and Lode angle; c are rocks cohesion and internal friction angle.
According to the report of geological survey, the classes of rock mass are divided by Basic
Quality. Initial ground stress in vertical direction only considers rocks weight stress, and the
horizontal ground stress is considered to be 0.8 times the vertical ground stress. Left and right
boundaries are fixed on x-direction, and bottom boundary is restrained in three directions(x,y,z).
The model is simplified to be plane strain problem, with restrain in y-direction restrain. Up
boundary is applied hydrostatic pressure. Pore water pressure is of graded distribution in the
scope of the model. For example, the thickness of rock cover is 20m, and the tunnel is excavated
using up and down steps method. The finite difference calculation gridding is showed in Fig.4.
Elastic-plastic numerical calculations are finished under different rock cover thicknesses. The
transformation regulars of key points and plastic zone around the tunnel are analyzed after the
tunnel is excavated. The distribution of key points is showed in Fig.5, with point 1 to record sink
of vault, point 2 and 3 to record convergence of top bars boundaries, point 4 to record upheaval
of inverted arch. The displacement change of key points around tunnels boundary is showed in
Fig.6. The change of plastic zone is showed in Fig.7. The distribution of plastic zone under
different rock cover thicknesses is showed in Fig.8.
From Fig.6, we can see that with the increase of rock cover thickness, creating a minimum
value. The rock cover thickness, corresponding to vault displacements minimum, is 16 m.
Upheaval of inverted arch presents to be linear trend, and convergence presents to be
approximate a linear trend of increase. Through analysis, the shape of tunnels cover rock is
thought to be similar to arch. If vault sinks, the stiffness of rock will become weaker when the
rock cover thickness is thin. So, the deformation will increase under the pressure of seawater.
When rock cover is thick, the initial ground stress around tunnel will become larger, and the
deformation will increase too. Convergence of tunnels boundary and deformation of inverted
archs upheaval have small relationship with rock cover thickness, but are influenced largely by
tunnels depth. From Fig.7, we can see, with the increase of rock cover thickness, the area of
plastic zone around the tunnel increases. When the rock cover thickness is 10m-20m, the area of
plastic zone around the tunnel is about 40 m2~60 m2. For other sections of Qingdao subsea
tunnel, similar results were found, too. In conclusion, for a certain section of subsea tunnel, with
the growth of rock cover thickness, vault displacement will decrease first and then increase.
There always exists a minimal value. When the rock cover thickness is determined by minimum
displacement, through the stability analysis of deformation, it is found that tunnels surrounding
rock is relatively more stable, too. Besides, the sink of vault due to the tunnel's excavation can
reflect the surrounding rocks deformation and stability most. Therefore, the rock cover thickness
corresponding to minimum vault displacement is thought to be a technically reasonable
minimum rock cover thickness.
ASCE
85
86
Fig.4 Calculation grids when the rock cover thickness is 20m
Fig.5 Distribution of calculated monitoring points

3.00
sink of vault
convergence of tunnel's boundary
upheaval of inverted arch
displacement(cm)
2.50
2.00
1.50
1.00
0.50
0.00
0
10
20
30
40
50
60
rock cover thickness(m)
Fig.6 Deformation curve of tunnels surrounding rock under different rock cover
thicknesses
area of plastic zone(m2)
180
160
140
120
100
80
60
40
20
0
0
10
20
30
40
50
rock cover thickness(m)
Fig.7 Plastic zones change curve of tunnels surrounding rock under different rock cover
thicknesses
ASCE
87
(a) minimum rock cover=10m (2) minimum rock cover=25m (3) minimum rock cover=40m
Fig.8 Plastic zone under different rock cover thicknesses
Factor of safety method
Zienkiewicz and others have presented strength reduction technology in 1975. With the spread
of numerical simulation software and the improvement of computers property, strength
reduction technology has been widely used in stability analysis of geotechnical engineering.
From the point of physical interpretation, strength reduction technology is a concept based on
material strength reserve. Therefore, the physical meaning and security coefficient of limiting
equilibrium method are almost the same. Duncan J M
1996Griffiths D V Lane P A
1999
Zheng Yingren2004Tang Chunan2006etc have been using strength reduction technology
to research a series of slope projects. Zhang Yiming2007 etc try to apply strength reduction
technology into an estimation of tunnels stability. Subsea tunnel using drilling and blasting
methods can adapt to the rocks of better geological conditions, with thin rock cover thickness
and flat shape of tunnel. The rock cover part is easy to be destroyed and would run through to
seabed, existing a potential slip plane. The paper uses strength technology to do stability research
with subsea tunnels rock cover stratum.
Subsea tunnels strength reduction technology is to reduce the strength parameters of
surrounding rock until the tunnel is unstable or destroyed, during the process of real
elastic-plastic numerical calculation. According to potential failure surface caused by plastic
strains cut-through, strength storage security coefficient of tunnels surrounding rock can be got
through calculating. Finite difference griding, constitutive model, yield criterion, flow method,
mechanical parameter, initial condition and boundary condition are the same as section 3.1. The
definition of FOS Fs based on the concept of strengt c = c / Fs h storage is thought to be: when
materials shear strength parameters c and t are replaced by critical intensity parameters
c and t respectively, then the tunnel will fall into critical equilibrium state. Where:
(2)
= / Fs
ASCE
(3)
88
t = t / Fs
(4)
At present, theres not a uniform unstability criterion of tunnels surrounding rock. That is to
say, now, we have no way to get an acceptable terminal condition of FOS. The unstability
criterions of tunnels surrounding rock are as the followings: regarding force or
displacements convergence as unstability criterion Tang Chunan 2006 regarding
displacements abrupt change of feature parts as unstability criterion. For example, Wang
Mingnian 2010 researched deformation regular of Xiamen subsea tunnels
completely-weathered and highly-weathered sections, and provided control standards of absolute
displacement. regarding plastic zone or plastic strains cut-through as unstability criterion.
The paper mainly uses plastic zones cut-through as subsea tunnels unstability criterion. If the
rock cover thickness is 8m, with the increase of strength reduction, the rock cover destroyed as
shown in Fig.9. Firstly, failure zones occur near the flanks. And then spreads to all the
boundaries of the tunnel. Next, failure zones expand to the upper part along both sides of
spandrel. Finally, plastic zones are formed between tunnel vaults both sides and seabed, which
means the tunnel is destroyed. Fig.10 shows shearing strain distribution regular at the first step of
destruction. Also, the destruction starts from spandrels both sides, spreads in an inclined upward
direction of tunnels outboard. The relation curve of FOS and subsea tunnels rock cover
thickness is shown in Fig.11. Obviously, the calculated subsea tunnels FOS has clear regularities.
For any section of subsea tunnel, with the growth of rock cover thickness, FOS will increase first
and then decrease. There always exists a maximum value. The rock cover thickness
corresponding to maximum FOS is 18 m.
From the analysis above, the reasonable rock cover thickness determined by vault
displacement method is 16 m. The value can be regarded as designs lower limit. The rock cover
thickness determined by FOS method is 18 m. The value can be regarded as designs higher limit.
Above Qingdao subsea tunnels deepest points section, there are completely-weathered and
highly-weathered parts with a thickness of about 10 m. The strength of completely-weathered
and highly-weathered parts is very low, and carrying capacity is very low, too. Therefore, the
rock cover thickness should be the summation of completely-weathered and highly-weathered
parts thickness and reasonable rock cover thickness. The value is suggested to be 23~30 m. In
practice, it is reasonable and reliable to set a designed value to be 30m.
ASCE
89
Fig.9 Failure progresses of rock cover partsminimum rock cover=8 m
Fig.10 Shearing strain of rock cover parts when destruction happens

minimum rock cover=8 m
Fig.11 Security coefficient curve under different rock cover thickness
ASCE
CONCLUSIONS
Through lots of numerical calculations and theory analysis, the minimum vault displacement
method and FOS are used to determine the minimum rock cover thickness. The two methods can
be proved by each other. Calculation results show that for any section of subsea tunnel, vault
displacement will decrease first and then increase, always existing a minimum value, with the
increase of rock cover thickness. Besides, the calculated FOS of subsea tunnel has obvious
regularities. For any section of subsea tunnel, FOS will increase first and then decrease, always
existing a maximum value, with the increasing of rock cover thickness. There is a good
corresponding relationship between vault displacements and FOS distribution trends. The
feasible technology and economically reasonable minimum rock cover thickness is a rational and
effective thickness, considering the maximum security coefficient and minimum vault
displacement at the same time. Under this numerical method, the determined minimum rock
cover thickness can give great reference values to the planning of tunnels vertical routes.
ACKNOWLEDGEMENTS
Financial supports for this paper provided by National Science Foundation of China
(No.50909056) and Provincial Science Foundation of Shandong Province (No.ZR2014EEM014,
ZR2014EEM029) , and Science and Technology Project Plan in 2015, Ministry of Housing and
Urban-Rural Development of China (No. 2015-K5-004) are gratefully acknowledged.
REFERENCE
Sun, J. (2006). Discussion on Some Key Technical Issues for Design and Construction of
Undersea Tunnels. Chinese Journal of Rock Mechanics and Engineering. Vol.25(8):
1514-1520.
Wang, M.S., Huang, F.M. (2005). Key Problems on Subsea Tunnel Construction. Journal of
Architecture and Civil Engineering. Vol.22(4): 1-4.
Li, S.C., Li, S.C., Xu, B.S., Wang, H.P., Ding, W.T. (2007). Study on Determination Method for
Minimum Rock Cover of Subsea Tunnel. Chinese Journal of Rock Mechanics and
Engineering. Vol. 26(11): 2289-2295.
Nilsen B. (1993). Empirical analysis of minimum rock cover for subsea rock tunnels. In:
Burger H ed. Options for Tunnelling. Amsterdam: Elsevier. 677-687
Lu, M., Grv, E, Nilsen, B., Melby, K. (2005). Norwegian Experience in Subsea Tunnelling.
Chinese Journal of Rock Mechanics and Engineering. Vol. 24(23): 4219-4225.
Li, S.C., Li, S.C., Zhang, J.W., Xu, B.S. (2006). Study on Numerical method for the Minimum
Rock Covers of Subsea Tunnels. Chinese Journal of Geotechnical Engineering. Vol. 28(10):
1304-1308.
Li, S.C., Xu, B.S., Ding, W.T., Zhang, Q.S. (2009). Weighted Function Method for Minimum
Rock Cover Thickness of Subsea Tunnel. Rock and Soil Mechanics. Vol. 30(4):989-996
ASCE
90
Zienkiewicz, O.C., Humpheson, C., Lewis, R.W. (1975). Associated and nonassociated
viscoplasticity and plasticity in soil mechanics. Getechnique. Vol. 25(4):671-689
Duncan, J.M. (1996). State of the art: limit equilibrium and finite element analysis of slopes.
journal of Geotechnical and Geoenvironmental Engineering. Vol. 122(7):577-569.
Griffiths, D.V., Lane, P. A. (1999). Slope stability analysis by finite elements. Geotechnique,
Vol. 49(3): 387-403
Zheng, Y.G., Zhao, S.Y. (2004). Application of strength reduction FEM in soil and rock slope.
Chnese Journal of Rock Mechanics and Engineering. Vol. 23(19): 3381-3388.
Tang, C.A., Li, L.C., Li, C.W., Ma, T.H. (2006). RFPA strength reduction method for stability
analysis of geotechnical engineering. Chnese Journal of Rock Mechanics and Engineering.
Vol. 25(8): 1522-1530
Zhang, L.M., Zheng, Y.R., Wang, Z.Q., Wang, J.X. (2007). Application of strength reduction
finite element method to road tunnels. Rock and Soil Mechanics. Vol. 28(1): :97-106.
Wang, M.N., Lu J.F., Liu, D.G., Zhang, J.G. Study of absolute deformation control criterion and
its application for large section subsea tunnel with "CRD" method.
Shao, G.J., Zhuo, J.S., Zhang, Q. (2003). Research on analysis method and criterion of
rockmass stability. Chinese Journal of Rock Mechanics and Engineering. Vol. 22(5):
691-696.
ASCE
91
92
A Study on the Strategy of Seepage Leakage Treatment of Highway Tunnels

under a Long-Term Groundwater Table
Y. L. Zhang1 and Q. Y. Tian2
1
Senior Engineer, Geotechnical Engineering Dept., Guangdong Hualu Transportation Science and
Technology Com. Ltd., Guangzhou 510032, China. E-mail: zhyl0802@163.com

2
Director, Geotechnical Engineering Dept., Guangdong Hualu Transportation Science and Technology
Com. Ltd., Guangzhou 510032, China. E-mail: 826875931@qq.com
Abstract: Under the condition of long-term action of groundwater table, especially in

the karst area, tunnels diseases, especially seepage leakage and lining deterioration,
which threatening the safety of traffic and tunnel structure, appear gradually due to the
changing groundwater flow path around tunnels with atmospheric precipitation in the
catchment area and physical and mechanical properties worsen of surrounding rocks and
soils. How to prevent and treat the diseases effectively have become an urgent problem
to be solved. A comprehensive strategy, with the example of the Pishuangao tunnel on
Beijing-Zhuhai expressway, combined with ground surface, surrounding rock and the
lining treatment, after architectonic geology, hydro- and engineering geology,
geophysics and engineering geological survey, was adopted by the authors, and it
achieved a satisfactory result after two years verification.
INTRODUCTION
Groundwater is stored in the rock void. The amount, size, connectivity and abundance
of the void is directly dependent on the lithology and structure of the rock. While the
amount of groundwater is dominated by the lithology and structure of the rock, the
terrain and climate conditions affect the supply, run-off and discharge of groundwater.
The highway tunnel is often designed to have the function of drainage, and it becomes
an important discharge way for the groundwater in the tunneling region. After the
long-term operation of the highway tunnel, the influence of the groundwater on the
surrounding rock can be indicated in several aspects as follows: softening the rock,
weakening the structural surface, and instability of the rock due to artesian water
induced sliding force, etc. The aforementioned influence can change the existing rule of
supply, run-off and discharge of groundwater during tunnel operation due to the
progressive change in geological structures, meteorologic condition, groundwater
migration and the direction of the maximum lateral principal stress. For the supply of
ASCE
93
groundwater, vertical charge from precipitation and river is mainly dependent on the
meteorologic condition, while the lateral charge is greatly dependent on the abundance
of rock joints and fractures. Due to the weakening of rock and structural and thus
resulting abundance of rock joint and fracture, the run-off of groundwater is changed. In
addition, the construction of the tunnel has great influence on the original geological
structure and further the run-off of groundwater. The leakage of groundwater on the
ground, walls and vaults of the tunnel happens very often. These leakage of groundwater
can further deteriorate the rock. As such, the failure of the tunnel structure can occurs if
no prevention is made during the tunnel operation.
Tunnels, which have accident hazard liability and is generally considered as a special
structure, should be paid much attention on its maintaining based on the mechanism and
feature of disaster. By doing so, it can ensure the safety of tunnel structure and
operation.
In this paper, taking the in groundwater leakage in the Pishuangao tunnel of the
Beijing-Zhuhai expressway as an example, it proposes a method for analyzing the cause
of groundwater leakage and its treatment.
PROJECT OVERVIEW
The Pishuangao tunnel, which is a section of Beijing-Zhuhai expressway and started
in service in 2003, has a length of 760 m, and it is constructed across the karst stratum.
In this region, the atmospheric precipitation and groundwater are rich. Between 2011
and 2012, the leakage of groundwater in the tunnel became serious. Even the spouting of
groundwater occurred in rainy season, which greatly affects the normal operation and
traffic safety of the tunnel. The leakage of groundwater in the tunnel is shown in Fig. 1.
and Fig. 2.
FIG. 1 Spouting of groundwater in

the tunnel
ASCE
FIG. 2 Spouting of groundwater on

lateral wall with the mud
REASONS OF THE SEEGAGE LEAKAGE

To prevent leakage in the tunnel, the first step is to find out the source, run-off and
discharge of the groundwater, and then make a suitable solution to the issue.
Based on the preliminary geological and geophysical investigation, the present study
provides large-scale map of hydro-geological profile in the tunneling region. The
groundwater connectivity test with ion tracer technique is carried out to the connection
between surface water and leaking groundwater in the tunnel. The high density
resistivity method and seismic imaging technique are used to probe the karst rock in this
region. Through principles of architectonic geology, fine hydro- and engineering geology,
geophysics and engineering geological survey, the origin of the leakage and rock mass
conditions are explored. The result of survey shows that there are three main reasions for
groundwater leakage in the tunnel.
Multiple infiltration paths
The tunnel space is connected to the water catchment, such as marsh land and ravines,
through pores, fractures, catavothres, karst caves. Through fine hydro-geological
investigation, it is found that the groundwater is carbonate and karst water. There is no
Perennial accumulating surface water, and it is mainly charged by the rain infilration. In
the dry season, groundwater recharges the surface water; while in the rainy season, the
river recharges the groundwater. In the bedrock mountainous and hilly area, vertical
infiltration is the main charge. In this kind of area, the depth of groundwater is shallow,
the run-off is short, hydraulic gradient is big. Therefore, it is the shallow circular mesh
fissure water and it is discharged to the valley in the form of bulk flow or falling flow.
Consequently, it becomes the resource of surface water in the dry season. And part
becomes the resource of the quaternary pore water. There are twenty-five position where
ground subsidence and karst doline happened. The subsidence occurred mainly in the
shallow-depth area, around surface gully and reversed fault (around old karst subsidence
or existing cut-off ditch). These subsidence becomes the main infiltration path of surface
water. Through trenching, it is found that there is a shallow 0.5-1.0 m layer of clayed soil
and bare rock in some area. A in-situ permeability test was conducted and the result
shows that the seepage velocity is big and permeability coefficient is up to 1~5m/d. The
surface water flows into the tunnel from the interface between the rock and surface
clayed soil.
High infiltration velocity
There is abundance of surface karst subsidence, karst doline and caven. There is rich
karst rock behind the lining, even the rock mass fracture occurs with dramatic water
leakage. In the shallow area, the surface water is directly irrigated into the tunnel, which
results in big amount of water behind the lining in the rainy season.
ASCE
94
95
The high density resistivity method and seismic imaging technique are used to find
out the developing state of karst dolines. And the connectivity test is used to determine
the connection between surface water and groundwater.
Karst dolines and cave distribution around the tunnel
A high density resistivity test was conducted from the surface of strata above the
tunnel with seven surveying lines. From the results, six karst dolines were found along
five lines and the biggest one is located along the tunnel trend and has a length of 80 m.
Another four karst dolines were found on the surface of strata above tunnel. Fig. 3.
shows a typical cross-section by the high density resistivity method. For the karst
dolines observed, most of them are located in the position close to tunnel top, and they
can provide the infiltration path of water in subsidence and karst cave into the tunnel.
This will cause dramatic water leakage in the karst area around the tunnel.
FIG. 3 PMX7-7 vertical section in the Pishuangao tunnel

A seismic imaging test was conducted from lining surface of the tunnel. There are
sixty-one karst dolines were found, of which fifty-three ones are assumed to be karst
caves, and eight one as karst fractures. The width of karst dolines is in the range of
2.8~32.4 m and the height is about 1.7~33.5 m. The connection of these karst dolines
to the surrounding karst provides a good path for the water infiltration and leads to a
disaster. The fracture development zone in the LK1905+378.8~+476.4 section have an
apparent dip of 75~79 to the small number section. In this zone, the rock mass is
cracked, and big or bead-like karst caves may exist. This provides infiltration path for
the groundwater. Fig. 4 shows the map obtained by the seismic imaging test. Theses
karst dolines and fractures may be connected to each other, which provides a good
infiltration path for groundwater and further leads to the disaster.
ASCE
96
FIG. 4 Karst distribution map using seismic imaging method

Connectivity of Karst dolines and cave
A series of connectivity tests were operated in the Pishuangao tunnel, nine injection
points were arranged on the ground of the tunnel, and six receiving points were arranged
in the tunnel and the surrounding. The test results are shown in Fig. 5. Judged from the
results, the groundwater toward the west from TF05-TF07 lines will join the Changxi
River all the ways and will not flow into the tunnel. The shallow groundwater between
TF01-TF04 lines is the main source of the gushing water in the tunnel. The time
histories of concentrations of TF01 (Fluorescence), TF01 (rhodamine) and TF03
(fluorescent bleacher) at the two ends of the tunnel are shown in Figs. 6 and 7.
FIG. 6. Time histories of concentrations of TF01, TF02 and TF03 at the end
from Beijing side
ASCE
FIG.7. Time histories of concentration of TF01, TF02 and TF03 at the end from
Zhuhai side
Defects of the tunnel
Due to the insufficient drainage capacity or weak position of drainage system of the
tunnel, groundwater leakage could happen in the tunnel in the rainy season.
STRATEGY OF SEEPAGE LEAKAGE TREATMENT
To prevent the leakage, draining treatment is adopted on the ground surface, in the
tunnel and behind the tunnel linings.
Treatment on the ground surface
The blocking, cut-off and draining methods are used on the ground surface in the
tunnel. Backfilling the aven and subsidence, circumferential cut can be employed. For
the small karst caves, tamping clayed soil is suitable; while for the big ones, the
backfilled soil should be paved by stones and covered using C20 concrete. When newly
observed karst caves exist in the surface binding terrain, tree-like ditches ought to be
arranged above the tunnel.
Backfilling the karst caves in the surrounding rock
The quick setting cement slurry is used to fill the karst caves behind the linings
within 5 m distance under the situation that void pace occurs due to soil erosion by
groundwater.
Draining treatments
Draining treatments along with buried pipes in cutting groove surrounding the
construction joints and draining holes arranged at the side wall corners are used in the
tunnel.
ASCE
97
EFFECT OF TREATMENT
Through the treatment to leakage in Pishuangao tunnel of Beijing-Zhuhai expressway
during 2011~2014, the situation of water leakage is significantly improved. In the rainy
season in 2015, no dramatic leakage was observed and only slight leakage occurred in
some wall corners. This slight leakage has no impact on the normal service of the tunnel.
In light of the significant improvement of the leakage in the tunnel, the technique used in
the present study is highly recommended.
CONCLUSIONS
Under long-term groundwater table, after a tunnel service of many years, the
infiltration path of ground water changes, which will lead to change in the recharge of
groundwater in the region. These changes may cause the leakage in the tunnel. Since
some changes are a progressive procedure, and some ones are a transient process, the
monitoring and regular inspection to ground surface and draining conditions during the
tunnel maintaining is necessary.
After the leakage happens, a fine hydro-geological and engineering geological
surveying should be first performed. And then, the connectivity test can be conducted in
the zone through which groundwater is collectively flowing into the tunnel. After the
connection between surface and ground water, the blocking, cutting-off and draining
treatment can be used in this zone.
The karst dolines should be investigated behind the tunnel linings using High density
resistivity method and seismic imaging technique.
The prevention to leakage in the tunnel should be implemented by considering
ground surface-surround rock-lining as an integration. Backfilling the subsidence and
ground fractures, and draining on the ground are the key measures for the ground surface.
For the surrounding rock, backfilling the gas behind the lining is the key method.
Reinforcement to the structure and consolidated draining groundwater are suitable for
the lining. Treatment to the ground surface, surrounding rock and lining is subsequently
taken in practice.
The potential impacts on the tunnel maintaining must be considered in the design and
construction plan. In the area with complex geological conditions, multiple geophysical
and geochemical methods are necessary and should be employed to investigate the
engineering geological and hydro-geological conditions.
ASCE
98
ACKNOWLEDGMENTS
This paper is supported by Transportation science and technology fund of Guangdong
province.
REFERENCES
Ita. (1991). Report on the damaging effects of water on tunnels during their working
life. Tunnelling and Underground Space Technology, Vol. 6(1): 11-76.
Li, G., Du, X., Zeng, Y.W. (2008). Analysis on the interaction between tunnel and
groundwater. China Water Transport, Vol. 9(8): 263-265.
Liu, H.J., Xia, C.C., Zhu, H.H., Luo, X. (2007) Recent research and development of
tunnel disasters. Chinese Journal of Underground Space and Engineering, 5:
948-953.
Jiang, Z.X. (2005). Interaction between tunneling and groundwater. Chinese Journal
of Rock Mechanics and Engineering, Vol. 24(1): 121-127.
Zhang, L.C. (2003). Study on the technique of water draining in highway tunnels.
Chinese Journal of Technology of Highway and Transport, 4: 68-69.
Zhao, S.Y. (2008). A study and treatment to the leakage in mountain tunnels. Master
thesis, Chongqing University.
Zhu, H.Q., Li, H.Y., Du, C.S. (2008). A study on the comprehensive method for water
leakage in the tunnel. Chinese Railways, 2: 42-45.
ASCE
99
A Case Study for Escape Route Optimization after Water Inrush in a Backward
Excavated Karst Tunnel
S. C. Li1; J. Wu2; Z. H. Xu3; L. P. Li4; X. Huang5; Y. G. Xue6; and Z. C. Wang7
1
Professor, Geotechnical and Structural Engineering Research Center, Shandong Univ., Jinan,
Shandong 250061, China (corresponding author). E-mail: lishucai@sdu.edu.cn
2
Ph.D. Student, Geotechnical and Structural Engineering Research Center, Shandong Univ., Jinan,
Shandong 250061, China. E-mail: wujing9516@163.com
3
Lecturer, Geotechnical and Structural Engineering Research Center, Shandong Univ., Jinan,
Shandong 250061, China. E-mail: zhenhao_xu@sdu.edu.cn
4
Associate Professor, Geotechnical and Structural Engineering Research Center, Shandong Univ.,
Jinan, Shandong 250061, China. E-mail: yuliyangfan@163.com
5
Ph.D. Student, Geotechnical and Structural Engineering Research Center, Shandong Univ., Jinan,
Shandong 250061, China. E-mail: hx19891018@yeah.net
6
Professor, Geotechnical and Structural Engineering Research Center, Shandong Univ., Jinan,
Shandong 250061, China. E-mail: xieagle@sdu.edu.cn
7
Associate Professor, Geotechnical and Structural Engineering Research Center, Shandong Univ.,
Jinan, Shandong 250061, China. E-mail: wanz@sdu.edu.cn
Abstract: With the development and utilization of underground space in karst area,
water-inrush is becoming a major geological disaster in tunnel engineering. The
investigation for escape routes optimization after water inrush in process of karst
tunnel excavation is particularly important. In the present study, a case of water inrush
from the working face is investigated in a backward excavated karst tunnel. Numerical
simulation is carried out by using the FLUENT software, the variation laws of velocity
and pressure along with distance are analyzed and discussed by selecting a series of
probing lines in a plane. Then flow characteristics of water after inrushing during
double-line tunnel excavation are summarized. Finally, the optimized escape routes
are made. The results show that: (1) the water velocity nearby the tunnel side wall is
the minimum, while it is the maximum in the middle position. So people in tunnels
should first move to the tunnel side wall and then escape quickly when water inrush
happens. (2) The water pressure at the intersection area of the cross passage and
tunnels is large. So people must not stay at the intersection area of the cross passage
and tunnels when water inrush happens. (3) When water inrush happens in the left
tunnel, if people meet the cross passage during escaping, they should pass through it
rapidly, turn to the right tunnel and run to the entrance. (4) If there is not enough time
to escape, people can run to the trolley and other equipment in the vicinity of the right
tunnel working face. In addition, some rescuing equipment can be set up at the high
ASCE
100
location of the cross passage. (5) Even if large flow water inrush occurs in the tunnel,
people can escape safely as long as the disaster is observed in time and the escape
routes are optimized scientifically and rationally. The results are of important practical
significance and engineering value to ensure the safety of tunnel construction.
INTRODUCTION
In recent years, China has increased investment in infrastructure construction, and
many tunnelling projects have been put on the agenda continually. The tunnel
engineering as a vital link of transportation seems to be more important. As everyone
knows, the excavation of underground and tunnel engineering is a very complex
process, and geological condition in mountains is also difficult to control, thus, any
careless will lead to major disasters, such as landslides, collapses, water inrush, etc.,
which causes heavy losses to the national economy and personal safety. Especially in
karst tunnels, the loss is more immeasurable due to the complexity of caves
development (Wang, 2004; Ivars, 2006; Qian, 2012). A large number of deep and long
tunnels in karst areas put Chinese tunnel construction scale onto a new climax. But
water inrush disaster is encountered by the probability of 80 percent during tunnel
construction in karst areas, and the downtime caused by water inrush disaster accounts
for 30 percent of the total construction period. Water inrush has become the most
commonly encountered geological disaster with great dangers during construction
(Zhang, 2005; Wang et al., 2006; Li et al., 2009; Liu et al., 2010; Wang and Wang,
2011; Wang et al., 2012). According to railway tunnel statistics since 1988, water
inflow of 31 tunnels has over l104m3/d. More than 1300 tunnels encounter water
seepage in operation, and the tunnels of encountering leakage problem account for 30
percent of the total tunnels.
It can be said, water inrush brings major disasters and incalculable economic losses
for construction safety of karst tunnels (Xu and Huang 2000; Jung-Woo Cho et al.,
2013; Xu et al., 2014; Li et al., 2014). Now researches often emphasis on how to use
comprehensive advanced prediction technique to carry out forecast and early warning
of geological disasters in karst tunnel construction (Qu et al. 2006; Wang et al. 2007;
Li et al. 2008; Shi et al. 2014). However, the study on flow characteristics of water
after inrushing during double-line tunnel excavation is often ignored. It is also seldom
related to the specific work on formulating optimized escape routes. Therefore, in
order to ensure the construction safety of high-risk karst tunnel and reduce disaster
losses caused by water-inrush, the investigation for escape routes optimization after
water inrush in process of karst tunnel excavation has important theoretical
significance and great practical engineering value.
In the present study, one case study of water inrush from the working face is
simulated in a backward excavated karst tunnel by using FLUENT software. In order
to show the change rules of velocity and pressure in tunnels more clearly, a series of
probing lines are selected in a plane, and the variation rules of velocity and pressure
along with distance are analyzed and discussed. Finally, the optimized escape routes
are formulated. Combined with this case study, the flow characteristics of water after
inrushing during double-line tunnel excavation are concluded and summarized, which
has important guiding significance for making scientific and rational escape routes and
rescue routes.
ASCE
101
102
NUMERICAL
L SIMULAT
TION
One
O case stu
udy of water inrush is sim
mulated in a backward exxcavated karrst tunnel.
In
I process off the left tun
nnel excavattion, the tunnneling cannot be carriedd out due too
the effects of geological
g
co
onditions an
nd other facctors. At thiss time, the right tunnell
shou
uld first be excavated, and
a then botth ends of thhe left tunneel are furtheer excavatedd
afteer passing thrrough the cross passage..
During
D
the leeft tunnel ex
xcavation, water
w
inrush occurs in thhe working fface close too
the left tunnel entrance,
e
i.ee., the workiing face closse to the lefft tunnel enttrance is thee
inleet of water inrush
i
(veloccity, 2 m/s),, and the rigght tunnel eentrance is thhe outlet off
watter inrush (prressure, 0 Paa). Boundary
y conditions are set as foollows, the w
working facee
closse to the leeft tunnel en
ntrance is velocity
v
inleet boundaryy, and the rright tunnell
entrrance is pressure outlet boundary.
b
Ana
alysis and Discussion
D
In
I order to sh
how the chan
nge rules off velocity andd pressure inn tunnels moore clearly, a
series of probin
ng lines are selected.
s
Thee details of thhe probing lines are show
wn in Fig. 1.
Thee probing lin
nes are researrched and an
nalyzed.
(a)
(b)
(c)
(d)
FIG.. 1 The locattion of the p
probing linees
Thee Left Tunnell
As
A for the fiive probing lines in the left tunnel (Fig. 1a), thheir velocityy curves andd
presssure curves are shown in
i Fig. 2.
ASCE
103
5
X=-14,Y=2
Velocity (m/s)
X=-16,Y=2
X=-18,Y=2
X=-20,Y=2
X=-22,Y=2
0
0
20
40
60
80
Distance (m)
(a) Velocity curves
194
X=-14,Y=2
190.8
Pressure (Pa)
100
X=-16,Y=2
187.6
X=-18,Y=2
184.4
X=-20,Y=2
181.2
X=-22,Y=2
178
0
20
40
60
80
100
Distance (m)
(b) Pressure curves
FIG. 2 The velocity and pressure curves of the left tunnel
1) On the whole, the velocity of five probing lines shows the tendency of
decreasing, and it is 2 m/s in the position of working face with water inrush (Z = 100),
while it reduces to 0 m/s in the position of the other working face (Z = 0). To be more
exact, when the variation range of Z is from 60 to 100, the velocity basically keeps
unchanged at 2 m/s. When the variation range of Z is from 40 to 60, the velocity
decreases sharply. When the variation range of Z is from 0 to 40, the velocity reduces
to 0 m/s. In addition, the velocity close to one side of the cross passage is the
maximum, and then it decreases gradually toward the other side, i.e., the velocity is
the maximum when X = -14, while it is the minimum when X = -22.
2) In general, the pressure of five probing lines shows the tendency of increasing.
When the variation range of Z is from 60 to 100, the pressure basically keeps
unchanged at 191 Pa. When the variation range of Z is from 40 to 60, the pressure
close to one side of the cross passage is the minimum, and then it increases gradually
toward the other side, i.e., the pressure is the minimum when X = -14, while it is the
maximum when X = -22. When the variation range of Z is from 0 to 40, the pressure
keeps unchanged at 193 Pa.
The Cross Passage
As for the five probing lines in the cross passage (Fig. 1b), their velocity curves and
ASCE
104
pressure curves are also drawn. The analysis process is the same as the left tunnel and
the main conclusions are as follows:
1) The velocity in the center of the section is the maximum, and then it decreases
gradually toward both sides of the section, i.e., the velocity is the maximum when Z =
50, while it is the minimum when Z = 51.6.
2) On the whole, the pressure decreases gradually with the increase of X, and it is
the maximum at the starting point of the cross passage.
The Right Tunnel
As for the five probing lines in the right tunnel (Fig. 1c), their velocity curves and
pressure curves are also drawn. The analysis process is the same as the left tunnel and
the main conclusions are as follows:
1) Among the five selected probing lines, the velocity is the minimum when X = 20
and the variation range of Z is from 60 to 100.
2) On the whole, the pressure changes greatly in the vicinity of the cross passage,
while it changes gently in other areas.
The Y-direction
As for the five probing broken lines along the height direction (Y-direction) of
tunnel centerline (Fig. 1d), their velocity and pressure curves are shown in Fig. 3.
15
Y=0.4
Velocity (m/s)
12
Y=1.2
Y=2
Y=2.8
Y=3.6
0
0
28
200
56
84
Distance (m)
(a) Velocity curves
112
140
Y=0.4
Pressure (Pa)
155
Y=1.2
110
Y=2
65
Y=2.8
20
Y=3.6
-25
0
ASCE
28
56
84
112
140
Distance (m)
(b) Pressure curves
FIG. 3 The velocity and pressure curves of Y-direction
105
The
T analysiss process is the
t same as the left tunnnel and the m
main concluusions are ass
follows:
1)
1 In the left tunnel, th
he velocity changes moore gently, aand the velocity is thee
min
nimum when
n Y = 0.4. In the processs of flowinng from the left tunnel to the crosss
passsage, the veelocity increeases obviou
usly. In the cross passaage, the veloocity in thee
centter of the secction is the maximum,
m
and
a then it deecreases graadually towarrd the upperr
and
d lower, i.e., the velocity
y is the max
ximum wheen Y = 2, w
while it is thhe minimum
m
wheen Y = 0.4 or 3.6. In th
he process of
o flowing ffrom the crooss passage to the rightt
tunn
nel, the velo
ocity decreasses rapidly. In
I the right ttunnel, the vvelocity channges gently,
and
d the velocity
y is lower When
W
Y = 0.4 or 3.6.
2)
2 In the left tunnel, th
he pressure basically keeeps unchannged at 1922 Pa. In thee
proccess of flow
wing from th
he left tunn
nel to the crross passagee, the presssure reducess
quicckly. Then, in
i the processs of flowing
g from the ccross passagee to the righht tunnel, thee
presssure contin
nues to reducce. On the whole,
w
the ppressure in tthe vicinity of workingg
facee with water inrush is the maximu
um, and the n it decreasses graduallyy along thee
direection of water inrush. In addition, the
t pressuree curves of ffive probing lines are inn
goo
od agreemen
nt, which sho
ows that thee pressure vvalues along the length direction off
tunn
nel are basiccally fixed in
n the height direction
d
of ttunnel (Y-diirection).
Thee Optimized
d Escape Ro
outes
(1)
( Accordin
ng to the results of velo
ocity curvess, the velociity in the ceenter of thee
secttion is the maximum,
m
while
w
it at boundaries
b
iis the minim
mum. Thereffore, peoplee
shou
uld first move to the tunnel side wall and thenn escape quicckly when w
water inrushh
happens.
(2)
( Accordin
ng to the results of pressure
p
cuurves, the ppressure chaange at thee
inteersection areea of the cro
oss passage and tunnels is large. Soo, people m
must not stayy
heree for a short time when escaping.
e
(3)
( Water inrrush occurs in
i the workin
ng face close to the left tunnel entraance. Duringg
the escaping, peeople in the left tunnel should
s
pass through crooss passage rrapidly, turnn
to th
he right tunn
nel without water
w
inrush and run to th
the entrance..
Combined
C
with
w this casee, the optimizzed escape rroutes are shhown in Fig. 4.
FIG
G. 4 The op
ptimized esccape routes
ASCE
A comparison between the simulated results and the field measurements should
have been done thereby verifying the validity of numerical model, but it is impossible
to achieve such the field measurements, and the reasons are as follows: On the one
hand, water inrush often occurs suddenly during tunnel excavation, and it is hard to
carry out field measurements timely in process of water inrush. On the other hand,
when water inrush happens in a tunnel, the tunnel constructors are eager to escape
with their lives, so there is no other time to consider the field measurements. Therefore,
this study emphasizes on the optimized escape routes when water inrush happens,
which purpose is to reduce economic losses and casualties, and ensure the safety of
tunnel construction.
CONCLUSIONS
In the present study, Qiyueshan high risk karst tunnel is taken as research
background. Numerical simulation is carried out by using the FLUENT software. One
case of water inrush from the working face is investigated in a backward excavated
karst tunnel. Combined with this case study, the water characteristics of water after
inrushing in process of karst tunnel excavation are concluded and summarized.
(1) The water velocity nearby the tunnel side wall is the minimum, while it is the
maximum in the middle position. Therefore, people should first move to the tunnel
side wall and then escape quickly when water inrush happens.
(2) The water pressure at the intersection area of the cross passage and tunnels is
large. Thus, people must not stay here for a short while when escaping.
(3) When water inrush occurs in the left tunnel, if people meet the cross passage
during escaping, they should pass through it rapidly, turn to the right tunnel without
water inrush and run to the entrance.
(4) When water inrush occurs in the left tunnel, the water velocity and pressure
nearby the working face of the right tunnel are small. Thus, if there is not enough time
to escape, people can run to the trolley and other equipment in the vicinity of the right
tunnel working face.
(5) When water inrush occurs in the left tunnel, the velocity and pressure at the
high location of the cross passage are both small, so some rescuing equipment can be
set up there.
ACKNOWLEDGMENTS
We would like to acknowledge the financial support from the National Basic Research
Program of China (973 Program, No.: 2013CB036000), the National Natural Science
Foundation of China (Grant No.: 51509147, 51479106), the promotive research fund
for excellent young and middle-aged scientists of Shandong Province (Grant No.:
2014GN028) and the China Postdoctoral Science Foundation (Grant No.:
2014M551908).
ASCE
106
REFERENCES
Ivars, D.M., (2006). "Water inflow into excavations in fractured rocka
three-dimensional hydro-mechanical numerical study." Int. J. Rock Mech. Min. Sci.
Vol. 43(5): 705-725.
Jung-Woo Cho, Seokwon Jeon, Ho-Young Jeong, Soo-Ho Chang, (2013). "Evaluation
of cutting efficiency during TBM disc cutter excavation within a Korean granitic
rock using linear-cutting-machine testing and photogrammetric measurement." Tunn.
Undergr. Sp. Tech. Vol. (35): 37-54.
Li, S.C., Xue, Y.G., Zhang, Q.S., Li, S.C., Li, L.P. Sun, K.G., Ge, Y.H., Su, M.X.,
Zhong, S.H., Li, X. (2008). "Key technology study on comprehensive prediction and
early-warning of geological hazards during construction in high-risk karst areas."
Chinese J. Rock Mech. Eng. Vol. 27(07): 1297-1307.
Li, L.C., Tang, C.A., Liang, Z.Z., Ma, T.H., Zhang, Y.B.(2009). "Numerical
Simulation on Water Inrush Process Due to Activation of Collapse Columns in Coal
Seam Floor." J. Min. Safe Eng. Vol. 2, 008.
Li, S. C., Xu, Z. H., Ma, G. W. (2014). "A Graph-theoretic Pipe Network Method for
water flow simulation in discrete fracture networks: GPNM." Tunn. Undergr. Sp.
Tech. Vol.42: 247-263.
Liu, H. L., Yang, T.H., Yu, Q.L., Chen, S.K., Wei, C.H.(2010). "Numerical analysis
on the process of water inrush from the floor of seam 12 in Fangezhuang coal
mine." Coal Geol. Explor. Vol. 38(3): 27-31.
Qian, Q.H. (2012). "Challenges faced by underground projects construction safety and
countermeasures." J. Rock Mech. Eng. Vol. 31(10): 1945-1956.
Qu, H.F., Liu, Z.G., Zhu, H.H. (2006). "Technique of synthetic geologic prediction
ahead in tunnel informational construction." Chinese J. Rock Mech. Eng. Vol. 25(6):
1246-1251.
Shi, L.Q., Qiu, M., Wei, W.X., Xu, D.J., Han, J. (2014). "Water inrush evaluation of
coal seam floor by integrating the water inrush coefficient and the information of
water abundance." Int. J. Rock Mech. Min. Sci. Tech. Vol. 24(5): 677-681.
Wang, J.H., Lu, C.C. (2007). "A semi-analytical method for analyzing the tunnel
water inflow." Tunn Undergr. Sp. Tech. Vol. 22(1): 39-46.
Wang, J.T., Wang, X.L. (2011). "Discussion on Water Inrush Coefficient Method
Applied to Predict Water Inrush Danger of Seam Floor Based on Gaojiata Mine as
Example." Coal Sci. Tech. Vol. 7: 027.
Wang, M.S. (2004). "Hydrogeological and geological forecast of tunnel construction
in the karst district." Railroad Survey. Vol. (1): 7-10.
Wang, Y., Yang, W., Li, M., Liu, X. (2012). "Risk assessment of floor water inrush in
coal mines based on secondary fuzzy comprehensive evaluation." Int. J. Rock Mech.
Min. Sci. Vol. 52: 50-55.
Wang, Y.J., Huang, P., Li, S.C. (2006). "Statistics of industrial accident in China from
January to February in 2006." J. Safe. Environ. Vol. 6(2): 138-141.
ASCE
107
Xu, Z. H., Ma, G. W., Li, S. C. (2014). "A Graph-theoretic Pipe Network Method for
water flow simulation in a porous medium: GPNM." Int. J. Heat Fluid Fl. Vol. 45:
81-97.
Xu, Z.M., Huang, R.Q. (2000). Deep and extra-long tunnel and geological hazards
during construction. Southwest Jiaotong University Press, Chengdu, China.
Zhang, J.C. (2005). "Investigations of water inrushes from aquifers under coal seams."
Int. J. Rock Mech. Min. Sci. Vol. 42(3): 350-360.
ASCE
108
109
Studies on the Characteristics of Traffic Accidents in Expressway Tunnels

Dingwen Zhong1; Ling Pan2; and Qingyan Tian
1
Engineer, Pavement Research Institute, Guangdong Hualu Transport Technology Co., Ltd., Guangzhou
510420, China. E-mail: 774905756@qq.com
2
Senior Engineer, Pavement Research Institute, Guangdong Hualu Transportation Science and
Technology Com. Ltd., Guangzhou 510032, China. E-mail: 45104117@qq.com
Abstract: With the rapid development of expressway construction, operating safety of

tunnels gradually become prominent issues. It is necessary to garner more knowledge
on characteristics of traffic accidents in expressway tunnels, which in favor of
improving the safety of tunnels and reducing traffic accident rate of tunnel sections.
Therefore, a study of traffic accidents from sixty-one tunnels of eleven expressways in
Guangdong Province between January 2012 and June 2015 has been conducted. Then
the distribution characteristics of traffic accidents are summarized respectively from
the aspects of accident types, types of accident vehicles, seasons, annual variation
trend, location, pavement type, skid resistance, alignment parameter, etc. The analysis
results are helpful to formulate a reasonable operation strategy for tunnel traffic safety.
INTRODUCTION
In recent years the number of expressway tunnels in Guangdong has continually
increased. Expressway tunnel is a special structure with long tube shape and closed
space. When accidents occur in tunnels they can have more severe consequences than
those on open road sections. Traffic accidents in the tunnels have the features of
complexity and diversify, great damage, easy to induce secondary traffic accident, etc.
Many scholars have attempted to study tunnel accidents. Investigating Norwegian
road tunnels, Amundsen and Ranes found that accident rates of the tunnel entrance
zone are higher and accident rates diminish as one proceeds inside the tunnel.
Salvisberg et al. investigated Swiss road tunnels and found that the probability of an
accident occurring in a tunnel is lower in longer tunnels than in shorter tunnels. In
addition, the probability increases with increasing the percentage of heavy vehicles or
the AADT. Zhang and Ma study accidents characteristics of Shaoguan tunnel group of
Jingzhu freeway, and recommend some measures to improve tunnel safety.
In view of this, it is worth making an investigation into accidents in Guangdong
expressway tunnels and understanding the characteristics of tunnel accidents.
ASCE
110
DATA BASIS
The analysis is based on sixty-one tunnels of eleven expressways in Guangdong
Province, and Table 1 gives the detail. During a 3.5-year period (from January 2012 to
June 2015) traffic accident data are collected. Due to the opening time, the accident
data of GLN expressway and GLS expressway are from September 2014 to June 2015,
the accident data of GZW expressway are from January 2013 to June 2015.
Table 1. Summary of Tunnel Sections Involved in Traffic Accident Survey
Expressway
Name
GLN
GLS
JZN
JZS
SSW
YG
SM
YW
GZW
GH
YY
Total Number
Tunnel Name Abbreviation

DYS1, DYS2, DYS3, CYT , CJL
ZSD , DLS , LJK , YZ1 , YZ2 , HS
PS , MZ , PSG , YP , WKB , WQ , SMG
BLS , DBS , KYS , WLI , DJS
HM , YDS
DLG , HP , HSH , SHT , LZS
TK , ZZL , SHH , YSX , LHS , LGD
DHS , SYSH , DD , WZS , CLD , QSD , EGJ , NCD
BT , LMH , SS , BM , NL , YH
SCK , MEL , NMK , GD , FK , WK , SHM , DJK , MTT
EBL , SYZ
Number of
Tunnels
5
6
7
5
2
5
6
8
6
9
2
61
CHARACTERISTICS OF TUNNEL TRAFFIC ACCIDENTS

Tunnel Accident Types
Figure 1 shows rear-end accident makes up the highest proportion 52% with 230
rear-end accidents annually. Other accident accounts second for 29%, which includes
fire accident, accident of hazardous chemicals, etc. Collision accident comes in third,
accounts for 15%. Scrape accident and rollover accident account for 2% respectively.
Types of Accident Vehicles
Figure 2 shows the car under seven seats have the maximum accidents, followed by
big truck and trailer. The proportion of types of accident vehicle in tunnel shows the
car under seven seats counts 55%, big truck and trailer counts 30%, medium bus and
large bus over seven seats counts 10%, small and medium-sized truck count 5%.
Annual Variation Trend of Tunnel Accidents
A study is made to ascertain the annual variation trend of tunnel accidents. The data
in 2015 is converted by first half year. Figure 3 shows Tunnel accidents in 2013 have a
smaller percentage of growth, while accident number of other years remains stable.
Season Distribution of Tunnel Accidents
Season distribution is analyzed from 2012 to 2014. Use the following season
division standard: spring (March to May), summer (June to August), autumn
ASCE
111
(September to November), and winter (December to following February). Figure 4

shows seasonal distribution ratio of tunnel accidents is spring 20%, summer 13%,
autumn 24%, winter 43%. Its highest in winter and lowest in summer.
FIG. 1. The annual average number and composition of accident types
FIG. 2. Accident numbers and composition of various vehicle types
FIG. 3. Annual variation trend of tunnel accidents
FIG. 4. Season distribution of expressway tunnel accidents
ASCE
112
Location Distribution of Tunnel Accidents

Zone Definition
A driver normally takes 4~9s to adapt the tunnel environment. In Guangdong the
speed limit is generally 80km/h in tunnel and 100~120km/h outside. Assuming 9s as
reaction time and 120km/h as driving speed, the braking distance is 300m. 300m is
chosen as tunnel transition length. The tunnel is divided into six zones along driving
direction (Figure 5): Zone 1 (I1), interior transition zone: first 300m into the tunnel
entrance; Zone 2 (O1), interior transition zone: first 300m into the tunnel exit; Zone 3
(N), interior zone: the remainder of tunnel; Zone 4 (I2), exterior transition zone: first
300m outside the tunnel entrance; Zone 5(O2), exterior transition zone: first 300m
outside tunnel exit; Zone 6 (I3,O3), open road sections: 300m to 1000m outside tunnel.
FIG. 5. Tunnel zone schematic diagram

Zone Distribution of Tunnel Accidents
The zone distribution of accident rates is calculated (Table 2). The average tunnel
accident rate is 2.62 per year per km. Zone 1 has highest accident rate, 3.65 per year
per km, Zone 4 is in the second, and Zone 6 is the lowest, 2.46 per year per km. It
shows that accidents are more likely to occur when entering the tunnel than exiting.
Table 2. Zone Distribution of Tunnel Accident Rates (Unit: per year per km)
Statistics
Range
61 Tunnels
Tunnel Accident Accident Accident Accident Accident Accident

Accident Rate of Rate of Rate of Rate of Rate of Rate of
Rate
Zone 1 Zone 2 Zone 3 Zone 4 Zone 5 Zone 6
2.62
3.65
2.56
2.68
2.92
2.64
2.46
Interior and Exterior Zone Distribution of Accidents

Zone 1, 2, 3 are divided into interior zone, and zone 4, 5, 6 are divided into exterior
zone. Table 3 shows the average accident rate of interior zone is 2.88 per km per year,
and that of exterior zone is 2.43 per km per year. The accident rate of the interior zone
is higher than exterior zone.
ASCE
113
Table 3. Accident Rates of Interior Zone and Exterior Zone

Interior
Interior Zone
Exterior
Zone
Accident Rate
Zone
Length(km) (per km per year) Length(km)
61 Tunnels
154.363
2.88
214.5295
Statistics
Range
Exterior Zone
Accident Rate
(per km per year)
2.43
Tunnel Type and Zone of Accidents

Expressway tunnel is divided into four types: Short tunnel (L500m), Medium
tunnel (500m<L<1000m), Long tunnels (1000mL3000m), Super long tunnel (L>
3000m). Figure 6 shows average annual accident ratio of different tunnel type. The
long tunnels are highest 57%. Medium tunnel and short tunnel come second and third,
respectively 20% and 16%, and super long tunnel is lowest. Presumably there's a
reason only four super long tunnels (DYS1 tunnel, CJL tunnel, SYSH tunnel, EGJ
tunnel) are surveyed. It shows longer tunnels trend to be more dangerous.
FIG. 6. Annual average traffic accident ratio of different tunnel type

Calculate accident rate of zone 1~6 in different tunnel types. Figure 7 shows all
tunnel types have relatively high accident rate for zone 1, zone 2, zone 4, zone 5. The
accident of zone 3 in the long tunnels and super long tunnel is higher than the medium
tunnel. For long tunnel, zone 1 has the highest accident rate, up to 5.4 per km per year.
For super long tunnel, the highest accident rate is zone5, 4.54 per km per year.
FIG. 7. Accident rate of zone 1~6 in different tunnel type
ASCE
114
The Impact on Accidents Caused by Pavement Conditions

The Factor of Pavement Type
Table 4 shows the annual average accident rate of tunnel concrete pavement is 3.2
per km, and that of asphalt pavement is 1.9 per km. The annual accident rate of
concrete pavement is higher than the asphalt pavement.
Table 4. Annual Average Accident Rate for Different Tunnel Pavement Type
Statistics
Range
61 Tunnels
Tunnel
Concrete
Pavement
Length
(km)
119.0
Annual Average
Accident Rate of
Tunnel Concrete
Pavement (per
km)
3.2
Tunnel
Asphalt
Pavement
Length
(km)
35.3
Annual Average
Accident Rate of
Tunnel Asphalt
Pavement
(per km)
1.9
The Factor of Skid Resistance

Combining with skid resistance performance data and the tunnel accidents data,
build the relationship between tunnel pavement skid resistance level and the average
accident rate. One direction of a tunnel is viewed as one unidirectional tunnel. With
the increase of skid resistance level the tunnel accident rate decreases. See Table 5.
Table 5. Skid Resistance Level and the Average Accident Rate
NO.
1
2
3
4
Skid
Resistance
Level
Defective
Average
Good
Excellent
Unidirectional Average Accident Rate

(per km per year per 10
Tunnel
Number
million vehicles
>27.5, 33.5
13
13.52
>33.5, 40
13
5.85
>40, 48
64
3.96
>48
30
3.69
The Scope
of SFC
The Impact on Accidents Caused by Road Alignment Parameter

The Road Longitudinal Slope Factor
Build the relationship between expressway tunnel longitudinal slope (uphill grade,
downhill grade) and the accident rates. Figure 8 shows the accident rate of uphill grade
sections is 2.54 per km per year, and that of downhill grade sections is 3.70 per km per
year. The accident rate of the downhill grade is 45.7% higher than the uphill grade.
The Road Curved Shape Factors
Build the relationship between tunnel curved shape (line and curve) and the
accident rates. Overall, the accident rate of line sections is 7.76 per km per year, and
the accident rate of curve sections is 2.49 per km per year. The accident rate of the line
sections is 3.1 times higher than the curve sections. See Table 6.
ASCE
115
FIG. 8. Expressway tunnel longitudinal slope and the accident rates

Table 6. Expressway Tunnel Curved Shape and the Accident Rates
Statistics
Range
Line
Section
Length
(m)
Accident rate of
Line Sections
(per km per
year)
Curve Section
Length (m)
Accident Rate of
Curve Sections(per
km per year)
61 Tunnels
20374
7.76
93184
2.49
CONCLUSIONS
The study aims to investigate the characteristics of traffic accident in expressway
tunnels, using the Guangdong province context. Study of traffic accidents from sixtyone tunnels of eleven expressways between January 2012 and June 2015 has been
conducted. Statistical analysis is carried out from the aspects of accident types, types
of accident vehicles, annual variation trend. Then aiming at the accident factors of the
tunnel environment in people-car-tunnel system, distribution characteristics of traffic
accidents are summarized from the aspects of season, location, pavement type, skid
resistance, alignment parameter, etc. The main conclusions are as follows:
(1) The overall situation of expressway tunnel accidents in Guangdong province
Tunnel accidents in 2013 have a smaller percentage of growth, while it remains
stable in the other years. In order of most occupancy for accidents types, it goes: rearend accident 52%, other accident 29%, collision accident 15%, scrape accident 2%,
rollover accident 2%. The proportion of accident vehicle types in the tunnel shows:
the car under seven seats counts 55%, big truck and trailer counts 30%, medium bus
and large bus over seven seats counts 10%, small and medium-sized truck count 5%.
(2) Season distribution characteristic of tunnel accidents
The number of tunnel traffic accidents is highest in winter, accidents of spring and
autumn fairly equal to each other, and summer is lowest. Seasonal distribution ratio of
tunnel accidents is spring 20%, summer 13%, autumn 24%, winter 43%. The reason
for winter accidents are higher than other seasons is mainly due to the significant
change of traffic flow in winter impacted by the Spring Festival holidays, etc.
ASCE
(3) Zone distribution characteristic of tunnel accidents

The average tunnel accident rate is 2.62 per year per km. Zone 1 has the highest
accident rate, 3.65 per year per km, being mostly attributed to multivehicle crashes
and "black hole effect", Zone 4 is in the second, its accident rate is 2.92 per year per
km, the accident rate of Zone 6 is lowest, 2.46 per year per km. It shows that accidents
are more likely to occur when entering tunnel than exiting. The average accident rate
of interior zone is 2.88 per km per year, and that of exterior zone is 2.43 per km per
year. The accident rate of the interior zone is higher than that of exterior zone.
(4) Traffic accident features of different tunnel types
The long tunnels have the highest accident rate 57%, medium tunnel and short
tunnel are respectively 20% and 16%, and super long tunnel is lowest 7%. There's a
reason that only four super long tunnels are surveyed and the sample size is small. The
longer tunnels trend to be more dangerous. All tunnel types have relatively high
accident rate for zone 1, zone 2, zone 4 and zone 5. The accident of zone 3 in the long
tunnels and super long tunnel is higher than in medium tunnel. For long tunnel, zone 1
has the highest accident rate. For super long tunnel, the highest accident rate is zone 5.
(5) Tunnel accident rate of different pavement type
The ratio between the tunnel cement pavement length and the asphalt pavement
length is 3.37:1. The annual average accident rate of tunnel concrete pavement is 3.2
per km, and that of asphalt pavement is 1.9 per km. It can be concluded that the annual
accident rate of concrete pavement is significantly higher than the asphalt pavement.
(6) The relationship between tunnel pavement skid resistance level and accidents
There is a negative correlation between skid resistance value and the accident rate.
(7) The impact on accident rate caused by road alignment parameter
The accident rate of uphill grade sections is 2.54 per km per year, and that of
downhill grade sections is 3.70 per km per year. The accident rate of line sections is
7.76 per km per year, and that of curve sections is 2.49 per km per year.
ACKNOWLEDGMENTS
The authors would like to thank the corresponding highway management department
for providing information relating to the investigated tunnels.
REFERENCES
Amundsen, F.H. and Ranes, G. (2000). "Studies on traffic accidents in Norwegian
road tunnels." Tunnelling & Underground Space Technology, Vol. 15(1): 3-11.
Salvisberg, U., Allenbach, R., Cavegn, M., Hubacher, M. and Siegrist, S. (2004).
"Verkehrssicherheit in Autobahn und Autostrassentunneln des
Nationalstrassennetzes." BFU- Report, Bern. ISBN 3-908192-17-X
Zhang, S.R., Ma, Z.L. (2006). "Traffic accident characteristics in freeway tunnels."
Proceedings of 5th International Conference on Traffic and Transportation Studies
(ICTTS 2006): 771779.
ASCE
116
117
Earth Pressure in a Head Chamber and Grouting Issues for EPB Tunneling in Soft
Ground: A Case Study in Tianjin Area
1
Jianjun Zhou1 and Dongyuan Wang, M.ASCE2
Deputy Director, State Key Laboratory of Shield Machine and Boring Technology, Zhengzhou, China.
Professor, Dept. of Geotechnical Engineering, Southwest Jiaotong Univ., Chengdu, China. E-mail:
Dongyuan_wang@163.com
Abstract: Based on a shield tunneling project using Earth Pressure Balance (EPB)
construction method in Tianjin, China, this paper builds a database of real monitored
data which includes ground surface settlement at different phases of tunneling,
chamber earth pressure, grouting pressure and volumes as well as excess pitch
information. A model to determine the earth pressure in head chamber to maintain the
face stability is proposed and verified with monitored data. Moreover, grouting
volumes are analyzed and compared with different considerations.
INTRODUCTION
Underground transportation networks are strongly needed to relieve traffic
congestions in populous metropolis worldwide. Even though shield tunneling has
advantages in mitigating the risks of ground movements over open excavations and
thus for has been widely used, ground surface settlement caused by shield tunneling
still can lead to angular distortion, sometimes severe damages to roadway, buildings
and utilities etc (Burd et al. 2008). Factors affecting ground surface settlement caused
by tunneling include the types of tunneling shield parameters, geometry of tunnels,
geological conditions, load conditions above tunnel, as well as construction
parameters, including earth cabin pressure, volume of excavated earth, grouting
pressure and volumes, etc (Liu et al. 2003).
Tunneling parameters are important to mitigate the risks associated with ground
settlement induced by shield tunneling. Based on a shield tunneling project using EPB
construction method in Tianjin, this paper builds a database consisting of ground
surface settlement at different phases of tunneling, chamber earth pressure, grouting
pressure and volume as well as the excess pitch of tunneling etc. A model to determine
the earth pressure in head chamber to maintain the face stability is proposed. The
method is verified with the data collected in real-time monitoring program. Moreover,
grouting volumes are analyzed and compared with different considerations.
ASCE
PROJECT DESCRIPTION AND DATABASE BUILT for THIS STUDY

Tianjin area of China is known for its soft ground, which typically consists of
marine sediments of silty clay formed in 4th Quaternary Period. Tianjin Subway Line
6, consisting of a twin tunnel approximately located below the sidewalks of the
surface street with a distance of 15 m, is built in such soft ground. The depth is about
10 m below the ground surface. From ground surface to approximate 40 m depth,
typical geological conditions and the geotechnical properties corresponding to each
stratum of the project site are summarized in Table 1.
Table 1. Geotechnical Properties of Each Stratum
Real-time monitoring system was adopted for this project. With shield advancing,
tunneling parameters, shield status, etc., were transmitted wirelessly and shown on the
monitors in the clients monitoring center. In the same time, these data are stored in
continued formats of images in the system. In addition, a stringent monitoring
program was carried out every day so that ground surface settlement at different
phases of construction can be measured. Figure 1 shows the monitoring plan for
ground surface settlement.
Aiming at determining and verifying the earth pressure in the head chamber of Earth
Balance Boring (EPB) tunneling as well as studying the grouting issues, 8 monitoring
points above the centerline of the right tunnel and 7 monitoring points above the
centerline of the left tunnel (data from one point abandoned due to bad documentation)
as well as the corresponding cut-out images of ground surface settlement, earth
pressure in head chamber, grouting pressure, grouting volumes and the excess pitch
information were selected. Table 2 summarizes these collected data.
MODELING EARTH PRESSURE in HEAD CHAMBER
The EPB method consists of a cutting chamber located behind the cutterhead. The
earth pressure in head chamber maintained by the mucking system under pressure is
critical to make the cutting face stable. Tremendous contributions have been done for
the face stability analysis, e.g., Leca and Domieux (1990) analyzed the upper and
ASCE
118
119
lower bounds of face stability followed by Anagostous and Kovari (1996), Chambon
and Corte (1994) and Huang et al. (2013). However, this contribution focuses on
determining the earth pressure in head chamber practically to maintain the stability of
the cutting face instead of scholarly research.
FIG. 1. Monitoring plan for ground surface settlement.

Table 2. Data Collected from the Real-Time Monitoring System
No.
Points
Cover
Depth (m)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
DBC-Y21
DBC16-6
DBC-Y20
DBC15-6
DBC-Y19
DBC14-6
DBC13-6
DBCY-17
DBC16-3
DBC-Z20
DBC15-3
DBC-Z19
DBC13-3
DBCZ-17
DBCZ-16
9.948
9.883
9.782
9.761
9.695
9.711
9.644
9.62
9.886
9.715
9.789
9.636
9.344
9.356
9.393
Overburden on
Crown (kPa)
Max. Settlement
Prior to Shield
Arrival (mm)
189.01
187.78
185.86
185.46
184.21
184.51
183.24
182.78
187.83
184.59
185.99
183.08
177.54
177.76
178.47
-0.48
-1.13
-0.82
-1.09
-0.79
-0.55
-8.48
-8.83
-6.47
-2.33
-1.49
-0.50
-0.65
-0.54
-0.52
Max.
Max.
Max.Settlement
Settlement
Settlement
Prior to Grouting Immediately
when Shield
(mm)
after Grouting
Passing (mm)
(mm)
0.27
0.02
-0.06
-0.26
0.01
0.35
-5.19
-5.88
-2.45
0.19
-1.49
-0.50
1.02
1.32
1.39
-21.69
-9.85
-5.05
-3.09
0.00
-4.07
-30.19
-5.88
-23.40
-0.19
-1.49
-0.50
-0.72
1.00
-48.90
-0.75
13.71
39.07
20.01
6.95
5.04
-18.99
-5.42
-17.35
10.81
28.13
33.31
30.25
10.37
-37.17
Earth
Pressure in
Chamber
(kPa)
200.73
214.23
219.43
181.50
185.15
198.15
219.18
203.95
240.05
193.05
309.43
249.60
229.60
225.53
207.10
Grouting Grouting
Pressure Volume
(kPa)
(m3)
393.00
475.00
476.80
469.00
483.70
586.10
695.10
688.30
389.00
164.75
262.10
279.85
317.50
325.05
284.75
5.3
7.1
7.1
8
4.1
6.2
8
6.5
7.1
8
6
6.6
7.4
6.6
6.4
Maximum ground surface settlement data prior to shield arrival and when shield
passing indicate that the cutting face is fairly stable (refer to columns 5 and 6 in Table
2). Statistical analysis for the earth pressure in head chamber indicate that it may
follow a normal distribution as the cover depth of tunnel is flat and the soil strength,
which usually follows a normal distribution, contributes to the earth pressure in head
chamber. The mean and the standard deviation of earth pressure for this study are
=218.4 kPa and =31.7 kPa respectively. Moreover, back analysis of the data
ASCE
Excess
pitch,
(deg)
0.3
0.04
0.16
0.07
0.07
0.29
0.44
0.6
0.11
0.15
0.1
0.3
0.22
0.19
0.3
indicate that the earth pressure in head chamber, which can be automatically slightly
adjusted by the shield machine, are largely scattered between the lower and upper
bound (, +) of the monitored earth pressure in head chamber.
Assume the soil pressure from soil mass works on the open face as an external
square to the tunnel cross-section (i.e, the square with a side length d), the following
equation can be derived to estimate the earth pressure in head chamber based on the
assumption that an equilibrium state is reached at the open face:
2
(1)
overburden = H = (2C + d )(1 sin ' )
where C is the cover, measured from the ground surface to the crown of the tunnel; d
is the diameter of the tunnel; is the unit weight of the soils above the tunnel center,
and is the effective friction angle of soils. In practice, the average of unit weight
and effective friction angle weighted by layer thickness shall be used for
calculation.
Anagostous and Kovari (1996) assume the soil behind the cutting-head is a wedge,
which generates the pressure on the face supported by the earth pressure in head
chamber to reach an equilibrium state. They found at equilibrium, the effective
support pressure depends on the tunnel diameter d, on the overburden C, on the
piezometric head, on the elevation of the groundwater table h0, as well as the effective
shear strength parameters c and ', and on the submerged unit weight and dry unit
weight d. Based on Mohr-Coulomb yield criterion, they proposed a method to
estimate the minimum effective support pressure in soils:
h
' = F0 ' d F1c'+ F2 ' h F3c'
(2)
d
where h denotes the difference of water elevations, i.e., the vertical distance from
the groundwater level to the center of the tunnel. Fi (i=0, 1, 2 and 3) is dimensionless
coefficients that depends on the friction angle ', on the ratio of the cover over the
tunnel diameter C/d, (h0 is groundwater level), and on the ratio of dry to the
submerged unit weight. Based on Tianjins most subway tunnels, groundwater level is
2 m below ground surface, tunnel diameter d=6.2 m, effective friction angle of soil
=20 deg., the dimensionless factors F0=0.45, F1=3.5, F2 =0.64, and F3 =0.052 (Note,
Fi is based on =20. F0, F1 and F2 will decrease while F3 will increase with an
increase of ). Substituting these dimensionless factors to Equation (2) obtains
h
' = 0.45 ' d + 0.64 ' h 3.5c'0.052
(3)
d
Hydrostatic pressure to the tunnel open face shall be added to Equation (2) to obtain
the total pressure:
h
at crown
(0.45 ' d + 0.64 ' h 3.5c'0.052 d ) + C w
(4)
A K =
h
(0.45 ' d + 0.64 ' h 3.5c'0.052 ) + (C + d ) w at bottom
d
3
where w =10 kN/m .
Earth pressure calculated with Equation (4) was plotted in Figure 2. It is observed
that the range of lower and upper bound in Figure 2 is quite wide; also for
ASCE
120
convenience purpose in practice, modification to the estimation of earth pressure in

head chamber based on A-K method is as follows:
h
d
A K = (0.45 ' d + 0.64 ' h 3.5c'0.052 ) + (C + ) w
(5)
2
d
Calculated earth pressure in head chamber using Equations (1) and (5) are plotted in
Figure 3 respectively, which are all in total stress. It is noted that the results calculated
with the two equations are very close, especially the trends by the two methods are
stringently in line with each other. However, the A-K method produces a relatively
low estimation than the overburden method we proposed.
FIG. 2. Comparisons of monitored earth pressure with determined values using

the proposed model and A-K method
GROUTING ISSUES
The earth pressure in head chamber of earth balanced shield tunneling can support
the soils; five phases of settlement are classified according to shield location as shown
by Figure 3. Settlement occurred from the time that the ground was cut to that the tail
of the shield passed the monitoring points which are defined as Phase III contributes
more to the total ground surface settlement (Leca and New 2007) and for this specific
project it is up to 80% for DBC 16-3 monitoring point.
Lee et al. (1992) stated that the settlement caused by tunneling depends on (i) the
ground and groundwater conditions, (ii) the tunnel depth and diameter, and most
important (iii) the construction details. Moreover, they studied the reasons for that and
evaluated the gap parameters which cause such settlement.
A shield boring machine with an outside diameter D is needed to construct a tunnel
with an inside diameter d (d here actually is the outside diameter of the segmental
ring). The relation between the shied outside diameter and d is D = d + 2 + , where
2 + is termed as physical gap, and and is the thickness of tailpiece and
clearance respectively for segmental ring erection as illustrated by Fig. 4(a). Grouting
cementitious material to the shadow area shown in Figure 4(b) which models the gap
ASCE
121
122
when the shield tail passing the settlement points and erecting the lining is necessary
to prevent the collapse of soil.
FIG. 3. Five phases of ground surface movement caused by shield tunneling.

The grouting volume to fill the physical gap for each segmental ring is
VPhysica _ gap = W (d + + )(2 + )

2
Lee et al. (1992) studied the gap parameter g. They proposed
*
g = G p + u 3d
+
(6)
(7)
*
represents the equivalent 3D elastoplastic
where G p = 2 + . The term u3d
deformation at the tunnel face; it equals approximately zero for EPB tunnel; and the
term takes the quality of workmanship into account, and = L , here
represents "excess pitch" (i.e., an upwards or downwards angle of the shield when
advancing) and L represents the length of the shield, which are illustrated in Figure 5.
Lee et al . (1992) also stated that for EPB tunneling the workmanship parameter is
up to 6Gp.
In practice, cementious material is grouted to the gap via 4 or 6 holes evenly
arranged along the tunnel segmental ring. The grouting volume to fill the shadow area
taking workmanship into account for each segmental ring in Figure 5(b) is
L
Vgap _ fill = W (d + + +
)(2 + + L )
(8)
2
2
However, more grouting volume than the calculated with Equations (7) and (8) are
conducted to eliminate the excessive settlement. Grouting volumes calculated using
Equations (6) and (8) as well as the real monitored grouting volumes are plotted in
Figure 6.
ASCE
123
FIG. 4. Illustration of gap at tailpiece of shield
FIG. 5. Illustration of excess pitch (workmanship) caused gap

High grouting pressure has to be used to grout more cementitous materials to the
ground. Unfortunately the relationship between the grouting pressure and the grouting
volume is too complex to be identified using simplified Equations like Equations (7)
and (8). Monitored grouting volumes are used in this contribution to compare with
those calculated by using Equations (7) and (8).
FIG. 6. Monitored grouting volumes vs magnitudes with different considerations.

Figure 6 shows the comparisons of grouting volumes: (1) only considering the
physical gap; (ii), consider the physical gap and workmanship and (iii) consider the
high grouting pressure (Note, the project of interest uses earth pressure balanced shield
boring machine TM634-PMX with a thickness of 40 mm of tailpiece, a clearance for
ASCE
segmental ring erection of 30 mm and a length of the shield of 8.68m). Figure 6 shows
that the grouting volumes considering the workmanship are between the magnitudes
that to fill the physical gap and using the upper limit workmanship (i.e., =0.6Gp by
Lee et al. 1992). However, the high grouting pressure grouted much more
cementitious materials. When biding a project, the appropriate model shall be selected
to calculate the volumes.
To study the effect of the extra grouting, normalized ground heave caused by
grouting (i.e, the settlement immediately after grouting minus the max. settlement
prior to grouting/max. settlement prior to grouting) is studied. Statistical analysis for
the 15 points indicated that approximately 50% normalized extra grouting lead to 40%
of the maximum ground heave.
CLOSING REMARKS
A model to determine the earth pressure in head chamber is proposed and verified
with the database. In addition, comparison with accepted A-K method indicates it is
valid but much simplified. Different phases of settlement caused by shield tunneling
and the gap in shield tail and clearance of erection of lining are discussed. Grouting
volumes are compared; extra grouting volume can cause ground heave, and a general
ratio of extra grouting volume to the ratio of ground heave is analyzed. The result of
the work is applicable to other similar projects, and design method is easily replicated.
ACKNOWLEDGMENTS
The authors are grateful for the support of the open research fund (#2014-01)
provided by State Key Laboratory of Shield Machine and Tunneling Technology,
China.
REFERENCES
Anagostous, G, and Kovari, K. (1996). "Face stability conditions with earth-pressurebalanced shields." Tunn. Underg. Spa., 11(2):165-173.
Burd, H.J., Houlsby, G.T., Augarde, C.E. and Liu, G.(2000). "Modeling tunnelinginduced settlement of masonry buildings." Proc. ICE, Geotechnical Eng., 143(1): 1729.
Chambon, P. and Corte, J.F. (1994). "Shallow tunnels in cohesionless soil: stability of
tunnel face." J. Geotech. Engrg., 120:1148-1165.
Clough G.W. and Schmidt B. (1981). "Design and Performance of Excavations and
Tunnels in Soft Clay." Elsevier, Netherland, pp. 569-634.
Huang, M. S., Song, C.X. and Lu, X.L.(2013). "Upper bound analysis for stability of a
circular tunnel in heterogeneous clay." Chinese J. Geotech. Eng., 35(8):1504-1512.
Leca, E. and Domieux, L. (1990). "Upper and lower bound solutions for the face stability
of shallow circular tunnels in frictional materials." Geotechnique, 40(4):581-606.
Leca, E. and New, B. (2007). "Settlement Induced by Tunneling in Soft Ground." Tunn
Undergr. Sp. Tech., 22, 119-149.
Lee, K.M., Rowe, R.K, and Lo, K.Y. (1992). "Subsidence owing to tunneling. I:
Estimating the gap parameters." Canadian Geotechnical Journal, 29(6):929-940.
Liu, Z., Wang, M. S. and Dong, X. (2003). "Analysis of ground surface settlement
caused by shield tunneling." Chinese J. Rock Mech. Rock. Eng., 22(8): 1297-1301.
ASCE
124
125
The Effects of Relative Tunnel Depth and Volume Loss on Vertical Settlements
above Tunnels in Dense Sands
Andrea Franza1; Bo Zhou2; and Alec M. Marshall3
1
Ph.D. Student, Dept. of Civil Engineering, Univ. of Nottingham, Nottingham NG7 2RD, U.K. E-mail:
evxaf2@nottingham.ac.uk
2
Geotechnical Engineer, Shanghai Civil Engineering Co., Ltd. of China Railway Group Ltd., 278 JiangChang-San Rd., Shanghai 200436, China. E-mail: zhou_bo@aliyun.com
3
Lecturer, Dept. of Civil Engineering, Univ. of Nottingham, Nottingham NG7 2RD, U.K. E-mail:
alec.marshall@nottingham.ac.uk
Abstract: Past studies have provided data on the variation of settlement distribution
above relatively shallow tunnels in sands. There is, however, a lack of research that
considers the case of relatively deep tunnels. This paper compares the results of a set
of plane-strain centrifuge tests in dry dense sands. The cover-to-diameter ratio, C/D, of
the tunnels ranged between 1.3 and 6.3, thereby including relatively shallow and deep
tunnels. Ground movements are measured using an image-based measurement
technique. Gaussian curves are fitted to the settlement data in order to evaluate the
characteristics of the settlement profiles. An assessment of greenfield settlement
trough shape, both at the surface and subsurface, is carried out. The effect that relative
tunnel depth and volume loss has on the settlement trough shape is demonstrated and
discussed. The results indicate a non-linear trend of settlement trough shape with C/D,
which suggest a transition between shallow and deep tunnels within the investigated
C/D range. To account for highlighted results, new relationships are proposed to
estimate settlement trough shape parameters in dense sands.
INTRODUCTION
Current needs for infrastructure and services in urban areas often require the
construction of tunnels. Because of environmental and economic reasons, tunnels may
be excavated close to the surface or at greater depth. However, tunnel excavation
inevitably induces ground movements that affect existing surface and buried
structures. To preserve structural serviceability and safety, it is necessary to assess the
magnitude and distribution of tunnelling-induced ground movements. In particular, the
prediction of the vertical settlement trough shape is important: a narrow settlement
trough (high curvature) with large maximum settlement poses potential for damage to
structures. Ground movements due to tunnelling in clay have been widely monitored
ASCE
126
and discussed. Recently, several cases of tunnel construction in coarse-grained soils

have been documented and experimental investigations have been carried out.
However, there is a lack of understanding for relatively deep tunnels.
The magnitude of tunnelling-induced movements is generally described with the soil
volume loss, Vl,s, that is the ratio between the volume of the settlement trough per unit
length of tunnel, Vs, and the excavated area of the tunnel cross section, V0. To model
the ground loss due to tunnel excavation, which causes Vl,s, analytical and
experimental studies use the tunnel volume loss, Vl,t, which is the ratio between the
over-excavated ground at the tunnel periphery, V, and V0.
The shape of vertical settlement trough is generally predicted by tunnel engineers
using simple empirical methods. In clays, the surface and subsurface settlement trough
shape agrees with a standard Gaussian curve, with the maximum settlement, umax, and
the horizontal distance of the inflexion point, i, defining the shape of the curve.
x2
u z = u max exp
2i 2
(1)
The value of i was found to be proportional to the vertical distance between the tunnel
and the depth of interest through the width parameter K;
i = K ( zt z )
(2)
where the parameter K was defined as a function of the ratio between depth and tunnel
depth, z/zt (Mair et al. 1993).
K=
0.175 + 0.325(1 z / z t )
1 z / zt
(3)
Jones (2010) displayed that the Mair et al. (1993) expression would overestimate the
width parameter K in case of deep tunnels. Based on field data measurements, the
author proposed a logarithmic formula for the prediction of the width parameter
depending on the height above the tunnel zt z rather than the relative depth z/zt
K = 0.25 ln ( z t z ) + 1.234
(4)
Interestingly, Eq. (4) predicts a decrease of the surface width parameter with zt.
Therefore, the settlement trough shape in clays is a function of depth, z, and tunnel
depth, zt; on the other hand it should not depend on Vl,t.
In sandy soils, the use of the modified Gaussian curve rather than a standard
Gaussian curve was recommended by Vorster et al. (2005) to obtain a better fit to
observed settlements in sands induced by shallow tunnels.
u z = u max
(n 1) + exp[ (x 2 / i 2 )]
; n = e
2 1
2 + 1
+1
(5)
where is an additional parameter controlling the profile shape whereas umax and i are
defined as in the Gaussian curve (Eq. 1). In particular, the additional degree of
freedom, , allows for more effective curve fitting in case of narrow settlement
troughs. Several studies highlighted that, in sandy soils, the width parameter i
ASCE
127
increases with the cover to diameter ratio, C/D, decreases with the magnitude of
volume loss, Vl,t, and decreases with relative density, Id (Marshall et al. 2012;
Sugiyama et al. 1999; Zhou et al. 2014).
Despite its versatility, the use of the modified Gaussian curve is not as user-friendly
as the standard Gaussian curve and the parameter does not have a physical meaning.
To overcome these drawbacks, Marshall et al. (2012) suggested to characterise the
shape of modified Gaussian curves through x* and x** that are, respectively, the
horizontal distance between the tunnel centre line and the point where uz = 0.606 umax
and uz = 0.303 umax. x* and x** are related to the depth through x* =K*(zt - z) and
x** = K**(zt - z). For standard Gaussian curves x coincides with i, therefore the use of
x and x allows for a reasonable comparison of the trough width of standard and
modified Gaussian curves.
Marshall et al. (2012) proposed the following equations, based on the modified
Gaussian curve, to predict surface and subsurface settlement troughs in dense sands.
K s* + (x * / z )( z / z t )
K =
1 z / zt
*
%
* slope
K s* = K s*,int
Vlt + Vl ,t K s ,Vlt
**
K s** + (x ** / z )( z / z t )
=
1 z / zt
(6)
K s** = K s*,*Vltint + Vl %,t K s*,*Vltslope
However the regression of centrifuge data was performed assuming a linear variation
** slope
**int
of the trough shape parameters ( x* / z ; x** / z ; K s*,slope
; K s*,int
Vlt ; K s ,Vlt
Vlt ; K s ,Vlt ) with
tunnel volume loss and C/D ratio.
To investigate the effect of the cover-to-diameter ratios (C/D) for relatively deep
tunnels, this paper presents the outcomes of a plane-strain tunnelling centrifuge test
performed using a dry dense silica sand for a C/D of 6.3 (Id=90%). The measurements
of surface and sub-surface soil movements are compared with the centrifuge test data
from Marshall et al. (2012) and Zhou et al. (2014) performed with the same soil
condition for C/D of 1.3, 2.0, 2.4, and 4.4. Tests are labelled according to their C/D
ratio (i.e. test with C/D of 6.3 is referred to as CD6.3). The qualitative and quantitative
assessment of the results shows that, in dense sands, the variation of the vertical
settlement trough shape with C/D ratio should be represented by a non-linear trend. A
new set of equations is proposed to estimate the trough shape parameter of Equation
(6) in relation to C/D.
EXPERIMENTAL SET UP AND SCALE EFFECTS
The test was conducted at the University of Nottingham Centre for Geomechanics
geotechnical centrifuge. The centrifuge test of was performed in plain strain condition
at 80 g with a 40 mm diameter model tunnel buried at 270 mm depth (to tunnel axis).
The centrifuge model replicated, at prototype scale, the excavation of a 3.2 m diameter
tunnel with a 20 m cover.
A dry silica sand known as Leighton Buzzard Fraction E with d50 = 0.122 mm was
used for testing. The sand was poured to achieve a relative density, Id, of 90%. The
tunnel comprised a rubber membrane filled with water. It consisted of a concentric
cylinder with enlarged ends covered by a latex sleeve sealed with O-rings. The tunnel
was designed with a concentric cylinder rather than a downwards eccentric one
ASCE
because, due to the high relative depth, the tunnel is most likely to experience a radial
contraction. However, because the tunnel volume loss is due to water extraction, a
concentric cylinder allows for both a radial and an oval-shaped contraction depending
on the equilibrium condition between the soil and the model tunnel. A tunnel volume
control system composed by a constant-head standpipe, a solenoid valve, a linear
actuator, a water-filled sealed cylinder and an LVDT was used to 1) compensate,
during the centrifuge spin up, the volume loss at the tunnel due to the compressibility
of the air trapped within the system; 2) extract water from the tunnel during the test to
replicate the ground loss due to tunnel excavation and 3) measure the extracted
volume of the water, which is proportional to Vl,t. Further details regarding the setup
were provided by Zhou et al. (2014).
Three digital cameras were used to take pictures of the soil at the front Perspex wall
of the centrifuge container. During the test, digital photos were taken at 0.2% Vl,t
increments up to 10%. To measure surface and subsurface soil displacements, the
geoPIV image-based measurement technique was used (White et al., 2003). To
improve measurement precision, the field of view of the soil was limited to one side of
the tunnel with respect the centreline, both above and beneath the tunnel. For this test,
the precision was of the order of 0.01 mm.
Scale effects in centrifuge tests should always be evaluated. In case of tunnelling,
the main consideration is the ratio between the model tunnel diameter, D, and average
grain size, d50. Kutter et al. (1994) and Marshall (2009) suggested, respectively, that
grain size impact on the test results decrease when D/d50 > 350 and that scale effects
should be negligible for D/d50 > 500 in collapse condition. For the test presented in
this paper, the ration D/d50 is 327 and the investigated range of Vl,t is within precollapse conditions. Therefore, although grain size effects may be present, they should
have a minor influence on the results.
RESULTS
Comparison of deformation patterns due to shallow and deep tunnels
Fig.1 presents the normalized vertical displacement fields measured at Vl,t equal to 2
and 5% in tests CD2.4 and CD6.3. Vertical settlements and spatial coordinates were
normalized, respectively, by Vl,t R and D to allow comparing the displacement fields
induced, at different Vl,t, by tunnels of different diameters. In case of shallow tunnels,
displacements are localized at the tunnel centreline especially at higher relative depth
z/zt. On the other hand, ground movements are highly spread out in case of deep
tunnels resulting in wide settlement troughs without a zone of major settlements, even
at relative depth z/zt greater than zero. In both cases, soil movements measured below
the tunnel springline depth were negligible across the range of volume losses
considered. It was not possible to measure soil movements close to the tunnel
periphery due to the presence of a hole in the Perspex which was necessary to host the
enlarged tunnel end and a fitting ring.
The ground loss at the shallow tunnel CD2.4 resulted in a chimney-like
displacement field with soil movements that tend to propagate from the tunnel more
and more vertically with Vl,t. For the deep tunnel CD6.3, it was not possible to identify
the chimney-like mechanism that was suggested for coarse soils by Cording (1991).
ASCE
128
The soil arching phenomenon probably plays a major role in 1) the transition from a
chimney-like displacement field to a wide one with C/D and 2) the decrease of the
width parameter K with Vl,t. In general, ground movements due to tunnel ground loss
tend to propagate vertically within the arching zone inducing narrow settlement
troughs in this area. On the other hand, above the arch, ground movements propagate
towards the surface, spreading in the horizontal direction, resulting in wide settlement
troughs. In case of deep tunnels, the soil arching is localized at the tunnel crown and
the soil deformation pattern is wide overall. In case of shallow tunnels the arching
affects a proportionally larger zone of soil above the tunnel, resulting in a chimneylike displacement field (i.e. narrow settlement troughs). With the increase of Vl,t, the
arching failure induces a narrowing of the settlement troughs in both the zones above
and below the initial soil arch because additional movements due to tunnel ground loss
tend to propagate vertically from the tunnel towards the surface.
FIG. 1. Normalized vertical soil movements for test CD2.4 and CD6.3.
Fitting curves and variation of through shape with C/D
For the test CD6.3, Fig. 2 shows experimental data for a relative depth, z/zt, equal to
0 and 0.5; these data are curve-fitted with both Gaussian and modified Gaussian
curves. Settlement troughs are shown to become narrower with depth. The goodness
of fit is assessed with the coefficient of determination R2. The modified Gaussian
curve is able to provide a better fit to subsurface ground movements for test CD6.3.
However, as also noted by Marshall et al. (2012), standard Gaussian curves are
suitable for the interpolation of wide settlement troughs in sands, which are generally
measured at the surface in case of relatively deep tunnels. In the following sections,
ground movements are curve-fitted with modified Gaussian curves.
ASCE
129
A qualitative assessment of the influence of the tunnel relative depth on the

settlement trough shape is obtained by plotting the experimental settlement troughs of
tests CD2.4, CD4.4 and CD6.3 in Fig.3 for Vl,t = 2; 5 % and z/zt = 0.1; 0.6. These data,
normalised by the maximum settlement at the tunnel centreline, are fitted with
modified Gaussian curves. The higher the C/D, the higher is the surface and
subsurface settlement width. Although the level of scattering in the data of CD6.3 is
relatively high, it is particularly interesting that, in test CD6.3, the settlement curves at
z/zt = 0.6 are almost identical. Overall, Fig.3 suggests that the influence of C/D and Vl,t
on the settlement shape decreases with C/D magnitude, obtaining similar curves for
the cases CD4.4 and CD6.3. These results suggest that C/D effects on greenfield
ground movements may have a non-linear trend.
FIG. 2. Fitting curves to settlement data from CD6.3.
FIG. 3. Comparison of settlement trough curves for CD2.4, CD4.4 and CD6.3.
Formulas for the prediction of x* and x**
The settlement trough data of the five centrifuge tests were interpolated with
continuous modified Gaussian curves for Vl,t = 0.5 5 % ( 0.8 5 % for CD6.3).
Experimental values of K* and K**, which are a set of points corresponding to specific
values of C/D, Vl,t and z/zt, were obtained. For each centrifuge test, K* and K**
represent scattered points in the space {Vl,t; z/zt; K}. K* and K** were curve-fitted with
the surface described by Equation (6) obtaining scalar values of the trough shape
parameters, which are plotted vs. C/D in Fig. 4. The variation of the vertical
ASCE
130
131
settlement trough shape with C/D ratio has a non-linear trend, probably due to a
transition between shallow and deep tunnels. In clays, Jones (2010) has previously
described the non-linear trend of K by means of a logarithmic curve. Therefore, the
regression of the trough shape parameter coefficients with C/D was based on
logarithmic formulas. This allows for a realistic extrapolation of these parameters in
case of C/D higher than the investigated range. The equations of the interpolating
curves plotted in Fig. 4 are reported in Eq. (7). The prediction of Eq. (7) compared
well with the data used for their statistical regression, as shown in Fig. 5.
x * / z = 0.094 ln(C / D ) 0.378
x ** / z = 0.064 ln(C / D ) 0.712
K s*,slope
= 0.005 ln(C / D ) 0.040
Vlt
K s*,*Vltslope = 0.016 ln(C / D ) 0.070
K s*,int
Vlt
int
K s*,*Vlt
= 0.180 ln(C / D ) + 0.424
= 0.255 ln(C / D ) + 0.779
FIG. 4. Outcomes of settlement trough parameter regression.
FIG. 5. Trough width parameters against depth for CD2.4 and CD6.3.
ASCE
(7)
CONCLUSIONS
To investigate relatively shallow and deep tunnels, a greenfield tunnelling centrifuge
test in plane-strain condition and in dry dense sand was carried out for a C/D ratio
equal to 6.3. Results were compared with a set of similar centrifuge tests in the same
soil condition with C/D of the tunnels that ranged between 1.3 and 4.4. The following
conclusions can be drawn. 1) Modified Gaussian curves provided a better fit to the
settlement data compared to Gaussian curves. However, standard Gaussian curves
may be representative of surface settlements induced by deep tunnels (C/D > 4) in
dense sands. 2) The settlement trough width increased with the increase in C/D ratio
whereas it decreased with increase in Vl,t. However, the interpolation of settlement
data with modified Gaussian curves indicated a non-linear trend of settlement trough
shape parameters with C/D, which suggests a transition between shallow and deep
tunnels within the investigated range. This phenomenon may be due to the arching in
the soil above the tunnel. 3) A linear interpolation of the width parameters may lead to
an overestimation of the settlement width, which is not conservative for soil-structure
interaction analyses, in case of C/D higher than the investigated range. Therefore, a set
of equations with a logarithmic trend was proposed to estimate settlement trough
shape parameters in dense sands. This work may be extended to include the effect of
soil relative density.
REFERENCES
Cording, E. J. (1991). "Control of ground movements around tunnels in soil." In: Proc.
9th Pan-American Conf. on Soil Mech. and Found. Engng.: 2195-2244.
Jones, B. (2010). "Low-volume-loss tunnelling for London ring main extension." In:
Proc. Inst. Civil Eng. Geotech. Eng., Vol. 163 (3): 167-185.
Kutter, B. Chang, J. and Davis, B. (1994). "Collapse of cavities in sand and particle
size effects." In: Proc. 94 Int. Conf. on Centrif.: 809-815.
Marshall, A.M. (2009). "Tunnelling in sands and its effect on pipes and piles." Ph.D.
thesis, Cambridge Univ., Cambridge, U.K.
Marshall, A.M. Farrell, R.P. Klar, A. and Mair, R. (2012). "Tunnels in sands: the
effect of size, depth and volume loss on greenfield displacements. " Gotechnique,
Vol. 62 (5): 385-399.
Mair, R.J. Taylor, R.N. and Bracegirdle, A. (1993). "Subsurface settlement profiles
above tunnels in clay." Gotechnique, Vol. 43 (2): 315320.
Sugiyama, T. Hagiwara, T. Nomoto, T. Nomoto, M. Ano, Y. Mair, R.J. Soga, K.
(1999). "Observations of ground movements during tunnel construction by slurry
shield method at the Docklands Light Railway Lewisham Extension-East London."
Soil & Found, Vol. 39 (3): 99112.
Vorster, T.E.B. Klar, A. Soga, K. and Mair, R.J. 2005. "Estimating the Effects of
Tunneling on Existing Pipelines." J. Geotechnical & Geoenv. Engng., Vol. 131
(11): 13991410.
White, D.J. Take, W.A. and Bolton, M.D. (2003). "Soil deformation measurement
using particle image velocimetry (PIV) and photogrammetry." Gotechnique, Vol.
53 (7): 619-631.
Zhou, B. Marshall, A.M. and Yu, H.S. (2014). "Effect of relative density on
settlements above tunnels in sands." In: Proc. 2014 GeoShanghai Int. Congress:
Tunn. Undergr. Constr., 242 GSP: 96-105.
ASCE
132
133
Performance of a Long Pipe Roof for a Large Span Shallow Tunnel below an Existing Highway
Yu-Feng Shi1,2; Da-Xin Geng1; Chang-Jie Xu; Jun-Sheng Yang3; and Jin-Yang Fu3*
1
Lecture, School of Civil Engineering and Architecture Dept., East China Jiaotong Univ., 808 Shuanggang East Ave.,
Nanchang 330013, China. E-mail: s074811156@126.com
2
Postdoctoral Fellow, Jiangxi Provincial Institute of Water Science, 1038 Beijing East Rd., Nanchang 330029, China. E-mail:
s074811156@126.com
3
Lecture, Civil Engineering School of Central South Univ., 68 Shaoshan South Rd., Changsha 410075, China. E-mail:
jy.fu85@gmail.com
Abstract: The pipe roof reinforcement is usually used to stabilize the tunnel arch for tunneling in weak ground
section or passing below highway. In this study, large scale pipe-roof (85m long and 159 mm) was used to
enhance the Shitougang tunnel where a railway tunnel passes below the Hengyang-Kunming Highway in China.
The deformation of the pipe was measured by a series of strain gauges, and meanwhile, a three dimensional
numerical analysis was carried out to investigate the mechanical performance of the long and larger pipe roof. The
measured results are in good agreement with the numerical results. They both show that the stress distribution of
the pipe roof can be divided into three sections: tension section ahead of tunnel face, compressive section behind
the tunnel face and tension section near the entrance casing arch. The peak stress value occurs close to the
entrance casing arch and a certain distance ahead and behind the tunnel face, respectively. This analysis illustrates
the mechanical response of large and long pipe roof reinforcement used for a shallow tunnel passing below
highway.
Keywords: Tunnelling; Long and large pipe roof; Field test; Numerical analysis.
1 Introduction
The method of pipe roof is used extensively because of its adaptability, simplicity for installation and low cost [1-4].
According to the Chinese design code [5], the lengths of the pipe roof are generally between 10-45m and the
diameter of the pipe mostly ranges from 79 to 108mm.In general, the design of pipe roof is still largely depended
on experience and engineering analogy in a qualitative way. In recent years, engineers attempt to make progress
from qualitative design to quantitative design for pipe roof support for underground excavation. Field
measurements and numerical analysis are usually carried out to testify the mechanism of pipe roof reinforcement.
Most numerical analyses used equivalent formation model or beam element method to simulate the pipe roof
[6-12]
. These analyses mainly focus on the effects of pipe roofs in restraining ground surface settlement. It was
revealed that the pipe roof can redistribute the surrounding rock stresses and give an arching effect to the tunnel
face. Consequently, the pipe roof can improve the tunnel face stability and reduce the tunneling-induced ground
surface settlement significantly. The working mechanism of pipe roof has been investigated by Gou et al who
used the theory of elastic foundation beams based on field tests. The results showed that the jacked steel pipes
roof works as beams not only to support overburden loads but also to enhance the properties of surrounding rock
[13]
. Some other scholars improved the elastic foundation beams model, such as the model considering the variable
coefficient of subgrade reaction in front of the tunnel face [14]. However, the beam theories model could not reflect
the working mechanism of pipe roof perfectly because the model parameters are highly uncertain and the model
cannot account for the construction process.
Although pipe roof reinforcement has been used widely in practice, the pipe diameter and length varied for
different projects and were determined mainly on the basis of field experiences of the engineers. There were few
ASCE
134
field measurements on the effects of the steel pipe roof reinforcement, especially long and large diameter pipe
roof reinforcement. This study aims to investigate the mechanical performance of pipe roof, based on field
measurements of large scale pipe-roof (85m long and 159 mm) in the project of Shitougang tunnel where the
railway tunnel passes below the Hengyang-Kunming Highway in China. The deformations of the steel pipes were
observed by the installation of a series of strain gauges inside the pipe. A three-dimensional numerical analysis
was carried out to improve the understanding of the reinforcement mechanism of pipe roof, especially for long
and large span pipe roof.
2 Project overview
The Shitougang Tunnel crosses the Hengyang-Kunming highway at section DK214+975~ DK215+060 about 6~8
m beneath the surface. The angle between the tunnel center line and the highway center line is about 41 degree, as
shown in Fig.1. The width of roadbed of Heng-Kun highway is 26 m while the width of concrete pavement is
11.25 m 2. The tunnel excavation section is 14.68 m in width and 13.09 m in height. The tunnel was excavated
with its upper part in soft expansive soil layer while its lower part in limestone, as shown in Fig. 2 (a). The
expansive soil is made up of silty clay and turns out grayish yellow and in semi-hard state.
1
About 40 m
Elevation (m)
Hengyang-Kunming highway
186.7
178.9
Expansive soil
T unnelprofile
Limestone
DK215+060
DK214+975
165.8
Centerline Intersection position

between highway and tunnel
D K 215+ 021
85 m
Limestone
159 mm pipe roof

60 mm pipe roof
1m
Top
1
R=
6.4
Expansive soil
Pre-support of 159 mm & 60 mm pipe roof

The first layer primary lining (30 cm thick
C25 sprayed concrete & I25a Steel arch)
The second layer primary lining (25 cm thick
C25 sprayed concrete & I20b Steel arch)
159 mm pipe roof spacing 40cm
60 mm pipe roof
spacing 40cm
13.09 m 5.75 m
Roadbed
(a) Longitudinal tunnel section under the highway
Bench
1 Permanent
lining
0.4 m
0.4 m
Invert
14.68 m
(c) Enlarged view of

pipe roof
(b) 1-1 Cross-section
Fig.1 3D view of the tunnel and highway
Fig. 2 Longitudinal and transverse section of the Project
The tunnel was excavated by top heading method. The round length of top heading is 0.5m leaving
behind about 4-6 m core soil to support the tunnel face. The first layer reinforced shotcrete was placed
immediately after the excavation to support the upper part of the tunnel, as shown in Fig. 2(b). Before
excavating the lower part of the tunnel, the second layer reinforced shotcrete was erected in the upper
part. Then, the bench and invert were excavated with a round length of 1 m and 2 m, respectively, and
consequently the second layer reinforced shotcrete was placed to support the bench and invert and
closed as a shotcrete support loop. The specific support parameters and tunnel dimensions are shown in
Fig. 2.
To ensure the safety of Hengyang-Kunming highway and the tunnel construction, a long and larger
pipe roof was installed to pre-reinforce the soil mass above the tunnel arch. The steel pipes were 85m
long and with a diameter of 159 mm. Three reinforced steel bars with a diameter of 18 mm were inserted
ASCE
135
into each pipe to enhance the stiffness and strength of the pipe roof. The pipes were installed with a
spacing of 0.40 cm and about 0.18 m outside the excavation profile. There are some diameter 60 mm
middle pipes among the large pipes with a spacing of 0.4 m as well. However, Large cracks appear (as
shown in Fig. 4) during tunnel construction because of the difficulty for long pipe-roof construction
process control. It Powerful explanation of the pipe-roof necessity as design, or the pavement will suffer
more serious influence and even catastrophic collapse.
3 Field Measurements
3.1 Instrumentation
To investigate the mechanical performance of a long and large pipe roof, the deformation of the pipe was
measured through a series of strain gauges installed inside the pipe. Since the pipe diameter is large and very long,
only half length (43 m) of pipe was instrumented. A typical pipe (pipe A) at the tunnel crown was selected to be
instrumented. The layout of the installed strain gauges is shown in Fig.3 as well as their distribution along the
pipe A.
A1
6
A2
5
A3
4
A4
6
A5
4
A6
5
A8
A7
4
A9
4
(a) Layout of strain gauges

strain gauge
steel 50 steel tube
(b) Local detail drawing
159 pipe roof

50 steel tube
Strain gauge
18 steel
(c) Cross sectional layout of gauge

Fig.4 Pavement cracking caused by tunnel construction
Fig.3 layout of the instrumentation and details of the strain gauges
As shown in the Fig.3, A1-A9 represent the position of the nine strain gauges. These strain measuring points
were embedded in pipe A during the installation of the pipe in late March 2010. The tunnel excavation began in
late July and it passes through the highway in January 2011. The measurement was ended on May 22, 2011,
which has lasted about 10 months.
3.2 Monitored results

3.2.1 The measured strain history
Fig. 5 shows the development of measured strain of each monitoring point along the pipe A. It was found that
only point A1 results in positive strain (in tension), the rest of points all results in negative strain (in compression).
The measured strain was varied with the progress of the tunnel face. The strain at each measuring point has
decreased to some extent when they passed the tunnel face, while some points even decreased significantly except
point A1. When the tunnel face gets closer to the measuring point, the strain begins to grow faster and then reach
the maximum value when the tunnel face passed the point in a certain distance. The measured strain starts to
decrease when the primary shotcrete support was closed as a loop. Finally, the strain stabilized at 15-69% of the
maximum value.
Fig.6 shows measured strain of point A6 which is 30m away from the casing arch. When the tunnel
face arrives at A4 which is 9m away from A6, small tension strain occurs at A6. When the tunnel face
arrives at A5, the tension strain at point A6 reaches its maximum value about 35 . This strain
decreases gradually from tension strain to compression strain, and gets to -33 when the tunnel face
ASCE
136
arrives at A6. With the tunnel face advancing, the compression strain increases and undergoes a sharp
increase when the tunnel faces close to A8 which is about 10 m away from A6. The strain stays at the
value of -194 temporarily due to the construction interval. Then, this compression strain begins to
decrease with the restart of the excavation and finally stabilizes at -70 after another 20m excavation.
100
300
A3
A6
A7
A8
-100
tunnel face reach gauge A4
A5
-100
A9
-200
Date
tunnel face reach gauge A6
Strain()
Strian()
A4
100
A9
pipe roof
A2
tunnel face reach gauge A8,and

with construction interval 1 month
200
A6
A1
A1
tunnel
face reach
gauge A9
A6
Date
-200
-300
Fig.5 Measured strain curve of pipe A
Fig.6 Measured strain curve of point A6
Due to the different location, the values of the measured strain are different at each point. With the
progress of the tunnel face, the strain at A1 which is nearest to the entrance casing arch stays increasing
as tension strain and stabilizes at about 230. The strain of A2 and A3 varies slightly and finally
stabilizes at a small value (about 10). As all other points are relatively far away from the entrance
casing arch, they have undergone similar trend of strains as point A6.
3.2.2 The longitudinal strain distribution
During the tunnel construction, the pipe roof should resist a complex loading condition as the tunnel
must pass below the highway. The measured strain of pipe roof turns out similar to a bimodal curve
distribution, as shown in Fig.7. The waveform of the curve can be divided into a tensile region before
tunnel face, a compressive region behind tunnel face and a tensile region near the entrance casing arch.
These three regions are obvious when the excavation distance was about 35m, however, for excavation
distance larger than 40 m, the tensile region before tunnel face cannot be measured because of the lack
of monitoring points. The monitored result shows that the compressive region behind the tunnel face and
the tensile region near the entrance casing arch tended to transmit forward with progress of the tunnel
face. No obvious increase of the strain of the pipe roof was observed after tunnel face passed 40 m.
0
10
20
Distance to entrance (m)

30
40
50
60
70
80
20
Distance to entrance (m)

30
40
50
60
150
2010-11-19Excavation 30m
Excavation direction
Excavation
direction
Tunnel face
Strain ()
50
-50
50
-50
Tunnel
face
-150
-250
-150
(a)
(b)
Fig.7 Development of pipe strain with different tunnel face position
ASCE
70
250
150
Strain ()
10
Tunnel face
80
137
4 Numerical performance of the long pipe roof

4.1 Numerical model
A three-dimensional model was built to simulate the real landform and ground conditions, as showed in
Fig. 8. The pipe roof and surrounding rock were simulated with solid element while the primary
shotcrete support was modelled with so called liner elements. Time-dependent strength effect of the
sprayed concrete was considered by gradually increasing the strength parameters during the computation.
The strength of the shotcrete is assumed to increase linearly and stabilizes after six round length of
shotcrete supporting.
Fig. 8 3D meshed numerical model
Fig.9 Detail view of meshed model
Since the diameter of the pipe roof is very small compared to the excavation section size, very fine
mesh should be used for the pipe roof and the surrounding rock mass. This leads to a total number of
450 thousand solid elements. The model was first built with Midas and then imported into FLAC 3D for
computation. Fig.9 shows some details of the meshed model in 3D.
The upper surface boundary of the model was free, while the four sides were horizontally
constrained and the bottom surface was fixed. The ends of the pipes were rigid constrained because of
they are embedded with the entrance casing arch. The highway pavement was considered as different
material layers in the model. The tunnel excavation was simulated by top heading and then bench and
invert excavation, with a round length of 1m, 2m and 4m, respectively. The length of core soil during the
top heading was maintained to be 4-6m. The bench face was about 12-14m behind the top heading face,
and the inverted face was about 12m behind the bench face. Each step of removing the soil of round
length was calculated to obtain numerical equilibrium, and consequently support has been placed before
the next step of excavation. Such excavation process was equivalent to the real situation and was
simulated by writing FISH script in FLAC3D. Other construction effects such as blasting of the lower
hard rock were not considered in this research.
The behavior of the tunnel surrounding rock was modelled by the MohrCoulomb failure criterion
[15]
. The primary support and the pipe roof were modelled as linear elastic. The physical and mechanical
material parameters used in this calculation are shown in Table 1 [16]. The stiffness of the pipe roof was
calculated from the steel pipe and the filled concrete, while the reinforcement in the primary support was
equivalent to an increase of shotcrete stiffness.
ASCE
138
Table 1 Material properties

(kg/m3)
2400
2400
2000
2000
2600
2400
2400
2500
Material
The pavement concrete layer
The pavement basic layer
Soil layer
Expansive soil
Limestone
First layer primary support
Second layer primary support
Grouting pipe roof
E(GPa)
30
0.6
0.06
0.03
5
26.4
26.32
76.5
0.25
0.35
0.35
0.30
0.2
0.2
0.2
c(kPa)
10
20
18
600
-
()
30
40
20
55
-
H(m)
0.3
0.4
0.3
11.2
33.8
0.55
0.25
-
NoteDensity, EElastic Modulus, Poisson, cCohesion, Friction angle.
4.2 Numerical results

4.2.1 The stress of the pipe roof
Fig. 10 shows the longitudinal stress distribution of the selected pipe varies with the location of tunnel
face. The stress of the pipe also appears to be a bimodal curve distribution. The two peak values of the
stress increase and their position moves forward with the tunnel face advancing. After a certain distance
of tunnel excavation, the two peak values no longer increase but move forward with the excavation of
tunnel face. The pipe roof has undergone tensile stress ahead of the tunnel face. This tension length is
about twice the height of top heading excavation with the maximum tensile stress at the half length. The
pipe roof has undergone compression behind the tunnel face and the maximum value was at a distance
equal to the top heading height behind the tunnel face. Moreover, the pipe roof near the entrance has
undergone tension. The maximum tensile stress occurred close to the entrance casing arch.
2.0E+06
5.0E+06
tunnel face advance 10m
1.5E+06
3.0E+06
Stress (Pa)
Stress (Pa)
1.0E+06
4.0E+06
5.0E+05
0.0E+00
-5.0E+05
2.0E+06
1.0E+06
0.0E+00
-1.0E+06
-1.0E+06
-2.0E+06
-1.5E+06
-3.0E+06
10
20
(a)
30
40
50
60
70
80
The 0-30m segment of excavation
10
(b)
20
30
40
50
60
70
80
The 30-65m segment of excavation
Fig.10 Horizontal stress longitudinal distribution for pipe roof
Compressive
Stress
Tension
Both the field measurements and the numerical analysis show that the performance of the pipe roof
can be divided into three areas along the longitudinal direction. They are the tension section ahead of the
tunnel face, the compression section behind the tunnel face, and the tension section near the entrance
casing arch, as shown in Fig. 11.
Excavation direction
Pipe roof
Tunnel face
Fig.11 Zoning of longitudinal stress for pipe roof
ASCE
139
Actually, the pipe roof in section I has undergone tension. The peak tensile stress occurred ahead of
the tunnel face at a distance which is equal to the excavation height of top heading. The pipe roof in
section II has undergone compression. The peak value of compressive stress occurred at a distance
behind the tunnel face. This distance is about 5 m in numerical analysis while it is about 10 m in field
measurements. The relative larger distance in field measurements is due to the lagging effect of the
closure of the primary shotcrete in a tope heading construction method. In section III the pipe roof has
undergone tension as well. The peak value of tensile stress occurred close to the entrance casing arch.
All the analysis shows that the calculated section length and the peak stress of each section vary
with the distance of excavation length. The length of section and increases with the excavation
of tunnel face, however, the length of section keeps to be a constant length. Besides, the peak stress
of pipe roof in each section also increases with advancing of the tunnel face. This increase is fast at the
beginning of excavation and after the excavation passed half length of the pipe roof the stress increases
very slowly.
4.2.2 Comparison between field measurements and numerical results
The measured strain of the pipe was converted to stress by using the equivalent stiffness of the pipe
roof, which is calculated from the elastic modulus of the steel pipe and the in-filled mortar. Fig. 12
shows the stress calculated both from field measurements and numerical analysis. The two results are
found in good agreement both in values and distribution. This results show that the numerical method
used in this study can serve as an effective method in modelling the performance of long and large pipe
roof in tunnelling.
4.E+06
4.E+06
3.E+06
Numerical result
Measured result
1.E+06
0.E+00
-1.E+06
-2.E+06
-3.E+06
Numerical result
2.E+06
Stress(MPa)
Stress (MPa)
2.E+06
Measured result
0.E+00
-2.E+06
-4.E+06
-4.E+06
-5.E+06
0
10
20
30
40
50
60
70
80
(a) Stress of the pipe with 30 m excavation
-6.E+06
0
10
20
30
40
50
60
70
80
(b) Stress of the pipe with 42 m excavation
Fig.12 Comparison of the pipe stress from numerical simulation and field measurements
5 CONCLUSIONS
The performance of long and large pipe roof in Shitougang tunnel was analyzed by field measurements and 3D
numerical analysis. Based on the mechanical response of the typical pipe at the tunnel crown, the following
conclusions were obtained:
The development of stress along the long and large pipe roof turns out to be a bimodal curve distribution.
The waveform of the distribution curve increases initially and then moves forward with the advance of the
tunnel face.
The stress distribution of the pipe roof can be divided into three sections: tension section ahead of the
tunnel face, compressive section behind the tunnel face and tension section near the entrance casing arch.
The peak tensile stress generally occurred near the entrance casing arch and at a distance equal to the
ASCE
excavation height ahead of the tunnel face. The peak compressive stress is related to the distance from
closed primary support to the tunnel face. The length of tension section which is close to the entrance
casing arch tends to be constant while the other two sections vary with the tunnel face position.
The stress of the pipe increases fast first and then it increases very slowly after the tunnel face passed the
half length of the pipe roof.
The calculated stress of pipe roof was in good agreement with the measured result. Solid elements with
consideration for the contact between the surrounding rock and pipe roof can serve as an effective tool to
analyze the long and large pipe roof.
Acknowledgement
The authors are grateful to the CCCC Fourth Highway Engineering Co., LTD. of China and Nanning Railway
Bureau of China Railway Corporation for the contributions and support regarding the field measurements
presented in this paper. The technology demonstration research project of China Ministry of Transport
(CXKJSF0106-1) and the National Science and Technology Support Program of China (No. 2012BAK24B02) are
also appreciated.
References
1. Barisone,
G., B. Pigorini, and S. Pelizza. Umbrella Arch Method for Tunnelling in Difficult ConditionsAnalysis of Italian
Cases. Proc.,4th Congress, International Association of Engineering Geology, 1982,New Delhi, India, pp. 15-27.
2. Ibrahim, O. Control of Surface Settlements with Umbrella Arch Method in Second Stage Excavations of Istanbul Metro.
Tunnelling and Underground Space Technology, Vol. 23, 2008, pp.674-681.
3. C.O. Aksoy, and T. Onargan. The Role of Umbrella Arch and Face bolt as Deformation Preventing Support System in
Preventing Building Damages. Tunneling and Underground Space Technology, Vol. 25, 2010, pp.553-559.
4. Volkmann, G., and W. Schubert. Optimization of Excavation and Support in Pipe Roof Supported Tunnel Sections.
Tunneling and Underground Space Technology, Vol. 21, 2006.
5. The Professional Standards Compilation Group of Peoples Republic of China (JTG D702004). Code for Design of Road
Tunnel. China Communications Press, Beijing, 2004.
6. Yoo, C. Finite Element Analysis of Tunnel Face Reinforced by Longitudinal Pipes. Computers and Geotechnics, Vol.29,
2002, pp.73-94.
7. Yoo, C., and Shin, H.K. Deformation Behavior of Tunnel Face Reinforced with Longitudinal Pipes Laboratory and
Numerical Investigation. Tunneling and Underground Space Technology, Vol.18, 2003, pp.303-319.
8. Tan, W. L., and P. G. Ranjith. Numerical Analysis of Pipe Roof Reinforcement in Soft Ground Tunneling. Proc., 16th
International Conference on Engineering Mechanics (CD-ROM), 2003, ASCE, Seattle, Wash.
9. Hefny, A. M., W. L. Tan, P. Ranjith, J. Sharma, and J. Zhao. Numerical Analysis for Umbrella Arch Method in Shallow
Large Scale Excavation in Weak Rock. Tunneling and Underground Space Technology, Vol. 19, 2004.
10. Galli G, Grimaldi A, and Leonardi A. Three-Dimensional Modelling of Tunnel Excavation and Lining. Computers and
Geotechnics, Vol. 31, No.3, 2004, pp.171-183.
11. Bae, G. J., J. S. Shin, C. Sicilia, Y. G. Choi, and J. J. Lim. Homogenization Framework for Three-Dimensional
Elastoplastic Finite Element Analysis of a Grouted Pipe-Roofing Reinforcement Method for Tunneling. International
Journal for Numerical and Analytical Methods in Geomechanics, Vol. 29, No. 1, 2005, pp. 1-24.
12. Shin, J.H., Y.K. Choi, O.Y. Kwon et al. Model Testing for Pipe-Reinforced Tunnel Heading in a Granular Soil. Tunneling
and Underground Space Technology, Vol. 23, 2008, 241-250.
13. Yang, J.S., D.M. Gou and Y.X. Zhang. Field Measurements and Numerical Analyses of Double-layer Pipe Roof
Reinforcement in a Shallow Multiarch Tunnel, Journal of Transportation Research Record (No. 2050/Structures): 2008,
pp.145-153.
14. ZHENG, J., R. ZHANG and Q. YANG. Mechanical Mechanism of Pipe Roofs with Variable Coefficient of Subgrade
Reaction in Shallow Tunnels. Chinese Journal of Geotechnical Engineering, Vol. 31, No. 8, 2009.
15. FLAC Users Manual, Version 5.0. Itasca Consulting Group, Inc., Minneapolis, Minn., 2005.
16. China Railway Eryuan Engineering Group CO.LTD. Survey and Design Report of Shitougang Tunnel of Bid Section
of Extension Project of Hunan and Guangxi Railway. Chengdu, 2009.
ASCE
140
141
Analysis of the Impacts of a Tunnel on a Normal Fault Rupture through

Uniform Soil Cover
Mehdi Sarayloo1 and Reza Mahinroosta2
1
Dept. of Civil Engineering, Zanjan Branch, Islamic Azad Univ., Zanjan, Iran. E-mail:
m_sarayloo@yahoo.com
2
Lecturer, Engineering, Engineering Faculty, Charles Sturt Univ., Panorama Ave BATHURST, NSW
2795, Australia. E-mail: rmahinroosta@csu.edu.au
Abstract: Damages due to a fault activity can be divided into dynamic rupture and
quasi-static rupture. In some special projects like roads, channels, and pipelines,
engineers have to design and construct them on the faults inescapably. In addition to
seismic design, the effects of fault rupture crossing the structures must be considered
in the design process. In this paper, four different computer models of a granular
alluvium with and without tunnel were made by a three-dimensional finite-difference
software. A normal fault offset with three different cross angles was applied to the
bottom of the models at the next step. Then, the shear bands propagation and ground
deformations were calculated. The results show that the effects of tunnel existence on
the ground deformations and shear band propagation depend on the cross angle.
INTRODUCTION
In a seismic event, the rupture of an earthquake fault generates two types of ground
displacement: permanent quasi-static offsets on the fault itself, and transient dynamic
oscillations away from the fault (Anastasopoulos et al. 2007). The permanent offset on
a fault affects the ground surface only in some cases, when the fault rupture extends
all, or nearly all the way to the surface. Naturally, over the last four decades, much
less effort had been devoted to understand how a fault rupture affects the overlying
soil, structures and facilities until the three notorious earthquakes of 1999 in Turkey
and Taiwan having offered numerous examples of detrimental effects of large surface
fault ruptures (Loukidis et al. 2009, Anastasopoulos and Gazetas 2010).
The existence of active faults in the path of tunnels implies a seismic hazard, which
can affect underground and on-ground structures in addition to the tunnel. To this
extent, seismic codes correctly require that the construction not be built within the
immediate vicinity of active faults, at least not without a specialized analysis and
design. However, the quantitative indications for the width of these set-back zones are
few and rather uncertain. Given that various conditions such as a structure existence
can deflect the path of shear bands in the overlying soil, the width of the set-back zone
cant be defined precisely. So for these cases, the following questions are raised:
ASCE
1) Will the fault rupture reach the ground surface and at which location?
2) How will the ground surface be distorted when a tunnel exists?
This paper studies the special case of a normal fault. The numerical analyses
presented herein are quasi-static, i.e. they do not account for the potential effects of
seismic wave propagation and ground response.
METHODOLOGY
The numerical analyses presented herein were performed using the finite-difference
code FLAC 3D, which employs the dynamic relaxation technique. The merits of the
dynamic relaxation technique, in connection with the numerical analysis of fault trace
propagation problems, have also been demonstrated in the previous studies (Roth et al.
1982). In this study, firstly a simple alluvium model was made, and fault displacement
was applied. Then, the same fault offset was applied to the same alluvium with a
tunnel for different cross angles. Finally, shear bands causing ground deformations
and distortions were calculated and compared. As well as the mentioned models, two
different models based on Lin et al. (2007) sandbox tests were conducted to verify the
modeling method.
Verification
In this part, the two numerical models with similar characteristics with that of Lin et
al. (2007) sandbox tests, were conducted in FLAC 3D and then the results were
compared with theirs. They also made numerical models with FE code software
(ABAQUS) the same as the physical tests. Faulting in a free-field alluvium was
depicted in Figure 1 and the one in alluvium with a tunnel was depicted in Figure 2.
FIG. 1. Deformation and development of fault zones of the overlying overburden

soil. (a) Sandbox test; (b) Sandbox test in detail; (c) FE model; (d) FD model.
ASCE
142
FIG. 2. Deformation and development of fault zones of the overlying overburden

soil with underground tunnel. (a) Sandbox test; (b) Sandbox test in detail; (c) FE
model; (d) FD model.
The similarity in shear bands development and the shape of settlements between the
illustrations verifies the modeling method of this paper. Even it can be seen in Figure
1 that the results of FD model are more accurate than those of FE model in this
problem.
Mesh and Boundary Conditions
The alluvium depth chosen in this study was 20 m while the length in both X and Y
axis was four times bigger than the depth (i.e. 80 m) to reduce the boundary condition
effects on results. In order to mesh the alluvium, a grid with a different element
density was used. The tunnel excavated in the second step has a 5 m diameter and 10
m depth. The tunnel is lined with concrete segments, which were modeled by the liner
elements and were connected to the grid through an interface.
FIG. 3 Problem geometry, mesh and imposed normal faulting boundary

conditions; =Cross angle =Fault dip angle (Sarayloo 2011).
The boundary walls of the model are fixed in their normal direction and are free in
other directions. The bottom boundary is split in two parts, one that remains firm and
the other follows the hanging wall movement of the fault (Figure 3). The displacement
specified to the bottom boundary and boundary walls of the moving block are parallel
to the fault plane in the bedrock, i.e. their direction is inclined at a dip angle =30
relative to the horizon. The modeling is conducted for the three cross angles of =0,
45, 90 and the one-meter downward fault displacement along the slide surface.
ASCE
143
144
Materials Behavior
Several experimental and numerical studies have shown that post-peak soil behavior
is a decisive factor in fault rupture propagation and its possible emergence on the
ground surface (Anastasopoulos et al. 2007).The constitutive model assigned to the
soil is the elastoplastic Mohr-Coulomb with isotropic strain-softening model.
Furthermore, the studied alluvium is made of sand with low cohesion. The features of
the alluvium material are mentioned in Table 1. The constitutive model assigned to the
concrete lining is also elastic and is shown in Table 2.
Table 1. Parameters of the MC model used for the soil.
Model
Constant
Value
(t/m3)
2
E
(kPa)
4104
0.3
cp
(kPa)
1
cres
(kPa)
0.75
Ini
res
Ini
res
20
45
30
-11
Table 2. Parameters of the MC model used for the concrete lining.

Parameter
Value
E (kPa)
2107
0.2
fc (kPa)
2.5104
fct (kPa)
2.5103
Thickness (m)
0.5
NUMERICAL ANALYSIS
Various studies (Roth et al. 1982, Bray et al. 1994, Lin et al. 2006, Anastasopoulos
et al. 2007, Loukidis et al. 2009) have been done about the fault rupture propagation in
the overlying soil and the effect of different parameters such as friction angle, dilation
angle, cohesion, soil type, fault type, etc. In most of these studies (Lin et al. 2006,
Anastasopoulos et al. 2007, Loukidis et al. 2009), centrifuge tests or sand box tests
have been used to verify the modeling. The results show the accuracy of numerical
modeling and indicate the ability of these methods in predicting the fault effects on
soil and structures. Some studies have been done (Shahidi and Vafaeian 2006, Lin et
al. 2007, Anastasopoulos et al. 2008, Anastasopoulos and Gazetas 2010, Mahinroosta
and Mirmoayed 2011) about the impact of fault rupture on structures located in the
alluvium cover of the fault. Valuable results have been conducted, but because of 2D
modeling, the probable cases of faulting have been limited and in fact, some cases
have not been investigated.
Free-Field Alluvium
Fault rupture can cause the development of the primary shear band and a secondary
shear band which are approximately symmetrical (Figure 4a).
ASCE
FIG. 4 One-meter fault offset with a 30 dip angle. (a) Shear bands propagation
through overlying soil; (b) Ground surface settlement and inclination plot.
Usually in normal faults with small dip angles, depending on the soil type, a wedgeshaped depression can be seen on the ground surface, made by the development of the
secondary shear band. This wedge failure can be seen in Figure 4a. According to the
fault dip angle, the fault trace is predicted to appear at X=74.6 m but based on Figure
4b, the fault trace appears at X=46 m, indicating an average angle of faulting of 73.
Actually, the rupture propagation path is deflected in soil and becomes closer to the
perpendicular. This deviation is due to the rupture path which passes through the soil.
According to Figure 4b the maximum inclination is 0.49, which occurs at X=46 m and
the maximum settlement is 1 m, which is 2 times larger than the faults vertical
displacement (0.5 meter), at X=37 m.
Alluvium with tunnel
Previous studies (Shahidi and Vafaeian 2005, Lin et al. 2007, Anastasopoulos et al.
2008, Anastasopoulos and Gazetas 2010) show that having a tunnel in the path of a
fault rupture changes the path and the fault trace. In the second part of this study, the
alluvium of the previous part is modeled with an excavated tunnel in it for different
cross angles of =0, 45, 90 (0 for faulting along the tunnel and 90 for faulting
perpendicular to the tunnel). For more accurate modeling, the tunnel has been modeled
using step by step excavation and lining method.
Faulting with a cross angle of =0
As shown in Figure 5a, the development of shear bands has been changed because of
the tunnel. After the primary shear band reaches the tunnel perpendicularly, it is
distributed on half of its circumference and then from the upper left corner it reaches
the ground surface with an angle near to perpendicular. The primary shear band affects
the bottom and the left walls of the tunnel. The secondary shear band is also developed
which reached the surface with a similar angle in the previous section. Figure 5a
shows that the wedge failure is formed above the tunnel on the left which has a
smaller base width than the wedge failure formed in the free-field alluvium.
ASCE
145
FIG. 5 One meter fault offset with a 30 dip angle and a 0 cross angle above the
tunnel. (a) Shear bands propagation through overlying soil; (b) Ground surface
settlement and inclination plot.
According to the tunnel existence and the deviation of the primary shear bands
development, the fault trace appears at Y=38 m on the ground (Figure 5b). The
maximum settlement is 1.48 m, which is 2.96 times larger than the faults vertical
displacement and appears at Y=33 m (Figure 5b). Since the wedge failure has a
smaller width in this state, a smaller zone is affected by the fault rupture and so the
ground surface inclination is larger than the free-field alluvium case. As shown in
Figure 5b, the maximum ground surface inclination magnitude is 1.08, which is 2.2
times greater than that of the free-field alluvium state and occurs at Y=38 m.
In this case, because of the specific cross angle, in order to reduce the boundary
conditions effects on the results, 20 meters were added to both sides of the model in
the X direction. For this condition, the development of shear bands after reaching the
tunnel, are distributed along the tunnel so that the fault rupture does not reach the
ground surface (Figure 6a). The cause of this phenomenon is the angle between the
tunnel and the fault trace (named cross angle), causing the tunnel to behave like a
continuous beam against the fault induced movement. Therefore, the stresses caused
by faulting are distributed in a longer area, and this situation does not allow the shear
bands to reach the ground surface. The primary shear band affects the walls and the
bottom of the tunnel when it reaches the tunnel, and like the previous state, it causes
an increase in the stresses, especially tensile stresses in the concrete lining. As shown
in Figure 6a, the wedge failure is only formed under the tunnel.
The maximum ground surface settlement recorded above the tunnel is 0.21 m, which
is 2.38 times smaller than the faults vertical displacement and occurs at X=43 m. The
maximum inclination in this path is 0.0078, which occurs at X=45 m. (Figure 6b)
ASCE
146
FIG. 6 One-meter fault offset with a 30 dip angle and a 45 cross angle above
the tunnel. (a) Shear band propagation through overlying soil; (b) Ground
surface settlement and inclination plot.
This case compared the development of shear bands, reaching the tunnel and
propagating along it. In the previous state, the fault rupture does not reach the ground
surface (Figure 7a) because the tunnel in the rupture path acts like a long continuous
beam, causing the distribution of the stresses along itself and it also reduces its effect.
The primary shear band affects the bottom when reaching the tunnel, walls and the
crest and causes an increase in the stresses, especially tensile stresses in concrete
lining. Moreover, in this case, the wedge failure is only formed under the tunnel,
where the shear bands are being developed. (Figure 7a)
FIG. 7 One-meter fault offset with a 30 dip angle and a 90 cross angle above
the tunnel. (a) Shear band propagation through overlying soil; (b) Ground
surface settlement and inclination plot.
ASCE
147
The maximum ground surface settlement recorded above the tunnel is 0.16 m, which
is 3.13 times smaller than the faults vertical displacement. Also the maximum
inclination along this direction is 0.0093, which occurs at X=44 m. (Figure 7b)
CONCLUSIONS
In this article, the effect of faulting on ground surface settlements and shear band
development was investigated for a simple alluvium and an alluvium with an
excavated tunnel. The point of interest in this paper is the investigation of the cross
angle. Based on the performed analyses, the following results are obtained:
1. The maximum settlement of 1.48 m is 2.96 times larger than the faults vertical
displacement and it occurs in the case =0. After that, the maximum settlement is
for the free-field alluvium case.
2. The maximum inclination magnitude is 1.08, which occurred in the case of =0.
After that the maximum inclination occurs in the case of free-field alluvium.
3. The fault rupture reaches the ground surface only in the case of =0.
4. An existing tunnel (in all the cases mentioned above) may cause a reduction in
surface settlement along the tunnel or may increase it. It depends on the situation
of the tunnel against the fault plane.
5. Based on the maximum surface settlements, it can be concluded that developing
the secondary shear band increases these amounts and increases the risk.
6. Faulting in all above cases produces high stresses in the concrete lining. The
tensile stresses are much bigger than the compressive stresses. In fact, the
concrete lining failed due to the tensile stress before it reached the maximum
compressive stress.
According to the results, it is necessary to consider all the conditions, which can affect
the surface settlement and also the width of set-back zone in a fault zone.
REFERENCES
Anastasopoulos, I., and Gazetas, G. (2010). Analysis of cut-and-cover tunnels against
large tectonic deformation. Bulletin of Earthquake Engineering, 8:283307.
Anastasopoulos, I., Gazetas, G., Bransby, MF., Davies, MCR., and El Nahas A.
(2007). Fault rupture propagation through sand: finite-element analysis and
validation through centrifuge experiments. Journal of geotechnical and
Geoenvironmental Engineering, 133(8):94358.
Anastasopoulos, I., Gerolymos, N., Drosos, V., Georgarakos, T., Kourkoulis, R., and
Gazetas, G. (2008). Behaviour of deep immersed tunnel under combined normal
fault rupture deformation and subsequent seismic shaking. Bulletin of Earthquake
Engineering. 6:213239.
Asghari, S., Memarian, H., Zare, M. (2009). Evaluating the stability of the twin
tunnels of Ahvaz subway in the crossing section with Ahvaz fault. 6th National
Conf. on Geology Engineering and Geoenvironmental Engineering. Vol I: pp
329-338.
Bray, J.D., Seed, R.B., Cluff, L.S., Seed, H.B. (1994). Earthquake fault rupture
propagation through soil. Journal of Geotechnical Engineering. 120, 543561.
ASCE
148
Lin, M.L., Chung, C.F., Jeng, F.S. (2006). Deformation of overburden soil induced by
thrust fault slip. Engineering Geology. 88, 7089.
Lin, M.L., Chung, C.F., Jeng, F.S., and Yao, T.C. (2007). The deformation of
overburden soil induced by thrust faulting and its impact on underground tunnels.
Engineering Geology. 92, 110132.
Loukidis, D., Bouckovalas, G.D., and Papadimitriou, A.G. (2009). Analysis of fault
rupture propagation through uniform soil cover. Soil Dynamics and Earthquake
Engineering. 29, 13891404.
Mahinroosta, R., Mirmoayed, S.A. (2011). Shear band development in embankment
dam and its alluvial foundation due to normal fault rupture deformation. 6th Int.
Conf. on Seismology and earthquake Engng, Vol I: pp 10879.
Roth, WH., Kalsi, G., Papastamatiou, D., Cundall, PA. (1982). Numerical modeling of
fault propagation in soils. 4th Int. Conf. on numerical methods in geomechanics.
Vol I: pp. 48794.
Sarayloo, M., (2011). Investigation of normal fault effect on instability of shallow
tunnels. M.Sc. dissertation, Islamic Azad University, Zanjan branch, Zanjan, Iran.
Shahidi, A.R., Vafaeian, M. (2005). Analysis of longitudinal profile of the tunnels in
the active faulted zone and designing the flexible lining (for Koohrang-III tunnel).
Tunnelling and Underground Space Technology. 20, 213221.
ASCE
149
150
Plasticity Solutions of an Undrained Cavity Contraction for the Prediction of Soil

Behaviour around Tunnels
Pin-Qiang Mo1 and Hai-Sui Yu2
1
Former Research Fellow, Faculty of Engineering, the Univ. of Nottingham, University Park,
Nottingham NG7 2RD, U.K. E-mail: mopinqiang@gmail.com
2
Professor, Faculty of Engineering, the Univ. of Nottingham, University Park, Nottingham NG7 2RD,
U.K. E-mail: hai-sui.yu@nottingham.ac.uk
Abstract: Cavity contraction method has been used for decades for the design of
tunneling and the prediction of ground settlement, by modelling the cavity unloading
process from the in-situ stress state. Analytical solutions of undrained cavity
contraction in a unified state parameter model for clay and sand (CASM) are used in
this paper to predict the soil behaviour around tunnels. CASM is a critical state soil
model with two additional material constants, which has the ability to capture the
overall behaviour of clay and sand under both drained and undrained loading
conditions. Large-strain and effective stress analyses of cavity contraction provide the
distributions of stress/strain within elastic and plastic regions around tunnels. The
effects of ground condition and soil model parameters are investigated from the results
of stress paths and cavity contraction curves. Comparisons are also provided between
the predicted and measured behaviour of tunneling, using data of centrifuge tunnel test
in clay. To account for the effect of free ground surface, Loganathan & Pouloss
method using undrained gap parameter is incorporated to estimate tunneling induced
ground surface settlement.
INTRODUCTION
Cavity expansion theory, concerning stress/displacement fields around cavities, has
been developed and applied to a variety of geotechnical problems, as described in Yu
(2000). By modelling the cavity unloading process from the in-situ stress state, cavity
contraction method has been used for decades for the design of tunneling and the
prediction of ground settlement (e.g. Hoek & Brown, 1980; Mair & Taylor, 1993). In
the past two decades, critical state solutions were increasingly proposed to account for
the dependence of soil strength with deformation history (e.g. Collins & Yu, 1996; Yu
& Rowe, 1999). Undrained solutions of cavity expansion are recently developed using
a unified state parameter model for clay and sand (CASM), which has the ability of
capturing the overall behaviour of clay and sand (Mo & Yu 2016).
ASCE
This paper provides analytical solutions of undrained cavity contraction in a unified

state parameter model for clay and sand (CASM) to predict the soil behaviour around
tunnels. Taking account of the effect of stress history by an effective stress analysis,
the predictions of stress fields and soil displacement are compared with previous
analytical results and centrifuge data. In addition, to account for the effect of free
ground surface, undrained gap parameter by Loganathan & Poulos (1998) is suggested
to be incorporated for estimation of tunneling induced ground surface settlement.
PROBLEM DEFINITION
The contraction of a spherical/cylindrical cavity with initial radius a0 in an infinite
soil under undrained condition is concerned in this paper. The geometry and
kinematics of cavity contraction are illustrated schematically in Fig. 1. Parameter m
is used to integrate both spherical ( m = 2 ) and cylindrical ( m = 1 ) scenarios. The
preconsolidation pressure is referred to as p' y 0 and R0 = p' y 0 / p'0 represents the
isotropic overconsolidation ratio in terms of the mean effective stress. To
accommodate the effect of large deformation in cavity contraction process, large strain
analysis is adopted for both elastic and plastic regions by using logarithmic strains.
Note that a compression positive notation is used in this paper.
FIG. 1. Geometry and kinematics of cavity contraction.

With the benefits of the concept of state parameter, Yu (1998) proposed a unified
state parameter model for clay and sand, which is referred to as CASM. It is a simple
constitutive model with two additional material constants introduced to the standard
Cam-clay model, whereas the overall behaviour of clay and sand can be satisfactorily
modelled by CASM under both drained and undrained loading conditions. The state
boundary surface of CASM (Fig. 2b) is described as:
n
ln( p' / p' y )

(1)
=
= 1
R
ln r *
M
where = q / p' is known as stress ratio; n is the stress-state coefficient;
R = ( ) ln r * , is the reference state parameter; and r * is the spacing ratio,
defined as p' y / p' x (see Fig. 2a). In addition, a non-associated flow rule based on
Rowes stress-dilatancy relation is adopted to better describing the deformation of
sands and other granular media.
ASCE
151
152
.C
N
C.S
.L.
L
S.
.
q / ( M p'y )
.
.L
C.
FIG. 2. Unified state parameter model for clay and sand (CASM).
PLASTICITY SOLUTIONS
Plasticity solutions are presented in this section, for a cavity contracted from a0 to
a until the soil around the cavity reaches the critical state (i.e. soil medium is
deformed to have elastic, plastic and critical-state regions). c is the radius of the
elastic-plastic boundary, and ccs is the radius where the soil starts to be in critical
state. Thus, for r > c , soil is in elastic region; whereas for ccs < r < c , soil is in plastic
region, and critical-state zone is for soil at a < r < ccs (see Fig. 1b). In terms of
undrained condition, the soil volume within an arbitrary radius ( r ) can be assumed as
constant, and the relation can be written as:
m+1
m +1
r0 r m+1 = a0 a m+1 = T
(2)
The following subsections describe the solutions in elastic and plastic regions, while
the details of derivations can be found in authors personal notes (i.e. Mo & Yu, 2016;
As these notes have not been published yet, readers are welcome to ask for relevant
document).
Solution in Elastic Region
The effective stresses, total stresses and strains in elastic region are expressed in
Equation (3).
m
A(r )
=
r
r = p0 + 2G0 mB(r )
' r = p' 0 mA(r )
2G0
' = p' 0 + A(r )

= p0 + 2G0 mB(r ) + (m + 1)A(r ) = 1 A(r )
2G0
where
k
2G0 r m +1 + T
1 T / r m +1
and B(r ) =
(3)
A(r ) =
ln
k2
m + 1 k =1
m + 1 r m +1
ASCE
153
The elastic-plastic boundary c can be written as:

1
m+1
c=
n
ln
R
Mp
'
0
0
1 exp
*
ln
r
2
G
(4)
Solution in Plastic Region:

Elastic volumetric strain and plastic volumetric strain:
p' y
p'
pp =
ln
ep = ln
p'
p'0
y0
Elastic deviatoric strain:
[1 + (m 1) ]M n [A + A ln p' ]1n +1 +
(m + 1)
qe =
A(c )
(m + 1)(1 2 ) (1 + n )A2 1 2
2G0
1
[A1 + A2 ln p' ]n
(5)
1
1
n
[A1 + A2 ln p'0 ]n +1 [A1 + A2 ln p'0 ]n
(1 + n )A2
where
ln R0 + 1 ln p'0
1
A
and
=
2
ln r*
ln r*
Plastic deviatoric strain:
n(m + 1) 2 M n
qp =
cn + 9 + 3M 2 M 2
n
9A2 M m n
0
k
n 1
n 1

M d = M k =0 n + k M
n 1 k

k =0 M k
n 1 k
A1 =
] (
(6)
n1
d
M
(c = M )
(c < M )
(7)
(c > M )
Total stresses can be calculated by numerical integration from Equation (8):

q
r = m r dr
(8)
RESULTS AND DISCUSSION

Comparisons with Results of Solutions by Yu & Rowe (1999)
In this section, the results of soil behaviour around tunnels are presented by using
the provided plasticity solutions of cavity contraction in undrained condition. As the
yield criterion of the original Cam-clay model can be recovered from CASM by
ASCE
154
selecting the material constants: n = 1.0 and r * = 2.7183 , the validation of the
solutions is carried out by comparing the results of original Cam-clay model with the
results of solutions by Yu & Rowe (1999). The values of the critical state parameters,
chosen to be relevant for London clay, are identical to Yu & Rowe (1999). It needs to
be noted that ambient pore pressure is not included in the results of total stresses (i.e.
= ' + u ).
Figures 3 and 4 present the results of soil behaviour around tunnels using cylindrical
and spherical scenarios, with the overconsolidation ratio of R0 = 1.001 . The final
contraction for both cylindrical and spherical tests is a0 / a = 1.95 and 1.12,
respectively. The results are found to be comparable with data from Yu & Rowe
(1999) when using non-associated flow rule, while identical results are shown for tests
using associated flow rule.
(b)
(a)
FIG. 3. Soil behaviour around tunnels using cylindrical cavity model.
(a)
(b)
FIG. 4. Soil behaviour around tunnels using spherical cavity model.

Comparisons with Results of Centrifuge Test by Mair (1979)
The proposed analytical solutions are related to soil behaviour around tunnels, with
comparisons to centrifuge results by Mair (1979). The selected centrifuge test is 2DP
with cover to diameter ratio: H / D = 1.67 . The tunnel test in clay can be assumed to
ASCE
155
be undrained condition. According to Mair (1979) and Yu & Rowe (1999), soil
properties are chosen as: = 3.92 , = 0.3 , = 0.05 , M = 0.8 , = 0.3 , su = 26 kPa .
The predictions of crown settlement ( uc ) and mid-surface settlement ( u s ) are shown
in Fig. 5. The crown settlement in Fig. 5a shows comparable results with previous
analytical results (Yu & Rowe, 1999) and centrifuge data (Mair, 1979). As shown in
Fig. 5b and noted by Yu & Rowe (1999), the cavity solutions tend to underpredict the
observed mid-surface settlement, probably owing to the shallow tunnel test of 2DP
with the effect of free ground surface.
(a)
(b)
FIG. 5. Predicted and observed settlements for a centrifuge test in clay.

By using a virtual image technique suggested by Sagaseta (1987), Verruijt & Booker
(1996) provided an analytical solution for a tunnel in a homogeneous elastic halfspace (Fig. 6a). However, Verruijt & Bookers method results in wider settlement
trough and larger horizontal movements, as reported by Loganathan & Poulos (1998).
The settlements caused by tunnelling are often characterized by ground loss, which is
defined as a percentage of the ratio of the surface settlement trough volume and the
tunnel volume per unit length. The undrained ground loss was suggested to be defined
based on the gap parameter (referred to as equivalent ground loss, introduced by
Lo & Rowe, 1982 and Rowe & Kack, 1983) by Loganathan & Poulos (1998), for an
analytical prediction for tunnelling-induced ground movements in clays.
Regards to the gap parameter g for the magnitude of the equivalent twodimensional void formed around the tunnel, the ovalisation (Fig. 6b) is attributed to
the combined effects of the three-dimensional elastoplastic ground deformation at the
surface, overexcavation of soil around the periphery of the tunnel shield, and the
physical gap that is related to the tunnelling machine, shield, and lining geometry.
Rowe & Kack (1983) defined the undrained gap parameter g as shown in
g = G p + U 3* D + , where G p = physical gap ( G p = 2 + ) that represents the
geometric clearance between the outer skin of the shield and the lining; = thickness
of the tailpiece; = clearance required for erection of the lining; U 3* D = equivalent 3D
elasto-plastic deformation at the tunnel face; and = value that takes into account the
quality of workmanship. The theoretical method for estimation of the gap parameter
can be found in Lee et al. (1992).
ASCE
156
FIG. 6. Circular- and Oval-shaped ground deformation patterns around tunnels.

Loganathan & Poulos (1998) defined the equivalent undrained ground loss 0 with
respect to the gap parameter: 0 = 4 gR + g 2 / 4 R 2 100% , where R = radius of the

tunnel. According to the ground deformation patterns and ground loss boundary
conditions, Loganathan & Poulos (1998) derived the nonlinear ground movement
around the tunnel soil interface, as shown in Equation (9).
1.38 x 2
4 gR + g 2
0.69 z 2
exp
x ,z =
(9)
2
4R2
H 2
(H + 2 R )
After incorporated into the closed form elastic solutions derived by Verruijt &
Booker (1996), the modified formula for the prediction of surface settlement can be
wright as shown in Equation (10).
1.38 x 2
H +R
4 gR + g 2
(10)
u z =0 = 4 (1 )R 2
exp
2
(H + R )2 + x 2 4 R 2
(H + 2 R )
As shown in Fig. 5b, Loganathan & Pouloss method with g = 0.5 uc ~2.0 uc is
incorporated to estimate tunneling induced ground surface settlement, which gives
better agreement with experimental observation, especially for g = 1.0 uc ~1.5 uc .
CONCLUSIONS
By modelling cavity unloading process, analytical solutions of undrained cavity
contraction in a unified state parameter model for clay and sand (CASM) were
proposed in this paper to predict the soil behaviour around tunnels, including stress
fields and crown/ground settlements. Taking the advantages of CASM with the ability
of capturing overall behaviour of clay and sand, large-strain and effective stress
analyses of cavity contraction provided the distributions of stress/strain within elastic
and plastic regions around tunnels. The results of soil behaviour around tunnels using
cylindrical and spherical scenarios showed identical results with previous analytical
solutions using original Cam-clay model. Comparisons were also provided between
the predicted and measured behaviour of tunneling, using data of centrifuge tunnel test
in clay. To account for the effect of free ground surface, Loganathan & Pouloss
ASCE
method was suggested to be incorporated to estimate tunneling induced ground surface

settlement, by using undrained gap parameter g = 1.0 uc ~1.5 uc .
REFERENCES
Collins, I.F. and Yu, H.S. (1996). Undrained expansions of cavities in critical state
soils Int. J. Numer. Anal. Meth. Geomech., 20(7), 489-516.
Hoek, E. and Brown, E.T. (1980). Underground excavations in rock The Institution
of Mining and Metallurgy, London, England.
Lee, K.M., Rowe, R.K. and Lo, K.Y. (1992). Subsidence owing to tunneling. I:
Estimating the gap parameter. Can. Geotech. J., Ottawa, Canada, 29, 929-940.
Lo, K.Y. and Rowe, R.K. (1982). Prediction of ground subsidence due to tunneling in
clays Res. Rep. GEOT-10-82, Facu. Of Engrg. Sci., Univ. of Western Ontario,
London, Ont., Canada.
Loganathan, N. and Poulos, H.G. (1998). Analytical prediction for tunneling induced
ground movements in clays Journal of Geotechnical and Geo-Environmental
Engineering, ASCE, 124(9), 846-856.
Mair, R.J. (1979). Centrifuge modelling of tunnel construction in soft clay PhD
thesis, Cambridge University.
Mair, R.J. and Taylor, R.N. (1993). Prediction of clay behaviour around tunnels using
plasticity solutions Predictive Soil Mechanics, Thomas Telford, London, 449-463.
Mo, P.Q. and Yu, H.S. (2016). Undrained cavity expansion analysis in a unified state
parameter model for clay and sand: CASM Personal note, paper submitted to
Gotechnique.
Rowe, P.K. and Kack, G.J. (1983). A theoretical examination of the settlements
induced by tunneling: Four case histories Can. Geotech. J., Ottawa, Canada. 20,
299-314.
Sagaseta, C. (1987). Analysis of undrained soil deformation due to ground loss
Gotechnique, London, England, 37(3), 301-320.
Verruijt, A. and Booker, J.R. (1996). Surface settlements due to deformation of a
tunnel in an elastic half plane Gotechnique, London, England, 46(4), 753-756.
Yu, H.S. (1998). CASM: A unified state parameter model for clay and sand Int. J.
Numer. Anal. Meth. Geomech., 22(8), 621-653.
Yu, H.S. (2000). Cavity expansion methods in Geomechanics Kluwer Academic
Publishers, The Netherlands.
Yu, H.S. and Rowe, R.K. (1999). Plasticity solutions for soil behaviour around
contracting cavities and tunnels Int. J. Numer. Anal. Meth. Geomech., 23, 12451279.
ASCE
157
Numerical Analysis of a Temperature Field within a Vertical Frozen Soil Wall

Jun Hu1; Yong Liu2; and Kai Yao3
1
Lecturer, College of Civil Engineering and Architecture, Hainan Univ., Haikou, Hainan 570228, China.
E-mail: 183633299@qq.com
2
Research Fellow, Dept. of Civil and Environmental Engineering, National Univ. of Singapore, 1
Engineering Dr. 2, Singapore 117576, Singapore (corresponding author). E-mail: ceeliuy@nus.edu.sg
3
Ph.D. Student, Dept. of Civil and Environmental Engineering, National Univ. of Singapore, 1
Engineering Dr. 2, Singapore 117576, Singapore (corresponding author). E-mail: yaokaisdu@gmail.com
Abstract: This paper presents an investigation on the generation and distribution pattern
of the temperature field within a vertical frozen soil wall by ADINA. The vertical
freezing closure time was 7 days, and after a freezing time of 30 days, the vertical
freezing formed a water-resistant frozen soil wall with a thickness greater than 1.6 m.
The area within 1 m of the vertical frozen soil wall was affected due to the exchange of
heat between the frozen and unfrozen zones. Thus, the spacing of vertical freezing pipes
should be less than 1 m. For design of the freezing process, the effect of temperature at
boundary zones should be taken into account. Marginal zones should be set aside
because the effective/workable zone of freezing pipes is about 1 m. The width and depth
of the vertical frozen wall should be larger than the designated sizes.
INTRODUCTION
A large diameter shield machine is usually required for the construction of cross-over
and cross-ocean tunnels, which results in increased risk during the break-in and
break-out at the end of a tunnel. Therefore, it is important to enhance the soil strength at
the end of a tunnel for the tunnel stability during boring process (Hu 2013). Ground
improvement is required for shield launching and receiving the large-diameter TBM.
The ground freezing approach has been widely used in tunnel (e.g. Nakamura 2004; Yu
et al. 2005; Yang et al. 2006; Ou et al. 2009; Pimental et al. 2012). For example, a triple
axis mixing machine is first used for soil consolidation and is then followed by a ground
freezing process to freeze the pore water to generate a frozen soil wall, which has high
strength and is impermeable (Andersland and Ladanyi, 2004). Sometimes, a dewatering
well is also necessary. In this paper, we analyzed the distribution regularity and
evolution of the temperature field in the freezing area during shield launching by
ADINA (see Bathe and Dong 2014) for guiding the design process.
A 21m19.4 m 30.5m (length width depth) launching shaft supported by the
diaphragm walls, was constructed for the cross-river tunnel for Nanjing Metro line 10.
This structure will be excavated by a tunneling boring machine (TBM) with an
excavation diameter of 12.03m and a length of 14.05m. The burial depth is 21.05m
above the tunnel center. The detailed information and parameters of soil layers around
the launching shaft were listed in Table 1. For this project, we combined two approaches
to improve the surrounding soils: overlapping cement-admixed columns and ground
ASCE
158
159
freezzing (see Fig

g. 1). The former
f
figuree is construucted using tthe deep miixing methodd
with triple-shaft mixers. Th
he diameter of each shaaft is 1 m, and the cenntre-to-centrre
m The stron
ng reinforcem
ment area refers to a higgher cement ccontent of 17%
distaance is 0.75 m.
and the
t weak rein
nforcement area has a ceement conteent of 7%. Inn order to furrther enhancce
the bonding
b
betw
ween the imp
proved soil and the retaaining structuure, two row
ws of verticaal
freezzing pipes were
w
arranged along the tunnel axis as shown inn Fig. 1. Thhe spacing oof
pipess was kept at
a 0.8 m with
h the triangu
ular layout ppattern. Therre are 29 andd 28 pipes inn
Row
ws A and B, respectively
r
y (see Fig. 2)). The distannce betweenn Row A andd the edge oof
diaph
hragm wall was
w 0.4m, and the spaciing between the adjacennt pipes was 0.8m. Figurre
2 sho
ows the in-situ layout off vertical freeezing pipess. The frozenn depth of eaach pipe waas
31.15
5 m and the total frozen
n depth was approximatee 1776 m. Thhe diameter of each pipe
was 127 mm and had thickn
ness 4.5 mm
m. The brinee was used aas the liquidd for freezing
with temperaturee around -29C for the fiirst 30 days aand around --27C thereaafter.
Table 1. Ph
hysical mecha
anical parameters of soil h
horizons at sh
hield launchin
ng spot
Plain fill
16.5
0.36
7.00
26.0
Friction
angle
(degree)
7.9
(2) 0.8 m 1.7 m
Clay
18.1
0.30
7.90
18.0
13.2
(3) 1.7 m 16.7 m
Silty clay
16.7
0.32
6.22
18.0
15.3
(4) 16.7
7 m 31.1 m
Silt
S and fine sand
18.8
0.26
19.97
8.0
31.5
(5) 31.1
1 m 48.2 m
Silt
S and fine sand
19.4
0.26
24.17
6.0
27.3
Depth
D
(1) 0 m 0.8 m
Soil Strata
Unit weight
(kN/m3)
Poissons rat
atio
Youngs moddulus
(MPa)
Cohesioon
(kPa)
(a) Plane view

w
(b
b) Cross-section
FIG. 1. Sch
hematic layou
ut of soil reinfforcement
and scope (unit: mm)
FIG. 2. L
Layout of verttical freezing
pipes
MERICAL MODEL
M
NUM
The
T basic asssumptions for
f the simu
ulations are as follows: (1) all soil layers in thhe
simu
ulation rangee are distribu
uted in a horrizontal dire ction; (2) eaach soil layeer is regardedd
to be
b homogen
neous and isotropic in thermodyynamic prooperties; (3)) the initiaal
temp
perature with
hin the soilss is the sam
me; (4) the innfluence off the pipes tthemselves iis
negliigible; in otther words, the temperrature load iis applied tto the soils directly; (55)
therm
mo-physical parameters are constantt throughout the simulatiion. The phaase change oof
transsient heat co
onduction waas used to caalculate the temperaturee field. Onlyy a section oof
ASCE
160
12 m in length was consideered due to the symmeetry of the pproblem. Thhe simulationn
modeel is illustraated in Fig. 3. There weere 15 and 14 equally sspaced freezzing pipes inn
Row
ws A and B,, respectivelly. The frozzen depth off each pipe was 31.2 m
m. All otheer
param
meters weree set as the same
s
of those in the in--situ projectt. The 4-nodde tetrahedraal
elem
ment was seleected for mo
odel discretizzation. The llocal mesh reefinement w
was applied to
the surrounding
s
area of the freezing
f
pipees. The moddel consists oof 815,788 eelements withh
149,0
000 nodes. In
I order to trrack the chan
nge in tempeerature, 22 rrecording pooints were seet
in thrree paths in the planes z = 0 and y = 0 (Fig. 3).
FIG. 3. Schematic diagrams

d
of ggeometric mod
del
The
T thermop
physical paraameters of ceement-admixxed soils aree summarizeed in Table 22,
and are
a based on
n the soils saampled from
m the construuction site (ssee Hu 20122, 2013). Thhe
initiaal temperatu
ure in simulaation was seet as 30C. This is beccause the tem
mperature oof
naturral soils usuaally varies frrom 19 to 20
0Cat a deptth of 10 m; inn addition, tthe formationn
temp
perature may
y rise by app
proximately 10C due too the heat of hydration of the cementt.
A teemperature of 30C was
w set consstant at thee model booundaries too reflect thhe
temp
perature of so
oils outside the model. The
T temperaature load waas applied oon the surface
of frreezing pipes and is giv
ven in Tablee 3. An incrrement of 244 hours wass used in thhe
proceess of tempeerature load
ding. In ordeer to verify tthe calculateed results obbtained from
m
finitee element modeling,
m
thee numerical results werre comparedd to the in-siitu data. Thhe
temp
perature at recording po
oints 1 and 2 were com
mpared to thoose at the coorrespondingg
pointts in the fielld as shown in Fig. 4. The
T temperatture curves for the num
merical resultts
weree relatively smoother
s
than those of the field daata. It is attrributable to the fact thaat
numeerical simullation idealiizes the pro
oblem param
meters and simplifies the complexx
boun
ndary and lo
oading conditions. Neveertheless, thee numericall simulation results werre
geneerally consisttent with thee field data.
Tablee 2. Parameteers of cement--admixed soills
Soils
Density
y
(kg/m3)
Moisture
content (%)
Cement soiil
2040
11.6
0
30
1
0
Thermal cond
ductivity
(KJm-1d-11C-1)
Unfrozen soil
Frozen
F
soil
180
205
Specific heat cappacity

(KJKg-1C-1)
Unnfrozen soil
Frrozen soil
1.60
1.66
Latentt heat of phase

change
(
108J/m3)
0.81
Table 3. Cooling program

m of brine
Time - day
Temperaature - C
ASCE
5
-12
10
-20
15
-25
20
-28
300
-300
40
-28
C
Cooling
tem
mperature
raange(C)
[-2,0]
161
FIIG. 4.The com

mparison of simulation ressults and meaasured data
RES
SULTS
The
T isotherm
ms at 0C forr the overalll model andd cross-sectioon z=0 at diifferent timees
are shown
s
in Fig
g. 5. As show
wn in Fig. 5,, the isotherm
m at 0C in the early freeezing period
(the first five daays) consisteed of circles centered aat the corressponding freeezing pipess,
out intersection betweeen adjacent circles. Thee cylindricall frost zone surrounding
witho
each freezing pip
pe in Row A was larger than that in Row B. Thiis can be attrributed to thhe
fact that
t the pipees in Row B were surrou
unded by moore soil, whicch offers moore resistancce
to tem
mperature ch
hange. Afterr freezing fo
or six days (F
Fig. 5b), thee frozen zonees around thhe
freezze pipes begaan to merge;; however, th
he isotherm at 0C for thhe overall m
model showed
that the
t frozen zo
ones had yett not fully co
onnected. Affter freezing for seven daays (Fig. 5c)),
a co
ontinuous veertical frozen wall had formed. Thhe growth of the frozen wall waas
contiinued until the 40th daay (Fig. 5d)). After the merge, thee arced isothherm at 0C
C
gradu
ually linearized. Additio
onally, the ex
xpansion of tthe frozen aarea progresssively slowed
but the
t thicknesss of the fro
ozen wall co
ontinued to increase. T
Thus, with reespect to thhe
freezzing scheme, it takes 7 days to form
m closure beetween freezze pipes givving rise to a
comp
plete frozen soil wall. In
n Fig. 6, the isotherms att 0C and -10C after freeezing for 30
and 40 days fo
or both the overall mo
odel and crross sectionn z=0areillusstrated. It iis
onstrated thaat the thickn
ness of the frrozen wall aassociated wiith the isotherm at -10C
C
demo
reach
hed 1.6m aft
fter freezing for 30 dayss; furthermorre, the thickkness associaated with thhe
isoth
herm at 0Cw
was 0.2-0.3m
m thicker thaan that at -100C. After frreezing for 440 days, therre
was only a sligh
ht increase of the thickneess associateed with the isotherm at -10C, up to
aroun
nd 1.8m; th
he thickness associated with
w the isootherm at 0C was 0.3-0.4m thickeer
than that at -10C. Thereforre, the thick
kness of the frozen walll was greateer than 1.6m
m
after vertical freeezing for 30 days.
a)freezing forr 5 days b)ffreezing for 6 days c)freeezing for 7 daays d)freezinng for 40 dayys
FIG
G. 5. Isotherm
m at 0C for the
t overall mo
odel and crosss section z = 0 at differentt time points
ASCE
162
a)freezin
ng for 30 dayss
b)freeezing for 40 ddays
FIG. 6. Isothermss at 0C (Curv

ve 1) and -10C (Curve 2) for the overaall
mo
odel and cross section z = 0 after 30 and
d 40 days of ffreezing.
To
T examine the cooling trend of th
he frozen sooils betweenn Rows A annd B, a pathh
betw
ween the two rows (deno
oted as Path 1) was speciified as paraallel to y-axiis. Recording
pointts 1-9 were arranged on
n Path 1, wiith Y coordiinates of 0, 1.6, 3.2, 4.88, 6.4, 8, 9.66,
11.2,, and 12m, respectively
y. Temperatture-time cuurves for eaach recordinng points arre
preseented in Fig. 7. It is ind
dicated that there
t
was a nearly identtical coolingg tendency oof
the curves
c
for reecording points 1-7.It was
w also fouund the tem
mperature on these pointts
dropp
ped below 0C
0 after freezing for 6 days, and thhe frozen soiils between tthe two row
ws
starteed to form closure
c
after freezing forr 7 days. Thiis implies thhe continuouus frozen walll
was completely constructed.. For points 8 and 9, it ttook 9 and 221 freezing ddays to reachh
0C, respectively
y; the changee in temperaature was sloower than thoose for otherr points. Thiis
he fact that there were less
l
freezingg pipes surroounding thesse two pointts
may be due to th
than for other po
oints as show
wn in Fig. 3b.For
3
exam
mple, the disttance from ppoint 9 to itts
neareest freezing pipe was 0.8
8 m. The ressults indicatte that the efffective workking distancce
of a freezing pip
pe is about 0.5 m; in other
o
words,, the distancce between two freezingg
pipess should not be larger thaan 1 m.
FIG.
F
7.The temperature cu
urves for poin
nts
19
FIG. 88.Temperaturre distributioon curves of

d
different pointts on Path 1. (d = day)
Th
he points in Path 1 were further exam
mined basedd on their disstance from the centre oof
the tu
unnel. The temperature
t
curves at different freezzing time aree plotted in F
Fig. 8. Usingg
thesee curves, two regions were
w
classifieed: points 1--7 were desiignated as reegion I whille
the rest
r
made up
p region II. As shown in Fig. 8, tthe rate of ttemperature decline waas
greattest during the
t first 15 days than the rate theereafter. This is mainly because thhe
ASCE
163
temp
perature of the brine was
w maintain
ned at -30C
C, and the heat exchaange is morre
inten
nsive in periiods with a larger temp
perature disccrepancy beetween the bbrine and itts
surro
ounding soilss. Thus, the decrease in temperaturee of the surroounding soills reduces thhe
rate of temperatu
ure decline. The temperrature in reggion 1 droppped to about -20C at thhe
15th day. Region
n 2 was located at the ed
dge of Row A
A, and the teemperature iin this regionn
chan
nges significantly slowerr than otherr points as a result of thhe phenomeena discussed
previiously. The distance
d
betw
ween points 7 and 8 wass approximat
ately 1/14 of the length oof
Row
w A. This imp
plies that a marginal
m
disttance of 1/144 of the freezing length should be seet
asidee during the design of vertical
v
freezze projects. T
Thus, one orr two additioonal freezing
pipess at each side boundarry should be
b installed to ensure the marginaal zones arre
accou
unted for. To
T examine the
t cooling trend
t
along the depth grradient, anotther part waas
chosen (denoted
d as Path 2)) and includ
ded points 11, 10-15, ass shown in Fig. 2c. Thhe
perature-timee curves for the 7 record
ding points aare shown inn Fig. 9. It iis shown thaat
temp
the cooling
c
rate was
w basically identical for
f points 1 aand 10 to 133. The tempeerature for alll
thesee points dropped below
w zero after freezing forr 6 days. Att this time, a continuouus
vertical frozen wall
w was form
med as the frozen
fr
zones around the freeze pipess of two row
ws
weree totally merg
ged. For poiint 14, the teemperature ddropped beloow zero afterr freezing foor
abou
ut 10 days. As
A it was furtther away fro
om the freezze pipe, poinnt 15 droppeed below zero
after freezing for 30 days. The
T distancee between pooint 14 and 15 was 1 m
m. The resultts
imply
y that a marrginal distan
nce of at leaast 1 m alonng the z-axiss should set aside for thhe
desig
gn. Similar finding
f
can be found by
y plotting thhe points in P
Part 2 accorrding to theiir
distaance to the grround level (Fig.
(
11).
FIG. 9. Thee temperaturee curves for

po
oints 1, 10 to 15
1
FIG. 100.Temperature distributioon curves of

diifferent pointts on Path 2. ((d = day)
To
T profile th
he cooling raate of the veertical frozenn wall alongg the tunnelling directionn
(i.e. x-axis in Fig. 3), 8 reco
ording points were seleccted (denoteed as Path 3)) at the crosss
2 with x-cooordinates 0, --0.4, -0.8,-1.2, -1.5, -2.00,
section z=0, inclluding pointss 1, 16 to 22
a shown in Fig.
F 3b. Pathh 3 was locaated in the central part oof
-2.5 and -3.0, resspectively, as
fr
wall perpendicula
p
ar to the plan
ne y=0. The cooling ratees for points in Path 3 arre
the frozen
show
wn in Fig. 11
1, in which, it can be seeen that the ccooling curvees differed ffrom those oof
Pathss 1 and 2. The
T cooling-rrate curve of point 17 w
was the samee as the coolling program
m
of brrine as it waas 0.4m awaay from the heading facce. For otherr points, thee cooling-ratte
ASCE
164
curvee demonstraated that the further awaay from the heading facce, the lowerr the cooling
rate and the high
her temperaature. The teemperature oof points 166, 1, and 18 between thhe
he heading face
f
droppedd below 0C
C after freeziing for 6 to 7
freezze pipes in row B and th
days forming a continuous
c
vertical
v
frozeen wall. Thee temperaturee of point 199, which waas
0.3m
m away from
m the freeze pipes in Ro
ow B along tthe x-axis, ddropped beloow 0C afteer
freezzing for 10 days.
d
Point 20
2 took23 daays to drop bbelow 0C. E
Even after freeezing for 40
days, the temperrature did not
n drop to below 0C for points 221 and 22, which had a
nd 1.8m, resspectively, frrom the freezzing pipes inn Row B aloong the x-axiis.
distaance of 1.3 an
Figure 12 gives the cooling rates at diffferent time ppoints. The ppoints alongg Path 3 werre
arran
nged based on
o their disttance from th
he heading face. These curves can be classified
into four regionss: I, II, III an
nd IV. The freeze
fr
pipes in rows A aand B were llocated at thhe
respeective centerrs of regionss I and II. In region I, thee temperaturre at differennt time pointts
was basically
b
sym
mmetrical arround the freeeze pipes inn row A. Hoowever, in coontrast to thhe
trend
ds exhibited in region I, the temperaature in regioon II was low
wer at the siide closest to
the freeze
f
pipes in row A du
ue to influen
nce of freezee pipes in roow A. Becauuse region IV
V
was far away fro
om the freezee pipes, the temperature
t
did not decrrease signifiicantly, and iit
can be
b seen thatt even pointts in this reg
gion closestt to the freeze pipes didd not displayy
signiificantly low
wer temperattures. Overall, the coolinng rate of sooils on Pathh 3 decreased
gradu
ually, especcially after freezing fo
or 20 days.. After freeezing for 110 days, thhe
temp
perature with
hin 1.5 m of heading facee dropped beelow 0C. A
After freezingg for 40 dayss,
the teemperature within
w
2.4 m of heading face droppeed below 0C
C.
FIG. 11
1. The temperrature curvess
for points 1, 16 to 22
FIG. 112. Temperatu

ure distributiion curves
of d
different poin
nts on Path 3. (d = day)
NCLUSION
NS
CON
The
T distributtion of the temperature field in a veertical frozeen soil wall for the TBM
M
break
k-in of a larrge diameterr shield cross-river tunnnel was studdied using A
ADINA. Thhe
concclusions can be
b summarizzed as follow
ws: (1) The ttrend of the calculated teemperature iis
in geeneral congru
uence with the
t measured
d data. Furthhermore, the temperaturee curve based
on th
he calculated
d data seemss to be relatiively smoothh. This indiccates that it is feasible to
use a numerical model to sim
mulate the temperature change of a vertical froozen soil walll
durin
ng the launcching of a sh
hield tunnel.. (2) In the designed freeezing schem
me, it took 7
ASCE
days to form the closure of the frozen soil wall. After a freezing time of 30days, a
water-resistant frozen soil wall with a thickness of greater than 1.6 m was formed. (3)
There was a significant impact due to the heat exchange between the unfrozen soil and
frozen soil wall on both sides of the frozen soil wall. The effective zone of a freezing
pipe was about 0.5 m, and as a result, the spacing of freezing pipes is recommended to
be less than 1 m. (4) A marginal zone should be set aside in design. One or two
additional freezing pipes along each row should be installed to ensure the generation of a
marginal zone at each side boundary. For the z-axis, the marginal zone should be at least
1 m in depth, which implies the freezing pipes should reach the depth at least 1 m deeper
than their designated positions.
ACKNOWLEDGMENTS
This research was financially supported by the China Postdoctoral Science Foundation
funded project (2015M580559), the Scientific Research Foundation for the Returned
Overseas Chinese Scholars ([2014]-240) and the Scientific Research Project of
Education Department of Hainan Province (Hnky2015-10).
REFERENCES
Andersland O.B. and Ladanyi, B. (2004). Frozen ground engineering. American Society
of Civil Engineering, John Wiley & Sons Inc., Hoboken, New Jersey.
Bathe K. J. and Dong, J. (2014). Component mode synthesis with subspace iterations
for controlled accuracy of frequency and mode shape solutions. Computers and
Structures, 139: 28-32.
Hu J. (2012) Study on the reinforcement methods of subway large-diameter shield
launching in the sandy clay with high water pressure. PhD thesis Nanjing: Nanjing
Forestry University. (in Chinese)
Nakamura T., Imaishi T., Eguchi H. and Nishida, Y. (2004). Reducing swelling pressure
on the freezing method of ground treatment by drilling and coring measures for
tunnelling works.Tunnelling and Underground Space Technology, 19: 499-450.
Ou C.Y., Kao C.C. and Chen C.I. (2009). Performance and analysis of artificial ground
freezing in the shield tunnelling.Journal of GeoEngineering, 4(1): 29-40.
Pimentel P., Papakonstantinou S. and Anagnostou G. (2012). Numerical interpretation
of temperature distributions from three ground freezing applications in urban
tunnelling.Tunnelling and Underground Space Technology, 28: 57-69.
Yang P. Ke J.M., Wang J.G., Chow Y.K. and Zhu F.B. (2006). Numerical simulation of
frost heave with coupled water freezing, temperature and stress fields in tunnel
excavation.Computers and Geotechnics, 33: 330-340.
Yu Z. K., Huang H. W. and Wang R. L. (2005). Application of the artificially ground
freezing method to shanghai metro engineering.Journal of Glaciology and
Geocryology, 27(4), 550556.
ASCE
165
166
Improvement of Off-Shore Immersed Tunnel Foundations in Soft Ground by

Composite Foundation Methods
Hongtao He1; Yougao Lin2; and Jianyu Li3
1
Engineer, CCCC-FHDI Engineering Co., Ltd, Guangzhou 510230, P.R. China. E-mail:
heht@fhdigz.com
2
Professorate Senior Engineer, CCCC-FHDI Engineering Co., Ltd, Guangzhou 510230, P.R. China. Email: linyg@fhdigz.com
3
Senior Engineer, CCCC-FHDI Engineering Co., Ltd, Guangzhou 510230, P.R. China. E-mail:
lijianyu@fhdigz.com
Abstract: The off-shore immersed tunnel is placed at a varying level and

consequently has to accommodate complicated foundation condition. The segmental
type of tunnel is sensitive to the differential longitudinal settlement. Therefore, a
number of ground improvement methods are considered in order to provide an
appropriate foundation for the type of immersed tunnel in soft ground. The composite
foundation method is one of techniques which can reduce settlement, increase bearing
capacity and slope stability. The composite foundation methods included sand
compaction pile (SCP), cement deep mixing (CDM), and jet grouting pile, etc. Based
on the Island and Tunnel Project of Hong Kong-Zhuhai-Macao Bridge in China and
the Busan-Geoje Fixed Link Project in South Korea, this paper describes design and
construction in soft ground by the SCP method mainly for off-shore immersed tunnel
foundation.
INTRODUCTION
The sand compaction pile (SCP) method is a technique by placing granular material
(usually sand) into a ground through a casing pipe, and the sand is compacted by
either vibration, dynamic impact or static excitation to construct a compacted sand pile
in soils (Kitazume, M., 2005). This method can be applied to soft clay soils which is a
reinforcement of composite ground consisting of compacted sand piles and
surrounding clay. The improvement purpose is to assure stability, reduce ground
settlement, enhance drainage function, and increase horizontal resistance of pile. This
method can be also applied to sandy soils or soft clay soils. In this paper, the
improvement patterns, design procedures, implementation, quality control and cases
study for clay soils are described in details.
The SCP method can be categorized into on-land construction and marine
construction, respectively. For on-land construction, the SCP method has been applied
ASCE
167
to embankments, storage yard, and building foundations, while the SCP method has
been applied to breakwaters, sea revetments and piers in marine construction. For the
marine construction, the maximum improvement depth of the method increases to
70m and the diameter ranges from 1.6m to 2.0m. The replacement area ratio for
applications on clayey soils is ranged from 0.3 to 0.8. Moreover, the SCP marine
construction method has been applied successfully to the immersed tunnel foundation
in the Island and Tunnel Project of Hong Kong-Zhuhai-Macao Bridge in China and
Busan-Geoje Fixed Link Project in South Korea.
IMPROVEMENT PATTERNS FOR CLAY GROUND
Improvement Patterns
Clayey soils improved by the SCP method can be considered to be a composite
ground consisting of compacted sand piles and surrounding clay as schematically
shown in FIG. 1.
pp
p
As
ps
Clay
Sand Pile
Clay
Sand Pile
Clay
A
FIG. 1. Schematic of composite foundation consisting of compacted sand piles

and surrounding clay.
The replacement area ratio, m, is defined as the ratio of the sectional area of the sand
piles, As, to the improving ground area, A, which is formulated as Equation (1).
m=
As
A
(1)
Because the compressive stiffness of sand piles is much higher than that of the
surrounding clay ground, more external load is applied on the top of sand piles, while
relatively less load is applied on the clay among the sand piles. The stress
concentration ratio, n, is defined as the ratio of the vertical stress acting on sand piles,
pp, to that on the surrounding clay, ps. A simple analytical approach provides a
formulation to calculate above stresses as expressed in Equations (2) to (4), which are
derived from the stress equilibrium between sand piles and surrounding clay.
n=
ASCE
pp
ps
(2)
np
pp =
= p p
1 + (n 1)m
(3)
168
ps =
p
= s p
1 + ( n 1) m
(4)
where m is replacement area ratio; A is cross sectional area of clay ground and sand
pile (m2); As is cross sectional area of sand pile (m2); n is stress concentration ratio; p
is average vertical stress (kN/m2); ps is vertical stress on clay ground (kN/m2); pp is
vertical stress on sand pile (kN/m2).
Improvement Principle
The improvement principle of the SCP method for soft clay ground is described in
the following aspects:
(1) Replacement:
For the soft clay ground, sand compaction piles are formed by driving a casing into
the soft ground to a predetermined depth. Sand is then forced through the casing into
the ground to displace the soft soils, and thereby a dense and large-diameter sand piles
are being formed.
(2) Compaction
The sand is then compacted in the casing. The effect of its density of the compacted
sand should be taken into account since it can affect the stability and bearing capacity
of the ground, especially for low replacement ratios lower than 40%.
(3) Consolidation
The compacted sand piles can function as a drainage path to accelerate the
consolidation process in the soft clay ground. Combined with the preloading method,
the ultimate settlement of the soft clay ground with compacted sand piles will be
reduced in long operation period.
DESIGN PROCEDURES OF IMMERSED TUNNEL FOUNDATION
Design Criterion
The design of SCP method improved clay ground should meet the requirement for
both total settlement and differential settlement for tunnel foundation, and consider the
load on tunnel structure and the deformation induced by stiffness difference thus
incurred.
Design Procedures
(1) The topographic, bathymetric survey strata, engineering geology and
hydrological data in this site area is needed to be investigated and collected.
(2) The loading condition, stiffness, bearing capacity and settlement requirements is
determined as the input design parameters.
(3) Preliminary calculation for bearing capacity and settlement of natural foundation
under operation loading condition is needed to determine the improved area.
(4) Based on the construction condition, preliminary scheme with diameter, layout,
replacement area ratio and improved depth is proposed. The stability, bearing
capacity, settlement and consolidation for preliminary scheme is analyzed.
ASCE
169
IMPLEMENTATION AND QUALITY CONTROL IN IMMERSED TUNNEL

FOUNDATION
Material
The most suitable material for the SCP method is well blended granular material.
However, in the case where suitable soil cannot be obtained due to economic and/or
environmental reasons, a soil with poor quality can be also used after confirming
acceptable performance and execution ability. Applicable grain size range of filling
materials for SCP method can be showed in FIG. 2.
FIG. 2. Applicable grain size range of filling materials.

Field Trial Test
It is advisable to conduct a field trial test in advance at a ground or adjacent to the
construction site, in order to ensure the quality of improvement and smooth execution
at the site.
Implementation Procedures of Sand Compaction Pile Method
Sand compaction piles are typically installed by inserting a closed-ended casing
filled with sand into the soil mass that needs to be treated. The casing is driven to its
required depth by means of a vibrator mounted on top. During partial withdrawal of
the casing, its closed-end opens up, releasing the sand. The casing is redriven and the
sand is compacted by the vibrations of the casing. This process is repeated until
completion of the pile. The procedure for constructing compacted sand piles is shown
in FIG. 3.
FIG. 3. Implementation procedure of the Sand Compaction Pile method.
ASCE
Quality Control
The pile shaft must be continuous and complete, and the pile diameter, depth and
pile strength must comply with design requirement. Similarly, the specifications of
pile material must comply with design requirement.
CASES STUDY
Hong Kong-Zhuhai-Macao Bridge
The Hong Kong-Zhuhai-Macao Bridge (HZMB), crossing the Pearl River Estuary
and linking Hong Kong to the East, and Zhuhai and Macao to the West, is a
combination of bridges, a tunnel and artificial islands. The immersed tunnel, with a
length of approximately 6 km, will become the longest one in the world. Currently the
Hong Kong-Zhuhai-Macao Bridge Link is under construction.
Foundation
The longitudinal profile of the tunnel is shown in FIG. 4. The tunnel will be placed
at a very deep depth, and consequently has to accommodate large water and ground
loads, varying soft soil conditions. Ground conditions at the project location (transition
between tunnel and artificial islands) are relatively poor, whereas the ground profile
varies considerably over a relatively short distance. Ground improvement is required
over a considerable part of the tunnel alignment.
FIG. 4. Indicative longitudinal profile of the immersed tunnel

Solution
To promote smooth transition from one tunnel part (e.g. piled cut and cover tunnel
at the islands) to the other (e.g. immersed tunnel), sand compaction piles (SCP), jetgrouting piles, and PHC piles are applied to ground improvement. The objective of the
ground improvement is to improve the foundation conditions for the tunnel. In this
way the ultimate settlements and differential settlements can be limited as well as the
internal forces in the tunnel.
In the transition between immersed tunnel and artificial islands, two SCP schemes
were adopted for the ground improvement. For additional stress less than zero at
bottom of immersed tunnel, the diameter of the sand compaction piles ranges from
1.5m to 1.6m, and the column grid was 1.8m 1.8 m with replacement area ratio of
55% to 62%. For additional stress greater than zero at bottom of immersed tunnel,
SCPs combined with sand drains preloading are adopted. Some soft soils were
replaced by SCP which was also considered as the drain path for consolidation. The
typical cross section of SCP is shown in FIG. 5.
ASCE
170
171
FIG. 5. Typical cross section of sand compaction pile

Busan-Geoje Fixed Link
The Busan-Geoje fixed link involves the construction of an 8.2 km motorway
connecting Busan, Koreas southernmost and second largest city, to the island of
Geoje. The connection includes an immersed tunnel - one of the longest and deepest in
the world. The total length of the immersed tunnel is 4 km with two 170 m long cut
and cover sections at both ends. About two thirds of the immersed tunnel (16 elements)
lies in an approximately 15 m deep trench below the seabed (see FIG. 6). The
maximum depth of the seabed is about 40 m below the sea level.
FIG. 6. Longitudinal cross section of immersed tunnel of Busan-Geoje Fixed

Link project
Foundation
The foundation of the tunnel is formed by a thick layer of marine clay down to
about 30 m below the seabed, followed by a layer of sand and gravel, underlain by
weathered rock. The properties of the marine clay require special attention. Normally,
clay deposits exhibit some degree of overconsolidation due to ageing. On the basis of
the available research data, such an effect does not exist for this marine clay. A
remarkable feature was that the clay exhibited fairly stiff behaviour at low stress levels,
whereas the deformations increased sharply at higher stresses (see Table 1).
Eventually the clay was classified as structured clay.
Table 1. Soil Parameters of Marine Clay
ASCE
Soil Type
Weight
(kN/m3)
Void
Ratio e0
Marine Clay
14.7
2.44
Compress
ion Index
Cc
1.25
Expansio
n Index
Cs
0.091
Cohesi
on c
(kPa)
3
Friction
Angle
()
25
However, a heavy backfill on and beside the tunnel is necessary to offer sufficient
resistance to wave pressures. Consequently, the soil stress below the tunnel increases
beyond the limit stress, thus significantly increasing the settlements of the tunnel. This
can be expected to be associated with the large differential settlement. These in turn
may lead to the opening of joints, possibly beyond the capacity of the waterstops.
Solution
To remove the uncertainties described above with respect to the settlements,
different soil improvement techniques had been used. At the ends of the tunnel, the
clay layer under the tunnel was relatively thin. Here the clay was replaced by the rock
fill. Along the major part of the tunnel route, the tunnel was founded on a grid of
columns formed in soils using the Cement Deep Mixing (CDM) method. At the
western end, the tunnel will be resting in a dam construction built on the sea floor. The
CDM method is considered less suitable here, as it may cause disks to be created,
which would form a potential slip surface during an earthquake. The solution chosen
herein is the use of sand columns according to the Sand Compaction Pile (SCP)
marine construction process. FIG. 6 shows the longitudinal cross section of immersed
tunnel of Busan-Geoje Fixed Link project.
FIG. 7. Typical cross section of immersed tunnel with SCP foundation

These sand columns, with diameters ranging from 1.2m to 2.0m, are inserted to a
record depth of approximate 60m. Two different diameters are adopted in each single
sand pile for upper layer and lower layer as shown in FIG. 7. The sand column has
diameter of 2.0m at upper layer, and the column grid is 3.0m 2.6m with replacement
area ratio of 40%; the sand column has diameter of 1.6 m at lower layer, and the
column grid is 3.0m 2.6m with replacement area ratio of 26%. The dam is built onto
the column grid with some excess height, so that the settlement will be very small
after placement of the tunnel elements.
ASCE
172
DISCUSSIONS
(1) The SCP method is the reinforcement of composite ground consisting of
compacted sand piles and surrounding clay, which is different from the sand drain
method in which sand piles without any compaction are constructed principally
for drainage. The improvement principle of the SCP method for soft clay ground
has replacement, compaction, consolidation, and etc.
(2) The primary advantages of the SCP method are, a) applicable to a variety of
ground, including sandy and clayey ground; b) wide range of filling materials are
usable; c) higher rate of construction than conventional methods; d) sand, which is
often considerably cheaper than stone, in used in construction compared to Stone
Column method.
(3) However, compared to Stone Column method, Cement Deep Mixing method, and
Rigid Concrete Column method, the SCP method also have several disadvantages:
a) the SCP method with a high replacement area ratio is quite expensive and
causes the problem of how to dispose of the upheaval soil; b) they have a lower
angle of internal friction and a lower stiffness than Stone Columns.
CONCLUSIONS
(1) The SCP method can be applied to soft clay ground which is a reinforcement of
composite ground consisting of compacted sand piles and surrounding clay.
(2) Some important issues of the SCP improved clay ground for immersed tunnel
have been discussed, which include the improvement patterns, design procedures,
implementation and quality control, etc.
(3) The SCP marine construction method has been applied successfully to the
immersed tunnel foundation in Hong Kong-Zhuhai-Macao Bridge in China and
Busan-Geoje fixed link project in South Korea. The SCP method is demonstrated
to be able to reduce the settlements and improve the stability of the tunnel.
(4) With some advantages and disadvantages, the SCP method is still an excellent
method for improving stability, reducing ground settlement, enhancing drainage
function, and increasing horizontal resistance applications.
REFERENCES
G. Meinhardt, and R.M.W.G. Heijmans. (2008). The off-shore immersed tunnel in
the Busan-Geoje Fixed Link project in South Korea. GEOtechniek, 2008:18-23.
GONG Xiao-nan. (2007). Theory of composite foundation and engineering
application (Second edition). Beijing: China Architecture and Building Press, (in
Chinese)
Kitazume, M. (2005). "The Sand Compaction Pile Method." Taylor & Francis.
Port and Harbours Bureau, et al. (2009). Technical standards and commentaries for
port and harbor facilities in Japan. Japan: The overseas coastal area development
institute of Japan.
Yeoward A, Sig Koo I, Fraser D. (2010). Design and Construction of Busan-Geoje
Fixed Link, S. Korea. Bridge Engineering, Vol.163, 2010 (BE2):59-66.
ASCE
173
174
The Excavation Damage Zone of a Subway Tunnel Considering Stress Release

Feng Huang1; Hehua Zhu2; Shuping Jiang3; and Bo Liang4
1
Associate Professor, School of Civil Engineering, Chongqing Jiaotong Univ., Chongqing 400074,
China. E-mail: huangfeng216@126.com
2
Professor, School of Civil Engineering, Tongji Univ., Shanghai 200092, China. E-mail:
zhuhehua@tongji.edu.cn
3
Professor, China Merchants Chongqing Communications Research & Design Institute Co., Ltd.,
Chongqing 400067, China. E-mail: Jiangshuping@ccrdi.cmhk.com
4
Professor, School of Civil Engineering, Chongqing Jiaotong Univ., Chongqing 400074, China. E-mail:
liang_laoshi@126.com
Abstract: The excavation damaged/disturbed zone (EDZ) is significantly important

for determining the loosening load of rock mass surrounding tunnel. While the main
search was focused on the hard rock in the past, this paper presents a study on the soft
ground of the longest subway tunnel constructed by subsurface excavation in China.
Based on the damage constitutive laws, the method of numerical analysis was funded
by finite element, which was validated by comparing with the laboratory tests. Some
results can be concluded by numerical analysis. The EDZ of rock mass is an
approximate circle around tunnel for the isotropic initial stress of the ground, and it
will initiate when the ratio of stress release is more than 0.5. Besides, the EDZ would
suddenly increase as the ratio of stress release is over 0.9. There will be an obvious
stress concentration and deformation gradient in the EZD.
INTRODUCTION
The support of tunnel is usually not followed so closely or enough strongly that the
rock mass (especially for soft rock) near tunnel surface would become unstable
severely after excavation during construction (Bizjak, 2004). In general, tunnel
excavation will lead to excavation damaged/disturbed zone (EDZ) and loosing load on
the structure in the meantime. In the excavation process, the load releases
continuously in the stress adjustment process till the structure and rock mass
stabilizing (Sakurai, 1978). Therefore, the assessment of EDZ around tunnel after
excavation is significantly important to construct tunnels well (Martino, 2004).
The characteristics of EDZ would greatly depend on the geological conditions
(Malmgren, 2007). In the past, we would consider that the damage zones mainly occur
in brittle or quasi brittle, which too many study of EDZ using damage mechanics paid
more attention hard rock mass (Cai, 2004). In fact, the damage in soft rock is very
evident, such as the porosity ratio reduction, the strain softening, even shear band
ASCE
175
phenomena. In other words, the failure of soft rock is a progressive process with a lot
of plastic strain, so the damage mechanics is greatly fit to studying the failure process
of soft rock (Wang, 2010). Therefore, the evolution process of EDZ of tunnel in soft
ground is obviously progressive, which can be described by the ratio of stress release.
E.g. Popp (2008) employed an elasto-plastic constitutive model considering the
hardening/softening to study the spatial development of EDZ around a drift at the
Mont Terri site. Li (2011) carried out numerical analysis and model test to research the
load releasing process of the whole section in the construction of ultra-large section
railway tunnel.
In this paper, some numerical simulations are carried out to study the EDZ of the
longest subway tunnel constructed by subsurface excavation in China and are made up
of some cases of different ratio of stress release by the finite element method based on
the elasto-plastic constitutive laws.
ENGINEERING BACKGROUND
60 m
The Tongluoshan tunnel locates in Chongqing City of China, which is a typically

interval tunnel and the part of subway NO.6. The whole length of tunnel is 5633*2 m,
which is longest tunnel of interval subway in China constructed by the method of
subsurface excavation. By the comparison of plans, the method synthesizing the tunnel
boring machine (TBM) and drill-blasting has been adopted. The section with TBM is
about 2736*2 m for the ground of relatively intact and hard ground. On the other hand,
the tunnel will go through the soft and weak ground of the length of about 2897*2 m
including karst zone and fault zone treated by the method of drill-blasting.
The section of fault fracture zone within the soft ground was chosen as the study
background, which is the length of about 50 m. The ground is mainly covered with
strongly weathered sandstone, which can be defined as the grade according to the
design codes of railway tunnel in China. The tunnel is a typically separated tunnel
with the crossing section of diameter of 6 m and altitude height from 268 m to 273 m.
By considering the boundary effects and Saint-Venant's Principle, the dimension of
being numerically simulated is finally determined to be what is shown in Fig.1.
6m
120 m
Fig. 1 The diagram of dimension of tunnel and ground

THE METHOD OF NUMERICAL SIMULATION
The constitution laws
The numerical simulations will be performed using the finite element software
ABAQUS. A plastic damage constitution model is adopted with a criterion of
ASCE
176
Drucker-Prager yield and recited as following, which has been proved to be reasonable
for study strain softening and progressive failure of soft rock mass (Huang, 2013).
If the stress estimation was beyond yield criterion, it appeared that the material
damage initiated with plastic stain. According to strain equivalent law proposed by
Lemaitre (Lemaitre, 1983), the Cauchy stress ij will be replaced by effective
stress ~ , while material stiffness is degraded progressively according to the specified
ij
damage evolution response. The relationships can be defined as following form:

~
~ij = (1 D ) ij , E ij = (1 D) Eij
(1)
where, E~ ij is the renewed elastic stiffness, D is the value of damage variable, which
generally ranges from 0 to 1. Considering the residual strength of rock mass in
uniaxial compression test, the maximum of damage variable is set as 0.8.
Thus, the initialization and evolution of damage variable is critical specification
of the model. The initiative damage is determined by the equivalent plastic strain 0pl ,
which is obtained by Equations (3). The damage variable evolutes in a linear form
defined by equivalent plastic strain, which is given as
D=
pl 0pl
fpl 0pl
(2)
where, fpl is equivalent plastic strain when the damage variable reaches the
maximum. In the failure process of the model material, there are linear elastic, plastic
hardening, and plastic softening phases. The relations between strain values at
different phases are proposed as following (Aydan, 1993):
(3)
opl = f ( c , e ) = 2 c0.17 e , fpl = f ( c , e ) = 5 c0.32 e
where, the uniaxial compression strength c can be obtained from Table 1 and the
maximum elastic strain e can be automatically passed by ABAQUS during
calculation iteration.
The parameters of rock from tests
There are 17 rock samples with the diameter of 50mm and height of 100mm from
the field rock mass near tunnel. They are tested in laboratory under unixial and
common triaxial compression loads for elastic and plastic parameters which are shown
in Table 1.
Table 1 The mechanical parameters of rock
Bulk weight Elastic modulus Poissons ratio Cohesion Friction angle
UCS
3
c /MPa
/ kN/m
E/ MPa
c/ MPa
/
7.32
20
1304
0.32
2.2
32
For validating the damage constitution laws, a numerical simulation of unixial
compression test (UCT) is carried out, which is made up of 5000 elements with the
same dimension of rock sample in laboratory test as shown in Fig.2. The comparison
between numerical simulation and test results are shown in Fig 3. The results error
between laboratory test and numerical calculation are acceptable, which indicates the
parameters and the constitution laws are reasonable and believable for the further
study.
ASCE
177
8
7
unixial stress /MPa
6
5
4
3
elastical strain
strain hardening
strain softening
plastic flow
lab test
2
1
0
0.0
0.2
0.4
0.6
0.8
1.0
1.2
unixial displacement /mm
Fig. 2 The numerical mode for UCT
Fig. 3 The comparison of test and

calculation
The FE model and scheme

A kind of 8-node linear brick elements was introduced for rock mass surrounding
tunnel as shown in Fig. 4, where none of tunnel supporting is considered. The
dimension of numerical simulation is 50m120m90m, which is the same as the field
ground shown in Fig. 1. The excavation will be completed by full section tunneling
and the only middle section of the model will be analyzed.
Fig.4 The FE mesh for numerical simulation Fig. 5 The diagram of stress release
The whole process of stress release of rock mass will be completed by 2 steps as
following: (i) the reaction forces of node denoted by RF in equation (4) are
outputted. (ii) The negative forces p(z) are loaded to the nodes by the ratio of stress
release denoted by r in equation (4), which is shown in Fig. 5.
(4)
p( z ) = [1 r ( z , p)] RF
ASCE
178
ASSESSMENT OF EXCAVATION DAMAGED ZONE

The evolutions of EDZ
The isograms of excavation damage zone with the stress release ratio of 1.0 or none
force of nodes reaction are shown in Fig. 6. The red zone means the rock mass is
complete damage zone/failure zone (D=0.8), the blue one means no damage of the
rock mass (D=0), and the others indicate the partial damage of rock mass (0.8>D>0).
So the EDZ (D>0) is equal to the damage zone including the complete damage
zone/failure zone and partial one. Accordingly, the ground after tunneling can be
distinctly divided into 3 parts: failure zone, damage zone and in-situ stress zone. Since
the initial ground stress is symmetrical and isotropic, the EDZ has the distribution rule
of an approximate circle, which is also found in plastic zone as shown in Fig.7.
Fig. 6 The EDZ for the ratio of 1.0
Fig. 7 The plastic zone for the ratio of 1.0
In fact, the formation of excavation damage zone (EDZ) is a progressive process

in the cases of different stress ratio, which is presented in Fig.8, where the R is the
radius of tunnel. As the ratio is more than 0.5, the damage zones will initiate. When
the ratio is more than 0.7, the failure zones will appearance. The damage zones and
failure zones suddenly also increase while the ratio of stress release is over 0.9.
1.4
damage zone
failure zone
the depth of zones/*R
1.2
1.0
0.8
0.6
0.4
0.2
0.0
0.5
0.6
0.7
0.8
0.9
1.0
the ratio of stress release
Fig.8 The depths of zones with the different ratio
When the ratio of stress release is respectively 0.7, 0.8, and 0.9, the distribution
isograms of damage zone are shown in Fig 9. The greater ratio of stress release is, the
more obvious and wider the degree and area of EDZ of rock mass respectively are.
ASCE
179
(a) r = 0.7
(b) r = 0.8
(c) r = 0.9
Fig. 9 The isograms of EDZ with the different ration
The mechanical reactions inner EDZ
The final distribution isograms of Tresca stress (namely the half of maximum of shear
stress) with the stress release ratio of 1.0 are shown in Fig. 10. The ground after
tunneling can be distinctly divided into 3 parts: stress loosening zone, stress bearing
zone (red parts) and in-situ stress zone.
Fig. 10 The isograms of Tresca stress with the ratio of 1.0

the nuerical results based on damage model
the analytical results based on plastic model
the analytical results based on elastic model
the ratio of stress concentration
2.0
1.8
1.6
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0
the distance to surface of tunnel /R
Fig. 11 The curves of stress ratio based on different method
When the ratio of stress release is 1.0, the curves of tangential stress at hance of
tunnel got from different model are presented in Fig.11, where the ratio of stress
concentration K is equal to value tangential stress divided by initial stress in ground.
The result of elastic model is obviously unreasonable for soft ground with none of
ASCE
180
peak value. The results of damage model and the ones of plastic have the analogue
rules, while the effects of excavation of damage model is about 3 times of diameters of
tunnel and remarkable bigger than the ones of plastic.
When the ratio of stress release is 1.0, the comparison between the ratio of
tangential stress concentration K and the value of damage variable D at tunnel hance
is presented in Fig.12. The stress loosening zone almost locates in the failure zone, the
partial damage zone offers the external ground bearing, and the stress gradually
transfers to the in situ state.
1.6
1.4
1.2
failure
zone
1.0
0.8
in situ
zone
partial
damage
zone
0.6
the ratio of stress concentration
0.4
the value of damage viarable D
0.2
0.0
-1
the distance to the surface of tunnel/R
Fig. 12 The comparison of curves of K and D
When the ratio of stress release is 1.0, the comparison between the radial
displacement Ur and the value of damage variable D at tunnel crown is presented in
Fig.13. It is apparent that the gradient of radial displacement in the failure zone is
distinctly greatest and then gradually decreases to a constant in the in situ zone.
the radial displacement/10cm
1.4
1.2
1.0
the value of damage vairable D
the radial displacement Ur
0.8
0.6
failure
zone
0.4
partial
damage
zone
in situ
zone
0.2
0.0
-1
the distance to surface of tunnel/R
Fig. 13 The comparison of curves of Ur and D

CONCLUSIONS
Based on the damage constitutive laws, the method of numerical analysis was
funded by the finite element, which was validated by comparing with the laboratory
tests. Some numerical simulations were carried out for studying the progressive
process and mechanical reaction inner the EDZ with the different ratio of stress release.
ASCE
The EDZ of rock mass is an approximate circle for the isotropic initial stress in the
ground and will initiate when the ratio of stress release is more than 0.5. The EDZ
would suddenly increase while the ratio of stress release is over 0.9. The ground after
tunneling can be divided into 3 parts: failure zone (completely damaged zone), partial
damage zone and in-situ stress zone. There will be an obvious stress concentration and
strong deformation gradient in the damage zone.
ACKNOWLEDGMENTS
This work is supported by the National Natural Science Foundation of China (No.
51308574), the post doctor Science Foundation of China (No. 2014M562286), the
natural science project of Chongqing education commission (No. KJ130404), the
project of Chongqing science and technology commission (No. cstc2013jcyjA30007),
the post doctor Science Foundation of Chongqing Xm2014092 , and the post
doctor daily funds of Chongqing Rc201428.
REFERENCES
Aydan ., Akagi T., Kawamoto T. (1993): The Squeezing Potential of Rocks Around
Tunnels Theory and Prediction, Rock mechanics and rock engineering, 26(2): 137163.
Bizjak K.F. and Petkovsek B. (2004): Displacement analysis of tunnel support in soft
rock around a shallow highway tunnel at Golovec, Eng. geology, 75(3): 89-106.
Cai M. and Kaiser P.K. (2004): Assessment of excavation damaged zone using a
micromechanics model, Tunnelling and Underground Space Technology, 20(6):
301-310.
Huang F., Zhu H.H., Xu Q.W., et al (2013). The effect of weak interlayer on the
failure pattern of rock mass around tunnel-Scaled model tests and numerical
analysis, Tunnelling and underground space technology, 35(4): 207-218.
Lemaitre J. (1983): A continuous damage mechanic model for ductile fracture, J. Eng.
Mater, Tech, 107(1): 83-89
Li S.C., Zhao Y., Li L.P. et al (2011). Study of construction section comprehensive
load releasing process of ultra-large section railway tunnel, Rock and Soil
Mechanics (in Chinese), 32(9): 2845-2851.
Malmgren L., Saiang D., Tyr J., et al (2007): The excavation disturbed zone (EDZ)
at Kiirunavaara mine, Swedenby seismic measurements. Journal of Applied
Geophysics ,61(8): 1-15.
Martino J.B. and Chandler N.A. (2004): Excavation-induced damage studies at the
Underground Research Laboratory, International Journal of Rock Mechanics and
Mining Science, 41(5) : 1413-1426.
Popp T., Salzer K., Minkley W. (2008). Influence of bedding planes to EDZ evolution
and the coupled HM properties of Opalinus Clay, Physics and Chemistry of the
Earth, 33: 374-387.
Sakurai S. (1978). Approximate time-dependent analysis of tunnel support structure
considering progress of tunnel face. International Journal for Numerical and
Analytical Methods in Geomechanics, 2(2): 159-175.
Wang S., Yin X.T., Tang H., et al (2010): A new approach for analyzing circular
tunnel in strain-softening rock masses, International Journal of Rock Mechanics
and Mining Sciences, 47(1): 170-178.
ASCE
181
182
Variation of Slurry Pipe Jacking Force in Coarse and Fine Soils

James C. Ni1; Louis Ge2; and Wen-Chieh Cheng3
1
Professor, Dept. of Civil Engineering, National Taipei Univ. of Technology (Taipei Tech), 1, Sec 3,
Chung-Hsiao E. Rd., Taipei 10608, Taiwan, R.O.C. E-mail: ckni@ntut.edu.tw
2
Professor, Dept. of Civil Engineering, National Taiwan Univ., No. 1, Sec. 4, Roosevelt Rd., Taipei
10617, Taiwan, R.O.C. E-mail: louisge@ntu.edu.tw
3
Formerly Research Associate, College of Engineering, National Taipei Univ. of Technology (Taipei
Tech), 1, Sec. 3, Chung-Hsiao E. Rd., Taipei 10608, Taiwan, R.O.C. E-mail: s2428030@gmail.com
Abstract: Pipe jacking force mainly consists of the penetration resistance from face
pressure and the friction resistance between the pipe and surrounding soil. Jacking
forces depends on following parameters: soil type, groundwater table, overburden,
overcut, lubricant, stoppage, misalignment, length of drive, and intermediate jacking
station (IJS). A slurry shield excavating through alluvium deposit to install a 1.5-m
diameter sewer line was selected in many straight drives with similar conditions of
groundwater table, overburden, overcut, and stoppage. The lubricant distributed among
injection outlets along the pipeline ensures a more uniform distribution of lubricant
around the jacked pipes and a more effective reduction in frictional resistance. This
leaves the soil type and steering correction standing out to reduce the jacking force
required. It can be concluded that the soil classification system developed in separate
study helps interpreting the local variation of jacking force that is resulted from
increasing penetration resistance due to either driving from coarse soil to fine soil or
driving into wood log. This penetration resistance increment is recoverable when
driving condition is reversed. The friction resistance increases significantly if
misalignment is larger than a threshold value of 60 mm, and remains unchanged even if
deviation is corrected. The misalignment effect on friction resistance is not recoverable.
For large diameter pipe jacking with stabilized overcut and sufficient amounts of
lubricant, the face resistance can be significant relative to the friction resistance.
INTRODUCTION
Designing and selecting a pipe jacking system, such as jacking cylinders, traction
frame, thrust block, intermediate jacking station (IJS), pipe material with proper
thickness and sufficient compressive strength, lubrication system, and distance of drive
depend on accurate thrust force estimations (Rahjoo et al. 2012; Khazaei et al. 1994).
Not only site related parameters, i.e., ground conditions, piezometric level, overburden
stress, and soil stress history, but also construction related parameters, i.e., overcutting,
use of lubricant and its volume, stoppage, misalignment, jacking speed, curved
alignment, length of drive, and intermediate jacking station are linked with the thrust
force estimation. However, in fact all the models presented in the literatures (Kastner et
al. 1996; Pellet-Beacour and Kastner 2002; Rogers and Yonan 1992; Rahjoo et al. 2012)
have not yet been fully taken the aforesaid factors into account. Therefore, for tunneling
practice the provision of sufficient jacking capacity is largely based on experiences
from previous projects with similar ground conditions.
ASCE
Jacking load mainly consists of the penetration resistance from face pressure and the
resistance from the friction between the pipe and surrounding soil. The friction force
usually constitutes the major component of jacking load. Since in most shields
penetration resistance is not measured, Pellet-Beacour and Kastner (2002) utilized load
cell attached to the head of shield to better understand the percentage of the face
resistance in comparison with the total jacking load. They concluded that the local
variations of total jacking load are mostly linked to the face resistance and the minima
of total jacking load correspond to very low face load. Despite the minor face resistance,
it can be significant for pipe jacking of large diameter under high groundwater level
(Auld 1982; Stein 2003). The face resistance should be greater than the active earth
pressure to prevent surface settlement and less than the passive earth pressure to exclude
surface heaving. Reducing the torque force based on the face pressure is quite important
at the beginning of pipe jacking from the launching shaft to maintain the correct steering
direction.
On the other hand, the friction resistance can then be determined using friction
coefficient between pipe and surrounding soil in conjunction with the overburden
pressure acting on pipe. The bigger the soil particles, the greater the dilatancy of the
soil, and the higher the friction resistance for the same overcut, i.e., annular space
between the excavated tunnel bore and the outer pipe wall. The friction resistance is
proportional to the length of drive and the use of the lubricant injected into overcut can
effectively reduce the friction resistance during pipe jacking. According to Ishibashi
(1988), a reduction of 30 to 50% for clayey soils and about 20% for sandy soils can be
achieved. The lubricant can only work effectively if it is maintained in between the
sliding surfaces of the pipe and the excavated soil surface. The injection of the lubricant
with high pressure may force it to flow only into certain areas, whereas the lubricant
viscosity may contribute a lot to the stability of the tunnel bore (Rahjoo et al. 2012;
Stein et al. 1989). The angular deviation between the central axes of successive pipes
can result in misalignment and a severe misalignment can further cause significant
additional interface friction between pipe and surrounding soil, resulting in an
increment of the total jacking load (Norris 1992; Ni and Cheng 2012). In addition, the
total jacking load may increase after stoppages that stand for the interruption of pipe
jacking during the night or weekend and this increase corresponds to the friction
resistance increase with the face resistance staying the same.
This study presents a case history where one sewer line of the Shulin distinct sewer
network was carried out using a slurry shield, and the drive data obtained were
investigated. The objectives of this study are: (i) to determine the significant factors that
lead to the variation of the total jacking load with the unified soil classification system
and (ii) to verify the critical influence of soil type and lubricant onto total jacking load.
SITE DESCRIPTIONS
Geological Conditions
During ground investigation phase, 7 15-m deep boreholes were constructed for better
understanding the geological conditions along the drive as shown in Fig. 1. The soils
mainly consist of three broad categories of soils: (1) silty sand occasionally mixed with
fill debris and, (2) dense to very dense, poorly graded to well graded sand and gravel
with cobbles or with zones of silt and clay, and (3) silty sand. Groundwater level is at
4.8m below the ground surface.
ASCE
183
184
BH-5
BH-6
BH-4
Fill debris
silty sand
9.64 m
11.14 m
LEGEND
Sewer line (D=1.5 m)
Fill
Gravel and sand
Gravel and sand with cobbles

or with zones of silt and clay
Silty sand
FIG. 1. Ground profile along the drive.

Design and Construction
A full face slurry shield has been utilized in this project to conduct the excavation.
This shield was chosen because of its proven track record of successfully excavating
through wood, sand, and gravel. The shield itself is in two sections (i.e., shield head and
trailing can) to allow for launch and recovery from minimal sized steel shafts with an
internal diameter of 3590 mm and a thickness of 20 mm.
The 2.65-m long shield head, at 9.7 tons, has a 1.455 m-long articulated steering
section, which is floating on a 1.02 m long can, up to 2.15-degree articulation in the
steering is controlled by 442 ton steering cylinders. The trailing can, at 0.92 m long
and some 2 tons, is bolted on to the shield head. Inside the trailing can is the power pack
for the steering cylinders and all other shield functions. The jacking system consisted of
2180 ton hydraulic cylinders. A laser beam apparatus targeted at a moving sensitive
plate attached to the back of the shield head, dispatches the alignment signals to the
control room on the ground surface. The collected data provides information to the
operator in order to maneuver the steering cylinder.
The pipe jacking technique utilizes bentonite drilling mud to maintain the equilibrium
pressure at the excavation face in addition to facilitating the collection of the excavated
material. The ground soil to be excavated is gathered in the crusher cone section and
spoil is continuously pumped to the separation facility (shaker screen and sedimentation
tank) at the ground surface as shown in Fig. 2. The openings of the coarse and fine
shaker screens of the separation plant are 2 mm and 0.3 mm, respectively.
Reinforced concrete pipe, with an internal diameter of 1200 mm and an outer
diameter of 1440 mm, in 1 m length is utilized. Each pipe is fed successively into the
hydraulic jack carrier and is pushed into the ground. An overcut annular of 15 mm on
radius is created by using slightly larger cutting wheel of 1500 mm in diameter than the
shield head of 1470 mm in diameter. Another overcut annular of 30 mm on radius is
established by smaller outside concrete pipe of 1440 mm in diameter. A special
lubricant of bentonite slurry with polymer (2%) particularly developed for the gravel
formation is utilized to fill in annular space. This special lubricant was injected in a
controlled sequence under appropriate pressure through a narrow pattern of outlets in
the string to ensure a uniform distribution of lubricant. The flow cone testing result is
more than 38 minutes indicating that the viscosity of this lubricant is high enough to
prevent it from seeping into permeable sand and gravel formation. This will ensure the
function of the lubricant to be maintained throughout driving length.
ASCE
185
Cutter heads are fitted with a combination of roller discs and scraper bits. The roller
discs are 30 mm ahead of a series of scraper bits as shown in Fig. 3. The roller discs
should grind down the hard stones and cobbles, whereas the scraper bits should
shave-off the weak materials such as sand/gravel, silt or clay.
Shaker
Di = 3.59 m
Sedimentation
tank
Di = 3.59 m
h=2m
Bentonite
mixing
P1
(l 1=l 2):15.26m
4m
1500 mm
10.36 m
Arrival
shaft
H = 11.14 m
Launching
shaft
P2
L = 126 m
FIG. 2. Description of shafts and shield dimension as well as drive length.

Monitoring of Jacking Parameters
In order to better understand the variations of penetration resistance and friction
resistance, a continuous monitoring of torque of cutting wheel, penetration speed,
lubricant distribution, stoppage, slurry pressure, pipe deviation and corrections was
conducted during pipe jacking. Samples of cuttings on shaker screen and bentonite
slurry in sedimentation tank were both collected for following laboratory tests to
classify the ground soil along the drive. This additional data helps interpreting the local
variation of jacking force. This paper presents one of many drives with drive lengths
varying from 73 to 125 m.
PIPE JACKING RESULTS AND DISCUSSION
Access of entering the control room was given for taking the spot monitoring readings
every 20 cm distance drive in addition to the above-mentioned continuous monitoring.
The soil samples obtained were used to track parameters such as jacking force, torque,
and penetration rate during the pipe jacking. Shield performance in different ground
conditions encountered along the drive was then evaluated.
Fig. 4 shows a 127-m long pipe jacking result together with the soil log. The fine soil
controlled ground is marked by four yellow shaded areas (i.e., between 21 and 26m,
between 45 and 65m, between 89 and 111m, and between 123 and 127m), otherwise the
ground will be sand and gravel with cobbles. The torque of cutting wheel and shield
deviation are plotted against vertical axis at left, and the jacking force and driving speed
are displayed against vertical axis at right.
ASCE
186
Shield head
Trailing can
Steering jack
Motor
=0
Cutting
head
Crusher
cone
chamber
Tilt
meter
L = 0
Centerline
Laser beam
Laser
beam
Target
plate
Motor
Steering jack
15 mm
Slurry tube
l=1455 mm
Steering gap
Laser beam
Laser beam
=0
Roller
disc
Strawberry
roller discs
FIG. 3. Cutting wheel and steering facility of slurry shield.

The jacking force maintains at 105 tons along the initial 20 m distance drive and
between 26 and 45 m drive. This force is caused by the face resistance and the friction
between heavier shield and soil. The smallest jacking force after 66 m is 120 tons. The
largest jacking force 150 tons occurs at the half way of drive and the end of drive, and
this happens to be in the fine soil controlled region. It appears that the jacking force does
not increase along the driving length, implying the bored hole may be stable enough to
keep the steering gap open due to the annulus area being full with viscous lubricant. The
highly viscous lubricant prevents it from seeping into coarse soil, i.e., sand and gravel.
Pellet and Kastner (2002) observed the importance of overcutting which, through
reducing the radial compression in soil, leads to a reduction in the normal stress acting
upon the pipe.
Standard roller discs were located within the central portion of the cutter head to spall
encountered cobbles. Scraper bits were included around the periphery of openings in the
cutter head to excavate finer grained silts and clays. The inset was used to protect these
teeth during pipe jacking in the cobbles.
Spikes in torque suggest that the sand and gravel formation with occasional cobbles is
encountered by cutting wheel in this alluvial deposit. The roller discs are used to break
larger particles and they are running 30 mm ahead of a series of scraper bits. When sand
and gravel formation is encountered, the only contact points are between roller discs and
excavation face, and the scraper bits do not contact with excavation face. If the bored is
stable at this moment, the face resistance will maintain steady, so does the jacking force.
But when cutting wheel runs into ground controlled by fine soil, the roller discs will
sink into excavation face and the entire cutting wheel will contact the excavation face.
This results in the increase of jacking force. This can be observed in those four yellow
ASCE
shaded areas shown in Fig. 4. The increase of face resistance due to fine soil will
disappear when shield runs into sand and gravel formation again. This fact is confirmed
by Pellet and Kastner (2002). They used a jack load cell attached to the head of shield
machine to measure face pressure. They concluded that the local variations of total
jacking load are mostly linked to the face resistance and the minima of total load
correspond to very low face loads.
The largest increase of face resistance is about 50 tons which is slightly smaller than
the passive soil resistance at the excavation face. Face pressure should be greater than
the active earth pressure to prevent ground surface settlement and less than the passive
earth pressure to exclude surface heaving. Therefore, slower driving speed not only
avoids the excessive face resistance, but also provides more time to slurry system to
separate fine soil from the bentonite slurry. The excessive face resistance may cause the
cutting wheel stalling.
The largest deviation is 30 mm which is smaller than the thresh hold value of 60 mm.
Therefore, the deviation of this drive will not likely cause the friction between pipe and
ground soil. The flow cone testing result is 38 minutes, indicating that the viscosity of
this lubricant is high enough to prevent it from seeping into permeable sand and gravel
formation. This ensures the function of lubricant to be maintained throughout driving
length. Lastly, the stoppage effect was not significant to be observed in this study
because the encountered ground is mostly non-cohesive soils.
FIG. 4. Jacking force versus soil type, torque, deviation, and driving speed.
SUMMARY AND CONCLUSIONS
From the pipe jacking result discussed in this paper, the following conclusions can be
drawn: (1) Bentonite-polymer slurry injected in the annular space at sufficient amounts
to compensate the losses into permeable sand and gravel leads to the stabilization of the
bore, which reduces the friction force around the pipes. (2) The jacking forces in the
first 20 m remain constant at 105 tons, and is caused by the face resistance and the
friction force between heavier shield and soil. (3) The local variations (increasing and
ASCE
187
decreasing) of jacking force are linked to the increasing face pressure due to driving
from coarse soil to fine soil. (4) The pipe deviation is smaller than a threshold value of
60 mm, and this will not generate any significant friction force between pipe and soil. (5)
The highly viscous lubricant distributed uniformly along the pipeline ensures a more
effective reduction in friction resistance during driving in sand and gravel formation. (6)
The classification of encountered soils developed in a separate study helps interpreting
the local variation of jacking force caused by increasing or decreasing of face pressure.
(7) For large diameter pipe jacking with stabilized overcut and sufficient amounts of
lubricant, the face pressure can be significant in comparison with the friction resistance.
REFERENCES
Auld, F.A. (1982). Determination of pipe jacking loads. Proceedings of the Pipe
Jacking Association, Pipe Jacking Association, Manchester, UK.
Ishibashi, N. (1998). Japans recent small diameter pipe jacking construction
methods. Proceedings of International NO-DIG'88, Washington DC.
Kastner, R., Pellet, A.L., Ouvry, J.F., and Guilloux, A. (1996). In-situ monitoring of
microtunneling projects. Proceedings of International NO-DIG'96, New Orleans:
171-182.
Khazaei, S., Shimada, H., and Mastsu, K. (2004). Analysis and prediction of thrust in
using slurry pipe jacking method. Proceedings of the 30th ITA-AITES World Tunnel
Congress, Singapore: 22-27.
Norris, P. (1992). The behavior of jacked concrete pipes during site installation. PhD
Thesis, Department of Engineering Science, University of Oxford, UK.
Ni, J.C. and Cheng, W.C. (2012). "Steering characteristics of microtunnelling in various
deposits." Tunnelling and Underground Space Technology, Vol. 28: 321-330.
Pellet-Beacour, A.L. and Kastner, R. (2002). "Experimental and analytical study of
friction forces during microtunneling operations." Tunnelling and Underground
Space Technology, Vol. 17 (1): 83-97.
Rogers, C.D.F. and Yonan, S.J. (1992). "Experimental study of a jacked pipeline in
sand." Tunnels and Tunnelling Int., Vol. 24 (6): 35-38.
Rahjoo, S., Najaf, M., Williammee, R., and Khankarli, G. (2012). Comparison of
jacking load models for trenchless pipe jacking. Pipelines 2012: Innovations in
Design, Construction, Operations, and Maintenance, Florida: 1507-1520.
Stein, D., Mllers, K., and Bielecki, R. (1989). Microtunneling: installation and
renewal of non-man size supply and sewage lines by the trenchless construction
methods. Ernst and Sohn, Berlin, Germany.
Stein, D. (2003). Practical guideline for the application of micro-tunneling methods.
Rademann, Ludinghausen, Germany.
ASCE
188
Research on the Comprehensive Geological Prediction Method on a Metro Tunnel Karst

Disaster and the Numerical Simulation of Corresponding Treatment Measures
Shucai Li1; Haotian Wu1; Ke Wu1*; Wen Zhang1; Maoxin Su1; and Yajun Wang1
1
Research Center of Geotechnical and Structural Engineering, Shandong Univ., Jinan, Shandong
250061, P. R. China.
*Shandong Univ., Jinan, Shandong 250100, P.R.China (corresponding author). E-mail: wk4223@163.com
Abstract: The construction of urban metro in high-risk karst areas is different from that of highway or
railway tunnel in many ways, such as having a great threat to the project stability, prone to induce
secondary geological disaster, damaging the surrounding existing structures and being vulnerable.
Based on the analysis of geological water level of internal and external tunnel project and various
geophysical prospecting, it puts forward a method to explore the geological disaster seepage channel
distribution regularity about urban metro in karst areas in an effective and accurate way. And the
implementation and comprehensive analytic process concerning this method are covered in detail,
which is exemplified by the geological disaster predication and seepage channel exploration of Yaman
interval in urban Rail Line 1, Guiyang city. Then the key technology for comprehensive treatment
towards seepage channel of karst tunnel is proposed and further verifying its effectiveness by utilizing
numerical simulation software.
Keywords: Urban metro; Karst tunnel; Engineering geology; Geophysical prospecting; Advanced
geological predication; Numerical simulation.
INTRODUCTION
With the increase of urban population density, urban underground space is developed on an
unprecedented scale [1]. For cities in karst areas, building underground projects will inevitably result in
sorts of karst geological disasters which undermine the construction and safety of builders [2, 3]. In this
condition, considerable achievements in researching how to forecast tunnel geological disasters in
karst areas has been made[4, 5]. Jetschny, et al[6] employed wave analysis on urban metro tunnel face to
predict the occurrence of geological disasters under metro construction. The distribution regularity of
tunnel geological disasters was revealed by Jang Hyongdoo, et.al[7] through analyzing the regional
geological parameters. Li Shucai, et al.[2, 4, 5, 8] put forward the four-color early-warning mechanism of
tunnel geological hazard by optimizing the comprehensive advanced geological prediction process.
Basing on the vast documents and engineering practice, Ye Ying[9] conducted a systematic study of
forecast methods by analyzing karst formation and distribution. He Chuan[10], through comparing
conventional methods and non-destructive testing methods, studied on applicability to detect water
seepage of highway tunnel by examining the situation of tunnel. However, research on technology of
geological disaster forecast and seepage channel exploration, which involved in urban subway in karst
areas, doesnt develop fully.
In view of the geological disaster predication and seepage channel distribution regularity of Yaman
interval in Guiyang urban Rail Line 1, a high-risk karst area, it brings forward the seepage-distribution
exploration method based on the geological water level analysis of internal and external tunnel project
as well as various geophysical prospecting, and then elaborates its implementation and comprehensive
analytic process.
ASCE
189
190
COMPREHENSIVE METHODS FOR PROSPECTING KARST SEEPAGE

CHANNEL
The main methods for advanced geological disaster predication and prospecting of underground
project are engineering geological method, geophysical prospecting method and empirical method, etc.
Every method has its advantages and disadvantages. Single method will lead to multiple results. Thus
the exploration precision and accuracy need utilize comprehensive methods[1, 11].
Application of Methods for Prospecting Geological Disasters of Urban Metro Tunnel
The implement of geological disasters forecasting mainly consists of field prospecting and data
explanation. For field prospecting, most forecast methods need adjust instrument parameters and
selection. For inner tunnel prospecting, the survey line should be arranged according to lithology and
structure of surrounding rock. For external tunnel prospecting, the line should be arranged on the top
of tunnel axis when transient electromagnetic method is chosen. For high-density resistivity method,
line is advised to be in each side of road instead of in the middle. Prospecting data explanation should
be carried out after analyzing the corresponding geological data and integrating results from inner and
outer tunnel prospecting. The process is shown in Figure 1.
Fig.1 Flow chart of exploration about karst geological disaster
CASE ANALYSIS
Adopting the comprehensive prospecting methods for outer tunnel (surface hydrogeology survey,
karst cave measuring, water-pumping observation, etc) and inner tunnel (TSP, geological radar and
transient electromagnetic method) ,the author, based on water inflow disaster in tunnel face of
ZDK19+200 at Yaman interval in Guiyang urban rail Line 1, explored the karst geological disaster and
its exact position , scale and damage degree are made clear.
Karst Water Inflow
After the water inflow occurred on August 18, four working faces of shaft at Manpo were filled
with water in less than twenty minutes. The peak of water inflow at tunnel face hit 700m3/min. It was
in only sixteen hours that the shaft was filled up. The permeable point is at right side of vault and
haunch in left tunnel. Furthermore, the frequent heavy rain in Guiyang, thick bottom silt and the depth
of shaft (36 meters) made it difficult to drain by pumps.
ASCE
191
Engineering Geology Analysis

The tunnel of Yaman interval is excavated by mining method. Its formation lies in mountain peak
valley and corrosion area of Guiyang basin, which falls into the dissolution slope and depression
landform. According to the geological data, there are three faults. F1 in the attitude of
150-16070-80 is the front thrust fault of thrust nappe. The attitudes of F9 and F3 are
24285and14082. Both of them are normal faults. Influenced by them, rock fracture in field is
from less developed to developed and the formation attitude is diverse.
Water Level Survey
In order to detect the distribution scale of karst seepage channel, first of all, confirming the water
diversion in field area according to geological map and field investigation. Its coverage ranges from
the Station Manpo to Tianyu City forest park.
Secondly, survey shows that there are three large surface water supply stations. (1) Machong ditch,
flowing from north to east, merges with rail line 1. The drain is 2.10~2.60m in width and 1.50~2.20m
in depth. Its water is 1143m in elevation.(2) The pond in the Guizhou University, about 120m2, with
water depth of 0.5~1.5m.(3) Xiaoguan Reservoir, flowing across field area. Interval between Yaguan
and Manpo Station is about 800m apart from the reservoir. The normal storage level is 1,121m, dead
water level is 1112m. Water source is mainly meteoric water.On the other hand, five underground
springs are found out after water-level geological investigation.
Finally, the statistical results of rainfall are shown in Table 1.
Table 1 Statistical table about water pumping record and rainfall
Item
Date
Rate of water
Trans-ver
Pumping
decreasing in
se area of
rate of
shaft
shaft
equipment
(m )
Rainfallmm
Guiyang
Begin
End
(m/h)
8.17
8.18
8.18
8.22
1.2
500
8.22
8.23
1.1
1200
8.23
8.31
1.3
9.1
9.4
9.4
9.5
(m /h)
107.3
Xiaog
uan
41.5
Forecast of
Yaguan
84.3
water inflow
m3/h
30000
400
30.2
16.6
64.1
1000
500
400
200
200
9.5
-0.6
200
9.8
0.5
200
87
21.2
21.2
8.4
300
150
Karst Cave Geology Survey

The exposed karst cave at ZDK19+200 tunnel face, measured by geological compass and infrared
ambulator, is about 12.5m in height, 16m in length and about 4m at its widest point. The watered point
inside tunnel is about 1117m in elevation. The cave extends from east to the right line of tunnel and
connects with spring. The length of visible part is about 10.4m. Its upper part is wide and ellipse in
shape while the lower part is V-shaped.
ASCE
192
TSP Detection
Using TSP203plus to detect the surroundings near the karst disaster areas in order to prove up
distribution characteristics of seepage channel. The result is in Figure 2:(1) surrounding rocks are
relatively broken from the section ZDK19+200.8 to ZDK19+190. The attitude of stratum changes
considerably. There is a 75% chance that dissolution fissure water or water-conductive muddy
intercalation exits. (2)The surrounding rocks are broken from section ZDK19+190 to ZDK19+160,
joint fissure develops and fillers are abundant. The surrounding rocks are not stable and corrosion
fissure develops.
Transient Electromagnetic Exploration Analysis
Considering geological and electromagnetic conditions, the common loop point array is given
preference for transient electromagnetic exploration. Edge of transmitting coil is 3m long, amounting
to 8 turns. The receiving device is probe, with the resonant frequency of 250K15%, effective area of
212m2. Its sensitivity is more than 0.9v/nT.Hz. The dot pitch of surveying line is 0.15m and the
number of measuring point is 20. The analysis result produced by IGGETEM-30A, which is invented
by Institute of Geophysical and Geochemical Exploration, Ministry of Land and Resource and
Shandong University, is shown in Figure 3. It can seen that there is an extensive karst geological
hazard body ahead of tunnel face with the distance of 30~45m.
Fig.2 TSP detectionFig.
Fig. 3 Contour map of apparent resistivity
Geological Radar Detection Analysis

The detection of tunnel vault by geological radar is to confirm the distribution and scale of karst
area. Figure 4 shows the detection result. With reference to it, there is a strong reflection of radar wave
located at 7m and 15m behind the vault, and probably existing karst anomaly. The event of 7m and
15m in figure is discontinuous which means the presents of fault fracture zone, and because of the
strong reflection of wave here, the fissure is probably filled by water.
DISTRIBUTION REGULARITY OF KARST SEEPAGE CHANNEL

The comprehensive prospecting methods for outer tunnel and inner tunnel are adopted to survey the
karst seepage channel in metro tunnel. The result indicates as below.
(1) The volume of karst cave is 600m3 and about 400m3 excluding the non-filling part. About 200
m3 of exposed karst cave was submerged by water. And there are at least 15000m3 karst area which
was estimated by the volume of pumping water.
(2) The karst cave at tunnel face of ZDK19+200 extends long and connects with No.1 Spring in
field area, deducing that the karst channel crosses the right tunnel. The channel would be exposed if
ASCE
193
right line excavating. The rocks around karst cave, affected by distinct structure, are broken. Because
strata is steep, rocks in vault are prone to fall down. The elevation of water in right karst cave is higher
than that of left water exit point, and water will seeps with right tunnel excavating.
(3) The tunnel karst seepage water is primarily supplied by surface sewage infiltration and
underground water. The water inflow in tunnel is severely affected by atmospheric rainfall.
Fig.4
Geological radar
Fig.5 Position of karst seepage channel
MEASURES FOR TREATING KARST SEEPAGE CHANNEL

Based on the similar engineering experience and waterproof conception in subway design,
comprehensive plugging measures, including the surface and underground ones, together with the
joint plugging measures are taken to seal underground water from the tunnel excavation scope and
meanwhile ensure the creation of permanent aquifuge for successful excavation of tunnels.
Firstly, the surface deep grouting measures are carrried out outside cave. In accordance with the
trend of tunnel and the relative plane relation of two lines, make field lofting to find external edge of
two lines. Five rows of deep holes are dug along the lateral excavation contour line. And the first row
is 1m away from external edge of right cave. The distance from the second, third, forth and fifth to the
first row is 2, 4, 6, 8 m. Deep grouting holes are made 9m away from the scope of excavation contour
line. Deep grouting hole are also made inside the excavation contour line, with the 2m spacing and 2m
row spacing. And all above is quincuncial shape. The sequence of grouting, as in Figure 6, is followed
from right to left in L shape. Deep-hole split grouting are adoptted.
Secondly, the mud clean-up measures inside cave. The silt becomes thin after being flushed. So, it is
difficult to dredge the thin silt and carry it out of tunnel. In such condition, mixing silt with 6-8%
cement or lime, it can be cleaning up after being consolidated. Water should be drained so as not to
flow over and reduce the effect of consolidation. As shown in the Figure 7.
Fig.6 Grouting reinforcement out of tunnel
Fig.7 Desilting inside Cave
Thirdly, reinforce the initial supports of excavated tunnel. After desilting, monitoring for section
convergence and vault subsidence are done for the immersed section. Whether to reinforce exist lining
and temporary reinforcement are decided by monitoring data. Geological worker should analyze the
property, trend, surrounding rock and influence scope concerning cave in tunnel face.
ASCE
Fourthly, treatment measures of plugging and reinforcement for left-line. Concrete 25 is used to seal
the cave and the filler is as thick as the distance from the entrance to 5 meters away from the bottom.
A 200mm diameter and 8mm thickness steel tube is pre-embedded to the bottom inside cave. Pressure
valve is set outside and the tube is connected to the sump pit of shaft. 10cm shotcrete is done to plug
the tunnel face and then reinforced concrete wall is built in the tunnel face and right karst cave. It is
advised that the wall is at least 2m in thickness and shaped in L, whose bottom and sides are
anchored with surrounding rock by 4m25 thread steel, with the spacing of 5050cm. After
construction of preventing-grout wall, karst cave plugging and advanced curtain grouting in left line
begin. For cave plugging, 6m50 steel flower tube radiates in sector and the mouth spacing is
0.5m0.5m, in which cement and water-glass grout is used. Advanced curtain grouting is done to
ensure the safe construction that behind tunnel face of left line. Reinforcing the excavation face and
rocks that 5~8 meters out of the contour line. At begining, a 2-3m thick grout-preventing wall is
erected. The technology for grouting is advancing phased grouting and cyclic grouting. On the other
hand, take advanced big pipe shed and small pipe grouting supplement strengthening measures for left
line. Advanced big pipe shed is taken on arch for tunnel face of left line. Seamless steel tube with
108mm outer diameter and 8mm thickness is used for big pipe shed. Each section is 2-3m in length.
Then, 42mm advanced grouting little pipe set around working face, which is 3.5m long and 20cm
in ring spacing with setting angle 15. As making consolidation grouting at tunnel bottom, the
geological condition of inverted arch should be scanned in time and backfill treatment need be
provided if cavern exists.
Finally, treatment measures for right tunnel are applied. For grout-preventing wall, its construct
scheme is similar to left line. Then, applying advanced curtain grouting in right line. Suggestting
strengthen primary supporting structure when right line is excavated. Arch centering, made of I22
steel, is modified to 0.5cm in spacing and the concrete is 30cm in thickness. The arrangement of
consolidation grouting for surrounding rocks is similar to left line. The branch tunnel is advised to be
built from the left to the right line when the construction of left line will not be affected.
NUMERICAL METHOD TO RESEARCH ON TREATMENT MEASURES OF

KARST TUNNEL
The treatment measures could be classified into two groups: treatments to supporting structure and
treatments to karst cave. This chapter would use Flac3D to analyze effect of above treatments.
Model Foundation
Based on the project, simplifying layer, tunnel and karst cave, establish a 2-D model. In order to
eliminate boundary effect, length of model is 60m, and the height of model is 80m. Divides model into
3 layer, and from upper to low: overlaying soil 10m; moderately weathered limestone 30m; fresh
limestone 40m. The tunnel shape is horseshoe with length 11m and height 8m. Its depth is 50m. On
the other hand, the karst cave is 8m above the tunnel and similar to elliptical shape. For purpose to
simplify calculation, establishing the cave with circle shape and located at weathered limestone layer
where 5m above tunnel vault. There are 2914 elements and 6070 nodes.
ASCE
194
195
calculation method and parameter selection

Essay ignores the influence of seepage field. Simply classify the treatments into two groups: cave
treatment and grouting reinforcement. Considering real effect, surface grouting, full-section grouting,
and curtain grouting all improve mechanical property of surrounding rock. So the essay adopt
full-section grouting method to simulate effect of these measures. On the other hand, the cave
treatment measures, as shown in above: construct 4m HRB40028 rebar as anchor, meanwhile
constructing 30mm thick steel plate as internal lining of cave, as shown in Fig 8.
The calculating method as below: (1) set top boundary of model as free surface; (2) the construction
of tunnel utilizes upper-lower bench method; (3) simulate tunnel and layer with mhor-columb model,
while simulate lining and supporting structure with elastic model. Anchor was simulated by cable
element. The mechanical property of materials are shown as table 2.
Fig 8 reinforcement system of tunnel and carve

Table 2 Mechanical Parameters of Model
Poisson
elasticity
ratio
modulusE
2.95
0.30
9.02
6.53
0.23
Material list
K/MPa
G/MPa
red clay
7.52
moderately weathered limestone
8.61
fresh limestone
13.05
tunnel lining
1.9110
grouting reinforcement area
25.78
supporting structure of cave
9.79
4
1.4410
19.5
5
1.3710
8.2410
Conhensi
Friction
on C/kPa
angle /
17.2
16
7.5
15.58
26.0
150
15
0.20
23.62
27.0
200
25
0.19
46.72
400
25
210
11
KN/m3
Analysis of Calculation Result

Set group without reinforcement measures as the control group to contrast with the full-section
grouting group and cave supporting group, and further analyze the deformation and stress.
(1) The layer settlement nephogram of three groups are shown in Fig 9. From the nephogram,
settlement mainly concentrate on the area between cave and tunnel vault. And both two methods could
effectively reduce eformation of tunnel and layer, which prove the effectiveness of measures
From the nephogram, the deformation trend of model after full-section grouting is simmilar to that
of normal condition, while it is different from the cave-reinforcement group: after full-section grouting,
settlement of tunnel and its surroundings obviously decrease. After cave reinforcement, the settlement
of tunnel decreases, but the settlement of surrounding zones doesnt clearly reduce.
ASCE
196
a. Nephogram of normal condition
b. Nephogram of full section grouting c. carve reinforcement using anchor
Fig 9 settlement nephogram of different condition
(2)Fig 10 has shown the plastic zones distribution of model under three reinforcement methods. The
distribute law of common condition and full-section grouting group are similar: V shape area above
the cave intend to failure. And in other area around cave, the element was both tensile failure and
shear failure; the plastic zones around tunnel mainly distribute on side wall, and tensile failure mainly
occurs on the haunch, while shear failure occurs on parts of arch foot. Furthermore, plastic zone of
reinforced model is less then un-reinforced one, especially around tunnel. After full-section grouting,
the load applied on supporting structure and surrounding rock decrease.
a. Normal condition
b. Full-section grouting
c. carve reinforcement using anchor
Fig.10 plastic zone distribution influenced by three groups of reinforcement measure
The plastic zone distribution after cave reinforcement is quite different from others. There are large
amount elements around the cave intend to broke and the failure zone around tunnel does not decrease
after reinforcement measures. That means adopting anchor to reinforce cave could reduce deformation,
but it would lead to increment of load on surrounding rock and supporting structures.
From above, both reinforcement measures could effectively control deformation. Meanwhile, using
anchor to reinforce the cave would lead to large burden on the supporting structure. Hence it should
cooperate with other treatment measures to ensure construction safety.
CONCLUSION
Taking example of geological disaster predication and seepage channel distribution regularity of
Yaman interval in Guiyang urban Rail Line 1, one new seepage exploration method is presented and
elaborates its implementation and comprehensive analytic process. The key technology for
comprehensive treatment of seepage channel is developed. Finially, get following conclusion: (1) the
new method to research into seepage channel of urban metro based on combining inside and outside
exploration technologies is effective. (2) According to the local project for researching into seepage
channel, some effective treatment measures are supported for similar projects.
ASCE
ACKNOWLEDGEMENT
This work was financially supported by National Natural Science Fund(51308323) and Open Fund of
Research Center of Dam safety and Disease Control in Ministry of Water Resources(2014008).
REFERENCE
[1] SU Maoxin, QIAN Qihu, LI Shucai, et al. An method of metra advanced geological prediction under karst geology
condition [J]. Chinese Journal of Rock Mechanics and Engineering, 2011, 30(7): 1428 - 1434.
[2] LI Shucai, XUE Xiangguo, ZHANG Qingcai, et al. Key technology study on comprehensive predication
earlywarning of geological hazards during tunnel construction in high-risk karst areas [J]. Chinese Journal of
Rock Mechanics and Engineering, 2008, 27(7): 1297-1307.
[3] WANG Meishu. Hydrogeological and geological forecast of tunnel construction in the karst district[J]. Railroad
Survey, 2004, (1): 7-10.
[4] LI Shucai, LI Shuchen, Zhang Qingsong, et al. Forecast of Karst-fractured groundwater and defective geological
conditions [J]. Chinese Journal of Rock Mechanics and Engineering, 2007, 26(2); 217 - 225.5
[5] ZHANG Wenjun. Study on cross-hole resistivity computed tomography method in survey of karst area[D].
Shandong University Thesis for Master Degree, 2014.
[6] Jetschny Stefan, Bohlen Thomas, Kurzmann Andr. Seismic prediction of geological structures ahead of the tunnel
using tunnel surface waves [J]. Geophysical Prospecting, 2011, 59(5):934-946.
[7] Jang Hyongdoo, Topal Erkan. Optimizing overbreak prediction based on geological parameters comparing multiple
regression analysis and artificial neural network[J].Tunnelling and Underground Space Technology, 2013,
38:161-169.
[8]
SHI Shaoshuai. Study on seepage failure mechanism and risk control of water inrush induced by filled disaster
structure in deep-long tunnel and engineering applications [D]. Shandong University Doctoral Dissertation, 2014.
[9] YE Ying. Tunnel construction ahead geological forecasting method research [D]. Beijing Jiaotong University
Doctoral Dissertation, 2006..
[10] LI Xing. Research on seepage diseases detection and evaluation system of highway tunnel [D]. Southwest Jiaotong
University Master Degree Thesis, 2014.
[11] GUO Ming. Study on concealed Karst caves influence on karst tunnel stability and treatment technology on
tunnels of E-xi mountainous[D]. Shandong University Thesis for Master Degree, 2014.
ASCE
197
Risk Assessment of a Tunnel Collapse in a Mountain Tunnel Based on the Attribute

Synthetic Evaluation System
Benshuo Wang; Shucai Li; Qianqing Zhang; Liping Li; Qian Zhang; and Fei Xu
Master, Geotechnical and Structural Engineering Research Center, Shandong Univ., Jinan, Shandong
250061, China. E-mail: wbs199203@163.com
Abstract: Collapse is one of the common disaster in tunnel construction. In order to

ensure the construction safety of mountain tunnels, the attribute recognition model of
risk assessment for tunnel collapse is established based on attribute mathematics
theory. By the comprehensive analysis for influencing factors of the tunnel collapse,
the surrounding rock grade, tunnel depth, angle of unsymmetrical pressure, rock mass
integrity, earthquake, and the excavation span are selected as the indexes of the risk
assessment attribute measurement functions were rigorously constructed to compute
attribute measurement of single index and synthetic attribute measurement. The
identification and classification of risk assessment of mountain tunnel samples were
recognized by the confidence criterion. An exemplification study of a tunnel showed
that the assessment results, obtained through attribute measurement analysis and the
date of geological forecast agreed well, with the tunnel construction. This risk
assessment methodology provides a powerful tool for systematically assessing the
risk of tunnel collapse.
INTRODUCTION
Tunnel collapse means that the rock lost control due to the presentation not in time
when it occurs the change of tunnel rock system like fault, crushed zone during
tunneling. Tunnel collapse can greatly endanger the safety of tunnel construction,
leading to huge casualties and economic loss. Therefore, prediction of the occurrence
probability of tunnel collapse is one of the most interesting areas of research in tunnel
construction, and there is a need to clarity the possibility of tunnel collapse during the
construction of tunnel. The risk assessment of tunnel collapse is attached more
attention.
Risk analysis is of signality for tunnel safety, various methods proposed originally
or derived from other fields have been used to evaluate the risk in tunnel.
Probabilistic models for tunnel construction risk assessment were established to
estimate tunnel construction time and guide the construction performance with data
from past tunnel projects. Bayesian probability networks have also been applied to
assess the risk in road tunnel, which is embodied in two aspects. Chen and Huang[1]
ASCE
198
199
presented a calculation method to access the probability and losses of the accidents
occurred during the shield construction in soft soil area.
As to the risk management and assessment on engineering, the International
Tunneling Association (TA) proposed the Guidelines for Tunnel Risk Management.
Based on the fuzzy analytical method and extension assessment method, many studies
were made to analyze the risk assessment of seismic damage at tunnel and tunnel
collapse. However, because of the random determination of membership function, the
fuzzy analytical method can not be used to obtain clear classification and reasonable
result.
Risk assessment can be used to forecast disaster happened in tunnel. Based on the
results of risk assessment, the possibility of the tunnel collapse can be judged, and
corresponding treatment measures also can be proposed. By using the attribute set,
attribute measure space and ordered partition class, Cheng [2]opposed a attribute
recognition model to analyze the comprehensive assessment in the 1990s. As
subsequently, the attribute mathematics theory was applied successfully in the rock
engineering. However, few studies are concerned on the attribute mathematics theory
of tunnel collapse. Therefore, this is a need to establish an attribute recognition model
by using the attribute mathematics theory to assess the possibility of tunnel collapse.
There are numerous influencing factors about tunnel collapse, such as the support
used and low resistance geological features. And in the present paper, the influencing
factors of the tunnel collapse, i.e., the surrounding rock grade, tunnel depth, angle of
unsymmetrical pressure, rock mass integrity, earthquake and the excavation span,
were used to assess the possibility of tunnel collapse, and to obtain the risk grade of
tunnel collapse. A case study of a mountain tunnel of Cheng-Lan railway was used to
verify the reliability of the present attribute recognition.
ATTRIBUTE MATHEMATICAL THEORY
An attribute synthetic assessment system consists of three components: single
index attribute measure analysis, multiple indices synthetic attribute measure analysis
and attribute recognition analysis.
Single index attribute measure analysis
Single index attribute measure function is established to compute the single index
attribute measure xjk = xij Ck
1kK. The data format of evaluation index is
presented in Table 1, where ajk (j =1, 2, , m) should satisfy aj0 < aj1 < < ajK or
aj0 > aj1 > > ajK.
Two parameters bjk and djk are used to establish the attribute measure functions.
The parameters bjk and djk can be expressed in the following forms:
b jk =
a jk 1 + a jk
d jk = min b jk a jk , b jk +1 a jk
ASCE
(1)
(2)
200
Table 1 The grade subdivision of single index

Evaluation index
Risk assessment grade

C1
C2
CK
a1K-1a1K
OR
><a1K
a20a21 a21a22
a2K-1a2K
OR
><a2K
amK-1amK
OR
><amK
Im
a10a11 a11a12
I2
I1
am0am1 am1am2
When aj0 > aj1 > > ajK, the function of xjk can be determined as:
t a j1 + d j 1
xj1 (t ) =
2 d j1
t < a j1 d j1
a j1 d j1 t a j1 + d j1
(3)
t > a j1 + d j1
0,
t < a jk d jk
t
a
d
+
jk
jk
, a jk d jk t a jk + d jk
2d jk
1,
xjk (t ) =
a jk + d jk < t < a jk 1 d jk 1
a +d
t
jk 1
jk 1
, a jk 1 d jk 1 t a jk 1 + d jk 1
2d jk 1
0,
t > a jk 1 + d jk 1
(4)
1,
t < a j1 d j 1
a j1 + d j1 t
, a j1 d j1 t a j1 + d j1
xj1 (t ) =
2 d j1
0,
t > a j1 + d j1
(5)
When aj0 < aj1 < < ajK, the function of xjk can be determined as:
ASCE
1,
t < a jK 1 d jK 1
a jK 1 + d jK 1 t
, a jK 1 d jK 1 t a jK 1 + d jK 1
xjk (t ) =
2d jK 1
0,
t > a jK 1 + d jK 1
(6)
0,
t < a jk 1 d jk 1
t a jk 1 + d jk 1 , a d
jk 1
jk 1 t a jk 1 + d jk 1
2d jk 1
xjk (t ) =
1
a jk 1 + d jk 1 < t < a jk d jk
a +d t
jk
jk
,
a jk d jk t a jk + d jk
2d jk
0,
t > a jk + d jk
(7)
201
0,
t < a jK 1 d jK 1
t
a
d
jK 1
jK 1
, a jK 1 d jK 1 t a jK 1 + d jK 1
xjk (t ) =
2
d
jK
1,
t > a jK 1 + d jK 1
(8)
where t is the value of evaluation index; k=1, 2, ..., K-1; and j=1, 2, ..., m.
The comprehensive attribute measure analysis
The comprehensive attribute xjk can be expressed in the following form:
m
xk = j xjk
(9)
j =1
where xjk is the weight of evaluation index Ij, which should satisfy the following
form:
0 j 1
j = 1
(10)
j =1
Attribute recognition analysis

The confidence criterion: (C1, C2, , CK) is made as one ordered assessment set of
attribute space F. is defined as the confidence coefficient and is in the range 0.5 to 1.
The value of is commonly adopted as 0.6 to 0.7. The risk grade of the evaluation
object is adopted as Ck0, and k0 can be expressed in the following forms.
When C1 >C2 >>CK and the value of k0 satisfies equation (11):
k
k0 = min k : xl 1 k K
l =1
(11)
When C1 <C2 <<CK , and the value of k0 satisfies equation (12):

k
k0 = max k : xl
1 k K
l =k
(12)
ATTRIBUTE EVALUTATION INDICES

In practice, it is difficult to analyze the tunnel collapse by considering all
influenced factors of the tunnel collapse. The important factors such as geological
factors and engineering factors should be selected to analyze the possibility of tunnel
collapse. For practical purposes, the surrounding rock grade, tunnel depth, angle of
unsymmetrical pressure, rock mass integrity, and earthquake can be clarified as
geological factors. The engineering factors are commonly used in the analysis of the
dynamic risk assessment of tunnel excavation. The landform, rock structure
conditions, natural factors and human factor are considered in the above-mentioned
factors. Therefore, these attribute evaluation indices can be used widely in
engineering practice.
The analysis of attribute evaluation indices
The surrounding rock grade
The strength and integrity have great influence on the stability of surrounding rock. It
is well known that the stability of surrounding rock decreases with decreasing rock
mass strength. The longitudinal wave can be regarded as a judgment criterion of rock
grade.
ASCE
202
Table 2 Grade division of longitudinal wave velocity

Surrounding rock grade
VP (km/s)
<1.5
1.5~2.5
2.5~3.5
3.5~4.5
>4.5
The rock mass integrity

The coefficient of the rock mass integrity can be used to reflect the grade about
integrality of rock mass, and can be expressed by:
v2
k v = pm
(13)
v pr2
where Vpm is the rock mass elastic longitudinal wave velocity, and Vpr is the rock
elastic longitudinal wave velocity. The value of Vpm can be estimated from tunnel
seismic prediction (TSP).
Table 3 Grade division of intactness index of rock mass
Complete situation
kv
Integrity
0.75~1
Relatively integrity
0.55~0.75
Relatively broken
0.35~0.55
Broken
0.15~0.35
Extremely broken
0~0.15
The tunnel depth

The stability of tunnel decreases with decreasing tunnel depth.
Angle of unsymmetrical pressure
The angle of unsymmetrical pressure may have great influence on the the stability
of surrounding rock. The possibility of tunnel collapse increases with an increase in
the angle of unsymmetrical pressure.
Excavation span
Generally speaking, the probability of tunnel collapse increases with increasing
excavation span. The influence of tunnel span on the collapse is shown in Table 4.
Earthquake
Earthquake has great influence on the possibility of tunnel collapse. In general, it
can be assumed that the possibility of tunnel collapse increases with an increase in the
seismic intensity.
ASCE
203
Table 4 Grade of tunnel excavation span

Excavation span
Degree
<7
7~10
10~12
12~15
>15
Criteria of risk assessment

Based on the previous analysis, the risk assessment of tunnel collapse can be
divided into five grade, and the grade of indexes are listed in Table 5.
Table 5 Indexes and criteria for risk assessment of tunnel collapse
,>4.5
3.5~4.5
2.5~3.5
1.5~2.5
<1.5
Rock ass integrity
0.9~1
0.75~0.9
0.5~0.75
0.2~0.5
0~0.2
Tunnel depth
>60
40~60
20~40
10~20
<10
Angle of unsymmetrical
0~10
10~20
20~30
30~40
>40
Earthquake
none
slight
medium
serious
high serious
Excavation span
<7
7~10
10~12
12~15
>15
Risk grade of tunnel collapse
Structure attribute measure functions

When the data of five indexes (I1~I5) (see Table 6) are met the requirements of
Table 1, Eqs. (1) ~ (8) can be used to establish the attribute measure functions of
single index. The attribute measure of qualitative index I6 is adopted as 0 or 1.
Weights of assessment indexes
The importance degree of assessment object can be reflected by the corresponding
index weight of risk assessment for tunnel collapse. Expert scoring method, binomial
coefficient method and analytic hierarchy process can be commonly used to
determine the weight of risk assessment for tunnel collapse. Equation (13) can be
used to compute weights of assessment indexes of surrounding rock grade, tunnel
depth, angle of unsymmetrical pressure, rock mass integrity, earthquake and the
excavation span. One obtains:
w = w11 + w22
ASCE
(13)
where w1 can be estimated by the statistical results of 100 tunnel collapse examples;
and w2 can be obtained from the simple correlation function method. In the present
paper, the values of 1 and 2 can be adopted as 0.5[15].
The weights of the above-mentioned indexes can be written as:
= ( 45 ,w2 , w3, w4, w5, w6) = (0.298, 0.142, 0.088, 0.155, 0.213, 0.104).
ASCE
204
ASCE
0,
1,
15-t
,
10
0,
x 51 (t ) =
1,
8.5-t
,
3
0,
1,
15-t
,
x 41 (t ) =
10
0,
x 31 (t ) =
x 21 (t ) =
0,
0.95-t
,
0.1
1,
1,
x11 (t ) = t -4,
C1
t > 8.5
5.5 t 8.5
t < 5.5
t > 15
5 t 15
t <5
t > 15
5 t 15
t <5
t > 0.95
0.85 t 0.95
t < 0.85
t<4
4t 5
t >5
C2
0,
t < 0.675
t -0.675
, 0.675 t 0.825
0.15
1
0.825 < t 0.85
0.95-t
0.85 < t 0.95
0.1
0,
t > 0.95
t<3
3t 4
4 < t 5
t >5
0,
t -5.5
,
3
x 52 (t ) = 1,
11-t
2
0,
t > 11
9 t 11
8.5 < t < 9
5.5 t 8.5
t < 5.5
t <5
0,
t -15
, 5 t 15
x 32 (t ) = 10
25-t 15 < t 25
10
0,
t > 25
t <5
0,
t -5
, 5 t 15
x 32 (t ) = 10
25-t
15 < t 25
10
t > 25
0,
x 22 (t ) =
0,
t -3,
x12 (t ) =
5-t
0,
x 33 (t ) =
x 33 (t ) =
x 33 (t ) =
0,
t<9
t -9
, 8.5 t 9
10
13-t
, 9 t 11
10
0,
t > 11
0,
t < 15
t -15
, 15 t 25
10
35-t
, 25 < t 35
10
0,
t > 35
0,
t < 15
t -15
, 15 t 25
10
1
35 < t 45
45-t
, 35 < t 45
0.15
0,
t > 45
t < 0.375
0,
t -0.375
, 0.375 t 0.625
0.25
0.625 < t 0.675
x 23 (t ) = 1
0.825-t
,
0.675 t 0.825
0.15
t > 0.825
0,
t<2
0,
t -2, 2 t 3
x13 (t ) =
4-t , 3 < t 4
t>4
0,
C3
C4
t <1
1 t 2
2 < t 3
t >3
t < 11
0,
t -11
, 11 t 14
x 54 (t ) = 2
13t
, 14 < t 15
2
t > 15
0,
t < 35
0,
t -35
, 35 t 45
10
45 < t 55
x 34 (t ) = 1
65-t
, 55 < t 65
10
t > 65
0,
t < 35
0,
t -35
, 35 t 45
10
45 < t 55
x 34 (t ) = 1
65-t
, 55 < t 65
10
t > 65
0,
t < 0.05
0,
t -0.05
, 0.05 t 0.35
0.3
0.35 < t 0.375
x 24 (t ) = 1
0.625-t
,
0.375 < t 0.625
0.25
t > 0.625
0,
0,
t -1,
x14 (t ) =
3-t ,
0,
Table 6 Attribute measure functions of single index
t <1
t>2
t < 15
0,
t -13
, 15 t 17
2
t > 17
1,
x 35 (t ) =
t < 35
0,
t -35
, 35 t 45
x 35 (t ) =
10
1,
t > 45
t < 55
0,
t -55
, 55 t 65
10
t > 65
1,
x 35 (t ) =
t<
1,
20
7
0.35-t 1
,
t
x 25 (t ) =
20
20
0.3
7
0,
t>
20
0,
1,
x15 (t ) = t -1, 1 t 2
C5

205
206
Project profile
Cheng-Lan Railway was located in Songpan county of Sichuan province, China. It
was a single two-way online tunnel with a length of 6601 m. The surface elevation
was 2570-2810m, with relative height of 240 m, and 25- 40 slope. Tunnel crossed
the Triassic formation in Jurassic Zhuwo group T [3] which was sandstone with
phyllite, local clipping carbonaceous phyllite. On section of the middle-lower Triassic
formation of T[2] z which was the sandstone with rock. Test zone belonged to thin
layer phyllite, with the peak acceleration of 0.30 g. This tunnel was distributed along
the right bank of Minjiang River upstream. Herein the attribute recognition model of
risk assessment was used to assess the possibility of tunnel collapse of the section of
D1K212+088-D1K212+090 of the tunnel (see Fig.1).
Export mileage Dk212+845
Altitude/m
2900
2800
2700
2600
2500
Sandstone with slate
Sandstone with phyllite
Tunnel body
Fig.1 Profile of the geological prediction

Ensure the measure of evaluation indexes
Based on the results of tunnel seismic prediction (TSP), geological penetrating radar
(GPR), drilling and excavation about the section of D1K212+088-D1K212+090 of the
tunnel, it can be concluded that the stability of surrounding rock is poor. The
surrounding rock was broken and can be clarified as grade.
Based on the correlation information, the geology condition about the section of
D1K212+063~ D1K212+200 of the tunnel was sandstone with slate and sandstone
with phyllite. Therefore, the value of Vpr can be adopted as 4300 m/s[23]. By using the
results of the tunnel seismic prediction (TSP), the value of Vpm can be adopted as 2395
m/s. Based on Equation (12), the rock mass integrity can be taken as 0.31.
For the whole tunnel, the maximum and minimum depth was 880 m and 53 m,
respectively. Based on the correlation information, the tunnel depth of the section
D1K212+088 ~D1K212+090 was adopted as 290 m.
According to the local investigation, the angle of unsymmetrical press was
adopted as 45.
Due to the monitoring measurement, the excavation of D1K212+090 was
13.8667 m.
ASCE
207
An earthq
quake with a seismic intensity oof 6.3 happpened in Kaangding onn
Nov
vember 22, 2014. Thiss earthquakee damages serious to the tunnel, leading too
casu
ualties and economic
e
losss.
Table
T
7
Asssessment reesult of tunn

nel collapsee samples
In
ndex
Measured valuee
Attribute measure
C1
C2
C3
C4
C5
2.2
0.2
0.8
Rock maass integrity
0.31
0.87
0.13
Tunn
nel depth
290
Angle of unsym
mmetrical pressu
ure
45
Excavation span
13.8667
Eartthquake
Serious
0.088
0.06
0.679
0.1173
Co
omprehensive in
ndex attribute Meeasure
Result
R
nnel collapse is
degree
The ppossibility of tunn
Analysis of attribu
ute recognition
The order of assesssment set (C1, C2, C3, C4, C5, and C6) is C1>C2> C3>C
C4> C5> C6. T
The value of
pressented in Eq. (11) can be ado
opted as 0.7, and
a Eq. (11) caan be rewrittenn as:
k
k0 = miin k : xl 0.70, 1 k 5
l =1
(14))
The stabiliity grade of this

t tunnel collapse
c
is addopted as C4, indicatingg a high risk.
Thee result derrived from the assessm
ment methood is consisstent with the presentt
calcculated resullt. The ration
nality and feasibility of aapplication oof attribute m
mathematicss
theo
ory in risk asssessment off tunnel collaapse is verifi
fied.
Casse study on th
he attribute recognition
r
Thrrough the ab
bove-mentio
oned result, the possib ility about tunnel collaapse of thee
secttion D1K212+088-D1K
K212+090 is high. On N
November 223, a weak iintercalationn
mix
xed with the surrounding
g rock appeaared at the riight side of tthe arch foott position off
D1K
K212+088. The
T collapseed cavity sizze was 10 m 5 m (hoorizontal) (hhigh) 4 m
(verrtical), leadin
ng to sprayin
ng concrete cracking onn D1K212 + 090. Excavaation resultss
of rock of tunneel face are sh
hown in Fig.2.
(a)
(b)
Fig.2 Ex
xcavation reesults of rocck of tunnel face
ASCE
CONCLUSIONS
Based on attribute mathematical theory, an attribute synthetic valuation system is
presented to systematically evaluate the risk of tunnel collapse. To meet the
requirement of the data format of attribute mathematical theory, a couple of evaluation
indices are modified and quantitatively graded according to five risk grades through
expert evaluation method. And surrounding rock gradetunnel depthangle of
unsymmetrical pressure, rock mass integrity, earthquake and the excavation span are
selected as the evaluation indices. A case study of a mountain tunnel of Cheng-Lan
railway was used to verify the reliability of the present attribute recognition.
For practical purposes, the risk assessment can be used to evaluate risk grade, and
drilling and excavation can be used to verify the assessment results. This method can
be referenced to other engineering project.
ACKNOWLEDGMENTS
This work was supported by the State Key Development Program for Basic
Research Program of China (No. 2013CB036000), the State Key Program of National
Natural Science Foundation of China (No.51139004), the National Natural Science
Foundation of China (No.51479106), the National Natural Science Foundation of
China (No.51408338), the China Postdoctoral Science Foundation Special Funded
Project (No.2014T70641), and the Shandong Provincial Natural Science Foundation
of China (No. ZR2014EEQ009). The authors would be grateful to the reviewers for
their valuable comments and suggestions that can help improve the quality of the
paper.
REFERENCES
Alan, N.B. (2010) "Tunnel safety, risk assessment and decision-making. "Tunnelling
and Underground Space Technology, 25:91-94.
Berkowitz, et al (2002) "Characterizing flow and transport in fractured geological
media: a review. Adv. " Water Resources. 25 (8-12):861-884.
Chen, L and Huang, H.W. (2006) "The practice of risk management in Shangzhong
road tunnel engineering. "Chinese journal of Underground Space and
Engineering, 2(1): 65-69+73.
Chen, J.J., Zhou, F and Yang, J.S. et al (2009) "Fuzzy analytic hierarchy process for
risk evaluation of collapse during construction of mountain tunnel." Rock and
Soil Mechanics, 30(12):3622-362.
Cheng, Q.S. (1997) "Attribute recognition theoretical model with application."
Universitatis Pekinnensis, Acta Scientianrum Naturalium, 33(1):12-20.
Cheng, Q.S. (1997) "Attribute sets and attribute synthetic assessment system. "
Systems Engineering Theory & Practice, 17(9):1-8.
Einstein, H.H. (1996) "Risk and risk analysis in rock engineering. " Tunnelling and
Underground Space Technology,11(2):141-155.
ASCE
208
Henrik, B and Terje, A (2013) "Treatment of uncertainty in risk assessments in the

Rogfast road tunnel project. " Safety Science, 55:34-44.
Li, L.P. et al (2014) "Cause Analysis of Soft and Hard Rock Tunnel Collapse and
Information Management." Polish Journal of Environmental Studies,
23(4):1227-1233.
Li, L.P. et al (2015) "An attribute synthetic evaluation system for risk assessment of
floor water inrush in coal mines." Mine Water and The Environment, DOI
10.1007/s10230-014-0318-0.
Li, L.P., Li, S.C. and Chen, J. et al (2011) "Construction license mechanism and its
application based on comprehensive risk evaluation." Chinese Journal of Rock
Mechanics and Engineering, 30(7):1345-1354.
Li, S.C. and Zhou, Z.Q. et al (2013) "Risk assessment of water inrush in karst tunnels
based on attribute synthetic evaluation system." Tunnelling and Underground
Space Technology 38:50-58.
Li, S.C., Shi, S.S. and Li, L.P. et al (2013) "Attribute Recognition Model and Its
Application of Mountain Tunnel Collapse Risk Assessment. " Journal Of Basic
Science And Engineering, 21(1):147-158.
Li, X.P. and Li, Y.N. (2014) "Research on risk assessment system for water inrush in
the karst tunnel construction based on GIS: Case study on the diversion tunnel
groups of the Jinping II Hydropower Station. " Tunnelling and Underground
Space Technology, 40:182-191.
Rita, L.S. and Herbert, H.E. (2011) "Risk analysis during tunnel construction using
Bayesian Networks Porto Metrocase. " Tunnelling and Underground Space
Technology, 27(1):1-15.
Qian, Q.H. and Gong, X.L. (2008) "State issues and relevant recommendations for
security risk management of Chinas underground engineering. " Chinese
Journal of Rock Mechanics and Engineering, 27(4):649-655.
Wang, Z.Z. and Zhang, Z (2013) "Seismic damage classification and risk assessment
of mountain tunnels with a validation for the 2008 Wenchuan earthquake. " Soil
Dynamics and Earthquake Engineering, 45:45-55.
Zhou, Z.Q., Li, S.C., Li, L.P. and et al (2013) "Attribute recognition model and its
application of fatalness assessment of water inrush in karst tunnels." Rock and
Soil Mechanics, 34(3):818-826.
ASCE
209
Gaussian Process Model of an Advanced Surrounding Rock Classification Based on

Tunnel Seismic Predictions
Peng He1; Li-Ping Li2; Qian-Qing Zhang3; Fei Xu4; Jie Hu5; and Jian Zhang6
1
Ph.D., Geotechnical and Structural Engineering Research Center, Shandong Univ., Jinan 250061,
Shandong, China. E-mail: hepenghank@163.com

2
Associate Professor, Geotechnical and Structural Engineering Research Center, Shandong Univ., Jinan
250061, Shandong, China (corresponding author). E-mail: 378246365@qq.com

3
Associate Professor, Geotechnical and Structural Engineering Research Center, Shandong Univ., Jinan
250061, Shandong, China. E-mail: 631589743@qq.com

4
Shandong, China. E-mail: sduxufly@126.com

5
Shandong, China. E-mail: 1031635601@qq.com

6
Shandong, China. E-mail: 471324300@qq.com
Abstract: In order to improve the prediction methods of geological hazards in the front
of a tunnel face. Gaussian process machine-learning is proposed and the classification
model is established for advanced surrounding rock classification based on TSP-203
system. This approach extracts the useful information from the detection results of
TSP-203, and establishes the quantitative index system of surrounding rock classification.
Then the unknown nonlinear mapping between the detected geological information and
the actual surrounding rock level is established by Gaussian process classification model
(GPC) ultimately. To verify the feasibility and superiority, GPC model is applied to
Jinpingyan tunnel on Chenglan railway in China and the results show that GPC performs
a higher prediction accuracy than support vector machine (SVM) and grey clustering
method in predicting surrounding rock level, providing a new approach for advanced
surrounding rock classification.
INTRODUCTION
The engineering quality classification of tunnel surrounding rock is an important
aspect of the safety assessment for underground engineering in complex conditions,
contributing to the design of tunnel excavation and support. To expedite the evaluation of
geotechnical structures as well as to prevent rockfall in excavated regions, many rock
mass classification methods have been developed constantly, including the rock quality
designation (RQD) system (Deere 1964), the rock mass rating (RMR) system
(Bieniawski 1984, 1989), the Norwegian Geotechnical Institute Q-system (Q) (Barton et
al 1974; Barton 2002) and the geological strength index (GSI) (Hoek and Brown 1997)
etc. However, these methods based on observations and experiments have many
ASCE
210
inadequacies, ignoring the impact of complexity, uncertainty and ambiguity features on

rock mass quality classification, and often lead to inconsistency with the actual
assessment results. Moreover, the tests (also called destructive methods) are very time
consuming and expensive. Besides, the sampling program is very sparse, and hence the
results cannot be representative of the entire structure.
Recently, non-destructive methods of prediction such as Ground Penetrating Radar
system (GPR) and Tunnel Seismic Prediction (TSP) are developed continually. Compared
to the above methods, nondestructive methods are more cost-effective and impact little on
the construction. A number of landmark projects such as Dafengyakou Expressway
Tunnel in China, Aibetsu Highway Tunnel in Japan and the Vereina Rail Tunnel in
Switzerland have utilized this method in their construction successfully (see Geotech Co.
2004). However, both TSP and GPR are based on the single factor judgment of the
longitudinal wave velocity, and lead to less accurate results in the surrounding rock
classification.
In recent years, some computational intelligence techniques such as the artificial
neural network (Suwansawat et al 2006; Wei et al 2013), support vector machines (Wang
et al 2012; Yao et al 2014), and the Gaussian process (Su and Zhang 2010; Liu et al 2011)
have been successfully applied for surrounding rock classification and some other related
subjects of time series analysis. Especially the Gaussian process model, a state-of-the-art
supervised learning approach, which derives the mapping from probability theory with
prior and covariance functions, can apply the mean of the distribution as point predictions
to avoid robust point predictions like that in artificial neural network and support vector
machine.
According to the above problems, this paper aims to establish the Gaussian process
model for advanced surrounding rock classification based on the geological information
detected by TSP system. To verify the feasibility and superiority, the Gaussian process
model is applied to predict surrounding rock classification of the Jinpingyan tunnel on
Chenglan railway in China, and the results are analyzed comparing with support vector
machine (SVM) and grey clustering method.
THE EVALUATION INDICES OF TSP
TSP is based on propagation and reflection of seismic waves through geotechnical
layers in front of an excavation face in a tunnels route and the response characteristics of
various typical geological defects to the elastic waves are studied. The accurate VP can be
obtained from the result of TSP, and the other parameters such as Poison's ratio and
Young modulus can be obtained by VP and empirical formulas. Moreover, the angle
between main structural surface and the tunnel axisdiscontinuous structural surface and
the state of groundwater can also be interpreted by maps of 2D tomography, depth
migration and reflector extraction (Barton 2002). Besides, qualitative description should
be quantified to match with the Gaussian process machine-learning algorithm. To be
ASCE
211
212
simplified, related literatures (Qiu et al 2010; Raoof et al 2013) can be consulted for the
specific criterion.
The faults and fractured zones can be identified according to the following
observations:
(1) The velocity of the vertical reflection wave increases slightly, and the velocities of
the vertical and horizontal reflection waves decrease;
(2) Depth migration begins with strong negative reflection and ends with strong
positive reflection, while there are many unorganized positive and negative reflective
layers in the reflection bands.
Besides, the water-bearing geological structures can be detected based on the
following evidences:
(1) In a 2D tomography map, the velocity of horizontal reflection wave decreases
while that of vertical reflection wave decreases or increases slightly; both the ratio of the
velocity of vertical reflection wave to that of horizontal reflection wave and the Poissons
ratio of rock mass increase suddenly; the density and Youngs modulus of rock mass
decrease greatly;
(2) The negative reflection is obvious and there are few reflection surfaces after the
primary reflection surface (Liu et al 2003).
GAUSSIAN PROCESS CLASSIFICATION MODEL
Gaussian process (Rasmussen C.E and Williams CKI 2006) is a nonlinear kernel
based learning method, focusing on modeling the posterior directly. In many applications,
it outperforms most other state of the art supervised learning methods. A Gaussian
process (GP) is fully specified by its mean function m(x) and kernel function k(x, x),
expressed as
f ( x ) ~ GP ( m ( x ), k ( x , x) )
(1)
For notational simplicity, it is common to consider the mean function of the GP to be

zero (Rasmussen C.E and Williams CKI 2006). A GP model is a non-parametric,
probabilistic (Bayesian) model in function space. One can think of a GP as defining a
distribution over functions, and inference taking place directly in the space of functions.
The kernel function characterizes correlations between different data points in the
Gaussian process, and can be learnt from the data. The inference can be carried out
directly under the GP framework by learning a kernel function from the training data. The
kernel function studied in this paper is the widely used Radial Basis Function (RBF),
which is defined as
2
k ( x, x) = 2 exp( x x / (2l 2 ))
(2)
where x and x are input vector pairs, l is the characteristic length scale and 2 is the
signal variance. The free parameters, i.e., l and 2, are called hyperparameters of GP
model.
ASCE
213
Assume that we have a dataset D with n observations: D={(xi,yi), i=1,2,,n}, where x

is the input vector of dimension d and y is the class label +1/1. The input dn data
matrix is denoted as X. Predictions for new inputs x are made out of this given training
data using the GP model. GP binary classification is performed by first calculating the
distribution over the latent function f corresponding to the test case (Rasmussen C.E and
Williams CKI 2006):
p ( f X , y , x) = p ( f X , x, f ) p ( f X , y ) df
(3)
where p ( f X , y ) = p ( y f ) p ( f X ) / p ( y X ) is the latent variable posterior, p ( f X , x, f ) is

the predictive posterior with respect to possible latent functions, and values of this could
lie anywhere within the range of (,+). So the probabilistic prediction is made by
= p ( y = +1 X , y , x) = s ( f ) p ( f X , y , x) df
(4)
where s can be any sigmoid function that squashes the prediction output to guarantee a
valid probabilistic value within the range of [0,1].
For multi-class classification problem, we can treat each one class as being
independent from the others, and apply binary classification individually to each (one)
class versus the rest classes as shown in Fig.1.
4
0.9
0.8
0.7
0.6
0.5
-1
0.4
-2
0.3
-3
0.2
-4
-4
-2
0.1
Fig.1 Sketch map of classification by GP model

Engineering situation
The Jinpingyan tunnel on Chengdu-Lanzhou railway, which has a total length of
12773m, locates in Songpan county of Sichuan province, the Tibet altiplano eastedge of
China. With weak and fractured rock mass, most of the surrounding rock of Jinpingyan
tunnel is grade III-V. Benching stepping method is adopted in the excavation, and
bolt-shotcrete primary support is immediately constructed after each excavation cycle
completed. During the construction process, the tunnel seismic prediction is applied to
detect the geological information in the front of tunnel face the whole time.
ASCE
214
Learning sample and model training

Taking 35 representative tunnel sections of Jinpingyan tunnel as samples, and the
geological information of different tunnel sections detected by TSP-203plus (i.e., 35 data
points) are randomly divided into training (25 data points) and testing data (10 data
points). Considering the limitation of space, only part of samples is listed in Table 1. The
GPC model seeks the nonlinear mapping between the detected geological information
and the actual surrounding rock level determined from direct observation and evaluation
after excavation.
Table 1 Original data of samples
No.
Vp / VS
State parameter of
structure plane
Ground water
condition
Grade
1
2
3
4
5
6
1.83
1.92
1.75
1.62
1.75
1.84
0.29
0.31
0.26
0.19
0.26
0.29
17
21
19
18
18
14
62
53
42
61
71
38
0.31
0.52
0.67
0.45
0.49
0.83
0.29
0.77
0.62
0.37
0.37
0.51
III
V
IV
IV
III
V
Before starting the training process, input and output data were scaled to be between
the upper and lower bounds of the transfer functions (usually between 1 and 1). To
transform and normalize the data among various normalization techniques, the minmax
normalization method (Jayalakshmi and Santhakumaran 2011) is adopted, as it is very
simple but also very sophisticated.
Engineering application
To verify the capability of GP classification model in engineering application, this
model was applied to Jinpingyan tunnel on Chenglan railway so as to forecast the
surrounding rock level in advance. In order to obtain the geological condition before
tunnel face, TSP-203plus is adopted to collect the reflection seismic wave data of a
certain tunnel section between D2K200+525 and D2K200+655 (total length is 130m).
The maps of 2D tomography, P, SH waves depth migration and reflector extraction
can be obtained by TSPwin (see Fig.2). Moreover, 2D pictures of prediction results can
also be interpreted as shown in Fig.3. The surrounding rock in predicted zone is divided
into several sections according to P, SH wave, poisson's ratio and young's modulus
obtained by TSP (see Fig.4). Besides, the angle between main structural surface and the
tunnel axisdiscontinuous structural surface and the state of groundwater can also be
interpreted according to Fig.3 and Fig.4 respectively. After quantified (Qiu et al 2010), all
the input data for training and testing is obtained as shown in Table 2.
ASCE
215
30
X /m
60
50
90
120
60
50
90
120
25
R/m
R/m
25
X /m
30
-25
-25
-50
-50
(a) P wave depth migration
(b) SH wave depth migration

X /m
30
60
90
120
50
R/m
25
0
-25
-50
(c) reflector extraction

Fig.2 Diagrams of P, SH wave depth migration and reflector extraction
0
X /m
90
45
50
120
180
X /m
90
45
120
180
Vp
R/m
25
Vs
0
Vp/Vs
-25
E
-50
R/m
X /m
90
45
120
180
10
0
-10
Fig.3 2D pictures of prediction results
Fig.4 Mechanical parameters of rock mass
Table 2 Physico-mechanical parameters of rock mass
ASCE
Distance
Vp / VS
State parameter
of structure plane
Groundwater
condition
YK200+525~546
YK200+546~591
YK200+591~620
YK200+620~638
YK200+638~655
1.75
1.83
1.94
1.64
1.51
2.41
2.42
2.34
2.43
2.46
0.26
0.29
0.32
0.20
0.11
18
18
11
20
24
63
52
27
35
25
0.35
0.56
0.81
0.62
0.77
0.31
0.37
0.73
0.65
0.52
216
During to the limitation of space, the detailed description and specific steps of GP
classification can be found in related research (Rasmussen C.E and Williams CKI 2006),
and the procedures are executed in Matlab2011a. With well-trained GPC model above, the
surrounding rock level of Jinpingyan tunnel between D2K200+525 and D2K200+655 in
front of the tunnel face can be predicted in advance as shown in Table 3. To verify the
superiority of GPC model, the surrounding rock in predicted zone is also classified by
grey clustering method and support vector machine (see in Table 3).
Table 3 Contrast of predicted classification results
Distance
Design level
Actual level
GPC
SVM
Grey cluster
YK200+525~546
YK200+546~591
YK200+591~620
YK200+620~638
YK200+638~655
IV
IV
IV
IV
IV-V
III- IV
IV
V
IV
V
III
IV
V
IV
V
IV
IV
V
IV
IV
III
IV
V
V
V
According to the results as shown in Table 3, we can conclude that the classification
results predicted by GP model are consistent with actual exposed surrounding rock level
completely, which means the GP classification model is totally feasible for predicting
surrounding rock level in advance. Moreover, we can get higher precision classification
result using GP machine-learning model than support vector machine and grey clustering
method, which performs the superiority of GP model to solve the highly nonlinear
problem with small samples and high dimension.
CONCLUSION
In this work, Gaussian process classification model (GPC) is established for advanced
surrounding rock classification based on the geological information detected by TSP-203
system. To verify the feasibility and superiority, GPC model is applied to Jinpingyan
tunnel on Chenglan railway in China, and performs high prediction accuracy, providing a
new approach for advanced surrounding rock classification. Moreover, as a newly
developed machine-learning method, Gaussian process shows a excellent capability than
support vector machine (SVM) and grey clustering method for solving the highly
nonlinear problem with small samples and high dimension. However, the performance of
the classification results depends on the quality of data and the ability of the mapping,
which directly affected by the parameters. Therefore, further work can be tried to analyze
the effects of parameters for the computational intelligence approaches.
ACKNOWLEDGEMENTS
This work was supported by the State Key Development Program for Basic Research
Program of China (No. 2013CB036000), the State Key Program of National Natural
Science Foundation of China (No.51139004), the National Natural Science Foundation of
China (No.51479106). The authors would be grateful to the reviewers for their valuable
comments and suggestions that can help improve the quality of the paper.
ASCE
REFERENCE
Barton N, Lien R, Lunde J (1974) Engineering classification of rock masses for the
design of tunnel support. Rock Mech 6(4):189236
Barton N (2002) Some new Q-value correlations to assist site characteristics and tunnel
design. Int J Rock Mech Min Sci39:185216
Bieniawski ZT (1984) Rock mechanics design in mining and tunnelling. A.A. Balkema,
Rotterdam
Bieniawski ZT (1989) Engineering rock mass classification. Wiley, Rotterdam, p 251
Deere DU (1964) Technical description of rock cores for engineer purposes. Rock Mech
Eng Geol 1(1):1722
Geotech Co., 2004. TSP 203: Case Histories, Amberg Co., Switzerland, p.26.
Hoek E, Brown ET (1997) Practical estimates of rock mass strength. Int J Rock Mech
Min Sci 34(8):11651186
Jayalakshmi T, Santhakumaran A (2011) Statistical normalization and back propagation
for classification. Int J Comput Theor Eng 3(1):17938201
Liu YK, Fang Y, Liu BG, Xu C (2011) Intelligent deformation prediction model of tunnel
surrounding rock based on Genetic-Gaussian process regression coupling algorithm.
Journal of the China Railway Society 33(12):102-106.
Liu ZGLiu XF (2003) TSP application and development in tunnel lead forecast. Chinese
Journal of Rock Mechanics and Engineering, 22 (8): 1 3991 402 (in Chinese).
Qiu DH, Li SC, Zhang QS et al. (2010) Prediction of surrounding rock classification in
advance based on TSP203 system and GA-SVM. Chinese Journal of Rock
Mechanics and Engineering 29(1): 3222-3226.
Raoof G, Vamegh R, Andisheh A (2013) Improved RMR rock mass classification using
artificial intelligence algorithms. Rock Mech Rock Eng 46:11991209
Rasmussen CE, Williams CKI (2006) Gaussian processes for machine learning. The MIT
Press, Cambridge.
Su GS, Zhang Y (2010) Forecasting Deformation Time Series of Surrounding Rock for
Tunnel Using Gaussian Process. CISE2010:1-4.
Suwansawat S, Einstein HH (2006) Artiifcia1 neural networks for predicting the
maximum surface settlement caused by EPB shield tunneling. Tunnelling and
Underground Space Technology 2l(2):133-150
Wang DD, Qiu GQ, Xie WB, Wang Y (2012) Deformation prediction model of
surrounding rock based on GA-LSSVM-markov. Nat Sci 4(2):8590.
Wei J, Qi J, Wu Y, Lu YL, Wang L (2013) Prediction of the deformation of the
surrounding rock around tunnels by GA-Bp network model. Applied Mechanics and
Materials 256:1157-1160.
Yao B, Yao J, Zhang M, Yu L (2014) Improved support vector machine regression in
multi-step-ahead prediction for rock displacement surrounding a tunnel. Scientia
Iranica. Transaction A, Civil Engineering 21(4):1309-1316.
ASCE
217
218
Energy Characteristics of Micro-Seismic Events Associated with Immediate

Rockbursts in Deep Tunnels
Yang Yu1; Bing-Rui Chen2; Chang-Jie Xu1; Xin-Hong Diao1; Li-Hong Tong1; and Yu-Feng
Shi1
1
College of Civil Engineering and Architecture, East China Jiaotong Univ., Nanchang 330013, China.
E-mail: lukeryuyang@163.com
2
State Key Laboratory of Geomechanics and Geotechnical Engineering, Institute of Rock and Soil
Mechanics, Chinese Academy of Sciences, Wuhan 430071, China.
Abstract: This study integrates micro-seismic data and data from hundreds of
rockbursts of different intensities that occurred in four deep headrace tunnels and one
water drainage tunnel at Jinping II hydropower station in Sichuan province, China.
The overburden of the studied tunnels is between 1900 m and 2525 m, largely
excavated in marble. The tunnels have a total length of 12.4 km. The micro-seismic
energy produced during the development of immediate rockbursts excavated with the
drill-and-blast method (DBM) and a tunnel boring machine (TBM) were studied. The
results indicated that the daily maximum micro-seismic energy can be used as a basis
for estimating the rockburst intensity: no rockburst, weak rockbursts, moderate
rockbursts, and intense rockbursts, and that the common logarithms of the microseismic energy of each level was <4, >4 and <5, >5 and <6, >6, respectively. For the
same intensity, rockbursts excavated by TBM and DBM have the same range but
different distribution characteristics. The micro-seismic energy can be used as a guide
to establish a dynamic warning system based on the evolution of micro-seismic events
to reduce the risk of immediate rockbursts during excavation of deep, hard-rock
tunnels.
Keywords: Drill-and-blast method (DBM); Tunnel boring machine (TBM);
Immediate rock burst; Micro-seismic events; Micro-seismic energy; Rockburst
intensity.
INTRODUCTION
Rockbursts that occur during the excavation of hard-rock tunnels at great depths
or very high stress can cause loss of life and construction delays as well as economic
loss. For example, more than 700 rockbursts occurred during the excavation of two
assistant tunnels, four headrace tunnels and one drainage tunnel in the Jinping II
hydropower project, China. (He et al. 2010, Zhang et al. 2012).
Rockbursts in deep, hard-rock tunnels can be divided into immediate and timedelay rockbursts based on their time of occurrence. The former, which occur
contemporaneously with microseismic events on an hourly or daily basis, are
ASCE
219
rockbursts within 030 m of the working face that happen approximately a few hours
to several days after excavation. The latter occur several days or several months after
excavation of tunnels with discontinuous occurrences of microseismic events (Chen et
al. 2012). In deep, hard rock tunnels most of the rockbursts are immediate rockbursts
(Zhang et al. 2008) which are also studied in this article. Based on the study of the
intensity of the rockburest (Tang et al. 1992, Wang et al. 1999), immediate rockbursts
in diversion tunnels of Jinping II hydropower station were of three levels of strength,
weak, moderate and intense, based on the sound, shape, and failure features of the
rockburst, as shown in Table 1.
Table 1. Intensity of Immediate Rockbursts in Deep Tunnels of the Jinping II
Hydropower Station
Intensity
Main failure type
Sound
Depth of
failure
Construction
Affected
Weak
rockburst
slight spalling and

slabbing in surface
of surrounding rock
mass; rock mass was
not ejected;
cracking sound can

be heard
occasionally
Less than
0.5 m;
barely affected
Moderate
rockburst
severe spalling and

slabbing of
surrounding rock
mass; rock mass was
slightly ejected;
cracking sound like

a detonator blasting;
slight cracking
sound lasting for
some time inside the
rock mass
Intense
rockburst
a great deal of rock

mass was suddenly
ejected with great
power;
the failure sounds

like an explosive
blast; it is loud and
lasts longer
more than
0.5 m
and<1.0 m
and failure
range is
obvious
more than
1.0 and
failure
range is
extensive
affected to
some extent
is seriously
affected
Stress waves generated by the process of rock mass failure are called microseismic. These can be detected by seismic instruments. Analysis and processing of
these micro-seismic waves can determine size, time, space, energy, and other
information about the microfracturing occurring in the process of rock mass failure.
Microseismic monitoring has been successfully applied during the construction of
deep tunnels (Feng et al., 2012, 2015). Micro-seismic energy is one of the most
important parameters describing the size of micro-seismic events (Chen et al. 2013).
In the high stress conditions during TBM and DBM excavation, the stress path
and extent of damage are different (Barton 2000). This paper will focus on the
relationship between the Micro-seismic energy and the rockburst during two different
excavation methods, TBM and DBM, in deep tunnels. Rockburst intensity and the
evolutionary mechanisms of the maximum energy micro-seismic events are studied,
respectively. The range and distribution characteristics of micro-seismic energy in the
evolutionary process of different intensity rockbursts involve contrastive analysis, and
few studies have explored these processes. This study thus addresses that gap.
ASCE
220
2 Project Introduction and Micro-seismic Monitoring

The case study in this paper is Jinping II Hydropower Station, which is located on
the Yalong River in Sichuan Province in southwest China. Figure 1 shows the
geographical location and geological section of diversion tunnels in the Jinping II
hydropower station, whose length is about 16.7 km on average.
Yalong river
N
Jinping II Hydropower
Station
4193 m
Jinping mountain
4448 m
1630 m
diversion tunnels
4309 m
Hai ganzi
Yalong river
Greenschist
Slate
Fault
2525 m
Jinping I Hydropower
Station
Marbl
17285 m
West
Eest
FIG. 1. Location and geological section of diversion tunnels for the Jinping II
hydropower station
DBM excavation
K6+168
TBM excavation
K7+344
K8+530
1#
2#
3#
4#
P#
A#
B2#
B3#
B1#
B#
3# K5+240 4# K5+097 P# K4+810 1# K9+138 2# K9+201 3# K9+003

4# K9+052P# K8+575 3# K11+165 3# K10+049
FIG. 2. Ranges of micro-seismic monitoring of TBM and DBM excavation of
diversion tunnels for the Jinping II hydropower station (1# 2#, 3# and 4# are the
indexes of the 4 headrace tunnels, P# is the drainage tunnel, A# and B# are traffic
tunnels, and B1#, B2# and B3# are branches of traffic tunnel B# that were used to
accelerate the construction of the four headrace tunnels).
ASCE
221
The rock surrounding the diversion tunnels is mainly comprised of

and
category Baishan marble (Poissons ratio is 0.21-0.33, Maximum principle stress is
46-73MPa, compressive strength is 55-114MPa, elasticity modulus is 25-40GPa,
modulus of deformation is 8-16GPa) (Zhang et al. 2012). The tunnels are covered by a
rock mass with a thickness of over 1500 m is about 75%, and the maximum depth of
the tunnels is about 2525 m.
Micro-seismic monitoring took place on the 5 diversion tunnels, the 4 headrace
tunnels, and the drainage tunnel. TBM excavation (3# headrace tunnel chainage
K10+049~11+165) and DBM excavation (1#, 2#, 3#, 4# headrace tunnels and drainage
tunnel) are both used in the course of construction, as shown in Fig. 2. The
construction process indicated that rockbursts occurred frequently and caused
significant safety problems, with primarily weak to moderate intensity and in local
areas, even intense rockburst occurred (Feng and Zhou 2006).
3 Micro-seismic Energy Characteristics of Immediate Rockbursts
3.1 Occurrence of immediate rockbursts
An intense rockburst with a loud sound like blasting occurred at the northern
sidewall of the No. 3 headrace tunnel at chainage K11+045 (centreline of the
rockburst) at approximately 16 p.m. on September 9, 2010, during TBM excavation.
The surrounding rock was coarse-grained white marble T2b with no geological
structural planes. The failure pit of the rockburst was approximately 1.4 m deep, 8 m
wide, and 9 m high, and the surface of the rockburst pit was shown in Fig. 3a. Microseismic events 10 metres ahead of and 30 metres behind the working face are
generally selected in the warning of immediate rockbursts (Feng et al., 2013). This
range is also used in this study of rockburst. The spatial distribution of microseismic
events in this zone (10 metres ahead of and 30 metres behind the centreline of the
rockburst) during the development of the rockburst with maximum and minimum
energy indices (logarithmic) of 6.33 and 1.10, respectively, is shown in Fig. 3b.
Rockburst zone
ASCE
222
North
10 m
30 m
South
Advancing direction
3# headrace tunnel
Centreline of the rockburst
Micro-seismic event
FIG. 3. (a) Photograph after the intense rockburst in the 3# headrace tunnel,
on September 9, 2010, (b) Micro-seismic events of the rockburst
3.2 Relationship between the micro-seismic energy and the intensity of rockburst
3# headrace tunnel chainage K10+632-11+165 (as shown in Fig. 2) with
overburden depths between 1964 m and 2029 m and lengths of 1.12 km, was
excavated by TBM. The surrounding rock (which has good integrity, Karst
nondevelopment, and moderate amounts of groundwater) is mainly comprised of
marble. Fig. 4 show the rockburst intensity and evolutionary mechanisms of the
maximum energy micro-seismic events in the warning zone. Fig. 4 indicates that
maximum energy micro-seismic events along the working face can be used as a basis
for estimating the immediate rockburst intensity. There were <4 zero maximum
energy (logarithmic) rockburst events, >4 and <5 rockbursts with weak maximum
energy (logarithmic) were, >5 and <6 rocketbursts with moderate maximum energy
(logarithmic), and >6 rockbursts with intense maximum energy (logarithmic).
9
None
rockbrust
No rockburst
Weak rockbrusts
Moderate rockbrusts
Intense rockbrusts
lglge
e
6
5
4
3
2
1
0
27-May
6-Jun
16-Jun
26-Jun
6-Jul
16-Jul
26-Jul
date
FIG. 4. Rockburst intensity and the evolutionary mechanism of maximum

energy micro-seismic events
ASCE
223
DBM excavation drainage tunnel chainage SK4+810~5+196 (as shown in Fig. 2)

has the same geological conditions and overburden depths as the TBM excavation.
The relationship between rockburst intensity and maximum micro-seismic energy
(logarithmic) have the same characteristics as those of the TBM excavation: in intense
rockbursts greater than 6, moderate rockbursts from 5 to 6, weak rockbursts from 4 to
5, and no rockburst less than 4, as shown in Fig .5.
9
8
None
rockbrust
No rockburst
Weak rockbrusts
Moderate rockbrusts
Intense rockbrusts
lg e
Lge
6
5
4
3
2
1
0
23-Jul
28-Jul
2-Aug
7-Aug
12-Aug
17-Aug
22-Aug
27-Aug
1-Sep
date
FIG. 5. Rockburst intensity and evolutionary mechanism of maximum energy

micro-seismic events during the DBM excavation
3.3 Micro-seismic energy characteristics of different types excavation methods
The micro-seismic energy (logarithmic) distribution of rockbursts of the same
intensity is shown in Fig. 6 (all of the micro-seismic events during TBM and DBM
excavetion in the monitoring range are shown in Fig. 2). For the same rockburst
intensity, the range of micro-seismic energy remains the same for either type of
excavation methods: no rockburst range from -1 to 4, weak rockbursts from -1 to 5,
moderate rockbursts from -1 to 6, and intense rockbursts from -1 to 7. The microseismic energy distribution during TBM excavation was higher than during DBM. Fig
6 a, b, c, and d also indicate that the energy distribution of micro-seismic events
during the evolution of rockbursts of different intensity rockburst is not same (intense
rockbursts > moderate rockbursts > weak rockbursts > no rockburst).
T BM excavation
DBM excavation
Curve of DBM excavation
Curve of T BM excavation
50
40
30
20
10
0
40
30
20
10
0
(-1,0)
(0,1)
(1,2)
Lge
(a)
ASCE
T BM excavation
DBM excavation
50
Rate of microseismic event%
60
(2,3)
(3,4)
(-1,0)
(0,1)
(1,2)
(2,3)
Lge
(b)
(3,4)
(4,5)
T BM excavation
DBM excavation
35
30
TBM excavation
DBM excavation
Curve of TBM excavation
35
30
40
224
25
20
15
10
5
0
25
20
15
10
5
0
(-1,0)
(0,1)
(1,2)
(2,3)
(3,4)
(4,5)
(5,6)
(-1,0)
(0,1)
(1,2)
Lge
(2,3)
(3,4)
(4,5)
(5,6)
(6,7)
Lge
(c)
(d)
FIG. 6. Energy distribution of micro-seismic events in immediate rockbursts of

different intensity during TBM and DBM excavation (a) no rockburst (b) weak
rockbursts (c) moderate rockbursts (d) intense rockbursts
Both the distribution range and distribution characteristics of the micro-seismic
energy move in the direction of high energy as the level of rockburst intensity rises. In
the evolution of processes of rockbursts of the same intensity during TBM and DBM
excavation, the range is the same but the distribution characteristics differ (TBM
excavation was larger than that of DBM). TBM excavation leads to Quasi-static
unloading, in which micro-seismic events are caused by the unloading process of the
surrounding rock. The micro-seismic energy is concentrated in the range from 101J to
105J (approximately 83.4 percent) which is higher. The initial in-situ stress of the
working face consists of dynamic unloading stress fields during DBM excavation. The
integrality and bearing capacity of the surrounding rock are reduced by blasting
effects. When combined with the impact of the micro-cracking caused by the blasting
stress wave, its micro-seismic energy is concentrated in a range of less than 104J
(approximately 87.1 percent), which is lower, as shown in Table 2.
Table 2. Energy Distribution of Different Types Excavation Methods
Excavation
Method
TBM
LgE
-1~0(%)
1.9
LgE
0~1(%)
4.6
LgE
1~2(%)
10.2
LgE
2~3(%)
23.2
LgE
3~4(%)
31.4
LgE
4~5(%)
18.6
LgE
5~6(%)
8.3
LgE
6~7(%)
1.8
DBM
26.2
18.6
18.5
13.4
10.4
7.6
3.9
1.4
4 CONCLUSIONS
This study investigates the micro-seismic energy in the evolutionary process of
immediate rockbursts in four deep headrace tunnels and a water drainage tunnel, with
overburdens of between 1900 m and 2525 m, at Jinping II hydropower station, China.
The following conclusions are drawn:
(1) The maximum energy of micro-seismic events of the working face can be
used as a basis for estimating the immediate rockburst intensity.
ASCE
(2) No rockburst is less than 104J, weak rockbursts are >104J and <105J,
moderate rockbursts are >105J and <106J, and intense rockbursts are >106J.
(3) In the evolutionary processes of same intensity rockbursts, the micro-seismic
energy has the same distribution range but different distribution characteristics for
each excavation method. For TBM excavation the micro-seismic energy (concentrated
in the range 101J to 105J) is higher than DBM excavation (concentrated in the range
less-than 104J). Both moving toward higher energy as the intensity the rockbursts
rises.
ACKNOWLEDGMENTS
The authors acknowledge the financial support from The National Natural Science
Foundation of China under Grant No. 51509092.
REFERENCES
Barton A. A. N. (2000). "TBM tunneling in jointed and faulted rock." Rotterdam:
Balkema, pp:6l-64.
Chen, B.R., Feng, X.T., Ming H.J. et al., (2012). "Evolution law and mechanism of
rockbursts in deep tunnels: time delayed rockburst." Chinese Journal of Rock
Mechanics and Engineering, Vol. 31: 561-569 (in Chinese).
Chen, B.R., Feng, X.T., & Li Q.P. (2013). "Rock Burst Intensity Classification Based
on the Radiated Energy with Damage Intensity at Jinping II Hydropower Station,
China." Rock. Mech. Rock. Eng., Vol. 48(1): 289303.
Feng X.T., Zhou H. (2006). "Research report on stability of surrounding rock mass
and structure design of diversion tunnels of Jinping II hydropower station at
Yalong River during invite public bidding and design phase." Institute of Rock
and Soil Mechanics, Chinese Academy of Sciences.
Feng, X.T., Chen, B.R., & Li S.J. (2012). "Study on evolution process of rockbursts in
deep tunnels." J. R. Mech. Geotech. Eng., Vol. 4: 289-295.
Feng, X.T., Chen, B.R., & Zhang C.Q. (2013). "Mechanism warning and Dynamic
Control of Rockburst Development Processes." China Social Sciences Publishing
House, Beijing(in Chinese).
Feng G.-L., Feng X.-T., Chen B.-R., Xiao Y.-X., Jiang Q. (2015). "Sectional velocity
model for microseismic source location in tunnels." Tunnelling and Underground
Space Technology. Vol. 45:7383.
He, M.C., Miao J.L., Feng J.L., (2010). "Rock burst process of limestone and its
acoustic emission characteristics under true-triaxial unloading conditions." Int J
Rock Mech Min Sci. Vol. 47: 286298
Tang, Y. (1992). "A new classification of rockburst intensity." Geol. Rev. Vol. 38(5):
439-443.
Wang L.S., Li T.B., Xu J. et al. (1999). "Study on rockburst and its intensity classifies
in the tunnel of Erlang Mountain road." Road, Vol. 2: 41-45.
Zhang J.J., FU B.J. (2008). "Rock and its criteria and control." Chinese Journal of
Rock Mechanics and Engineering, Vol.27: 2034-2042 (in Chinese).
Zhang CQ, Feng XT, Zhou H, Qiu SL, Wu W.P. (2012) "Case histories of four
extremely intense rockbursts in deep tunnels." Rock Mech Rock Eng. Vol.45:
275288.
ASCE
225
Numerical Study on Tunnel Instability in a Fault Zone: A Case Study of

the Wuzhuling Tunnel in China
Yingchao Wang1; Hongwen Jing2; Haijian Su3; and Jiangyue Xie4
1
Associate Professor, State Key Laboratory for Geomechanics and Deep Underground Engineering,
China Univ. of Mining and Technology, Xuzhou, Jiangsu 221116, China. E-mail:
wangyingchao@cumt.edu.cn
2
Professor, School of Mechanics and Civil Engineering, China Univ. of Mining and Technology,
Xuzhou, Jiangsu 221116, China. E-mail: hwjing@cumt.edu.cn
3
Assistant Researcher, State Key Laboratory for Geomechanics and Deep Underground Engineering,
China Univ. of Mining and Technology, Xuzhou, Jiangsu 221116, China. E-mail: 631441969@qq.com
4
Graduate Student, School of Mechanics and Civil Engineering, China Univ. of Mining and
Technology, Xuzhou, Jiangsu 221116, China. E-mail: 1414396349@qq.com
Abstract: The impact of fault fracture zone on the stability of surrounding rock in
tunnels is a key scientific topics of geologic engineering. In this study, a threedimensional finite element model was established for Wuzhuling tunnel of Zhuyong
highway in Zhejiang Province, China. By simulating the dynamic process of tunnel
excavation through the fault fracture zone, the deforming performance and stress
distribution law of surrounding rock were studied. It indicates that fault fracture zone
in the tunnel reduced the stability of surrounding rock. When constructing tunnels in a
fault fracture zone, three-dimensional numerical analysis can be performed to find the
areas where collapse occurs easily. The study could provide useful information on the
potential collapse areas for strengthening the engineering supporting structure and
preventing the relative disaster.
Keywords: Tunnel; Fault fracture zone; Surrounding rock; Collapse; 3D numerical
simulation.
INTRODUCTION
Sometimes, mountain tunnels have to go through fault fracture zones to meet the
requirements of proper linetype. When rock mass is broken and loose, it has poor selfstability in fault fracture zones, where collapse occurs easily, causing heavy economic
losses, injuries and deaths. Above all, the key problems in tunnel engineering is that
fault fracture zones affect the stability of surrounding rock.
Many scholars research the impacts of faults on tunnels and have made gratifying
achievements in recent years. K. Y. Geun and H. B. Hyun (2011) studied a important
ASCE
226
case for the tunnel collapse and reinforcement during the construction for the
waterway tunnel at large thrust fault zone in schist. K. Zhao and M. Janutolob (2014)
presented a model of the fault zone and of the TBM excavation process. S. Dalgc
(2003) proposed failure mechanisms and other phenomenon encountered during the
tunnel excavation. In addition, dyke rocks and karstic systems collapsed by faulting
caused some problems during the tunnel excavation. P. K. Shrestha (2014) studied the
groundwater effect on faulted rock mass. Mollon (2010) proposed a translational
three-dimensional multi-block failure mechanism in order to determine the face
collapse pressure of a circular tunnel driven by a pressurized shield. Fraldi (2009)
introduced an exact solution for the plasticity of the evaluation on collapse
mechanisms in rectangular cavities and some numerical examples. He also reported a
comparison with numerical procedures to predict plastic collapse in circular rock
tunnels based on a straightforward analytical approach. However, the reliable
prediction on the collapse state in tunnels and natural cavities is still one of the most
difficult tasks in rock engineering. In aspects of numerical simulation, researchers
have simulated characteristics of tunnel lining force and deformation and supporting
effect of spray anchor affected by faults. By using Plane Finite Element, S. W. Huang
(2006) simulated the response of surrounding rock construction mechanical in
different fault location. Lee and Osman (2006) evaluated tunnel stability and
investigated the so-called arching effect both for soft soils.
From previous studies, it can be seen that methods of numerical simulation or
numerical test is the most important methods that can be used to research the effect of
fault on tunnel. Three-dimensional finite element models was established in this article
to stimulate the dynamic process of tunnel excavation, analyzing characteristics of
surrounding rock force and deformation when tunnel passes through fault fracture
zone, to reveal the mechanism that expresses how fault fracture zone effects
surrounding rock stability. The research achievement can be used to provide necessary
reference and guidance to rock mass stability evaluation and supporting design in fault
fracture zone.
Project Background
Wuzhuling tunnel is located in Zhuyong super highway in China. During left hole
termini, stake is K58+372K58+990, and its length is 618 m. The tunnel has separate
two-lane one-way roads, its clear width is 10.5m, and its clear height is 7.1m. During
right hole termini, stake is K58+355K58+990, and its length is 635 m. The large
span tunnel has three-lanes in it. Its clear width of which is 14.25 m and clear height is
7.69 m. The tunnel was constructed according to NATM principle. It was supported in
composite support structure form. Combined support was applied in the tunnel, which
includes bolt, steel mesh and shotcrete. At the same time, a molded concrete structure
was used as secondary lining. There is a waterproof layer between initial support and
secondary lining.
A fault fracture zone existed where the construction of Wuzhuling tunnel was being
conducted. The fault fracture zone was 400m long, 15m wide, and intersected with the
tunnel axis at the angle 75. Rock mass in fault fracture zone was cracked, brecciated
and earthy. Surrounding rock fissure of both sides is extremely obvious. Affected by
ASCE
227
228
fault fracture zone (F8), when the tunnel was constructed to the point of K58+713.5, a
large-scale roof collapse occurred on Oct. 29, 2005. According to engineering
geological survey report, eluvial gravel was distributed upon the tunnel and strongly
weathered tuff was below the tunnel between section K58+738 and K58+713. As time
went on, the collapse scope expanded gradually. Finally, Accumulation range of
collapse body existed from K58+733 to K58+713.5. Original ground of the section
also appeared to collapse and formed a collapse cavity that was 10.3m long, 6.5m
wide, 15m deep. Cross-sectional view at collapse and its scene photos are shown in
Figure 1.
(a) vertical sectional profile
(b) scene photo
FIG. 1. The collapse of Wuzhuling tunnel

Three-dimensional finite element model
In order to research the law of surrounding rock stress and deformation when
tunnels go through the fault fracture zone, Wu Zhu Ling tunnel was taken as an
example and a three-dimensional numerical model was built, as shown in Figure 2.
We have chosen a small portion of the whole tunnel to use as the model. The area we
chose was about 100 meters long in the direction of the Z-axis, 200 meters long in the
direction of the X-axis, and 100 meters long in the direction of the Y-axis, as shown in
Figure 2. Surrounding rock and fault fracture was simulated using 8-node hexahedral
isoparametric element. Shell element was used to simulate the support structure. The
boundary conditions of the numerical model was setted as upper boundary with no
constraints, the rest of the sides and bottom boundary were constrained in normal
direction. The tunnel has two lanes in the left and three in the right. Main mesh
generation of left and right holes are shown in Figure 3.
According to the engineering geological survey report of Wuzhuling tunnel and
highway tunnel design specifications (JTG D70-2004), physical and mechanical
parameters of surrounding rock and supporting structure were obtained, as shown in
Table 1.
ASCE
229
FIG. 2. The finite element model
FIG. 3. Tunnel
Table 1. Physico-mechanical parameters of surrounding rock and lining

Structure
Category
Surrounding
Rock
Fault Fracture
Zone
Supporting
Structure
Elastic
Modulus
GPa
Poisson's
Ratio
Severe
kN/m3
Cohesion
/ MPa
Friction
Angle
/
10.0
0.25
26.0
1.0
50
0.30
0.40
17.0
0.07
30
30.0
0.20
27.0
The tunnel was excavated by 5-meter construction segments, which in total is 20

construction segments. We selected four sections'''' in the
fault fracture zone, and selected the dome of each section as a sampling point to do
analysis, as shown in Figure 4.
100 m
15 m
Tunnel
fault fracture zone
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
100 m
' '''
D
B
'cross section
FIG. 4. A side view of the calculation model
ASCE
230
Analysis of calculated results

Regular pattern of displacement variation
(1) Regular pattern vertical displacement variation in key points.
The curve of vertical displacement in tunnel vault and bottom, section'',
changed with the construction step, as shown in Figure 5. As can be seen from the
figure, most vault displacement in each section increases first and tends to stabilize
later. When the tunnel face passes through, displacement mutates. In each section,
vault displacement in the right tunnel is greater than that in the left. The peak value of
vault displacement is near the center of fault fracture zone, as shown in section '.
Analyzing the changing trend of bottom displacement, it is easy to see that the "floor
heave" phenomenon appeared after tunnel excavation; the displacement value
decreases first and tends to stabilize later. Displacement in bottom of tunnel is obvious
with construction step before the tunnel face goes through, which is different from
displacement variation in vault of tunnel.
(a) Section
(b) Section
(c) Section
(d) Section
Fig.5 The change of vertical displacement of cross-sections with the construction
step
(2) Displacement field characteristics.
Figure 6 is the vertical displacement contour plot after the sixth construction step
has finished. As can be seen from the figure, a "global slip" phenomenon appears in
the displacement field due to the fault fracture zone(Figure 6.b). There is a big
ASCE
difference between the front and the rear side of the fault fracture zone due to the
influence of the fault. Due to tunnel excavation, larger displacement appears in upper
chambers of excavated section, and vault of the right tunnel has a maximum
displacement.
(a) Three-dimensional map of vertical displacement (Unit: m) (b) Side view of

vertical displacement (Unit: m)
Fig. 6 Cloud chart of vertical displacement after tunnel excavation
The change law of stress
(1) Stress field characteristics.
Calculation results show that after the tunnel excavation, there is maximum
principal stress (1) concentration in the vault of the left and right tunnels in the fault
fracture zone. As shown in Figure 7.a, tensile stress occurs in the vault of the right
tunnel, and it becomes even greater closer to the interior of the fault fracture zone.
Because the rock and soil mass in the fault fracture zone has low tensile strength,
collapses easily happen under the greater tensile stress in the construction of the right
tunnel. Minimum principal stress increases along the sidewall of the tunnel but
decreases in the vault, as shown in Figure 7.b.
(2) Analysis of principal stress of key points.
Figures 8 shows changes in vault principal stress at each construction step in the
cross-section ' of left and right tunnels. As can be seen from the figures, principal
stress at each key point fluctuates when the tunnel goes through the fault fracture zone.
At cross- section', stress in the tunnel vault caused by excavation starts to
fluctuate from about 10m before the tunnel face reaches the key stress point to 10m
behind the tunnel face. Because the tensile strength of the rock mass is low in the fault
fracture zone, the rock mass can easily collapse after undergoing such large stress
fluctuations.
ASCE
231
(a) cloud chart of stress 1
232
(b) cloud chart of stress 3
Fig.7 Cloud chart of principal stress in fault fracture zone after tunnel
excavation
(a) The left tunnel vault cross-section
(b) The right tunnel vault cross-section
Fig.8 The change of vault principal stress of cross-section with the

construction step
CONCLUSIONS
In summary, it can be seen that the fault fracture zone reduces the stability of the
surrounding rock. The displacement of each cross-section increases first and stabilizes
later during tunnel excavation. Displacement fluctuates as the tunnel face goes
through. In the fault fracture zone, the vertical displacement field exhibits a "global
slip" phenomenon. Due to the slanted formation of the fault fracture zone,
displacement in front of the fault fracture zone is greater than that in the rear. After
excavation of the right Wuzhuling tunnel, tensile stress occurs in the vault, which is
greater than the tensile strength of the rock mass in the fault fracture zone, and finally
leads to collapse. When constructing tunnels in a fault fracture zone, threedimensional numerical analysis can be performed to find the areas where collapse
occurs easily. Then preventive measures can be taken to strengthen support earlier .
ASCE
ACKNOWLEDGMENTS
Financial supports from the NationalNatural Science Foundation of China (No.
51309222, 41572263), the Natural Science Foundation of Jiangsu Province (No.
BK20130193), the Specialized Research Fund for the Doctoral Program of
HigherEducation (No. 20130095120016) and the Fundamental Research Funds for the
Central Universities (No. 2014QNB37) are sincerely acknowledged.
REFERENCES
Dalgc, S. (2003) "Tunneling in fault zones, Tuzla tunnel, Turkey." Tunnel. Underg.
Space Technol., Vol. (18): 453-465.
Fraldi, M. and Guarracino, F. (2009) "Limit analysis of collapse mechanisms in
cavities and tunnels according to the HoekBrown failure criterion." Int. J. Rock
Mech. Min. Sci., Vol. (46): 665-673.
Fraldi, M. and Guarracino, F. (2010) "Analytical solutions for collapse mechanisms in
tunnels with arbitrary cross sections." Int. J. Solids Struct., Vol. (47): 216-223.
Fraldi, M. and Guarracino, F. (2011) "Evaluation of impending collapse in circular
tunnels by analytical and numerical approaches." Tunnel. Underg. Space Technol.,
Vol. (26): 507-516.
Geun, K. Y. and Hyun, H. B. (2011). "A Case Study of Collapse and Reinforcement
for Large Span Waterway Tunnel at Thrust Fault Zone." Tunnel. Underg. Space
Technol., Vol. 9 (4): 251-263.
Huang, F. C and Qin, B. (2013) "Determination of minimum cover depth for shallow
tunnel subjected to water pressure.", J. Cent. South Univ., Vol. (20): 2307-2313.
Hang, S. W., SI, T. H., and CHEN, W. S. (2006) Finite element analyses of influence
of fault on large-span tunnel surrounding rock stress., Chin. J. Rock Mech. Rock
Eng., Vol. 25(S2): 2845-2852.
Lee, C.J. (2006) "Tunnel stability and arching effects during tunnelling in soft clayey
soil.", Tunn. Undergr. Sp. Tech., Vol. 21(2): 119-132.
Mollon,G. and Dias, D. (2010) "Face stability analysis of circular tunnels driven by a
pressurized shield." J. Geotech. Geoenviron. Eng., Vol. 136 (1): 215-229.
Shrestha, P. K. and Panthi, K. K. (2014) "Groundwater Effect on Faulted Rock Mass:
An Evaluation of Modi Khola Pressure Tunnel in the Nepal Himalaya." Rock
Mech. Rock Eng.., Vol. (47): 1021-1035.
Wand, Y. C., Shang, Y. Q., and Jing, H. W. (2011) "Optimization of construction
scheme of tunnel collapse and treatment effect". Rock Soil Mech., Vol. 32(S2):
514-519.
Xiong, W. and Fang, W. (2010) "Numerical analysis of effect of normal fault activity
on road mountain runnel project.", Chin. J. Rock Mech. Rock Eng., Vol. 29(S1):
2845-2852.
Zhao K. and Janutolob, M. (2014) "3D simulation of TBM excavation in brittle rock
associated with fault zones: The Brenner Exploratory Tunnel case." Eng. Geol.,
Vol. (181): 93-111.
ASCE
233
The Effect of Foundation Pit Excavations on Adjacent Open-Cut Tunnels for

High-Speed Railways
Zhong Zhou1; Xiang-Can Wang2; Hao Yang3; and Qi-Fang Zhang4
1
Associate Professor, School of Civil Engineering, Central South Univ., Changsha 410075, China. E-mail:
369144091@qq.com
2
Master Student, School of Civil Engineering, Central South Univ., Changsha 410075, China. E-mail:
979135676@qq.com
3
1204952285@qq.com
4
916472775@qq.com
Abstract: Based on the project of deep foundation excavation of No. 15 Tianjian in

Tianjin City, this paper used the finite element software to simulate the entire
construction procedure from foundation excavation to applying the load of upper
structure, and analyzing the effects of every step on open-cut tunnels of high-speed
railway. The results show: the effects on tunnel of horizontal displacement are much
greater than the vertical displacement during the deep foundation excavation; with the
depth of excavation increasing, the horizontal displacement of the tunnel is much greater,
and the tunnel will be leaned; however, after applying load of upper structure, the effects
tend to be smaller; as a reinforcement measure, uplift pile can effectively limit the uplift
of bottom soil, on the other hand it can reduce the effects on the tunnel during the
foundation excavation.
INTRODUCTION
With the development of urbanization in China, the deep excavation projects due to
high-rise buildings and underground stations are increasing. The construction process of
the pits excavation unavoidably has impacts to various degrees on the surrounding
existing structures, especially on the adjacent operational tunnel. It has been an
important research topic which is significant to solve the contradiction between urban
development and urban tunnel protection.
In recent years, some scholars have led research on such issues, and have achieved
some results: Huangs paper discussed the influence of pits excavation on tunnel about
following aspects including the relative position between tunnel and excavation, tunnel
ASCE
234
diameter, pit size and pit protection measures (Huang et al. 2013). The displacement of
the soil is counteracted to some degrees by the tunnel owing to the stiffness of the tunnel
itself, and this counteraction further caused greater tunnels deformation (Jiang and Huo
2002). Someone put forward the two-stage method for determining the longitudinal
deformation of metro tunnels caused by adjacent foundation pits excavation. It may
provide certain theoretical basis to come up with correctly protective measures for metro
tunnels influenced by adjacent excavation (Zhang et al.2011). Combining the Longpan
Tunnels excavation which is right beyond two existing shield tunnels of Nanjing No.1
subway, the paper found that the tunnels upwarping deformations were too great
without stabilizing and could be safe with stabilizing (Yu et al. 2007).
The above researches mostly analyze the impacts of pits excavation on the subway
tunnel, and the study about open-cut tunnels of high-speed railway is few. The
deformation of high-speed railway tunnel is more strict. Based on Tianjian 15 #
excavation in Tianjin, the article used the finite element software to simulate the entire
process, and analyzed the effects of every step on open-cut tunnels of high-speed railway.
Furthermore, the article analyzed horizontal and vertical displacement trends of open-cut
tunnel structure by simulating different reinforcement measures and thereby may
provide certain basis for construction of pits excavation around adjacent open-cut
tunnels of high-speed railway.
Engineering Situation
There are two foundation pits. The shallow pits depth is 4.9m, and the deep one is
8.6m. The foundation pit construction is excavated open and orderly sequenced, and the
construction of shallow pit is prior to the deep one. Bored piles are used as enclosure
structure at the boundary between shallow pit and deep pit. The east wall of the
foundation pit is approximately parallel to adjacent open-cut tunnels. The distance range
between them is about 20-27m, so the foundation pit is in the protected area of
high-speed railway tunnel.
Soil in construction area is divided from top to bottom into following layers:
miscellaneous fill, the thickness is 1.0-4.9m; plain fill, the thickness is 2.6-5.8m;
silty clay, the thickness is 0.9-2.1m, mucky clay, the thickness is 1.9-3.6m, silty
clay, the thickness is 6.9-12.5m, silt and sand, the thickness is 10.6-17.5m; silty
clay, the thickness is more than 20m. The physical and mechanical parameters of soil
layer are shown in Table 1.
ASCE
235
236
Table 1. The Physical and Mechanical Parameters of Soil Layer

Soil
Miscellaneou
s Fill
Plain Fill
Silty Clay
Mucky Clay
Silty Clay
Silt and Sand
Silty Clay
Bulk
Density
(KN/m3)
Cohesion c
(kPa)
Internal
Friction
()
Deformation
Modulus E
(MPa)
Permeability
Coefficient K
(cm/s)
17
16
14.4
1.010-6
15.3
18.9
17.8
19.8
19.8
19.9
11.5
13.2
8.4
14.5
8.2
18.4
18
15
10
18
33
21
20.3
23.0
20.2
30.3
60.0
35.3
3.510-6
4.110-5
2.010-7
4.710-5
7.010-5
3.510-6
The Establishment of Three-dimensional Numerical Model

Assumptions and Size of Model
For the convenience of analysis, there are some assumptions: all materials are
isotropic; soils are analyzed as elastic-plastic material and concrete structures are
analyzed as elastic material; simulating the entire construction process step by step
and considering changes of the spatial displacement and excluding time effect.
To reduce the influence of the boundary constraint on the calculation results, the
distance between model boundary and enclosure structure is more than 3H (H is the
structural buried depth). Considering the depth of foundation pit and buried depth of
adjacent tunnel, model width is 350m, length is 400m and depth is 60m. Figure 1 is the
model of foundation pit.
FIG. 1. Three-dimensional model.
FIG. 2. Model of enclosure structure and

tunnel.
Mohr-Coulomb elastoplastic model was used as soil constitutive model in the

simulation. The enclosure structure was simulated by plate element and the support
structure was simulated by beam element, whose constitutive models were all linear
ASCE
237
elastic model. Horizontal displacement was restricted around the model as well as the
vertical and horizontal displacement of model bottom.
Numerical Simulation
The deformation requirements of high-speed railway tunnel are very strict.
Considering the safety requirements and deformation of additional longitudinal
curvature about tunnel, the maximum vertical displacement of structure must not exceed
3mm, and horizontal displacement is no more than 3mm. During the excavation, in order
to ensure the safety operations of high-speed railway, the excavation of foundation pit is
subdivided into the following steps and effects on adjacent open-cut tunnel are analyzed
one by one.
Table 2. Steps of Simulating Foundation Pit Excavation
Step
S0
S1
S2
S3
S4
S5
S6
S7
Content
Forming initial stress field, resetting displacement;
Activating shield excavated well, open-cut tunnel, shield tunnel, enclosure
structure and so on, resetting displacement;
Excavation of shallow foundation pit;
Constructing floor of shallow foundation pit;
Excavation of deep foundation pit;
Constructing floor of deep foundation pit;
Constructing basement structure;
Constructing upper building
Numerical Results Analysis

Displacement Analysis about Surrounding Soil of Foundation Pit
In the process of excavation of foundation pit, there is the maximum upwarping
deformation which is 37.6mm at the bottom of shallow pit, the maximum upwarping
deformation of deep pit is 41.4mm, and the maximum horizontal displacement of
surrounding soil is 17.0mm. After the removal of soil inside the foundation pit, stress
equilibrium of surrounding soil has been destroyed, which means that loads on the free
surface of foundation pit are removed, and soil deformation occurs in accord with the
direction of reaction force. So, soils at the bottom of foundation pit uplift, and
surrounding soils squeeze out from the direction toward foundation pit.
Horizontal Displacement Analysis about Open-cut Tunnel Structure
The east wall of the foundation pit is approximately parallel to adjacent open-cut
tunnels. The distance range between them is about 20-27m. By numerical simulated
calculation, we got structure displacement of every step which was shown in Table 3.
ASCE
238
Table 3. Displacement of open-cut tunnel structure

Step
S2
S3
S4
S5
S6
S7
Z-Displacement
0
0
0
0
0
-0.1
0.2
-0.2
0.2
-0.2
0
-0
X-Displacement
0
-0.1
-0.7
-1.0
-1.4
-4.0
-3.4
-6.1
-2.9
-5.0
-1.8
-2.3
Y-Displacement
0
0
0
0
0
-0.6
0
-0.8
0
-0.6
0
-0.1
Figure 3 is the cloud picture of horizontal displacement about open-cut tunnel

structure and Figure 4 is the curve. The direction of tunnel displacement is overall
secund toward the foundation pit and the maximum of displacement is 6.1mm, which
located at the bottom of junction between open-cut tunnel and shield excavated well.
Since the open-cut tunnel structure is far away from the shallow pit, excavation of
shallow pit has little effect on the displacement of tunnel at the stage S2 and S3 and the
maximum of displacement is 1.0mm. However, the displacement of tunnel increases
sharply at the stage S4 and S5, and the displacement of S5 increases to maximum 6.1mm.
Then, with the superstructure load being applied, tunnels horizontal displacement is
gradually reduced, the final maximum of displacement is 2.3mm. Overall, the horizontal
displacement of tunnel structure increases with the increasing of excavation depth, and
closer to foundation pit structure is, larger the displacement is. In the same section, the
upper displacement is different from the lower one, which shows that tunnel structure
becomes leaned in the space.
FIG. 3. The picture of horizontal displacement about open-cut tunnel structure.
ASCE
FIG. 4. The curve of horizontal displacement about open-cut tunnel structure.

Vertical Displacement Analysis about Open-cut Tunnel Structure
Figure 5 is the curve of vertical displacement about open-cut tunnel structure. The
maximum of displacement is 0.21mm at the stage S5, which located at the junction
between open-cut tunnel and shield excavated well. The structure has upward
displacement on the side that is closer to foundation pit, but the displacement is
downward on the other side, so the tunnel rotates. On the whole, the vertical
displacement of open-cut tunnel structure is small, and it increases with the increasing of
excavation depth.
FIG. 5. The curve of vertical displacement about open-cut tunnel structure.

After the removal of soil inside the foundation pit, stress equilibrium of surrounding
soil has been destroyed, so soil deformation occurs in accord with the direction of
reaction force which drives the movement of tunnel. The tunnel is located at the right of
foundation pit, therefore, the main direction of tunnels movement is horizontal. The
effect of foundation pits excavation on the tunnel structure horizontal displacement is
significantly greater than the vertical displacement.
Comparison and Analysis The Results of Reinforcement Scheme
In the process of foundation pits excavation, cumulative displacement of each step
in the tunnel cannot exceed the control value of displacement, but from the result of the
ASCE
239
above analysis, the horizontal displacement of the tunnel structure in step S4, S5 and S6
are beyond the control value. It needs to reinforce the foundation pit, so that the
displacement could be controlled within the safety value. There are two reinforcement
schemes:
Scheme 1: doing uplift piles of 1000mm, length 35m in the bottom of the
foundation pit, and closer to tunnel the place is, the denser uplift piles we make;
Scheme 2: on the basis of scheme 1, replacing the plain concrete with reinforced
concrete of 200mm cushion at the bottom of foundation pit.
Figure 6 is the cloud picture of horizontal displacement about open-cut tunnel
structure. The horizontal displacement of each section is markedly lower than the one of
original scheme, but there is no change about displacement law. From Figure 12, the
maximal horizontal displacement reduced from 6.1mm to 2.5 mm. Comparing with the
original scheme, horizontal displacement of every step decreased a lot after being
reinforced. Due to constraint function of uplift pile, the range of stress redistribution of
surrounding soils narrowed, and the plastic zone of soil narrowed. So horizontal
displacement of tunnel structure decreased at the same point.
FIG. 6. The picture of horizontal displacement about open-cut tunnel structure.

On the basis of scheme 1, scheme 2 replaced the plain concrete with reinforced
concrete at the bottom of foundation pit. From Figure 7, tunnel structure displacement
further decreased comparing with scheme 1. The uplift displacement of soil was 26mm,
and the maximal horizontal displacement of tunnel structure was 2.4mm, which are all
within the safe value. Reinforcement measures can satisfy the requirement of foundation
pit construction safety and effectively reduce the effect of foundation pits construction
on open-cut tunnel of high-speed railway.
ASCE
240
FIG. 7. The curve of horizontal displacement about open-cut tunnel structure.

CONCLUSIONS
Through analyzing the effects of deep foundation pit excavation on adjacent
open-cut tunnel of high-speed railway, we get the following conclusions:
1. With the increasing of excavation depth, the deformation of surrounding soil
increased, the upwarping displacement increased at the bottom of foundation pit and the
influence on displacement of adjacent tunnel increased. With the superstructure load
being applied, the displacement of soil and adjacent tunnel decreased.
2. With the excavation of foundation pit, the direction of tunnel displacement is
overall second toward the foundation pit. The main effect of deep foundation pit
excavation on adjacent open-cut tunnel of high-speed railway is the horizontal
displacement of tunnel, and there is little effect on vertical displacement. In the vertical
direction of sections, closer to foundation pit structure is, larger the horizontal
displacement is, so tunnel structure becomes leaned in the space.
3. The construction of uplift piles effectively reinforced surrounding soils and
reduced the effect of foundation pits construction on open-cut tunnel of high-speed
railway which showed the great role of uplift piles.
4. For safe construction of pits excavation around adjacent open-cut tunnels of
high-speed railway, it must take some reinforcement measures to reduce the effects.
Cumulative displacement of every step in the tunnel cannot exceed the control value of
displacement.
ACKNOWLEDGMENTS
This research was funded by the National Natural Science Foundation of China
(Grant No.50908234), and the Major State Basic Research Development Program of
China (973 Program, Grant No.2011CB710604).
ASCE
241
REFERENCES
Huang, X., Schweiger, H.F., Huang, H.W. (2013). Influence of Deep Excavations on
Nearby Existing Tunnels. International Journal of Geomechanics, Vol. 13(2):
170-180.
Huo, Z.L., Yan, S.W., Zhang, Q.B. (2014). Numerical Analysis of Effects of Foundation
Pit Excavation on Metro Tunnels. Tunneling and Underground Construction, Vol.
802-807.
Jiang, H.S., Hou, X.Y. (2002). The Influence of Deep Excavation on Adjacent Metro
Tunnel in Soft Ground. Industrial Construction, Vol. 32(5): 53-56.
Qi, K.J., Wang, X.D., Jiang, G., Chang, Y.S., Chen, Y.D. (2005). Analysis of Deep Pit
Excavation Adjacent to Tunnel. Chinese Journal of Rock Mechanics and
Engineering, Vol. 24(S2): 5485-5489.
Ruan, S.L., Hu, S.B., Lou, Y.L. (2012). Effects of a Foundation Pit on Adjacent
Subway Tunnels and Control Measures. Morden Tunnelling Technology, Vol. 49(1):
100-104.
Yu, J., Xu, Q.H., Xing, W.W., Ding, Y., Cheng, W.Z. (2007). Numerical analysis of
upwarping deformation of existing tunnels under a new tunnels excavation. Rock
and Soil Mechanics, Vol. 28(S): 653-657.
Zhang, Z.G., Zhang, M.X., Wang, W.D. (2011). Two-stage method for analyzing effects
on adjacent metro tunnels due to foundation pit excavation. Rock and Soil
Mechanics, Vol. 32(7): 2085-2092.
ASCE
242
Analysis on the Response of Ground Surface Settlement for a Metro Station during
Construction Using the Pile-Beam-Arch (PBA) Method Subjected to Pore Pressure
Fei Xu1; Li-Ping Li2; Qian-Qing Zhang3; Zhe-Chao Wang4; and Peng He5
1
China. E-mail: 869957556@qq.com
Associate Professor, Research Center of Geotechnical and Structural Engineering, Shandong Univ.,
Jinan, China (corresponding author). E-mail: sduytzx@126.com
Associate Professor, Research Center of Geotechnical and Structural Engineering, Shandong Univ.,
Jinan, China. E-mail: wanz@sdu.edu.cn
Abstract: A simple approach to predict the settlement of ground surface during the
construction of metro tunneling using pile-beam-arch (PBA) method, subjected to pore
pressure is presented. As to the analysis on the response of a single pile installed
beneath the tunnel, a softening nonlinear model was adopted to predict the relationship
between unit skin friction and pile-soil relative displacement developed along the
pile-soil interface and a bilinear base load-displacement model was used to predict the
relationship developed at the pile end. To analyze the settlement of ground surface
during the construction of metro tunneling using PBA method, a highly effective
iterative computer program was developed. In order to demonstrate the effectiveness
and accuracy of the proposed simple method in this paper, the result is compared with
one well-instrumented previous test. The proposed simple analytical approach is
economical and efficient, and suitable for the analysis of the settlement of ground
surface during the construction of metro tunneling using pile-beam-arch (PBA)
method subjected to pore pressure.
ASCE
243
244
INTRODUCTION
Nowadays, the metro with a larger transport capacity and a larger share in
guaranteeing the smoothness of traffic, have been building to ease their heavy traffic
in more and more cities. Ground settlement (surface vertical movement) is a critical
threat to both the surface (Melis et al., 2002) and subsurface facilities (Vorster et al.,
2005), especially in the shallow metro station. How to control the excessive ground
movement and reduce the influence of excavation on the existing nearby facilities is a
difficult question. As a main construction method, widely used in the construction of
metro station, especially the arch cover located in soft soil stratum and weak-fracture
stratum, the pile-beam-arch (PBA) method was introduced in this paper. The main
steps of PBA method are summarized as below: (a) two small cross-section pilot drifts
(left-and right-drift) located on each side of the metro tunnel, are excavated above the
tunnel crown using a short bench-heading method; (b) two rows of bored
reinforced-concrete piles are emplaced to a suitable depth below the tunnel floor.
Based on the mechanism of load transfer of PBA method, there are two aspects to be
figured out: one is the vehicle load acted on the arch cover subjected to pore pressure;
and the other is the displacement of pile induced by overlaying load.
LOAD ACTED ON THE ARCH COVER SUBJECTED TO PORE PRESSURE
Sine Hoek-Brown failure criterion was proposed by Hoek and Brown (1980), this
nonlinear failure criterion had been widely applied in a variety of geotechnical
engineering (Fraldi and Guarracino 2010; Huang et al. 2013). For deep tunnels, Fraldi
and Guarracino (2010) constructed a curved kinematically admissible failure
mechanism and introduced Hoek-Brown failure criterion to calculate the energy
dissipation of the failure surface. Huang et al. (2013) developed this solution
considering the effect of pore water pressure for the shallow cavities.
sym
L1
L2
Hock-Brown
rock mass
f (x)
n
Failure block
ru
n
Arch cover
c(x)
Pile
FIG. 1.
ASCE
Curved failure mechanism of shallow metro station constructed

with PBA method.
245
As described by Huang et al. (2013), considering the effect of pore pressure, and
using a new curved failure mechanism within the framework of upper bound theorem
and nonlinear Hoek-Brown failure criterion, the shape of failure surface of collapse
block can be drawn by the following form (Figure 1):
1 B
(1 + ru ) B
c1 B
f ( x) = A
(1)
x
+ c2
c (1 + ru )
The weight of the collapse block P can also be obtained by integral calculation:
L2
L2
P = c( x)dx- f ( x)dx
(2)
L1
0
1
B
where is the unit weight of the rock mass, c(x) is the equation describing the circular
tunnel profile which can be calculated by:
c( x) = H + r r 2 x 2
(3)
where H and r are the buried depth and radius of the circular tunnel, respectively.
As shown in Figure 1, the value of L1 and L2 can be obtained from the following
equations:
f(x=L1)=0
(4)
(5)
f(x=L2)=c(x=L2)
Substitute equation (4) into Eq. (1) and the values of constants c1 and c2 are
determined.
c1=L1
(6)
c2=0
(7)
Furthermore, one equation can be derived by equating external rate of work and
internal energy dissipation:
B 1
t ( L2 L1 ) + A B c B [ (1 + ru )] B
(1 + ru ) L2 ( H + r ) +
(1 + ru ) L2
2
L1 1+BB
L 1+ B
1
) ( L1 1 ) B
( L2
1 + ru
1 + ru
B +1
r 2 L22 +
(1 + ru )r 2
2
(8)
L
L
arcsin 2 + rq arcsin 2 + s L1 = 0
r
r
where n is the normal stress, n is the shear stress, A and B are the material constants,
c is the uniaxial compressive strength of the soil, ru is the pore pressure coefficient,
and t is the tensile strength of the soil (t>0).
Combining the Eqs. (5) and (8), and using the numerical software, the values of L1
and L2 can be given.
SOFTENING NONLINEAR MODEL OF SKIN RESISTANCE
As reported by Zhang et al. (2015), the relationship between skin friction and the
corresponding shear displacement follows a softening model when skin friction is
fully mobilized. Thus, in this work, a simple softening model, may be conveniently
adopted to describe the shaft shear stress and the shear displacement developed along
the pile-soil interface.
ASCE
246
This softening behavior between the skin friction and the pile-soil relative
displacement at a given depth z has been described by the following equation:
s ( z) =
Ss ( z )[a + cSs ( z )]
[a + bSs ( z )]2
(9)
where s(z) is the shaft shear stress at a given depth, z; Ss(z) is the relative
displacement along the pile-soil interface at a given depth, z. The parameters a, b, and
c are empirical coefficients, and be obtained from Zhang et al. (2015).
A series of field tests on bored piles under compression loading (Zhang et al. 2010)
demonstrated that values of s applied for the bored piles have been found to be in the
range 0.83-0.97. The value of Ssu is determined experimentally or by back-analysis of
field load test results. A series of field tests (Zhang et al. 2010) show that the values of
parameter Ssu have been found to be in the range 5-25 mm for bored piles of 0.7-1.1 m
diameter in different soils (e.g., mud, clay, sandy silt, silty clay).
Based on the results of laboratory and in situ tests, the ultimate unit shaft resistance,
su, can be determined using a formula corresponding to soil parameters. The effective
stress method is employed to predict su in the drained condition. Based on Coulombs
friction law, the following equation can be used to calculate the limiting unit shaft
resistance, su:
su = K v tan
(10)
where K is the lateral earth pressure coefficient; v is the effective overburden

pressure at the depth under consideration; and is the friction angle of the pilesoil
interface. For bored piles, the angle is commonly assumed to be equal to the angle of
shearing resistance of the surrounding soil, , for practical purposes. The coefficient
K depends on various factors including the soil state, the pile installation method, and
the pile geometry, and is related to the in situ earth pressure coefficient, K0, whose
value is approximately estimated by K0=1-sin for normally consolidated soils.
Suggested values of lateral earth pressure coefficient K and friction angle of the
pile-soil interface are given by Zhang et al. (2010; 2015).Therefore, Eq. (10) can be
written in another form (Yang et al. 2006):

K
tan v
K0

su = K 0
(11)
A BILINEAR BASE LOAD-DISPLACEMENT MODEL
Assuming that the load-displacement relationship developed at the pile base follows
a bilinear hardening model, the pile base settlement induced by the mobilized base
load can be written as the following:
(SbSbu )
k1 Sb
k1 Sbu + k2 ( Sb Sbu ) (Sb Sbu )
b =
ASCE
(12)
247
where b and Sb are equal to unit end resistance and pile-end settlement, respectively;
k1 and k2 represent the compressive rigidity of the pile-tip soil in the first and second
stage of the load-displacement curve, respectively; and the parameter Sbu is defined as
the pile-end settlement related to the limiting end resistance in the first stage. In the
bilinear hardening model of the soil below the pile base, three parameters (k1, k2 and
Sbu) are required to define the load-displacement response at the pile end, and the
calculation method could be found in Randolph and Wroth (1979) and Zhang et al.
(2010).
ALGORITHM FOR LOAD-SETTLEMNT BEHAVIOR OF A SINGLE PILE
EMBEDDED IN LAYERED SOILS
The theoretical method for a single pile embedded in multilayered soils can be
analyzed with the following procedure:
1. Assume a single pile is divided into m segments.
2. Assume a small pile end settlement Sbm.
3. Calculate the pile base load, Pbm, induced by the assumed pile base displacement
Sbm, using Eq. (12).
4. A vertical movement, Scm, at the middle height of pile segment m is assumed (for
the first trial, assume Scm= Sbm).
5. From the load-transfer function as given in Eq. (9), the unit shaft resistance of pile
segment m, sm, is obtained based on the assumed value of Scm.
6. The load at the top of pile segment m, Ptm, can then be calculated from the
following:
(13)
Ptm=Pbm+ dLmsm
where Lm= length of pile segment m; and d = pile diameter.
7. Elastic deformation at the midpoint of pile segment m (assuming a linear variation
of load in the segment), Sem, is calculated from the following:
Sem =
( Ptm + 3Pbm ) (
4
Lm
)
2 Ap E p
(14)
, is then given by the

8. The updated midpoint displacement of segment m, Scm
following:
Scm = Sbm + Sem
(15)
with the assumed value of Scm

9. Compare the updated midpoint displacement Scm
does not agree with Scm within a
from step 4. If the computed displacement Scm
as the new value of Scm. Repeat steps 4-9
specified tolerance, e.g., 110-6m, use Scm
) is within the assumed tolerance.
until the value of (Scm- Scm
ASCE
248
10. Calculate the load and displacement at the top of pile segment m, Ptm and Stm,
respectively, using the following equations:
Ptm = Pbm + dLm sm
(16)
S tm = Sbm + Scm
(17)
where sm is derived from Eq. (9) and an updated midpoint displacement Scm
.
11. Repeat steps 4-10 from pile segment m to pile segment 1 until the load-settlement
relationship developed at the pile head is obtained.
12. The procedure from steps 2-11 is then repeated using a different assumed pile end
settlement Sbm until a series of load-displacement values are obtained. The proposed
simple analytical approach is economical and efficient, and suitable for analysis of a
single pile embedded in layered soils.
CASE STUDY AND PARAMETRIC ANALYSIS
To analyze the influence of the parameters related to the proposed method on the
settlement of ground surface during the construction of metro tunneling using
pile-beam-arch (PBA) method subjected to pore pressure, a further study was
conducted.
Arch cover
8.0
FIG. 2.
17.0
2.5
Third transverse support
3.5
First transverse support

Second transverse support
Floor
Pile
Schematic diagram of cross-section and transverse supports of the main

chamber (units: m).
Analyses were conducted on a classical case history analyzed, regarding ground

surface settlement during the Beijing Guanghua Road metro station construction using
the PBA method, reported by Wang (2007). The overlying stratum is the mixture of
miscellaneous fill and gravel. As reported by Fraldi and Guarracino (2010), the
ASCE
material constants A and B can be taken as 2/3, 0.7, respectively. As illustrated by

Wang (2007), the minimum buried depth H of this metro station was taken as 7.4m,
and the radius b of the arch cover was adopted as 8.56m. The angle of shearing
resistance of the soil, , was taken as 30o. A value of 2.0 MPa was reported for the
shear modulus of the soil and the elastic modulus for the reinforced concrete pile was
taken as 30 GPa. The uniaxial compressive strength c , the tensile strength t ,
Poissons ratio b and unit weight can be adopted as 2.5MPa, 24 kPa, 0.32 and 18.3
KN/m3.The other parameters can be in founded in Figurer 2.
The 17-m-long concrete pile with a diameter r0 of 0.8 m employed was embedded
primarily in a layer of clayey silt to silty clay. The single pile is divided into 17
segments with each pile segment of 1.0 m in length, except of two pile segments near
the pile top where the pile segment lengths are assumed to be 1.2 m and 0.8 m,
respectively.
The value of / is assumed to be 0.9, and the value of K/K0 is adopted as 2.0. The
value of the parameter s and Ssu are adopted as 0.90 and 5.6 mm (0.7%d), respectively.
The value of pore pressure coefficient ru is assumed to be 0, 0.1, 0.3, respectively.
Based on the construction scheme of the main chamber, the value of effective pile
length is assumed to be 17 m, 14.5 m, 11.5 m, and 8 m, respectively.
Figure 3 indicates that the value of settlement of ground surface increases with
increasing pore pressure coefficient ru. Moreover, it can be conclude that the
calculated results estimated from the present approach is generally in good agreement
with the measured values and calculated values given by Wang (2007).
FIG. 3.
Analysis results of the ground surface settlement for different value of

parameter ru.
CONCLUSIONS
In this paper, a simple approach to predict the settlement of ground surface during
the construction of metro tunneling using pile-beam-arch (PBA) method, subjected to
ASCE
249
pore pressure is presented. In order to demonstrate the effectiveness and accuracy of

the proposed simple method in this paper, the result is compared with one
well-instrumented previous tests. Moreover, the influence of pore pressure to the
settlement of ground surface was analyzed. The value of settlement of ground surface
increases with increasing pore pressure coefficient ru. The proposed simple analytical
approach is economical and efficient, and suitable for the analysis of the settlement of
ground surface during the construction of metro tunneling using pile-beam-arch (PBA)
method subjected to pore pressure.
ACKNOWLEDGMENTS
This work was supported by the State Key Development Program for Basic
Research Program of China (No. 2013CB036000), the National Natural Science
Foundation of China (No.51479106) and the National Natural Science Foundation of
China (No.51408338).The authors would be grateful to the reviewers for their
valuable comments and suggestions that can help improve the quality of the paper.
REFERENCES
Fraldi, M. and Guarracino, F. (2010). Analytical solutions for collapse mechanisms in

tunnels with arbitrary cross sections. International Journal of Solids and Structures,
Vol. 47 (2): 216-223.
Hoek, E. and Brown, E.T., 1980. Empirical strength criterion for rock masses.
Journal of Geotechnical Engineering Division, Vol. 106 (9): 1013-1035.
Huang, F., Qin, C. and Li, S.C. (2013). Determination of minimum cover depth for
shallow tunnel subjected to water pressure. Journal of Central South University,
Vol. 20: 2307-2313.
Randolph, M. F. and Wroth, C.P. (1979). An analysis of the vertical deformation of
pile groups. Geotechnique, Vol. 29 (4): 423-439.
Vorster, T.E., Klar, A., Soga, K., and Mair, R.J. (2005). Estimating the effects of
tunneling on existing pipelines. Journal of Geotechnical and Geoenvironmental
Engineering, Vol. 131 (11), 1399-1410.
Wang, S.J. (2007). Study on ground settlement due to subway construction with
cavern-pile and CRD method. Doctor Degree Thesis, Harbin Institute of
Technology, China. (In Chinese)
Yang, J., Tham, L.G., Lee, P, K.K., Chan, S.T. and Yu, F. (2006). Behavior of jacked
and driven piles in sandy soil. Gotechnique, Vol. 56 (4): 245-259.
Zhang, Q.Q., Li, S.C. and Li, L.P. (2015). Field and theoretical analysis on the
response of destructive pile subjected to tension load. Marine Georesources &
Geotechnology, Vol. 33 (1): 12-22.
Zhang, Q.Q., Zhang, Z.M., Yu, F. and Liu, J.W. (2010). Field performance of long
bored piles within piled rafts. Proceedings of the Institution of Civil Engineers:
Geotechnical Engineering, Vol. 163 (6):293-305.
ASCE
250
251
Tunneling beneath Existing Buildings Supported on Shallow Foundations

C. F. Leung1; J. Yu1; Y. K. Chow1; Y. S. Ng2; H. T. Tan2; and C. J. Hua2
1
Centre for Soft Ground Engineering, National Univ. of Singapore, Singapore 117676. E-mail:
ceelcf@nus.edu.sg
2
Building & Infrastructure Group, Housing and Development Board, Singapore 310480. E-mail:
Hcj1@hdb.gov.sg
Abstract: The traditional modification factor approach employed to evaluate the

building movements due to tunneling is for plane-strain situation. Such approach
could not cater for the effect of building dimension along the tunnel axis. In the
present study, three-dimensional finite element analyses were conducted to determine
the movements of buildings with various plan dimensions and locations induced by
tunneling. A new modification factor is proposed to cater for the true threedimensional nature of the problem. A relationship between the new modification
factor and building bending stiffness is proposed to determine the relative movement
between the shallow foundation and Greenfield ground settlement due to tunneling.
INTRODUCTION
Assessing the influence of tunneling on existing buildings is essential in the design
and construction of tunnels in urban environment. To evaluate the movements of a
shallow foundation due to geotechnical activities, Burland and Worth (1974)
employed terminologies such as settlement, differential settlement and deflection
ratio. In a follow-up study, deflection ratio has been identified as the most critical in
the assessment on the risk of potential building damage due to tunneling.
Potts and Addenbrooke (1997) presented a tunnel-foundation interaction model (Fig.
1(a)) to evaluate the deflection ratio between Greenfield ground movements and
foundation movements. By combining the characteristic of the Greenfield settlement
trough, the analytical model introduced the concept of hogging (DRhog) and sagging
(DRsag) deflection ratios (Fig. 1(b)) as follows:
DRsag =
sag
Lsag
DRhog =
hog
Lhog
(1)
where the foundation length was separated by the inflection of the surface
settlement into sagging length
and hogging length
and
were
the differential settlements in the sagging and hogging zones respectively, as
ASCE
252
illustrated in Fig. 1(b). The building modification factors (MDRsag and MDRhog) were
then presented to denote the maximum deflection ratio of a building due to tunneling
in relation to the Greenfield condition:
M DRsag =
sag Lsag
g
sag
g
sag
M DRhog =
hog Lhog
g
hog
Lghog
(2)
The deflection ratios

and
were calculated in the sagging and
hogging zones corresponding to the building geometry and location using the
Greenfield settlement trough. In their two-dimensional finite element (2D FE) study,
the building was modeled as a weightless elastic beam with appropriate bending and
axial stiffness properties. Its contact with the ground surface was taken as rough. By
changing the building geometry and location in relation to the tunnel, a design chart
2
of modification factors in terms of building relative bending stiffness
is the soil elastic modulus) was proposed
( is the building bending stiffness;
such that the building deflection ratio can be estimated from the Greenfield condition.
(a)
(b)
FIG. 1 Interaction diagram and definition of deflection ratios.
Owing to 2D FE analysis, Potts and Addenbrooke (1997) only took the onedimensional geometrical characteristic of the building length into account. Based on
the modification factor approach, Franzius et al. (2006) subsequently applied 3D FE
method to extend the above model incorporating the building dimension in the
direction of tunnel axis and building weight. New design charts were proposed in
( is the building
terms of building relative bending stiffness redefined as
length in the direction of the tunnel axis; z is the depth of tunnel center).
It can be established from the above mentioned design charts that the eccentricity
ratio is an independent variable. Based on 2D FE results, a new parameter
relative bending stiffness (
and
) was presented by Goh and
Mair (2011) for buildings subjected to excavation-induced settlements.
and
are the sagging and hogging lengths in the Greenfield corresponding to the building
location, respectively. A series of parametric studies involving building length and
location demonstrated that the relationship between the building modification factor
and the relative bending stiffness is reasonably unique.
ASCE
253
Following the approach by Goh and Mair (2011) on foundation movement due to
excavation, the present study investigates the foundation movement due to tunneling.
In this paper, the relative bending stiffness will be further evaluated to study the
effect of raft foundation dimension in the direction of the tunnel axis and the building
weight on the modification factor using 3D FE analysis.
FINITE ELEMENT MODELLING
In the present study, the finite element program Plaxis 3D is employed to analyze
the effects of tunneling beneath a raft foundation. An undrained total stress analysis is
conducted using a linear-elastic perfectly plastic soil model with Tresca parameters of
strength and stiffness for a saturated clay. This is a commonly used geotechnical
analysis owing to the large database and experience in applying the Tresca soil model.
The simulation of the tunnel volume loss (see Fig. 2, the percentage tunnel crosssectional area loss upon tunneling) is modeled using the gap method proposed by Lee
and Rowe (1991). After the geostatic stresses reach equilibrium, the soil elements are
deactivated and the plate elements, representing the tunnel lining, will be activated
along its interface. The function of the contraction of the plate elements is applied to
simulate the volume loss. Since it is difficult to control the volume loss accurately in
a real case, the parameters of the tunnel and the soil in this paper are based on the
centrifuge model tests conducted by Loganathan et al. (2000). The depth of the
tunnel center is 18m, and the tunnel diameter is 6m. The soil submerged unit weight
is 6.5kNm . The undrained shear strength of the saturated soil is 35kPa at the
ground surface increasing linearly to 75kPa at the tunnel center elevation. Following
the recommendation by USACE (1990), a consistent ratio of 400 is adopted for
all the analyses. To simulate undrained condition, the Poissons ratio is set to 0.495.
The coefficient of earth pressure at rest,
= 1, is applied over the entire soil profile.
In the centrifuge test, the volume loss magnitude is 1%. A typical mesh used for the
3D FE analyses is depicted in Fig. 3.
R 2 r02
R2
FIG. 2 Volume loss.

Fig. 4 shows the ground surface settlement trough obtained from the FE analysis
which is in fairly good agreement with that measured from the corresponding
centrifuge test. Since the distance from the inflection to the tunnel centerline is about
10m for the current result, the foundation length is designed to be 20m such that the
building is solely located within the sagging zone without eccentricity.
ASCE
254
FIG. 3 3D Finite element model used in analysis.
Settlement (m)
0.000
-0.005
Inflection
-0.010
-0.015
-0.020
-30
Centrifuge test (Loganathan et al, 2000)

This study
-20
-10
10
20
30
Distance from tunnel centreline (m)
FIG. 4 Ground settlement induced by tunneling.

The foundation has a width in the direction of tunnel axis and an eccentricity
between the foundation and the tunnel center line. The width and the eccentricity
are varied independently to assess the influence on the modification factor. The plate
element of 1-m thick is modeled as the raft foundation and a rough contact is adopted
between the foundation and the soil.
FINITE ELEMENT RESULTS
Effect of Building Width
In this parametric study, the weightless foundation width is chosen to range from
10m to 70m and no eccentricity is included. Considering the boundary condition and
the model size, the foundation width of 70m can be considered as a plane-strain case.
For each foundation width, the modification factors are calculated with the raft
having an elastic modulus varying from 1.2104 N/m2 to 1.21010 N/m2.
Fig. 5(a) shows the comparison of modification factors among four buildings with a
length-width ratio of zero (plane-strain), 0.5, 1 and 2. For the same bending stiffness,
a smaller length-width ratio would result in higher modification factors, implying
more severe differential settlements. The plane-strain condition serves as the upper
ASCE
255
bound for all foundation length-width ratios. In addition, it should be noted that for
the foundation with a low bending stiffness, the maximum settlements are greater
than those from Greenfield. This observation is consistent with that reported by Potts
and Addenbrooke (1997).
By replacing the bending stiffness with the bending stiffness per unit foundation
width , the relationship between
and is illustrated in Fig. 5(b). As
such, the plane-strain condition can serve as the standard case to estimate
of a
3D case based on the concept of the unit bending stiffness as long as the foundation
eccentricity is not taken into account.
20 40 (LB)
20 20 (LB)
20 10 (LB)
Plane strain
20 70 (LB)
0.8
0.6
20 40 (LB)
20 20 (LB)
20 10 (LB)
Plane strain
Strandard
1.0
0.8
0.6
DRsag
1.0
1.2
DRsag
1.2
0.4
0.4
0.2
0.2
0.0 3
4
5
6
7
8
9
10
11
10 10 10 10 10 10 10 10 10
0.0 3
10
EI
10
(a)
FIG. 5 Effect of building width on modification factors.
10
10
EI/B
10
10
10
(b)
Relative Stiffness
Goh and Mair (2011) presented a dimensionless relative bending stiffness to unify
the relationship between the building modification factor and eccentricity for 2D
plane-strain structures. In the present study, the 2D dimensionless relative bending
stiffness is combined with the concept of unit bending stiffness. New parameters of
relative bending stiffness

and
are introduced by
considering the effect of eccentricity and width on the 3D foundation building
modification factors. The representative soil stiffness is defined as the weighted
average of the elastic modulus of the soil above the tunnel center.
Based on the new relative bending stiffness, a parametric study is performed for
eccentricity =5m, 10m, 15m and 20m. The weightless foundation has a lengthwidth ratio of 2 (20m by 10m in length and width). For =5m and 15m ( =
0.25 and 0.75), the inflection point is placed at the quarter length of foundation to the
edge. For = 0.5, the raft foundation is equally divided into the sagging part and
hogging part. The eccentricity of = 1 is a special case for the foundation entirely
located in the hogging zone.
Fig. 6 shows the relationship between modification factors and relative bending
stiffness of the 3D foundation with foundation eccentricity in relation to the tunnel
centerline. From Figs. 6(a) to (c), it can be seen that as bending stiffness increases,
the inflection point in the foundation gradually moves from the inflection location
corresponding to the Greenfield condition to the foundation edge along the direction
of the shorter distance to the edge and eventually disappears. For the same
eccentricity,
and
follow a similar phenomenon with the
and
.
corresponding relative bending stiffness
ASCE
256
1.2
DRhog
DRsag
0.4
0.6
0.4
0.2
0.0 -5
-4
-3
-2
-1
0
10 10 10 10 10 10
3
10
EI/EsBLsag EI/EsBLhog
0.0 -5
-4
-3
-2
-1
0
10 10 10 10 10 10
10
1.2
LB=20m10m
e=15m (Sagging)
e=15m (Hoagging)
1.0
1.0
DRhog
0.8
(a)
1.2
DRhog
0.8
DRsag
0.6
0.2
10
10
(b)
LB=20m10m
e=20m (Hogging)
0.8
0.6
0.6
0.4
0.4
DRsag
LB=20m10m
e=10m (Sagging)
e=10m (Hoagging)
1.0
0.8
DRhog
1.2
LB=20m10m
e=5m (Sagging)
e=5m (Hoagging)
1.0
0.2
0.2
0.0 -5
-4
-3
-2
-1
0
10 10 10 10 10 10
3
10
10
0.0 -5
-4
-3
-2
-1
0
10 10 10 10 10 10
3
EI/EsBLhog
10
10
(c)
(d)
FIG. 6 Distribution of modification factors for different eccentricities.
Fig. 7 shows that the modification factors for each foundation eccentricity fall
within a narrow cluster near the plane-strain case (the standard case). As such, the
adoption of this relative bending stiffness could take the foundation eccentricity into
account such that the magnitude of modification factor is close to unity.
Effect of Building Weight
All the above parametric studies do not consider the superstructure weight.
However, the building load may be a critical factor for foundation settlement. An
attempt is made to investigate whether the building load could severely affect the
deflection ratio. In the present study, the building loading pressure is taken as 25kPa.
In addition, since the foundation is usually assumed to be wished-in-place, the
displacement induced by the load would reset to zero after the initial geostatic
equilibrium stage. The foundation size and locations are the same as those adopted
earlier.
Fig. 8 illustrates the distribution of modification factors corresponding to different
building locations. Fig. 8(d) demonstrates that similar relationship between the
modification factor and the new relative bending stiffness can be obtained when the
building is located within the sagging zone or hogging zone. It can be established
from Figs. 8(a) and (b) that for a foundation with the same eccentricity, the
distributions of
and
versus the corresponding relative bending
stiffness are consistent. However, they are slightly higher than those of a building
without eccentricity.
ASCE
257
1.2
e/B=0.25 (Sagging)
e/B=0.25 (Hogging)
e/B=0.5 (Sagging)
e/B=0.5 (Hogging)
e/B=0.75 (Sagging)
e/B=0.75 (Hogging)
e/B=1 (Hogging)
Plane strain
Strandard
0.8
DRhog
1.0
0.4
DRsag
0.6
0.2
0.0 -5
-4
-3
-2
-1
0
10 10 10 10 10 10
3
10
10
10
10
Fig. 7 Effect of building eccentricities on modification factors.
DRsag
0.6
DRhog
0.8
0.4
0.6
0.4
0.2
0.0 -5
-4
-3
-2
-1
0
10 10 10 10 10 10
3
10
0.0 -5
-4
-3
-2
-1
0
10 10 10 10 10 10
10
1.2
10
10
(b)
LB=20m10m
e=0m (Sagging)
e=20m (Sagging)
1.0
0.8
0.8
DRhog
LB=20m10m
e=0m (Sagging)
e=15m (Sagging)
e=15m (Hogging)
1.0
(a)
1.2
DRhog
0.8
0.2
DRsag
0.6
0.4
0.6
0.4
DRSag
LB=20m10m
e=0m (Sagging)
e=10m (Sagging)
e=10m (Hogging)
1.0
1.0
DRsag
1.2
LB=20m10m
e=0m (Sagging)
e=5m (Sagging)
e=5m (Hogging)
DRhog
1.2
0.2
0.2
0.0 -5
-4
-3
-2
-1
0
10 10 10 10 10 10
3
EI/EsBLsag EI/EsBLsag
10
10
0.0 -5
-4
-3
-2
-1
0
10 10 10 10 10 10
3
EI/EsBLsag EI/EsBLsag
10
10
(c)
(d)
FIG. 8 Distribution of modification factors for building with weight
The comparison between the plane-strain standard case and the modification factors
for a building with weight is shown in Fig. 9 where
and
for various
foundation locations fall into a band close to the standard case. Therefore, the
building weight is not a critical factor affecting the deflection ratio of the building due
to tunneling. The plane-strain standard case can be used to provide guidance on the
influence building stiffness on its differential settlement behavior.
ASCE
258
1.2
BL=10m20m
e/B=0 (Sagging)
e/B=0.25 (Sagging)
e/B=0.25 (Hogging)
e/B=0.5 (Sagging)
e/B=0.5 (Hogging)
e/B=0.75 (Sagging)
e/B=0.75 (Hogging)
e/B=1 (Hogging)
Plane strain
Strandard
0.8
DRhog
1.0
0.4
DRsag
0.6
0.2
0.0 -5
-4
-3
-2
-1
0
10 10 10 10 10 10
3
10
10
10
10
FIG. 9 Effect of building weight on modification factors

CONCLUSIONS
In this paper, a new relative bending stiffness parameter is proposed to unify the
distribution of modification factor of the deflection ratio for various building widths
and locations in relation to the tunnel axis. The effect of building weight on the
deflection ratio is examined and the results indicate that the effect of building weight
is insignificant. As such, the modification factors versus the relative bending stiffness
obtained from the plane-strain case can be employed to estimate the deflection ratio
of a building supported on shallow foundation due to tunneling using the building
bending stiffness and the Greenfield ground movement.
REFERENCES
Burland, J.B. and Wroth, C.P. (1974). "Settlement of buildings and associated
damage: Session V review paper." Conference on Settlement of Structures,
Cambridge: 611-654, London: Pentch Press.
Franzius, J. N., Potts, D. M. and Burland, J. B. (2006). "The response of surface
structures to tunnel construction." Proc. Institution of the Civil Engineers, Geotech
Eng, Vol. 159(1): 3-17.
Goh, K.H. and Mair, R.J. (2011). "Building Damage Assessment for Deep
Excavations in Singapore and the Influence of Building Stiffness. " Geotech Eng J
SEAGS & AGSSEA, Vol. 42(3): 1-12.
Lee, K.M. and Rowe, R.K. (1991). "An analysis of three-dimensional ground
movements: the Thunder Bay tunnel." Can Geotech J, Vol. 28(1): 25-41.
Loganathan, N., Poulos, H. G. and Stewart, D. P. (2000). "Centrifuge model testing of
tunnelling-induced ground and pile deformations". Geotechnique, Vol. 50(3): 283294.
Potts, D.M. and Addenbrooke, T.I. (1997). "A structures influence on tunnellinginduced ground movements." Proc. Institution of the Civil Engineers, Geotech
Eng, Vol. 125(2): 109-125.
USACE (1990). Settlement Analysis, Enineer Manual EM 1110-1904. U.S. Army
Corps of Engineers: Washington, D.C.
ASCE
259
A Winkler-Based Method for the Assessment of Tunnelling-Induced

Deformations on Piled Structures
Andrea Franza1; Twana K. Haji2; and Alec M. Marshall3
1
evxaf2@nottingham.ac.uk
2
evxtkha@nottingham.ac.uk
3
Lecturer, Dept. of Civil Engineering, Univ. of Nottingham, Nottingham NG7 2RD, U.K. E-mail:
alec.marshall@nottingham.ac.uk
Abstract: In urban areas it is often required to assess the deformation distributions

induced by tunnel construction on piled buildings. In this paper, a simple analysis
method is presented for estimating these deformations. The proposed method is based
on a two-stage procedure: (1) estimation of the greenfield ground movements caused
by the tunnel excavation, and (2) analysis of the foundation/building on elastic springs
subjected to a system of forces induced by the greenfield ground movements. Simple
closed-form expressions are used for the evaluation of spring stiffnesses and
tunnelling-induced forces. The proposed method is validated by comparing its
predictions with those from 3D finite element analyses. It is shown that the simplified
analysis method provides a good assessment of building deformations for several
cases of tunnelling beneath piles and relative soil-structure stiffnesses. The proposed
method represents a useful tool for preliminary parametric analyses due to the reduced
computational cost and simple yet versatile implementation compared to 3D numerical
analyses.
INTRODUCTION
Urban development often involves the construction of tunnels. To satisfy the
structural serviceability state, engineers need to assess the effect of deformations
induced by tunnel construction on piled buildings. Several researchers have
investigated the effects of tunnel construction adjacent and beneath piles, leading to a
satisfactory confidence in the assessment of pile group displacements and internal
forces (Basile 2014; Huang et al. 2009). Although several studies have investigated
the tunnel-building interaction (Franzius et al. 2006; Mair et al. 1996), the
understanding of the role of piles in the global tunnel-pile-building interaction is
limited. As mentioned by Mair and Williamson (2014), despite the importance of
predicting structural settlements caused by tunnelling beneath piled foundations, there
ASCE
is a lack of methods for performing preliminary damage risk assessments of piled

buildings. It is well known that soil behaviour and soil-structure interactions are nonlinear, however many useful methods for tunnel-structure interaction analyses have
been developed using simplified elastic solutions (e.g. Huang et al. 2009). The method
presented in this paper also adopts an elastic framework, using a two-stage simplified
Winkler analysis. This method provides the displacements resulting from tunnel
construction in a generic structure; it is able to highlight the main tunnel-piled
foundation interaction mechanisms.
METHOD OF ANALYSIS
The problem studied in this paper considers an elastic structure on a foundation
composed of q circular vertical piles embedded in an elastic layered deposit where a
tunnel is constructed beneath the piles. The method is based on a two-stage procedure:
(1) estimation of the greenfield ground movements caused by the tunnel excavation,
and (2) analysis of the building on elastic springs subjected to a system of forces
induced by greenfield ground movements. The method considers only the vertical
degrees of freedom of the piles. This simplifying assumption is compatible with the
fact that tunnel construction induces negligible horizontal strains in structures with
continuous footings (Burland et al. 2004, Goh and Mair 2014). Moreover, the
conservative assumption of elastic soil medium, which neglects soil nonlinearities,
should lead to realistic predictions as long as the tunnel volume loss, Vl,t, does not
exceed a value of 1% (Basile 2014, Zhang et al. 2011).
FIG. 1. Simplified Winkler model for tunnel-pile-structure interaction: (a) single

pile case; (b) piled structure case.
Estimation of greenfield ground movements
In the analysis presented here, the closed-form expression proposed by Loganathan
and Poulos (1998) for the prediction of vertical greenfield tunnelling-induced
displacements in clays is implemented. It is important to note that any input of
greenfield displacements can be used in the method, thereby enhancing the versatility
of the approach.
ASCE
260
261
Global tunnel-piled structure interaction

Tunnel-single pile interaction
The schematic representation of the problem for the single pile case is shown in Fig.
1a. For the Winkler assumption, vertical linear springs distributed along the pile shaft
and base are used to simulate a layered soil deposit with no slip at the pile-soil
interface. The pile is discretized into n+1 nodes. If each pile is assumed rigid, it is
possible to provide the following simple closed-form expressions for the stiffnesses of
the equivalent pile-soil spring, Kp, and the resultant tunnelling-induced vertical force,
Fp, at the p-th pile head.
Lp
n +1
K p = k z ( z ) dz + K b k z ( z i ) z i + K b
Lp
Fp =
n +1
(1)
( z ) k z ( z ) dz + s z (L p )K b s z ( z i ) k z ( z i ) z i + s z (L p )K b
where kz and Kb are, respectively, the stiffness of shaft springs (per unit-length of pile)
and the base spring stiffness of the pile; zi is the effective pile length corresponding
to the i-th node, and sz is the greenfield vertical soil movement induced by tunnel
excavation at the pile axis line. As suggested by Randolph et al. (1978), kz and Kb are
assumed equal to
kz =
Kb =
2G s
2.5 L p (1 s )
ln
dp /2
d p Eb
(2)
1 b2
where Gs and s are the shear modulus and the Poissons ratio of the soil layer at the
node depth; Eb and b are the shear modulus and the Poissons ratio of the soil deposit
at the pile toe; dp is the pile diameter; Lp is the total pile length.
Global tunnel-piled structure interaction
If each pile is treated as an isolated case, the global tunnel-pile-structure interaction
problem is idealized as an elastic structure supported by independent vertical springs,
which account for the soil deformability (see Fig. 1b). The structure is restrained in the
horizontal direction. The only degrees of freedom of this system are the vertical
displacements of structural nodes connected to the springs (pile heads). The tunnel
excavation induces a system of vertical forces at the foundation level. The equilibrium
equation is formulated by adding the contribution of soil stiffness to the condensed
stiffness matrix of the structure. The equilibrium equation is
(K s + K g )u p = f p
(3)
where Ks is the condensed stiffness matrix of the structure, Kg is the stiffness matrix
of the soil-pile group system, up is the displacement vector of the piled foundation and
ASCE
262
fp is the vector of the tunnelling-induced forces. The condensed stiffness matrix of the
structure is a full matrix, whereas the stiffness matrix of the pile-soil system is a
diagonal matrix because pile-pile interaction is neglected. The elements up,i and fp,i are,
respectively, the displacement and the tunnelling-induced force at the i-th degree of
freedom (dof). If the structure degrees of freedom are fixed, Ks,ij is the structural
reaction force in the i-th dof due to a unit displacement of the j-th dof. Kg,ii is the
equivalent stiffness of the i-th pile.
Once the equilibrium equation is solved, structural deformations can be computed by
displacing the constrained structure according to the solution displacement vector up.
In this paper, the condensed stiffness matrix of the elastic structure was obtained with
a finite element code using the Euler-Bernoulli beam theory. Alternatively, the second
stage may be solved directly through a 3D numerical modelling software, simulating
the structure on vertical springs subjected to the tunnelling-induced force system.
VERIFICATION BY COMPARISON WITH 3D NUMERICAL ANALYSES
In this section, the Winkler-based method predictions are compared against more
rigorous 3D numerical analyses performed with Abaqus (Simulia 2010). This section
demonstrates that the proposed method allows for a reliable assessment of piled
building deformations due to tunnel construction. The influences of tunnel location as
well as soil and structure stiffnesses are also investigated.
Lp
db
Configurations
The validation of the proposed method was performed by analysing the deformation
profile of a beam with a foundation comprising a row of either 5 or 11 piles. The
beam, with a stiffness representing a foundation and superstructure, was located with
its centre at a horizontal distance Xt from the tunnel centreline. The pile spacing was
fixed, hence the 5-pile building model has a width B=20m, whereas the 11-pile
foundation has a width B=50m. Fig. 2 summarizes the considered tunnel-pile-structure
configurations.
FIG. 2. Studied configurations.
ASCE
Several relative soil-structure stiffness ratios, given by Es/Eb, where Es and Eb are the
Youngs modulus of the soil and foundation beam, respectively, and tunnel locations,
given by Xt, were investigated.
Model details
The ABAQUS model simulated both the soil and piles using 3D 8-node linear brick,
reduced integration solid elements (C3D8R). The mixed analytical-numerical
approach for soil-structure interaction analysis used by Klar and Marshall (2008) was
adopted here. This ensured that the input of soil displacements due to tunnelling in the
numerical model were consistent with those used in the Winkler method.
The mixed analytical-numerical analysis consists of two stages. In the first stage, all
nodes of the soil model are forced to displace vertically according to a chosen input
for greenfield settlements, in this case the closed-form expression proposed by
Loganathan and Poulos (1998), and the reaction forces of the nodes (nodal forces
required to produce the applied displacements) are recorded. In the second stage, the
model is returned to its original condition (before deformation) and the selected
structure, in this case the piles and superstructure (beam), are then added to the model.
The nodal reaction forces recorded in the previous stage are then applied to the model
which includes the added structure. Any difference in soil displacements between the
two stages of the model is due to the existence of the added structure (piles and
superstructure). All other aspects of the ABAQUS model were consistent with the
assumptions adopted in the Winkler method: tie connections, elastic linear isotropic
materials, no contact between the soil and beam, and weightless materials. The model
dimensions were set to ensure that boundary conditions did not affect results.
Results and discussion
A comparison of the numerical and Winkler model results of vertical surface
settlement are presented in Fig. 3 and 4 for structures with Eb = 30 and 600 GPa,
respectively, and for varying soil stiffness and building locations. Especially for the
more flexible building in Fig. 3, the results agree well with the observations of Mair
and Williamson (2014): piles with their toes above the tunnel settle more than the
surface level settlement and less than the surface level settlement otherwise. This
happens because the piles transfer a proportion of the displacements occurring along
their length (i.e. at greater depths where greenfield displacements near the tunnel are
larger) to the ground surface.
The following observations can be made:
Compared to the numerical analysis, the proposed method gives a good and
generally slightly conservative assessment of the tunnelling-induced
deformation profile.
Examination of Fig. 3c, 3d, 4c and 4d indicates that the superstructure shows
more flexible behaviour when the ratio between soil and structure Youngs
modulus and/or the ratio between structure width and tunnel depth is high.
The effect of superstructure stiffness is to reduce the maximum relative

deflection of the piled structure (in Fig. 4, the settlement of the 5-pile building varies
almost linearly with transverse distance), where the relative deflection is the distance
between the settlement curve and the segment connecting two points of the curve.
ASCE
263
Vertical settl.(mm)
Vertical settl. (mm)
264
-1
-1
Xt = 0m
Xt = 0m
0
0
1
1
2
2
3
3
4
4
5
5
6
Es = 24 MPa 6
7
Eb = 30 GPa 7
(a)
(b)
8
8
-1 -25 -20 -15 -10 -5 0 5 10 15 20 25-1 -25 -20 -15 -10 -5 0
Xt = -15m
Xt = -15m
0
0
1
1
2
2
3
3
4
4
5
5
6
Es = 24 MPa 6
7
(d)
Eb = 30 GPa 7
(c)
8
8
-1 -25 -20 -15 -10 -5 0 5 10 15 20 25-1 -25 -20 -15 -10 -5 0
Xt = -25m
Xt = -25m
0
0
1
2
3
4
5
6
7
8
(e)
-25 -20 -15 -10 -5 0
1
2
3
4
5
6
Es = 24 MPa
Eb = 30 GPa 7
(f)
8
5 10 15 20 25 -25 -20 -15 -10 -5 0
5 10 15 20 25
Es = 100 MPa
Eb = 30 GPa
5 10 15 20 25
Es = 100 MPa
Eb = 30 GPa
5 10 15 20 25
x (m)
x (m)
Greenfield Surface
Settlements
Es = 100 MPa
Eb = 30 GPa
Abaqus B=50m
Abaqus B=20m
Winkler B=50m
Winkler B=20m
FIG. 3. Comparison of numerical and Winkler model results: Eb = 30 GPa.

Results highlight the role of piles in the tunnel-pile-structure interaction. Because of
the interaction of the piles with subsurface ground movements, which become
increasingly narrow with depth, piles increase the relative deflection of the structure
compared to shallow foundations. Furthermore, piled foundations are generally stiffer
than shallow foundations, therefore the piled foundation will have a more dominant
role in a structure-foundation interaction analysis. To evaluate the potential damage of
piled buildings, the limiting tensile strain method (Mair et al., 1996) may be used with
the building settlement profile obtained with the Winkler-based method rather than the
greenfield surface settlement trough. Although the proposed analysis method does
account for the effect of horizontal strains, these should be negligible in case of
foundations with continuous footings.
ASCE
265
-1
-1
Xt = 0m
Xt = 0m
0
0
1
1
2
2
3
3
4
4
5
5
6
Es = 24 MPa 6
7
(a)
Eb = 600 GPa 7
(b)
8
8
-1 -25 -20 -15 -10 -5 0 5 10 15 20 25-1 -25 -20 -15 -10 -5 0
Xt = -15m
Xt = -15m
0
0
1
1
2
2
3
3
4
4
5
5
6
Es = 24 MPa 6
7
Eb = 600 GPa 7
(d)
(c)
8
8
-1 -25 -20 -15 -10 -5 0 5 10 15 20 25-1 -25 -20 -15 -10 -5 0
Xt = -25m
Xt = -25m
0
0
1
2
3
4
5
6
7
8
(e)
-25 -20 -15 -10 -5 0
1
2
3
4
5
6
Es = 24 MPa
7
(f)
Eb = 600 GPa
8
5 10 15 20 25 -25 -20 -15 -10 -5 0
x (m)
Greenfield Surface
Settlements
Es = 100 MPa
Eb = 600 GPa
5 10 15 20 25
Es = 100 MPa
Eb = 600 GPa
5 10 15 20 25
Es = 100 MPa
Eb = 600 GPa
5 10 15 20 25
x (m)
Abaqus B=50m
Abaqus B=20m
Winkler B=50m
Winkler B=20m
FIG. 4. Comparison of numerical and Winkler model results: Eb = 600 GPa.

CONCLUSIONS
This paper presented a simple elastic Winkler based method of analysis for piled
buildings subjected to tunnelling-induced ground movements. Results obtained from
the model compared very well to finite element analyses for cases where both soil and
structure stiffness as well as the location of the structure were varied. Results indicate
that the proposed method is suitable for preliminary assessment of building/foundation
deformations. The method has the advantage that it permits implementation of a
framed or a 3D structure rather an equivalent beam or plate at the foundation level.
ASCE
Results highlighted that neglecting the pile contribution in the tunnel-pile-structure

interaction is not conservative. To account correctly for the relative structure-piled
foundation stiffness and the pile transfer of subsurface movements to the surface level,
the structural damage may be assessed with the limiting tensile strain method using the
building settlement profile obtained with the Winkler-based method.
Finally, it is worth emphasising that this Winkler-based method has a simple
implementation and allows for a remarkable reduction of the computational cost
compared to complete 3D numerical analysis. Thus, it would be a suitable tool for
parametric analysis. This work may be extended to more complex structural
configurations in the future.
REFERENCES
Basile, F. (2014). "Effects of tunnelling on pile foundations." Soils & Found., Vol. 54
(3): 280-295.
Burland, J.B. Mair, R.J. and Standing, J.R. (2004). "Ground performance and building
response due to tunnelling." In: Proc. Int. Conf. on Advances Geotech. Engrg., Vol.
1: 291-342.
Goh, K.H. and Mair, R.J. (2014). "Response of framed buildings to excavationinduced movements." Soil & Found., Vol. 54 (3): 250-268.
Franzius, J.N. Potts, D.M. and Burland, J.B. (2006). "The response of surface
structures to tunnel construction." Proc. Inst. Civ. Eng. Geotech. Eng., Vol. 159 (1):
317.
Huang, M. Zhang, C. and Li, Z. (2009). "A simplified analysis method for the
influence of tunneling on grouped piles. " Tunn. Undergr. Space Technol., Vol. 24
(4): 410-422.
Klar, A. and Marshall, A.M. (2008). "Shell versus beam representation of pipes in the
evaluation of tunneling effects on pipelines." Tunn. Undergr. Space Technol., Vol.
23 (4): 431437.
Loganathan, N. and Poulos, H.G. (1998). "Analytical prediction for tunneling-induced
ground movements in clays." J. Geotechnical & Geoenv. Engrg., Vol. 124 (9): 846856.
Mair, R.J. Taylor, R.N. and Burland, J.B. (1996). "Prediction of ground movements
and assessment of risk of building damage due to bored tunnelling." In: Proc. Int.
Symp. Geotech. Aspects Undergr. Constr. Soft Ground: 713-718.
Mair, R.J. and Williamson, M.G. (2014). "The influence of tunnelling and deep
excavation on piled foundations." In: Proc. Int. Symp. Geotech. Aspects Undergr.
Constr. Soft Ground: 21-30.
Randolph, M.F. Wroth, P.C., & Wroth, C. (1978). Analysis of deformation of
vertically loaded piles. J. Geotechnical & Geoenv. Engrg. Division, Vol. 104 (12):
1465-1488.
Simulia, D.S. (2010). Abaqus analysis users manual.
Zhang, R. Zheng, J. Pu, H., & Zhang, L. (2011). "Analysis of excavation-induced
responses of loaded pile foundations considering unloading effect." Tunn. Undergr.
Space Technol., Vol. 26 (2): 320-335.
ASCE
266
Effectiveness of Buttress and Cross-Wall in Deep Excavations

Shong-Loong Chen1; Cheng-Tao Ho2; Chia-Yu Yeh3; and Meen-Wah Gui, M.ASCE4
1
Professor, Dept. of Civil Engineering, Taipei Tech, No 1, Sec 3, ZhongXiao E Rd., Taipei 10608,
Taiwan. E-mail: f10391@ntut.edu.tw
2
Senior Engineer, Chao-Wei Engineering Consultant, 5F, No 185, Sec 1, DunHua S Rd., Taipei 106,
Taiwan. E-mail: s4679005@ntut.org.tw
3
Graduate Student, Dept. of Civil Engineering, Taipei Tech, No 1, Sec 3, ZhongXiao E Rd., Taipei
10608, Taiwan. E-mail: a0905a0905@yahoo.com.tw
4
Professor, Dept. of Civil Engineering, Taipei Tech, No 1, Sec 3, ZhongXiao E Rd., Taipei 10608,
Taiwan (corresponding author). E-mail: mwgui@ntut.edu.tw
Abstract: Buttress and cross-wall have been widely adopted with success in soft
ground excavation work. The interaction between the soil, buttress/cross-wall and
diaphragm wall is a three dimensional (3D) problem but in practice the problem is
commonly simplified into a two-dimensional problem. This study compared the
contribution of buttress and cross-wall on the displacement of diaphragm wall via a
series of 3D finite element (FE) analysis. The numerical procedure was first calibrated
against the field data obtained from a 16 m deep excavation project in Taipei city. The
model was then used in the subsequent parametric study to evaluate the efficiency of
buttress and cross-wall for a diaphragm wall supported excavation. Result showed that
cross-wall and buttress were equally efficient in restraining the lateral displacement of
diaphragm wall.
INTRODUCTION
The lateral displacement of retaining walls is dealt with in many ways, such as via
cross-wall, buttress, ground improvement, thicker diaphragm wall and a stiffer
strutting system. Cross wall and buttress are used extensively as they provide better
retaining result (Chen et al., 2011; Chen et al., 2014). In practice, the analysis of these
retaining structures is often simplified to provide contribution similar to ground
improvement. For example, Hsieh and Ou (1998) converted the contribution of
buttress and cross-wall to the equivalent shear strength increase of the ground for
analysis carried out using the program such as RIDO and TORSA. The simplification
may be convenient but it does not provide the true picture of three-dimensional (3D)
excavation.
ASCE
267
268
Certain semi-empirical formulas can be used to account for the 3D effect of

excavation; for example, Ou et al. (1996) proposed the plane strain ratio (PSR) to
estimate the 3D displacement of a retaining wall from a series of 2D numerical
analysis. Finno et al. (2007) modified the PSR concept to relate their 2D excavation
behavior to the 3D one. Ou et al. (2008) pointed out the use of 3D finite element
analysis allows effective simulation of displacement in diaphragm walls braced with
buttress. When restrained at the bottom with buttress, the diaphragm walls experience
less displacement. Ou et al. (2011) concluded that the use of cross-wall enable them to
reduce the maximum lateral displacement of a diaphragm wall, constructed for a deep
excavation project in Taipei, to as much as 75%~82% of that without the use of crosswall. Field data obtained by Wu et al. (2013) from the 22 monitoring cases, with and
without cross-walls, also suggested that cross wall effectively restrained the wall
displacement.
This study aims at numerically examining the effects of using cross-wall and
buttress in an excavation project using the program Plaxis3D (Brinkgreve and Swolfs,
2007). The numerical result was first compared with field monitored data to justify
the numerical procedures. A series of parametric study on the influence numbers of
cross wall and buttress on diaphragm wall displacement was also conducted.
STUDY PROJECT
The excavation site was the construction site for a 15-storey building with 4-level of
basements in Taipei City. The excavation site was 52.8 m long and 36 m wide with an
excavated volume of 30,400 m3. The 16 m deep excavation was supported by
diaphragm wall with a thickness of 80 cm and 32 m long; the layout of the excavation
is shown in FIG. 1. Five cross walls were installed across the excavation site together
with seven buttresses installed on the east and west sides of the site. Bottom-up
construction technique was adopted, excavation was performed in six stages with five
levels of strutting.
FIG. 1. Layout of diaphragm walls,

buttress (BW) and cross (CW) walls; and
the locations of inclinometer.
ASCE
FIG. 2. Mesh used in FE analysis.
269
Soil and Structure Parameters

The Mohr-Coulomb (MC) model is the most common soil material combination
law used in practice. The parameters required for the MC model are unit weight of soil
t, undrained shear strength cu, and internal friction angle , all of which can be
obtained from the tests conducted for boring report. The soil parameters obtained for
the site were tabulated in Table 1. The undrained modulus of elasticity Eu of clay in
Taipei Basin has been widely adopted as 600Su. The modulus of elasticity for sand
was calculated from E = 2 VS2 (1 + ) , where is the mass density of soil; (0.3<<0.5)
is the modifier for large strains generated during excavation was taken to be 0.5, as
recommended by Ou et al. (2006). The shear wave velocity, VS (m/sec), was calculated
using the empirical formula of Lee (1990): Vs = 74.44N0.16H0.25, where N is the SPT-N
value obtained from standard penetration test, and H the depth of soil layer. The
modulus of elasticity of diaphragm walls is a function of the 28-day compressive
strength of concrete.
Table 1. Soil parameters for the FEM model.
Soil
layer
SF
SM
CL-ML
CL
CL
CL-ML
SM
CL-ML
CL-ML
GM
Depth
(m)
3.3
10.7
15.6
20.4
24.1
28.5
30.8
41.3
55.6
56.6
SPT
N
4
7
3
3
5
7
16
10
18
>50
t
(kN/m3)
17.8
19.5
18.6
18.1
18.7
18.8
19.1
18.9
18.9
22.1
cu
(kPa)
0
0
36
47
56
65
0
93
122
10
(deg)
29
30
28
28
29
30
32
32
33
40
(deg)
0
0
0
0
0
0
2
2
3
10
E' or Eu
(kPa)
14700
69000
21600
28200
33600
39100
180000
55800
73200
615000
0.35
0.30
0.35
0.35
0.35
0.35
0.30
0.35
0.35
0.25
The reduction in effective moment of inertia due to concrete cracking was also
considered as its stiffness experiences changes under large bending moment. The
design stiffness was obtained by multiplying the stiffness of concrete with a reduction
factor ranged between 0.6 and 0.8. For this study, the reduction factor was taken as
0.7 for the diaphragm wall, cross-wall and buttress. A reduction in strutting stiffness
was also considered as the quality of installation, the fitness and fatigue of steel
materials would be imperfect on site. The reduction factor is generally ranged between
0.5~0.7; a reduction factor of 0.6 has been taken in this study. Beam element was used
to simulate the H-shaped steel struts and it has a yielding strength of 256 MPa. Table 2
shows the material parameters used in this study.
Numerical Model
The site was an asymmetric excavation site and thus full cross section analysis was
carried out. A roller boundary should be at least 3 times the excavation depth from the
ASCE
270
diaphragm wall while for ground settlement, Peck (1969), Clough and ORourke
(1990), Hsieh and Ou (1998) and Wang et al. (2010) suggested that the influence zone
caused by excavation in soft soil could be between 2~3.5He. Thus, vertical boundaries
were at 4 times the excavation depth from the diaphragm wall while the bottom
horizontal boundary was set at 5 times the excavation depth, i.e. 56.5 m deep at the
bottom of the diaphragm wall, in the hard soil layer. The 3D FE analysis mesh used in
the study is shown in Fig. 2. Bottom-up construction was adopted for this building.
The site was excavated in 6-stage with 5-level of strutting.
Table 2. Material parameters for diaphragm wall (DW), buttress (BW), crosswall (CW) and struts.
Material
property
DW
CW
Depth
from (m)
0~32
13.6~26
0.8
0.8
23.5
23.5
23.5
77.1
24.6
0.15
24.6
0.15
24.6
0.15
206
0.30
Thickness
(m)
Unit wt.
(kN/m3)
E (MPa)
BW
Struts
Level 1: H3003001015
13.6~32 Level 2: H3503501219
Level 3: 2H3503501219
Level 4: 2H4004001321
0.8
Level 5: 2H4004001321
Table 3. Construction sequence

Steps
1
2
3
4
5
6
7
8
9
10
11
12
13
ASCE
Items
Initial equilibrium of stress
Installing diaphragm walls, cross walls, buttresses and posts
(H4004001321, L=32m)
1st stage excavation to GL-2.05m
Installing 1st level of struts H3003001015, preloaded to 294kN/strut
2nd stage excavation to GL-3.40m
Installing 2nd level of struts H3503501219, preloaded to 588kN/ strut
3rd stage excavation to GL-7.30m
Installing 3rd level of struts 2-H3503501219, preloaded to 588kN/ strut
4th stage excavation to GL-10.30m
Installing 4th level of struts 2-H4004001321, preloaded to 785kN/ strut
5th stage excavation to GL-13.40m
Installing 5th level of struts 2-H4004001321, preloaded to 1178kN/ strut
6th stage excavation to foundation level at GL-16.00m
271
Numerical result and field observed data

Four inclinometers SI-1~SI-4 were installed in the diaphragm wall (Fig. 1). A typical
comparison of the numerical and field monitored results at SI-3 for each of the six
construction stages is shown in Fig. 3. The numerical profile was in good agreement
with the observed data in each excavation stage. The comparison between the
maximum diaphragm wall displacement and that observed from the field monitored
data was summarized in Table 4. The maximum lateral displacements obtained from
the FE analysis (3Dm) were between 0.84~4.30 cm, while that from the monitored data
(m) was between 0.64~5.20 cm. The inclinometers data were 97~123% of that from
the FE analysis. Both the results/data at locations SI-3 and SI-4 were in particular
good agreement, about 96% of the monitoring data (Table 4). In addition, the location
of the maximum lateral displacement in the diaphragm walls was also very close to
that observed in the field, indicating that the FE procedures used in this study was
effective for the analysis of buttress- and cross-walls supported excavation.
-4
-4
-4
-8
-8
-8
-12
-12
-12
-16
-20
-28
Depth (m)
-16
-20
SI-3
-28
FEA-stage 1
SI-3
-20
-28
Lateral Displacement (cm)

-2 0 2 4 6 8 10
-32
-2 0 2 4 6 8 10

-2 0 2 4 6 8 10
0
-4
-4
-4
-8
-8
-8
-12
-12
-12
-20
-24
-28
-32
Depth (m)
Depth (m)
-16
-16
-20
-28
-16
-20
-24
-24
SI-3
SI-3
-28
-32
SI-3
FEA-stage 6
FEA-stage 5
FEA-stage 4
SI-3
FEA-stage 3
FEA-stage 2
-32
-32
-16
-24
-24
-24
Depth (m)

-2 0 2 4 6 8 10
0

-2 0 2 4 6 8 10
Depth (m)
Depth (m)

-2 0 2 4 6 8 10
-32
FIG. 3. Lateral displacement: comparison between FE analysis and inclinometer

SI-3 data for each excavation stage.
ASCE
272
EFFECTIVENESS OF BUTTRESS AND CROSS-WALL

Chen et al. (2011) used the so-called displacement reduction ratio (DRR) to
indicate the efficient use of buttress:
DRR =
hm hmb
hm
(1)
where hm and hmb is the maximum lateral displacement of the diaphragm wall
without and with buttress, respectively. In this study, the contribution of cross wall in
diaphragm wall supported excavation is also analyzed using the analogous definition,
with simulation conducted with 0, 1, 3 or 5 numbers of buttress and cross-walls.
Table 4. Maximum displacement: FE analysis and monitored data.
No.
m (cm)
3Dm (cm)
m /3Dm
SI-1
5.20
4.30
121%
SI-2
0.64
0.84
131%
SI-3
3.39
3.53
96%
SI-4
0.94
0.97
97%
Maximum Wall Disp (cm)
N.B.=0
N.B.=1
N.B.=3
N.B.=5
SI-3
6
4
2
0
0
12
16
20
24
Distance from the corner (m)
28
32
36
(a)
N.B.=0
N.B.=1
N.B.=3
N.B.=5
SI-3
Maximum Wall Disp (cm)
8
6
4
2
0
0
12
16
20
24
Distance from the corner (m)
28
32
36
(b)
FIG. 4. Effect of 0~5 numbers of buttress on the west side of DW on the DWs
lateral displacement: (a) with 5 CW in N-S direction; (b) the 5 CW were replaced
by buttress.
ASCE
273
The displacement profile result obtained from the analysis with five N-S crosswalls and various numbers of buttress on the west diaphragm wall are shown in Fig.
5(a). The displacement profile of the west diaphragm wall reduces as the number of
buttress (NB) increases from 0 to 5. As for case where the five 36 m long cross-walls
were replaced by five 7.5 m long buttresses on each of the north and south diaphragm
walls, the diaphragm wall displacement is shown in Fig. 5(b). Clearly, the diaphragm
wall displacement was larger this time, albeit not very drastically. This is perhaps the
buttress itself was sufficiently stiff in preventing the wall displacement.
Table 5. DRR of buttress on west diaphragm wall.
Item
No. of buttress
walls (NB)
BW spacing (m)
hmb (cm)
DRR(%)
5 cross-walls across the

excavation site
No cross-wall but with 5

buttresses at North and South
diaphragm walls
N/A
7.06
0
18.0
5.23
26.0
11.4
4.67
33.9
6.6
3.57
49.7
N/A
6.48
0
18.0
4.63
28.5
11.4
4.05
37.4
6.6
2.85
55.9
As shown in Table 5, with five cross-wall across the excavation site, the
maximum lateral displacement of diaphragm wall and DRR for NB=0, 1, 3 and 5 on
the west diaphragm wall was 7.06 cm, 5.23 cm, 4.67 cm and 3.57 cm, respectively;
while the DRR was 0%, 26%, 33.9% and 49.4%, respectively. However, with five 7.5
m long buttresses to replace the cross-wall, the maximum lateral displacement of the
diaphragm wall for NB=0, 1, 3 and 5 became 6.48 cm, 4.63 cm, 4.05 cm and 2.85 cm,
respectively; while the DRR was 0%, 28.5%, 37.4% and 55.9%, respectively. This
must be due to the high stiffness contributed by the cross-wall in the north and south
diaphragm walls than that in east and west walls supported only by buttresses. This
must be due to the compliancy of the soil in the latter case
CONCLUSIONS
A series of 3D FE analysis had been conducted to examine the displacement
behavior of diaphragm wall during an excavation in Taipei City. The diaphragm wall
was supported by five cross-walls running across N-W of the excavation site while on
the west side various numbers of buttress (0~5) was provided to examine the
displacement of the west wall. The following conclusions have been reached:
The 3D FE result was with 97%~123% of the field observed data; in addition, the
location where the maximum lateral displacement in the diaphragm wall was also
close to that observed in the field, indicating that the numerical procedures and
materials parameters adopted were reliable.
Both the buttress and cross-wall were effective in reducing lateral displacement of
diaphragm wall. In either cases, the displacement reduction ratio DRR increases as
the spacing of the buttresses decreases, thus it is more efficient in reducing the lateral
displacement of a diaphragm wall. When the buttress spacing was about 20% (=6.6
m/36 m), the DRR could be as high as 50%.
ASCE
The different between whether to use cross-wall or buttresses in the current

configuration of excavation layout was not very clear. Both choices attracted
comparable DRR; this could also mean that it might be uneconomical to prefer the 7.5
m buttress to the 36 m cross-wall. A huge saving in materials and cost could have been
served had buttress was chosen in the first place. More studies are required to further
quantify this issue.
ACKNOWLEDGMENTS
The authors would like to acknowledge the assistance given by Kai Gu Engineering
Co., Ltd. who has kindly provided the field data that made this study possible.
REFERENCES
Brinkgreve, R. B. J. and Swolfs, W. M. (2007). "PLAXIS 3D Foundation: Material
Models Manual Version 2." PLAXIS bv, Delft, The Netherlands.
Chen, S. L., Ho, C. T. and Gui, M. W. (2014). "Diaphragm wall displacement due to
creeping of soft clay." Proc. of the Institution of Civil Engineers- Geotechnical
Engineering, Vol 167(3): 297-310.
Chen, S. L., Ho, C. T., Li, C. D. and Gui, M. W. (2011). "Efficiency of Buttress wall
in Deep Excavations." J of GeoEngineering, Vol. 6(3): 143-155.
Clough, G. W. and ORourke, T. D. (1990). "Construction induced movements of in
situ walls." Proc. ASCE Conf. on Design and Performance of Earth Retaining
Structure, Lambe P. C. and Hansen L. A. (Eds.), (GSP 25), ASCE, Reston/VA:
439470.
Finno, R. J., Blackburn, J. T. and Roboski, J. F. (2007). "Three-Dimensional Effects
for Supported Excavations in Clay." J. Geotechnical & Geoenv. Engrg., Vol.
133(1): 3036.
Hsieh, P. G. and Ou, C. Y. (1998). "Shape of Ground Surface Settlement Profiles
Caused by Excavation." Canadian Geotechnical Journal, Vol. 35(6): 1004-1017.
Lee, S. H. (1990). "Regression models of shear wave velocities in Taipei basin." J of
the Chinese Institute of Engineers, Vol. 13(5): 519-532.
Ou, C. Y., Chiou, D. C., and Wu, T. S. (1996). "Three-dimensional finite element
analysis of deep excavations." J of Geotechnical Engrg, Vol. 122(5): 337345.
Ou, C. Y., Hsieh, P. G. and Lin, Y. L. (2011). "Performance of Excavations with
Cross Walls." J. Geotechnical & Geoenv. Engrg., Vol. 131(1): 94104.
Ou, C. Y., Lin, Y. L. and Hsieh, P. G. (2006). "Case record of an excavation with
cross walls and buttress walls." J of GeoEngineering, Vol. 1(2): 79-86.
Ou, C. Y., Teng, F. C., Seed, R. B. and Wang, I. W. (2008). "Using buttress walls to
reduce excavation-induced movements." Proc. of the Institution of Civil
Engineers-Geotechnical Engineering, Vol. 161(4): 202-209.
Peck, R. B. (1969). "Deep excavation and tunneling in soft ground." Proc. of the 7th
Int. Conf. on Soil Mechanics and Foundation Engineering, Mexico City, Mexico,
225-290.
Wang, H., Xu, Z. H. and Wang, W. D. (2010). "Wall and Ground Movements due to
Deep Excavations in Shanghai Soft Soils." J. Geotechnical & Geoenv. Engrg.,
Vol. 136(7): 985-994.
ASCE
274
Wu, H. H., Ching, J. Y. and Ou, C. Y. (2013). "Predicting Wall Displacements for
Excavations with Cross Walls in Soft Clay." J. Geotechnical & Geoenv. Engrg.,
Vol. 139(6): 914-927.
ASCE
275
Effects of Cut-and-Cover Tunnel Construction on Adjacent Piles of an Overpass in

Bangkok Soft Clay
Zaw Zaw Aye, M.ASCE1; Thayanan Boonyarak, Ph.D.2; Nutthapon Thasnanipan3; and
Sereyroath Chea4
1
Executive Vice President. E-mail: zaw@seafco.co.th

Chief of Engineering Division. E-mail: thayanan@seafco.co.th
3
Executive Vice President. E-mail: nutthapon@seafco.co.th
4
Geotechnical Engineer. E-mail: sereyroath.chea@gmail.com
Seafco Public Company Limited, 144 Prayasuren Rd., Bangchan, Klongsamwah, Bangkok, Thailand.
2
Abstract: To mitigate heavy traffic in Bangkok, a cut-and-cover tunnel is often

constructed at a congested intersection beneath an existing overpass. The effects of
stress relief due to long and narrow excavation can cause adverse impact on deep
foundation and super structure of the overpass. However, this adverse impact is still not
fully understood. To evaluate the tunnel-pile interaction, measured and back-analyzed
data collected from a cut-and-cover tunnel construction across beneath an adjacent
flyover is interpreted. The tunnel was supported by diaphragm wall, designed for 7 m
deep excavation in thick soft clay. Bored piles supported the overpass are located about
8 m away from the diaphragm wall. Numbers of instruments were installed to observe
ground and structural responses due to the excavation. Interaction between the cut-andcover tunnel and deep foundation of the overpass was back-analyzed using finite
element method. For numerical simulation, an advanced hardening soil model with small
strain stiffness was adopted. Effects of construction time on movement of diaphragm
wall was modeled and compared with the measured results. Difficulties and practical
construction method adopted under the existing flyover are discussed. Back-analyzed
responses of force in piles, movement and stress in soil due to excavation and
construction time are reported.
INTRODUCTION
Excavation of cut-and-cover tunnel results in stress relief and soil movement. When a
pile group is located within the excavation influence zone, additional axial force,
bending moment and displacement is likely to be induced. In addition, capacity and
serviceability of the pile should also be evaluated. From literature, influence zone of
excavation is estimated to be 2.5 times of the excavation depth (H) proposed by Aye et
al. (2006). Wang et al. (2010) also reported that distance of ground surface settlement
induced by excavation is ranging from 1.5 to 3.5 H, depends on soil type and stiffness of
retaining wall.
This paper presents back-analyzed responses of pile group due to construction of
nearby cut-and-cover tunnel. The numerical analysis results were validated with
measured results from the retaining wall of the tunnel. Time dependent behavior of the
ASCE
276
wall and pile is reported. Interpretation of induced axial force, bending moment and
displacement of pile due to changes in stress is given.
SITE INFORMATION
Figure 1a shows plan of a tunnel construction project at one of the major intersections
in the eastern part of Bangkok. The four-lane underpass along Srinakarin Road was in
the form of 758 m long cut-and-cover tunnel. At the intersection, the tunnel was
constructed underneath across the existing flyover of Udomsuk Road.
The design in the initial stage for the tunnel adopted bottom-up construction method
with temporary bracing to complete the excavation. Each roof-slab with dimension of
2.5 m wide and 18.8 m long made of precast pre-stressed concrete was initially planned
to be used. However, installation of heavy precast panels underneath the existing flyover
was considered difficult and high risk involved. After reviewing different options with
consideration of the constraints imposed on site and constructability, top-down
construction method with cast-in-place post-tension roof slabs was selected.
To facilitate top-down construction, roof slab openings were provided at an evenly
spacing of about 40 m for access of soil excavation underneath the slab. Adjacent to the
cut-and-cover tunnel, a pile group on each side of the underpass was located with
minimum clear distance of 8.0 m and 12.5 m
Section of the underpass is shown in Figure 1b. The thickness of the diaphragm wall is
0.8 m and depth is 25 m. At the middle of the tunnel, barrette size of 0.8 by 2.7 m with
depth of 29 m at 4.5 m spacing was used for replacing bored piles. This is because the
machine for barrette was modified for construction under the limited height of 6 m. Each
adjacent pile group of the flyover consists of six bored piles with diameter of 0.8 and
depth of 42 m.
Construction sequences
To construct the underpass (refer to Fig 1b), diaphragm wall and barrette with
embedded temporary stanchion were constructed, followed by casting of roof slab
supported by the stanchion. Before excavation started, post-tension force was applied on
the roof slab. Excavation using top-down method was performed to final level or 7 m
deep from the existing ground surface. The base slab was constructed afterward,
followed by casting of median wall between the temporary stanchions. The stanchions
were removed and the remaining median wall was constructed.
Soil profile
A soil profile at the site, derived from five soil investigation boreholes, is shown in
Figure 1b. A typical soil profile at the site is characterized by thick Bangkok soft clay at
the top, followed by thin layer of medium clay underlying by alternating stiff clay and
dense sand layers. Soil properties used for numerical analysis are given in the Section
Constitutive model and input parameters.
Geotechnical instrumentation
To monitor the behavior of the underpass construction and to ensure safety,
instruments (see Fig. 1a) consisting of inclinometer embedded in the diaphragm wall,
surface settlement markers and strain gage in the diaphragm wall were installed.
ASCE
277
278
Existing flyover
See Section in Fig. 1b
Pile group
of flyover
Inclinometer and
Strain gage
Srinakarin Road
8.0 m
G
Roof slab for cut-and-cover

tunnel section
40 m
Diaphragm wall
of underpass
4-Lane underpass
G
S1
S2
S3
S4
S5
S6
I
P
Surface
settlement
markers
Opening for
excavation
12.5 m
Udomsuk Road
(a)
Existing flyover
Left pile group
0
Fill
Depth from ground surface (m)
10
15
20
25
30
35
40
45
50
Median wall / stanchion

Roof slab
Soft clay
Medium stiff clay

Very stiff
clay
Dense sand
Very stiff to
hardclay
Very dense sand
Right pile group
Base slab
10
15
20
8.0 m
12.5 m
Excavation
depth is 7.0 m
25
30
35
40
Diaphragm
wall 0.8 m
thick 25.0 m
deep
Barrette
0.8x2.7 m
29.0 m deep
Bored pile
diameter 0.8 m
42.0 m deep
45
50
(b)
FIG. 1. Site information (a) plan; (b) section.

NUMERICAL BACK-ANALYSIS
To assess the effects of excavation on the existing bored pile, numerical back-analysis
was carried out using finite element program PLAXIS2D 2015. Although threedimension numerical analysis can be used for soil-structure interaction problem, it is
more practical to perform plane strain analysis in design and calculation for deep
excavation works. To simplify bored pile in two-dimensional analysis, embedded beam
elements which allow load to be transferred to the surrounding soil and take the threedimensional effects into account were used.
ASCE
279
Mesh and boundary conditions

Figure 2 shows the finite element mesh used for back-analysis of the monitored
results. The dimension of the mesh was identical to that in the field. The boundary
conditions of the two vertical sides were roller support while hinge support was applied
to the bottom boundary. The diaphragm wall and base slab were modeled using plate
element. Bored pile and barrette were simulated using embedded beam row with spacing
shown in the figure. Fixed-end anchor was used for the roof slab.
Water table
= -3.0 m
Surcharge 10 kN/m2
-7.0 m
Column load
12,250 kN equivalent to 5,110 kN/m
Roof slab 0.5 m thick
Base slab 0.7 m thick
Diaphragm wall
0.8 m thick x
25 m deep
Bored pile diameter
0.8 m x 42 m deep at
spacing of 2.4 m
Depth of model 50.0 m

Width of excavation 18.0 m
Barrette size
0.8 m x 2.7 m x 29 m
deep at spacing of 4.5 m
Width of model 80.0 m
FIG. 2. Finite element mesh.

Constitutive model and input parameters
Hardening soil model (HSS) with small strain stiffness considered (Schanz et al.,
1999; Benz 2006) was adopted for numerical analysis. Unlike conventional elasticperfectly plastic soil model, HSS model is able to simulate soil stiffness dependency on
stress, strain and path. Input parameters for Bangkok soil were adopted and modified
from Surasak et al. (2012) and Likitlersuang et al. (2013). All parameters are
summarized in Table 1.
Parameters for reinforced concrete were assigned to diaphragm wall, roof slab, base
slab, barrette and bored pile with Youngs modulus and Poissons ratio of 27 GPa and
0.15, respectively.
ASCE
280
Table 1. Summary of input parameters

Soil Type
Medium
stiff clay
Soft clay
Medium
stiff clay
Stiff to very
stiff clay
Hard clay
Dense sand
Hard clay
Very dense
sand
c'
'
(kPa) ()
5.0 27
'
()
0
(kPa)
5,150
(kPa)
6,425
(kPa)
19,000
ur
0.7
(%)
Rf
0.9
1.0
0.74
0.2
0.0010
Gmax
(kPa)
16,000
1.0
27
800
850
8,000
0.9
1.0
0.74
0.2
0.0005
5000
5.0
27
5,150
6,425
19,000
0.9
1.0
0.74
0.2
0.0010
16,000
11.5
28
5,150
6,425
19,000
0.9
1.0
0.5
0.2
0.0010
48,000
11.5
28
9,500
12,000
30,000
0.9
1.0
0.5
0.2
0.0015
124,000
0.1
31
27,000
27,000
162,000
0.9
0.5
0.5
0.2
0.0140
200,000
11.5
28
9,500
12,000
30,000
0.9
1.0
0.5
0.2
0.0015
124,000
0.1
31
27,000
27,000
162,000
0.9
0.5
0.5
0.2
0.0140
200,000
Numerical modeling procedures

Basically, the numerical back-analysis procedures followed that in the field. Initial
stress condition was simulated according to K0 of each soil layer. In order to generate
existing load, a surcharge of 10 kN/m2 was imposed on each side of the underpass and
point load of 12,250 kN or equivalent to 5,110 kN/m (at a spacing of 2.4 m) was applied
on the center of each pile group. To allow dissipation of excess pore water pressure due
to surcharge and load from pile cap, consolidation analysis was carried out. After fully
dissipation, the soil inside the underpass to depth of 0.5 m deep was excavated, using
undrained effective stress analysis (Undrained A in PLAXIS). Afterward, the roof slab
was activated, followed by excavation to final level (7 m) in undrained condition. To
analyze time dependent behavior, consolidation for six months was carried out after base
slab was constructed.
INTERPRETATION OF RESULTS
Time dependent movement of diaphragm wall and bored pile
Figure 3a shows measured and computed lateral displacement of diaphragm wall. The
results are shown at the final excavation stage and 6 months after the final excavation.
The maximum measured horizontal movement of the wall at final excavation was 20.2
mm, equivalent to 0.29%H (i.e., H = 7 m). According to data collected by Phienwej et
al. (1995) and Wang et al. (2010), average lateral movements of retaining wall for topdown construction are 0.20%H and 0.27%H, respectively. One of possible reasons of
wall movement of this project larger than an average range for top-down construction is
the excavation length of cut-and-cover tunnel is long (i.e., similar to plane strain
conditions). Comparing with other basement excavation projects, three-dimensional
effects of basement excavation caused smaller lateral wall movement than plane strain
condition in this study. The maximum measured displacement at 6 months after reached
the final excavation became 24.4 mm or increased by 21%.
The computed horizontal wall displacements were in the same trend as the measured
one. The discrepancy between the two results maybe because the parameters adopted
from literature are from similar type of Bangkok soil, but not from the site.
ASCE
281
Computed lateral displacement of pile is shown in Figure 3b. As expected, the

displacement of the pile on the left side of underpass was 55% larger than that on the
right side. The maximum lateral pile displacement was 13 mm or 1.6% of pile diameter
(D). At 6 months after final excavation, there was a rebound of lateral displacement of
both pile groups. The maximum lateral movement of pile was still within the allowable
serviceability limit of 38 mm (ONeil and Reese, 1999).
To further investigate the time dependent behavior, excess pore water pressure
computed at the final excavation stage from soil elements along the diaphragm wall and
the two bored piles on each side of the underpass is shown in Figure 3c. At 6 months
after the final excavation, the excess pore water pressure was fully dissipated and the
results are not shown for clarity. The soil elements along the depth of diaphragm wall
were located inside the underpass. It can be seen that from depth of 7 m to 13 m, excess
pore water pressure up to -50 kPa was induced. This is because of unloading due to
excavation causing negative excess pore water pressure. When the excess pore water
pressure of soil located underneath the excavation dissipated at 6 months after the final
excavation, soil heave occurred. This heave caused increase in lateral movement of
diaphragm wall (see Fig. 3a). For soil elements along the shaft of bored piles, negative
excess pore water pressure was induced (Fig. 3c). After dissipation, there was a rebound
of soil movement outwards from the underpass due to swelling, resulting in rebound of
lateral pile displacement (refer to Fig. 3b).
20
30
5
10
15
15
20
20
Depth (m)
Depth (m)
10
25
25
30
30
35
35
40
40
45
Mea sured [Fina l exca va tion]

Mea sured [6 months a fter fina l exc]
Computed [Fina l exca va tion]
Computed [6 months a fter fina l exc]
(a)
45
10
20
-100
-50
0
Diaphragm wall
[Final excavation]
5
30
Left pile
[Final excavation]
10
Left pile
[6 months after]
Right pile
[Final excavation
Right pile
[6 months after]
50
Left pile
[Final
excavation]
15
Depth (m)
10
Excess pore water pressure
Pile lateral displacement (mm)
Wall lateral displacement (mm)

0
20
25
30
35
Right pile
[Final
excavation]
40
Computed results
Left pile [Fina l exca va tion]
Left pile [6 months a fter fina l exc]
Right pile [Fina l exca va tion]
Right pile [6 months a fter fina l exc]
(b)
45
Computed results
Dia phra gm wa ll [Fina l exca va tion]
Left pile [Fina l exca va tion]
Right pile [Fina l exca va tion]
(c)
FIG. 3. Time dependent behavior (a) Lateral displacement of diaphragm wall; (b)
pile horizontal movement; (c) excess pore water pressure.
Effects of excavation on bored pile
Due to excavation of the cut-and-cover tunnel, bending moment was induced in
adjacent bored pile as shown in Figure 4a. The results were computed from the pile
group located on the left side of the underpass at the final excavation stage. As expected,
the largest bending moment was induced on the front pile which was the closest to the
underpass. The maximum bending moment was 27 kN.m and the bending moment
extended up to depth of 33 m (about 5H).
ASCE
282
Figure 4b shows induced axial force in each pile due to excavation. Maximum
induced axial force of 130 kN occurred in the front pile at depth of 18.5 m. The reason
of increase in axial force in the bored pile is due to relative downward movement of soil,
causing negative skin friction. Due to excavation, the maximum total axial force (i.e.,
existing and induced axial force) in the pile became 2,200 kN.
To assess the effects of excavation on pile, allowable bending moment and axial
compressive force were estimated to be 400 kN.m and 4500 kN, respectively. The
summation of bending moment ratio (BMinduced/BMallowable) and axial force ratio
(Finduced/Fallowable) was 0.56, still within the allowable capacity of 1.
The effects of excavation caused stress relief due to removal of soil weight and
inward soil movement. This stress relief in the underpass resulted in horizontal stress
reduction in each pile as shown in Figure 4c. As a result, bending moment was induced
in pile (refer to Fig. 4a).
Induced pile bending moment (kN.m)
-25
25
Induced pile axial force (kN)

0
50
50
100
Incremental horizontal stress in soil (kPa)

150
-20
10
10
10
15
15
15
20
20
20
25
30
Depth (m)
Depth (m)
Depth (m)
-50
25
30
35
35
40
40
45
45
Computed [Front pile]
Computed [Middle pile]
Computed [Rea r pile]
10
20
30
40

25
35
45
-10

(c)
(b)
(a)
FIG. 4. Computed results of bored piles on the left of the underpass (a) induced
axial force; (b) induced axial force; (c) incremental horizontal stress in soil.
CONCLUSIONS
Based on field monitoring and numerical back-analysis results, the following
conclusions may be drawn.
(a) Maximum measured lateral displacement of diaphragm wall of the underpass was
0.29% of the excavation depth. Measured time dependent movement of the wall at 6
months after final excavation increased by 21% compared with that at the final
excavation stage. This increment was due to heave of soil inside the underpass
caused by dissipation of negative excess pore water pressure, resulting in further
lateral displacement of the diaphragm wall. The maximum computed lateral pile
head movement was 13 mm or 1.6% of pile diameter. This movement was still
under the serviceability limit given by ONeil and Reese (1999).
ASCE
(b) Induced bending moment and increase in axial force occurred in adjacent bored pile.
The reason of induced bending moment was reduction in horizontal stress along the
pile shaft. The increase in axial force was caused by relative downward soil
movement or negative skin friction.
(c) The extent of induced bending moment and axial force was about five times of the
excavation depth of the underpass, given that the distance between the pile and
diaphragm wall was within three times of the excavation depth. The induced
bending moment and axial force of the pile were still within the allowable limit.
(d) A permanent diaphragm wall coupled with effective design of associated structures
and suitable selection of construction method could offer a practically and
technically attractive solution in construction of vehicle underpass in the congestive
traffic area of Bangkok.
REFERENCES
Aye, Z. Z., Karki, D. and Schulz, C. (2006). Ground Movement Prediction and
Building Damage Risk-Assessment for the Deep Excavations and Tunneling Works
in Bangkok Subsoil. in the International Symposium on Underground Excavation
and Tunneling, Urban Tunnel Construction and Protection of Environment, EIT,
Bangkok, Thailand, pp. 281-297
Benz, T. (2007). Small-strain Stiffness of Soils and Its Numerical Consequences.
Volume 55 of Institut fr Geotechnik Stuttgart, Mitteilung. Inst. fr Geotechnik
Likitlersuang, S., Teachavorasinskun, S., Surarak, C., Oh, E. and Balasubramaniam, A.
(2013). Small Strain Stiffness and Stiffness Degradation Curve of Bangkok Clays.
Soils and Foundations. 53(4): 498509
ONeil, M. and Reese, L. C. (1999). Drilled Shafts: Construction Procedures and
Design Methods. Federal Highway Administration. FHWA-IF-99-25. Virginia
Phienwej, N., Akawanlop, K. and Balasubramaniam, A. (1995). Comparative
Evaluation of Ground Movements Associated with Braced-excavation in Bangkok
Soft Clay. in 10th Asian Regional Conference on Soil Mechanics and Foundation
Engineering, Beijing, 341-344
Schanz, T., Vermeer, P.A. and Bonnier, P.G. (1999) Formulation and verification of the
Harening soil model, in Beyond 2000 in computational Geotechnics, A.A. Balkema,
Roterdam, Netherlands, pp. 281-290
Surarak, C., Likitlersuang, S., Wanatowski, D., Balasubramaniam, A., Oh, E., Guan, H.,
(2012). Stiffness and strength parameters for hardening soil model of soft and stiff
Bangkok Clays. Soils and Foundations 52 (4), 682697.
Wang, J. H., Xu, Z. H. and Wang, W. D. (2010). Wall and ground movements due to
deep excavations in Shanghai soft soils. Journal of Geotechnical and
geoenvironmental Engineering, 136 (7), 985-994
ASCE
283

Asce 7690 9780784480038

Transféré par

Informations du document

Titre original

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

Asce 7690 9780784480038

Transféré par

Droits d'auteur :

Formats disponibles

Selected Papers from the Proceedings of the

Fourth Geo-China International Conference

James C.. Ni, Ph.D.., P.E.

Published by American Society of Civil Engineers

Geo-China 2016 GSP 260

Geo-China 2016 GSP 260

Geo-China 2016 GSP 260

Study on the Reasonable Cover Thickness of a Subsea Tunnel with the

Geo-China 2016 GSP 260

The Excavation Damage Zone of a Subway Tunnel Considering Stress

Geo-China 2016 GSP 260

Effectiveness of Buttress and Cross-Wall in Deep Excavations ......................... 267

Geo-China 2016 GSP 260

A Jet-Grouted Wall in Mitigating Tunnelling Effects on Adjacent Structures

Abstract: Technical measures, such as underground partition wall, could sever as

Geo-China 2016 GSP 260

2.1 Construction techniques

Geo-China 2016 GSP 260

East boun d tunnel

centreline of L-off-ramp of Xinzhong Road viaduct

clayey sandy grevel with pebble

Geo-China 2016 GSP 260

Clayey sandy gravel

Geo-China 2016 GSP 260

Geo-China 2016 GSP 260

Temporary panel house

Figure 3 Scheme of field measurements

Geo-China 2016 GSP 260

Horizontal displacement at C2 tube (mm)

Vertical depth below surface (m)

Horizontal displacement at C1 tube (mm)

Figure 4 Horizontal displacement at the two side of partition wall

Geo-China 2016 GSP 260

Distance from the tunnel centreline (m)

Figure 5 Surface settlement profiles

Distance of tunnel face to the monitoring section

Groundwater depth (m)

Shield passing the

Figure 6 Groundwater variation at tube C1 and C2 during tunnel passing

SUMMARY AND CONCLUSIONS

Geo-China 2016 GSP 260

Besides to redistribute the ground displacements of the structures to be less

Geo-China 2016 GSP 260

The Kriging Overlapping Method of Numerical Simulation for Complex

Geo-China 2016 GSP 260

Geo-China 2016 GSP 260

Let estimation variance minimal in unbiased conditions, that is

Conditional extreme value can be obtain by Lagrange multiplier method

where, is Lagrange multiplier.

Variogram is defined as:

Structural analysis of regionalized variables

Geo-China 2016 GSP 260

fitting can be performed for selecteed theoreticaal model. T

FIIG. 1. Modeel parameterr diagram

Geo-China 2016 GSP 260

Geo-China 2016 GSP 260

ideaal elastic-plaastic constitu

Geo-China 2016 GSP 260

Results comparrison of Krig

Geo-China 2016 GSP 260

Geo-China 2016 GSP 260

Geo-China 2016 GSP 260

Simple Method for the Safety Evaluation of Underground Pipelines Induced by