As 2015 Opt

With unparalleled underground expertise
and worldwide experience, aeCom is always

a trustworthy partner to deliver innovative
solutions to overcome complex geology
challenges faced by our clients.
nearly 100,000 employees serving clients

in more than 150 countries, we are a leader
in all the key markets we serve, including
transportation, facilities, environmental,
energy, water and government.
aeCom, recently joined by Urs, is a premier,

fully integrated support services firm. With
www.aecom.com
Worlds largest underground terminal aerial view of the

West Kowloon Terminus under the Hong Kong Section of the
Express Rail Link project
The HKIE Geotechnical Division 35th Annual Seminar, 2015

Soil-Structure Interaction: From Modelling to Observations
Designing m
ore
innovatively
The HKIE
Geotechni
cal
Division
35th
Annual
Seminar,
2015
SoilStructu
re
Interact
ion:
From
Modelli
ng to
Observa
tions
30
24
2421-25
Rm. 2421-25, 24/F., Sun Hung Kai Centre, 30 Harbour Road, Wanchai, Hong Kong
Tel: (852) 2511 9001 Fax: (852) 2580 0697
Distinctly Different
Lambeth brings together the diverse experience of operations, planning,

commercial and design professionals from within Gammon and Balfour Beatty
ensures it remains at the forefront of new techniques and new thinking. We set
the standard for both innovation and technological advancement in building, civil,
environmental, foundations, geotechnical and safety disciplines.
A member of the
Group
28/F Devon House, Taikoo Place, 979 Kings Road, Hong Kong Tel: 2516 8823 Fax: 2516 6352
Proceedings of the 35th Annual Seminar

Geotechnical Division, The Hong Kong Institution of Engineers
Soil-structure Interaction:
From Modelling to Observations
22 May 2015
Hong Kong
Jointly organised by:
Geotechnical Division, The Hong Kong Institution of Engineers
The Hong Kong Geotechnical Society
Supported by:
Civil & Structural Divisions, The Hong Kong Institution of Engineers
Captions of Figures on the Front Cover
Left Top to Bottom:
1)
Ground Movement Vectors below Proposed Grade Separated Road and the Predicted
Transverse Lining Deformation of a Nearby Tunnel in UK
(By courtesy of Jacobs China Ltd and Jacobs UK Ltd)
2)
Aerial View of Debris Flow Event above Yu Tung Road on Lantau Island
(By courtesy of GEO of HKSAR)
Right Top to Bottom:
3)
Excavation and Lateral Support Works for Underground Stormwater Storage Scheme in
Happy Valley
(By courtesy of DSD of HKSAR, ONLY Geotechnics Ltd and Black & Veatch Hong
Kong Ltd)
4)
Flexible Debris-resisting Barrier and the Simulated Debris-Barrier Interaction
(By courtesy of GEO of HKSAR, Department of Civil and Environmental Engineering of
HKUST and Ove Arup & Partners Hong Kong Ltd)
Soft copy of the proceedings can be downloaded from the HKIE Geotechnical Divisions
website http://hkieged.org/geodiv/annualseminar.aspx
The HKIE Geotechnical Division Annual Seminar 2015
ORGANISING COMMITTEE
Chairman
Ir Patrick Chao
Members
Ir Dr Johnny Cheuk
Ir Dr Julian Kwan
Ir David Lai
Ir Chris Lee
Ir Rupert Leung
Ir David Mak
Ir Clifford Phung
Ir Vincent Tam
Dr Ryan Yan
Ir Patrick Yong
Ir Irene Yu
Technical Sub-committee
Ir Dr Johnny Cheuk
Ir Dr Julian Kwan
Ir David Lai
Ir Chris Lee
Ir Dr S W Lee
Dr Andy Leung
Ir Clifford Phung
Ir Dr H W Sun
Dr Ryan Yan
Ir Patrick Yong
Any opinions, findings, conclusions or recommendations expressed in this material do

not reflect the views of the Hong Kong Institution of Engineers or the Hong Kong
Geotechnical Society
Published by:
Geotechnical Division
The Hong Kong Institution of Engineers
9/F., Island Beverley, 1 Great George Street, Causeway Bay, Hong Kong
Tel: 2895 4446
Fax: 2577 7791
Printed in Hong Kong
FOREWORD
Soil-structure Interaction: From Modelling to Observations is set as the theme of the 35th Annual
Seminar of the Geotechnical Division of the Hong Kong Institution of Engineers. The last time
soil-structure interaction was chosen as the theme of HKIE Geotechnical Divisions Annual Seminar
was in 1988. At that time, the industry was just starting to embrace the use of personal computers in
their day-to-day work. Progress in technology since then has allowed much more powerful tools at
our finger tips. Yet key to advancement in the understanding of soil-structure interaction behaviours
are the development of theories and constitutive models, design methodologies, and their verification
through testing and observations, which required the concerted effort of engineers both in the
academia and in professional practice.
This volume of proceedings contains a paper by Prof. H.G. Poulos, who is invited to deliver the
keynote lecture, and twenty papers contributed by both academics and practicing engineers. The
topics of the papers span across a wide spectrum, covering material behaviour, design methodologies
and case studies. While the classical soil-structure interaction problems of deep excavation,
foundation and tunneling are well represented, we have also seen the expansion of the study of
soil-structure interaction into a new area - the behaviour of landslide debris-resisting barriers.
Soil-structure interaction is a cross-discipline engineering problem, which requires engineers to have
a good understanding of behaviours of both structural systems and earth materials. Its role in the
design of civil engineering works cannot be overstated as essentially all civil engineering works are
in touch with the ground. I am delighted of contribution of papers from engineers of the civil and
structural disciplines. And I trust this Seminar will serve as an occasion for exchange of ideas among
fellow civil, structural and geotechnical engineers.
On behalf of the Geotechnical Division, I would like to thank the Hong Kong Geotechnical Society
for jointly organising this seminar, and HKIE Civil Division and Structural Division for their support.
I am grateful to our Guest-of Honour, Mr HUI Siu-wai JP, Director of Buildings of the Government
of the Hong Kong Special Administrative Region, and our Keynote Speaker, Prof. H.G. Poulos. I
would also like to express my gratitude to the authors and speakers contributing to this seminar and
the Organising Committee, under the leadership of Ir Patrick Chao, for their hard work in making
this seminar possible.
Ir Rupert Leung
Chairman, Geotechnical Division
Hong Kong Institution of Engineers (2014/2015 Session)
May 2015
ACKNOWLEDGEMENTS
The Organising Committee would like to acknowledge the support of Civil and
Structural Divisions of HKIE and the following sponsors for their generous support of
the Seminar:AECOM Asia Co Ltd.
Arup
China Geo-Engineering Corporation
Earth Products China Ltd.
Bachy Soletanche Group Limited
Dextra Pacific Ltd.
Hyder Consulting Ltd.
Mott MacDonald Hong Kong Ltd.
Tai Kam Construction Engineering Co Ltd.
C M Wong & Associates Ltd.
Lambeth Associates Limited
TABLE OF CONTENTS
Keynote Lecture
1
Soil-structure Interaction in Tall Building Foundation Design

H. G. Poulos
Page No.
1 - 26
Papers
2
The Role of Pile Structural Capacity in Pile Lateral Load Effect on

Slope Stability
B. K. L. Ieong, R. K. Y. Leung & R. N. N. Keung
27 - 35
Lateral Load Tests on Four Large Diameter Piles in Hong Kong

J. M. Shen
37 - 42
Subgrade Reaction vs Elastic Solutions Theoretical Considerations

J. M. Shen
43 - 46
Piled Raft Foundation Design for a Supertall Tower

J. W. C. Sze & A. K. M. Lam
47 - 56
Does 3D Finite Element Analysis Allow Better Design of Piles?

A Case Study for Temporary Marine Pile Design
Gavin S. H. Toh & Chao Li
57 - 64
Application of Modern Analytical Tools to a Four Decade Old

Metro Rail Station
S. P. Chin, A. K. L. Ng, L. J. Endicott & M. Ramanathan
65 - 71
Geotechnical Challenges in Construction of an Underground

Stormwater Storage Scheme in Happy Valley
W. H. Luk, C. L. Leung, C. Y. Lam, M. S. Hendy, S. S. Fang,
J. Premchitt & K. H. Lum
73 - 84
Recent Experiences of Numerical Prediction & Assessment

Excavation over a Tunnel of Unbolted Segmental Tunnel Lining
J. B. Wang, L. Swann & S. Reynolds
85 - 93
10
Outward Wall Movements Induced by Jet Grouting

L. W. Wong & R. N. Hwang
95 - 104
11
Effects of Soil-structure Interactions on the Design of ELS Systems

Albert T. Yeung & Iris C. Y. Ng
12
Numerical Modelling of Flexible Barriers Subject to Landslide 115 - 122

Debris Impact
C. Lam, R. C. H. Koo, Z. H. Zhou, Y. P. Liu & S. L. Chan
13
Effects of Positioning of Debris-resisting Baffles on Impediment of

Debris Flows
R. P. H. Law
14
Study of Interaction between Landslide Debris and Debris-resisting 133 - 139

Structures
J. S. H. Kwan, W. K. Pun, Y. K. Shiu, C. W. W. Ng, D. Song & J. Yiu
15
Performance Monitoring of a 50 m High 75 Cut Slope Reinforced 141 - 149

with Soil Nail Bars Made from Glass Fibre Reinforced Polymer
(GFRP) at Ho Man Tin Station
A. K. L. Kwong & J. S. S. Chim
16
Considerations for Consolidated Undrained Triaxial Testing of 151 - 160

Saprolites for Soil Structure Interaction Design
Ada Chan, A. D. Mackay & N. R. Wightman
17
Interface Modelling - A Comparison Study Using Finite Element

and Boundary Element Methods
H. C. Mark Chan & Chao Li
161 168
18
Evaluation of the Consolidation Behavior of a Settling Ground at

Reclaimed Ground of a Container Port in South China
S. L. Chiu, T. T. Fong & S.H. Yung
169 - 178
105 - 113
123 - 131
Keynote Lecture
Soil-structure Interaction in Tall Building Foundation Design

H.G. Poulos
Coffey Geotechnics, Australia & University of Sydney, Australia
ABSTRACT
This paper reviews some of the challenges that face designers of foundations for very tall
buildings, primarily from a geotechnical and a soil-structure interaction viewpoint. Some
characteristic features of such buildings are reviewed and then a three-stage process of foundation
design and verification is described. Design procedures for ultimate and serviceability limit state
design are outlined, and the importance of proper ground characterization and assessment of
geotechnical parameters is emphasized. Consideration is given to a number of soil-structure
interaction issues, including pile-soil-pile interaction, superstructure stiffness effects and the
effects of incorporating raft stiffness into the design analyses. The application of these effects is
illustrated via three high-rise projects.
1 INTRODUCTION
The last two decades have seen a remarkable increase in the rate of construction of super-tall buildings in
excess of 300m in height. Figure 1 shows the significant growth in the number of such buildings either
constructed (to 2010) or projected (2015 and beyond). A large number of these buildings are in the Middle
East or in China. Dubai has now the tallest building in the world, the Burj Khalifa, which is 828m in height,
while in Jeddah Saudi Arabia, the Kingdom Tower is currently under construction and will eventually exceed
1000m in height.
Super-tall buildings are presenting new challenges to engineers, particularly in relation to structural and
geotechnical design. Many of the traditional design methods cannot be applied with any confidence since they
require extrapolation well beyond the realms of prior experience, and accordingly, structural and geotechnical
designers are being forced to utilize more sophisticated methods of analysis and design. In particular,
geotechnical engineers involved in the design of foundations for super-tall buildings are leaving behind
empirical methods and are increasingly employing state-of-the art methods.
Figure 1: Total number of buildings in excess of 300m tall (after CTBUH, 2011)
This paper will summarize, relatively briefly, some of the challenges that face designers of foundations for
very tall buildings, primarily from a geotechnical viewpoint. Some characteristic features of such buildings
will be reviewed and then the options for foundation systems will be discussed. Some important aspects of
Keynote Lecture
soil-structure interaction will be described, and then some case histories will be presented to illustrate the
importance of considering soil-structure interaction in foundation design.
2 CHARACTERISTICS OF TALL BUILDINGS
There are a number of characteristics of tall buildings that can have a significant influence on foundation
design, including the following:
The building weight, and thus the vertical load to be supported by the foundation, can be substantial.
Moreover, the building weight increases non-linearly with height, and so both ultimate bearing
capacity and settlement need to be considered carefully.
High-rise buildings are often surrounded by low-rise podium structures which are subjected to much
smaller loadings. Thus, differential settlements between the high- and low-rise portions need to be
controlled.
The lateral forces imposed by wind loading, and the consequent moments on the foundation system,
can be very high. These moments can impose increased vertical loads on the foundation, especially on
the outer piles within the foundation system. The structural design of the piles needs to take account
of these increased loads that act in conjunction with the lateral forces and moments.
The wind-induced lateral loads and moments are cyclic in nature. Thus, consideration needs to be
given to the influence of cyclic vertical and lateral loading on the foundation system, as cyclic loading
has the potential to degrade foundation capacity and cause increased settlements.
Seismic action will induce additional lateral forces in the structure and also induce lateral motions in
the ground supporting the structure. Thus, additional lateral forces and moments can be induced in the
foundation system via two mechanisms:
o Inertial forces and moments developed by the lateral excitation of the structure;
o Kinematic forces and moments induced in the foundation piles by the action of ground
movements acting against the piles.
The wind-induced and seismically-induced loads are dynamic in nature, and as such, their potential to
give rise to resonance within the structure needs to be assessed. The risk of dynamic resonance
depends on a number of factors, including the predominant period of the dynamic loading, the natural
period of the structure, and the stiffness and damping of the foundation system.
3 FOUNDATION OPTIONS
The common foundation options include the following:
1.
2.
3.
4.
5.
Raft or mat foundations;

Compensated raft foundations;
Piled foundations;
Piled raft foundations;
Compensated piled raft foundations.
The majority of recent high rise buildings are founded on the latter three foundation types. In particular,
piled raft foundations have been used increasingly. Within a piled raft foundation, it may be possible for the
number of piles to be reduced significantly (as compared with a fully piled system) by considering the
contribution of the raft to the overall foundation capacity. In such cases, the piles provide the majority of the
foundation stiffness while the raft provides a reserve of load capacity. In situations where a raft foundation
alone might be used, but does not satisfy the design requirements (in particular the total and differential
settlement requirements), it may be possible to enhance the performance of the raft by the addition of piles. In
such cases, the use of a limited number of piles, strategically located, may improve both the ultimate load
capacity and the settlement and differential settlement performance of the raft and may allow the design
requirements to be met. It has also been found that the performance of a piled raft foundation can be
optimized by selecting suitable locations for the piles below the raft. In general, the piles should be
Keynote Lecture
concentrated in the most heavily loaded areas, while the number of piles can be reduced, or even eliminated,
in less heavily loaded areas (Horikoshi and Randolph, 1998).
4 THE DESIGN PROCESS
There are commonly three broad stages employed in foundation design:
1. A preliminary design stage, which provides an initial basis for the development of foundation
concepts and costing;
2. A detailed design stage, in which the selected foundation concept is analysed and progressive
refinements are made to the layout and details of the foundation system. This stage is desirably
undertaken collaboratively with the structural designer, as the structure and the foundation act as an
interactive system;
3. A final design phase, in which both the analysis and the parameters employed in the analysis are
finalized.
It should be noted that the geotechnical parameters used for each stage may change as more knowledge of
the ground conditions, and the results of in-situ and laboratory testing, become available. The parameters for
the final design stage should also incorporate the results of foundation load tests.
5 DESIGN ISSUES
The following issues will generally need to be addressed in the design of foundations for high-rise buildings:
1. Ultimate capacity of the foundation under vertical, lateral and moment loading combinations;
2. The influence of the cyclic nature of wind, earthquakes and wave loadings (if appropriate) on
foundation capacity and movements;
3. Overall settlements;
4. Differential settlements, both within the high-rise footprint, and between high-rise and low-rise areas;
5. Possible effects of externally-imposed ground movements on the foundation system, for example,
movements arising from excavations for pile caps or adjacent facilities;
6. Dynamic response of the structure-foundation system to wind-induced (and, if appropriate, wave)
forces;
7. Earthquake effects, including the response of the structure-foundation system to earthquake excitation,
and the possibility of liquefaction in the soil surrounding and/or supporting the foundation;
8. Structural design of the foundation system, including the load-sharing among the various components
of the system (for example, the piles and the supporting raft), and the distribution of loads within the
piles. For this, and most other components of design, it is essential that there be close cooperation and
interaction between the geotechnical designers and the structural designers.
The analyses required to examine the above design issues require consideration of soil-structure interaction
effects. Such analyses range from hand calculation methods for preliminary design to detailed three
dimensional finite element analyses for final design. However, the latter analyses should always be checked
for reasonableness by comparison with simpler methods and with available prior experience.
6 DESIGN ANALYSES
6.1 Ultimate limit state
There is an increasing trend for limit state design principles to be adopted in foundation design, for
example, in the Eurocode 7 requirements and those of the Australian Piling Code AS2159-2009. In terms of
limit state design using a load and resistance factor design approach (LRFD), the design criteria for the
ultimate limit state can be stated as follows:
Keynote Lecture
Rs* S*
(1)
Rg* S*
(2)
where
Rs* = design structural strength = s. Rus,

Rg* = design geotechnical strength = g. Rug,
Rus = ultimate structural strength,
Rug= ultimate strength (geotechnical capacity),
s = structural reduction factor,
g = reduction factor for geotechnical strength, and
S* = design action effect (factored load combinations).
The above criteria are applied to the entire foundation system, while the structural strength criterion
(equation 1) is also applied to each individual pile. It is not considered to be good practice to apply the
geotechnical criterion (equation 2) to each individual pile within the group, as this can lead to considerable
over-design. Rs* and Rg* can be obtained from the estimated ultimate structural and geotechnical capacities,
multiplied by appropriate reduction factors.
The structural and geotechnical reduction factors are often specified in national codes or standards. The
selection of suitable values of g requires engineering judgment and should take into account a number of
factors that may influence the foundation performance. As an example, the Australian Piling Code AS21592009 specifies an approach involving a subjective risk assessment, with lower values of g being associated
with greater levels of uncertainty and higher values being relevant when ground conditions are reasonably
well-known and a significant amount of load testing is to be carried out (see Section 6.6).
If any of the design requirements are not satisfied, then the design will need to be modified accordingly to
increase the strength of the overall system or of those components of the system that do not satisfy the criteria.
6.2 Cyclic loading
In addition to the normal design criteria, as expressed by equations 1 and 2, it is suggested that an additional
criterion be imposed for the whole foundation of a tall building to cope with the effects of repetitive loading
from wind and/or wave action, as follows:
Rgs* Sc*
(3)
where Rgs* = design geotechnical shaft capacity,

Sc* = maximum amplitude of wind loading,
= a reduction factor.
This criterion attempts to avoid the full mobilization of shaft friction along the piles, thus reducing the risk
that cyclic loading will lead to a degradation of shaft capacity. In most cases, it is suggested that can be
taken as 0.5, while Sc* can be obtained from computer analyses which give the cyclic component of load on
each pile, for various wind loading cases.
6.3 Serviceability
The design criteria for the serviceability limit state are as follows:
max all
(4)
max all
(5)
where max = maximum computed settlement of foundation,

all = allowable foundation settlement,
max = maximum computed local angular rotation,
Keynote Lecture
all = allowable angular rotation.

For the serviceability analysis, the best-estimate (unfactored) values of foundation resistances and
stiffnesses are employed and the serviceability limit state (SLS) loads are applied. The design will be
satisfactory if the computed deflections and rotations are within the specified allowable limits (equations 4
and 5).
Values of all and all depend on the nature of the structure and the supporting soil. Table 1 sets out some
suggested criteria from work reported by Zhang and Ng (2006). This table also includes values of intolerable
settlements and angular distortions. The figures quoted in Table 1 are for deep foundations, but Zhang and Ng
also consider allowable settlements and angular distortions for shallow foundations, different types of
structure, different soil types, and different building usage. Criteria specifically for very tall buildings do not
appear to have been set, but it should be noted that it may be unrealistic to impose very stringent criteria on
very tall buildings on clay deposits, as they may not be achievable. In addition, experience with tall buildings
in Frankfurt suggests that total settlements well in excess of 100mm can be tolerated without any apparent
impairment of function. It should also be noted that the allowable angular rotation, and the overall allowable
building tilt, reduce with increasing building height, both from a functional and a visual viewpoint.
Table 1: Suggested Serviceability Criteria for Structures (Zhang and Ng, 2006)
Quantity
Value
Limiting Tolerable Settlement

mm
Observed Intolerable Settlement
mm
Limiting Tolerable Angular
Distortion rad
Limiting Tolerable
Distortion rad
Angular
Comments
106
Based on 52 cases of deep foundations.
349
1/500
1/250 (H<24m)
From 2002 Chinese Code
H = building height
to
1/1000 (H>100m)
Observed Intolerable Angular

Distortion rad
1/125
6.4 Dynamic loading

Issues related to dynamic wind loading are generally dealt with by the structural engineer, with geotechnical
input being limited to an assessment of the stiffness and damping characteristics of the foundation system.
However, the following general principles of design can be applied to dynamic loadings:
The natural frequency of the foundation system should be greater than that of the structure it supports,
to avoid potential resonance phenomena. The natural frequency depends primarily on the stiffness of
the foundation system and its mass, although damping characteristics may also have some influence.
The amplitude of dynamic motions of the structure-foundation system should be within tolerable
limits. The amplitude will depend on the stiffness and damping characteristics of both the foundation
and the structure.
The dynamic stiffness and damping of the foundation system can be estimated via solutions and analyses
such as those provided by Gazetas (1983, 1991). For high-rise structures, the first natural period is relatively
large, and so dynamic effects on stiffness may be relatively small, while the radiation damping may also be
small, especially for rotational modes of excitation. Thus, damping may arise primarily from internal damping
of the soil. Higher modes of vibration may then become more significant.
Keynote Lecture
6.5 Earthquake loading

Soil deposits at a site subjected to an earthquake may experience the following effects:
Increases in pore pressure;
Time-dependent vertical ground movements during and after the earthquake;
Time-dependent lateral ground movements during and after the earthquake.
In foundation design, consideration must therefore be given to possible reductions in soil strength arising
from the build-up of excess pore pressures during and after the earthquake. In extreme cases, the generation of
pore pressures may lead to liquefaction in relatively loose sandy and silty soils.
As a consequence of the earthquake-induced ground movements, piles and other deep foundations will be
subjected to two sources of additional lateral loading:
a. Inertial loadings these are forces that are induced in the piles because of the accelerations generated
within the structure by the earthquake. Consideration is generally confined to lateral inertial forces
and moments, which are assumed to be applied at the pile heads;
b. Kinematic loadings these are forces and bending moments that are induced in the piles because of
the ground movements that result from the earthquake. Such movements will interact with the piles
and, because of the difference in stiffness of the piles and the moving soil, lateral stresses will be
developed between the pile and the soil, resulting in the development of shear forces and bending
moments in the piles. These actions will be time-dependent and need to be considered in the structural
design of the piles.
Thus, in addition to the usual design considerations for static loading, the above factors of strength
reduction, inertial loadings, and kinematic loadings, need to be incorporated into the design process.
When considering both the strength and stiffness of soils, consideration should also be given to the effects
of the rapid rate of loading that occurs during a seismic event. In addition to generating an undrained response
in finer grained soils, loading rate effects tend to increase both the strength and stiffness of such soils.
Appropriate assessment of the geotechnical parameters for earthquake response is a critical component of
geotechnical design for seismic actions, as it is for other types of imposed loadings. Reference can be made to
sources such as Kramer (1996) who discusses such issues as the effects of strain, cyclic loading and loading
rate effects on soil stiffness and damping.
6.6 Summary of design analysis process
A summary of the analyses that are recommended to be carried out for building foundation design are
summarized in Table 2. These analyses involve various combinations of factored/unfactored geotechnical
strengths and Ultimate Limit State (ULS) or Serviceability Limit State (SLS) loadings.
The assessment of the geotechnical reduction factor g is an important part of the design process.
Procedures are described in various codes and standards, for example Eurocode 7, and Standards Australia
Piling Code (AS2159-2009). Various attempts have also been made to rationalise the selection of g based on
probabilistic methods and the achievement of a target reliability index.
In practice, a series of factors need to be considered in making an assessment of g, including the
following in AS2159-2009:
The geological complexity of the site;

The extent of ground investigation;
The amount and quality of geotechnical data;
Experience with similar foundations in similar geological conditions;
The method of assessment of geotechnical parameters for design;
The design method adopted;
The method of utilizing the results of in-situ test data and pile installation data;
The level of construction control;
The level of performance monitoring of the supported structure during and after construction.
Keynote Lecture
g can typically range between 0.4 for conservative designs involving little or no pile testing and
uncertain ground conditions prevail to 0.8 for cases in which a significant amount of testing is carried out and
the ground conditions and design parameters have been carefully assessed.
Table 2: Summary of Design Analyses
Case
Purpose
Geotechnical
Design Capacity
ii
Structural
Design Capacity
iii
Serviceability
Factor applied
to Geotechnical
Strength
Parameters
Load Case
Comment
ULS
Geotechnical reduction factor, g, applied to

strength parameters to assess overall stability of
the pile group
1.0
ULS
Unfactored geotechnical strength parameters

are adopted to assess maximum pile axial load
and pile bending moment using short term pile
modulus.
1.0
SLS
Unfactored geotechnical strength parameters

are adopted to assess pile head deflections and
rotations
7 DESIGN ISSUES RELATED TO SOIL-STRUCTURE INTERACTION

7.1 Pile-Soil-Pile interaction
There is ample evidence of the fact that the behaviour of a pile within a group is influenced by the
surrounding piles, via transmission of stresses and strain through the supporting soil. The main effects of such
interaction are:
1. The settlement of a pile in a group is influenced by the settlement of other piles within the group;
2. Consequently, there is generally an increase in settlement due to interaction among the piles, and
therefore a reduction in stiffness of each pile within the pile group as compared with a single isolated
pile carrying the same average load.
3. The interaction among piles within the group may lead to a re-distribution of load within the piles,
and a change in the ratio of the load carried by shaft friction and by end bearing. In general, there will
be a tendency for the end bearing component to increase while the shaft load component decreases.
Various methods of analysis can be employed to incorporate such interaction effects. Some of the earlier
methods employed the concept of superposition of two-pile interaction factors, while more modern numerical
methods such as the finite element method automatically include such effects, depending on the soil model
adopted.
The consequences of including pile-soil-pile interaction into a design analysis will be discussed later in this
paper in relation to the Emirates Twin Towers project.
7.2 Stiffening effect of superstructure
It is common in geotechnical design to analyse a raft or piled raft without considering the stiffening effect of
any structure that is supported by the raft. Methods of incorporating the stiffness of a structure into a raft
analysis have been examined by several authors including Lee and Brown (1972), Poulos (1975) and Brown
and Yu (1986). Zhang and Small (1994) analysed 3-dimensional framed buildings on raft foundations, and
demonstrated that the larger the relative stiffness of the building frame, the smaller the differential deflections
in the raft. Such approaches can be extended to piled raft foundations.
Keynote Lecture
Gusmao Filho and Guimaraes (1997) have looked at construction sequence and have noted that the loads in
columns reach a maximum (or minimum) value as more stories are added to the building, leading to the idea
of the building reaching a limit stiffness.
It may be concluded that the stiffness of a structure will influence the calculated settlements and differential
settlements of a raft or piled raft foundation, but this depends on the stiffness of the structure relative to the
raft. For buildings with rigid shear walls, the stiffening effect on the raft will be significant. However, for
flexible light framed structures, the effect of the structure on a thick raft, will be small.
When undertaking a piled raft analysis, it may be convenient to represent the stiffness of the structure by
using thicker raft elements at locations where are walls and larger columns. While not providing any
information on the structural behaviour, such an approach can provide a more realistic assessment of
differential settlements within the footprint of a structure (Russo et al, 2013).
A convenient approach to foundation-structure interaction is for the piles to be represented by springs, the
stiffness of which are computed by the geotechnical engineer and which include the important effects of
interaction among the piles and the raft. Such interaction can significantly reduce the axial and lateral stiffness
of piles within a group, as compared with the values for an isolated single pile. In this way, a more reliable
analysis can be undertaken to compute not only the structural forces, but also the pile loads, the raft moments
and the distribution of settlement within the foundation system.
An example of the effects of incorporating superstructure stiffness into a geotechnical settlement analysis is
given later in this paper via the Burj Khalifa project.
7.3 The effects of raft flexibility, and the estimation of pile load distribution
In checking the structural loads within the piles in a piled raft system, it is essential to give proper
consideration to the flexibility of the raft. Making the common assumption that the raft is rigid can lead to
very misleading outcomes, as it tends to over-estimate the loads in the outer piles within the system. In
addition, consideration of the superstructure stiffness in a piled raft analysis can also have a significant
influence on the computed distribution of axial pile loads. As a consequence, the computed values of pile head
stiffness may also be affected. The case of the Incheon Tower in South Korea is an example of the importance
of incorporating the flexibility of the raft, and this is presented later in this paper.
7.4 Factoring of resistances
When considering soil-structure interaction to obtain foundation actions for structural design (for example, the
bending moments in the raft of a piled raft foundation system), the most critical response may not occur when
the pile and raft capacities are factored downwards. For example, at a pile location where there is not a
column, load acting, the negative moment may be larger if the pile capacity is factored up, rather than down.
For this reason, in the structural design of the raft and the piles, the results of the ULS overall stability
analysis are not considered to be relevant, because the loads that can be sustained by the piles would then be
artificially reduced by the geotechnical reduction factor. Consequently, it is suggested that the most rational
approach is one in which a separate ULS analysis is carried out using the various ULS load combinations but
in which the unfactored resistances of the foundation components are employed. The consequent computed
foundation actions (i.e. pile forces and, if appropriate, raft moments and shears) are then multiplied by a
structural action factor (for example 1.5) to obtain the values for structural design.
8 ASSESSMENT OF GEOTECHNICAL DESIGN PARAMETERS
8.1 Key parameters
For contemporary foundation systems that incorporate both piles and a raft, the following parameters require
assessment:
The ultimate skin friction for piles in the various strata along the pile;
The ultimate end bearing resistance for the founding stratum;
The ultimate lateral pile-soil pressure for the various strata along the piles;
The ultimate bearing capacity of the raft;
Keynote Lecture
The stiffness of the soil strata supporting the piles, in the vertical direction;
The stiffness of the soil strata supporting the piles, in the horizontal direction;
The stiffness of the soil strata supporting the raft.
It should be noted that the soil stiffness values are not unique values but will vary, depending on whether
long-term drained values are required (for long-term settlement estimates) or short-term undrained values are
required (for dynamic response to wind and seismic forces). For dynamic response of the structure-foundation
system, an estimate of the internal damping of the soil is also required, as it may provide the main source of
damping. Moreover, the soil stiffness values will generally vary with applied stress or strain level, and will
tend to decrease as either the stress or strain level increases.
8.2 Methods of parameter assessment
The following techniques are used for geotechnical parameter assessment:
1. Empirical correlations these are useful for preliminary design, and as a check on parameters
assessed from other methods.
2. Laboratory testing, including triaxial and stress path testing, resonant column testing, and constant
normal stiffness testing.
3. In-situ testing, including various forms of penetration testing, pressuremeter testing, dilatometer
testing, and geophysical testing.
4. Load testing, generally of pile foundations at or near prototype scale. For large diameter piles or for
barrettes, it is increasingly common to employ bi-directional testing to avoid the need for substantial
reaction systems.
8.3 Geophysical testing
Geophysical testing is becoming more widely used in geotechnical investigations. At least three major
advantages accrue by use of such methods:
1. Ground conditions between boreholes can be inferred.
2. Depths to bedrock or a firm bearing stratum can be estimated.
3. Shear wave velocities in the various layers within the ground profile can be measured, and
tomographic images developed to portray both vertical and lateral inhomogeneity.
4. From the measured shear wave velocity, vs, the small-strain shear modulus, Gmax, can be obtained as
follows:
Gmax = vs 2
(6)
where = mass density of soil.

For application to routine design, allowance must be made for the reduction in the shear modulus because of
the relatively large strain levels that are relevant to foundations under normal serviceability conditions. As an
example, Poulos et al (2001) have suggested the reduction factors shown in Figure 3 for the case where Gmax /
su = 500 (su = undrained shear strength). This figure indicates that:
The secant modulus for axial loading may be about 20-40% of the small-strain value for a practical
range of factors of safety;
The secant modulus for lateral loading is smaller than that for axial loading, typically by about 30%
for comparable factors of safety.
An important outcome of the strain-dependence of soil stiffness is that the operative soil modulus below the
foundation system will tend to increase with depth, even within a homogeneous soil mass. When modelling a
foundation system using a soil model that does not incorporate the stress- or straindependency of soil
stiffness, it is still possible to make approximate allowance for the increase in stiffness with increasing depth
below the foundation by using a modulus that increases with depth. From approximate calculations using the
Boussinesq theory to compute the distribution of vertical stress with depth below a loaded foundation, it is
Keynote Lecture
possible to derive a relationship between the ratio of the modulus to the small strain value, as a function of
relative depth and relative stress level. Such a relationship is shown in Figure 4 for a circular foundation, and
may be used as an approximate means of developing a more realistic ground model for foundation design
purposes. When applied to pile groups, the diameter can be taken as the equivalent diameter of the pile group,
and the depth is taken from the level of the pile tips.
0.7
G m ax
-------------= 500
su
0.6
Fahey & C arter (1 99 3),
(w ith FS= / m ax )
Gsec /Gmax
0.5
Ax ially loaded
pile
0.4
0.3
0.2
L aterally
loaded pile
Shallo w
Footing
0.1
0
4
3
2
1
Factor Safety A gainst U ndrain ed
Soil Failure
Figure 3: Example of ratio of secant shear modulus to small-strain value (Poulos et al, 2001)
Figure 4: Ratio of operative modulus to small-strain modulus versus relative depth

Note: p/pu = ratio of applied pressure to ultimate pressure.
Haberfield (2013) has demonstrated that, when allowance is made for strain level effects, modulus values
derived from geophysical tests can correlate well with those from pressuremeter tests. Figure 5 reproduces
such an example in which a reduction factor of 0.2 has been applied to the small-strain modulus values
10
Keynote Lecture
derived from cross-hole seismic test results. The modulus values so derived were found to be consistent with
values obtained from subsequent pile load tests.
Figure 5: Comparison of modulus values from pressuremeter and cross-hole seismic tests (Haberfield, 2013)
9 TYPICAL HIGH-RISE FOUNDATION SETTLEMENTS

It can be useful to review the settlement performance of some high-rise buildings in order to gain some
appreciation of the settlements that might be expected from two foundation types founded on various deposits.
Table 3 summarizes details of the foundation settlements of some tall structures founded on raft or piled raft
foundations, based on documented case histories in Hemsley (2000), Katzenbach et al (1998), and from the
authors own experiences. The average foundation width in these cases ranges from about 40m to 100m. The
results are presented in terms of the settlement per unit applied pressure, and it can be seen that this value
decreases as the stiffness of the founding material increases. Typically, these foundations have settled between
25 and 300mm/MPa. Some of the buildings supported by piled rafts in stiff Frankfurt clay have settled more
than 100mm, and despite this apparently excessive settlement, the performance of the structures appears to be
quite satisfactory. It may therefore be concluded that the tolerable settlement for tall structures can be well in
excess of the conventional design values of 50-65mm. A more critical issue for such structures may be overall
tilt, and differential settlement between the high-rise and low-rise portions of a project.
Table 3: Examples of Settlement of Tall Structure Foundations
Foundation Type
Founding
Condition
Location
No. of Cases
Settlement per Unit

Pressure mm/MPa
Raft
Stiff clay
Houston; Amman;
Riyadh
227-308
25-44
Stiff clay
Frankfurt
218-258
Dense sand
Berlin; Niigata
83-130
Weak Rock
Dubai
32-66
Limestone
Frankfurt
38
Limestone
Piled Raft
11
Keynote Lecture
10 THE EMIRATES TWIN TOWERS, DUBAI

10.1 The project
The Emirates Project is a twin tower development in Dubai, one of the United Arab Emirates. The towers are
triangular in plan, with a face dimension of approximately 50 m to 54 m. The taller Office Tower has 52
floors and rises 355 m above ground level, while the shorter Hotel Tower is 305 m tall. These towers are more
than double the height of the nearby World Trade Centre, which was once the tallest building in Dubai. The
twin towers are located on a site of approximately 200000 m2, which also incorporates low level retail and
parking podium areas. Poulos and Davids (2005) have presented details of the towers, the geotechnical
investigations, and the foundation design.
The main geotechnical investigation involved the drilling of 23 boreholes, to a maximum depth of about
80m. It was found that the stratigraphy was relatively uniform across the whole site, so that it was considered
adequate to characterize the site with a single geotechnical model. The groundwater level was relatively close
to the surface.
Because of the relatively good ground conditions near the surface, it was assessed that a piled raft system
would be appropriate for the foundation of each of the towers. The design of such a foundation system
required information on both the strength and stiffness of the ground, and as a consequence, a comprehensive
series of in-situ tests was carried out, together with both conventional and advanced laboratory testing.
The geotechnical model developed for foundation design under static loading conditions was based on the
relevant available in-situ and laboratory test data, and is shown in Fig. 6.
10.2 Load testing
As part of the foundation design process, a program of pile load testing was undertaken, the main purpose
being to assess the validity of the design assumptions and parameters. The test program involved the
installation of three test piles at or near the location of each of the two towers. All piles were drilled under
bentonite slurry support, with steel casing being provided in the upper 3-4m of each shaft. Because of the very
large design loads on the piles, it was not considered feasible to test full-size piles in compression, and as a
consequence, the maximum pile diameter for the pile load tests was 0.9m.
Figure 7 shows the test setup for the compression tests on 0.9 m diameter test piles. For the compression
tests, the loading was supplied by a series of jacks, while the reaction was provided by 22 anchors drilled into
the underlying calcisiltite. The anchors were connected to the test pile via two crowns (a larger one above a
smaller unit) located above the jacks and load cells. For the tension tests, the reaction was supplied by a pair
of reaction piles 12 m long, with a cross-beam connecting the heads of the test and reaction piles. In the lateral
load tests, the test pile was jacked against the adjacent 0.9m diameter compression test pile.
In order to provide some guidance on the expected behaviour of the piles during the test pile program,
Class A predictions of the load-deflection response of the test piles were carried out and communicated to
the main consultant prior to the commencement of testing. The geotechnical model was similar to that used
for design (Fig. 6), with some minor modifications to allow for the specific stratigraphic conditions at the test
pile locations, as revealed during installation of the test piles.
Comparisons between predicted and measured test pile behaviour were made after the results of the tests
were made available. Figure 8 compares the measured and predicted load-settlement curves for Test P3(H),
and reveals a fair measure of agreement in the early stages. The predicted settlements exceed the measured
values, and the maximum load of 30 MN reached exceeded the estimated ultimate load capacity of about 23
MN. The corresponding comparison for the Office Tower test pile P3(O), revealed excellent agreement, with
the predicted ultimate load capacity of 23 MN again being exceeded.
12
Keynote Lecture
Eu
MPa
E
MPa
fs
kPa
fb
MPa
pu
MPa
Unit
SILTY SAND, some

calcarenite bands
40
30
0.2
18
0.15
0.1
As above
125
100
0.2
73
1.5
1.5
CALCAREOUS
SANDSTONE
700
500
0.1
200
2.3
2.3
SILTY SAND
125
100
0.2
150
1.9
1.9
CALCISILTITE
500
400
0.2
450
2.7
2.7
As above
90
80
0.3
200
2.0
2.0
As above
700
600
0.3
450
2.7
2.7
10
20
Depth (m)
30
40
50
60
70
80
Figure 6: Geotechnical model adopted for design.
Working platform
(-0.50)
(-1.50)
Ground anchors
(-2.00)
(-5.00)
Unit 1 - Silty sand

900
(-10.0)
1285
203
0
(-16.0)
Unit 2 - Calcareous sandstone

Reference beams
(-20.0)
Footprint of the ground anchors

at the ground level
(-25.0)
No. 1 Extensometer
(-30.0)
Unit 3 - Silty sand
(-36.0)
(-40.0)
No. 4 Strain gauges
22 Nos of ground anchors
Unit 4 - Calcisiltite
Figure 7: Setup for compression pile tests.
13
Keynote Lecture
30000
Predicted
Measured
Applied Load (kN)
25000
20000
15000
10000
5000
0
10
20
30
Settlement (mm)
40
Figure 8: Predicted and measured load-settlement behaviour for pile P3(H).
10.3 Soil-Structure interaction aspects

In the final design, the piles were primarily 1.2 m diameter, and extended 40 or 45 m below the base of the
raft. In general, the piles were located directly below 4.5 m deep walls which spanned between the raft and the
Level 1 floor slab.
In additional to the conventional design analyses, more complete analyses of the settlement of the
foundation system were undertaken with the computer program GARP (Poulos, 1994). GARP (Geotechnical
Analysis of Raft with Piles) utilizes a simplified boundary element analysis to compute the behaviour of a
rectangular piled raft when subjected to applied vertical loading, moment loading, and free-field vertical soil
movements. The raft is represented by an elastic plate, the soil is modelled as a layered elastic continuum, and
the piles are represented by hyperbolic springs which can interact with each other and with the raft. Beneath
the raft, limiting values of contact pressure in compression and tension can be specified, so that some
allowance can be made for non-linear raft behaviour. In addition to GARP, the program DEFPIG (Poulos and
Davis, 1980) was used for the pile stiffness values and pile-pile interaction factors, and for computing the
lateral response of the piles.
For the analysis of settlements under the design loads, the same values of Youngs modulus were used as
for the single piles, whose behavior had been quite well-predicted. For the non-linear GARP analysis, the
unfactored values of estimated raft bearing capacity and ultimate pile load capacity were used.
Table 4 summarizes the computed maximum settlement and angular rotation under serviceability loading
conditions. These computed values were relatively large, but were nevertheless acceptable for the project. It
was found that the settlements showed a dishing pattern, with the settlements near the centre being
significantly greater than those near the edge of the foundation. Fig. 9 shows contours of the computed longterm settlement of the Hotel tower.
Table 4: Computed Maximum Settlement and Angular Rotation Serviceability Limit State
Tower
Max. Settlement mm
Max. Angular
Rotation
Office
134
1/384
Hotel
138
1/378
14
Keynote Lecture
55
50
105
45
1
0
115
40
115
5
10
125
25
125
105
30
115
y axis (m)
35
10
5
125
20
10
5
11
5
15
115
10
105
5
0
10
15
20 25 30
x axis (m)
35
40
45
50
Figure 9: Foundation layout and computed settlement contours for Hotel Tower.
The generally good agreement between measured and predicted performance of the test piles gave rise to
expectations of similar levels of agreement for the entire tower structure foundations. Unfortunately, this was
not the case. Measurements were available only for a limited period during the construction process, and these
are compared with the predicted time-settlement relationships in Fig. 10, for typical points within the Hotel
Tower. The time-settlement predictions were based on the predicted final settlement, an assumed rate of
construction, and a rate of settlement computed from three-dimensional consolidation theory.
Fig. 10 shows that the actual measured settlements were significantly smaller than those predicted, being
only about 25% of the predicted values after 10-12 months. Figure 11 shows the contours of measured
settlement at a particular time during construction for the hotel tower. Although the magnitude of the
measured settlements was far smaller than predicted, the distribution was similar to that predicted. Thus,
despite the considerable thickness of the raft and the apparent stiffness of the structure, the foundation
experienced a dishing distribution of settlement more characteristic of a flexible foundation.
Time (months)
1998
9 10 11 12
Settlement (mm)
T4
10
Measured
20
Predicted
T15
30
40
50
Figure 10: Predicted and measured settlement vs time Hotel Tower.
15
Keynote Lecture
T1
-7.0
T9
T3
T2
-6.5
-6.9
-7.4
-7
T16
-7
-6.3
T10
-8.3
T5
T4
-7
T13
-7.9
-6
T14
T11
-8.3
T15
-5.3
-7
T18
-8
-8
-8
-6.2
-8.7
-8.2
-7.2
T19
T20
-8.0
-7.5
-7
-6.0
T17
-7
T6
-6
T7
-7.4
-7.3
-5.8
T12
T8
-6.5
Figure 11: Measured settlement contours Hotel Tower.
10.4 Assessment of predictions - possible reasons for discrepancies.

The disappointing lack of agreement between measured and predicted settlement of the towers prompted a
post-mortem investigation of possible reasons for the poor predictions. At least four reasons were
suggested:
1.
2.
3.
4.
Some settlements may have occurred prior to the commencement of measurements;

The assumed time-load pattern may have differed from that assumed;
The rate of consolidation may have been much slower than predicted;
The interaction effects among the piles within the piled raft foundation may have been over-estimated.
Of these, based on the information available during construction, the first two did not seem to be likely, and
the last was considered to be the most likely cause. Calculations were therefore carried out to assess the
sensitivity of the predicted settlements to the assumptions made in deriving interaction factors for the piled
raft analysis with GARP. In deriving the interaction factors originally used, it had been assumed that the soil
or rock between the piles had the same stiffness as that around the pile, and that the rock below the pile tips
had a constant stiffness for a considerable depth. In reality, the ground between the piles is likely to be stiffer
than near the piles, because of the lower levels of strain, and the rock below the pile tips is likely to increase
significantly with depth, both because of the increasing level of overburden stress and the decreasing level of
strain. The program DEFPIG was therefore used to compute the interaction factors for a series of alternative
(but credible) assumptions regarding the distribution of stiffness both radially and with depth. The ratio of the
soil modulus between the piles to that near the piles was increased to 5, while the modulus of the material
below the pile tips was increased from the original 70 MPa to 600 MPa (the value assessed for the rock at
depth). The various cases are summarized in Table 2.
Figure 12 shows the computed relationships between interaction factor and spacing for a variety of
parameter assumptions. It can be seen that the original interaction curve used for the initial predictions lies
considerably above those for more realistic assumptions. Since the foundations analyzed contained many
piles, the potential for over-prediction of settlements is considerable, since small inaccuracies in the
interaction factors can translate to large errors in the predicted group settlement (for example, Poulos, 1993).
Revised settlement calculations, on the basis of these interaction factors, gave the results shown in Table 5.
The interaction factors used clearly have a great influence on the predicted foundation settlements, although
they have almost no effect on the load sharing between the raft and the piles. The maximum settlement, for
Case 4, is reduced to 29% of the value originally predicted, while the minimum settlement is only 25% of the
original value. If this case was used for the calculation of the settlements during construction, the settlement at
16
Keynote Lecture
Point T15 would be about 12 mm, which is in much closer agreement with the measured value of about 10
mm than the original predictions.
Curve No.
Modulus of
Layer below
MPa
Modulus of Soil
between Piles
to Near-Pile Values
1
2
3
4
5
90
90
200
700
700
1.0
5.0
5.0
5.0
1.0
0.4
1
Interaction Factor
0.3
2
0.2
5
3
0.1
0
1
10
s/d
20
50
100
Figure 12: Sensitivity of computed interaction factors to analysis assumptions.
The importance of proper assessment of the geotechnical model and the pile interaction factors in order to
compute the effects of group interaction has been emphasized by this case history. In addition, there was
probably an inadequate appreciation of the potential problems of applying interaction factors to large pile
groups, at the time of the predictions. In addition, in contrast to the single pile tests, the calculations were done
for the purposes of design, rather than prediction, and consequently tended to be conservative.
Case
Original calculations
Case 2
Case 3
Case 4
Case 5
Table 5: Summary of Revised Calculations for Hotel Tower

Modulus
Ratio of max.
Max.
Min.
below
to near-pile
Settlement
Settlement
53 m MPa
modulus
mm
mm
80
1
138
91
80
200
600
600
5
5
5
1
122
74
40
58
85
50
23
32
% Load on
Piles
93
93
92
92
92
11 THE BURJ KHALIFA, DUBAI

11.1 The project
The Burj Khalifa project in Dubai comprised the construction of a 160 storey high rise tower, with a podium
development around the base of the tower, including a 4-6 storey garage. The Burj Khalifa Tower (originally
denoted as the Burj Dubai prior to completion and opening) is currently the worlds tallest building at 828m. It
17
Keynote Lecture
is founded on a 3.7m thick raft supported on 196 bored piles, 1.5 m in diameter, extending approximately 47.5
m below the base of the raft. Figure 13 shows the completed tower.
The key challenges in this case were to undertake an economical foundation design for the worlds tallest
building, where the founding conditions were relatively weak rock and where significant wind loadings were
to be resisted. The foundation design was undertaken by Hyder Consulting UK, with peer review by Coffey
Geosciences. The final design involved the use of advanced three dimensional finite element analyses. A
detailed description of this case is given by Poulos and Bunce (2008).
The geotechnical investigation was carried out in four phases and involved the drilling of 33 boreholes,
with SPT testing, pressuremeter testing and geophysical testing being undertaken.
Figure 13: The Burj Khalifa
11.2 Load testing

Two programs of static load testing were undertaken for the Burj Khalifa project:
Static load tests on seven trial piles prior to foundation construction.

Static load tests on eight works piles, carried out during the foundation construction phase (i.e. on
about 1% of the total number of piles constructed).
In addition, dynamic pile testing was carried out on 10 of the works piles for the tower and 31 of the 750
piles for the podium, i.e. on about 5% of the total works piles. Sonic integrity testing was also carried out on a
number of the works piles.
Both the preliminary test piling program and the tests on the works piles provided very positive and
encouraging information on the capacity and stiffness of the piles. The measured pile head stiffness values
were well in excess of those predicted, and those expected on the basis of the experience with the nearby
Emirates Towers. The capacity of the piles also appeared to be in excess of that predicted, and none of the
tests appeared to have fully mobilized the available geotechnical resistance. The works piles performed even
better than the preliminary trial piles, and demonstrated almost linear load-settlement behaviour up to the
maximum test load of 1.5 times working load.
18
Keynote Lecture
11.3 Soil-Structure interaction aspects

The original settlement prediction carried out by the author employed two analyses, one using an in-house
program PIGS (Pile Group Settlement) and another using the commercially-available program FLAC (in axisymmetric mode). The maximum settlement predicted from each analysis was almost identical (74mm and
73mm respectively), and these values also agreed with the values derived from an ABAQUS analysis carried
out by the foundation designer, Hyder.
The measured maximum settlement towards the end of construction was around 43mm, indicating that the
final long-term settlement would probably be in the order of 55 5mm. Subsequently, a re-assessment of the
settlements was undertaken in an attempt to understand the possible reasons for the (relatively modest) overprediction of settlement (Russo et al, 2013). The majority of the foundation settlement re-assessment was
carried out using linear elastic analyses with the computer program NAPRA, developed in Italy by Viggiani
and Mandolini (1997). In developing the geotechnical model, lessons learned from the Emirates project were
employed to try and avoid gross inaccuracy in the predicted foundation settlements.
The mesh adopted for the NAPRA analyses is shown in Figure 14, and in this mesh, the columns were
spaced 1.7 m apart. Preliminary analyses indicated that using a finer mesh than this produced no change in the
results. The actual shape of the raft was modelled by adopting a piecewise approximation. The majority of
analyses were undertaken using the best-estimate modulus values derived from a careful interpretation of the
load test data, taking into account the effects of pile-soil-pile interaction between the test piles and the reaction
piles. Among the parameters investigated was the effect of the assumed loading pattern, and the effect of
incorporating the stiffness of the superstructure.
Fig.14: Model adopted in the NAPRA analyses
11.4 The influence of the loading pattern

The initial results were obtained by assuming that the average design load (32.21 MN) was applied to each
pile location. In reality, the loads will be applied via wall and column locations, and consequently, NAPRA
was used to examine the influence of the loading pattern on the computed settlement profile for two cases:
a. Equal loads on all the piles;
b. The actual design loadings are applied at the wall and column locations.
The computed settlement profiles along Wing C of the foundation system are shown in Figure 15, and
reveal a difference in the computed settlement patterns, with the equal load assumption giving smaller
maximum settlement than the other case. Thus, it would appear desirable to employ the actual load pattern in
design calculations.
19
Keynote Lecture
Figure 15: Influence of Assumed Loading Pattern on Computed Settlement Profile
In order to investigate the effect on the computed settlement and differential settlement, and to try and
obtain a more accurate estimate of the pattern of settlement via a geotechnical analysis, the stiffening effect
exerted by the superstructure on the raft was taken into account, in various ways, by increasing the bending
stiffness of the raft in each wing (estimated by the structural designers to be equivalent to an increase of 25200
kNm2 per wing). Six alternative methods of incorporating this increased bending stiffness were adopted:
a. Increasing the thickness of the whole raft to reflect the bending stiffness of the entire tower (Model 1);
b. Increasing the raft thickness over the central part of the wings and on the core, as shown in Figure 16,
to reflect the bending stiffness of the entire tower. This is denoted as Model 2;
c. Increasing the raft thickness only below the shear walls (see Figure 17), to reflect the bending
stiffness of the entire tower; this case is denoted as Model 3;
d. Model 1, with only 10% of the stiffness of the tower considered (Model 1M);
e. Model 2, with only 10% of the stiffness of the tower considered (Model 2M);
f. Model 3, with only 10% of the stiffness of the tower considered (Model 3M).
In each case, the actual pattern of loading via the columns and walls was applied, with only the dead load
component considered
Figure 18 compares the various computed profiles of settlement across the tower, together with those in
which no account is taken of superstructure stiffness. Also shown is the design profile developed by Poulos
and Bunce (2008), which was for combined dead plus live load, and therefore not directly comparable.
Clearly, there is a considerable difference between the extreme profiles, those taking all the superstructure
stiffness into account, and that in which no account is taken of the superstructure stiffness. It would appear
reasonable to assume that the profiles from Models 1M, 2M and 3M may be more reasonable approximations
to reality, and this appears to be borne out by the comparisons with the measured settlements, as described
below.
Figure 16: Raft model 2
Figure 17: Raft Model 3
20
Keynote Lecture
Figure 18: Comparison between Various Calculated Settlement Profiles
Detailed settlement measurements were only available up to February 2008, before the end of construction
and well before the commissioning and occupation of the building in January 2010. Nevertheless, anectodal
evidence indicated that the additional settlements between February 2008 and January 2010 were relatively
small, of the order of 1-2mm.
Figure 19 shows comparisons between the latest available measured profile of settlement in February 2008,
and the calculated settlement profiles from Models 1M, 2M and 3M. The following observations are made
from an examination of Figures 18 and 19:
1. Without allowance for superstructure stiffness, the calculated maximum final differential settlement is
about 23mm which is considerably larger than the measured value of about 14mm. The computed
maximum settlement is also larger than the measured value, although some additional settlement
would be expected after the building has been in operation for some years;
2. When allowance is made for the superstructure stiffness, the computed maximum settlement is similar
in magnitude to the measured value. However, for Models 1, 2 and 3, in which the full superstructure
stiffness is incorporated (albeit approximately), the computed settlement pattern differs somewhat
from the measured pattern, and the computed differential settlements are significantly smaller than
those measured. It seems clear that it is not appropriate to allow for the bending stiffness of the entire
structure when trying to modify the foundation stiffness;
3. When the allowance for superstructure stiffness is reduced by a factor of 10 (Models 1M, 2M and
3M), there is better agreement between the computed and measured profiles, with a computed
maximum differential settlement of about 22 mm, similar to the measured value. In this case, the
stiffness of the raft is approximately 53 times its original value, and this latter value is much larger
than the value of 10 times adopted by Hooper (1973) for the Hyde Park Barracks in London and by
Sales et al. (2010) for the Skyper Building in Frankfurt. Interestingly, and almost certainly
coincidentally, the profile for this case is rather similar to that obtained for the case when the average
load is imposed on each pile;
4. There remain some differences between the measured and computed settlement profiles in the vicinity
of the edge of the wing. There may well be scope to refine the method by which the superstructure is
incorporated into the geotechnical foundation analysis;
5. The calculated settlements from the design phase are considerably greater than those obtained from
the analyses in this paper. The main reason for these larger settlements is that the settlements were for
both dead and live load acting, and in addition, conservative values of Youngs modulus were used in
these analyses, with a somewhat different distribution of ground stiffness with depth being adopted in
those calculations. Once again, this comparison emphasises the importance of appropriate selection of
the ground stiffness values if accurate foundation settlement predictions are to be made.
21
Keynote Lecture
Figure 19: Measured and Computed Settlement Profiles along Wing C
The settlements measured during construction were consistent with, but comfortably smaller than, those
originally predicted (between 70 and 80mm), with a maximum settlement of about 44mm being measured.
Overall, the performance of the piled raft foundation system exceeded expectations.
As with other high-rise projects, the Burj Khalifa involved, as well as soil-structure interaction
considerations, close interaction between the structural and geotechnical designers in designing piled raft
foundations for the complex and significant high-rise structures. Such interaction has some major benefits in
avoiding over-simplification of geotechnical matters by the structural engineer, and over-simplification of
structural matters by the geotechnical engineer, thus promoting more effective and economical foundation and
structural designs.
12 INCHEON 151 TOWER, SOUTH KOREA
12.1 The project
A 151 storey super high-rise building project is currently under design, to be located in reclaimed land
constructed on soft marine clay in Songdo, South Korea. This building is illustrated in Figure 20, and is
described in detail by Badelow et al (2009) and Abdelrazaq et al (2011). The challenges in this case relate to a
very tall building, sensitive to differential settlements, to be constructed on a reclaimed site with very complex
geological conditions.
The site lies entirely within an area of reclamation, and comprises approximately 8m of loose sand and
sandy silt, over approximately 20m of soft to firm marine silty clay. These deposits are underlain by
approximately 2m of medium dense to dense silty sand, which overlie residual soil and a profile of weathered
(soft) rock.
The footprint of the tower was divided into eight zones which were considered to be representative of the
variation of ground conditions, and geotechnical models were developed for each zone. Appropriate
geotechnical parameters were selected for the various strata based on the available field and laboratory test
data, together with experience of similar soils on adjacent sites. One of the critical design issues for the tower
foundation was the performance of the soft silty clay under lateral and vertical loading, and hence careful
consideration was given to the selection of parameters for this stratum.
The foundation comprised a concrete raft 5.5m thick with 172 piles, 2.5m in diameter, with the number and
layout of piles and the pile size being obtained from a series of trial analyses through collaboration between
the geotechnical and structural designers. The piles were founded a minimum of 2 diameters into the soft
rock, or below a minimum toes level of El -50m, which was deeper.
22
Keynote Lecture
The use of a suite of commercially available and in-house computer programs allowed the detailed analysis
of the large group of piles to be undertaken, incorporating pile-soil-pile interaction effects, varying pile
lengths and varying ground conditions in the foundation design. During final design, an independent finite
element analysis was used to include the effect of soil-structure interaction and to include the impact of the
foundation system on the overall behaviour of the tower.
Figure 20: Incheon 151 Tower (artists impression)
12.2 Load testing

A total of five pile load tests were undertaken, four on vertically loaded piles via the Osterberg cell (O-cell)
procedure, and one on a laterally loaded pile jacked against one of the vertically loaded test piles. For the
vertical pile test, two levels of O-cells were installed in each pile, one at the pile tip and another at between the
weathered rock layer and the soft rock layer.
The vertical test piles were loaded up to a maximum one way load of 150MN in about 30 incremental
stages, in accordance with ASTM recommended procedures. The lateral pile load test was performed after
excavation of about 8m of the upper soil, to simulate a similar ground condition as for the tower foundation.
The lateral test pile was subjected to a maximum lateral load of 2.7MN.
The results of pile load tests indicated that the actual performance, under both vertical and lateral loads, was
superior to that predicted initially, thus providing scope for the development of a more cost-effective design.
Presently the tower site is fully reclaimed and fenced, and enabling works are being planned.
12.3
Soil-Structure interaction aspects
The overall settlement of the foundation system was estimated during all three stages of design, using the
available data at that stage, and relevant calculation techniques. The predicted settlements ranged from 75mm
from a simple equivalent pier analysis to 56mm from a PLAXIS 3D finite element analysis.
The detailed design analyses were carried out using an in-house computer program CLAP (Combined Load
Analysis of Piles) for the ultimate limit state load cases (ULS) and the program GARP (Small and Poulos,
2007) for serviceability (SLS) loadings. As part of the design process, estimates were required of the
maximum axial loads in each pile within the foundation system, and initially, the program CLAP was used.
CLAP implicitly assumes that the raft supporting the piles is rigid, and as a consequence, the computed axial
loads on some piles were found to be very large.
To investigate the effect of the rigid raft assumption, the foundation performance was re-assessed using
GARP and taking the flexibility of the raft into account. The serviceability load case (i.e dead and live loads)
was considered and the loads were applied at column and core locations.
23
Keynote Lecture
Table 6 presents a summary of foundation settlement, axial loads and stiffness on the corner, centre edge
and centre piles of the foundation (see Figure 21). The maximum predicted settlement occurs within the
heavily loaded core area, while the computed pile stiffness values are greatest for the outer piles. As the
analysis considered non-linear pile behaviour, therefore the higher stiffness (and hence larger loads) for the
outer piles degrade more rapidly under loading than the central piles.
Figure 21: Foundation Layout for Incheon Tower
Considering a rigid raft for the foundation, the total and differential settlement was predicted to be smaller,
with higher pile head loads for corner and centre-edge piles, thus resulting in higher vertical pile stiffness
values, especially on the outer piles when compared with those for a flexible raft.
When the flexibility of the raft was incorporated, the pile load distribution was found to be fairly uniform,
with slightly higher pile loads being predicted at the centre of the foundation where the heavily loaded core is
located. The loads on piles for a rigid raft case are approximately two times the loads for a flexible raft, except
for the centre piles.
Pile Load (MN)
Pile Stiffness (MN/m)

Raft Settlement (mm)
Table 6: Summary of foundation performance

Rigid Raft
Centre Pile
24
Centre Edge Pile
65
Corner Pile
85
Centre Pile
511
Centre Edge Pile
1418
Corner Pile
1604
Maximum
52
Minimum
26
Flexible Raft
49
33
43
726
932
1292
67
28
As the thickness of the raft (t) is 5.5m and the average centre-tocentre pile spacing (s) is approximately
5m, the ratio of t/s is greater than one, and the raft would be expected to behave as a semi-flexible raft (i.e.
behavior in-between a flexible and rigid raft) with the loads on the outer piles then being significantly less
than those for a perfectly rigid raft.
Based on the assessment, it is concluded that it is important to model the flexibility of the raft to avoid
having to design for unrealistically large loads in the outer piles within the group.
13 CONCLUSIONS
This paper has outlined a process for the design of high-rise building foundations and has identified a number
of soil-structure interaction aspects that need to be incorporated into this process. In particular, three aspects
have been discussed in detail:
24
Keynote Lecture
1. Pile-soil-pile interaction, i.e. the fact that a pile within a group environment is influenced by the loads
and deformations of the other piles within the group and also, within a piled raft foundation, by the
raft behaviour;
2. The stiffness of the superstructure and also the way in which the loads are represented in the analysis;
3. The flexibility of the raft.
The influence of these three aspects has been illustrated via three projects:
1. The Emirates twin towers in Dubai, for which over-simplification of the geotechnical model led to a
significant over-estimate of the pile-soil-pile interactions and a consequent significant over-estimate
of the foundation settlements;
2. The Burj Khalifa in Dubai currently the worlds tallest building, for which over-simplifying the
representation of the loads and ignoring the superstructure stiffness led to some inaccuracy of the
predicted distribution of settlement across the foundation;
3. The Incheon 151 Tower in Incheon, South Korea, for which the initial assumption of a rigid raft led to
very large estimated axial loads in some of the outer piles. Inclusion of the effects of raft flexibility
led to significantly smaller maximum pile loads and a more uniform distribution of pile loads.
It is clear that soil-structure interaction effects should be incorporated into the design of foundations for tall
buildings. Failure to do so can lead to inaccurate assessments of foundation settlements and pile loads. In
particular, failure to account for raft flexibility can lead to serious over-estimates of axial loads in some piles
and to a consequent over-conservative structural design.
The value of pile load testing, in conjunction with advanced methods of analysis and design, has been
emphasized, as has the importance of constructive interaction between the structural and geotechnical
designers.
ACKNOWLEDGEMENTS
The author gratefully acknowledges the contributions of Frances Badelow, Tristan McWilliam, Helen
Chow and Patrick Wong in relation to the analyses for the towers described in the paper. Grahame Bunce led
the foundation design for the Burj Khalifa. S.H. Kim, Ahmad Abdelrazaq and their teams in Korea had a
major involvement in the Incheon Tower foundation design. Professor John Small has been pivotal in
developing the GARP program and implementing it in a user-friendly form.
REFERENCES
Abdelrazaq, A., Poulos, H.G., Badelow, F. and Sung-Ho Kim (2011). Foundation Design and Pile Testing
Program for the 151 Story Incheon Tower in a Reclamation Area, Incheon, Korea. Paper TS21-02, Int.
Conf. Tall Buildings & Urban Habitat (CTBUH), Seoul, Korea.
Badelow, F., Kim, S., Poulos, H.G. and Abdelrazaq, A. (2009). Foundation design for a tall tower in a
reclamation area. Proc. 7th Int. Conf. Tall Buildings, Hong Kong, Ed. F.T.K. Au, Research Publishing,
815-823.
Brown, P.T. and K.R. Yu, 1986. Load sequence and structure-foundation-interaction. J. Struct. Eng., 112:
481-488.
CTBUH (2011). Council for Tall Buildings and the Urban Habitat, Press release, December 8, 2011.
Gazetas, G. (1983) Analysis of Machine Foundation Vibrations: State of the Art. Int. Jnl. Soil Dynamics &
Earthquake Eng., Vol. 2 (1): 2-42.
Gazetas, G. (1991). Foundation Vibrations. Chapter 16 of Foundation Engineering Handbook, 2nd Ed.,
Edited H-Y Fang. Chapman & Hall, New York.
Gusmao Filho, J.A. and Guimaraes, L.J. (1997). Limit Stiffness in Soil Structure Interaction in Buildings.
Proc. 14th Int. Conf. Soil Mechs. Found. Eng., Hamburg, Vol. 2, 807-808.
Haberfield, C.M. (2013). Tall Tower Foundations from concept to construction. Advances in Foundation
Engineering, Ed. K.K. Phoon et al, Research Publishing Services, 33-65.
Hemsley, J.A. (2000).Design Applications of Raft Foundations. Thomas Telford, London.
25
Keynote Lecture
Horikoshi, K. and Randolph, M.F. (1998). Optimum Design of Piled Rafts. Geotechnique 48(3): 301-317.
Katzenbach, R., Arslan, U. Moorman, C. and Reul, O. (1998). Piled Raft Foundations: Interaction Between
Piles and Raft. Darmstadt Geotechnics, Darmstadt University of Technology, 4: 279-296.
Lee, I.K. and Brown, P.T. Structure-Foundation Interaction Analysis. Journal of Structural Division,
American Society of Civil Engineers, Vol.98, No. 11, pp2413-2431.
Mandolini, A. and Viggiani, C. (1997). Settlement of Piled Foundations. Geotechnique, 47(3): 791-816.
Poulos, H.G. (1975) "Settlement Analysis of Structural Foundation Systems". Proc. 4th St.-East Asian Conf.
Soil Eng., Kuala Lumpur.
Poulos, H.G. (1993). Settlement Prediction for Bored Pile Groups. Bored and Auger Piles, BAPIII, Ed W.
van Impe, Balkema, Rotterdam, 103-118.
Poulos, H.G. (1994). "An Approximate Numerical Analysis of Pile-Raft Interaction". Int. Jnl. Num. Anal.
Methods in Geomechs., Vol. 18, 73-92.
Poulos, H.G. (2001). Piled Raft Foundations Design and Applications. Geotechnique, 51(2): 95-113.
Poulos, H.G. (1994). Settlement Prediction for Driven Piles and Pile Groups". Vert. and Horizl. Deformns. of
Foundns. and Embankments, Geotech. Spec. Publ. No. 40, ASCE, New York, Vol. 2, 1629-1649.
Poulos, H.G. and Bunce, G. (2008). Foundation design for the Burj Dubai the worlds tallest building.
Proc. 6th Int. Conf. Case Histories in Geot. Eng., Arlington, VA, Paper 1.47, CD volume.
Poulos, H.G., Carter, J.P. and Small, J.C. (2001). Foundations and Retaining Structures-Research and
Practice. Proc. 15th Int. Conf. Soil Mechs. Geot. Eng., Istanbul, 4: 2527-2606.
Poulos, H.G. and Davis, E.H. (1980). Pile Foundation Analysis and Design. John Wiley, New York.
Poulos, H.G., Small, J.C. and Chow, H.S.W. (2013). Foundation Design for High-Rise Towers in Karstic
Ground. ASCE GSP229, Foundation Engineering in the Face of Uncertainty, Ed. J.L. Withiam, K-K.
Phoon and M.H. Hussein, 720-731.
Randolph, M.F. (1994). Design Methods for Pile Groups and Piled Rafts. Proc. XIII Int. Conf. Soil Mechs.
& Found. Eng., New Delhi, Vol.5, 61-82.
Russo, G., Abagnara, V., Poulos, H.G. and Small, J.C. (2013). Re-Assessment of Foundation Settlements for
the Burj Khalifa, Dubai. Acta Geotecnica, 8(1): 3-15.
Small, J.C. and Poulos, H.G. (2007). A Method of Analysis of Piled Rafts. Proc. 10th Australia-New Zealand
Conf. Geomechs., Aust. Geomechs. Society, 1: 550-555.
Zhang, L. and Ng, A.M.Y.(2006). Limiting tolerable settlement and angular distortion for building
foundations. Geotech. Special Publication No. 170, Probabilistic Applications in Geotechnical
Engineering, ASCE (on CD Rom).
Zhang, and Small, J.C. (1994). Finite layer analysis of soil-raft-structure interaction. Proc. 13th Int. Conf.
Soil Mech. Foundn Engng, New Delhi. India. Jan. 1994. 2, 587-590. A. A. Balkema. Rotterdam. 1994.
26
The Role of Pile Structural Capacity in Pile Lateral Load Effect on

Slope Stability
B.K.L. Ieong, R.K.Y. Leung & R.N.N. Keung
Hyder Consulting Limited, Hong Kong
ABSTRACT
Piles under lateral load and embedded within a slope are often considered to have a load effect on
the slope and hence affect its stability. As a common design approach, soil-structure analysis with
piles modelled as beams and lateral soil resistance as elastic springs is undertaken to work out the
distribution of soil resistance along pile shafts. Limit state equilibrium analysis is then undertaken
for the slope to ensure that, with the lateral load action on the soil mass considered, the slope still
achieves the required factor of safety. The intrinsic inconsistence of this approach, however, is
that the forces from an elastic analysis are assumed to apply continuously into the plastic
deformation stage of the slope when the slope fails, i.e. the increase in resistance to the lateral load
applied to the pile group through internal forces of the piles (shear and, in case of raking piles, axial
forces) associated with increase in deflection has not been taken into account.
This paper serves, with the aid of numerical modelling of a typical problem of pile lateral load effect
on slope, to present a design approach taking into account the available soil resistance, without
jeopardizing slope stability, and the pile structural capacity. It also aims to demonstrate that,
though it is a common practice to provide pile sleeves to reduce pile lateral load effect on slopes, the
pile sleeves can actually be eliminated for the benefit of reducing pile deflection while not
sacrificing slope stability, provided the piles are designed properly.
1
INTRODUCTION
In Hong Kong, many buildings are constructed on slopes and most of these buildings are supported by piles.
The stability of the slopes under the effect of piles under lateral load needs to be assessed by the geotechnical
engineers. Limit-equilibrium methods are the most commonly used approach for analyzing the stability of
slopes. The popularity of limit equilibrium methods is primarily due to their relative simplicity, ready ability
to evaluate the sensitivity of stability to various input parameters, and the experience that geotechnical
engineers have acquired over the years in interpreting calculated factor of safety values. The factors of safety
given by the limit equilibrium analyses help geotechnical engineers to guard against uncertainties associated
with geological model, soil strength parameters and loadings, etc., and the possibility that identified failure
mechanisms differ from actual behavior. Also, the recommended factor of safety values for slopes generally
ensures that deformations are within acceptable range.
For calculating the minimum factor of safety of slopes in Hong Kong under the effect of laterally loaded
piles, it is a common practice to adopt limit equilibrium approach using of method of slides. Using this
approach, the distribution along the pile of lateral load transfer from the pile to the soil mass is first calculated.
The lateral load transfer distribution is then modeled as horizontal load, either in the form of a series of point
loads or line loads, acting in the soil mass in the limit equilibrium slope stability analysis programme. Loads
above any defined slip surface are taken onto account in calculating the factor of safety against failure of the
slip surface. By adopting this approach, the calculated minimum factor of safety of the slope is destined to
decrease with a pile under lateral load in place. This approach, however, is conservative since the increase in
structural resistance of the pile to the applied lateral load through bending and shear associated with increase
in deflection is not taken into account. As a result, either pile sleeving, aiming at minimizing the transfer of
lateral load from the pile to the slope, particularly at shallow depth, or raking piles are required to be adopted
in order to avoid rendering the factor of safety of the slope below the acceptable limit.
27
Advances in computer technology have pushed the finite element method (FEM) and other numerical
analysis approaches to the forefront of geotechnical practice. It can accommodate practically all kinds of
geometry and can model elasto-plastic behaviour typical of soil materials. FEM is increasingly being applied
to slope stability analysis. One of the most popular techniques for performing FEM slope analysis is the shear
strength reduction approach. The shear strength reduction approach is simple in concept - systematically
reducing the shear strength parameters of materials by a factor and carrying out FEM analysis of the slope until
deformations are unacceptably large or the analysis does not converge due to the formation of a global failure
mechanism. In this paper, the shear strength reduction method is adopted to assess the factor of safety of a
slope under the same lateral load for different pile types in order to offer some insight into the behaviour of
laterally loaded piles on slopes.
2
SHEAR STRENGTH REDUCTION (SSR) APPROACH
By definition, the factor of safety of a slope is the ratio of actual soil shear strength to the minimum shear
strength required to prevent failure, or the factor by which soil shear strength must be reduced to bring a slope
to the verge of failure. In the SSR finite element technique, elasto-plastic behavior, such as Mohr-Coulomb
model, is assumed for slope materials. The material shear strengths are progressively reduced until collapse
occurs. The SSR technique for slope stability analysis involves systematic use of finite element analysis to
determine a stress reduction factor (R) that brings a slope to the verge of failure. The shear strengths of all the
materials in a FE model of a slope are reduced by dividing them with R, starting from unity and then
progressively increased or decreased. FE analysis of this model is repeated until a critical R value that induces
instability is attained. A slope is considered unstable in the SSR technique when its FE model does not converge
to a solution (within a specified tolerance). The key advantage of SSR method over traditional
limit-equilibrium analysis for investigating slopes stability is that there is no need to pre-define potential slip
surfaces as the SSR technique automatically establishes the critical failure mechanism.
Typical to slope stability problems where the Mohr-Coulomb soil model is used, the SSR method involves
reducing the cohesion, c and friction angle, , of soil materials by a strength reduction factor, R, using
Equations (1) and (2) below until the slope fails quasi-naturally:
(1)
(2)
The highest value of the strength reduction factor that still satisfies equilibrium and is kinematically
acceptable is the Factor of Safety, FOS, of the slope:
(3)
Figure 1 extracted from Ng et al. (2001) shows a typical relationship between the strength reduction factor
and the displacement at a nodal point adjacent to the pile from an analysis using an elasto-plastic model. When
the strength parameters are factored down continuously, the stress state of the slope would arrive at the critical
point A, beyond which isolated plastic zones would merge and expand rapidly. Plastic deformations would
then start to accelerate upon a further increase in the strength reduction factor. This critical point A symbolizes
initiation of slope instability, whereas Point B signifies global failure of the slope and the strength reduction
factor corresponding to this point is taken as the slope stability factor of safety.
28
Figure 1 Identification of Slope Failure from Calculated Nodal Displacement, Ng et al. (2001)
STUDY MODEL
In this paper, a single pile-slope system in an idealized elastic, perfectly plastic soil is investigated using the
commonly used finite element program PLAXIS 2D. A 10-m high soil cut slope with slope angle of 35 is
modelled. A pile is assumed to be of 20 m long to installed at 2 m setback away from the slope crest line. An
inclined bedrock is assumed to be located at 20m below the pile head at ground level of the slope crest. For
simplicity, the slope is assumed to be dry. The slope-forming material is modelled using the Mohr-Coulomb
soil model and assumed to possess a Youngs Modulus E of 50MPa, typical of dense Completely Decomposed
Granite encountered locally in Hong Kong. Apparent cohesion c and friction angle of the soil are assumed
to be 5 kPa and 35 respectively. The pile material is assumed to be linearly elastic. In PLAXIS 2D, the SSR
method is called Phi/C Reduction and the strength reduction factor, R, takes the notion of Msf (the Total
Multiplier Summation).
The factor of safety of the slope before pile installation, determined by limit equilibrium method with the
analysis programme SLOPE/W, is 1.471. Using SSR with PLAXIS 2D, the factor of safety of the slope is
found to be 1.502, which is 2% from that giving by SLOPE/W, thus confirming the two approaches give
consistent results. The SLOPE/W analysis and PLAXIS 2D analysis are presented in Figure 2 and Figure 3
respectively.
Figure 2 SLOPE/W analysis for slope without pile
Figure 3 PLAXIS analysis for slope without pile
Four types of pile under an assumed lateral load of 200kN at pile head as below are considered for the
SSR analysis:
29
a) 1m dia. bored pile with 2 m rock socket;

b) 1m dia. bored pile with pile base on bedrock level;
c) Socketted H-pile (305 x 305 x 223 kg/m UBP in 610-mm dia. grouted hole with 5 m socket into rock;
and
d) Driven H-pile of 305 x 305 x 223 kg/m UBP terminating at bedrock level.
Sectional properties for the 3 types piles are presented in Table 1 below:
Table 1 Sectional properties of piles
EA(kN/m)
1m dia. bored pile (Grade 45 concrete)
20734511.5
Socketted H-pile
11679395.8
Driven H-pile
5822000.0
EI (kNm2/m)
1295907.0
247217.4
108033.0
For simplicity, the pile stiffness and lateral load are applied directly into the plain strain FE model without
consideration of 3D load spread effect. The pile is modelled as a continuous wall into-the-plane.
Figure
4 shows the finite element model of Case (a)
Figure 4 PLAXIS Model of Case (a)
30
RESULTS OF ANALYSES
The calculated factor of safety of the slope is around 1.5 for Case (a), as presented in Figure 5 below. Similar
result is given by analyses of the other cases.
Figure 5 Calculation of FoS of Slope under Case (a)
The predicted slope movement mechanism under Case (a) after SSR to FoS of 1.4 is presented in Figure 6
below.
Figure 6 Predicted Slope Movement Mechanism under Case (a) after SSR
Figures 7 and 8 show the pile bending moment diagram of Case (a) respectively before strength reduction
process, i.e. with R equals 1, and after strength reduction process, with R equals 1.5. Comparing the two
bending moment diagrams, the one with soil strength parameters reduced by a factor of 1.5 shows at a greater
31
depth below pile head a higher maximum bending moment, implying there is less transfer of lateral load to the
soil mass at shallower depth and more lateral load is carried down the pile shaft through shear in the pile.
Figure 7 Pile Bending Moment Diagram for Case (a) before SSR
Figure 8 Pile Bending Moment Diagram for Case (a) after SSR
A plot of strength reduction factor against the calculated lateral displacement of soil immediately in front
of the pile head under Case (a) is presented in Figure 9. It shows that as R approaches Rmax = 1.5,
displacements become very large, meaning that the slope is yielding.
32
Figure 9 - Plot of Strength Reduction Factor against the Calculated Lateral Displacement of Soil Immediately in front of
the Pile Head under Case (a)
The bending moment and shear force diagrams of the pile in Case (a) for R from 1.0 to 1.5 are shown in
Figure 10. The plots further demonstrate that with and increasing R, lateral load transfer from the pile to the
soil mass takes place at greater depth and the pile bending moment becomes greater.
Figure 10 - Plots of Pile Bending Moment and Shear Force in Case (a)
33
Table 2 summarises the pile head displacement and maximum pile bending moment from the FE model
based on design strength parameters (i.e. R=1.0) and the maximum pile bending moment from the FE model
when R=1.4 is applied for the four cases.
Table 2 Summary of FE Analyses Results for the Four Study Cases
Case
Pile Head Displ. for
Max M for R=1.0
R=1.0 (mm)
(kNm)
Case (a) Bored pile with rock socket
368/470*
45.7
Case (b) Bored pile on rock
46.4
365
Case (c) Socketted H-pile
62.5
291
Case (d) Driven H pile
86.0
284
*M at top of rock socket
Max M for R=1.4

(kNm)
777
758
703
700
For verification purpose, the bored pile in Case (a) is re-modelled as elasto-plastic element and the
required bending moment capacity of bored pile for Case (a) (i.e. 777 kNm) corresponding to that of FOS
=1.4 is inputted as strength properties. After SSR, the FoS is found to be 1.4 as presented in Figure 11 below
and which is consistent with the result presented in Figure 5.
Figure 11 Calculation of FoS for Slope under Case (a) with Moment and Axial Strength Properties as Input
OBSERVATIONS
From the analysis results as shown above, the following observations can be made:
a) The stability of a slope, i.e. its factor of safety, is not impaired by a pile installed in it and subjected to
lateral load, provided the pile has sufficient embedment to carry the load to deeper soil stratum and does
not fail structurally.
b) With progressive reduction of soil strength, the pile bending moment increases and hence the required
pile structural capacity. Hence, the pile can be designed to a lower structural capacity if a lower slope
stability FoS (i.e. R) is required to be achieved. The study model has a factor of safety of 1.5, if 1.4 is
adopted as the design factor of safety, then the pile is only required to be designed to resist the bending
moment corresponding to an R of 1.4. The soil resistance provided to the pile will have intrinsically a
factor of safety of 1.4.
c) Whilst the four cases studied show the slope stability is not impaired provided the pile has the adequate
structural and geotechnical capacity, the pile and slope movements differs according to the pile stiffness
in the four cases. This has to be taken into account in serviceability consideration.
34
SUGGESTED DESIGN APPROACH
It is suggested for the classical problem of pile under lateral load embedded within a slope, the following
approach based on the Shear Strength Reduction (SSR) Method be adopted:
a) Carry out slope stability analysis taking into account the pile and pile lateral load using SSR.
b) Check that the pile has sufficient embedment that Rmax is greater than the required FoS of the slope.
c) Check serviceability limits are complied with, e.g. that the pile head movement is within allowable
limit under the analysis of R=1.0. If not, the pile design has to be revised by, say, increasing pile
stiffness or reducing pile lateral load.
d) Carry out structural design for the pile based on the shear force and bending moment corresponding to
the R equaling to the required slope FoS, typically 1.4 in most cases of slope stability problems in
Hong Kong.
ACKNOWLEDGEMENTS
The authors would like to thank Mr. Ellis Lam, also of Hyder Consulting Limited for his assistance in the
numerical analyses described in this paper.
REFERENCES
PLAXIS 2D (2015) PLAXIS 2D Manual
Ng, C. W. W, Zhang, L.M. & Ho, K.K.S. (2001) Influence of laterally loaded sleeved piles and pile groups on
slope stability, Canadian Geotech. J., 38, 553-566.
35
36
Lateral Load Tests on Four Large Diameter Piles in Hong Kong

J.M. Shen
Geotech Engineering Ltd, Hong Kong
ABSTRACT
This paper describes the test arrangements, the test results and the associated analysis for lateral
load tests carried out on two pairs of large diameter (1.2m and 1.8m) bored piles during the
construction of the Extension of Hung Hom Railway Terminus in the 1980s. The top soil stratum
which dominated the pile behaviour during test is CDG fill in a typical reclaimation. One
important feature of the tests is that besides deflection, rotation was also measured. This permits
two parameters of soil stiffness to be back calculated instead of one, and the distribution of
subgrade reaction with depth can thus be investigated. The method of analysis is described and
observations on the test results are discussed. Conclusions are drawn and recommendations made.
1 INTRODUCTION
During the construction of the Extension of Hung Hom Railway Terminus in the 1980s under the management
of the then Railway Development Office, lateral load tests were carried out on two pairs of large diameter
bored piles (of 1.2m and 1.8m diameter respectively). The author was with the GEO then and was not directly
involved in the project, but played an advisory role through a colleague, Mr W.K. Lai. He was then a Civil
Engineering Graduate. He had been posted to the GEO for training under the author before being posted to the
Railway Development Office working on the project under his supervisor Ir Y.C. Wong. This paper describes
the site in general, the test arrangements, the test results obtained and the associated analysis by the author.
2 DESCRIPTION OF THE LATERAL LOAD TEST.
2.1 Geology of the site
Figure 1: Geological Section of the Site
37
Typical reclaimation geology existed in the site, consisting of CDG fill overlaying marine clay, alluvium, insitu CDG and finally bedrock, which was encountered at depths varying from 20m to over 40m. The thickness
of various strata also varied greatly over the site. Figure 1 shows a typical geological section of the test site.
The top 6 to 10m was CDG fill with SPT values in the range of 4 to10 (with one exception at 38, probably
caused by boulder). As the soil near the ground surface has dominant effect over the stiffness of the pile soil
system in this test, the results would reflect mainly the stiffness of the top fill layer.
2.2 Details of the piles
Piles in the project are either 1.2m or 1.8m diameter bored piles with 8mm steel casings left in place. For the
purpose of calculating the EI of the pile in this analysis, it is assumed that the casing and the pile act as
composite section without shear slip. The piles are spaced at about 8 diameters centre to centre, and the piles
under test are about 40m long.
2.3 Test arrangement
Two piles were pulled towards each other using steel strand and hydraulic jack as shown schematically in
Figure 2. The line of force was set at ground level. Vertical extension rods about 3 meters long were fixed at
the top of the test piles. Horizontal deflection of the extension rods were measured with dial gauges against
fixed datum at heights of 825 and 2825 mm above ground. As the extension rods are not under any load and
remained straight, the deflection and rotation at the pile head at ground level can be calculated by linear
extrapolation.
Figure 2: Test Arrangement
Loadings were applied at 100 kN increments until a load of 800 kN was reached for the 1.2m diameter
piles, and at 200 kN increments until 1400 kN is reached for the 1.8 diameter piles. Loads were then
38
decreased at the same intervals to zero. Only one cycle of loading was carried out. Both the deflection and the
rotation at the top of the 4 piles were measured for each load increment and decrement.
2.4 Test results
The deformations of the four piles at the end of each loading/unloading stage are presented in Figure 3
(deflection) and Figure 4 (rotation).
Figure 3: Load-deflection Relationship
Figure 4: Load-rotation Relationship
3 ANALYSIS OF THE TEST RESULTS

3.1 Theoretical model
The secant modulus approach was adopted to handle the highly non-linear load-deflection/rotation
relationship. Two popular model used for analysis of laterally loaded piles are the subgrade reaction approach
and the elastic solution approach. The subgrade reaction approach is simpler to use, and does not involve L/D
ratio of the pile. Complete solutions in table form were published, for example, in Sheriff (1974). Complete
elastic solutions were also published for piles of selected L/D ratio, for example, in Poulos and Davies (1974).
Test piles described in this paper has a L/D ratio close to 25, and this solution is used for analysis.
3.2 Basic approach
Normally only deflections are measured in lateral load tests on piles (sometimes as the only acceptance
criterion) and hence only one stiffness parameter can be back calculated to fit the observed result.
Assumptions are usually made on the distribution of horizontal subgrade reaction with depth so that it can be
defined with one parameter, being constant with depth for hard clays or soft rock, and increasing linearly with
depth with zero stiffness at the ground surface for sand and normally consolidated clays. Since two parameters
(rotation and deflection) are observed this time, it is possible to back calculate two soil stiffness parameters.
The first step is to suggest a soil stiffness model so that it can be defined with 2 parameters. There are many
ways to do this, depending on the sophistication of the analytical tools and quality of soil data available. For
example, one can assume the SPT N-value to reflect the spring stiffness through a scale factor (first
parameter) and a depth influence factor (second parameter) to be back calculated. To do this, the analytical
tool need to be able to assign spring stiffness along pile depth according to SPT values, and SPT data at the
pile location have to be available.
In this analysis, it is assumed that the coefficient of horizontal subgrade reaction varies linearly from a
value of K0 at the pile top to K1 at the pile bottom, where K0 and K1 are to be back-calculated for each load
increment from the rotation and deflection observed. This is a simple extension of the conventional approach,
39
and can be carried out with published solutions. To facilitate repeated analysis, dimensionless charts are
prepared. The test results are then superposed on the chart to read off the required parameters of stiffness.
Elastic solutions for piles with a L/B ratio of 25 were also included in the chart for comparison.
3.3 Dimensionless parameters
The following dimensionless parameters are adopted for analysis:
i)
Distribution Shape Factor = K0/K1 (subgrade reaction) or Es0/Es1 (elastic solution)
ii) System Rigidity KC = 12EI/(K1.B.L4) for subgrade reaction solution and KR = EI/(Es1L4) for elastic
solution
iii) Rotation Factor RF = /(L.)
iv) Deflection Factor DF = 12EI/(L.P) where P is the applied force.
Charts were prepared with DF and RF as x and y axis respectively. Contours of constant and constant
KC/KR obtained from theoretical solutions are then plotted on the chart. This is the master chart for analysis
of real test results. The test results shown in Figures 3 and 4 were then transformed into dimensionless form
RF and DF and superposed on to the chart, as shown in Figures 5 and 6. Only the loading results are
presented.
Figure 5: Chart with Results of 1.2m Piles Superposed
3.4 Results
As shown in Figure 5, the results of the 1.2m diameter piles followed closely the trend predicted by the
subgrade reaction model, except for points at 300 kN and 600 kN for one of the piles. The results then entered
the zone of elastic solutions at 700 and 800 kN. On the contrary, the results of the 1.8m diameter piles (Figure
6) did not follow the trend predicted by the model initially, the rotation being much smaller then predicted for
a given deflection. However, as deflection (and therefore loading) increased, the test results entered (at 600
kN for one pile and 1200 kN for the other) into the zone predicted by the subgrade reaction model. They
finally entered the zone predicted by the elastic solution model.
40
When the test results fall within the predicted zone, it is possible to back calculate the values of the two
parametersand KC/KR defining the distribution of subgrade reaction with depth. Three pairs of such results
are presented in Table 1.
Figure 6: Chart with Results of 1.8m Piles Superposed
Load P (kN)
K0 (MN/m3)
K1 (MN/m3)
m (MN/m4)
Table 1 Subgrade Reaction Back Calculated From Pile Test Results

1.2m pile
1.8m pile
200
400
800
600
800
31
25
8
46
43
170
114
127
1150
434
3.5
2.2
3.0
27.5
9.8
1000
22
87
1.6
3.5 Observations and discussions

Two observations, which are not related to the test results, are made on the theoretical model. The first
observation is that the = 0 contour of elastic solution lies very close to the= 1 contour of the subgrade
reaction model. If test results fall on this region (and they do), there may be heated arguments on whether
should be 0 or 1, depending on the model used. This at first sight may seem to be an inconsistency between
the two models, but it is actually not. Shen (2015) shows that for the case of a smooth rigid wall embedded in
a uniform elastic half space (and therefore = 1), the corresponding subgrade reaction distribution is close to
= 0. This is a similar situation.
The second observation relates to how pile interaction would affect the single pile solution. The pile
interaction factors (this time with opposing load) given in Poulos and Davies (1974, Figures 14.22 and 14.26)
were used to modify the single pile solution for KR = 10-5. As a result, the point B in Figures 5 and 6 are
shifted to B. Apparently this would not significantly affect the analysis
As for the test results, those of the 1.2m diameter piles fit in generally with the models, but not those of the
1.8m piles. In the initial (stiff) stage of the loading, the relative value of rotation and deflection is not
compatible with the prediction of models assumed, and values of K0 and K1 cannot be back calculated. This
41
observation applies to both piles, although more pronounced in one of them. The author cannot offer any
explanation, but is interested to know whether others have similar experience. When the loading exceeds
about 50% of the final load, the results start to fall into the zone predicted by the subgrade reaction model, and
values of K0 and K1 can then be obtained and found to be much higher than conventionally associated with
DG fill.
As can be seen from Table 1, there is a drastic decrease in soil stiffness at depth (K1) for the 1.8m pile and
the author tried to seek an explanation. The elastic curves of the pile at 600 kN and 1000 kN are plotted out
according to the subgrade reaction model. It is found that, although having a diameter of 1.8m, it actually
behaves as a flexible structure when it is 40m long. At 600 kN, the soil is very stiff and the elastic curve
consists of over 5 half waves. At 1000 kN, the soil is less stiff and the elastic curve consists of less than 4 half
waves. As a consequence, the soil at the bottom 40% of the pile has undergone stress reversal, even though
the loading was monotonic increasing. Since the soil on unloading would not provide the same support as
loading as assumed by the subgrade reaction model, there would be a drastic drop in soil support at depth, as
observed in Table 1. This explanation is, of course, based on the subgrade reaction model. When most of the
results do not fit in with the model, one may consider any explanation based on the model superfluous.
Other factor that may cause results to deviate from the model include soil variability near the ground
surface, such as a rigid inclusion in the form of a boulder in the fill near the pile.
4 CONCLUSIONS AND RECOMMENDATIONS
Lateral load tests and the associated analysis carried out on two pairs of large diameter bored piles have been
described in this paper. The measurement of both deflection and rotation enables 2 parameters on soil stiffness
to be back calculated instead of one, and permitted a more in depth analysis of the results. It may also expose
the deficiency of existing models, such as those shown in this paper. If only deflections were measured, one
would be happy to back calculate one soil stiffness parameter without knowing the deficiency of the model.
As the measurement of rotation is relatively simple and would enable more in-depth analysis, it is
recommended that it be done routinely for lateral load test on piles through the test specification, even if it is
not part of the acceptance criteria.
Information on the pile top only does not provide sufficient data to thoroughly analyse the situation and the
case in this paper is an example. For research projects, instrumentation should be provided to monitor the pile
behaviour at depth.
The method of analysis and the dimensionless charts presented in this paper can be used to analyze other
lateral load tests where rotation as well as deflection is measured.
Even 1.8 diameter piles would behave as flexible structures in CDG fill, if it is 40m long, and stress and
contact pressure reversal may occur at depth, even though the load is monotonic increasing.
Many test results, such as those shown in this paper, are hidden in project files and never made known to
the public. The author appeals to project engineers to publish them.
ACKNOWLEDGEMENTS
The author is grateful to the project team of the Railway Development Office for making the test results
available, without which this paper could not have been written. Opinions expressed in this paper are those of
the author only.
REFERENCES
Poulos, H.G. & Davis, E.H. 1974. Elastic solutions for soil and rock mechanics. John Wiley & Sons.
Shen, J.M. 2015. Subgrade reaction vs elastic solutions theoretical considerations. Proceedings of the 35th
HKIE Geotechnical Division Geotechnical Division Annual Seminar, May, Hong Kong, pp.
Sherif, A. 1974. Elatically fixed structures. Earnest & Sohn, Berlin.
42
Subgrade Reaction vs Elastic Solutions

Theoretical Considerations
J.M. Shen
Geotech Engineering Ltd, Hong Kong
ABSTRACT
In this paper comparisons are made between the subgrade reaction approach and the elastic
solution approach in solving soil structure interaction problems. Based on theoretical solutions
obtained from theory of elasticity, inferences are made on the magnitude and distribution of
subgrade reaction which would render the two approaches compatible. Both vertical subgrade
reaction and horizontal subgrade reaction problems are studied and cases with particular interest
(such as Gibson soil) are highlighted. It is concluded that if educated assumptions on the
magnitude and distribution of the subgrade reaction is made according to the size, shape and
rigidity of the structure, as well as the nature of the soil, realistic solutions similar to those in
elastic solutions could be obtained with the simple subgrade reaction approach.
1 INTRODUCTION
Subgrade Reaction and Elastic Solution are two popular tools for handling soil structure interaction problems.
Those in favour of one approach often criticize the other. This paper tries to point out, from a practical point
of view based on theoretical solutions, the similarities rather than the differences between the two approaches,
hoping to bridge the gap between the two sides. The reader is assumed to be familiar with the fundamental
assumptions of the two approaches, which are mentioned only very briefly below. Also the term subgrade
reaction is used interchangeably with the term coefficient of subgrade reaction to facilitate reading.
In the subgrade reaction approach, the soil reactions are modeled as independent springs. This makes the
formulation of the problem as well as its solution relatively simple. Soil variability can be handled readily as
different spring stiffness can be assigned to different soils without affecting one another. However, the
stiffness of these springs are not true soil properties, and depend on the size and rigidity of the structure. On
the other hand, soils are modeled as coupled continuum in elastic solutions in theory of elasticity, in which the
Youngs modulus Es and Poissons ratio are true soil properties independent of the structure. This is
obviously a great advantage over the subgrade reaction approach. The costs of these advantages are more
complicated problem formulation and solution, and less versatility in handling soil variability.
Terzaghi (1955) provided classical guidelines on the evaluation of coefficient of subgrade reaction, usually
in terms of semi-empirical formulae based on his vast experience and in-depth analysis. He pointed out, using
the zone of influence of the stress bulk, the effect of foundation size on clay is different from that in sand, due
to the increase soil stiffness with depth for sand, and gave empirical formula for the effect in sand. Many
elastic solutions have emerged since that time, providing further theoretical bases for the evaluation of the
correct sub-grade reaction. Some are discussed in this paper.
2 VERTICAL SUBGRADE REACTION
2.1 Particular cases of similarity
There are at least two particular cases where the two approaches are compatible. The first case is the elastic
solution for a surface load on a semi-infinite mass of Gibson soil (one with modulus increasing linearly with
depth) with Poisson ratio = 0.5 and zero Youngs modulus Es at the surface. In this case the two approaches
are entirely compatible for any size and shape of vertical surface load, with the value of the vertical subgrade
reaction kv given by:
kv = 2m/3
(1)
43
where m is the rate of increase in Youngs modulus Es with depth.

The second case involves a point load on an infinitely long structural strip with bending stiffness EpIp
resting on a finite layer of uniform soil with Youngs modulus Es (situation often used to simulate a railway
track under the wheel load of the train). Vesic (1961) showed that the two approaches are practically
equivalent (some numerical approximations involved) provided that the value of subgrade reaction is
calculated with the following formula:
kv.B = 0.65 Es.12(Es.B4/EpIp)/(1-2)
(2)
where B is the breadth of the structural strip.

This time kv is affected by the size and rigidity of the structural strip.
2.2 General Cases
Information in general can be obtained by considering the two limiting cases of loading stiffness on a soil :
flexible loading (constant pressure independent of deformation) and rigid loading (imposed deformation
irrespective of soil reaction). Actual loading would lie between these two limits. If a uniform kv is assumed for
the design of a flexible loading such as a large water tank, it will result in uniform settlement without any
bending on the floor slab, while in reality the centre of the tank will settle more, leading to bending. This is
the usual criticism on the subgrade reaction approach. The settlement profile predicted by elastic solutions for
a flexible load will give indication on the correct distribution of subgrade reaction that would make the two
approaches compatible. On the other limit, the contact pressure distribution of a rigid loading would reflect
the value of subgrade reaction that need to be used to make the two approaches compatible, based on the
definition of subgrade reaction.
The case of a circular raft resting on a semi-infinite mass under a uniform load will be used for illustration.
The system rigidity in this case is defined by a dimensionless parameter K given by
K = (Ep/Es)(t/a)3(1 -2)
(3)
Where t and a are thickness and radius of the raft respectively.

Figures 1 to 3 below are reproduced respectively from Figures 12.22, 12.26 and 12.23 of Poulos & Davis
(1974). Figure 1 gives the settlement at the centre; Figure 2 gives the maximum differential settlement
(between centre and edge) and Figure 3 gives the contact pressure for a range of values of K. From these
figures the following observations can be made:
i)
ii)
iii)
iv)
v)
Practically K = 10-2 can be considered as flexible load and K = 102 rigid load (figure 1).
The centre of a flexible raft settles 27% more than that of a rigid raft under the same load (figure 1).
For flexible load, the settlement at the edge is 0.32 of that at the centre (figure 2), implying the
subgrade reaction at the edge should be about 3 times that at the centre in order to produce such result.
There is good consistency of contact pressure at 0.8 radius despite the large range of K. This contact
pressure is 83% of the applied pressure (figure 3).
Poissons ratio has little effect on differential settlement (figure 2).
From the elastic solution for a rigid loading, the contact pressure is given by
z = pav/{2. (1 - r2/a2)}
(4)
where z is the contact stress, pav the average applied pressure, a the radius and r the distance from the centre.
The contact stress at the centre is 50% of the average contact stress and the contact stress at the edge is
theoretically at infinity. The contact stress is 3 times that at the centre at a distance of 94.3% of the radius
44
from the centre (or 5.7% from the edge). It seems surprising that an edge/centre ratio of 3 for spring stiffness
would seem appropriate for both a flexible and a rigid loading.
For intermediate values of K shown in Figure 3, both the settlement and the contact pressure have to be
evaluated in order to arrive at an edge/centre spring stiffness ratio. Using the results of Figures 2 and 3,
calculation revealed a ratio of 2.7 (at 5% radius from edge) for K = 1. This is close to 3 found for the two
extremes. Absolute value of spring stiffness at the centre can be evaluated to complete the picture of subgrade
reaction distribution.
Figure 1: Central Vertical Displacement
Figure 2: Differential Settlement
Figure 3: Contact Pressure
2.3 Observations
The above paragraphs indicate that provided the value of subgrade reaction is chosen properly, the two
approaches can be highly compatible. For the circular raft, the edge/centre spring stiffness ratio is surprisingly
consistent for a wide range of K. The case of Gibson soil shown in paragraph 2.2, although academic and
unrealistic in real life, is particularly interesting in that it demonstrates complete compatibility between these
two concepts is possible.
3 HORIZONTAL SUBGRADE REACTION
3.1 Distribution with depth
In the case of vertical subgrade reaction, one is not directly involved in distribution of stiffness with depth as
only surface loading is involved, although it has effect indirectly on the soil stiffness or subgrade reaction as
shown in paragraph 2.1 above, or pointed out in Terzaghi (1955). However, when horizontal subgrade is
considered, one need to know the distribution of subgrade reaction with depth, as it has profound influence on
the results. For piles, where the critical dimension is the breadth B, it is conventionally assumed that the
stiffness is constant with depth in stiff clays and increasing linearly with depth for sands. For walls where the
critical dimension is the height L, it is conventionally assumed as suggested in Terzaghi (1955) that the
distribution of subgrade reaction is increasing linearly with depth, somewhat like earth pressure. One can
apply the rigid loading approach described in paragraph 2.2 to see what would be the correct distribution
with depth according to elastic solutions in these cases.
3.2 Rigid loading approach
Various elastic solutions have been collected and published in Poulos & Davis (1974). Solutions for
practically rigid piles, plates and wall (mainly defined by the L/B ratio) are selected. By superposing the right
combination of force and moment at the top of the rigid structure, the structure can be made to cast a uniform
45
displacement into the soil, and the resultant pressure distribution would represent the subgrade reaction, by
definition. Such solutions were worked out for a pile (L/B = 25), a plate (L/B = 2/3) and a wall (L/B = 0)
embedded in a semi-infinite mass. These are shown in Figures 4, 5 and 6 respectively.
Figure 4: Pile (L/B = 25)
Figure 5: Plate (L/B = 2/3)
Figure 6: Wall (L/B = 0)
3.3 Observations
From the above analysis and Figures 4 to 6, it is seen that for a uniform soil, the distribution of subgrade
reaction with depth is approximately constant with depth for a pile (with slight deviation at the two ends) and
increasing linearly with depth like earth pressure for a wall (Kh.L = 1.8 Es is a good approximation over top
70% of the wall height). A rigid plate is somewhere in between, with a finite value at the top and increasing
linearly with depth. Thus although the soil is uniform, the subgrade reaction is not necessarily uniform, but
depend on the shape of the rigid structure. If the soil has inherent trend of stiffness increasing with depth
(most soils do), the subgrade reaction may be expected to further increase with depth accordingly.
4 CONCLUSIONS
Subgrade Reaction and Elastic Solution are two popular tools for handling soil structure interaction problems.
Some comparisons between the two approaches are made in this paper. Elastic solutions are used to work out
the correct distribution and magnitude of the subgrade reaction to render the two approaches compatible. By
making an educated assumption on the distribution of subgrade reaction, one can then make use of the
simplicity of the subgrade reaction approach to solve the problem, without deviating too far from the more
sophisticated elastic solution approach.
REFERENCES
Poulos, H.G. & Davis, E.H. 1974. Elastic solutions for soil and rock mechanics. John Wiley & Sons, New
York.
Terzaghi, K. 1955. Evaluation of coefficient of subgrade reaction. Geotechnique, 5(4): 297-326.
Vesic, A.S. 1961. Bending of beams resting on isotropic elastic solid. Journal of Engineering Mechanics
Division, ASCE, 87(2): 35-53.
46
Piled Raft Foundation Design for a Supertall Tower

J.W.C. Sze and A.K.M. Lam
Ove Arup and Partners Hong Kong Limited
ABSTRACT
High-rise or super high-rise buildings usually accompanied by a deep basement to accommodate
car park spaces and plant rooms. With favorable ground condition, the basement may already be
rested onto rock and a raft foundation can simply be adopted. In contrast, some project sites may
be located in alluvial plain area where excessive weak deposits are encountered. In such case, pile
foundation is demanded and common pile design process is followed which may not require
significant geotechnical input.
A special case could be that the underlying soil or rock is marginally competent to support the
tower load or to assure that the movement performance of the tower can be met. In such situation,
a piled raft foundation might be the best option to pursue in order to achieve cost effectiveness.
However, the design of a piled raft foundation is somewhat different from the traditional way of
designing building foundation as it involves soil-structural interactions amongst the raft, piles,
soil and/or rock.
This Paper uses a supertall tower in South Korea resting on soft rock with complex geological
setting as a case history to illustrate the geotechnical design process of a piled raft foundation.
The specialties are attributed by the way in establishing the possible geological models, three
dimensional soil-structural interaction analyses and non-linear stiffness were adopted in the
analyses to better represent the non-linearity of the ground response under the tower loads. The
findings of the three dimensional soil-structural interaction analyses using Arup in-house program
GSRaft computer models and a cross-checking exercise using PLAXIS 3-D model will be
discussed. Another critical element during the design process is the active input from the designer
during the construction phase apart from the routine supervision by the construction supervisor.
Instrumentation for monitoring the performance of the foundation and hence the tower will also
be highlighted.
1 INTRODUCTION
The combination of raft and pile foundations as a unified supporting system to support super high-rises
involves soil-structure interactions amongst the raft, piles and the underlying soil. The estimation of the load
sharing among the raft and the piles and the degree of mobilization of the pile capacity are the key challenges
for a piled raft design. Normally, a safety factor of 2 and 3 are applied to the design of the pile and the raft
foundations respectively. When the raft alone does not possess adequate safety factor, the piles are introduced
to reduce the bearing pressure of the raft. However, as the piles are much stiffer than the soil, if the
conventional safety factor is adopted, most of the load will be taken by the piles and the contribution from the
raft under the piled raft system is substantially decreased to less than 15 to 25% of the total load based on the
past experience. Although the raft has a lot more spare capacity in this situation, the use of a lower safety
factor for the pile in the piled raft system is not widely accepted.
An alternative is to design the piles as settlement reducers to control the settlement and/or differential
settlements of the raft. This Paper uses a supertall tower in South Korea resting on soft rock with complex
geological setting as a case history to illustrate the geotechnical design process of a piled raft foundation using
the concept of piles as settlement reducers.
2 DESIGN PHILOSOPHIES
The design philosophies with respect to piled raft foundation have been clearly defined by Randolph (1994)
and summarized by Poulos (2001a) as follows: -
47
(a)
(b)
(c)
the conventional approach, in which the piles are designed as a group to carry the major part of the
load, while making some allowance for the contribution of the raft, primarily to ultimate load capacity;
creep piling, in which the piles are designed to operate at a working load at which significant creep
starts to occur, typically 70 to 80% of the ultimate load capacity; sufficient piles are included to reduce
the net contact pressure between the raft and the soil to below the pre-consolidation pressure of the soil.
differential settlement control, in which the piles are located strategically in order to reduce the
differential settlements, rather than to reduce the overall average settlement substantially.
Design philosophy (a) is predominately governed by the pile group behavior. Design philosophy (b) is
considered to be effective to control the raft settlement with piles designed with a lower safety factor. Design
philosophy (c) is believed to be the most cost effective piled raft system as the piles are designed to fully
utilize their ultimate capacity. It is however difficult to precisely predict the load sharing between the piles and
the raft in view of the complex soil-structural interaction of a pile group, the non-linear stiffness behavior of
the soil and the actual pile deformation behavior in practice.
As far as the overall stability of a piled raft system is concerned, the most effective application of such
system is that the raft alone can provide adequate load bearing capacity and the piles are only designed to
control the settlement and/or differential settlements of the raft. In this regard, the piles can be treated as
settlement reducers or part of the ground stiffening element and a lower safety factor can be adopted. This
design concept is acceptable to most design engineers as the overall stability is also increased in associated
with the introduction of the piles. Nevertheless, the structural design of the piles shall comply with the
respective design code in order to prevent any structural damage. Taking the design philosophy (b) as
discussed above but with the raft alone being capable to provide adequate safety factor against bearing failure,
piles are designed with factor of safety of about 1.3 to 1.4 against the geotechnical capacity. With this design
concept, Davies et al (2009) applied the piled raft system to control the settlement and the differential
settlement for a high-rise development and discussed its financial benefits.
3 DESIGN PROCESS AND METHOD
The design process of piles as settlement reducers or ground stiffening element and the geotechnical design of
the raft foundation is shown in Figure 1. For the case of raft foundation bearing on rock (i.e. soft rock) with
complex geological setting, phased rock face mapping is essential to establish a credible geological model.
Sometimes, several geological models are developed for design in respond to the variability of the ground.
Borehole data
Credible geological
models
Initial rock face

mapping above
Raft level
Ground stiffness
parameters
Field &
laboratory testing
data
Update geological
models
Soil -structure
interaction
analysis
Ground stiffening
pile configuration and
parameters
Sensitivity
analysis
Raft, corewall & mega

column configuration
Movement
criteria
Structural loads
Final face mapping at

foundation level
Ground stiffening piles

configuration
Raft settlement profiles
Spring constant for
structural design of Raft
Figure 1: Geotechnical Design Process of the Raft Foundation with Ground Stiffening Piles
Apart from the geological model, the stiffness of the ground is another critical factor that influences the
analysis results. Soil and soft rock are well known as non-linear materials but in routine foundation designs
they are usually treated as elastic material to facilitate an easier computational analysis. However, the ground
with complex geological setting renders difficult estimation of a reasonable average stiffness of the ground. In
48
addition, the use of non-linear soil and/or rock stiffness do not require a well-defined rigid boundary beyond
which the settlement becomes minimal. The degradation curve of the non-linear modulus for sandy soils can
be referred to Seed & Idriss (1970) and Pappin et al (1989). The initial modulus at very small strain (i.e. about
0.0001%) can be correlated with the in-situ shear wave velocities measurement whilst the modulus at higher
strain values (i.e. about 0.01% to greater than 1%) can be correlated with the deformation modulus obtained
from the small strain triaxial tests or Pressuremeter tests for soil and Goodman Jack tests for rock. The radial
strain measured from these tests were converted into axial strain based on the relationship of axial = 2 radial /
3 as proposed by Jardine (1992).
The essential part of a piled raft design is soil-structure interaction. Katzenbach et al (1999) described the
interactions amongst different elements of a piled raft design, which is much more complex than the
traditional raft foundation design. These included raft-pile interaction, raft-soil interaction, pile-soil interaction
and pile-pile interaction. Poulos (2001b) described different methods of analysis for piled raft foundations.
Amongst these methods, numerical analysis using the plate-on-springs approach and 3D finite element
analysis are potentially the most accurate method available. It is however very time consuming to carry out a
3D finite element analysis. Arup has developed an in-house program namely GSRaft, which is part of the
General Structural Analysis (GSA) program, using similar concept of plate-on-springs approach with
consideration of the soil-structure interaction by conducting iterative process to achieve the compatibility of
settlement between the structural elements and the underlying soil elements. The program takes into account
the structural stiffness of the raft, piles as well as the stiffness of the underlying soils. The interaction between
the raft and the soil is simulated by a group of springs at the base of the raft while the interaction between the
pile and the soil is simulated by a group of soil interaction springs as illustrated in Figure 2 and the level of
these springs can be defined as the base of an equivalent raft level of the piles, which is usually taken as twothird of the pile depth for pile friction developed linearly with depth or half of the pile depth for constant pile
friction. Since the pile in GSRaft is modeled as elastic beam element, when the pile approaches its ultimate
capacity, the pile spring stiffness may be overestimated. Therefore, GSRaft is primarily applicable for design
philosophies (a) & (b). Furthermore, the assigned equivalent raft level of the piles will affect the results and in
layered soil strata and soil with varying stiffness, it is not easy to determine the appropriate level.
Figure 2: GSRaft Conceptual Model
4 CASE HISTORY
4.1 The project
The site is located at Jamsil-dong of Seoul, South Korea with a footprint of 87,000m2 and to be developed into
an integrated commercial, residential and entertainment complex. As a landmark of development as well as
Seoul, a 123-storey, 555m high supertall tower with a 6-level basement is situated at the western end of the
site. The initial ground investigation boreholes at the Tower were sunk in early 2006 to provide the basic
geotechnical information for the design of the Tower foundation. The Tower was subsequently replanned and
Arup was appointed as the geotechnical engineer of the international design team to provide the geotechnical
designs for the newly planned tower in view of the suspected complex ground condition as revealed from the
initial boreholes.
49
4.2 Structural form

The Tower has a footprint of approximately 72m x 72m. The structural form comprises concrete core wall
with 8 peripheral mega columns to resist both the gravity and lateral forces. To enhance the structural rigidity,
two belt trusses are positioned at 218m and 347m from the base of the raft. The total serviceability gravity
load is around 6,700MN.
4.3 Ground investigations & geological conditions
There were three phases of borehole investigation carried out at the Tower location. The first phase was
carried out in early 2006 and consisted of 16 boreholes accompanied by in-situ Pressuremeter/Goodman Jack
tests and laboratory strength tests. Most of these boreholes were terminated at around 20m below the
foundation level, which were relatively short comparing to the Tower footprint. Additional ground
investigation consisted of 7 numbers of deep boreholes were sunk in mid 2009 with depth as deep as 80m and
signs of faults running underneath the Tower were identified. With the mega-column locations and the
corewall dimensions became more certain, the third phase boreholes, which consisted of 10 numbers of
boreholes, was carried out in late 2009 accompanied by additional in-situ Pressuremeter tests and Goodman
Jack tests to further supplement the existing borehole information. Figure 3 illustrates the variability of the
rock quality and strength at distances underneath the foundation level.
The ground investigation works revealed that the bedrock at the Tower consisted of Gyeonggi Gneiss
complex overlaid by Quaternary colluvium and alluvium deposits. The Gneiss had banded structures and
medium grain to coarse grained texture. Hydrothermal intrusions were observed along fractures resulted from
faulting. Due to the variability of the rock strength over the depths underneath the Tower, for engineering
purpose, the rockmass could be classified into several categories using Geological Strength Index (GSI)
classification suggested by Hoek et al (1995) and the corresponding rockmass modulus could be estimated as
shown in Table 1. The rock condition at the excavation face of around 10m above the foundation level was
also inspected and mapped by the Korea Society of Engineering Geology (KSEG) in late 2009 and reviewed
by Arup, in order to better infer the extent of the individual rockmass over the footprint of the Tower. The
final rock face mapping at final formation level is illustrated in Figure 4.
Figure 3: A Sample Rock Core
Figure 4: Final Rock Face Mapping Results
50
Table 1: GSI Classification of the Rockmass

Rockmass Category
GSI/Em (GPa)
upper bound
lower bound
Banded Gneiss A
50/8.7
35/3.7
Banded Gneiss B
36/3.2
21/1.3
Banded Gneiss at Shear Zone
31/2.1
13/0.8
Fault zone
18/0.5
5/0.3
average
42/5.5
30/2.2
23/1.3
13/0.4
4.4 Proposed foundation system

In view of the loading condition of the Tower and the bearing capacity of the ground even assuming lower
conservative parameters, a single raft foundation was considered to be a technically feasible after an initial
assessment. Compared to individual footings, a single 6.5m thick raft foundation over the Tower footprint was
more robust in tolerating local weak spots due to its high stiffness. However, the presence of faults, shear
zones and highly foliated soft to hard rock would have a significant influence on the total and differential
settlements of the raft, which were carefully considered in order not to impair the Tower performance. The
uncertainties associated with the estimation of raft foundation settlement included: 1.
2.
3.
The rock conditions were highly variable. The stress and strain levels were concentrated locally at the
good rock areas under the core and mega columns. According to the Pressuremeter test results, the
difference of elastic modulus between good and poor rocks could be varied by a factor greater than 20.
The risk of unacceptable differential settlement would be high especially under lateral forces.
The geology condition beneath the formation level was highly complex. It would be difficult to develop
a full or complete geological model to represent the ground. The uncertainty of the geological model
would impose a high risk in predicting the actual response of the ground under the applied pressure.
Computation analysis, irrespective to the degree of sophistication, involved simplifications and the
inferred geological models could not be fully reflected in the analysis.
To mitigate such ground risk and enhance the performance of the raft foundation, ground stiffening piles
were placed at strategic locations to balance the foundation cost, ground variation and the building
performance. Using different credible geological models and after several rounds of studies, a total of 108 nos.
1m diameter Percussion Rotary Drilling (PRD) piles were designed to be positioned underneath the perimeter
corewall and inner zones to maximize the benefit. PRD pile was selected due to its readily easy maneuvering
and quick installation upon bulk excavation had reached 1 to 2m above the foundation level. The piles did not
require a temporary steel casing to be used and the down-the-hole hammer drilling could be applied directly to
Figure 5: Typical Section
51
enable a fast construction. These piles had a net length of 20m and 30m located at the highly stressed
influence zone and distributed based on the inferred geological setting. Figure 5 shows a typical section
showing the extent and depth of the proposed ground stiffening piles across the Tower footprint.
4.5 Design parameters & models
Non-linear stiffness of the banded Gneiss and brecciated Gneiss (shear zone) were applied in order to better
reflect the strain dependent behavior of these rockmasses, as shown in Figure 6. The initial moduli at very
small strain was correlated with the in-situ shear wave velocities measurement whilst the moduli at higher
strain values were correlated with the deformation moduli obtained from the Pressuremeter and Goodman
Jack tests. In view of the ground variation risk, a lower bound design line was adopted for design and
prediction of the Tower performance. For fault zones, since the long term creeping of the highly weathered
soil dominated the non-linear behavior rock, linear deformation modulus was adopted in the design.
Figure 6: Non-linear Modulus Design Curves
With different phases of borehole information and the rock face mapping at 10m above the foundation
level during the bulk excavation, two possible geological models were established to be the most credible for
the design purpose. One of the simplified geological models is shown in Figure 7.
In the GSRaft analysis, the raft was modeled as grillage and the cross points formed the common nodes to
iterate with the soil model. Piles could also be attached to these nodes. The piles were modeled as bar
elements with axial stiffness only and the top of the piles were pin-jointed to the raft. Soil-pile interaction
springs were placed at the end of the equivalent length of the piles. The loads were simulated as patches of
pressure loading over the mega columns and core walls footprint. A number of GSRaft models were set up to
carry out sensitivity check based on a range of design parameters and different load cases. A GSRaft model is
shown in Figure 8, which was coupled with the soil model as shown in Figure 7 during the analysis.
The result of analysis along one section at a particular load case of the proposed raft is shown in Figure 11.
The maximum and differential settlements can be derived accordingly from the displacement at individual
node. The corresponding spring constant could also be derived by dividing the reaction by the displacement at
each node for individual load case and a spring contour map could be produced for the use of the Project
Structural Engineer for the detailed structural analysis and design of the raft foundation.
52
In addition, from the GSRaft models, the envelope of reaction forces at individual ground stiffening piles
could be obtained. As discussed above, the raft foundation alone could sustain the total tower load. The
inclusion of the ground stiffening piles was to reduce the raft settlement and to bridge over fault zones or
localized weak zones. In this regard, these piles were allowed to settle plastically relative to the surrounding
soil (i.e. creeping) such that the exceptionally high load could not be developed in piles. In order to allow the
pile settled plastically relative to the surrounding soil/rock following the design philosophy (b) as mentioned
in Section 2, a factor of safety of 1.3 was adopted for the derivation of allowable geotechnical capacity of the
ground stiffening pile. A trial pile with load testing was carried out prior to the commencement of the working
pile construction and a bond capacity between the concrete and the brecciated Gneiss was proven to be
400kPa. Since the piles were generally socketed into banded Gneiss and/or brecciated Gneiss, with such a
high bond capacity, the pile capacity was not governed by the geotechnical capacity. Instead, the structural
design governed the pile capacity. The structural design of the pile complied with the Korean Building Code
(KBC) - Structural 2009. The ground stiffening piles had a concrete cylinder strength of 60MPa, which had
the same concrete strength as the raft after considering a tremie factor of 0.85.
Tower Raft
Remark: Half-length of piles modelled in GSRaft to reflect

the pile-soil interaction level
Figure 7: Geological Model for Input into Computer

Analysis
Figure 8: Raft and Piles in GSRaft Model
Figure 9: Pile Head Soil Cushion
A key feature of these piles was the pile heads being disconnected from the raft, with a 200mm soil
cushion as shown in Figure 9, which separates the piles entirely as part of the ground. This also reduced the
maximum pile reaction by 30% as the load from the raft was transferred to the ground before reaching the
53
pile. This allowed the load from the raft to transfer more uniformly into the ground instead of forming
extremely hard spots at the individual pile locations.
When the excavation approached around 1 to 2m above the foundation level, a final rock mapping was
carried out and the geological model was finalized.
The GSRaft analysis was then refined and further compared with a 3-D Plaxis model.The mesh set-up, the
ground model and the raft simulation are shown in Figure 10. Non-linear moduli of the rockmass were also
adopted in the Plaxis model. Based on the results of the analysis, similar peak settlement was found in both 3D Plaxis and GSRaft models but the predicted differential settlement by 3-D Plaxis was about 17% smaller
than that predicted by GSRaft model. It might be explained by the fact that the raft in Plaxis was modeled as
solid continuum element, which allowed dispersion of load through the physical dimension of the elements
resulting in more uniform load distribution beneath the raft while the raft in GSRaft was modeled as shell
element with no physical thickness despite the stiffness equivalent to the raft was incorporated.
Figure 10: 3-D Plaxis Model to Cross-check GSRaft Analysis
4.6 Construction stage support

It is a common practice in South Korea that the construction supervisor was independent from the design team
and the supervision team might have limited knowledge on the geotechnical design. In view of the variability
of the rock condition, the active input from the Designers during the construction phase apart from the routine
supervision by the Construction Supervisor is a critical element to promptly review if the actual ground
condition matches with the design assumptions. Periodic site supervisions were conducted by Arups designer
and engineering geologist to carry out rock face mapping and to verify or update the geological models at the
interim stages of excavation and before the final formation level was reached. Another important issue is the
potential blast induced damage to the rockmass immediately beneath the raft, which may result in undue
settlement during recompression of the opened joints under the imposition of the building loads. It is therefore
advised that no blasting shall be applied to the rockmass within 1m above the foundation level.
4.7 Design performance
Monitoring of the raft response during the imposition of loads throughout the construction phase is important
to verify the design assumptions and hence to secure the Tower performance. Instrumentations at raft
foundation level included the strain gauges in selected ground strengthening piles, earth pressure cells
underneath the raft, extensometers in rockmasses and settlement markers at the top of the raft. The bearing
stress, settlements and hence the differential settlements obtained from the site were compared to the
prediction at different stages of the Tower construction. The Tower is still under construction at the time of
54
preparing this Paper. Figure 11 shows the measured settlement of the raft when the Tower construction was up
to about three quarter of the total building height comparing with the design prediction. The projected
settlement when the total building load is applied is also estimated and is well within the design prediction.
This reflects that the assumed ground stiffnesses are reasonably conservative and the actual ground stiffness
may be close to the average ground stiffness as shown in Figure 6.
Besides, the bearing stress measured by the earth pressure cells underneath the mega column was found to
be slightly less than the theoretical value. This may be due to the actual concrete stiffness of the raft being
higher than the design prediction to spread the column load.
Figure 11: Comparison of Measured, Predicted and Projected Settlement of the Tower
5 CONCLUSIONS
The use of piles as settlement reducers (i.e. ground stiffening pile) to control the settlement and/or differential
settlement of a raft foundation is geotechnically viable. Using the modern computer modeling tools, the soilstructural interaction among the raft foundation, the ground stiffening piles and the ground can be efficiently
analysed. Furthermore, with the consideration of non-linear ground stiffness, a cost-effective piled raft
foundation solution can be derived. The case history presented in the paper has demonstrated the successful
application of ground stiffening piles for the raft foundation design in a complex ground condition to support a
high super tall landmark tower in Seoul. The Designers involvement to calibrate the ground models
throughout the construction process is one of the essential elements for the success of the project.
ACKNOWLEDGEMENTS
The authors would like to express gratitude towards the client Lotte Moolsan in allowing the publication of
this Paper. The authors would also like to express thanks to the project design team in particular the Structural
Engineer Lesile Robertson Associates and the local geotechnical consultant Jin Young Consultants for the
collaborated effort in pursuing this foundation design approach.
REFERENCES
Davies J., Lam K.L.A. & Junaideen S.M. 2009. Piled Rafts to Support High Rise Developments, Proceedings
of The 3rd Asian Seminar on ATC 18 Mega Foundation: Recent Advance of Deep Excavation and
Foundations, Bangkok, Thailand.
Hoek E., Kaiser P.K. and Bawden W.F. 1995. Support of underground excavations in hard rock, Rotterdam,
Balkema, 213pp.
Jardine R.J. 1992. Nonlinear stiffness parameters from undrained pressuremeter tests, Canadian Geotechnical
Journal, Vol. 29, pp. 436-446.
55
Katzenbach R., Schmitt A. and Turek, J. 1999. Co-operation Between Geotechnical and Structural Engineers
Experience Gained from Projects in Frankfurt, COST Action C7, Soil-Structure interaction in urban civil
engineering Proceedings of the Workshop in Thessaloniki, 1./2. Oct. 1999
Pappin J.W., Heidenbrecht A.C., Henderson P. and Naumoski N. 1989. Site response study soil modeling,
McMaster University, EERG Report 89-02, 52pp.
Poulos, H.G. 2001a. Piled Raft Foundations Design and Applications. Geotechnique, Vol. 50, (2): 95-113.
Poulos, H.G. 2001b. Methods of Analysis of Piled Raft Foundations. A Report Prepared on Behalf of
Technical Committee TC18 on Piled Foundations.
Randolph, M.F. 1994. Design Methods for Pile Groups and Piled Rafts. S.O.A. Report, 13 ICSMFE, New
Delhi, 5: 61-82.
Seed H.B. and Idriss I.M. 1970. Soil moduli and damping response analysis. EERC Report No. 70-10,
Berkeley, California.
56
Does 3D Finite Element Analysis Allow Better Design of Piles?

A Case Study for Temporary Marine Pile Design
Gavin S.H. Toh & Chao Li
Lambeth Associates Limited, Hong Kong
ABSTRACT
Large diameter steel tubular piles have been widely used by contractors in Hong Kong
as temporary support to working platform for construction of permanent bridge bored
piles. In Hong Kong design of piles has been normally practiced according to GEO
Publication No. 1/2006 (GEO, 2006), which reflects a combination of local experience
with state-of-the practice design philosophy. In general, GEO 1/2006 has been a very
useful design manual for practitioners and has been widely used in Hong Kong
successfully. However it does not particularly differentiate the design according to the
expected function/nature of the piles. Particularly temporary steel tubular piles have not
been explicitly dealt with in sufficient details. This paper uses PLAXIS 3D as an
analytical tool and to investigate the pile behaviours with soil-structure interaction and
its implication in design. Sensitivity analyses regarding drained/undrained behaviour of
alluvium as well as Rint (Interface Factor) were also carried out. For temporary works,
the authors recommend the use of observation method, with the aid of advanced finite
element analysis. General observations and recommendations on the practicality of
using advanced method in design are also provided.
1 INTRODUCTION
A series of temporary steel tubular piles has been designed/constructed as part of some bridge foundation
project carried out by Gammon Construction Limited. The function of these piles is mainly for support of a
working platform used for permanent works, which includes both installation of permanent bored pile and
construction of bridge piers. This paper presents a series of 3D finite element modelling works which were
used as basis for temporary tubular pile design considering the intended function of these piles. For simplicity
and clarity, only axial loadings are of concern in this paper.
GEO Publication 1/2006 has been routinely used in checking pile capacity. For drained soils, Equations
6.3 to 6.5 are frequently used (GEO, 2006). The shaft resistance equation based on semi-empirical method is
repeated as below
s = v'
(1)
where is the shaft resistance coefficient.

It is noticed that according to Table 6.3 (GEO, 2006), value can have a large spread. For driven small
displacement piles, the value varies from 0.1 to 0.5. For driven large displacement piles, the range is 0.2 1.5.
This large spread simply means that the shaft resistance can vary up to 5 times for small displacement piles
and 7.5 times for large displacement piles. Local experience is therefore critically important for establishment
of the shaft resistance coefficient .
The local experience for temporary steel tubular piles has not been fully established. To help to understand
the implications of value, a series of 3D simulation were carried out to investigate the performance of the
large steel tubular piles under axial loadings. The results appear to point to a more efficient way of doing
temporary steel tubular pile design, which, if implemented in full detail, can lead to more efficient design,
improved performance prediction and establishing local experience in steel tubular pile as temporary pile
foundation.
57
2 THE PROBLEM SETUP

2.1 Geological profile and pipe pile modelling
For the concerned project, the soil deposit mainly consists of Marine Deposit (MD) and Alluvium. Alluvium
can be either Alluvium Clay or Alluvium Sand. The design shear parameters for the Marine Deposit and
Alluvium adopted in the 3D FE model is presented in Table 1. Mohr Coulomb model has been used as the
material constitutive model.
Soil Type
Marine Deposit
Alluvium
Table 1: Soil Parameters Adopted in FEM Analysis for Location A

Friction
c'
Elevation
Density
Youngs Modulus
Cu
Angle
3
(kPa)
(mPD)
(kN/m )
(kPa)
(')
-5
-15
15
Eu = 400 Cu
----5
-15
-50
19
E = 1N (MPa)
30
4
50
dCu/dz
(kPa/m)
1.1
2.5
* N is the SPT N-value
Dimension and parameters of the pile is given as following:
Pipe pile diameter D = 1.2m

Pipe pile thickness = 16 mm
Material: Steel,
Elastic properties of steel: E = 205GPa, v = 0.2
2.2 Model geometry and analysis settings

The photograph in Plate 1 shows a picture of the construction project during the installation of the 1.2m
diameter steel tubular pile. Figure 1 illustrates the associated application of this problem the investigation
was started with a real construction problem where the steel tubular pile has been installed to refusal but is not
able to reach the design toe level. To verify the load capacity of the tubular pile, loading test is commonly
adopted. However, under the marine condition physical loading test is difficult with safety concern. PLAXIS
3D is therefore adopted as a simulation tool to model the soil-structure loading behaviour of the tubular pile
with the following considerations:
Soil parameters including drained/undrained alluvium and the interface factor;

The open tubular pile behaves as plugged or unplugged mode; and
The allowable settlement of the tubular pile that will provide a safe temporary platform for the
permanent piling work.
Given the complexity of soil pile interaction, this paper primarily focuses piles under static axial loading
and the driven process is not simulated.
58
Plate 1: Installation of the Steel Tubular Pile
Figure 1: Configuration of the Temporary Platform
3 RESULTS OF PLAXIS 3D MODELS

3.1 Analysis cases
Four cases of the PLAXIS 3D model representing different assumptions is summarized in Table 2. Case 1
considers the alluvium as undrained and no reduction of interface strength (which means the interface has
the same properties as the surrounding soils). This case is considered as the base design condition for
comparison. The model predicts a settlement of around 10mm at top of the steel tubular pile. In Case 2, the
alluvium as drained material is considered. In Case 3 a reduction of 50% on the interface material properties
was performed. In Case 4, an artificial steel plate is set at the pile base to verify the result of Case 3 where the
59
tubular pile is expecting to behave as a plugged mode. The vertical displacements at the top of the tubular pile
is presented in Table 2.
Loading displacement curves up to 50mm settlement of four cases is shown in Figure 2. A typical result of
the PLAXIS 3D model showing the vertical displacement contour is presented in Figure 3.
Table 2: Summary of PLAXIS 3D Models and Predicted Settlements
Interface
Interface
Length of
Pile Settlement
Alluvium
factor
Factor, Rint
Pile (m)
(mm)
1
1
25
Undrained
10.0
1
1
25
Drained
11.6
0.5
0.5
25
Undrained
15.5
0.5
0.5
25
Undrained
14.3
CASE
No.
1
2
3
4*
* The soil-plug condition is simulated by including an artificial steel plate at the bottom of the tubular pile (this case is
for reference only).
9000
8000
Case 1
7000
Loading (KN)
6000
5000
Case 2
4000
3000
2000
Case 4
1000
Case 3
0
0
10
15
20
25
30
35
40
45
Settlements at Pile Top (mm)
Figure 2: Load-displacement Curves for Cases 1 to 4
Figure 3: Vertical Displacement Contour under Working Load, P = 3000kN
60
50
3.2 Discussion on soil parameters

The difference of the load-displacement curve between drained and undrained of alluvium by comparing Case
1 and Case 2 is notable in Figure 3. However, the differences of the predicted displacement between two sets
are still rather small at the working load, P of 3000kN (only 1.6mm difference). This set of sensitivity analysis
is not to make judgment about whether drained or undrained behaviour actually shall take place, particularly
considering the uncertainties and variations in actual soil conditions.
In this paper, however, the computer model is mainly used as a simulation tool which can fit with the
observed data and provide guidance to future performance. For this purpose either drained or undrained loaddisplacement curves satisfies the purposes. For simplicity, the set of analysis which fit with the observed
monitoring data is considered. It should be noted that the general trend between drained and undrained
analyses are similar.
In the PLAXIS 3D model the interface elements are used to model the soil-structure interaction behaviour
where the material properties of the interface is in-general follows the adjacent soils with a reduction factor
(PLAXIS, 2012). It is important to realize that Rint (Interface Factor) in PLAXIS 3D is not strictly a physical
material property. Rather it is the proportion of strength reduction to the strength of adjacent soil clusters
(PLAXIS, 2012). The comparison between Cases 1 and 3 suggests that Rint can change the load-displacement
relationship significantly. It is noted that when interface factor Rint drops from 1.0 to 0.5, pile top settlement
increase 50%. The comparison between prediction model and observation suggests that modelling
assumptions in Cases 1 and 2 are closer to observed behaviour. However, this should not be interpreted that
the analyses has confirmed that the Rint should take the value 1.0; but on the other hand, it has confirmed that
the interfaces properties are close to c/cu or phi values as presented in Table 1.
3.3 Discussion on formation of soil plug
GEO 1/2006 has make references to the idea of formation of soil plug when making difference between small
displacement and large displacement driven piles type. If the open-ended tubular pile behaves in a plugged
mode the pile will effectively become large displacement pile type. It has therefore drawn special interest
from the authors to investigate whether the performance of the open-ended pile is close to the plugged or
unplugged mode. The comparison between Cases 3 and 4 has confirmed that the plugged assumption is valid.
3.4 Choice of prediction model based on comparison with measured settlements
Figure 4 shows the following measured settlement at the working platform under a design loading of 3000kN.
It is noted that Case 2 prediction is closest to the measured value. Therefore it is considered most appropriate
to adopt Case 2 as the prediction model.
Settlement (mm)
-5
-10
-15
-20
4-Jun-14
24-Jun-14
14-Jul-14
3-Aug-14
23-Aug-14
12-Sep-14
2-Oct-14
Figure 4: Measured Vertical Settlement at the Temporary Platform Near to the Concerned Pile
61
4 REVIEW OF PILE ULTIMATE LOAD AND SHAFT RESISTANCE

Using Case 2 as the preferred model, Figure 5 shows the load-displacement curve of the pile loaded up to
12,000kN with unloading-reloading at 6000kN, 9000kN. The tubular pile did not reach a physical failure load,
beyond which the load cannot be further increased. However, it should be noted that the displacements has
become more than 500mm under the axial loading of 12000kN.
For normal design purpose, GEO 1/2006 made reference to BSI (1986), in which the ultimate pile load is
defined as the loading mobilized when the settlement has reached about 10% pile diameter. From the loaddisplacement curve in Figure 5, the ultimate pile load is determined as 6800kN at 120mm settlement. The
axial load distribution along the pile shaft from the PLAXIS 3D model is shown in Figure 6 for different
loading magnitude. The results indicate that the load bearing at the pile base varies from 10% to 25%
(summarized in Table 3) depending on the magnitude of the pile load at the top.
Figure 5: Load-displacement Curve of the Pile (Case 2)
Figure 6: Pile Load Distribution with Depth (Case 2)
62
Table 3: Percentage of Load Bearing at Pile Base (Case 2)

Total Load (kN) End bearing (kN)
Percentage (%)
12000
2950
25%
9000
2100
23%
6000
1145
19%
3000
295
10%
1500
155
10%
Since PLAXIS provides a detailed distribution of axial loading along the pile shaft, it is possible to back
calculate value of Equation (1) at the alluvium layer. For the ultimate load of 6800kN, the distribution of the
shaft loading at the marine deposit, alluvium and end bearing is summarized in Table 4. For alluvium, the
back-calculated value of = 0.78, which falls within the large displacement driven pile range of 0.2 1.5
(GEO 1/2006). This is consistent with the observation of a plugged mode behaviour discussed earlier.
Table 4: Distribution of Loading along the Pile (Case 2 with 68000kN)
Load Type
Load Value (kN)
Percentage (%)
Ultimate Load
Shaft Load in MD
Shaft Load in Alluvium
6800
800
4640
12%
68%
End Bearing
1360
20%
The analysis approach using PLAXIS 3D presented in this paper has been used to investigate the tubular
piles at other temporary platforms of the project. Summary of the piles details, the average modulus of the
alluvium, depth of rock head beneath the pile toe and back-calculated values are tabulated in Table 5. The
ultimate load is taken as load at a displacement 10% of pile diameter. The results of the pile at location A have
been described in detail in this paper. The value around 0.4 at location B is smaller than the value at A is
likely due to the lower modulus of the alluvium at B. When the rock head is closer to the pile toe at location D,
the lower value compare to A suggests a higher distribution of load at pile toe due a stiffer bearing material.
However, the value at location C is notably high despite a lower modulus and higher rock head. It is likely
due to a higher mobilised shaft friction at the alluvium layer. Nonetheless, the back-calculated values are
within the large displacement driven pile range indicating a plugged mode behaviour.
Table 5: Summary of PLAXIS 3D and Back-calculated Values
Location
ref.
Tubular
pile ref.
A1
B1
B2
B3
B4
C1
D1
C
D
Ultimate
load
(kN)
6800
6200
6400
5400
5700
3500
5000
Pile Length
in soils
(m)
24.0
35.5
36.0
34.0
34.5
19.0
23.5
Pile length Rock head

in alluvium depth below
(m)
pile toe (m)
14.0
30
16.0
15
16.5
15
14.0
15
15.0
15
10.0
7
15.0
8
63
Ave. E of
alluvium
(MPa)
26.3
12.2
12.5
11.4
11.8
18.0
28.6
Backcalculated
0.78
0.41
0.40
0.43
0.41
0.80
0.52
5 CONCLUSIONS
It is possible to directly determine an acceptable working load of temporary tubular piles from finite element
PLAXIS 3D. With the load-displacement relationship from PLAXIS 3D, one would be able to estimate the
behaviour of the piles. However, careful calibration with observed settlements must be carried out in view of
variations in the input parameters. Alert, action, alarm (AAA) levels need to be set up in case of unforeseeable
ground conditions. The level of AAA levels, however, will depend on the acceptable settlement of the
temporary work.
From design point of view, direct use of GEO 1/2006 is preferable because of its simplicity in calculation.
However, since the value has a relatively large spread, the variation of working loading is also very large. A
conservative design may result from the simplified calculation based on GEO 1/2006 requiring a deeper pile
toe. Verification of the load capacity of the pile using physical loading test is a challenge in a marine condition.
PLAXIS 3D can be an alternative method for verification of the pile load.
It can be concluded that finite element using PLAXIS 3D together with site observation can help
to review the behaviour of temporary tubular piles with consideration of soil-structure and geological
variations, to assess the observed displacement and to predict future movement as the construction progress.
REFERENCES
BSI, 1986. British Standard Code of Practice for Foundations (BS 8004: 1986). British Standards Institution,
London.
GEO. 2006. Foundation design and construction (GEO publication No. 1/2006). Geotechnical Engineering
Office, Civil Engineering and Development Department, The Government of Hong Kong SAR
PLAXIS 3D, 2013. PLAXIS Reference Manual 3D-Version 2013. Delft.
64
Application of Modern Analytical Tools to a

Four Decade Old Metro Rail Station
S.P. Chin, A.K.L. Ng, L.J. Endicott & M. Ramanathan
AECOM Asia Co. Ltd., Hong Kong
ABSTRACT
The existing Diamond Hill Station in Kwun Tong Line (KTL-DIH) was built in mid-seventies
with provision for future expansion to either side. Thus a special form of external wall comprising
1500mm diameter hand dug caisson shafts with steel king posts and concrete jack arch panels
were adopted, which could facilitate removal of the walls for future expansion. However, the
KTL-DIH station does not expand sideways as originally planned. Now, the station in its fourth
decade is being linked to a new Diamond Hill Station (SCL-DIH) as a part of the Shatin to
Central Link expansion. When SCL commences its service DIH Station will be transformed into
a major railway hub for East Kowloon. The new and old DIH stations will be interconnected with
adits and unpaid links, and at these locations jack arch panels and kingposts would be removed.
Also, lifts and escalators would be added to KTL-DIH and existing entrances modified to connect
with SCL-DIH.
To carry out these modifications and to assess the impact due to deep excavation and
construction SCL-DIH structure in close proximity, the designers need to review and understand
the design principles and behaviour of the KTL-DIH structure. Back in the seventies, there were
few computers in use and hand calculations adopted. The Terzaghi and Peck envelopes were used
to determine the strut forces. A plastic design method, which was not recognised in any Code of
Practice until 1990, was adopted taking each stage of loading independently. The main walls
comprising king posts with external lagging are still in use today. PLAXIS, a modern-day
geotechnical software based on the finite element method has been used to replicate and backcalculate the conditions that were encountered during construction of KTL-DIH and establish the
locked-in-stresses considering the soil-structure interaction. SAP2000, state-of-art structural
design software has been used to evaluate the transition stresses from the transient active stage to
a permanent static stage. The forces and deformations from these two analyses are combined and
the walls as built are found acceptable based on the then prevailing codes. This method paved
way for further design leading to modification and adjacent excavation works, and has made the
connection from Shatin to Central Link possible.
1 INTRODUCTION
1.1 Diamond Hill Stations - old and new
In the year 1975, Mass Transit Railway (MTR) commenced the construction of the then called Modified
Initial System from Hong Kong Island, under the harbour, through Kowloon and outlying suburbs to terminate
at Kwun Tong: a distance of about 16km with 15 stations. The Modified Initial System (now part of the Kwun
Tong Line(KTL) and Tsuen Wan Line) comprised of 12 underground stations linked by bored and cut-andcover tunnels, and three elevated stations connected to 2.4km of viaducts. MTR decided that the initial system
should be let on a multi-contract basis, of which 25 were civil engineering works. The individual contracts for
the underground works were let on a design and construct basis and for a lump sum. Paul Y. Construction Co.
Ltd., Hong Kong was successful in bagging four civil engineering tenders which formed a continuous site of
2km with two major stations, a cut-and-cover tunnel and a bored tunnel. Choi Hung and Diamond Hill are the
two stations built by Paul Y (Benjamin et al. 1978). The detailed design of these works was carried out by
Maunsell Consultants Asia Ltd. (During the year 2009 Maunsell merged with AECOM).
65
Diamond Hill Station in Kwun Tong Line (KTL-DIH) which began its operation in the year 1979 was built
with provision for future expansion to either side (MTR 1974). However, the KTL-DIH station does not
expand sideways as originally planned, as a better alignment has now been planned to suit future transport
integration, and also to respect subsequent cityscape developments.
Shatin to Central Link (SCL) was gazetted in the year 2010 and construction commenced in 2012. It
comprises of two sections, namely Tai Wai to Hung Hom Section and Hung Hom to Admiralty Section.
The former Section will extend the existing Ma On Shan Line from Tai Wai to the West Rail Line via East
Kowloon to form the East West Corridor with the following stations: Tai Wai, Hin Keng, new Diamond
Hill, Kai Tak, To Kwa Wan, Ma Tau Wai, Ho Man Tin and Hung Hom. Refer to Figure 1.
Diamond Hill Station will be transformed into a major railway hub for East Kowloon when Tai Wai to
Hung Hom Section is completed. Modification works began in 2013 to upgrade Diamond Hill Station to
serve as the future interchange between the SCL and the KTL. A new Diamond Hill station (SCL-DIH) with
separate concourse and platforms would be built and a wide array of enhanced facilities added there including
the installation of three new lifts, 12 escalators, additional ticket issuing machines, automatic teller machines,
shops, and new public toilet facilities. Upon completion of SCL, passengers will be able to travel directly
from Diamond Hill Station to the East and West New Territories, Kai Tak, Kowloon City, and Hung Hom
without needing to change trains.
The MTR Corporation Ltd. appointed AECOM among other consultants to provide design services for
SCL. AECOM's work involves civil, structural and building services engineering design of a new interchange
station at Diamond Hill, and also upgrade of an existing rail depot and an extensive footbridge network.
Unlike KTL-DIH Station which was a design and construct contract, the new SCL-DIH Station is an
Engineers Design contract. Whilst AECOM (formerly Maunsell) worked as the detailed design consultant for
the contractor for KTL-DIH station, it is servicing as the clients detailed design consultant for SCL-DIH
station.
Figure 1: MTR Part Route Plan
Figure 2: Diamond Hill Stations-Location Plan
2 DESIGN OF OLD KTL-DIH STATION

2.1 Station box
KTL-DIH station is located under the carriageway of Lung Cheung Road (Figure 2). The existing roof,
concourse and track slabs are approximately 4m, 10m and 16.6m lower than the road level (Figure 4). Lung
Cheung Road dips from +12.8mPD (west) to 10.6mPD (east)
In the original design, provision was made to expand the station to either sides to accommodate an
additional track to the north and one to the south of the existing tracks. For such a future expansion main walls
along the sides of the station box were required to be removed with trains in service.
66
2.2 Tender design

The engineer's notional tender design scheme was to drive steel sheet piling to a cut-off below the station, then
excavate and build the station from the bottom upwards. Sheet piles as a semi-permanent wall could be readily
dismantled at a later date when the station widened. However, this scheme was later discarded as hard ground
conditions would result in damage to the sheet piles during driving with a corresponding loss of strength and
water-tightness. Also, sheet piles with limited structural depth are inefficient in bending over large spans. The
diaphragm wall option was also found unsuitable as it would have been difficult to break open at a later date.
The contractor had access to some 1500 hand dug caisson workers. Typically shafts of about 1.5m
diameter were excavated, if a boulder or bed rock was encountered it could be excavated by pneumatic tools.
Following exchange of ideas with the Contractors site staff, it was decided to use hand dug caissons
extensively. For the external walls, primary shafts sunk at 2.8m centres were used to place steel beams
vertically to act as king posts. Steel beams 911 x 418 x 342kg/m in Grade 43A in general were set to form the
main structural element, spanning vertically between the slabs. Between king posts, intervening soil excavated
and concrete jack arches spanning horizontally were provided extending from the level of the top of the roof
of the station to the underside of the track slab. Similar inner lining were also provided, but only as a
temporary support of the soil until the main excavation was done. Refer to Figure 3.
Figure 3: Outer Wall of KTLDIH Station
Figure 4: Typical Cross Section of KTL-DIH Station
Hand dug caissons were also used for the end walls where all the caissons were completely filled with
reinforced concrete to construct permanent walls. However, these caissons were not founded on bedrock and
the station box in effect is a floating structure. The station box is also supported by two rows of 900 x
1200mm internal columns at every 8.4m grid.
With the walls and columns in place top down construction commenced with casting the roof slab and then
excavating underneath the roof slab down to the level of the concourse and so on. This process was then novel
to Hong Kong. At the time there was only one deep excavation which was for the basement of the New World
Hotel which used diaphragm walls and tie back anchors for bottom up construction.
In 1975 there were few computers in use in the industry and very few engineering programs. Hand
calculations were commonly in use. For excavations with several strutting levels it was common to use
Terzaghi and Peck envelopes to determine strut forces. Moreover designers tended to determine bending
moments in the walls by using the same envelope but not taking into account the built in deformations
resulting from the staged construction.
67
In order to minimise costs, the tender design attempted to consider soil/structure interaction and to consider
the built in deformations. Instead of adopting the Terzaghi and Peck envelope, the active and passive soil
pressures were determined for each stratum on the site and the displacements needed to reduce the external
soil pressure from the at rest pressure to the active pressure and on the inside the forwards movement required
to develop the passive were included in the design calculations. This was a novel approach at the time.
However the approach was not readily accepted by the checkers and, for the detailed design for construction,
the Terzaghi and Peck envelope method was adopted.
Detailed design by traditional methods led to an estimated overstress which was overcome by adopting a
plastic design method (Horne 1968). Plastic design permitted assumptions of plastic hinges which resulted in
redistribution of bending moments. To accommodate shear stresses it was necessary to stiffen the webs of the
steel beams at the level of the concourse. By this means the design was approved for construction. Plastic
design was not recognised in a Code of Practice in U.K. until 1990, some 15 years later (BS5950-1990).
2.3 Geology and ground conditions
The area typically comprises 2m to 10m thick fill, underlain by 2m to 7m thick alluvium, in turn
underlain by saprolite, a completely decomposed granite ranging from 22m to 57m in thickness. Isolated
pockets of colluvium up to 9m in thickness were also identified locally. Level of bedrock (i.e. moderately
decomposed rock or better) is variable, between -25mPD to -40mPD.
2.4 Inspection
To verify the jack arch wall system the station was inspected during 2010. King posts are found to be well
covered by fire protection cement, no spalling noted, no significant rust identified, no excessive deflection
found nor no sign of distress. Jack arch panels are found dry in general, no major seepage or cracks identified,
and found generally in good condition. The RC diaphragms used as a tie member for king posts are also found
without any sign of cracks or distress. The station is well maintained.
3 IMPACT OF NEW WORKS TO KTL-DIH
3.1 Modification works
Two new interchange adits in the middle and an unpaid link in the western side KTL-DIH would be added to
serve the concourses of both SCL-DIH and KTL-DIH stations. The adits will be integrated with KTL-DIH,
and with movement joints at the interface with SCL-DIH. The existing jack arch walls and king posts on the
south side between roof and concourse would be removed at adit locations, and its impact should be assessed.
The existing public subway above KTL-DIH serves Entrances A1 & A2 and also provides entry down to
the KTL-DIH concourse. Entrance A2 would now be extended and configured to suit the layout at SCL-DIH.
A new lift will be provided at Entrance A1. A new shaft will be built and breakthrough of the subway wall is
necessary for entry at the public subway level.
3.2 Excavation of SCL-DIH
Owing to deep excavation of SCL-DIH structure, KTL-DIH, as a floating structure, will be subjected to lateral
movements and settlement. There will also be changes in forces acting by the surrounding soils to the
structure. In order to assess the structural adequacy of KTL-DIH when subjected to these movements and
changes in forces, it is first necessary to establish the current utilization of the structural members of KTLDIH, predict its future movements due to the excavation of SCL-DIH station and then estimate the
corresponding impact to the KTL-DIH structure.
As the KTL-DIH structure was built by top down method, the assessment of locked in stresses during its
construction stages plays a vital role. To assess the impacts of modification and new works, the existing
station has to be reassessed by modern computing methods. It was decided to adopt a two-computer software
approach and the outputs were coupled. More details are given in the following section.
68
4 DESIGN CHECK OF KTL-DIH USING MODERN TOOLS

PLAXIS (2D), a modern-day geotechnical software based on the finite element method has been used to
replicate and back-calculate the conditions that were encountered during the construction of KTL-DIH
structure and establish the locked-in-stresses considering the soil-structure interaction. The top-down
construction sequence of KTL-DIH structure and groundwater drawdown were simulated using PLAXIS to
find out the stresses and deformations locked-in the structural members at each key construction stage. This
analysis can determine the stresses and deformations induced by all temporary works, and provide the
conditions of the structural members at several critical sections at every stage of construction. The lateral
stress from the surrounding soil can be denoted as Kplaxis, which may be close to Ka conditions depending on
how much movement the soil has experienced.
SAP2000, state-of-art structural design software, on the other hand was then used to further replicate the
station structure by applying the increment of loading from the surrounding soils (Kplaxis) to the final loading
in permanent (Ko) condition. Simple two dimensional models are generated from the as-built records at
various sections. The models provide the current utilization of the structural members of the KTL-DIH
structure, setting a bench mark for the subsequent analyses for investigation impact to the structure due to
SCL DIH construction. Table 1 summaries the methods adopted for analysing the KTL-DIH structure in the
original design and present methodology.
Water Drawdown Analysis

ELS/Construction Sequence
Analysis
Permanent Case Analysis
Table 1: Comparison of Verification Methods

1970s
Seepage Analysis
Using Flow net
Beam Frame Analysis by Stage by
Stage approach
Plastic Analysis
Present
PLAXIS
PLAXIS
(PLAXIS + SAP2000)
The superposition of stresses, namely Kplaxis and (Ko-plaxis) and the forces derived from the two
software are illustrated in Figure 5. The design checks in accordance with BS449-1969 (elastic design)
indicated that the structural capacity of the king posts estimated in terms of P/A + M/Z is close to the unity.
However, in order to further compare with the modern codes, king posts when checked against
BS5950:1990 (limit state design), have been found well within the limit. Nevertheless, before
application of further forces to KTL-DIH structure, by modification and adjoining new works, it was decided
to strengthen the kingposts based on the then prevailing elastic design (AECOM 2011).
5 EXTENDING THE FINDINGS
Reassessing the adequacy of the existing station by the above mentioned two-fold method has led the
designers to understand the behaviour and the conditions of the structural members; the application is further
extended into stimulating the global and local effect from the new SCL-DIH construction.
To assess the impact of the KTL-DIH structure due to SCL-DIH construction, PLAXIS models comprising
KTL-DIH and SCL-DIH structures would be set up at critical sections, and the excavation of the SCL-DIH to
its formation level, including installation of lateral supports would be modeled. Excavation of interconnecting
adits, removal of king posts and jack arch walls of KTL-DIH, construction of adits and back filling would also
be modeled. These PLAXIS models would provide predictions on the movement of the KTL-DIH structure at
each critical stage of SCL-DIH and adit construction, and respective changes to the soil pressures acting on
the KTL-DIH structure.
The predicted movements of the KTL-DIH from PLAXIS would then be fed into the SAP2000s three
dimensional model of KTL-DIH structure to study its structural behavior. Furthermore, multiple stageconstruction analyses of SAP2000 model of KTL-DIH simulating with strut propping forces and soil
pressures from PLAXIS model would be carried out. By looping the above process, several iterations would
be required to achieve converged movements of the KTL DIH station as well as soil pressures acting on the
station. While the iteration is a lengthy process, it would capture the soil-structure interaction in a more
rigorous way where the changes in the soil pressures acting on the station can be modeled more realistically.
69
Figure 5: Superposition of Results from Two Softwares
6 CONCLUSIONS
The condition of the walls, both steel and concrete was found to be well maintained and in good conditions.
The check would have been very difficult if it were not for the detailed records held by the MTR Corporation
Ltd. including design calculations dating from 1976 and as-built drawings dating from 1978. The check
determined that the walls as built are acceptable when checked to current design standards. To be compatible
with present day design and construction methods, and to meet the challenge of construction a new station and
interconnecting adits next to the old one, two different computer software have been used to model the old
structure; and this would be extended for assessing the impact on old structure owing to new construction.
ACKNOWLEDGEMENTS
The authors wish to thank the MTR Corporation Ltd. for providing them an opportunity to be part of this
project and also permitting to publish this paper. The authors are grateful to Clement Ngai, Chief Design
Manager and Jason Wong, Project Manager of the SCL project team of the MTR Corporation Ltd. for their
support and guidance. The authors are indebted to AECOMs Jacky Wong, Managing Director, Building
Engineering and Ian Whitton, Sector Director-Railways for their encouragement.
70
REFERENCES
AECOM. 2011. Shatin to Central Link, Consultancy Agreement C1103 Diamond Hill Station and Stabling
Sidings, Works Contract 1106 A3.13A2 BD and RDO Consultation Submission for KTL DIH
Structural Calculation, Rev.B, December 2011.
Benjamin, A.L., Endicott, J.L. & Blake, R.J. 1978. The design and construction of some underground stations
for the Hong Kong mass transit railway system. The Structural Engineer, January 1978/No. l/Volume 56A,
pp11-20.
Horne, M.R. 1968. The Full Plastic Moments of Sections Subjected to Shear Forces and Axial Forces. British
Welding Journal.
Mass Transit Railway. 1974. Contract 207, Diamond Hill Station, Tender Design Statement.
71
72
Geotechnical Challenges in Construction of an Underground

Stormwater Storage Scheme in Happy Valley
W.H. Luk, C.L. Leung & C.Y. Lam
Drainage Services Department,
the Government of Hong Kong SAR
M.S. Hendy, S.S. Fang & J. Premchitt

ONLYgeotechnics Limited, Hong Kong
K.H. Lum
Black & Veatch Hong Kong Limited, Hong Kong
ABSTRACT
Happy Valley Underground Stormwater Storage Scheme (HVUSSS) is under construction by
Chun Wo Construction and Engineering Company Limited and supervised by the Drainage
Projects Division of the Drainage Services Department, the Government of the HKSAR,
supported by an Independent Geotechnical Engineer. The Scheme includes construction of a
storage tank, which is located below the existing football pitches at Happy Valley. Phase One
work including the southern portion of the scheme was constructed to a very tight programme.
As a part of this work, temporary excavations supported by strutted sheet pile wall have been
carried out in close proximity to sensitive structures. Soil-Structure Interaction modelling was
conducted as a part of the design of temporary support for sheet pile wall. The project requires
extensive geotechnical instruments monitoring with daily feedback to the Project Manager,
Contractor, temporary works designer and the IGE. This paper presents the geotechnical
challenges of the Excavation and Lateral Support works of the Phase One project (include 30,000
m3 storage tank, pump house, box culvert, etc.) together with description of notable events during
construction.
1 INTRODUCTION
Happy Valley was first developed as a race-course in 1846 on a piece of swampy land in a valley and was
often affected by floodwater. The area around Happy Valley has long been prone to flooding problems.
During the major rainstorms in August 2000, April 2006 and June 2008, severe flooding occurred in Happy
Valley and adjacent areas including Sing Woo Road, Wong Nai Chung Road, Morrison Hill Road, Lap Tak
Lane and the Happy Valley Racecourse and the Recreation Ground.
In this area drainage improvement by upgrading the existing major stormwater drains is disruptive and very
difficult. It would involve open trenching in the busy main roads. To avoid serious disruption to the public
and to minimize complicated diversion of the underground utilities, an underground storage tank was
proposed.
The underground storage tank will temporarily store part of the stormwater collected from the upstream
catchment. The addition of the storage tank would enable the drainage system to handle major rainstorm
events effectively, thus minimizing flooding problems in the area. The stormwater will subsequently be
discharged via pumps and gravity drains to the outfall once the stormwater flow volume in the existing drains
has subsided.
The Contract was awarded to Chun Wo Construction & Engineering Co., Ltd. (Chun Wo) in September
2012 who submitted, and obtained approval for a cost-saving design with raft foundations and under-drainage
system. The permanent and temporary works design is by Black & Veatch Hong Kong Limited (BV HK).
This contract is one in the initial batch of Government contracts to use New Engineering Contract (NEC)
instead of the traditional General Conditions of Contract (GCC). As a part of this contract, an Independent
Geotechnical Engineer (IGE), ONLYgeotechnics Ltd, was appointed to provide advice and assistance to the
Client, DSD, in their supervision and control with regard to geotechnical aspects of the contractors work. To
73
ensure timely control and minimizing any damages during Excavation and Lateral Support works,
geotechnical instruments monitoring was carried out with daily feedback to the Project Manager (the Engineer
as referred in GCC), Contractor, temporary works designer and the IGE.
2 SITE LOCATION
The site is situated on the estuary area downstream of the Wong Nai Chung Valley (ERM, 2010). An old map,
Figure 1, indicated that villages including Wong Nai Chung Village had been established on the valley
upstream of the estuary area. The original swampy ground was infilled over 150 years ago to form the race
course with filling probably sourced from the surrounding hills. The infield area of the race course was mainly
used as a resting area for recreation and leisure, up to the present. As revealed in old maps and aerial
photographs, there have been no significant earthworks within the infield area since then. Happy Valley
Underground Stormwater Storage Scheme (HVUSSS) is located beneath the infield area (football pitches) in
the centre of the Hong Kong Jockey Clubs Happy Valley Racecourse Figure 2.
Figure 1: Old map of Happy Valley Area
THE SCHEME
The project (two phases) involves the construction of a 60,000 m3 stormwater storage tank, an inlet structure,
a twin cell box culvert and a movable crest overflow side-weir system, see Figure 2. It also includes the
construction of a pump house with a discharge rate of 5,400m per hour. At the southern end of the scheme,
towards the hills, at the stilling basin, is a short length of hand-excavated tunnel which carries the new culvert.
The culvert then passes under the race-course and through the centre of the football pitches area, and exits at
the north western end of the race-course. The catchment area served by this scheme is about 130 hectares.
The infield area of the Happy Valley Racecourse is the lowest spot in the area, therefore it provides an ideal
location for the scheme from the hydraulic point of view.
The design calls for the use of a movable crest overflow side weir system incorporating supervisory control
and data acquisition (SCADA) for real time monitoring of water levels at the existing drainage system and the
74
tidal levels at the outfall to control up and down movement of the weir automatically. This innovative and
intelligent design would reduce the volume of the storage tank by 30% and save electricity consumption up to
60%. The pumping station will be remotely monitored and controlled, with outgoing data stream of water
levels, status of pumps and penstocks, any equipment failure etc.
Figure 2: Key features of the HVUSS
4 DESIGN SCHEME
The original design for foundations of the scheme indicated that pre-bored H-piles socketed into bed rock
would be required. Due to the lighter unit weight of the storage tank than that of the existing ground, the
foundation was designed to be capable of resisting the buoyant force. The Contractor proposed a cost saving
design using a raft foundation with subsoil drainage system including a cut-off wall (BV, 2012). The system
would maintain groundwater level to below a limit by conveying any excess groundwater through a drainage
blanket to be discharged or re-used. A series of groundwater monitoring points was installed along the
perimeter of the storage tank within the groundwater cut-off wall. An array of pressure relief wells was
constructed at the base slab of the storage tank as a part of emergency relief of any high groundwater pressure
within the groundwater cut-off wall. This alternative design was agreed and adopted.
The storage tank itself is a large concrete structure, L-shaped in the overall layout plan as viewed from
above, see Figure 3. The structure is some 7 metres deep. It is built in two phases according to the layout of
the existing football pitches in order to maintain the use of some of the football pitches throughout the
construction duration. The design allowed for soil cover of up to 800 mm thick on top of the storage tank for
re-turfing works, which would enhance the resistance to the uplifting force.
The alignment for the stormwater culvert (Figure 3) runs parallel to the storage tank and lies between the
LCSD building and the main excavation. The building is 2-storey LCSD changing rooms building for the
users of the infields sports facilities, with the large Diamond Vision Screen mounted on its roof. In order to
minimize disturbances to nearby sensitive structures, the construction sequence required early completion of
the stormwater culvert, before the start of the excavation for the adjacent storage tank. This would minimise
the effect of the main excavation on ground movements at the LCSD building.
5
GROUND CONDITIONS
Ground investigations were conducted in the 1990s associated with development of the Hong Kong Jockey
Club. Later, a two phased ground investigation was carried out in 2010 for USSS by DRILTECH comprising
36 drillholes, 1 trial pit, 30 inspection pits and 6 trial trenches and 8 GCO probes. A third phase of
75
investigation was undertaken following award of the contract to verify the design parameters for the cost
saving design. This included an additional 10 boreholes and 10 Cone Penetration Tests (CPTs).
Overall the investigation showed the ground conditions comprise several metres of sandy clayey Fill
underlain locally by Pond/Marine Deposits (soft clays) up to 5 metres thick. Below the soft clays is Alluvium
of predominantly sandy silty CLAY but locally clayey silty SAND up to 17m thick. The underlying bedrock is
Granite. Figure 4 is a summary of soil strata, and a section through the site is shown in Figure 5. Geotechnical
properties of the strata are summarized in Figure 6.
Figure 3: Underground Tank and Culvert Alignment
Figure 4: Summary of Soil Strata Extracted from GEO (2012)
Review of the Phase I ground investigation results indicated that the superficial deposit comprises fill,
pond deposit, estuary deposit, marine deposit, alluvium and colluvium. The fill layer is found in all drillholes
and the trial pit/trial trenches with a maximum thickness of 4.20m in drillhole BH15, which composed mainly
of clayey silty sand and sandy clayey silt with some angular gravel and occasional cobbles of rock fragments
76
materials. Alluvium is found in all drillholes with a maximum thickness of 16.20m in drillhole BH1, locally
interbedded with pond deposit in BH3 and IBH2, interbedded with estuary deposit in BH6 to BH8, BH11 and
IBH2 and interbedded with marine deposit in BH10 and IBH2, which composed mainly of sandy silty clay
and clayey silty sand with gravel materials. Colluvium is found in drillhole BH4 only with a thickness of
approximately 9.20m, which composed mainly of clayey silty sand with gravel materials. The distribution of
subsoil strata is rather complex, hence in model simulation a large number of different strata has to be
included and their thicknesses vary greatly in different vertical sections.
Figure 5: Ground Profile at Section A-A of Figure 3
Figure 6: Geotechnical Properties of Various Soil Strata Extracted from GEO (2012)
Groundwater monitoring records show the highest level of groundwater to be close to ground level ~
+5mPD with a lower level at around +2.9mPD. The water table generally follows the ground profile which
falls gently towards the north west.
6 SOIL-STRUCTURE INTERACTION MODELLING
As a part of the design for the Excavation and Lateral Support system, extensive Soil-Structure Interaction
modelling was carried out. Two dimensional finite element analyses were conducted using the BD approved
program PLAXIS version 9.02. Plane strain conditions are assumed. The analyses results were used to
assess the ground movements at and around the excavation face, the sheet pile and the nearby sensitive
structures. Initial analyses were conducted from early 2013. Five vertical sections within the area under
construction were analysed, see Table 1. The analysis for each of the sections was performed through all the
construction stages (phases) relevant to section from start to finish in a step by step procedure. The number of
77
ground strata included in the analysis varied from 24 to 10 depending on the section considered. Additionally,
relevant structural elements were also included, such as sheet pile wall, concrete walls and slabs of storage
tank, box culvert and LCSD Building. The total numbers of construction stages, time steps, ground strata and
structural elements for each of the sections analysed are presented in Table 1. Examples of the finite element
grid used for Section 10-10 are shown in Figure 7 for the initial ground condition before construction work
and in Figure 8 for near completion stage with strutted support to the sheet pile wall.
Table 1 Summary of the Sections for PLAXIS Analyses Carried Out Initially
Section
Total
Total time steps Ground Strata1 Structural Elements2
No.
Construction Stages
37
717
24
14
8-8
23
343
19
12
10-10
25
304
22
10
11-11
24
634
24
10
12-12
15
303
10
2
X
Note 1. Ground strata such as Marine Clay, Alluvial Clay, CDG each has assigned engineering
properties such as unit weight, permeability, Youngs Modulus, etc.
Note 2. Structural elements such as sheet pile wall, concrete walls and slabs of storage tank, box
culvert, LCSD Building, etc and elements acting as anchors.
For finite element analysis, the important values of Young Modulus for soils were derived from relevant
considerations including initial values from early studies (Figure 6) and those estimated from later
investigation as a part of the design including CPT results for soil layers above CDG and SPT N values
from the tests on CDG and HDG layers. It is generally recognized that the Youngs Modulus for unloading
and reloading could be much higher than that for the virgin loading. In the ground movement analysis, the
Youngs Moduli for soil within the excavated area of the storage tank, where effective stress would be
reduced, were taken to be 2 times the respective virgin moduli (unloading & reloading E = 2 x virgin E).
However, for determination of member forces and moments for structural design (under ULS) the virgin
values were used.
Supplementary PLAXIS simulation was carried out and reported in early 2014 after initial part of the
work has been done. This is to take account of the agreed change in the strut support arrangement in view of
the much smaller measured ground movements than expected. This included further analyses on Sections 1010, 11-11 and 12-12.
Figure 7: The PLAXIS Finite Element Grid for the Ground Before Excavation
78
The total ground settlement around an excavation comprises two part, settlement due to wall deflection
and that due to groundwater drawdown. PLAXIS analysis was able to simulate the wall deflection and the
associated ground settlement as well as the groundwater drawdown and the related settlement. For the area
around the LCSD Building the results from PLAXIS for the maximum total ground settlement is 26 mm. The
maximum estimated at the most critical high mast for lighting (out of the 10 masts) is about 22 mm. The
range of possible maximum settlements at the edge of race track was estimated to be from 5 to 14 mm. For 10
m wide zone surrounding the outside of the sheet pile cut-off wall, the estimated ground settlement is more
than 25 mm. Part of the jogging path and part of paved area at the east of the LCSD Building are in this zone
and could be affected. It is considered that the impact to the race track and the area east of LCSD Building are
much less critical than the impact to LCSD Building itself and the high masts.
Figure 8: The PLAXIS Finite Element Grid for the Excavated Ground with Strutted Support (Stage 21)
It is considered that the results from PLAXIS simulation are within the settlement limit generally adopted
for settlement control at or near structures. Therefore the Action Level was set at 25 mm, with Alarm and
Alert Levels set at 20 and 10 mm respectively.
7
GEOTECHNICAL CHALLENGES
The construction works started from September 2012, and reached substantial completion of Phase One by
March 2015. The works commenced with construction of the pump house followed by excavation for the
storage tank. As the excavation progressed, temporary support is provided by installation of sheet piles using
a silent piler to minimize any noise impact to the public in the vicinity. Thickness of the soft layer is critical
in the control of movements during the work, and the thickness contour is shown in Figure 9. The progress of
work up to May 2014 is illustrated in Figure 10.
The cost saving design requires the sheetpile wall around the storage tank to act as groundwater cut-off
wall as a part of permanent work to minimize the water pressure under the tank. The sheetpile wall has to
perform this function well or else there could be a high uplift force, nevertheless there is a contingency
measure for such an event by activating the relief wells. Pumping test for demonstrating water tightness of the
sheetpile wall was conducted. Although some seepage was observed from the sheetpile wall, the amount is
considered small and well below the limit allowed for in the design.
The design also requires excavation of about 7 metres of the superficial deposits, with much of it below the
level of the ground water table. The LCSD building with Diamond Vision Screen on top, and the high mast
supporting the flood lighting (> 20 m high) are both quite close to the excavation. According to ArchSD asbuilt drawings, the building is founded on a raft footing, a hollow box structure embedded about 1.6m below
ground (BV, 2012). The hollow box reduced the load on the soil strata below. The high mast (visible in
79
Figure 12) poses a particular challenge. It is founded on shallow spread footing, size 4.3m x 4.3 m at about 1.6
m depth, and only a few metres from the excavation. Local softening due to heavy rain exacerbated the
potential movement and will be discussed later. The challenge for the construction was to have adequate
excavation support system, with control on construction sequence, and continuous monitoring / alert system
for on-going detection of any excessive movement and timely remedial action.
Figure 9: Contours of Soft Deposits
Figure 10: Phase One Construction in Progress May 2014
The design of the Excavation and Lateral Support is a typical arrangement with multiple struts required as
excavation proceeds. One of the particular challenges arose at the sheet pile wall immediately in front of the
LCSD building. This is the most sensitive area but made more complex because of the location of the box
culvert between the storage tank and the building. Programme requirements dictated early construction of the
box culvert between the pair of sheet piles shown in Figure 11. The design indicated that some movement of
the LCSD building and area to the east of the building would occur during trenching and installation of the
box culvert and this movement would be further developed during excavation for the adjacent stormwater
storage tank. As a cost-effective alternative for extensive ground treatment works, pre-loading of the struts
was introduced to mitigate the predicted movement of the building. The lateral support work in progress is
shown in Figure 10.
(a)
(b)
Figure 11: The Arrangement for Excavation and Support in Two Areas
80
8 INSTRUMENTATION AND MONITORING

Assessment of the results from the finite element modelling could be made by comparing the results with the
measured movements from instrument monitoring including those on sensitive facilities located close to the
race course compound. This was done based on the monitoring report of cumulative readings of all
instruments up to the end of February 2015. For ground settlement, it is mostly below Action Level of 25 mm,
with only about 10% of ground settlement markers (19/189) having settlement readings exceeded the level.
The percentage is also relatively low, about 18%, for building settlement markers (12/68), and 5% for tilt
markers (4/83). Most other readings show little movements. Apart from SSI modelling, other factors during
construction have to be taken account of, such as the mis-match in the sequences of excavation work and
lateral support work, and heavy rainfall at vulnerable stages of work. For the exceedance cases, all parties
were aware and take appropriate actions since after Alert Level has been reached. Some minor damages such
as cracked paving and surface drainage channels did occur and were immediately rectified but no major
damages were observed. Two cases of exceedance are presented below to show the geotechnical challenges
encountered in actual construction work.
Figure 12: Excavation and Lateral Support Woks in Progress
The contour plan in Figure 9 shows the clay is up to 3 metres thick below the building. The LCSD
building is on a spread raft foundation with the existence of soft clay layer below it. Geotechnical instruments
installed around LCSD Building are summarized in Table 2.
In the area near LCSD Building, up to 4mm of ground settlement was recorded on by October 2013 due to
nearby excavation for the box culvert. Settlement continued but gradually reduced, stabilizing from end of
2013 to June 2014, see Figure 15(a) for ground settlement marker GS34, the location of which is shown in
Figures 3 & 13. This marker is located near a surface u-channel in the area east of LCSD Building. Starting
from July 2014 substantial settlement occurred and progressed at a high rate concurrent with the rapid
excavation in the area, while strutting support work was not yet catching up. The settlement also appeared to
be accelerated at times of heavy rainfall. High ground settlement (26mm) in this area is anticipated according
to PLAXIS results discussed earlier. During this period, the building settlement markers at LCSD Buildings
recorded fairly uniform settlements along the edge parallel to the excavation indicating very small tilting and
distortion of the building. Immediate response to readings included increasing the monitoring rate, reviewing
the construction sequence and immediately repairing gaps between the LCSD building and the adjacent
ground to avoid water ingress. It became apparent during the review that the building itself was not subject to
significant movement but that adjacent channels at ground level had moved during excavation. This
movement was not critical to the building and was resolved by an adjustment to the construction sequence.
81
Instruments
Ground Settlement Marker
GS33 GS41 (10 Nos.)
Building Settlement Marker
BS1 BS15A(15 Nos.)
Inclinometer
IN8 - IN9
Piezometer
SP4
Extensometer
E1 - E2
Tilt Marker
T9 T20
Table 2 Summary of Instrumentation around LCSD Building

Purpose
Ground Level, mPD
To monitor the ground settlement around LCSD Building
+4.371 - +4.796
To monitor the building deformation/tilting
+4.862 - +6.530
To monitor the deformation profile inside soils
+4.362 - +4.450
To monitor the ground water level and pore water

pressure around the LCSD building
+4.45
To monitor the deformation between soil layers
+4.473 - +4.535
To monitor LCSD light mast
The teams involved were alerted to the situation and the increased observation and rectification of existing
facilities and newly built structures (e.g. box culvert) were carried out. No significant damage was observed
on the newly built structures and the LCSD building while there was localised damage to the drainage channel
near the building. The settlement stabilized by the end of August and during September 2014, as the full strut
supporting work was completed. There have been no significant settlements since September 2014. The
progress and magnitude of ground settlements around the LCSD Building (see also Figure 14) at three
settlement markers can be seen in Figure 15(a).
Figure 13: Instruments around LCSD Building
Figure 14: The Diamond Visual Screen on top of LCSD Changing Rooms Building
82
(a)
(b)
Figure 15: Monitoring Results (a) Ground Settlement at LCSD Building (b) Tilt Marker at the High Lighting Mast
Close attention was paid to the high mast for lighting T15 as it is located near the construction area as
shown in Figure 3. The data from the reading of tilt marker T15 are shown in Figure 15(b). The tilt marker
T15 is of the type recommended by GEO (2012) with a vertical distance of 5 m between the two markerpoints. The location of T15 is indicated in Figure 3. The first significant movement of about 12 mm occurred
in August 2013, related to the excavation for underground storage tank, possibly also included unknown
measurement anomaly due to the change in monitoring team at the time. Responses to rainfall events similar
to that at GS34 were also observed for T15. The most significant movement of up to 20 mm was observed in
April 2014 during the construction work of the storage tank in the vicinity. The relevant teams were alerted
and prompt attention and collaboration was initiated. The movements were quickly stabilized within about a
month (Figure 15 (b)) at the completion of base slab of the tank and the lateral support system nearby.
The construction of HVUSSS is well on the way to completion on schedule for both Phases One and Two.
9
CONCLUSIONS
An innovative project to alleviate the serious flooding problems in Wan Chai Happy Valley area is in
progress with Phase One substantially completed. Temporary excavations had been carried out in close
proximity to existing sensitive structures. A plan of continuous monitoring and construction control has been
implemented from the start within the frame work of the newly introduced NEC Conditions of Contract. The
monitoring and control work has been well conducted with effective co-operation among the Project Manager,
Contractor, Temporary Works Designer and the Independent Geotechnical Engineer (IGE). Analyses to
determine ground movement, using PLAXIS finite element simulation as a Soil-Structure Interaction
modelling, was carried out from the design stage and well into the construction stage so that adjustments can
be made for the modelling to be more representative of the actual conditions. The simulation results
compared with the actual measured movements showed that only 5% to 18% of all instrumented locations
have readings exceeding the maximum limit specified. Two exceedance cases are described. They showed
that compliance with the designed construction sequences in all stages is important in reducing any potential
problems during the work. The presented cases illustrate the vital role and effectiveness of continuous
instrument monitoring and reporting in the control and management of geotechnical works, so that any
possible problems are detected early and prompt co-ordination of actions can be carried out to prevent more
serious problems from occurring.
ACKNOWLEDGEMENTS
This paper is published with the permission of Drainage Services Department, the Government of the Hong
Kong Special Administrative Region (HKSAR).
83
REFERENCES
BV(2012), Happy Valley Underground Stormwater Storage Scheme Innovative Design Enhancement, No.
382903/B&V/GDR/001, Dec 2012, Black & Veath Hong Kong Ltd., produced for Chun Wo Construction
& Engineering Co. Ltd.
ERM (2010), Archaeological Impact Assessment for Happy Valley Underground Stormwater Storage
Scheme Phase A Desktop Research Report, Ref.0122176, Nov 2010, Environmental Resources
Management, produced for DSD
GEO (2012), Item No. 160CD Happy Valley Underground Stormwater Storage Scheme (HVUSSS)
Foundation Design, ADR 5/2012, Aug 2012, Geotechnical Engineering Office, produced for DSD.
84
Recent Experiences of Numerical Prediction & Assessment

Excavation over a Tunnel of Unbolted Segmental Tunnel Lining
J.B. Wang & L. Swann
Jacobs (China) limited
S. Reynolds
Jacobs (UK) limited
ABSTRACT
A 700 m long Grade Separated Road (GSR) was proposed to link the new Terminals and on to a
new set of remote stands at an Airport in UK. Construction of the GSR involved excavation
adjacent to and over the existing live underground tunnels. The excavation had a depth upto 11m,
and was only 6m above the crown of existing tunnel. To investigate the impacts on the unbolted
concrete segmental tunnel lining from the proposed GSR work, a 3D model together with a series
of 2D models were developed considering the site history and construction sequences
proposed. Both short and long term behaviour of the existing underground tunnels during and
after the GSR construction was investigated, showing the proposed GSR construction method
was a viable solution to minimise the impacts on the existing tunnels.
1 INTRODUCTION
As part of the airport development, a Grade Separated Road (GSR) is proposed to link new Terminals and on
to a new set of remote stands. The road has an overall length of 700 m and comprises two taxilane
underpasses to avoid conflict between road vehicles and aircraft. Construction of the proposed GSR involves
the installation of embedded secant pile retaining walls, casting bridge decks across the underpass sections,
and subsequent excavation within the walls to depths of up to 11m with temporary propping, in places, to
reach formation level for casting the base slab.
The GSR passes over and runs sub-parallel to the existing underground tunnels which were built in the
1970s using an unbolted concrete segmental tunnel lining. The excavation for the GSR, over and adjacent to
the live tunnels, causes great concern over safe operation of the trains. ELS scheme was developed to limit the
impacts on the existing tunnels associated with the GSR construction consisting of installing 900mm diameter
secant pile walls as temporary retaining walls during the excavation stage and also as permanent retaining
walls for the GSR.
FEM 2D software Plaxis 2-D Numerical analyses were carried out at selected critical sections to estimate the
associated ground movements and to assess the impacts to the underground line. The Plaxis 2D was adopted at
locations where excavation is along / parallel to the existing tunnels, hence a plain strain condition is generally
applicable. A 3Dmodel was established to investigate the impacts where the proposed GSR crosses over the
existing tunnels at a skew direction. In addition, a Plaxis 2D analysis was also performed at the 3D region to
illustrate the inaccuracy of 2D analysis for solving such 3D problem. The 3D model was analysed using 3D
software FLAC 3D which adopts finite difference method for its lower computation memory demand.
This paper presents the case study from the numerical modelling aspect, concentrating on the 3-D analysis in
particular. Compared with 2-D analysis, the benefits of the 3-D modelling for prediction with higher accuracy
are highlighted.
2 SITE AND GROUND CONDITIONS
Ground conditions at the site, comprise Made Ground, underlain by River Terrace Deposits overlying London
Clay. The existing ground level is generally flat, approximately at 23.1 m AOD. Groundwater levels recorded
on site were typically at 3m below ground level.
85
The Made Ground is relatively thin having a thickness of 0.8m to 1.6m, and is generally described as
medium dense silty sand with a base level at 21.65 to 22.25 m AOD. The thickness of the River Terrace
Deposits varied between 3.5 m and 4.7 m, and are described as medium dense to very dense sandy gravel with
occasional cobbles. London Clay has typically been described as firm to stiff, grey extremely closely to
closely fissured clay with occasional silt laminate and partings and occasional medium sand to fine gravel size
shell fragments and is heavily overconsolidated. The ground conditions on site are illustrated below:
Ground level at 23.1m AOD
Made Ground(base level at 21.65 to 22.25 m AOD)
Design Groundwater at 21m OAD
River Terrace Deposits (base level at 17.2 to 18.5 m AOD)
London Clay
The GSR is about 700 m in length and 12 m wide, lying above the existing live tunnels. The GSR ramp
crosses over the tunnels at an angle of 57 degrees and then runs approximately parallel to the tunnels with a
minimum 4.3 m spacing. At the crossing area, the GSR ramp is about 3 m deep and is approximately 8 m above
the tunnels. The depth of the GSR increases to about 8 m at the down ramp near the proposed left taxilane
underpass. The internal diameter of the Piccadilly Line tunnels is 3.81 m, formed by 22 numbers of precast
concrete segments. The tunnel rings are approximately 0.6 m long and 152.5 mm thick. Airbus A380 / Boeing
747 aircraft loading of total 9000 kN is considered in the design of the bridge and retaining walls.
3 METHOD AND SEQUENCE OF CONSTRUCTION
It is proposed that the excavation for the GSR be supported by secant pile walls and that the secant pile walls
will also form the permanent walls of the GSR. The secant piles above the tunnels will terminate at
approximately 6 m above the tunnels and the piles shall also be a minimum of 3 m away from the tunnel
structure as required by the tunnel exclusion zone.
One to two layers of temporary props are proposed to support the secant pile walls during GSR
construction. At the underpass area, with a bridge deck at the top, no props are proposed during excavation.
Construction of the GSR shall adopt top-down method under the bridge deck and bottom-up construction
sequence for other parts. The typical construction sequence using bottom up sequence for the GSR sections
outside the Taxilane Underpass is presented below:
(1)
(2)
(3)
(4)
(5)
Install secant piles

Excavate to 0.5m below bridge deck
Install /casting bridge deck
Excavate to final excavation level
Cast base slab and skin walls and connect it with the secant pile walls
4 3-D NUMERICAL MODELLING

Excavation for the construction of the GSR will change the stresses and induce ground movement around the
existing tunnels. The western ramp of the GSR crosses over the existing tunnels at a skew, which requires a
more sophisticated study on the likely impacts on the existing tunnels. A three-dimensional numerical model
was developed and analysed with Finite difference software FLAC 3D version 3.1.
4.1 Modelling domain and boundary conditions
The 3D model developed is a rectangular block of 102 m long x 100 m wide x 63 m deep comprising a total of
around 60,000 soil zones. A general view of the domain is presented in Figures 1 & 2 below.
86
Proposed GSR
Figure 1: General View of FLAC 3D Model
Figure 2: Plan View at 13.9 mbgl
The model simulates the western portion of the proposed GSR which runs above the existing twin tunnels at
an acute angle, between chainage of 25 m and 135 m as shown in Figure 2. A 100 m length of the twin tunnels
are covered in the model, running parallel to the y-axis from y = -50 m to y = +50 m.
For the four vertical external boundaries, horizontal movement perpendicular to the plane is restricted while
movement in the plane of the vertical boundary is allowed. The vertical movement is not permitted at the base
boundary of the 3D model.
4.2 Soil elements and soil properties in the model
The ground under consideration comprises London Clay overlain by Made Ground (MG) and River Terrace
Deposit (RTD). The ground model has been simplified to combine the thin layer of MG and RTD into a single
stratum of drained soil in the 3D model since the thin layer MG has similar granular material properties to the
RTD. A constitutive model of liner elastic perfectly-plastic with Mohr-Coulomb failure criterion was adopted
for the Made Ground and the River Terrace Deposit. The nonlinear elastic properties of the London Clay shows
a strong dependency of the soil stiffness on the strain levels experienced, i.e. higher stiffness at lower strains.
The nonlinear tangent elastic properties of the London Clay can be described by the following equations, given
by Jardine et al (1986):
BI 1
Eu = cu A + B cos I
sin I
ln
(
10
)
( )
( )
(1)
where:
cu = undrained shear strength , I = log10 ( / C), A, B, C, , = material-specific constants,
a =
2
( 1 2 )2 + ( 2 3 )2 + 3 1
3
)}
2 0.5
,1 , 2, 3 = principal strains (or principal deviatoric strains)
87
The soil parameters adopted in the 3D model are summarized in Tables 4.1 and 4.2 below:
Table 4.1: Soil Parameters
(kN/m3)
Made Ground &

River Terrace
Deposits
20
0.2
0 5.5mbgl: 0.4
London Clay
20
0.49
5.50 11.25mbgl: 2.6

11.25 31.75mbgl: 2.2
31.75 63.00mbgl: 1.2
Soil Type
ko
c /cu
(kPa)
()
E
(kPa)
k
(m/s)
36
37,500
2 x 10-4
2 x 10-9
Increase with
depth (86 kPa at
London Clay top
Table 4.2: Jardine Model Parameters for London Clay

A
C (%)
1350
1350
0.001
1.319146
0.66336
min (%)
0.0011
max (%)
0.3
4.3 Tunnel lining in the model

The existing tunnel lining comprises 22 unbolted pre-cast concrete segments with a ring length of 600mm.
Since it is impractical and unnecessary to build a 3D model with those lining segments modelled precisely, an
equivalent stiffness method by Muir Wood (1975) is adopted, which suggests:
Ie = Ijoint + (4 / n)2 x Isegment
(2)
Where:
Ie = equivalent moment of inertia for a continuous liner element in model,
Ijoint = moment of inertia of joints between lining segments (= 0 if no structural connection),
Isegment = moment of inertia of segments,
n = number of segments (n = 22 in current case)
The tunnel lining were modelled as a liner structure element in the FLAC 3D model that can take the tunnel
hoop stress, bending moment and shear stress and also friction between the soil and the tunnel lining. The
secant pile wall and base slab of the GSR were also modeled as liner structure elements, having isotropic linear
elastic properties of the concrete material. Beam structure elements were adopted for the struts modelling.
4.4 Proposed works and modelling sequence
(a) Establishing initial conditions
To establish the existing conditions of the ground and the existing underground structures, the 3-D model was
set up to include the major construction history of the existing tunnels including the construction year and tunnel
face volume loss during tunnelling. The West Bound (WB) Tunnel was constructed first and then East Bound
(EB) Tunnel. The tunnel face volume loss was simulated by excavating the tunnel area allowing a 2% volume
loss in term of tunnel cross-sectional area. Tunnel lining is then installed to support soil stresses around.
The tunnel construction stage is followed by long-term consolidation in which any excess pore pressure
generated due to tunnel excavation would be fully dissipated, assuming a permeable tunnel lining. After full
consolidation, present (or existing) condition of the domain has been arrived at. All displacement vectors are
then set to zero before commencement of the proposed works.
(b) Modelling of construction Sequence
The excavation and GSR construction are designed to be carried out in 3 bays. Subsequent to the installation of
secant pile wall which is assumed as wished-in-place, excavation to 1mbgl and installation of the first layer of
711 x 14 CHS struts and bridge deck are modelled. Construction of the GSR is then carried out bay by bay in
three stages as shown in Figure 3. Further excavation, installation of second layer of 711 x 14 CHS struts and
casting of base slab in one bay would then be completed, before a similar operation for the next bay.
88
Upon completion of base slab construction for the Bay 3, all struts would be removed. Construction of the
airfield pavement would then be modelled. Finally, a long-term consolidation is then carried out. The
construction stages modelled are detailed in Section 3.
Figure 3: 3-bay Construction in FLAC 3D Model
5 RESULTS AND DISCUSSIONS

5.1
Predicted tunnel movement
Long term ground movement in a horizontal plane is presented in Figure 4 below. It can be observed that the line
of zero transverse (in x-direction, see Figure 4) ground movement vector generally matches the centreline of
GSR. The ground heave effect including ground movement towards the excavation from both sides is illustrated
in Figure 5 below.
Figure 4: Contours of Transverse Ground Movement in Horizontal Plane
89
Figure 5: Typical Ground Movement Vector below Proposed GSR
(a) Eastbound tunnel

The predicted maximum tunnel lining deformation of the crown and invert along the longitudinal direction for
the EB Tunnel is given in Figure 6. The tunnel lining deforms in a similar way to the surrounding ground.
Maximum deformation of the tunnel lining occurs at the area where the proposed GSR runs across the EB
Tunnel. It experiences an upward movement of around 8.5 mm and 1.8 mm at the crown and invert respectively
at this critical location, inducing a differential vertical movement of approximately 6.6 mm (moving apart). It is
also worthwhile to notice that such maximum deformation occurs at the end of excavation to the final
excavation level in Bay 2, when the excavation is directly above the tunnel.
The predicted tunnel lining deformation of the critical section for the EB Tunnel is plotted in Figure 7. The
tunnel is squeezed by approximately 4.8 mm at the end of excavation to final excavation level in Bay 2.
Y Direction (m)
-45
-40
-35
-30
-25
-20
-15
-10
-5
10
15
20
25
30
35
40
45
10
11
12
EB EB EB EB EB EB EB EB -
Original Crown
Original Invert
Deformed Crown (exc. to FEL - bay 2)
Deformed Invert (exc. to FEL - bay 2)
Deformed Crown (exc. to FEL - bay 3)
Deformed Invert (exc. to FEL - bay 3)
Deformed Crown (final consolidation)
Deformed Invert (final consolidation)
13
14
Z Direction (m)
-50
Differential vertical movement at y = -1.25m

= 6.64mm (move apart)
15
16
17
Figure 6: Predicted Lining Deformation along EB Tunnel (exaggerated by 50 times)
90
50
X Direction (m)
4
10
11
11
12
Z Direction (mbgl)
13
14
15
16
EB - Original
EB - Deformed - y=-1.25m (excavation to FEL - Bay 2)
17
EB - Deformed - y=-1.25m (excavation to FEL - Bay 3)

EB - Deformed - y=-1.25m (consolidation)
18
Figure 7: Predicted Transverse Lining Deformation (exaggerated by 100 times)
(b) Westbound tunnel

Maximum lining deformation of 4.1 mm and 1.4 mm are predicted respectively at the crown and invert along the
longitudinal direction for the WB Tunnel, inducing a differential vertical movement of about 2.7 mm (moving
apart). Again, the maximum deformation occurs at the end of the excavation to the final excavation level in Bay
2, when the excavation is directly above the tunnel. The predicted maximum differential horizontal movements
of the tunnel lining side walls is 1.9 mm with a maximum lateral movement of 2 mm occurring at the eastern
side wall of the tunnel after excavation to final level in Bay 2.
5.2 Predicted tunnel lining forces
The predicted bending moment of the tunnel lining in the transverse (hoop) direction after excavation
completion remains small throughout the GSR construction as expected. The hoop force of the tunnel lining is
reduced slightly after GSR excavation, from average 403 kN/m of the tunnel lining compression at the existing
condition down to average of 397 kN/m. Therefore the overall structural force change of the existing tunnels
associated with the proposed GSR construction is expected to be small.
5.3 Predicted long term behaviour
Figure 8(a) below presents the contours of total pore pressure along critical section (i.e. y = -1.25m) at the end of
proposed construction. Negative excess pore pressure is developed in London Clay due to clay swelling upon
unloading in the undrained condition.
Subsequent to completion of all proposed construction activities in the model, consolidation is carried out for
long term behavior. Excess pore pressure generated in previous construction stages would be dissipated within
around 10 years time. Contours of total pore pressure at the end of full consolidation are given in Figure 8(b).
91
(a) End of GSR Construction
(b) Long Term Consolidation
Figure 8: Predicted Total Pore Pressure Distribution at Different Stages
Figure 9 below shows a development of the lateral movements of the secant-pile wall at various locations
during and after GSR construction. The restraining effects on the wall deflection from the bridge deck of the
taxilane underpass are shown in the Figure 9. The predicted wall top deflection at the end of excavation from 3D
modelling is similar at various locations as indicated in Figure 9 (a). However, the long term wall top deflection
developed from less than 2mm at the decking area, to around 9.5 mm and 22 mm respectively at 4.5 m and 13 m
away from the underpass decking (see Figure 9(b)).
Predicted Horizontal Wall Deflection
at the End of Excavation(mm)
10
11 12
Depth (mbgl)
Depth (mbgl)
Predicted Horizontal Wall Deflection

at the End of Final Consolidation(mm)
10 12 14 16 18 20 22 24
10
10
12
11
11
y = 23m (13m aw ay from Deck Portion)

y = 32m (4.5m aw ay from Deck Portion, close to Section B)
y = 41m (w ithin Deck Portion)
(a)
During GSR Construction
12
y = 23m (13m aw ay from Deck Portion)

y = 32m (4.5m aw ay from Deck Portion, close to Section B)
y = 41m (w ithin Deck Portion)
(b)
After GSR Construction
Figure 9: Development of lateral movements of the secant-pile wall
92
2D analysis at section B which is about 4.5m away from the decking with a y value of 32 m (see Figure 2)
was also carried out using geotechnical software Plaxis. The numerical model boundary was placed
sufficiently remote from the excavation region and the existing tunnel structures to minimise boundary effects,
having a model of 120m wide and over 60m high with lower boundary setting at -40m AOD. The model
was restrained laterally on the 2 vertical boundaries but fixed at both horizontal and vertical directions at the
lower boundary.
The Plaxis 2D analysis adopted Hardening Soil Model with small strain stiffness (HSsmall) for London
Clay while Mohr-Coulomb soil model having properties as given in Table 4.2 was used for other soil stratus.
Conventional bottom-up construction sequence with one layer strut was followed at section B. The 2D
analysis results give a wall top deflection of 23 mm at long term compared with less than 10mm predicted by
3D analysis. The 2D analysis of the section B is unable to consider the restraining effects from the adjacent
underpass deck, hence significantly overestimate the wall movements. It is also interesting to note that, from
3D modelling, the wall top deflection reached 22 mm at the location of about 13m away from the underpass
deck, which is similar to the deflection predicted by 2D analysis at section B. Thus the restraining effect on
the wall deflection from the underpass deck decreases to negligible at a distance of about 1.5 times of the
retaining height away from the restraint. Maximum of 22 mm and 23 mm of the wall deflection were
predicted by 3D analysis and 2D modelling respectively, both occurring at wall top and having similar values.
6
CONCLUSION
Impacts on the existing live underground tunnels caused by the proposed GSR construction were assessed with
numerical modelling including a 3D modelling. The 3D model analysis results suggest that the deformation of
the segmental tunnel lining is small and the structural force change of the tunnel lining is also
insignificant. Therefore, with the ELS scheme developed and the construction in sequential bays, the likely
impact on the existing tunnels associated with the proposed GSR construction is expected to be small. Though
much more computation efforts is required for 3D modelling compared with a simpler 2D analysis, adoption of
3D modelling would be necessary for some 3D problems which is large in scale or has sensitive properties.
Comparison of the 3D and 2D analysis results for the long term behaviour of the retaining wall near the
underpass deck also illustrated the 3D modelling ability to analysis problems in a 3D environment.
REFERENCES
Itasca 2006. FLAC 3D Ver 3.1, Reference Manual.
Jardine R J, Potts D M, Fourie A B and Burland J B 1986. Studies of the influence of non-linear stress-strain
characteristics on soil-structure interaction. Geotechnique, 36(3): 377-396.
Muir Wood, A.M. 1975. The circular tunnel in elastic ground, Gotechnique, 25(1): 115127.
93
94
Outward Wall Movements Induced by Jet Grouting

L.W. WONG
SMEC Asia Ltd, Hong Kong
R.N. HWANG
Moh and Associates, Inc., Taiwan
ABSTRACT
The model on distribution of lateral earth pressure due to jet grouting is reviewed and verified
by numerical analyses. The result of studies demonstrates that the pressure increments induced
by jet grouting depend on the grouting technique adopted and range from 0.6 to 1.2 times the
total vertical soil pressure. The pressure increments due to jet grouting are inversely
proportional to the radial distance to the jet grout columns.
1
INTRODUCTION
Jet grouting has been extensively used as the supplementary measure for cut-and-cover construction of
underground structures. The soil mass could achieve undrained shear strengths ranging from 1 MPa to 4 MPa.
Jet grout columns are used as grouted slabs to act as an internal strut below the bottom excavation level to
reduce the magnitude of wall and ground movements caused by the excavation. Grout curtains formed by jet
grout columns are installed to seal the gaps where diaphragm walling is obstructed by existing utilities.
There are case histories recording that jet grouting would induce ground movements. Berry et al. (1987)
reported ground heave as large as 550 mm observed during jet grouting for the Singapore MRT bored tunnel
sections. Wong et al. (2000) and Poh et al. (2001) reported ground heave of 24 mm and lateral soil movements
as large as 35 mm due to installation of the jet grout slab for Singapore Post Centre. Chen et al. (1997)
reported jet grouting causing maximum lateral wall movements of 59 mm and 35 mm at Station BL13 (City
Hall Station) and at Station BL14 (Yong Chun Station) of Taipei MRT respectively.
This paper reviews the mechanism of ground movements caused by jet grouting. The lateral earth pressures
inducing the wall movements are assessed from the jet grouting trials. Numerical analyses are conducted on
the case history for Station BL13 to verify the magnitude of lateral earth pressures.
2
PRESSURES IN THE GROUTHOLE
Jet grouting uses high speed water jet to undermine soils then replaces and mixes the disturbed soils with
cement grout. The grouting techniques are described by Kauschinger et al. (1989). With the jetting nozzles of
2 mm to 3 mm in diameter, the jet stream has the sub-sonic speed of about 200 m/s. Although the pressures at
the grout pump range from 20 MPa to 40 MPa, the potential energy is transformed into the kinetic energy for
imparting the soil and for mixing it with cement grout. Once the fluid leaves the nozzles, the pressure
promptly drops to the atmospheric. As long as the return slurry is freely flowing to the surface, the pressure
inside the grouthole is balanced by the atmospheric pressure. Collapsing of the grouthole would however
block the passage of the return grout. Under these circumstances the pressure in the grouthole will then be
equal to the pump pressure. Buttling et al. (1988) pointed out that such blockage will cause the hydraulic
fracture mode of failure in the surrounding ground.
The magnitude of the cavity pressure in the grouthole could be assessed from the principle of fluid
mechanics. Wong et al. (1994) proposed that the cavity pressure, Pc, could be expressed by Eq. 1 to Eq. 3:
Pc = Psl + Pfr
(1)
Psl = sl H
(2)
Pfr = f H
(3)
95
Figure 1: Distribution of Pressures in the Grouthole
where Psl is the hydrostatic pressure of the return slurry, Pfr is the head loss due to friction f between the
grouthole wall and the return slurry, sl is the density of the return slurry and H is the height of the slurry in the
grouthole. Figure 1 depicts the distribution of the fluid pressures in the grouthole during jet grouting. The Pc
value may exceed the total vertical stress, vo , and cause cylindrical expansion along the grouthole wall. The
Psl and the Pfr values have been estimated from the performance of trial grouting. Sections 3 to 5 present three
case histories to verify the cavity pressures estimated by various methods.
3
CASE 1 CHUNGHO LINE
3.1 Ground conditions

Wong et al. (1994) reported the result of the jet grouting trial panels conducted at a MRT station of Chungho
Line, Taipei MRT. The ground is composed of the Sungshan Formation of 45 m in thickness overlying the
Gingmei Formation of cobbles and gravel deposits. The Sungshan Formation located at the southern portion
of Taipei Basin is mainly composed of silty clay that the N values vary with depth and range from 2 to 20. At
the depths of 22 m to 27 m, there is a silty sand layer with the N values ranging from 9 to 20. The groundwater
level is located at 1 m below ground surface.
3.2 Trial grouting
Three trial panels using the single-fluid, double-fluid and triple-fluid jet grouting techniques were conducted.
The jet grout columns were installed to as deep as 24 m. The panel using the single-fluid method composed of
32 jet grout columns of 0.9 m in diameter arranging in two rows. For trial panels using the double-fluid and
the triple-fluid methods, each panel composed of 9 inclined jet grout columns of 1.5 m and 1.6 m in diameter
with the angles to the vertical varying from 44o to 59o.
The friction loss values along the grouthole wall have been estimated with the velocity of the upward slurry
flow, the dimension of the annular void between the grouthole wall and the monitor as well as the viscosity of
the return slurry. Table 1 summarizes the jet grouting parameters for the various methods. As the triple-fluid
technique uses compressed air to sheath the jet stream of cement grout, its return slurry had lower unit weight.
Jet grouting
method
Single-fluid
Double-fluid
Triple-fluid
Table 1: Estimation of Friction Loss in Grouthole during Jet Grouting

Unit weight of
Diameter of
Diameter
Rate of
Velocity of
return slurry
grouthole
of monitor
fluid injected
slurry flow
kN/m3
mm
mm
l/s
m/s
15.7
180
90
2.9
0.15
15.3
146
90
30
1.4
13.4
146
90
33
1.6
96
Friction along
grouthole
kN/m2/m
9.5
18.3
10.2
The viscosity for the return grout for the triple-fluid, measured with the Marsh funnel, was 53 centi-poise.
The return slurry for the single and the double-fluid were too viscous to flow through the funnel. The
estimated initial shear was 0.13 kN/m2. The estimated frictions per metre length of the groutholes were 9.5
kN/m2, 18.3 kN/m2 and 10.2 kN/m2 for the single, double and the triple-fluid methods respectively. It is noted
that the flow rate for the single-fluid was 2.9 l/s and that for the triple-fluid was 33 l/s. The uncased grouthole
for the single-fluid was 180 mm in diameter, which is larger than the 146 mm for the triple-fluid system.
Therefore the return slurry of the single-fluid system has a bit lower friction loss.
3.3 Horizontal pressure increment
The Psl and the Pfr values are normalized with the total vertical stress, vo , of the ground. As summarized in
Table 2, the Pc /vo ratio for the single, double and the triple-fluid methods are about 1.4, 1.9 and 1.3
respectively. Results of this trial grouting show that the cavity pressure, Pc , in the groutholes during jet
grouting range from 1.3vo to 1.9vo , which are far less than the pumping pressure ranging from 20 MPa to 40
MPa. The in-situ horizontal pressure, ho , is defined as ho = K vo where K is the earth pressure coefficient.
The lateral pressure increment, Ph , due to jet grouting immediately next to the wall of the grouthole is
expressed in Eq. 4:
Ph = Pc - ho
(4)
The K value could be taken as 0.5 for ground prior to jet grouting. However, after the first jet grout column
is installed, the stiffness of the surrounding ground is improved. The K value of 0.9 could be adopted. Relating
the Ph value with the vo value and adopting an average K value of 0.7, the Ph /vo ratio due to jet grouting
using single, double and triple-fluid methods would be 0.7, 1.2 and 0.6 respectively. The average Ph /vo
ratios for various types of jet grouting are summarized in Table 2.
Jet grouting
method
Single-fluid
Double-fluid
Triple-fluid
Table 2: Estimation of Cavity Pressure in Grouthole during Jet Grouting

Unit weight of
Friction along
Unit weight
Pc /vo =
return slurry
grouthole
of soil
( sl + f ) /s K = 0.5
f , kN/m2/m
sl , kN/m3
s , kN/m3
15.7
9.5
18
1.4
0.9
15.3
18.3
18
1.9
1.4
13.4
10.2
18
1.3
0.8
Ph /vo
K = 0.9
Average
0.5
1.0
0.4
0.7
1.2
0.6
CASE 2 SINGAPORE POST CENTRE
4.1 Observation of pressure increments

The basement excavation of the Singapore Post Centre involved extensive jet grouting to improve the soft
marine clay presence within the excavation of 88 m in width. Poh et al. (2001) reported the results of
monitoring on the jet grouting trial. As depicted in Fig. 2, the field trial comprised 18 jet grout columns of 1.8
m in diameter and 20 m in depth. The top and the bottom levels for the jet grout mass were 93 m and 84 m
respectively. Arranged in 5 rows, the jet grout columns had a triangular spacing of 1.55 m. The centerline for
the Row 1 columns had a distance of 0.45 m to the three diaphragm wall panels of 0.8 m in thickness, 32 m in
depth and 18.6 m in total length. The triple-fluid system was used for installing the jet grout columns.
The soil consists of 4 m of fill overlying a 13 m thick soft marine clay stratum. The marine clay is
underlain by a 5 m thick stiff silty clay, a 3.5 m thick loose clayey sand and a 3.5 m thick stiff silty clay layer
that is underlain by hard clayey sandy silt of the Old Alluvium Formation. The unit weights for fill and for the
marine clay are 17 kN/m3 and 15 kN/m3 respectively. The groundwater level was at 102.0 m.
The ground movements induced by trial grouting were monitored by eleven inclinometer casings in soil
and by one inclinometer in diaphragm wall. The lateral pressures were measured by eleven total pressure cells,
six pneumatic piezometers and two standpipe piezometers. As shown in the section in Fig. 2, the total pressure
cells and the pneumatic piezometers were installed at the mid-depth of the jet grout columns and at the level
of 88.5 m. The instruments were installed on both sides of the diaphragm wall. It is noted that four pressure
97
Figure 2: Plan and Section for Jet Grout Columns, Pressure Cells and Piezometers - Case 2
cells, located at 20 m and 30 m east and west from the centre of the mid diaphragm wall panel, are not shown
in the plan for sake of clarity.
The numbers of the jet grout columns are the sequence of their installation. Two columns were installed in
each day. Columns no. 1 and 2 were installed on Day 1 and columns no. 17 and 18 were on Day 9. Instrument
readings were taken after completion of jet grouting for each column.
4.2 Distribution of pressure increments
During the jet grouting trial variation in lateral earth pressures and in piezometric pressures were observed. As
the readings were taken after completion of each jet grout column, the peak pressure increments occurred
during grouting may not be observed. The monitoring results are presented in Fig. 3, showing the pressure
increments are inversely proportional to the radial distances to the jet grout columns. The pressure increments
versus the radial distances could be expressed by Eq. 5 and Eq. 6:
Pr = P0 r0 / r
(5)
ur = u0 r0 / r
(6)
where Pr and ur are the lateral earth pressure increment and the piezometric pressure increment at the radial
distance r to the jet grout column respectively, P0 and u0 are the lateral earth pressure increment and the
piezometric pressure increment respectively at the radius, ro , which is the radius of the jet grout column.
98
100
JGC no.
80
Pore pressure increment, ur , kPa
Earth pressure increment, Pr , kPa
1 to 6
7 to 18
60
Pr = P0 ro/ r
P0 = 200 kPa, r0 = 0.9 m
40
P0 = 50 kPa
20
0
0
5
10
15
20
Normalized radial distance, r / r0
100
JGC no.
80
1 to 6
7 to 18
ur = u0 ro/ r
u0 = 200 kPa, r0 = 0.9 m
60
40
u0 = 50 kPa
20
0
0
5
10
15
20
Normalized radial distance, r / r0
Figure 3: Variation of Earth and Piezometric Pressure Increment with Distance to Jet Grout Column
The jet grout columns installed in the early stage induced larger pressure increments than those in the later
stage. Columns no. 1 to 6 were installed on Day 1 to Day 3 while columns no. 7 to 18 were installed on Day 4
to Day 9. As shown in Fig. 3, jet grouting for columns no. 1 to 6 and for 7 to 18 induced the P0 values of 200
kPa and 50 kPa respectively. Similarly, installing jet grout columns in the early and in the later stages induced
u0 values 200 kPa and 50 kPa respectively. It appears that as jet grouting proceeds, the ground becomes
stiffer and stiffer. In the later stage the pressure increments are about 25 % of those induced in the early stage.
The same value of P0 and u0 is an indication that the ground deformation is in undrained condition. This
finding agrees with Poh et al. (2001) interpretation that there is a linear relationship between the earth pressure
increment and the pore water pressure increment.
The pressure increments are measured at the depth of 15 m. The vo value at that depth is 233 kPa.
Adopting the unit weight and the friction loss values established from Case 1 and summarized in Table 2, the
Psl and the Pfr values for the triple-fluid method in Case 2 would be 201 kPa and 153 kPa respectively.
Deducting the ho value of 116 kPa, the Pc value induced by jet grouting would be about 238 kPa.
The P0 value of 200 kPa for early stage jet grouting as inferred from Fig. 3 would be an indication of
dissipation of cavity pressure. Since the pressure increments were presumably measured at 1 hr to 2 hr after
completion of jet grouting, the increments had been dropped by 16 %, from the estimated 238 kPa to the
inferred pressure of 200 kPa.
5
CASE 3 CITY HALL STATION
5.1 Ground conditions

The case history of using the jet grout slab to minimize ground movements outside diaphragm wall was
reported by Chen et al. (1997). Station BL13, the City Hall Station, Nankang Line of the Taipei MRT, is
located on the eastern side of the Taipei Basin. The ground level is 106.8 m. The soil deposits belong to the K1
Zone of the Sungshan Formation underlain by the Gingmei Formation. In this project site the Sungshan
Formation is dominated by soft clay of 50 m in thickness. The underlying Gingmei Formation, encountered at
the level below 56 m, is composed of sandy gravel and cobbles.
Station BL13 is 24 m in width and 278 m in length. The excavation of 18.8 m in depth was supported with
diaphragm walls of 1.2 m in thickness and 44.5 m in length. Five levels of struts preloading to 50% of the
design load were installed. According to Chen et al. (1997), it was anticipated that the cut-and-cover
construction would cause lateral wall deflection of about 80 mm. In order to reduce wall deflections and to
minimize the settlement effect to the adjacent buildings, a jet grout slab of 4 m in thickness was installed.
Figure 4 shows the extent of the jet grout slab for Station BL13.
Thirteen inclinometers casings were installed at the Station BL13 area. Seven of them were installed in
diaphragm wall and six were in ground. The toes of the inclinometers installed in wall were 3 m below the
toes of the diaphragm wall. Inclinometer no. SID7 was installed in the diaphragm wall at Station BL13.
99
104.8 m
SID7
-20
Wall deflection, mm
0
20
40
60
80
110
Movement toward
wall back
100
Elevation, m
Jet grout slab
90
80
70
60
50
Figure 4: Soil profile for Station BL13 - Case 3
Jet grout
slab
Station BL 13
SID7
Install JGP
Figure 5: Observed wall deflection profile
5.2 Jet grout slab

Jet grouting for Station BL13 was reported by Wong et al. (1997). The jet grout slab was mainly formed by ten
rows of JSG columns spacing at 2.2 m centre-to-centre with a triangular pattern. JG columns of 1.2 m in
diameter were installed along the diaphragm wall at 1.04 m spacing to fill the gap between the wall and the
JSG columns. The JSG and the JG columns were installed by the double and the single-fluid methods
respectively. During installation of the JSG and JG columns, the pumping pressures were as high as 40 MPa
with the rate of injection 200 l/min. Chen et al. (1997) reported that due to traffic diversion arrangement, the
jet grout slab was installed in bays. Each bay was divided into 3 zones. Grouting commenced at the northern
zone, followed by the central and finally at the southern zone. In order to avoid the return slurry from flowing
outside the works area, 2 m to 2.5 m depth of excavation was conducted prior to jet grouting.
5.3 Pressures due to jet grouting
As shown in Fig. 5, installation of the jet grout slab induced outward wall deflections as large as 59 mm in
SID7. Apparently the pressure increments due to jet grouting caused the wall deflections. Since the wall
deflections are proportional to the cavity pressure, the total wall deflection could be computed by
superimposing the movement effect due to all jet grout columns taking into account the various distances to
the diaphragm wall, the pressures in the groutholes and the soil strengthening effect caused by jet grouting.
Calculation for wall deflections caused by jet grouting would comprise the following procedures:
(1) Estimation on the cavity pressure in a jet grout column due to jet grouting. The cavity pressure induced
by a single jet grout column has been established in Case 1and summarized in Table 2.
(2) Estimation on the pressure transmitted from a jet grout column to the diaphragm wall during jet
grouting. The pressure increment versus radial distance relationship has been developed in Case 2 and
expressed by Eq. 5.
(3) Superimposing the cavity pressures acting on the diaphragm wall caused by jet grout columns with
various distances to the wall.
(4) Using the cumulative pressures to conduct the numerical analysis to obtain the total wall deflection.
Due to various reasons, the procedure in Item 3 would be difficult. Superimposing the jet grout columns in
the entire grouted slab would come to large cumulative pressure or large magnitude of wall deflection. Take
Case 3 as an example, the 12 rows of JSG and JG columns would end up with unrealistic large cumulative
pressure increment. With the ground level at 104.8 m inside the diaphragm wall and the bottom level of the jet
grout slab at 84 m, the vo value at the base of the slab was about 381 kPa. Adopting the Ph /vo ratios of 0.7
100
and 1.2 summarized in Table 2 for the single and the double-fluid methods, the minimum Ph values in a
single grouthole would be 267 kPa and 457 kPa for the JG and the JSG columns respectively. Superimposing
the 12 rows of columns would get the total pressure increment as large as 5,100 kPa. Inputting such a large
cumulative pressure for numerical analysis would obtain wall deflection 10 times of the observed value.
In line with the procedure given in Item 2 above, it is proposed that reduction factors could be applied on
the cumulative pressure. As observed in Case 2 and shown in Fig. 3, the pressure transmitted from the
grouthole to the point of interest is reduced in the later stage. Among the 18 jet grout columns, 1/3 of them (no.
1 to 6) induced P0 and 2/3 of them (no. 7 to 18) induced 0.25P0. The weighted average for the entire 18
columns would be 0.5P0. Further than that, the jet grout slab in Station BL13 was implemented in 3
sub-zones. During installation of the columns in the later zones, the surrounding soil had already been
improved by jet grouting in the earlier stages. The larger in ground stiffness, the smaller in radial strains
would be induced. The pressure reduction factor F1 = 0.5 is allowed for the ground stiffening effect.
The jet grout slab in Station BL13 is 4 m in thickness. The groutholes above the top level of the slab for JSG
is 146 mm in diameter. At the radial distance of 1.27 m, the influence for pressure increment applying Eq. 5 is
146/1270 = 0.11, which is negligible small. As depicted in Fig. 8, the area for the trapezoidal pressure of Ph
and 4 m in height is equivalent to a triangular distributed pressure of 0.4Ph and 21 m in height. The pressure
reduction factor F2 = 0.4 is allowed for the effect of jet grout slab with limited thickness.
Thirdly, the pressure imposing on the wall would be reduced as the wall moved backward. The reduction
factor however could be assessed by numerical analyses. It is proposed that the pressure reduction factor F3 =
0.5 shall be allowed for the wall movement effect.
Applying the reduction factors F1 to F3 , the total reduction in cumulative pressure increment would be 0.5 x
0.4 x 0.5 = 0.1. The estimated cumulative pressure increment of 5,100 kPa is then reduced to 510 kPa. In the
numerical analysis, the Ph values of 400 kPa and 500 kPa acting on the diaphragm wall have been adopted.
The triangular distribution of earth pressure increment adopted for analysis is depicted in Fig. 6.
5.4 Numerical analyses
The wall deflection profile for inclinometers no. SID7 located at Station BL13 has been selected for numerical
analyses. Back-analyses have been conducted to estimate the earth pressure increment due to jet grouting.
This is achieved by comparing the results obtained in back-analyses with the observed performance of wall.
For the case of interest, analyses have been performed by using the finite element program PLAXIS (PLAXIS
BV, 2011) developed at Delft University. Figure 6 shows the finite element mesh adopted in the analyses.
Since Gingmei Formation is a competent stratum and ground movements therein are expected to be small, the
base of the finite element mesh is placed at a depth of 5 m below the surface of this formation.
The ground strata are modeled by 15-node elements. The Mohr-Coulomb model is adopted for simulating
the behavior of the soil of the Sungshan Formation and the Gingmei Formation. The clayey sub-layers II, III
and IV of the Sungshan Formation are modeled as the Undrained Type C material. The Gingmei Formation is
modeled as the Drained Type material.
Triangular
pressure Ph
400 kPa &
500 kPa
106.8 m
IV-1
IV-2
IV-3
84.0 m
Sublayer IV
IV-4
IV-5
IV-6
Jet grout slab
IV-7
IV-8
Diaphragm wall
Sublayer III
Sublayer II
Gingmei Formation
Figure 6: Finite Element Mesh Adopted in Numerical Analysis
101
30 m
Soil strata
Sub-layer
IV
IV-1
IV-2
IV-3
IV-4
IV-5
IV-6
IV-7
IV-8
Sub-layer III
Sub-layer II
Gingmei layer
Table 3: Soil Parameters Adopted in Numerical Analysis

Unit
Effective strength
AverUndrained
Undrained
weight Cohesion Friction
age
shear
Youngs
N value
strength
modulus
angle
kN/m3
Eu, MPa
su, kPa
kPa
degree
18.6
0
28.0
3
55
11
18.6
0
28.0
2
59
12
18.8
0
28.5
3
63
13
18.8
0
28.5
3
78
16
18.8
0
29.0
4
105
21
19.4
0
29.0
5
131
26
19.4
0
29.5
6
158
32
19.4
0
29.5
14
184
37
19.4
0
30
20
211
42
19.4
0
31
26
237
48
20.0
0
42
100
100
Poissons
ratio
u ,
0.495
0.495
0.495
0.495
0.495
0.495
0.495
0.495
0.495
0.495
0.35
The diaphragm walls are simulated by plate elements and the Youngs modulus for concrete, the Ec value,
of 25,000 MPa is adopted with the characteristic strength of 42 MPa. As suggested by Gaba et al. (2003), the
stiffnesses of the diaphragm wall, the Ec I and the Ec Ac values, where I and Ac are the moment of inertia and
the sectional-area of concrete respectively, are reduced by 30 %, giving a value of 2,520 MN-m for the former
and 21,000 MN/m for the latter to account for the influence of tremieing and degradation of concrete.
The Youngs modulus of the soil, Eu , is assessed by the empirical relationship with the su value of the
clayey Sungshan deposits:
Eu / su = 200
(7)
The soil parameters adopted in the analyses are summarized in Table 3.

5.5 Piezometric pressures
Extensive dewatering in the Taipei basin was conducted during the period between 1960 and 1990. The
groundwater drawdown conditions have been summarized by Hwang et al. (2013). Due to past pumping in the
Taipei Basin, the piezometric level in the Gingmei Formation was as low as 55 m in 1975. Since then the
piezometric level started to recover. In 1994 the piezometric level in the gravel layer recovered to about 90 m.
Piezometric pressure, kPa

100 200 300 400
110
Equivalent
triangular
pressure at r0
1995
Hydrostatic
1994
100
Elevation, m
500
Pressure
increment at
distance r0
90
Jet grout slab
80
70
60
0.4Ph
BL13
P h
50
Figure 7: Water Pressure Outside Excavation area
Figure 8: Pressure Increment at Distance r0 to a Grouthole
102
-20
Wall deflection, mm
20
40
60
80
110
500 kPa
Elevation, m
100
Ph = 400 kPa
90
80
SID7
400 kPa
500 kPa
Observed
70
60
Figure 9: Observed and Computed Wall Deflection Profiles Due to Jet Grouting
Chen et al. (1997) reported the distribution of piezometric levels in the Sungshan Formation. The
groundwater at the uppermost strata between the elevations of 106.8 m and 99.0 m had the piezometric level
of 105 m in the early stage of excavation in around 1994. Below the elevation of 99.0 m, the groundwater
pressure was non-hydrostatic, gradually reducing from the piezometric level of 105.0 m (65 kPa) at the
elevation of 90 m to the piezometric level of 90 m at the base of Sungshan Formation at the elevation of 56.8
m (332 kPa). Figure 7 depicts the water pressures adopted in the numerical analyses.
5.6 Result of numerical analyses
The calculated wall deflection profiles under Ph values of 400 kPa and 500 kPa are presented in Fig. 9.
The observed wall deflection profile for SID7 could be best matched with the calculated deflection profile
using the Ph value of 450 kPa. The reduction factors proposed in Section 5.3, in particular, the factor F3, are
verified by the numerical analysis.
It is however noted that more case histories on jet grouting should be collected and studied so that the
various contributing effects to wall movements could be identified.
6
CONCLUSIONS
Case histories on jet grouting causing lateral ground movements have been critically reviewed. A model on the
distribution of lateral earth pressure has been established. The factors for transmitting the pressure increment
from the grouthole wall to the point of interest have been verified by numerical analyses. Based on results of
this study, the following concluding remarks could be drawn:
(1) The lateral ground movements caused by jet grouting is induced by static pressure of the return slurry
emerging from the grouthole and by friction loss between the return slurry and the grouthole wall.
(2) Depends on the grouting techniques adopted for installing the jet grout columns, the total pressure in
the grouthole during jet grouting would range from 1.3 to 1.9 times the total vertical stress. Therefore
the pressure along the wall of the jet grout column is far less than the pumping pressure ranging from
20 MPa to 40 MPa.
(3) The pressure increment that will cause the lateral ground or wall movements depends on the jet
grouting technique. Assuming the in-situ horizontal earth pressure is about 0.7 times the total vertical
pressure of the surrounding ground, the pressure increments induced by the single, double and the
triple-fluid methods would be about 0.7, 1.2 and 0.6 times the total vertical pressure.
(4) The pressure increments due to jet grouting are inversely proportional to the radial distances to the jet
grout columns.
(5) Due to ground improvement effect, the pressure increments transmitting from the groutholes installed
103
in the later stage would be about 25 % of those installed in the early stage.
(6) The less thickness of the jet grout slab, the less cavity pressure would be transmitted from the
grouthole to the point of interest.
(7) Result of preliminary analysis suggests that the outward wall movement would cause pressure
reduction transmitting from the groutholes.
Understanding the magnitude of the lateral earth pressure increment induced by jet grouting, the ground
movement effect to adjacent ground and to nearby underground structures such as diaphragm wall, sheet pile
wall, basements, pile foundations and tunnels could then be reliably assessed. With the aid of numerical
analyses, application of jet grouting for jacking the diaphragm wall outward and to minimize the ground
settlements behind diaphragm wall could be properly controlled.
REFERENCES
Berry, G.L., Shirlaw J.N., Hayata, K & Tan, S.H. 1987. A review of grouting techniques utilised for bored
tunneling with emphasis on the jet grouting method. Proc. Singapore Mass Rapid Transit Conference, April.
Buttling, S. & Shirlaw, J.N. 1988. Review of ground treatment carried out for tunnels of the Singapore Mass
Rapid Transit System. Proceedings, Tunnelling 88, Institution of Mining and Metallurgy, London, 1988.
Chen Y.K., Huang C.C. & Wang, F.G. 1997. Use grouted raft for building protection of excavation in soft clay.
Proceedings of 7th Conference on Current Research in Geotechnical Engineering, Chishan, Taiwan, August,
Vol. 1, pp 593-600. (in Chinese)
Gaba, A.R., Simpson, B., Powrie, W. & Beadman, D.R. 2003. Embedded Retaining Walls Guidance for
Economic Design, CIRIA Report, C580, London.
Hwang, R.N., Moh, Z.C. & Hu, I.C. 2013. Effects of consolidation and specimen disturbance on strengths of
Taipei Clays. Geotechnical Journal of the SEAGC & AGSSEA, 44(1): 9-18.
Kauschinger, J.L. & Welsh, J.P. 1989. Jet grouting for urban construction, Proc. of the 1989 Seminar, Design,
Construction and Performance of Deep Excavations in Urban Areas, MIT Cambridge, MA.60pp.
PLAXIS BV 2011. Reference Manual PLAXIS BV: Amsterdam, the Netherlands.
Poh, T.Y. & Wong I. H. 2001. A field trial of jet grouting in marine clay. Canadian Geotechnical Journal, 38:
338-348.
Wong I. H. & Poh, T.Y. 2000. Effect of jet grouting on adjacent ground and structures. Journal of
Geotechnical and Geoenvironmental Engineering, 126(3): 247-256.
Wong, L.W. & Hwang R.N. 1997. Evaluation of jet grouting by in-situ tests. Proceedings of the International
Conference on Ground Improvement Techniques, Macau, May, pp641-647.
Wong, L.W. & Kao, H.S. 1994. Failure mechanism induced by jet grouting in cohesive soils,
Sino-Geotechnics, 47, pp107-120 (in Chinese).
Wong, L.W., Shirlaw J.N. & Kao, H.S. 1994. Application of jet grouting in geotechnical engineering. Proc. of
International Structure and Foundation Symposium, Hanzhou, October (in Chinese).
104
Effects of Soil-structure Interactions on the Design of ELS

Systems
Albert T. Yeung & Iris C.Y. Ng
University of Hong Kong, Hong Kong
ABSTRACT
The conventional design methodology for the excavation and lateral support (ELS) system
maintains the equilibrium of lateral earth pressures and support forces in the design. It does not
account for the stiffness of the lateral support system and its effects on the magnitudes of the
lateral earth pressures. However, it is well known that the magnitude of the lateral earth pressure,
either active or passive, depends on the lateral movement of the lateral support system. Therefore,
using the conventional design methodology may not yield the most efficient or appropriate
design. More importantly, it may not be able to give a reasonable estimate of the lateral
movement of the lateral support system and the resulting ground settlement in the vicinity of the
excavation. In this paper, the effects of these soil-structure interactions on the design of ELS are
evaluated and presented. In particular, the effects of the stiffness of the lateral support system on
the required embedment depth, strut loads, and bending moments in the lateral support system are
evaluated.
1 INTRODUCTION
The excavation and lateral support (ELS) system is an essential component for the construction of
foundations, basements or other underground structures. The system provides temporary support to maintain
stability of the excavation. The conventional design methodology adopts the concept of limit equilibrium
among the active earth pressures, passive earth pressures and strut loads. The required embedment depth, and
bending moments and shear forces in the lateral support system, i.e. sheetpiles, pipe piles etc. are determined
and designed accordingly. The stiffness of the lateral support system is not explicitly taken into account.
However, the stiffness of the lateral support system may affect the magnitudes of the lateral earth pressures, as
the lateral earth pressure is a function of the magnitude of deflection of the lateral support system. Therefore,
a conservative approach is normally taken in practice. Current design software can take into account of the
lateral earth pressure as a function of the deflection of the lateral support system, and iterate the lateral earth
pressures to obtain the equilibrium forces. As a result, the effects of soil-structure interactions are taken into
consideration for the design implicitly.
In this paper, the effects of these soil-structure interactions on the design of ELS are evaluated and
presented. In particular, the effects of the stiffness of the lateral support system on the required embedment
depth, strut load, and bending moment in the lateral support system are evaluated.
2 BACKGROUND
2.1 Traditional approach
Typical temporary ELS systems include cantilever walls, anchored or propped walls and braced excavations,
as classified by the earth retaining mechanism (Knappett & Craig 2012).
Cantilever walls are used only when the retained height of soil is relatively low. The stability of the wall is
provided entirely by the passive resistance of soil mobilized in front of the wall. The mode of failure is by
rotation about a point O near the wall toe as shown in Figure 1(a). Passive resistance acts in front of the wall
above O and behind the wall below O as shown in Figure 1(b), thus providing a restoring moment. The design
is generally based on the simplification shown in Figure 1(c), assuming the net passive resistance below point
O is represented by a concentrated force R acting at a point C which is located slightly below O, at depth d
below the excavation level. Using the traditional method of analysis, the embedment depth d is determined by
105
equating the moments about C. A factor of safety F is applied to the available passive resistance in front of the
wall. The value d so determined is then increased by 20% to allow for the simplification taken in the method
of analysis. The concentrated force R can be obtained by equating the horizontal forces. It is advisable to
check that the net passive resistance available over the additional 20% embedded depth is equal to or greater
than R. The shear force and bending moment distributions in the wall are then determined for subsequent
selection of the appropriate structural section for the wall.
Figure 1: Cantilever Sheetpile Wall
Anchored and propped walls are usually constructed of steel sheetpiles, steel pipe piles or reinforced
concrete diaphragm walls. Additional support to embedded walls is provided by a row of tie-backs (anchors)
or props near the top of the wall, as shown in Figure 2(a). It is assumed that the depth of embedment below
the excavation level is insufficient to produce fixity at the wall toe. The wall is thus assumed to be free to
rotate about its toe, the resulting bending moment distribution is shown in Figure 2(b). To satisfy the
equilibrium conditions, the sum of the restoring moments about the anchor or prop must be greater than or
equal to the sum of the overturning moments. The anchor or prop force is determined from the equilibrium of
horizontal forces. The shear force and bending moment distributions in the wall are then determined for
subsequent selection of the appropriate structural section for the wall.
Figure 2: Anchored Sheetpile Wall: Free Earth Support Method
Figure 3: Relationship between Lateral Strain and

Lateral Pressure Coefficient
In both cases, the structural section for the wall is selected after the shear force and bending moment
distributions in the wall have been determined from limit equilibrium conditions. Therefore, the stiffness of
106
the wall has no effect on the lateral earth pressure, shear force and bending moment distributions, and soilstructure interactions are not taken into consideration at all.
The coefficients of active and passive lateral pressures used in these limit equilibrium analyses are
obtained from the lower bound limit analyses, assuming the soil mass is subjected to lateral expansion (active
case) or compression (passive case) and reaches a state of plastic equilibrium. In the active case, a specific
(minimum) value of lateral expansion strain would be necessary for the development of the active state within
the soil mass. If the deformation of the wall cannot satisfy this minimum deformation requirement, the soil
adjacent to the wall would not reach a state of plastic equilibrium and the lateral pressure would be between
the active and at-rest values. In the passive case, a specific (minimum) value of lateral compression strain
would be necessary for the development of the passive state within the soil mass. In practice, only part of the
potential passive resistance would normally be mobilized. The relatively large deformation necessary for the
development of full passive resistance would be unacceptable. As a result, the lateral earth pressure under
working conditions would be between the at-rest and passive values. Experimental evidence indicates that the
mobilization of full passive resistance requires a wall movement of the order of 2-4% of embedded depth in
the case of dense sands and of the order of 10-15% in the case of loose sands. The corresponding percentages
for the mobilization of active pressure are of the order of 0.25 and 1%, respectively. Generally, for any
condition intermediate to the active and passive states, the value of the coefficient of lateral earth pressure is
unknown. The form of the relationship between the strain and the coefficient of lateral earth pressure is shown
in Figure 3. It is evident that the lateral earth pressures acting on the ELS system depend on the lateral
movement of the ELS system which depends on the stiffness of the lateral support system. These important
effects of soil-structure interactions are not taken into account in the traditional design of the ELS system
using limit equilibrium.
2.2 FREW
FREW (Flexible REtaining Walls) is a computer software developed by Oasys Ltd., which enables engineers
to define and solve complex embedded flexible retaining wall design problems efficiently. The program
performs two-dimensional pseudo-finite element analyses to allow the user to study the deformations of, and
stresses within, the structure through a specified sequence of construction (Oasys Ltd. 2014). The wall is
modeled as a series of nodal points connected by elastic material elements. Lateral earth pressures acting the
lateral support system, displacements, bending moments, shear forces and strut (or anchor) forces occurring in
the system during each stage in construction can be calculated by the program. The wall deflection is
governed by the wall stiffness and the lateral earth pressure (Wang et al. 2013), while the lateral earth
pressure is controlled by the wall deflection as shown in Figure 3. These soil-structure interactions are thus
inter-related. However, a realistic analysis should take these soil-structure interactions into consideration. The
approach of the pseudo-finite element analysis is completely different from that of the limit equilibrium
analysis. A detailed comparison of the two different analyses is tabulated in Table 1.
3 CASE STUDIES
The results of a few case studies on ELS systems using sheetpiles with or without lateral struts are presented
in this paper to demonstrate the effects of soil-structure interactions on the design of ELS systems. The
geometry of the ELS system used in these case studies is shown in Figure 4. The depth of excavation is 6 m
and the width of the excavation is 18 m. The groundwater table is at 1.5 m below the existing ground level.
The lateral struts, if installed, are located at 1 m below the existing ground level. The embedment depth of the
sheetpiles required to maintain stability of the ELS system is denoted as d in the figure. The king posts for
internal support of the lateral struts are omitted for clarity.
In the analyses, the coefficients of active and passive lateral earth pressure are taken to be 0.3 and 4.2,
respectively. The bulk unit weight of soil and water are taken to be 20 kN/m3 and 9.81 kN/m3, respectively.
Sheetpiles of different flexural rigidities as tabulated in Table 2 are used in the analyses.
107
Table 1. Comparison between Limit Equilibrium and Pseudo-Finite Element Analyses

(Excerpted From Gaba Et Al. 2003)
Type of
Analysis
Advantages
Limitations
Limit
equilibrium
Needs only the soil strength

Simple and straightforward
Does not model soil-structure interaction, wall

flexibility and construction sequence
Does not calculate deformations. Hand
calculations of deformations possible by
relating mobilized strength, soil shear strain
and wall rotation (rarely done); or through
empirical databases
Statically indeterminate systems (e.g. multipropped walls), non-uniform surfaces and
berms require considerable idealization
Can model only drained (effective stress) or
undrained (total stress) conditions
Two-dimensional only
Results take no account of pre-excavation
stress state
Pseudofinite
element
e.g. FREW
Full soil-structure interaction analysis is

possible, modeling construction sequence, etc.
Soil modeled as an elastic solid with soil
stiffness matrices calculated using a finite
element program
Soil-structure interaction taken into account
Wall movements are calculated
Relatively straightforward
Takes account of pre-excavation stress state
Two-dimensional only
Limited to linear elastic soil model, with active
and passive limits
Berms and certain structural connections are
difficult to model
Global effects not modeled explicitly
Ground movements around wall are not
calculated
Type
of Sheetpile
FSP III
FSP IV
FSP V
Table 2. Flexural Rigidity of Sheetpiles

Second Moment of Area I
Young's Modulus E (kPa)
(cm4/m)
205,000,000
205,000,000
205,000,000
Flexural Rigidity EI
(kN-m2/m)
16,800
38,600
63,000
Figure 4. Geometry of ELS System Used in Case Studies (Not to Scale)
108
34,440
79,130
129,150
The levels of the groundwater tables on the two sides of the sheetpile wall are different after excavation.
Assuming steady state seepage conditions have developed and maintained, the pore water pressure
distributions on the two sides of the wall will be unbalance as shown in Figure 5(a). The dotted lines indicate
the hypothetical and unrealistic hydrostatic pressure distributions. A flow net analysis on the steady state
seepage condition can be performed to determine the net pore water pressure distribution. A simplified pore
water pressure distribution as shown in Figure 5(a) was proposed by Padfield and Mair (1984), assuming the
total hydraulic head dissipates linearly along the back and front wall surfaces between the two groundwater
table levels.
(a) Pore Water Pressure across a Wall
(b) Net Pore Water

Pressure under Seepage
Figure 5: Distribution of Pore Water Pressure across a Retaining Wall
The maximum net pore water pressure occurs at the level of the lower groundwater table level as shown in
Figure 5(b) and is given by
=
2( +
)( )
(1)
2 +
where w = unit weight of water.

The seepage force j can thus be estimated by
j=
( +
)
(2)
2 +

The effective unit weight of the soil below the groundwater table is increased to (' + j) behind the wall
where seepage is downwards, and reduced to (' j) in front of the wall where seepage is upwards. These
values should be used in the calculation of active and passive pressures if steady seepage is maintained. It
should be noted that the assumption of hydrostatic pressures on both sides of the wall is generally
conservative as the net pore water pressure acting on the wall is excessive. However, the approach
underestimates the active lateral earth pressure behind the wall and overestimates the passive lateral earth
pressure in front of the wall simultaneously. Therefore, the use of hydrostatic pressures for ELS design may
not be conservative all the times.
3.1 Case study 1 cantilever retaining wall
Sheetpiles are installed for a cantilever retaining wall as shown in Figure 1. The required embedment depth d
determined by limit equilibrium using the Burland-Potts-Walsh approach (Burland et al. 1981) is 8.85 m when
the factor of safety is unity. The deflections of the wall constructed of different types of sheetpiles computed
by FREW for the embedment depth of 9 m are depicted in Figure 6. The maximum deflections of FSP III, IV
and IV sheetpiles are 954.5 mm, 451.8 mm and 295.0 mm, respectively. These deflections are unacceptably
large. The relative maximum deflection is thus 3.24 : 1.53 : 1.00. The relative flexural rigidities of these
109
sheetpiles as tabulated in Table 2 is 1.00 : 2.30 : 3.75. If the active lateral earth pressures on these different
types of sheetpiles are identical, the deflection should be inversely proportional to the flexural rigidity of the
sheetpile, i.e. the product of the relative maximum deflection and the relative flexural rigidity should be a
constant. However, this is not the case as revealed by the results of the analyses. In fact, the products for FSP
III, IV and V sheetpiles are 3.24, 3.52 and 3.75, respectively, indicating the relative maximum deflection
increases with the flexural rigidity of the sheetpile.
It can be observed in Figure 3 that the coefficient of active lateral earth pressure decreases with deflection
of the wall. When the wall is relatively more rigid, the active earth pressure acting behind the wall is
increased, resulting in a relatively larger deflection. Similarly, the passive earth pressure acting in front of the
wall is smaller, resulting in less passive resistance supporting the cantilever retaining wall. Therefore, the
effect of soil-structure interactions makes a more rigid sheetpile less efficient in terms of deflection control.
However, it should be noted that the absolute deflection decreases with increase in flexural rigidity of the
sheetpile.
The results also demonstrate the deficiencies of the limit equilibrium approach. Firstly, the required
embedment depth so determined is independent of the flexural rigidity of the sheetpiles used. It is evidently
demonstrated in Figure 6 that the deflection of the sheetpile depends heavily on the flexural rigidity of the
sheetpile. Secondly, the approach does not give an estimate of the lateral movement of the ELS system, which
may cause excessive settlement and/or distress of nearby structures. As demonstrated by the results of this
case study, the minimum embedment depth may yield a maximum deflection of 295 mm even when FSP V
sheetpiles are used.
It is recommended that the factor of safety of 1.2 should be applied to the calculated embedment depth
(Knappett & Craig 2012). As a result, the design embedment depth would be 1.2 8.85 = 10.62 m. The
deflections of the wall constructed of different types of sheetpiles computed by FREW for the embedment
depth of 10.6 m are depicted in Figure 7. It can be observed that the extra embedment depth can only reduce
the deflection slightly. The design embedment depth may yield a deflection of 289.1 mm even when FSP V
sheetpiles are used. The impact of embedment depth on wall deflection is minimal.
Figure 6: Displacement of Cantilever Walls of

Embedment Depth of 9 m
Figure 7: Displacements of Cantilever Walls of

Embedment Depth of 10.6 m
The distributions of lateral earth pressures as a function of depth acting on the cantilever sheetpile retaining
wall are depicted in Figure 8. The solid lines denote the lateral earth pressure behind the wall which is taken to
be negative. The dotted lines denote the lateral earth pressure in front of the wall which is taken to be positive.
As the movement of soil behind the wall is quite large above the excavation level, the active condition has
been developed. As a result, the lateral earth pressure is practically independent of the flexural rigidity of the
sheetpile. Similarly, the soil in front of the wall down to approximately 4 m below the excavation level is in
110
the passive plastic equilibrium state. The distribution of lateral earth pressure further down the excavation
level depends on the flexural rigidity of the sheetpile. The effects of soil-structure interactions are also
depicted in the bending moment distributions shown in Figure 9. It should be noted that maximum bending
moments occur well below the excavation level. It can also be observed that the bending moment in the
sheetpile close to the wall toe increases slightly with increase in flexural rigidity of the sheetpile. The
phenomenon can be attributed to the differences in lateral earth pressure distributions as shown in Figure 8.
FSP III
FSP IV
FSP V
Figure 8: Distributions of Lateral Earth Pressures as a Function of Depth in the Cantilever Retaining Wall
3.2 Case study 2 propped wall

Sheetpiles are installed for a propped cantilever retaining wall as shown in Figure 1. The required embedment
depth d determined by limit equilibrium using the free earth support analysis is 4.2 m when the factor of
safety is unity. Applying the factor of safety of 1.2 to the calculated embedment depth, the design embedment
depth becomes 5.04 m. Therefore, an embedment depth of 5 m was adopted in this case study. The horizontal
struts used in the analyses are 305305137 kg/m UC at 4 m center-to-center. The Young's modulus of steel
is taken to be 205,000,000 kN/m2 and the cross-sectional area of the strut is 0.0174 m2. Therefore, the stiffness
of the strut in the analysis is determined to be (205,000,0000.0174) / (94) = 99,083 kN/m/(m run of
sheetpiles).
Figure 9: Distributions of Bending Moment in the

Cantilever Retaining Wall
Figure 10: Lateral Displacements of the Propped

Cantilever Wall
111
The lateral displacements of the propped cantilever retaining wall as a function of depth are shown in
Figure 10. The maximum deflections of the wall occur at approximately 1 m above the excavation level. The
maximum deflections for FSP III, IV and V sheetpiles are calculated to be 39.49 mm, 18.27 mm and 12.31
mm, respectively. The relative maximum deflection is thus 3.21 : 1.48 : 1.00. The relative flexural rigidities of
these sheetpiles as tabulated in Table 2 are 1.00 : 2.30 : 3.75. If the active lateral earth pressures on these
different types of sheetpiles are identical, the deflection should be inversely proportional to the flexural
rigidity of the sheetpile, i.e. the product of the relative maximum deflection and the relative flexural rigidity
should be a constant. However, this is not the case as revealed by the results of the analyses. In fact, the
products for FSP III, IV and V sheetpiles are 3.21, 3.41 and 3.75, respectively, indicating the relative
maximum deflection increases with the relative flexural rigidity of the sheetpile. It can also be observed from
the deflection of the sheetpile wall that passive condition has been developed in front of the wall. The
phenomenon is consistent with the lateral earth pressure distributions shown in Figure 11. However, the
shortening of the strut decreases with the flexural rigidity of the sheetpile. As a result, the strut load decreases
with increase in flexural rigidity of the sheetpile. In this case study, the strut loads are determined to be 262.61
kN/m, 211.66 kN/m and 182.33 kN/m for FSP III, IV and V sheetpiles, respectively.
FSP III
FSP IV
FSP V
Figure 11: Distributions of Lateral Earth Pressures as a Function of Depth in the Propped Cantilever Retaining
Wall
The distributions of lateral earth pressures for the propped cantilever sheetpile retaining wall are depicted
in Figure 11. It can be observed that the soil behind the retaining wall is generally in the active plastic
equilibrium state regardless of the flexural rigidity of the sheetpile, as the deflection of the wall is adequate to
mobilize the coefficient of active lateral earth pressure Ka as shown in Figure 3. The lateral earth pressure in
front of the retaining wall remains in the passive state, as the wall deflects towards the excavation. However,
the coefficient of lateral earth pressure decreases with depth when the deflection of the wall decreases. When
the lateral displacement is adequately large to mobilize the soil in front of the wall to the passive plastic
equilibrium, the coefficient of lateral earth pressure equals KP as depicted in Figure 3. The resulting lateral
earth pressure thus increases linearly with overburden pressure as shown in Figure 11. As the deflection of the
retaining wall below the excavation level decreases with depth as shown in Figure 10, the coefficient of lateral
earth pressure decreases below a threshold depth which decreases with the flexural rigidity of the wall as
shown in Figure 11.
The distributions of bending moments in the propped cantilever sheetpile retaining wall as a function of
depth are depicted in Figure 12. It can be observed that the bending moment in the sheetpile below the strut
decreases with increase in flexural rigidity of the sheetpile. The phenomenon can be attributed to the decrease
in strut load with increase in flexural rigidity of the sheetpile.
112
Figure 12: Distributions of Bending Moments in the Propped Cantilever Retaining Wall
4 CONCLUSIONS
The following conclusions can be drawn from this study:
(1) The limit equilibrium approach in the design of ELS systems without considering any soil-structure
interactions is conservative in the determination of embedment length, strut load, bending moment in
the sheetpile etc. However, the resulting design may not be practical. For example, the deflection of
the retaining wall may be excessive.
(2) The engineering behavior of a propped cantilever retaining wall is different from that of a cantilever
retaining wall.
(3) When soil-structure interactions are considered, the flexural rigidity of the sheetpile and the stiffness
of the strut have significant impacts on the engineering behavior of the ELS system.
(4) The design of ELS system can be significantly improved by taking soil-structure interactions into
consideration.
ACKNOWLEDGEMENTS
The first author is grateful to Ir James W.C. Sze of Ove Arup & Partners Hong Kong Ltd. for the detailed
discussion on the software FREW.
REFERENCES
Burland, J.B., Potts, D.M. & Walsh, N.M. 1981. The overall stability of free and propped embedded cantilever
retaining walls. Ground Engineering, 14(5):28-38.
Gaba, A.R., Simpson, B., Powrie, W. & Beadman, D.R. 2003. Embedded retaining walls guidance for
economic design. Report No. C580, CIRIA, London.
GEO 1990. Review of design methods for excavations. GCO Publication No. 1/90, Geotechnical Engineering
Office, Civil Engineering & Development Department, Hong Kong SAR Government, Hong Kong.
Knappett, J.A. & Craig, R.F. 2012. Craig's soil mechanics. 8th Edition, Spon Press, New York, New York.
Oasys Ltd. 2014. FREW - Version 19.2. Oasys Ltd., London.
Padfield, C.J. & Mair, R.J. 1984. Design of retaining walls embedded in stiff clay. Report No. 104, CIRIA,
London.
Wang, S.-T., Vasquez, L. & Xu, D. 2013. Application of soil-structure interaction (SSI) in the analysis of
flexible retaining walls. In J. Hu, J. Ma, J. Meneses, T. Qui, X. Yu & X. Zeng (ed.), Challenges and recent
advances in geotechnical and seismic research and practices, Proc., IACGE 2013, Geotechnical Special
Publication No. 232, ASCE, Reston, Virginia, 567-577.
113
114
Numerical Modelling of Flexible Barriers Subject to Landslide

Debris Impact
C. Lam & R.C.H. Koo
Geotechnical Engineering Office, Civil Engineering and Development Department,
the Government of Hong Kong SAR
Z.H. Zhou, Y.P. Liu & S.L. Chan

Department of Civil and Environmental Engineering,
The Hong Kong Polytechnic University, Hong Kong
ABSTRACT
Steel flexible barriers are commonly used as defence measures to intercept rockfalls in Hong
Kong. Due to their ability to absorb high impact energy through elasto-plastic deformation, they
have also been found capable of retaining landslide debris in different site settings. However, thus
far there is no widely accepted test procedure to verify the performance of these barriers under
debris impact. To facilitate the design of flexible debris-resisting barriers, a finite element
program named NIDA-MNN has been developed under the collaboration between geotechnical
and structural engineering professionals. A special feature of this program is that it can simulate
the sliding action of the metallic nets along the support cables, i.e., the so-called curtain effect.
This program was recently used to back-analyse the structural behaviour of a flexible barrier
intercepting landslide debris in Jordan Valley, Hong Kong. It was found that the back-analyses
can reproduce some of the salient field observations and provide insights into the behaviour of the
barriers under impact loading. In this paper, the key features of NIDA-MNN and its use for the
back-analysis of the Jordan Valley case history are discussed.
1 INTRODUCTION
1.1 Background
Steel flexible barriers are commonly used to intercept rockfalls and snow avalanches around the world. Due to
their ability to sustain large deformations and absorb impact energy, these barriers have also been found
capable of containing landslide debris in different site settings (Roth et al. 2004). The use of flexible barriers
to mitigate landslide hazard is highly desirable since they are less intrusive to the surrounding environment
than the conventional rigid barriers which are made of reinforced concrete. Flexible barriers also have the
advantage of relatively easy construction since they are made of pre-fabricated components that can be
transported individually and assembled on site. No heavy machinery is typically required during installation.
1.2 Design of flexible debris-resisting barriers
When flexible barriers are used as rockfall fences, their energy-absorbing capacities are typically verified by
full-scale tests performed in accordance with standardized procedures (Duffy and Badger 2012; European
Organisation for Technical Approvals 2013). However, at present there is no widely accepted test procedure
to verify the performance of flexible barriers to withstand landslide debris. Testing a full-scale barrier under
debris impact loading is intrinsically difficult as it will involve producing an artificial debris flow or preinstalling a barrier at a location that is prone to debris flows. Full-scale testing is thus relatively rare, and so far
only has been carried out by barrier manufacturers and research institutions (e.g. Wendeler et al. 2006;
Bugnion et al. 2008; Luis-Fonseca et al. 2011).
In light of the above, numerical simulations present an attractive alternative because they can be performed
on modern computers and the results can be used to identify critical barrier components in the design.
However, the program must be able to simulate the dynamic behaviour of all barrier components, and the
115
numerical modeller must possess the relevant knowledge in order to correctly interpret the simulation results.
Previous authors such as von Boetticher et al. (2011) and Volkwein et al. (2014) have discussed the use of
numerical modelling for the design of flexible debris-flow barriers. This paper further demonstrates the value
of numerical analyses by presenting a landslide case history in Hong Kong, where the falling debris was
arrested by a flexible barrier originally designed to intercept rockfalls. Numerical simulations were carried out
using the finite element program NIDA-MNN to understand the structural behaviour of the barrier upon debris
impact. Before the simulation results are presented, information about the landslide and the computer program
is given as follows.
2 LANDSLIDE IN JORDAN VALLEY, HONG KONG
2.1 Location and barrier information
Figure 1 shows the locations of the landslide and the flexible barrier in Jordan Valley. The landslide scar is 10
m wide and 7 m long and the source volume is about 110 m3. The flexible barrier, which is a proprietary
product and has a designed energy capacity of 1000 kJ, was originally designed to retain boulders with a
volume up to 1 m3. The barrier consisted of 5-m tall steel posts made of 140-mm square hollow sections. The
spacing between the posts was about 10 m. The posts were anchored to the ground by uphill and lateral wire
ropes. Metallic ring nets and secondary wire mesh were hung down from wire ropes attached between the
steel posts. The posts were founded on 400-mm square concrete pad footings with an embedment depth of
about 500 mm. Energy-dissipating devices were fitted to some of the uphill cables to absorb impact energy.
Figure 1: Location of the Landslide and the Posts of the Flexible Barrier (Kwan et al. 2014)
2.2 Post-landslide barrier conditions

When the landslide occurred, the debris struck post P1 and two adjacent netting panels spanning across posts
P1-P2 and posts P1-P15. Figure 2 shows the barrier retaining the debris and the severely damaged post P1.
Post P15 was also bent though it was not subjected to any direct impact by the debris. Both posts P1 and P15,
116
together with their concrete footings, were found to have moved forward by 1 to 2 m. It is worth noting that
the construction details of post P15 were slightly different from those of posts P1 and P2. The netting was
attached to post P15 by nine layers of stainless steel bands, whereas posts P1 and P2 were not tied to the net.
This difference might have contributed to the failure of post P15 as shown in Figure 3.
Damaged
post P1
Ring net and

secondary mesh
Figure 2: Landslide Debris Intercepted by a Flexible Barrier at Jordan Valley, Hong Kong (Kwan et al. 2014)
Damaged
post P15
Wire ropes
Foundation
block
Figure 3: The Damaged Steel Post P15 and Its Uplifted Foundation Block
117
2.3 Debris mobility analysis

A mobility analysis of the landslide debris was carried out using the computer program 2d-DMM (Kwan and
Sun 2006). The results showed that the debris could have reached the barrier at about 4 s after the onset of the
landslide, and that the maximum frontal debris velocity could be about 4 m/s at the location of the barrier. The
calculated debris thickness ranges from 0.8 to 1.2 m.
3 NUMERICAL ANALYSIS
3.1 Finite element program NIDA-MNN
The numerical back-analysis was conducted using the non-linear finite element program NIDA-MNN (Chan
et al. 2012). This program can simulate elasto-plastic deformation of the energy-dissipating devices by the use
of non-linear structural elements. It can also simulate the sliding action of the metallic nets along the support
cables, that is, the so-called curtain effect. This modelling capability has proven to be essential in the realistic
modelling of the structural behaviour of flexible barriers. For example, Figure 4 shows the original and the
deformed shape of a net subject to the impact of a falling rock. It can be seen that the deformed net elements
concentrate around the region of impact as would be expected in a real physical test. The stretching of the net
is also realistically simulated in the computer model.
NIDA-MNN is on the pre-accepted computer program list maintained by the Buildings Department in
Hong Kong. This program is not computationally demanding so that it is suitable for routine design use.
Nodes slide
towards the sag
of the cable
Before impact
After impact
Figure 4: Deformation of Metallic Net due to Rockfall
3.2 Modelling of the flexible barrier

The finite element model was set up according to the physical dimensions of the various structural
components (e.g. steel posts, netting, wire ropes) as shown in the as-built drawings of the barrier. The energydissipating devices were not included in the model because the post-landslide inspection showed that they
were not mobilised. The foundations of the steel posts were modelled as pin supports because shallow
footings cannot provide high rotational restraint. This assumption agrees with site observations that the
footings of posts P1 and P15 failed by rotation after the landslide impact. The computer model of the barrier is
shown in Figure 5.
118
No connection between
steel posts and ring net
Net attached
to post P15
Wire ropes
Lateral wire rope
Uniformly distributed pressure

applied on ring net and post P1
Steel post
Anchor
Figure 5: Computer Model of the Flexible Barrier for Structural Analysis
3.3 Modelling of debris impact load

The debris impact load was modelled as a uniformly distributed pressure acting perpendicular to the barrier.
Site observation indicates that the landslide debris could have impacted on post P1 and the netting on the both
sides of the post. From the results of the post-landslide inspection and the debris mobility analysis, the
thickness of the debris hitting the barrier was about 1.0 m thick. Thus, the height of the loaded area is assumed
to be 1.0 m. The width of the loaded area is taken as 10 m, which is based on the result of post-landslide
inspection. A single impact was considered in the analysis because there was no evidence of multiple impacts.
In view of the fact that posts P1 and P15 failed both structurally and geotechnically, the magnitude of the
debris impact pressure was adjusted to replicate this in the numerical model. It was found that some of the
steel posts would fail when the debris impact pressure reached 50 kPa and the wire ropes would reach their
ultimate tensile strength (270 kN) when the pressure reached 66 kPa.
3.4 Results
Table 1 summarises the calculated maximum axial compressions (P), bending moments (M), shear forces (V)
in posts P1, P2 and P15 for the two load cases. It should be noted that these member forces and moments
considered nonlinear effects in the deformed location. The calculated maximum horizontal reactions at the
post bases and the cable forces are also given in the table. The bending moment and shear force diagrams for
the posts for a debris pressure of 50 kPa, together with the computed deformation of the barrier, are shown in
Figures 6 and 7 respectively. The analysis shows that maximum calculated deformation of the netting is about
2.5 m.
Table 1: Numerical Analysis Results from NIDA-MNN (Kwan et al. 2014)
Uniformly
distributed
pressure
(kPa)
50
66
Post
no.
P
(kN)
Mx
(kNm)
My
(kNm)
P2
P1
P15
P2
P1
P15
21
68
92
25
83
107
0
13.9
4
0
19.8
5.5
0
0.2
20.4
0
0.1
26.6
Section
capacity
R
(-)
0.1
1.0a
1.6a
0.1
1.4a
2.1a
Posts fail in buckling
119
Vx
(kN)
Vy
(kN)
0
1.2
15.6
0
1.6
20.7
0
86.6
3.4
0.1
123.3
4.8
Cables fail in tension
Post base
horizontal
reaction in
y-dir. (kN)
47
333
51
62
438
67
Wire
rope
force
(kN)
227
227
227
271b
271b
271b
Figure 6: Bending Moment Diagrams for the Steel Posts under a Debris Pressure of 50 kPa
Figure 7: Shear Force Diagrams for the Steel Posts under a Debris Pressure of 50 kPa
3.5 Discussion
The analysis results show that the steel posts could have been subjected to a combination of axial load and
bending moments about the two principal axes of the cross section. According to Buildings Department
(2011), buckling failure of a steel column would occur if the section capacity factor (R) as defined in Equation
(1) is greater than unity:
R=
My
Mx
P
+
+
Psq M CX M CY
(1)
where P is the calculated axial force, Mx and My are the calculated bending moments about x- and y-axes, MCX
and MCY are the bending moment capacities about x- and y-axes (= 17 kNm), and Psq is the squash load (=
120
critical yield strength cross-sectional area = 372 kN). The steel post has a critical yield strength ( ) of
174 MPa and a cross-sectional area of 2.14 103 mm2.
By applying Equation (1) to examine the results given in Table 1 and Figures 6 and 7, it can be deduced
that both posts P1 and P15 probably failed by buckling when the debris pressure reached 50 kPa. The large
bending moment about x-axes (Mx) in post P1 could have been caused by the debris pressure acting directly on
the post, whereas the large bending moment about y-axes (My) in post P15 could have been caused by the
lateral deformation of the netting attached to it. Failure of both posts P1 and P15 could be due to the combined
actions of bending moment and axial compression load, as shown by their high R values. In addition, post P15
could have experienced an axial compressive load (P) higher than that in post P1. This is thought to be due to
the downward forces brought about by both the anchored uphill and the lateral wire ropes attached to the top
of the post. The shear forces (Vx and Vy) induced in the posts were all lower than the shear capacity (= 263 kN
in this case), so they were not critical in terms of structural capacity.
As regards the effect of the debris impact on post foundations, it can be seen from Table 1 that the
calculated horizontal reaction force at the base of post P1 is in the order of 330 to 440 kN for the two load
cases. It is very likely that this reaction force exceeded the sliding resistance of the footing. This numerical
result is in line with the site observation that the footing of post P1 was displaced from its original location.
The footing of post P15 was also displaced (Figure 3) and this is probably due to the bending failure of the
post. However, the post-failure mechanism is not modelled in the present analysis.
3.6 Dynamic pressure coefficient
The dynamic pressure coefficient () can be deduced from the back-analysis results. In the current design
practice (Kwan and Cheung 2012), debris impact pressure (p) can be estimated by using the following
equation:
p = d v2
(2)
where d is density of debris in Mg/m3 and v is debris velocity in m/s. As discussed previously, the numerical
analyses indicate that posts P1 and P15 would buckle when the debris pressure reaches 50 kPa, and that the
wire rope cables would fail in tension when the pressure reaches 66 kPa (Table 1). It follows that the actual
debris pressure should lie within this range, because the posts were indeed severely damaged by the landslide
but the cables did not fail. By assuming a debris density of 2.0 Mg/m3 (typical value) and a debris velocity of
4 m/s (from debris mobility analysis), the back-calculated dynamic pressure coefficient () therefore ranges
from 1.6 to 2.1.
3.7 Limitations of the analysis
The numerical analysis discussed above is subject to the following limitations:
(1) the analysis does not consider the post-failure behaviour of the barrier and the foundation, so that the
possibility of progressive failure is not considered;
(2) the weight of the debris retained in the bulged portion of the netting is neglected in the present analysis.
However, if the self-weight of the debris is considered important, this can be modelled by adding a
downward loading to the bulged portion of the net;
(3) the strength and stiffness of the landslide debris are not considered; in the computer model the debris
impact pressure is modelled as a pseudo-static pressure.
(4) the debris pressure is assumed constant and uniformly distributed throughout the impact. Debris-barrier
interaction is therefore not considered.
4 CONCLUSIONS
The structural behaviour of a flexible barrier under debris impact has been analysed by using the finite
element program NIDA-MNN. The numerical analysis successfully reproduces the key salient field
observations and provides useful insights into the conditions of major barrier components. The dynamic
pressure coefficient () for debris impact has been back-calculated from the analysis results.
121
With the continuing increase of computing power and technology, it is expected that the use of numerical
tools for the analysis of flexible barriers will become more popular in the future. For the study of coupled soilstructure interaction problems, other numerical tools such as LS-DYNA may also be used (Huang et al. 2014;
Kwan et al. 2015).
ACKNOWLEDGEMENTS
This paper is published with the permission of the Head of Geotechnical Engineering Office and the Director
of Civil Engineering and Development of the Government of the Hong Kong Special Administrative Region.
REFERENCES
Bugnion, L., Denk, M., Shimojo, K., Roth, A. & Volkwein, A. 2008. Full-scale experiments on shallow
landslides in combination with flexible protection barriers. In Proceedings of the 1st World Landslide
Forum, 18-21 November 2008. International Consortium on Landslides.
Buildings Department. 2011. Code of practice for the structural use of steel. Buildings Department, the
Government of Hong Kong Special Administrative Region.
Chan, S.L., Zhou, Z.H. & Y.P. Liu. 2012. Numerical analysis and design of flexible barriers allowing for
sliding nodes and large deflection effects. In C.K. Lau, E. Chan & J. Kwan (eds.), Proceedings of the One
Day Seminar on Natural Terrain Hazards Mitigation Measures. Hong Kong, 16 October 2012. The
Association of Geotechnical and Geoenvironmental Specialists (Hong Kong) Ltd.
Duffy, J.D. & Badger, T.C. Flexible Rockfall Fences. In A.K. Turner & R.L. Schuster (eds.), Rockfall:
Characterization and Control: 526-553. Transportation Research Board.
European Organisation for Technical Approvals. 2013. Guideline for European technical approval of falling
rock protection kits, ETAG 027. European Organisation for Technical Approvals.
Huang, Y., Yiu, J., Pappin, J., Sturt, R., Kwan, J.S.H. & Ho, K.K.S. 2014. Numerical investigation of
landslide mobility and debris-resistant flexible barrier with LS-DYNA. In Proceedings of the 13th
International LS-DYNA Users Conference. Dearborn, Michigan, 8-10 June 2014. Livermore Software
Technology Corporation.
Kwan, J.S.H. & Sun, H.W. 2006. An improved landslide mobility model. Canadian Geotechnical Journal,
43(5): 531-539.
Kwan, J.S.H. & Cheung, R.W.M. 2012. Suggestions on design approaches for flexible debris-resisting
barriers. Discussion Note DN 1/2012, Geotechnical Engineering Office, Hong Kong.
Kwan, J.S.H., Chan, S.L., Cheuk, J.C.Y., Koo, R.C.H. 2014. A case study on an open hillside landslide
impacting on a flexible rockfall barrier at Jordan Valley, Hong Kong. Landslides, 11(6): 1037-1050.
Kwan, J.S.H., Pun, W.K., Shiu, Y.K., Ng, C.W.W., Song, D. & Yiu, J. 2015. Study of interaction between
landslide debris and debris-resisting structures. In Proceedings of the HKIE Geotechnical Division 35th
Annual Seminar. Hong Kong, 22 May 2015. Hong Kong Institution of Engineers.
Luis-Fonseca, R., Ramat, C., Hrlimann, M., Abanc, C., Moya, J. & Fernndez, J. 2011. Debris-flow
protection in recurrent areas of the Pyreness. Experience of the VX systems from output results collected in
the pioneer monitoring station in Spain. In Proceedings of the 5th International Conference on Debris-flow
Hazards Mitigation: Mechanics, Prediction and Assessment. Padua, Italy, 7-11 June 2011. Casa Editrice
Universita La Sapienza.
Roth, A., Kstli, A. & Frenez, T. 2004. Debris flow mitigation by means of flexible barriers. In Proceedings
of the 10th International Congress INTERPRAEVENT. Riva del Garda, 26 May 2004. INTERPRAEVENT.
Volkwein, A. 2014. Flexible debris flow barrierDesign and application. Swiss Federal Institute for Forest,
Snow and Landscape Research WSL.
von Boetticher, A., Hbl, J., Wendeler, C. & Volkwein, A. 2011. Modeling the impact of shallow landslides
on flexible protection barriers. In R. Marschallinger & F. Zobl (eds.), Mathematical Geosciences at the
Crossroads of Theory and Practice, Proceedings of the IAMG 2011 Conference. Salzburg, Austria, 5-9
September 2011. International Association of Mathematical Geosciences.
Wendeler, C., McArdell, B.W., Rickenmann, D., Volkwein, A., Roth, A. & Denk, M. 2006. Field testing and
numerical modeling of flexible debris flow barriers. In C.W.W. Ng, L.M. Zhang & Y.H. Wang (eds.),
Proceedings of the 6th International Conference on Physical Modelling in Geotechnics. Hong Kong, 4-6
August 2006. Taylor & Francis.
122
Effects of Positioning of Debris-resisting Baffles on

Impediment of Debris Flows
R.P.H. Law
Geotechnical Engineering Office, Civil Engineering and
Development Department, HKSAR Government, Hong Kong
ABSTRACT
Debris flow is a common form of natural terrain landslides in Hong Kong. Because of its potential of high
mobility, even a small-scale debris flow that occurs in a dense urban setting is liable to result in severe
consequences. Mitigation of debris flow hazard has been one of the important initiatives in Hong Kongs slope
safety system. Option and design of debris flow mitigation measures have evolved in response to experience
and through continuous research and development effort and technology advances. With an improved
understanding of the dynamic soil-structure interactions between debris flows and mitigation measures,
largely through numerical modelling and simulations, more innovative designs of landslide mitigation
measures are anticipated.
In particular, recent research using numerical simulation by discrete element method (DEM) shows that
debris-resisting baffles, which act as one form of debris straining structures, can be tactically positioned to
substantially impede debris velocity. This paper will present the use of DEM in simulating landslide mobility
and soil-baffles interactions, using the flow path of a historical landslide as a demonstration, and discuss the
performance of debris-resisting baffles in impeding debris velocity at various locations along the flow path.
Important considerations and insights from the simulation will also be discussed.
1 INTRODUCTION
Debris flow is a mass wasting process of poorly sorted sediments surging down a slope in response to
gravitational attraction (Iverson 1997). The landslide debris travels along the natural recessed drainage
channel at high velocity as it mixes with the channel water. Debris flow is a common form of natural terrain
landslides in Hong Kong. Because of its potential of high mobility, even a small-scale debris flow that occurs
in a dense urban setting is liable to result in severe consequences. Mitigation measures, such as rigid and
flexible barriers, are deployed to intercept debris flows. The structural requirements of the mitigation
measures depend heavily on the design load exerted by the impacting landslide debris, which increases with
the impact velocity of debris (e.g. Kwan, 2012).
The dynamic interaction of debris flow with baffles is still an emerging area of research to both the
academics and the practitioners. Both continuum modeling (e.g. Pudasaini & Hutter 2007; Pudasaini &
Domnik 2012) and discrete element method (DEM) modeling (e.g. Ng et al. 2013) are adopted to analyze the
flow characteristics of granular debris flows. The interactions (e.g. development of dead zone and deposition
profile) between debris and protection structure such as breaking mound (Hakonardottir & Hogg 2005), rigid
walls (Teufelsbauer et al. 2011) and individual obstacle (Gray et al. 2003) are elaborated in the literature.
Effect of granular debris flow baffles installed immediately upstream of rigid barriers on dissipation of flow
energy has also been studied (Choi et al. 2014a, 2014b; Law 2015; Ng et al. 2014). Further to these, this paper
will present the use of DEM in simulating landslide mobility and soil-baffles interactions, using the flow path
of a historical landslide as a demonstration, and discuss the performance of debris-resisting baffles in
impeding debris velocity at various locations along the flow path. Important considerations and insights from
the study will also be discussed.
2
DEM SIMULATION
DEM was introduced in geomechanics by Cundall & Strack (1979) and was later adopted as a research tool by
many other researchers (e.g. Labra et al. 2008; Sibille et al. 2008; Thompson et al. 2009). It is an appropriate
tool to model debris flow and snow avalanche because of their granular flow nature (Zwinger 2000; Nicot
2004; Hutter et al. 2005; Pudasaini & Hutter 2007). In DEM modelling, the discrete elements displace
123
independent of one another and interact only at contacts between the particles and the boundary. The particle
motion of each discrete element is calculated from the forces acting on it by the Newtons laws of motion. The
displacements and rotations of the discrete elements are computed as the calculation progresses. New contacts
and complete detachment between discrete elements are recognized automatically. The contact forces and
displacements of a stressed assembly of particles are found by tracing the movements of the individual
particles. The simulation is implemented by a timestepping algorithm in which the velocities and accelerations
are constant within each timestep. The recent advancement of computer hardware and parallel computing
technology enables the analysis of granular flows using three-dimensional DEM efficiently.
The capability of DEM to simulate the soil-baffles interactions is verified by benchmarking with flume
model test results. The capability of DEM to carry out debris mobility modelling is verified by
benchmarking with the field evidence of the debris flow event that occurred above Yu Tung Road, Lantau in
June 2008. The open source 3D software called LIGGGHTS (Kloss & Goniva 2010) is adopted to model
landslide debris. The software is developed based on Large-scale Atomic/Molecular Massively Parallel
Simulator (LAMMPS), which is developed for molecular research by Sandia National Laboratories, US
Government. The verifications using flume model test and the historical debris flow event are elaborated
below.
Details of the calibration using flume model test results are reported in Choi et al. (2014b). Flume model
tests were carried out to validate DEM in carrying out analysis of debris-baffles interactions. The input
parameters adopted by the DEM is given in Choi et al. (2014b). The comparison of the flow profile between
the flume model test and DEM is given in Figure 1, which indicates consistent outcomes.
Figure 1: Comparison of Flume Experiments and Computed Flow Kinematics using DEM (Choi et al. 2014b):
(a) t = 0 s; (b) t = 0.03 s; (c) t = 0.06 s; (d) t =0.09 s
Field data obtained from the June 2008 Yu Tung Road event is used to validate DEM in terms of debris
mobility modelling. Figure 2 shows the aerial view of the debris flow above Yu Tung Road taken on 9 June
2008 (AECOM 2012). The debris flow involved a single landslide source area. The landslide source
comprised subangular to subrounded cobbles with many gravels and occasional boulders in a silty/clayey
sandy matrix. The landslide debris reached Yu Tung Road, resulting in blockage of both westbound lanes and
flooding of the adjacent Cheung Tung Road. About 2,600 m of debris was involved which had a runout
distance of about 600 m, and all the debris reached Yu Tung Road. The lower portion of the channelized debris
flow was captured on video by a member of the public, although the exact time of the failure is not known.
According to AECOM (2012), the velocity of the debris was estimated to be about 12 m/s at chainage =
100 m based on the super-elevation data measured (Hungr et al., 1984). The measured velocity data beyond
chainage = 400 m is based on both the super-elevation data and the video footage. Further details of the field
evidence is documented in AECOM (2012). The computed frontal velocity from the DEM is compared with
the field data. The frontal velocity is calculated by the average tangential velocity of the most frontal 10% of
124
the discrete elements that simulate the landslide debris. Discrete elements which have a coordination number
of zero are considered to be isolated particles and are not counted in the calculation of the frontal velocity.
The input parameters adopted by the DEM are given in Table 1. Derivation of the input parameters is
discussed in Section 4 in detail. No baffle is placed on the flow path in this validation exercise. Figure 3 shows
the comparison between the computed velocity profile and the measured data point. A close match between
the computed velocity profile and measured data point is observed. It is interesting to note that the measured
frontal velocity is increasing from Point B to E, but the DEM calculations give an opposite trend. According to
AECOM (2012), the flow path from B to E becomes less channelized and the gradient decreases to less than
15 , which explains the computed decrease in velocity in DEM. The relatively high measured frontal velocity
from Point B to E might deserve further research.
Figure 2: Aerial View of the Debris Flow Event above Yu Tung Road Taken on 9 June 2008
a)
20
Calculated frontal velocity using DEM

Field observations
Fontal velocity (m/s)
16
12
D
B
4
0
0
50
100
150
200
250
300
Chainage (m)
350
400
450
500
550
Figure 3: (a) Comparison between the Computed Velocity Profile using DEM and Field Evidence of Debris Flow
above Yu Tung Road in June 2008 and (b) the Flow Profile at time = 32 s; the Dashed Lines represent
the locations where field evidence of flow velocities are available
MODEL CONFIGURATION
The same DEM software, i.e. LIGGGHTS, is also adopted to model the motion of debris impacting on baffles.
Figure 4 shows the numerical model setup. The topography is developed based on the data obtained using the
Light Detection and Ranging technology. The computation domain is 650 m in length and 180 m in width.
125
The grid size of the terrain is 2 m x 2 m. The natural terrain above Yu Tung Road is represented by mesh solid
which is specified using interconnected triangulated meshes. The stereolithography (STL) file format is used
to store the location and dimension of the triangles. The source area of the debris flow is indicated in Figure 4.
Two rows of baffles are installed at staggered pattern at four different locations along the flow path (i.e.
locations 1, 2, 3 and 4 in Figure 4). The transverse spacing between each adjacent baffle (D) is 3 m. The baffle
height is equal to 3 m. The row spacing (L) between two adjacent rows of baffles is 9 m. The ratio between the
row spacing and transverse spacing (L/D) is equal to 3. The L/D =3 corresponds to the optimum row spacing
for baffle arrays to maximize the impedance effect on impacting debris (Law 2015).
Input parameter
Number of discrete elements
Particle diameter (m)
Particle stiffness (N/m)
Discrete element friction angle (o)
Coefficient of restitution
Coefficient of rolling friction
Table 1: DEM Input Parameters

Value
200, 000
0.3
1 108
11
0.5
0.7
In each simulation, only one location is installed with baffles. The chainage of baffle locations 1,2,3 and 4
are 100 m, 200 m, 340 m and 450 m respectively. A terminal rigid barrier is assumed to be installed at
chainage = 500 m to retain landslide debris. Baffle location 1 is characterized by its proximity to the source
area. Baffle location 2 is characterized by the high frontal velocity (i.e. ~18 m/s) of debris (see Figure 3a).
Baffle location 3 is approximately halfway from the location of peak frontal velocity to the terminal barrier.
Baffle location 4 is characterized by its proximity to the terminal barrier. The performance of
debris-resisting baffles in impeding debris velocity at the four locations along the flow path is studied.
Figure 4: Numerical Model Setup
126
INPUT PARAMETERS
The simulation of the motion of landslide debris using DEM requires the use of a number of key input
parameters, including the size of discrete element, the contact friction angle, the coefficient of restitution, the
contact stiffness, as well as the coefficient of rolling friction. The input parameters for this study (see Table 1)
are determined through a review of the literature, except the coefficient of contact friction for any two
contacting bodies, which is obtained through benchmarking using the June 2008 Yu Tung Road event.
Due to the constraint of computation capacity, it is not possible to model the motion of individual grains in
debris flow. Based on the study by Law (2015), the size of discrete elements has to be at least 10 times smaller
than the baffle spacing in order to avoid unrealistic clogging. Further reduction of particle does not result in
any change of the computed results. The spacing between baffles in one row is set to be 3 m. The diameter
of each discrete element should therefore be less than 0.3 m. The required number of discrete element of 0.3 m
in diameter is around 200,000 to give the volume of the landslide debris of 2,600 m3.
The local rheology of the flow material is simulated through the assignment of contact behaviour of the
discrete elements and the flow path. Frictional granular flow is the focus of this study. The relative
translational and rotational motions between the discrete elements are mainly resisted by the contact friction.
Assuming that the two objects in contact have the same roughness, the frictional force (Ff) between two
contacting bodies is calculated using the following equation:
F f = f * Fs
(1)
where f is the coefficient of friction between the two objects. Based on benchmarking using the June 2008 Yu
Tung Road event, the coefficient of friction is set as 0.2.
Discrete elements experience repulsive force when they are in contact with each other. The stiffness model
calculates repulsive forces between discrete elements. Given the contact stiffness is not unreasonably low, the
contact stiffness of the discrete element has negligible influence on the computed mobility of granular
material (Crosta et al. 2001). The stiffness of discrete element and barrier are both chosen to be 1x108 (N/m).
Calvetti & Nova (2004) observed that, as a consequence of the non-angular shape of the particles, the
macroscopic friction angle of the granular mass is generally very low, typically much less than 30,
irrespective of the value of the inter-particle friction angle adopted. Calvetti et al. (2003) and Tamagnini et al.
(2005) emphasized the need to inhibit rolling resistance in the discrete spherical particles. The directional
constant torque models (e.g., Zhou et al. 2003) apply a constant torque on a particle to represent the rolling
friction. The direction of the torque is always against the relative rotation between the two contact entities.
The torque is applied in pairs on each pair of particles in contact. An example of the directional constant
torque model could be found below, which is a simplified version of the formula reported by Zhou et al.
(1999):
= /| |
(2)
=
(3)
where i and j are the angular velocities of sphere i and j respectively, rel is the relative angular velocity
between them, r is the rolling friction coefficient, Fs is the normal contact force, and Rr is the radius of the
discrete element. The value of the rolling friction coefficient r is an unknown and can be determined through
calibration with experiments. Based on the parametric study presented in Law (2015), the coefficient of
rolling friction is set to be 0.7.
Based on the benchmarking results discussed in Section 2, the input parameters are therefore considered
to be appropriate to simulate the motion of debris in this study.
5
MODELLING PROCEDURES
Each numerical analysis is divided into two stages, namely preparation and flow stages. In the preparation
stage, an assembly of discrete elements with random packing is positioned on the source area under the action
of gravity. In the flow stage, the discrete elements are allowed to move beyond the source area and travel
127
downslope under the action of gravity. The velocity and the displacement of the frontal debris are recorded
and analyzed throughout the process. In each simulation, only one location (either location 1, 2, 3 or 4) is
installed with baffle arrays. The performance of debris-resisting baffles in impeding debris velocity at various
locations along the flow path is observed and evaluated.
6
IMPEDENCE EFFECTS OF BAFFLE ARRAYS AT DIFFERENT LOCATIONS
Figure 5 shows the comparison between the computed velocity profile without baffles (i.e. the control case,
indicated by solid line) and those with baffles at different positions (indicated by dotted line, dashed line,
dot-dashed line and hollow line at location 1 (chainage = 100 m), location 2 (chainage = 200 m), location 3
(chainage = 340 m) and location 4 (chainage = 450 m) respectively).
The key observations are given below.
(a) No significant drop of frontal velocity is observed right at the baffle location.
Based on the findings reported by Law (2015), the frontal debris that passes through the baffle arrays is not
instantaneously decelerated due to the absence of debris deposit behind the baffle arrays during the frontal
impact. As a result, the frontal velocity of debris right at the baffle location is found to be similar to the case
without baffles.
(b) Debris slows down after passing through the baffle arrays, as compared to the control case.
As compared to the control case, the slow-down process is due to by the reduction of discharge due to the
run-up and pile-up processes of debris behind the baffles (Law 2015). The relationship between the discharge
and frontal velocity of debris can be explained by the following equation, which calculate the lateral driving
force (P) acting on the frontal debris due to the lateral earth pressure of the upstream debris. The equation is
derived based on the work done by Hungr (1995).
P = k
H
H cos
s
(4)
where is the unit weight, k is the lateral earth pressure coefficient, H is the thickness, H is the difference in
thickness of two adjacent blocks of landslide debris, s is the tangential distance between the two adjacent
blocks of landslide debris and is the slope inclination. Graphical illustration of each term in Equation 4 is
given in Figure 6. As shown in Equation 4, the lateral driving force (P) acting on the frontal debris decrease
when the thickness of the upstream debris is reduced. After passing through the baffle arrays, the thickness of
the debris reduces as its discharge from the baffles decreases, resulting in a drop in its frontal velocity.
(c) Debris further slows down at bend, as compared to the control case.
The difference in the frontal velocity between the control case and those with baffles at location 1, 2 and 3
increases after chainage = 400 m, where a relative sharp bend of the recessed drainage path is present. In other
words, debris travelling through the bend has larger deceleration when the velocity of debris is lowered by the
baffle arrays.
(d) Impedance effect of baffle arrays increases with the impact velocity to baffle arrays.
The percentage reductions of frontal velocity at the terminal barrier, as compared with the control case, are
49% (location 1), 56% (location 2), 30% (location 3) and 6% (location 4) respectively. The baffles at location
2 area are able to achieve a higher reduction of frontal velocity in the end probably because they intercept the
landslide debris at a higher impact velocity.
128
20
Control case (No baffle installed)

Baffle array installed at location 1
16
Fontal velocity (m/s)
Field observations
12
4
Bend
location
Location 3
Location 2
Location 1
Location 4
Barrier
0
0
50
100
150
200
250
300
Chainage (m)
350
400
450
500
Figure 5: Comparison between the Computed Velocity Profile without Baffles and those with Baffles
at Different Positions
(e) Baffle arrays that are located too close to the source area or the terminal barrier are relatively ineffective
in reducing the frontal velocity at the terminal barrier.
The frontal velocity close to the source area is far below its peak velocity. There is still certain distance
along the flow path for the debris to pick up its velocity before slowing down. Therefore, the baffle arrays
close to the source area are not in optimal performance in reducing the frontal velocity of debris reaching the
terminal barrier. As there is no significant drop of frontal velocity right at the baffle location, and there is
insufficient distance for the debris to further slow down, baffle arrays located too close to the terminal barrier
are also not in optimal performance in reducing the frontal velocity of debris reaching the terminal barrier.
Baffle arrays
P = k
Block of
debris
H
H cos
s
Flow path
P
Direction of
motion
Figure 6: Lateral Driving Force (P) Acting on the Frontal Debris Due to the Lateral Earth Pressure of the Upstream
Debris (Hungr, 1995) (please refer to Equation 4 for the definition of each term)
129
CONCLUSIONS
The performance of debris-resisting baffles in impeding debris velocity at various locations along the flow
path is studied using DEM. A number of key observations are presented. The baffles are found to reduce the
frontal velocity of landslide debris that reaches the terminal rigid barrier. Since the impact pressure of
landslide debris on a rigid barrier depends on the impact velocity (Kwan, 2012), and the peak impact force
experienced by a rigid barrier is contributed mainly by the frontal impact (Law, 2015), the reduction of frontal
debris velocity will reduce the peak impact force on a rigid barrier. The actual performance of baffle arrays
are affected by a number of factors, such as the channel topography, debris rheological characteristics,
entrainment behaviour, etc. The observations may not be fully applicable to all debris flow events. Also,
the frontal velocity of impacting debris is not the only controlling factor that determine the structural
requirements of structural protection measures. Other factors, such as discharge rate and deposition
mechanisms, are equally important.
ACKNOWLEDGEMENTS
This paper is published with the permission of the Head of the Geotechnical Engineering Office and the
Director of Civil Engineering and Development, Government of the Hong Kong Special Administrative
Region.
REFERENCES
AECOM 2012. Detailed Study of the 7 June 2008 Landslides on the Hillside above Yu Tung Road, Tung
Chung. GEO Report No. 271. Geotechnical Engineering Office, Hong Kong.
Calvetti, F. & Nova, R. 2004. Micromechanical Approach to Slope Stability Analysis. In F. Darve & I.
Vardoulakis (eds), Degradation and instabilities in Geomaterials: 235-254. Springer, Verlag.
Calvetti, F., Viggiani, G. & Tamagnini, C. 2003. A numerical investigation of the incremental behavior of
granular soils. Rivista Italiana di Geotecnica, 37(3): 11-29.
Chiou, M.C. 2005. Modelling Dry Granular Avalanches Past Different Obstructs: Numerical Simulations and
Laboratory Analyses. Dissertation, Technical University Darmstadt, Germany.
Choi, C.E., Ng, C.W.W., Song, D., Kwan, J.H.S., Shiu, H.Y.K., Ho, K.K.S. & Koo, R.C.H. 2014a. Flume
investigation of landslide debris baffles. Canadian Geotechnical Journal, 51(5): 540-533.
Choi, C.E., Ng, C.W.W., Song, D., Law, R.P.H., Kwan, J.H.S., & Ho, K.K.S. 2014b. A computational
investigation of baffle configuration on the impedance of channelized debris flow. Canadian Geotechnical
Journal, published online 14 July 2014 (In press).
Crosta, G.B., Calvetti, F., Imposimato, S., Roddeman, D., Frattin, P. & Agliardi, F. 2001. Granular Flows and
Numerical Modeling of Landslides. Contract No EVG1CT-1999-00007, Universita delgi Studi di Milano,
Bicocca.
Cundall, P.A. & Strack, O.D.L. 1979. A discrete numerical model for granular assemblies. Geotechnique,
29(1): 47-65.
Gray, J.M.N.T., Tai, Y.C. & Noelle, S. 2003. Shock waves, dead zones and particle-free regions in rapid
granular free-surface flows. Journal of Fluid Mechanics, 491: 161-181.
Hkonardttir, K.M. & Hogg, A. 2005. Oblique shocks in rapid granular flows. Physics of Fluids, 17(7): 1-10.
Hungr, O., Morgan, G.C. & Kellerhals, R. 1984. Quantitative analysis of debris torrent hazards for design of
remedial measures. Canadian Geotechnical Journal, 21: 663-677.
Hutter, K., Wang, Y. & Pudasaini, S.P. 2005. The Savage-Hutter avalanche model: how far can it be pushed?
Philos Transact A Math Phys Eng Sci., 363(1832): 1507-1528.
Iverson, R.M. 1997. The physics of debris flow. Review of Geophysics, 35(3): 245-296.
Kloss, C., Goniva, C., Hager, A., Amberger, S., Pirker, S. 2012. Models, algorithms and validation for
opensource DEM and CFD-DEM. Progress in Computational Fluid Dynamics, An Int. J., 12(2-3):
140-152.
Kwan, J.S.H. (2012). Supplementary Technical Guidance on Design of Rigid Debris-resisting Barriers. GEO
Report No. 270. Geotechnical Engineering Office, Hong Kong, 91 p.
Labra, C., Rojek, J., Onate, E. & Zarate, F. 2008. Advances in discrete element modelling of underground
excavations. ActaGeotech., 3(4): 317-322.
130
Law, R.P.H. 2015. Computational Study of Granular Debris Flow Impact on Rigid Barriers and Baffles.
Doctorial thesis. Hong Kong University of Science and Technology, Hong Kong.
Ng, C.W.W., Choi, C.E. & Law, P.H. 2013. Longitudinal spreading mechanisms of channelized granular flow.
Geomorphology, 194: 84-93.
Ng, C.W.W., Choi, C.E., Song, D., Kwan, J.H.S., Koo, R.C.H., Shiu, H.Y.K. & Ho, K.K.S. 2014. Physical
modelling of baffles influence on landslide debris mobility. Landslides, Published online on 28 February
2014.
Nicot, F. 2004. Constitutive modelling of snow as a cohesive granular material. Granular Matter, 6(1): 47-60.
Pudasaini, S.P. & Domnik, B. 2012. Full two-dimensional rapid chute flows of simple viscoplastic granular
materials with a pressure-dependent dynamic slip-velocity and their numerical simulations. Journal of
Non-Newtonian Fluid Mechanics, 173-174: 72-86.
Pudasaini, S.P., Hsiau, S., Wang, Y. & Hutter, K. 2005. Velocity measurements in dry granular avalanches
using particle image velocimetry technique and comparison with theoretical predictions. Physics of Fluids,
17(9): 1-10.
Pudasaini, S.P. & Hutter, K. 2007. Avalanche Dynamics: Dynamics of Rapid Flows of Dense Granular
Avalanches. Springer, Berlin.
Sibille, L., Donze, F.-V., Nicot, F., Chareyre, B. & Darve, F. 2008. From bifurcation to failure in a granular
material: a DEM analysis. Acta Geotechnica, 3(1): 15-24.
Tamagnini, C., Calvetti, F. & Viggiani, G. 2005. An assessment of plasticity theories for modeling the
incrementally nonlinear behavior of granular soils. Journal of Engineering Mathematics, 52(1-3): 265-291.
Teufelsbauer, H., Wang, Y., Pudasaini, S.P., Borja, R.I. & Wu, W. 2011. DEM simulation of impact force
exerted by granular flow on rigid structures. Acta Geotechnica, 6(3): 119-133.
Thompson, N., Bennett, M.R. & Petford, N. 2009. Analyses on granular mass movement mechanics and
deformation with distinct element numerical modelling: implications for large-scale rock and debris
avalanches. Acta Geotechnica, 4(4): 233-247.
Zhou, Y.C., Wright, B.D., Yang, R.Y., Xu, B.H. & Yu, A.B. 1999. Rolling friction in the dynamic simulation of
sandpile formation, Physica A: Statistical Mechanics and its Applications, 269(2-4): 536-553.
Zhou, Y.C., Xu, B.H., Zou, R.P., Yu, A.B. & Zulli, P. 2003. Stress distribution in a sandpile formed on a
deflected base, Advanced Powder Technology, 14(4): 401-410.
Zwinger, T. 2000. Dynamik einer Trockenschneelawine auf beliebig geformten Bergha ngen. Ph.D. Thesis.
Vienna University of Technology, Austria.
131
132
Study of Interaction between Landslide Debris

and Debris-resisting Structures
J.S.H. Kwan, W.K. Pun & Y.K. Shiu
the Government of the Hong Kong SAR
C.W.W. Ng & D. Song

Department of Civil and Environmental Engineering,
Hong Kong University of Science and Technology
J. Yiu
Ove Arup and Partners Hong Kong Limited, Hong Kong
ABSTRACT
Natural terrain hillside catchments are systematically selected and their landslide risk dealt with
under Government's Landslip Prevention and Mitigation Programme. Experience shows that
landslide debris-resisting structures can provide practical and effective means for mitigating the
natural terrain landslide risk. While state-of-the-art engineering approach is being adopted, there is
still much uncertainty in the design of such structures. Concerted efforts of the geotechnical
profession and experts in other engineering disciplines have been made to enhance understanding
of the dynamic interaction between landslide debris and debris-resisting structures as well as the
performance of landslide debris-resisting structures for optimisation of the design approach. A
series of studies including physical tests and numerical investigations have been carried out. This
paper introduces the studies and presents some initial key findings.
1
INTRODUCTION
1.1 Landslide Risk in Hong Kong

In the past 30 years, resources had been devoted to upgrading substandard man-made slopes close to
developments. While landslide risks posed by man-made slopes have been substantially reduced, the landslide
risk associated with natural hillside failures increases progressively due to encroachment of more urban
development or redevelopment on steep natural hillsides.
Disruptions caused by the landslides during and after the rainstorm on 7 June 2008 are the vivid examples
which indicate the risk that Hong Kong is now exposed to. The rainstorm on that day hit the western part of
Lantau Island resulting in more than 2,400 natural terrain landslides in the area. Many of the landslides affected
important facilities. For example, one of the major landslides developed into debris flow which affected the
North Lantau Expressway and as a result the road had to be closed for 16 hours.
To deal with the natural terrain landslide risk, the Landslip Prevention and Mitigation Programme (LPMitP)
was commenced in 2010 to dovetail the Landslip Preventive Measures Programme that focused on upgrading of
man-made slopes. The LPMitP commits a substantial part of its annual funding to the study and risk mitigation
works of 30 vulnerable catchment areas each year.
1.2
Natural terrain hazard mitigation measures
Natural terrain catchments are sizable; slope stabilisation works as a means to mitigate landslide hazard are
often impractical and environmentally undesirable. The preferred approach is to provide defense measures to
mitigate the hazard caused by landslide debris. In most cases, the defense measures adopted in Hong Kong
consist of the provision of a rigid (concrete) barrier or a flexible barrier at the toe of a natural hillside
catchment (see Figure 1).
133
(b) Flexible debris-resisting barrier
(a) Rigid debris-resisting barrier
Figure 1: Debris-resisting Barriers
2
2.1
DESIGN OF DEBRIS-RESISTING BARRIERS

Current design practice
Rigid barriers are typically constructed using reinforced concrete to intercept channelised debris flows. They
are designed to resist the impact force of the debris and occasional boulders in the debris front. Lo (2000)
recommends the use of the hydrodynamic pressure equation (i.e. p = v2 where p = debris impact pressure,
= dynamic pressure coefficient, = debris density and v = debris impact velocity) to estimate the debris
impact load. For estimation of the boulder impact load, Lo (op cit) suggests using the Hertz Equation (with an
appropriate load reduction factor of 10). Kwan (2012) updated the above recommendation in respect of the
value of dynamic pressure coefficient, with the coefficient revised down from 3 to 2.5. The current
guidelines recommend the consideration of multiple phases of landslide debris impacting on a barrier, and the
use of the maximum calculated debris impact velocity for design.
Proprietary rockfall barriers are available from many suppliers. They are mostly based on empirical design
with capacity verified by full scale testing. Rockfall barriers are rated by the energy that a set of panels can
absorb without being breached. They are tested and certified against various national or European standards
(EOTA 2008). Over the past years, flexible rockfall barriers have occasionally been hit by debris flows and
landslide debris in different countries. A number of field cases and physical tests have demonstrated that
flexible barriers would be capable of arresting a certain amount of landslide debris (Duffy 1998).
The use of flexible rockfall barrier as landslide risk mitigation measure is a highly attractive option, as it is
visually less intrusive compared to rigid barriers e.g. built using reinforced concrete. Flexible barriers also
have the advantage of relatively easy construction, as they are usually made of relatively light-weight
materials. However, there lacks well-established design guidelines for flexible debris-resisting barriers.
Lo (2000) reviewed the use of flexible debris-resisting barriers. Implicit in its recommendation is the
design of the barrier to resist the full energy of landslide debris. This is likely to be conservative because as
the debris deposits behind the barrier, the deposited debris will dissipate some of the impact energy through
base friction or internal deformation. Sun & Law (2012) proposed analytical solutions for calculating the
energy loading with consideration given to energy loss caused by basal resistance experienced by landslide
debris. Wartmann & Salzmann (2002) described a design approach under which the energy from the first 4
seconds of the estimated peak flow of debris is to be resisted. Subsequently, WSL (2009) proposed to design
flexible barriers based on the dynamic impact pressure and hydrostatic pressure from debris disposition.
GEO put forward an empirical design approach for the prescriptive use of flexible rockfall barriers to
mitigate risk of open hillside failures (GEO 2014). It is based on the consideration of landslide risk and a
statistical assessment of the scale and the mobility of previous landslides. A number of qualifying criteria have
to be satisfied before such prescribed flexible barriers can be adopted.
Interim design guidelines based on the force approach were proposed by Kwan & Cheung (2012).
Hydrodynamic pressure equation is employed for estimation of the debris impact pressure. The recommended
value of for flexible debris-resisting barriers design is 2.0. This value includes the effects of impact boulders
134
of an equivalent dimension less than 2 m in diameter.

2.2
The need to understand debris-barrier interaction
When landslide debris is intercepted by a barrier, landslide debris gradually fills up the space behind the
barrier. This filling-up process involves rebound of a certain amount of debris materials and this leads to
turbulent mixing. Debris deposited against the barrier could provide cushioning effects conducive to alleviate
the impact of debris from behind. The debris-barrier interaction is highly dynamic. For flexible
debris-resisting barriers, the interaction is even more complicated since the barriers deform non-linearly under
the impact load.
The present design methodology does not consider the debris-barrier interaction. The design debris impact
velocity is determined with consideration of a 'free field' condition. A number of R&D initiatives are being
carried out with a view to driving technical advances in design of natural terrain mitigation measures.
3
RECENT TECHNICAL DEVELOPMENT WORKS
3.1 R&D initiatives

The ongoing R&D initiatives include physical tests and numerical investigation viz. (a) physical flume tests of
dry sand impacting on rigid barriers, (b) centrifuge tests of dry sand and viscous fluid impacting on flexible
barriers, and (c) numerical modelling using LS-DYNA.
3.2
Physical flume tests of dry sand impacting on rigid barriers
A 5 m long flume with a channel base width of 0.2 m was used for the tests. Side walls are about 0.5 m in
height, perpendicular to the channel bed. Dry sand was used in the test to simulate a debris flow. The angle of
repose of the sand was 33o and the interface friction angle between the sand and flume bed was 23o. The initial
bulk density of the sand mass was about 1,680 kg/m3. At the top end of the flume was a sand storage tank.
Sand can be released into the flume by opening a flip gate that was attached to the tank. A 0.1 m high vertical
barrier model was installed in the flume. A plastic film was placed on the flume bed to reduce the basal
resistance so that the sand flow can attain a velocity at the position of barrier model. The interface friction
Figure 2: High-speed Camera Images
135
angle was measured following the procedures described by Savage and Hutter (1989), of which the sand was
filled into a cylinder placed inside the flume, the interface friction angle was measured by increasing the
inclination of the flume until the cylinder starts to slide.
The flume inclination was set to be 26o to facilitate the development of sand flow corresponding to a
Froude number of 2.6 (Fr = v / (g h)0.5, where v is debris velocity, g is gravitational acceleration and h is debris
depth), similar to the local design scenarios where Fr 3.
The flume test was recorded with the use of a high-speed camera which could capture 100 images per
second. Figure 2 shows the image records in sequence.
Figure 2(a) shows the instant when the dry sand flow arrived at the barrier location. For discussion
purposes, the time of this instant is denoted as t = 0.0 s. After 0.1 second, the sand flow filled up the retention
zone behind the barrier and splashing of the sand is also evident. At t = 0.25 s, debris that was not trapped
behind the barrier started the overtopping process and launched into a ballistic flight. Subsequently, debris
overflowing from the crest of the barrier travelled along a projectile path and landed on the flume bed at a
distance of about 0.2 m downstream of the barrier (see Figure 2(d)). This process continued for about
16 seconds. At the end of the test, sands piled up behind the barrier and the deposition angle was about 33o.
The study did not only reveal the dynamic debris deposition mechanism behind rigid barriers but also
provided parameters for design of multiple barriers (Kwan et al. 2015).
3.3
Centrifuge tests
A series of centrifuge tests of landslide debris impacting on flexible barriers has been conducted by the Hong
Kong University of Science and Technology. A specially designed spring system, which exhibited a bilinear
stiffness profile, was developed to replicate the load-deformation characteristics of energy dissipation devices
in flexible barriers. Tests using dry sand and viscous fluid as landslide debris had been carried out. The
density of the dry sand was 1530 kg/m3 and the internal friction angle was 31o. The viscous fluid was a
mixture of zinc chloride solution and glycerol and carboxyl methyl cellulose with a density of 1580 kg/m3.
The tests were carried out under 25-g. During the test, the testing materials were released on to a ramp
inclined at 25o. A flexible barrier model comprising 4 cables attached to the specially designed spring system
was installed at the end of the ramp. Control experiments undertaken without the flexible barrier model
indicated that the Froude number of the debris flow at the position of the flexible barrier model was about 4.
The scaling laws stated in Bowman et al (2010) are followed, of which a factor of unity is applied to the flow
velocity.
Instrumentation included high-speed cameras, load cells and laser sensors. High resolution images, loading
of cables and extension of cables were obtained. Figure 3 shows the image of the viscous fluid at the instant
when it was impacting on the flexible barrier model as well as the vector plot of the velocity of the viscous
fluid.
450
vmax= 6.26 m/s

400
327
350
300
250
200
280
300
320
340
360
380
400
420
440
460
480
Figure 3: Viscous Fluid Impacting on the Flexible Barrier Model
136
Interpretation on the measured cable forces is in progress. The cable forces would be used to back calculate
the dynamic impact pressure and the corresponding value of . The test results could be used for further
calibration of the numerical analyses introduced in Section 3.4.
3.4 Numerical modelling using LS-DYNA
Depth-averaged continuum numerical models, viz. 2d-DMM (Kwan & Sun 2006) and DAN/W (Hungr 1995),
prove practical for debris mobility analysis to produce design debris velocity and thickness. They are
commonly adopted in local design practice where free field conditions are considered. In the depth-averaged
formulations, debris is discretised into a series of inter-connected slices or columns. The use of these models
to investigate the debris-barrier interaction is constrained by the connectivity requirement. Moreover,
vertical debris movements within the debris slices or columns are not considered by the depth-averaged
formulations, and thus simulations of turbulent mixing at the debris-barrier impact location are not possible.
A more advanced numerical model LS-DYNA has been used for numerical investigation purpose. It is a
multi-purpose finite element program used to analyse the nonlinear response of structures. A special module
has been developed to incorporate into LS-DYNA for debris mobility analysis. It handles scalar advection in
an Eulerian grid and solves equations of motion based on an Arbitrary Lagrangian-Eulerian description of
finite-element method. Landslide debris is assumed to be elasto-plastic which follows Drucker-Prager yield
criteria. The computational domain is discretised into an array of hexahedral elements which allows for a
genuine 3-dimenional debris mobility analysis. The ground surface on which landslide debris travels on is
modelled using rigid shell elements. Coulomb frictional rule is assumed at the interface between the landslide
debris and the shell surface.
20
LS-DYNA
Field observations
Frontal Velocity (m/s)
15
10
5
Rehological Parameters used:
Basal fiction angle = 8 o
Turbulent Coefficient = 500 m/s2
0
0
50
100
150
200
250
300
350
400
450
500
550
Chainage (m)
Figure 4: Frontal Velocity of Yu Tung Road Debris Flow
The LS-DYNA model had been benchmarked against several well-documented laboratory and field studies.
Kwan et al. (2015) reported the back analysed results of Yu Tung Road Debris Flow. In the analysis, an
additional damping force proportional to the square of the debris velocity was applied to retard the debris
motion. This damping force accounted for the energy loss due to the turbulence of debris flows similar to
Voellmy rheology. Figure 4 shows the calculated debris frontal velocity compared with the velocity
estimated using the super-elevations observed in the field.
Simulations of flexible barriers subject to debris flow impact had also been carried out. The debris flow
impact test of flexible barriers in Illgarben reported by Wendeler et al. (2006) was numerically replicated.
Figure 5 shows the simulation results. LS-DYNA was capable of capturing the debris-barrier interaction
reasonably well. The calculated maximum cable forces of the barrier are also comparable with the
137
measurement results. ARUP (2013) reported details of the study. Further calibration using the centrifuge data
(see Section 3.3) and more well-documented case histories of flexible barriers hit by landslide debris, e.g. the
Jordan Valley cases reported by Kwan et al. (2014), will be carried out. The technical development work aims
to promote use of advanced numerical tools for design of natural terrain hazard mitigation measures.
Figure 5: Simulated Debris-barrier Interaction by LS-DYNA
CONCLUSIONS
Technical development work to enhance the understanding of the dynamic interaction between landslide
debris and debris-resisting structures has been initiated with the ultimate goal of optimizing the current design
approach and possibly developing new design methods. Notable areas of recent technical development have
been presented in this Paper. The subject involves study of debris dynamics and structural response of barriers
through physical tests and numerical analyses. It calls for the collaboration between geotechnical profession
and experts in other engineering disciplines. Much remains to be learnt and developed, and further
knowledge-based and technology-driven advances are anticipated.
ACKNOWLEDGEMENTS
This paper is published with the permission of the Head of the Geotechnical Engineering Office and the
Director of Civil Engineering and Development, the Government of the Hong Kong SAR.
REFERENCES
ARUP 2013. Pilot Numerical Investigation of the Interactions between Landslide Debris and Flexible
Debris-resisting Barriers. Report prepared for Geotechnical Engineering Office, Hong Kong. ARUP, 143p.
Bowman, E.T., Laue, J., Imre, B. & Springman, S.M. 2010. Experimental modelling of debris flow
behaviour using a geotechnical centrifuge. Canadian Geotechnical Journal, vol 47, pp 742-762.
Duffy, J.D. 1998. Case studies on debris and mudslide barriers systems in California. Proceedings of the
One Day Seminar on Planning, Design and Implementation of Debris Flow and Rockfall Hazards
Mitigation Measures, Hong Kong, 27 October 1998. Edited by C.K. Lau, R.P. Martin and K.T. Chau, The
Association of Geotechnical Specialists (Hong Kong) Ltd. and The Hong Kong Institute of Engineers
(Geotechnical Division), pp 77-90.
EOTA 2008. Guideline for European Technical Approval of Falling Rock Protection Kits. European
Organisation for Technical Approvals (EOTA), 53 p.
138
GEO 2014. Guidelines on Enhanced Approach for Natural Terrain Hazard Studies. GEO Technical Guidance
Note No. 36 (TGN 36), Geotechnical Engineering Office, Hong Kong, 18 p.
Hungr, O. 1995. A model for the runout analysis of rapid flow slides, debris flows and avalanches. Canadian
Geotechnical Journal, vol. 32, pp 610-623.
Kwan, J.S.H. 2012. Supplementary Technical Guidance on Design of Debris-resisting Barriers (GEO Report
No. 270), Geotechnical Engineering Office, Hong Kong, 88 p.
Kwan, J.S.H., Chan, S.L., Cheuk, J.C.Y. & Koo, R.C.H. 2014. A case study on an open hillside landslide
impacting on a flexible rockfall barrier at Jordan Valley, Hong Kong. Landslides, vol. 11, pp 1037-1050.
Kwan, J.S.H. & Cheung, R.W.M. 2012. Suggestions on Design Approaches for Flexible Debris-resisting
Barriers (GEO Discussion Note 1/2012). Geotechnical Engineering Office, Hong Kong, 91p.
Kwan, J.S.H., Koo, R.C.H & Ng, C.W.W. 2015. Landslide Mobility Analysis for Design of Multiple
Debris-resisting Barriers. Canadian Geotechnical Journal, accepted (in press).
Kwan, J.S.H. & Sun, H.W. 2006. An improved landslide mobility model. Canadian Geotechnical Journal, vol.
43, pp 531-539.
Lo, D.O.K. 2000. Review of Natural Terrain Landslide Debris-resisting Barrier Design. GEO Report No. 104,
Geotechnical Engineering Office, Hong Kong, 91 p.
Savage, S.B. & Hutter, K. 1989. The motion of a finite mass of granular material down a rough incline.
Journal of Fluid Mechanics, vol 199(1), pp 177215.
Sun, W.H. & Law, R.P.H. 2012. A Preliminary Study on Impact of Landslide Debris on Flexible Barriers
(GEO Technical Note No. TN 1/2012). Geotechnical Engineering Office, Hong Kong, 43 p.
Wartmann, S. & Salzmann, H. 2002. Debris flow and floating tree impacts on flexible barriers. In Proceedings
of the Conference on Natural Terrain - A Constraint to Development, Hong Kong, 14 November 2002.
Institution of Mining and Metallurgy, Hong Kong Branch, pp 125-131.
Wendeler, C., McArdell, B.W., Rickenmann, D., Volkwein, A., Roth, A. & Denk, M. 2006. Field testing and
numerical modelling of flexible debris flow barriers. Proceedings of the Sixth International Conference of
Physical Modelling in Geotechnics, Hong Kong, 4-6 August 2006. Edited by C.W.W. Ng, L.M. Zhang, and
Y.H. Wang, pp 1573-1604.
WSL 2009. Full-Scale Testing and Dimensioning of Flexible Debris Flow Barriers. Swiss Federal Institute for
Forest, Snow and Landscape Research (WSL), 22 p.
139
140
Performance Monitoring of a 50 m High 75 Cut Slope Reinforced

with Soil Nail Bars Made from Glass Fibre Reinforced Polymer
(GFRP) at Ho Man Tin Station
A.K.L.Kwong* & J.S.S.Chim
MTR Corporation, Hong Kong
ABSTRACT
The MTR Corporation (MTR) Kwun Tong Line Extension (KTE) project is a 2.5 km long
underground extension of the existing Kwun Tong Line (KTL) running from the current terminus
at Yau Ma Tei (YMT) station to a new station terminus in the Whampoa District. An
intermediate station at Ho Man Tin (HOM) will serve as an interchange station with the future
Shatin to Central Link (SCL).
Site formation for the HOM station was carried out by open cut excavation with a maximum
depth of 50 m in soil/weathered rock, and 60 m in weathered rock/competent rock at sections
where the KTE caverns span across the station. As there is a future development proposed to the
north of the station box, and given that piling works will be required for this development, glass
fibre reinforced polymer (GFRP) bars were used as a substitute for high yield steel deformed bars
in the soil nailing system.
The station boxs excavation outline was formed by phased cutting of slopes in fill,
completely decomposed granite (CDG) and highly decomposed granite (HDG). The cutting and
strengthening works were carried out in stages as excavation proceeded downward. Steep slopes
ranging from 45 to 75 were formed and soil nails were used to reinforce these slopes.
A total of 35 nos. of soil nails (both high yield steel and GFRP) have been strain gauged and
have surface prisms installed at the proximal ends. Twenty inclinometers were installed and
numerous ground settlement markers and piezometers covered the whole site, targeting the crest
of the formed slopes. Utility monitoring points were used to monitor any nearby sensitive
receivers. The instrumentation was used to assess the state of deformation of the soil cut slopes
and the buildup of loads in the soil nails.
This paper describes the detailed monitoring system to validate the soil nail system during
construction. 5% of the installed soil nails were selected for monitoring of their movement and
tensile forces of the soil nail system by setting AAA response levels based on Plaxis modeling.
Surveying tools such as prism targets on soil nail heads and on the slopes shotcreted surface
were also used. The motivation for this comprehensive instrumentation scheme was to adhere to
an observational method to monitor and validate the performance of the soil nail system as the
excavation progresses. After significant horizontal movement of about 40 mm was observed
during the intermediate stage of the excavation, additional nails with load cells were installed and
nail installation method refined. This paper discusses the benefit of such observational approach
that made use of the measured movement and interactive forces between the nail and the soil to
calibrate the finite element models for subsequent validations and predictions.
*Note: Formerly MTR Corporation, Hong Kong
1 INTRODUCTION
1.1 Project background and description
The KTE project is an extension of the existing MTR Kwun Tong Line from Yau Ma Tei Station to
Whampoa, with two new stations at HOM and Whampoa. Passengers can interchange at the proposed HOM
141
Station with the future Shatin to Central Link. It will provide convenient, fast and reliable means of public
transport between Yau Ma Tei and Whampoa, and will enable residents in Ho Man Tin, Hung Hom and
Whampoa area to have direct access to MTR service, saving time for interchange from road transport to the
railway network. The alignment of the KTE will extend from the existing overrun tunnel of Yau Ma Tei
Station, run along Gascoigne Road cross Wylie Road and reach HOM Station at the site of the ex-Valley Road
Estate. It will then run through Chatham Road North via Wuhu Street and Tak Man Street, and extend to
Whampoa Station at Tak On Street.
HOM Station is located at the site of the former Valley Road Estate (Figure 1). The station comprises a
large box that will contain the KTE station cavern (KTE platforms) and the SCL station box, which is located
above the KTE station. The station is to be constructed beneath the filled platform (+37 mPD) occupied by
the former Hong Kong Housing Authoritys Valley Road Estate and is bounded by Fat Kwong Street (to the
North), Chung Hau Street (to the West) and Yan Fung Street (to the East). The site is also allocated to a MTR
property development including two level of semi-basement parking and 10 numbers of 25-storey residential
towers. There are no buildings or structures adjacent to the western slopes of this study but there is a live 900
mm WSD freshwater main running parallel and offset 3 m from the slope crest.
Figure 1: Site Location Plan
1.2 Geological setting

The station lies almost entirely within the Kowloon Granite which comprises a mantle of decomposed granite
overlying bedrock to varying depths. The granite is fine to medium grained in the HOM and Whampoa areas.
The dominating feature for the northern HOM station box is a NW-SE trending fault zone, which has resulted
in the formation of a deep valley with extensive weathering and shearing of the underlying rock mass. It is
noted that in the deepest part of the valley, weathered rock extends to an approximate depth of -35mPD or
more. This zone is about 10 m wide at -12 mPD (formation level of HOM) and extends through Chatham
Road to the Wuhu Street playground area to the SE. There is evidence of small amounts of alluvium and
colluvium along the valley floor.
The natural catchment (Lo Lung Hang) of the HOM area, prior to backfilling in the 1960s, has been
interpreted as having a SE descending valley passing through the present Chatham Road North and ending in
Wuhu Street and discharging to the sea to the south.
Fill materials up to 30 m thick are present to the north of HOM Station, and they are comprised of
predominantly sand with varying degrees of silt, clay and gravel with SPT varying from 5 to 40. Underlying
the fill is localized alluvium of up to 5 m thick which comprises of clay materials with silt of
reddish/yellowish brown in color with SPT varying from 5 to 30. Saprolite has a thickness varying from 5 m
to 20 m in general, which is described as CDG comprised mainly of sand, with silt and gravel with SPT
varying from 10 to 100. Grave IV (HDG) is comprised mainly as gravel size rock fragments and quartz, with
generally sand and silt materials.
142
2 DESIGN BASIS
The station boxs excavation outline was formed by phased cutting of slopes in fill, and in weathered rock
(CDG/HDG). The slopes ranging from 45 to 75 were strengthened with soil nails as the excavation
proceeds down from the existing road elevation (+37 mPD) to the final excavation level at about -12 mPD,
which results in an overall total excavation height of about 50 m in this study within the fault zone.
In the implementation of slope strengthening works, conventional high yield deformed steel bars and glass
fibre reinforced polymers (GFRP) bars were used as soil nails. Given that piling works may be required for
the future development to the north, the use of GFRP bars was warranted as a substitute for conventional high
yield deformed steel bars in the soil nail system so that they could be broken out later. Material properties and
testing of GFRP bars were described in detail by Swann et al. (2013).
The slope under this study has cut through the Fill material which has a thickness of about 28 m, whereas
those in CDG, HDG and MDG are 7 m, 10 m and 5 m respectively. The slope angle formed in Fill was 45,
in CDG and HDG was 75 and that in MDG was 90. In order to maintain the slope in stable condition with a
global factor of safety in limiting equilibrium greater than 1.2, 40 mm GFRP soil nails were installed typically
2 m vertical spacing in Fill and 1.3 m vertical spacing in CDG and HDG. The horizontal spacing of all the
nails was 2 m. The soil nail was typically 20 m long. Figure 2 shows the simplified geological Section 20 for
this study.
Figure 2: Typical Geology in Section 20 (Section A-A in Figure 1)
The ground movements were estimated for Section 20 in this study by using the computer program Plaxis.
Soil nails and shotcrete were modelled to be in place prior to excavation for the next level of soil nails. The
settlement along the crest of the west portions of the cut slope was calculated as 20 mm. Details of the
material properties and design can be found in Jacobs China Limited (2011).
143
The following response levels (ALERT, ACTION and ALARM) based on general empirical rule were
applied:
ALERT LEVEL
ACTION LEVEL
ALARM LEVEL
Monitoring Type
Ground Settlement
Building Settlement and
Tilting
Utilities Settlement and
Tilting
Groundwater Drawdown
Inclinometer
- set at 50% of maximum allowed movement.

Table 1: Summary of Action Response Levels
ALERT
ACTION
25 mm
40 mm
12.5 mm
20 mm
1:600
1:375
10 mm
16 mm
1:600
1:375
500 mm below reference
GWL
GWL
25 mm
40 mm
ALARM
50 mm
25 mm
1:300
20 mm
1:300
GWL
50 mm
3 CONSTRUCTION METHOD AND SEQUENCE

Construction was based on excavation in alternating panels of limited width (12 m to 15 m) in fill such that
any induced slope movements will be minimised. The construction sequence (Figure 3) was as follows:
Figure 3: Nail Installation Sequence
i)
Excavation to the level 500 mm below the 1st row of soil nails to form a temporary slope profile,
followed by installation of A393 wire mesh and shotcreting (100 mm thick).
144
ii)
iii)
iv)
v)
vi)
vii)
Installation of 1st row of soil nails.

Applied second layer of A393 wire mesh and shotcreting (100 mm thick) to the 1st row of soil nails.
Excavation to the level 500 mm below the 2nd row of soil nails to form a temporary slope profile,
followed by installation of A393 wire mesh and shotcreting (100 mm thick).
Installation of 2nd row of soil nails.
Applied second layer of A393 mesh and shotcreting (100 mm thick) to the 2nd row of soil nails.
Repeat until all rows of nails were formed.
4 EXCEEDANCE OF AAA LEVELS

Excavation and installation of the soil nails began in the middle of March 2012 at elevation of +37 mPD.
After the 4th layer of soil nails was completed at elevation of +30 mPD in early May 2012, horizontal
movement and vertical settlement started to increase between Sections 19 to 21 (horizontal length of about 50
m) to a magnitude that became a safety concern if further movement continued at the same rate.
The monitoring instruments between Sections 19 and 21 comprise inclinometers, settlement markers,
prisms mounted on the soil nail head, strain gauges along the nail and piezometers.
Section 20 (half way between Sections 19 and 21) was considered the most critical section because of the
largest thickness of fill here.
The monitoring instrument at this section showed the following readings when excavation reached +26
mPD:
i)
ii)
iii)
iv)
v)
Inclinometers 40 mm horizontal movement at the middle and 16 mm at the crest of the slope, had
exceeded the Action Level (set at 40 mm) as indicated by IN-03B and IN-16A.
Settlement Markers 30 mm to 35 mm vertical settlement at the crest, had exceeded the Alert Level
(set at 25 mm) as indicated by GSM2-37B.
Prisms 10 mm to 20 mm horizontal movement towards the open face and 5 mm to 15 mm vertical
downward movement, had reached the Alert Level (set at 10 to 20 mm for vertical).
Strain Gauges recorded 60 to 100 kN in tension in mid May 2012 at the row at +30 mPD.
Piezometers - showed groundwater table was below the intermediate excavation level.
The predicted horizontal movement at the excavation level (+24 mPD, 7th layer) was about 20 mm. Based
on the monitoring results, at one third of the current excavation level, the recorded horizontal movement was
already 40 mm which had exceeded 2 times the prediction. The concern was that if excavation continued
downward to the final formation level of -12 mPD, and with the current rate of horizontal movement and
settlement, excessive movement of the slope could cause failure and no remedial works would be considered
effective in stabilizing it.
5 ACTION PLAN
The use of GFRP soil nails in loose to medium dense fill slope at angle of 45 and a thickness of about 28 m
was unprecedented. Together with the site observations that washout was observed in one of the nail heads
during heavy rain in April 2012 and higher than normal grout intake was recorded in 2 to 3 nos. of nails, a
number of upgrading works were implemented as follows:
i)
ii)
iii)
iv)
v)
vi)
vii)
Entire area above the slope crest to have concrete paved.

Acceptance tests carried out on those nails which had high grout intake.
Additional settlement monitoring points installed at the 900 mm water mains.
Additional raking drains installed to relieve the build-up of water pressure within the slope.
Change unsupported air-lifting drilling to wash boring with temporary casing supported during
drilling.
Soil nails along 8th layer increased to 24 m long from 20 m so that bondage could be set in the CDG
layer.
Installed one additional layer of 24 m long, 40 mm GFRP nails between 6th and 7th layer, pre-stressed
to 120 kN and monitored with a load cell.
145
viii)
Based on the observed movement and nail force recorded in the nail, Plaxis modelling was refined and
arrived with a new prediction of horizontal movement to about 120 mm, vertical settlement to about
100 mm under the following scenario when the final formation level was reached.
a. Frictional angle of fill reduced from 37 to 33.
b. Shotcrete thickness increased from 200 mm to 300 mm.
c. Case A. Fill constitutive model using Mohr-Coulomb (=37)
d. Case B. Fill constitutive model using Hardening soil (=37, unload modulus Eurref=2.5xE50ref),
where Eurref refers to unloading modulus and E50ref refers to loading modulus at 50% peak load.
e. Case C. Fill constitutive model using Hardening soil (=37, unload modulus Eurref=2.0xE50ref)
f. Case D. Fill constitutive model using Hardening soil (=33, unload modulus Eurref=2.0xE50ref)
g. Case E. Same as Case D but slope surface without immediately meshed and shotcreted.
Since the Mohr-Coulomb model predicted heave in the early stages of unloading which was unrealistic, the
hardening model in Case E was considered appropriate in the refined modelling for prediction of AAAs
levels down to the final formation level because it closely matched the measured movement at IN-03B and
nail force up to the study level of +20 mPD. Details of the hardening model parameters and modelling
sequence can be found in Jacobs China Limited (2011).
Table 2 shows the horizontal movement of IN-03B at each stage of unloading for each of the scenarios
studied.
Table 2: Sensitivity of horizontal movement (mm) under Different Cases Studied
Stage
Case A
Case B
Case C
Case D
Apply Surcharge
2.07
1.60
1.62
2.76
Create Initial Benched
10.87
13.46
16.01
18.01
Profile
Install 1st Row
10.87
13.70
16.28
18.69
Install 2nd Row
10.84
13.42
15.93
18.36
Install 3rd Row
10.83
13.18
15.56
18.02
Install 4th Row
10.98
15.64
18.17
20.66
Install 5th Row
11.45
18.67
20.90
24.32
Install 6th Row
13.99
22.28
24.68
29.12
Install 7th Row
16.09
27.48
29.81
35.57
Grout additional Row
18.26
27.47
29.84
35.67
Install 8th Row
18.74
30.72
33.25
41.23
Install 9th Row
20.52
35.38
37.74
46.09
Install 10th Row
21.62
39.58
42.34
52.19
Install 11th Row
23.12
39.58
42.34
52.19
Install 12th Row
25.35
44.03
46.29
57.04
Install 13th Row
27.64
48.94
50.40
62.56
Install 14th Row
29.44
52.23
53.95
68.08
Install 15th Row
30.53
54.05
56.03
69.33
Install 16th Row
34.46
56.85
58.08
73.88
Install 17th Row
37.00
61.47
63.11
76.70
Install 18th Row
42.45
67.77
68.77
81.83
Install 19th Row
44.80
72.04
73.37
86.87
Install 20th Row
47.56
75.81
77.03
88.62
Install 21st Row
51.02
81.89
81.17
95.83
Install 22nd Row
53.54
85.30
85.63
99.20
Install 23rd Row
58.19
94.35
94.57
107.66
Install 24th Row
62.58
101.11
103.65
116.69
Install 25th Row
66.30
103.28
108.20
122.26
Install 26th Row
70.06
109.36
114.73
124.06
Install 27th Row
71.50
110.74
114.83
125.66
Rock Excavation
71.80
111.47
115.46
126.28
146
Case E
31.68
32.54
38.38
38.51
43.27
48.77
54.98
59.82
59.82
65.46
70.57
72.21
74.23
79.34
84.74
89.72
93.27
96.04
101.02
109.52
118.43
119.50
130.17
130.28
130.89
Comparing the lateral movement between Cases D and E at the stage Install 5 th Row, it was noted that
excavation to a lower bench without immediate application of shotcrete would result in 30% additional
movement caused by relaxation.
Results of the sensitivity on thickening the shotcrete layer indicated no obvious reduction of movement can
be detected and therefore shotcrete thickness was not increased for the subsequent layers.
For Case E (Figure 4), there was no apparent shear strains developed within the fill material and it can be
concluded that preferential shear plane cannot be developed when reaching the final formation level.
Figure 4: Shear Strain Concentration Down to Final Excavation Level
Revised AAA values as shown in Table 3 were re-established based on the worst case scenario of backanalysis carried out from Case E, which assumed a delayed in shotcrete applied for the subsequent layer.
ALERT
(80% of FEM Prediction)
ACTION
ALARM
Table 3: Summary of Revised AAA Levels

For IN-03B, to For IN-03B, to base For GSM2-37B, to
base of fill
of CDG
base of fill
71 mm
88 mm
75 mm
For GSM2-37B, to
base of CDG
89 mm
75 mm
93 mm
79 mm
93 mm
83 mm
102 mm
87 mm
103 mm
For Case E, the predicted horizontal movement to the base of CDG is presented in Figure 5.
Figure 6 shows a comparison of the FEM prediction versus the actual readings of vertical settlement up to
the study level of +20 mPD and down to the base of the fill (~+8 mPD).
147
Figure 5: Predicted Horizontal Movement from Case E to base of CDG.
Figure 6: Case E, Predicted Vertical Settlement Compared to Actual Reading.
148
6 CONCLUSIONS
Close monitoring continued as excavation continued down to the final formation level of -12 mPD. Due to
the additional nails installed together with improved drilling process, there were very little additional
movement and settlement recorded after the AAA values were reset. IN-03B recorded a final horizontal
movement of about 70 mm (predicted 120 mm) and GSM2-37B recorded a final vertical settlement of about
90 mm (predicted 100 mm) was achieved when the final formation level was reached in August 2013. The
horizontal movement was over predicted about 50 mm, but the vertical settlement was very close when the
final level was reached. There were no distresses observed on the slope and all the nail forces were well
within the working load. Station structures and slope backfilling was completed in early 2015.
The motivation for this comprehensive instrumentation scheme was to allow validation of the computer
model in assessing the performance of the soil nail system as the excavation progresses. In this study,
additional nails with load cells were installed and installation method refined based on re-assessment of
different constitutive models adopted and material shear strength. The study provided a good case history
which reinforced the benefit of such observational approach, allowing economical and safe design and
construction of high and steep slopes reinforced with GFRP nails.
ACKNOWLEDGEMENTS
The writers are grateful for the support of MTR and Jacobs China Limited. Their support contributed in
expediting the approval process. However, the contents of this paper do not necessarily reflect the views and
policies of these supporting organizations, nor does the mention of trade names and commercial products
constitute endorsement or recommendation for use.
REFERENCES
Jacobs China Limited November 2011. Temporary Works Design Report, Package B1, Temporary Works for
Soil Cut Slopes in HOM North Station Box, 206 p.
Swann, L.H., Ng, A. and Mackay, A.D. 2013. The Use of Glass Fibre Reinforced Polymer Bars as Soil Nails
to Permit Future Housing Development, Hong Kong Special Administrative Region. HKIE Geotechnical
Division Annual Seminar, pp. 165 to 171.
149
150
Considerations for Consolidated Undrained Triaxial Testing of

Saprolites for Soil Structure Interaction Design
Ada Chan
Leighton Asia, India and Offshore (LAIO)
A. D. Mackay
ALYSJ joint venture, Doha, Qatar
N. R. Wightman
Aquaterra Consultants Limited
ABSTRACT
Hong Kongs sub-tropical climate results in weathering of in-situ rock to a saprolite. These have
unique physical characteristics with similarities to both rock and soil, influencing the shear
strength values determined from the Consolidated Undrained (CU) triaxial test. The parent rock
forming the saprolite retains a relict structure increasing the shear strength at relatively small
strains, of say 3% or less, represented by the peak shear strength during the CU test. After this
structure, termed bonding, is ruptured the shear strength decreases and is mobilised by the interparticulate interaction. This paper summarises the main test stages for the CU test in Hong Kong
and how these are influenced by relict structure from random test specimens of granitic saprolite.
The difference in the peak and residual shear strength at different strain ranges is presented. It is
concluded that the upper bound shear strength parameters of saprolite representing the relict
structure prior to rupture should be adopted for engineering works where small strain
displacement may be anticipated.
1 INTRODUCTION
Numerous Consolidated Undrained (CU) triaxial tests are routinely carried out in Hong Kong to determine
effective shear strength parameters. Saprolites have variable and unique physical characteristics that need to be
identified to better understand the influence on the engineering works. The presence of the relict structure may
lead to inconsistent saturation, swelling and / or volume reduction during the consolidation stage of the CU
testing process, rupturing the relict structure (bonds) and, in extreme situations, restructuring the particles. This
paper provides examples from random specimens where the shear strength has been reduced due to rupture of
the specimen during testing. Recommendations for more sensitive testing, such as the use of side drains to
provide a more even permeation is made. The effects of variable shear strength values, representing strain
ranges within the in-situ and failed (residual) saprolite conditions is given.
2 TEST PROCEDURE
2.1 Hong Kong laboratory testing
In Hong Kong great importance is given to representative accurate soil test data to support proposed site
investigation, design and construction (BA, 2005). This requirement is emphasised in the Building Ordinance,
CAP 123, Section 17 (1) subsection 6(b), which requires material testing to be carried out in order to gain
approval for the building plan submission. If these requirements are not fulfilled; as set out in GEO, 2001 &
HKAS, 2009; the Building Authority (BA) may refuse approval of plans or consent for commencement of
building works (BA, 2009). Hong Kong therefore sets very high levels of quality and stringent testing
standards for construction works to proceed. The testing procedure and standards for Hong Kong laboratory
testing is set out in GEO, 2001.
151
2.2 The Consolidated undrained triaxial test

The CU test determines effective shear strength parameters from soil specimens using triaxial testing apparatus,
presented in Figure 1, and follows stringent accreditation procedures (HKAS, 2009).
Figure 1 - Triaxial test apparatus (GEO, 2001)
The main stages of the CU test procedure are saturation, consolidation and compression, each of which need
consideration for the type of ground being tested to ensure representative results are obtained. A summary of the
aim, procedure and relevant considerations for each test stage is summarised in Table 1 with considerations how
each stage may influence the shear strength.
152
Aim
SATURATION ensure voids between
particles in the test
specimen are saturated
CONSOLIDATION Ensure desired state of

effective stress is
achieved prior to
compression
COMPRESSION Shear failure.
Table 1 Considerations for Each Stage of the CU Test

Test Procedure (GEO,
Important Considerations
2001)
Increase in back pressure
The effective pressure increase in the specimen may exceed the in(pore water pressure) to cell
situ effective pressure,
pressure application (B) not The rate of back pressure application ensuring consistent
exceeding increments
permeation throughout the specimen;
between 5 to 25kPa or until Ensuring the volume increase is not excessive and or localised.
B exceeds say 0.95.
Increase to the desired
The rate of cell and back pressure application,
effective stress by raising
Suitable incremental increases in the cell and back pressure
the cell pressure, if
as the confining pressure is applied,
required, and / or
The rate of steady pore water pressure change to determine
dissipating back pressure
the completion of the consolidation,
during the test. The back
Dissipation of excess pore pressure. Typically taken as 95%
pressure does not fall below
dissipation,
the final saturation pressure Ensuring volume increase is not excessive and / or localised.
Increase the axial pressure
Rate of axial strain (typically based on the time taken to
by a consistent increase in
achieve pressure)
axial strain
Defining failure, either by volume increase or rapid decrease
in compression
For soils, which have shear strengths sensitive to their in-situ environments, such as saprolites, particular
attention for the test scheduling and supervision may include:
The use of side drains to enhance permeation during saturation;

The location(s) of pore-water (back) pressure application and measurement, typically taken from the
top and base of the test specimen respectively, refer to Figure 1;
Effects of variations within the test specimen;
A three day testing period is considered normal (MTRCL, 2009). Whilst a suitable test period is needed
this must not influence the time needed to obtain representative results, particularly for a fine grained soils
with a low permeability, typical of saprolites.
2.3 Key roles, responsibilities and interaction for the testing
Given the importance of identifying variations in the test specimen, particularly effects on the permeability,
key individuals, roles and particular tasks for individuals responsible for the laboratory test schedule and
supervision are outlined in GEO, 2001; HKAS, 2009; BA, 2006 and BA, 2010. This is summarised in Table 2.
Lead Role
Specifier
Supervising
Engineer
Table 2 Summary of Key Individuals, Roles and Particular Tasks in Laboratory Testing
Role
Ref.
Tasks (GEO, 1992 & 2001)
Laboratory
BA,
Prepare a soil description, including the possible origin and whether the soil
scheduling
2006 &
particles are susceptible to crushing;
2010
Request details of the Supervising Engineer;
Supervision HKAS, Verify the field description is appropriate, in particular presence of fabric,
& reporting 2009
discontinuities, and other geological effects affecting the test
Upon completion specimen description, including fabric and unusual
features affecting the result
The qualifications for both the Specifier and Supervising Engineer are Qualified Geotechnical Engineer
and / or Engineering Geologist (GEO, 2001) and a Corporate Member of a professional body and a
Registered Professional Engineer, Geotechnical with designations RPE/G (BA, 2001; BA, 2010 and HKAS,
2009).
The GI and laboratory contractors often have different contractual requirements and reporting streams. It is
therefore vital that the laboratory test scheduling, typically prepared by the ground investigation (GI)
supervisor, and the laboratory technicians, headed by the Supervising Engineer, have a common understanding
153
of the specimens being tested and flexibility in testing procedure as more information becomes available. This
needs to be incorporated in the contractual set up.
As the initial soil description is typically based on a jar sample from the GI contractor, this should be
updated during laboratory test specimen preparation, with feedback provided to the Specifier (GEO, 2001). An
update of the drillhole log and final GI report following specimen preparation and ideally test completion,
should be carried out to ensure consistency between the GI log description and the laboratory test findings.
Notwithstanding as GI reports typically require prompt submission the additional time taken to incorporate
laboratory test findings can be problematic.
3 CHARACTERISTICS OF SAPROLITES
3.1 Engineering geological properties influencing shear strength
Weathering has a significant effect on the engineering properties of rock, therefore careful description of
the state of weathering is required (GEO, 1992). Weathering can either be disintegration (mechanical) and /
or decomposition (chemical) with saprolites being the product of more intense chemical weathering. For rocks
with a plutonic and volcanic origin, common in Hong Kong, decomposition effects are dominant. However, if
disintegration processes have affected the specimen, ultimately influencing the shear strength, this needs
inclusion in the sample and laboratory specimen description (GEO, 2007). Important influences on the
physical properties of the rock from both mechanical and chemical weathering processes are summarised in
Table 3:
Table 3 Weathering Effects Affecting Engineering Properties (Shear Strength) of Saprolites
Processes
Relevant descriptions and / or effects (GEO, 2007)
MECHNICAL - water
Micro-fracture growth. Typically resulting from de-stressing from the decomposition
absorption and release; stress
of quartz and feldspar components in the rock, becoming more obvious in coarse
and temperature changes
grained granite
Increase in more stable low strength clay minerals;
DECOMPOSITION Increase in pore space and associated void ratio and porosity;
breakdown of minerals to a
Variable weathering front with associated permeability variation;
more stable chemical state
Reduction in grain bonding, decreasing material strength,
Clay mineral concentration along discontinuities; particularly in saprolites close to
rock and soil interfaces.
The stress history imposed by influences such as groundwater fluctuation, recent erosion and recent
deposition needs consideration when judgments for the test schedule.
3.2 Physical parameters for saprolites
Typical ranges of physical parameters including void ratio, unit weight, effective shear strength and
permeability for different types of granitic saprolites from published data are summarised in Table 4.
Table 4 Typical Ranges of Values for Geotechnical Parameters for Completely Decomposed Granite (CDG)
(GEO, 1992, 1997, 2000 & 2007 & Martin, 2003)
Reference
GEO, 2000
Martin,
2003
GEO, 1997
Description
(GEO, 1992)
Void
Ratio
Upper CDG limit

Lower CDG limit
HDG, Mod weak
HDG, Very weak
weak
CDG, Very dense
CDG, Dense
CDG, Loose
0.4 0.6
0.7 1.1
Unit Weight (kN/m3)

Sat. Bulk
Bulk
Dry
16-21
14-19
16-18
12-15
22 25
20 24
20.5
17 20
22.5
20 21
16 19
18.5 20
14 16
17.5 18.5 12 14
154
Shear Strength Parameters

Angle friction
Cohesion
35 44
5 - 15
38-44
0-10
33-36
2-6
Permeability
10E-5 10E-7
As shown a reduced unit weight or density corresponds with decreased shear strength, increased
permeability and greater weathering grade.
An example of the variability of particle size distribution (PSD) from a saprolite with decreased weathering
grade with depth is shown in Figure 2 (Wightman, 2009). The borehole retrieved samples with an increase in
gravel in less weathered saprolite (highly decomposed rock layer) and an increase in silt and clay in more
intensely weathered saprolite, located immediately below the residual soil layer. The increased density and
decreased fines content corresponded was increased strength accordingly.
Increase in weathering
grade.
Figure 2 - Change in PSD of Saprolite with Depth Showing Increasing Silt and Clay Content (Wightman, 2009)
As referenced in GEO, 2000, the Completely Decomposed Volcanics, comprising Tuffs and Rhyolites,
have the same bulk and dry unit weight and permeability ranges as CDG with shear strength parameters
ranging from 5 to 15kN/m2 (cohesion) and 32 to 38 degrees (angle of internal friction). The clay content,
fabric and grading associated with the upper and lower limits of CDG decomposition grade are summarised in
Table 5.
Table 5 Grading of Remolded Samples of CDG (Martin, 2003).
Limit
Upper
Lower
Clay, % by
weight
10 25
30 - 50
Grading (2)
Description
Silty clayey, very sandy gravel or very gravelly sand
Slightly gravelly, sand silt / clay
Gravel (%)
30 - 50
10-30
Sand (%)
30-50
30-50
Silt / Clay (%)

10-20
30-45
Many of the saprolites physical properties referenced above, particularly fabric, mineralogy, grading clay
content, density can be approximately estimated from an accurate geological description. This can then be
used for advance considerations in preparing the test schedules and appropriate procedures.
4 EFFECTS OF THE TRIAXIAL TEST ON SAPROLITES
A number of Hong Kong laboratory tests were checked as part of the routine input to site investigation, design
and construction works. In almost all reports the saprolite laboratory tests included a brief geological
description sent from the Specifier obtained from jar samples with no subsequent update. It is noted that
GEO, 2001 requires the following important information:
Representation of the geological units and update of inaccurate description to the specifier;
A detailed test specimen description, including the features that may render the test results invalid;
155
Incorporation of classification tests results (particle size distribution and Atterberg limits), carried out
on the sample into the geological description;
Presentation of both laboratory and jar sample descriptions, including detailed soil description, soil
fabric and any unusual features, in accordance with GEO, 1992.
Random test specimens of CDG from a laboratory test report, which included CU triaxial test on CDG for
an MTRCL project, dated mid-2009, was checked in further detail by the authors, as summarised in Table 6.
Table 6 Depth (metres below ground level, mbgl), Sample Type and Soil Descriptions
for Randomly Selected CDG Specimens
No.
1
2
3
Specimen
mbgl
17.2 17.45
23.8 24
34.3 34.5
Soil Description
Dark brown, very silty, very gravelly SAND
Orangish brown, slightly gravelly, clayey SILT
Brown, very silty, very gravelly SAND
(17.0 17.45 mbgl)

(23.0 24.10 mbgl)
(33.5 34.60 mbgl)
For the specimen and sample disturbance, affecting the results, it was revealed that Specimen 1 had been
taken from a U76 sample; contrary to the requirements in Table 9, GEO, 1997 and the test specimens were
generally taken at or near the base of the sample which have greater sampling disturbance. A summary of the
test results and divergence from the each test stage (GEO, 2001) is summarised in Table 7.
Table 7 Soil Properties, during Saturation and Consolidation
Spec
No.
1
2
3
Saturation*(1)
Ratio
Pressure
(B)
(kN/m3)
(min)
Consolidation*(2)
Pressure
Volume
(kPa)
(mm3)
Cell Back Pore

205 200 200
205 205 205
205 200 200
150
2880
90
Cell
370
430
540
0.97
1
1
Time
Effective
170
230
340
Notes:
*(1)
Small effective pressure, typically taken as 5kPa, not
implemented.
The cell and back pressures often being equal and the
pressure ratio reaching unity.
The pore pressure (u) kept to the maximum allowable
(200kPa), in specimen 1 this is greater than the in-situ
effective stress.
Effective pressure exerted on specimens retrieved from the
greater depths was significantly less than in-situ effective
pressure.
Before After
680 643
606 547
606 557
Strain
Rate,
(mm/
min)
%
5.4 0.05
9.6 0.046
8.0 0.047
Compression*(3)
Dry Density
Wet Density
(kN/m3)
(kN/m3)
Before
14.6
12.5
15.0
After
15.5
13.8
16.4
*(2)
Rate of volume change increased
in Specimen 1 towards
completion, suggesting rupture of
the specimen.
Volume change during
consolidation, which approaches
10% in some specimens, is
significantly greater than the
change that cemented bonds can
withstand thereby causing rupture
Before
19.0
17.6
19.4
After
19.6
18.6
20.1
*(3)
Time taken to reach
consolidation for all
specimens was
significantly
different, yet the
strain rate used in
compression for all
specimens was
similar, say 3mm/ hr
A summary of the parameters recorded for each test specimen at different presumed failures, including
report reference, greatest stress ratio (1/ 3) and maximum pore water pressure, are presented in Table8.
Table 8 Soil Parameters at Different Test Failure Stages
No
1
Failure stage
Reported
Greatest u
Greatest 1 / 3
Reported
Greatest u
Greatest 1 / 3
Reported
Greatest u
Greatest 1 / 3
u
250
307
294
337
346
345
360
429
420
19.5
2.69
6.06
7.9
3.97
5.07
20
3.3
5.6
1
454
254
339
369
342
356
673
449
521
3
120
62
75
92.9
83.4
84.7
179
111
120
156
s
287
158
207
231
212
220
426
280
321
t
167
96
132
138
129
136
247
169
201
p
231
126
163
158
169
175
344
224
254
q
334
192
264
277
258
271
493
338
401
1/ 3
3.8
4.1
4.5
4
4.1
4.2
3.8
4
4.3
The reported specimen failures for 1, 2 and 3, had reached a reported strain () of approximately 20%, 8%
and 20.5% respectively which had drastically exceeded the rupture of the cemented bonds between particles,
in the order of 20% to 100% greatest pore water pressure value (u) at about 4% to 5% strain. The results
were therefore unrepresentative of the reported peak shear strength values. The failures referenced in Table
7 recorded at the maximum pore pressure and stress ratios occurred at relatively similar strains and could be
more representative of the peak failure at or prior to the bond failure and show a significant increase in
effective cohesion and internal angle of friction when compared with the values reported.
Barrelling effect
in specimen
Horizontal, linear failure along discontinuity in specimen
Linear features
Decomposed
feldspar
Localised
kaolin present
Sampling disturbance and smearing influences towards outer edge
Plate 1: Specimen 1 - Failure
Shearing
Plate 2: Specimen 1 - Failure Cut Section
Coarser grained constituents
Fine grained constituents corresponding with failure
Movement over platen
157
Barrelling effect
Failure along finer grained constituents
Decomposed
feldspar
Kaolin
The specimen photographs are presented in Plates 1 to 6 and generally show failures corresponding with
finer grained constituents, possibly resulting from inconsistent permeation through the specimen. The clays are
generally derived from decomposed feldspar and kaolin, which could not have been described from a jar
sample. Specimen 1 shows a horizontal failure, which may be due to the sample disturbance during retrieval
using the U76 tube.
5 CONSIDERATIONS FOR FUTURE TRIAXIAL TESTING ON SAPROLITES
Considerations on improvements, most of which require a closer review of the Hong Kong standards, are set
out below:
Detailed soil descriptions including effects on the laboratory testing. Whilst Hong Kong sets high
standards for the supervising and vetting of staff (BA, 2009 and HKAS, 2009), the qualifications are
ultimately derived from a Corporate Membership of a recognised body (mainly civil engineering
background). As this lacks the knowledge needed for soil descriptions, relevant to the parameters
derived, a more appropriate requirement would be Chartered Geologist with relevant engineering
experience. This would still satisfy the GEO (2001) requirements;
Greater collaboration and interfacing between the GI works and laboratory test team representatives,
ensuring consistency in documentation. Due to the demand for rapid submission of the GI
documentation, early incorporation of the laboratory data, possibly during specimen preparation, should
be carried out;
For saprolite testing, a realistic strain and effective pressure should be adopted during saturation and
consolidation. This prevents rupture of the bonds prior to compression;
Representative effective pressures during schedule preparation, accounting for the soil stress history;
Different permeability ranges within specimens often prevent a consistent saturation. As a result
localised swelling and volume reduction may occur during consolidation. These changes may also
rupture the cemented bonds and, in extreme situations, allowed particles to restructure resulting in a
significant shear strength reduction;
The test procedure should consider routine use of side drains to provide more rapid pore water
permeation and back pressure application and pore water measurement through the top and base of the
specimen; and
Accurate volume measurement, recording localised specimen failure, should be considered.
158
6 CONCLUSIONS
A random review of CU triaxial test results reveal the tests have minimal, recorded geological detail and that
the results obtained may not represent the peak shear strength results and the strength generated between the
cemented bonds from the parent rock. A focus is needed to understand the significance of geological features
on the test result, which may require a greater interaction between the GI and laboratory test sub-contractors
teams and a greater understanding of the significance of the findings from the designer. Although the results
may provide a robust approach to the design they have a danger of being misrepresentative and / or
misunderstood.
ACKNOWLEDGEMENTS
The contents of this paper do not necessarily reflect the views and policies of these supporting organizations,
nor does the mention of trade names and commercial products constitute endorsement or recommendation for
use.
REFERENCES
British Standards Institute (BSI). 1990. Methods of Test for Soils for Civil Engineering Purposes (BS 1377,
1990, Parts 1 to 9). British Standards Institute, London, 1990, 406p.
Building Authority (BA). 2005. Practice Notes for Authorised Persons, Registered Structural Engineers and
Registered Geotechnical Engineers, PNAP 167 (APP 64), Methods for Testing Hong Kong Soils
(GEOSPEC 3 - Model Specification for Soil Testing), Buildings Department (BD), Hong Kong Special
Administration Region (HKSAR) Government.
BA. 2006. PNAP 132 (APP 49), Site Investigation and Ground Investigation. BD, HKSAR Government.
BA. 2010. Code of Practice for Site Supervision, BD, HKSAR Government.
Chen, Y.J. & Kulhawy, F.H. 1993. Undrained strength interrelationships among CIUC, UU and UC tests. J. of
Geotechnical Engineering, American Society of Civil Engineers, vol. 119, no. 11, pp 1732-1750.
Cheung, C.K., Greenway, D.R. & Massey, J.B. (1988). Direct shear testing of a completely decomposed granite.
Proceedings of the Second International Conference on Geomechanics in Tropical Soils, Singapore, vol. 1,
pp109-118.
Eurocode. 1994. ENV 1997-1, 1994, Eurocode 7 Geotechnical Design - Part 1, General Rules. 318
Eurocode. 1999. ENV 1997-2, 1999, Eurocode 7 Geotechnical Design - Part 2, Design Assisted by Laboratory
Testing.
Geotechnical Control Office (GCO). 1987. Guide to Site Investigation (Geoguide 2). Geotechnical Control
Office, Hong Kong, 362p.
GCO. 1988. Guide to Rock and Soil Descriptions (Geoguide 3), Hong Kong, 189p.
Geotechnical Engineering Office, (GEO). 2007. Engineering Geology Practice in Hong Kong, Government of
HKSAR.
GEO. 1992. Guide to Rock and Soil Description, Geoguide 3. Geotechnical Engineering Office, Civil
Engineering Development Department, Government of HKSAR.
GEO. 1993. Guide to Retaining Wall Design (Geoguide 1). GEO, Hong Kong, 276p.
GEO. 1994. Geotechnical Manual for Slopes. GEO, CEDD, Government of the HKSAR.
GEO. 1997. Mineralogical and Fabric Characterisation and Classification of Weathered Volcanic Rocks in
Hong Kong. GEO Special Report No. SPR 1/97. GEO, Civil Engineering Development Department
(CEDD), Government of HKSAR.
GEO. 2000. Guide to Retaining Wall Design, Geoguide 1. GEO, CEDD, Government of the HKSAR.
GEO. 2001. Model Specification for Soil Testing. GEO, CEDD, Government of the HKSAR.
Head K.H. (1992). Manual of soil laboratory testing, Volume 3: Effective stress tests, second edition, John
Wiley & Sons, New York.
HKAS. 1998. Regulations for Laboratory Accreditation (2002, Fifth Edition). Hong Kong Accreditation
Service, 25p.
Hong Kong Accreditation service (HKAS). 2009. Hong Kong Laboratory Accreditation Scheme (HOKLAS),
003 (9th Edition).
Martin, R.P. (2003). Review of geological aspects of slope engineering. Transactions of the Hong Kong
159
Institution of Engineers, Vol. 10, No. 3, pp26-33.

Mass Transit Railway Corporation Limited (MTRCL). 2009. Materials and Workmanship Specification for
Civil Engineering Works, Volume 3 of 3, Sections 2225.
Parry, S. & Hart, J.R. 2007. Challenges Facing the Profession. Engineering Geology in Risk Management;
Geological Society of London Bicentennial Celebration 1807-2007, 10 November 2007, pp4144.
Wightman, N.R. 2009. Destabilisation of natural slopes relating to soil density loss strength reduction. HKIE
Geotechnical Division, Proceedings of 29th HKIE Annual Seminar entitled Natural Hillside Study and
Risk Mitigation Measures, 17 April 2009, pp85-90
160
Interface Modelling - A Comparison Study Using Finite Element

and Boundary Element Methods
H. C. Mark Chan
Hong Kong SAR, China
Chao Li
Lambeth Associates Limited, Hong Kong SAR, China
ABSTRACT
Interfaces are often formed during soil-structure interactions. Examples include localised
failure within the soil/rock and slippage between the ground and structure during loading. The
interface usually has a large length-to-width ratio and non-linear slip-stress relation, which
cause substantial difficulties in its modeling. Case studies have been carried out to investigate
how the interface can be properly modeled by examining the behaviour of a Direct Shear Test
on a rock sample with a joint. The Test is first modeled by a Boundary Element Method using a
computer program named FROCK, which incorporates specialised elements to represent
interfaces/fractures. Then the Test is modelled using PLAXIS 2D (based on a Finite Element
Method), which incorporates interface elements to handle slippage. The results from the two
studies are compared. Interesting observations on the sliding behaviour of the interface have
been made and the pros and cons of using the two methods are discussed. Practical applications
in the conducting of laboratory tests and numerical modeling involving interfaces are also
suggested.
1 INTRODUCTION
Interfaces are often formed during soil-structure interactions. Examples include localised failure within the
soil/rock and slippage between the ground and structure during loading. The interface usually has a large
length-to-width ratio and non-linear slip-stress relation, which cause substantial difficulties in its modeling.
Common numerical methods to deal with this problem include the Finite Element Method (Bathe 1982;
PLAXIS 2012), the Displacement Discontinuity Method (Crouch and Starfield 1983; Curran and Vadamme
1983) and the Direct and Indirect Boundary Element Methods (Brebbia et al. 1984; Crouch and Starfield
1983).
Case studies have been carried out to investigate how the interface can be properly modeled by examining
the behaviour of a Direct Shear Test on a rock sample with a joint. The Test is first modeled by a Boundary
Element Method which employs a hybridized scheme using the Displacement Discontinuity and the Indirect
Boundary Element Methods (Chan et al. 1990; Vsrhelyi and Bobet 2000). The scheme has been
implemented in a computer program named FROCK, which incorporates specialised elements to represent
interfaces/fractures. Then the Test is modelled using PLAXIS 2D (based on a Finite Element Method), which
incorporates interface elements to handle slippage. The results from the two studies are compared. Interesting
observations on the sliding behaviour of the interface have been made and the pros and cons of using the two
methods are discussed in this paper.
2 ROCK JOINTS AS INTERFACES
Rock joints (or bedding planes) in otherwise sound rock represent weak zones in which shear failure or
slippage could occur when the rock mass is loaded. A common observation of the shearing behavior of a rock
joint is shown in Figure 1 (Hoek 2015). The rock joint can be viewed as an interface which has a large lengthto-width ratio and non-linear slip-stress relation.
161
(a) Rock Joint under Shearing
(b) A Common Slip-stress relation
Figure 1: Shearing behavior of a rock joint
The relation governing the shear behavior of the joint can be represented by:
= n f ( ) + c
(1)
where , n = shear and normal stresses on the joint surfaces respectively, = slip (or shear displacement)
across the joint surfaces, f ( ) = friction coefficient as a function of and c = cohesive resistance.
3 MODELLING OF THE INTERFACE USING THE DIRECT SHEAR TEST
3.1 Schematic set-up
A basic case of the modeling of an interface is the Direct Shear Test on a rock sample with a joint, which can
be shown schematically in Figure 2. The intact rock is assumed to be linear elastic with elastic constants given
by Table 1 below.
Table 1: Material Properties of Intact Rock Sample
7
Youngs Modulus (E)
2 x 10 kPa
Poissons Ratio ()
0.25
The joint is assumed to have a small aperture and f ( ) is approximated by a multi-linear function shown
in Figure 2(b).
The bottom of the rock sample is kept fixed. It is noted that the stress/displacement conditions at the left
and right ends of the joint (i.e. no shear stress on the vertical and horizontal planes if the end point is regarded
to be on the vertical edge and having a shear stress on the vertical and horizontal planes according to the slipstress relation if it is on the joint) could be ambiguous. However, use of the approximate slip-stress relation at
the end of the joint will not introduce additional external forces or moments at the end point, when compared
with the case where the true slip-stress relation is used. It is considered that only the stress distribution in a
small zone around the end point is affected by the approximation. Therefore, by Saint-Venants Principle
(Sternberg 1954) or the more general Principle of Asymptotic Proportionality (Chan 1990), the effect of the
approximation on the overall behavior of the sample is negligible. A seating load of p = 1000 kPa is applied
and the top of the sample is slowly sheared. The stresses and displacements in the shearing test are modeled
using a Boundary Element Method in Section 3.2 and then modeled using a Finite Element Method in Section
3.3.
162
(a) Problem Set-up
(b) Friction Coefficient

Figure 2: Schematic Set-up of the Direct Shear Test (Chan 1986)
3.2 Modelling using the computer program FROCK

The computer program FROCK (Chan et al. 1990; Vsrhelyi and Bobet 2000; Chan 1993; Chan 1986)
basically uses a Boundary Element Method to model a rock mass with discontinuities such as joints and
fractures. The intact rock is assumed to be homogeneous and linear elastic while the discontinuities can
possess non-linear slip-stress relations. Four Linear Stress Discontinuity Elements are used to model each
163
edge of the rock sample and the rock joint is modeled by four Linear Displacement Discontinuity Elements
with a Linear Displacement Discontinuity Surface Element each at the right end and the left end of the joint.
After application of the seating load, a uniform displacement u is applied slowly to the top edge of the
sample. At smaller values of u the joint does not slide and the shear stress along the joint takes an
approximately parabolic shape. For example, at u = 8 x 10-6 m the shear stress is shown in Figure 3. The
normal stress along the joint is shown in Figure 4. It is noted that the normal stress is higher on the right side
due to induced stresses which maintain moment equilibrium of the sample.
450
FROCK 2e-5
PLAXIS 2e-5
FROCK 8e-6
PLAXIS 8e-6
400
350
Shear Stress (kPa)
300
250
200
150
100
50
0
0
0.05
0.1
0.15
X
0.2
0.15
0.2
0.25
0.3 (m)
Figure 3: Shear Stress along Joint at u = 8 x 10-6 m and u = 2 x 10-5 m
0
0
0.05
0.1
0.25
0.3 (m)
-200
FROCK 2e-5
PLAXIS 2e-5
FROCK 8e-6
PLAXIS 8e-6
Normal Stress (kPa)
-400
-600
-800
-1000
-1200
-1400
-1600
Figure 4: Normal Stress along Joint at u = 8 x 10-6 m and u = 2 x 10-5 m
As the shear load increases sliding occurs first in the central part of the joint and then spreads towards the
-5
two vertical edges of the sample. At u = 2 x 10 m the entire joint has slid. The normal stress along the joint is
plotted in Figure 4. Again it is higher on the right side as required by moment equilibrium. The shear stress is
also higher (in magnitude) on the right side partly due to the higher normal stress there (Figure 3). The slip (
or DS) across the joint is plotted in Figure 5. The slip magnitude is smaller on the right mainly because the
164
higher normal stress there hinders sliding. The slip magnitude is higher in the middle than on the left because
sliding starts in the central part, despite the lower normal stress on the left. As the shear load increases further
-5
the slip on the left catches up. The x-displacement ux along the mid-vertical section at u = 2 x 10 m is plotted
in Figure 6. It can be seen that a jump occurs at the middle due to the slip at the joint. The continuous increase
of ux with y (apart from the joint) is due to the shear deformation of the intact rock. Larger shear loads are not
considered in this paper.
9.00E-06
PLAXIS
FROCK
8.00E-06
6.00E-06
5.00E-06
4.00E-06
3.00E-06
2.00E-06
1.00E-06
0.00
0.05
0.10
0.15
X
0.20
0.25
0.30 (m)
Figure 5: Slip along Joint at u = 2 x 10-5 m

2.00E-05
PLAXIS
1.80E-05
FROCK
1.60E-05
1.40E-05
Ux (m)
DS (m)
7.00E-06
1.20E-05
1.00E-05
8.00E-06
6.00E-06
4.00E-06
2.00E-06
0.00E+00
0.00
0.05
0.10
0.15
0.20 (m)
X
Figure 6: Horizontal Displacement along Mid-vertical axis at u = 2 x 10-5 m
165
3.3 Modelling using the computer program PLAXIS 2D

The computer program PLAXIS 2D (PLAXIS 2012) employs a Finite Element formulation to model general
solid mechanics problems. Using interface elements, slip surfaces such as the rock joint in Section 3.2 can be
modeled. 15-noded triangular elements are used to represent the intact rock while interface elements are used
for the joint. The same seating and shearing loads are applied to the sample (Figure 7).
Figure 7: Finite Element Mesh in PLAXIS 2D
Horizontal and vertical fixity is specified at the bottom of the sample. Pressure and horizontal
displacements are applied at the top of the sample. The two vertical sides are stress-free except for the top and
bottom points.
It is noted that the interface element in PLAXIS is based on a formulation similar to a continuum element.
which is convenient for Finite Element Formulation but also implies a number of limitations. Similar to a
continuum element, a constitutive model can be directly assigned to an interface. However, post-yielding
behavior, and the evolution of yield surface is not straightforward to implement. As a result, the curve in
Figure 2 cannot be directly included in PLAXIS 2D.
Interface elements with Mohr Coulomb constitutive law are used in PLAXIS to represent the rock joint. As
a result, the strength parameters of the interface element are constant under shear loading. Slippage is
developed a result of plastic deformation. After the entire surface has yielded, the load-displacement
relationship can only be specified by a hardening-law, which is not included in the Mohr Coulomb
formulation. The elastic parameters of the interface elements are the same as the rock while the strength
parameters taken from Figure 2(b) are used as in Table 2 below.
Table 2: Material strength properties of the interface element at small slip
cohesion (interface)
100 kPa
f ( )
0.2
Note that f ( ) = 0.2 was implemented by specifying a Mohr Coulomb material with a friction angle of
14.3 degree.
166
3.4 Comparison between PLAXIS and FROCK and observations

The comparisons between PLAXIS and FROCK are made in this section.
At u = 8E-6 m, the output normal and shear stresses at the joint are compared with those of FROCK in
Figures 3 and 4. As u increases, it is interesting to note that yielding first occurs near the middle of the joint,
which is consistent with the observation in the simulation using FROCK. At about u = 1.2E-5 m yielding
occurs at the middle portion of the joint and at u = 2.0 E-5 m the entire joint has yielded. At u = 2.0E-5 m, the
output normal and shear stresses at the joint are compared with those of FROCK in Figures 3 and 4.
It is noted that the normal stress distributions from the two programs are comparable while the shear
stresses show fairly good comparison. It is noted that near the left and right ends of the joint the computed
shear stresses are not reliable for the reason given in Section 3.1 above, and thus the stresses there may not be
shown in Figures 3 and 4. To show the effect of the joint slip on the behavior of the rock sample, the xdisplacement along the mid-vertical section of the sample is plotted in Figure 6. The displacement
distributions from the two programs compare fairly well. The slip distributions along the joint from the two
programs are compared in Figure 5. The slips on the left side compare satisfactorily while near the right side
there are lower slip values from both programs but higher relative differences. This phenomenon is considered
to be caused by the two different modeling techniques.
3.5 Observations on the use of the interface element in PLAXIS
It is noted that the normal way of using a PLAXIS interface element is to assume that its strength
parameters are proportionally reduced from those of its adjacent materials. For example, Page 29 of the
PLAXIS 2D 2012 Tutorial Manual suggets that the interaction between the wall and the soil be modelled at
bothsides by means of interfaces. The interface allows for the specification of a reduced wall friction
compared to the friction in the soil. The proportion of reduction is called Rint, the Interface Factor (PLAXIS
2012). The parameter Rint relates the strength of the soil to the strength of interface, according to the
equations:
tan int erface = Rint tan soil
(2)
cint erface = Rint c soil
(3)
In this paper, the interface element is specified directly with independent cohesion and friction coefficient.
This would reflect the physical reality in a better fashion.
The users of PLAXIS should also be aware of the limitation that post-yield strength is assumed to be
constant, which is an intrinsic characteristic of the Mohr Coulomb model.
To model the slippage development after yielding, a plasticity model with appropriate hardening law could
be used. But it appears that this type of interface element has not been recommended in PLAXIS 2D. Also the
robustness of the problem and calculation speed may be significantly affected if large plastic zones are
involved. It is more desirable to be able to use special elements to simulate an interface, rather than using
continuum elements. These special elements do not involve plastic zones or numerical difficulties and are
more suitable for large-scale simulation.
We do recognize that the development of a realistic interface element is never an easy task, but we would
encourage continuous development which imparts more diversity and flexibility to the program. Alternatively,
the authors of PLAXIS may consider representing the interface with Boundary Elements.
4 CONCLUSIONS, FURTHER COMMENTS AND PRACTICAL APPLICATIONS
Modelling of a rock joint with a non-linear slip-stress relation has been carried out using Boundary Element
and Finite Element formulations. The computed stresses, displacements and slips from the two programs used
are largely in agreement, indicating that the two approaches are sound.
Interesting observations on the sliding behaviour of the interface have been made: As shearing is applied,
slippage occurs first in the middle portion of the joint and spreads towards the right and left ends of the joint.
167
The horizontal displacement at the top of the sample can be considered to comprise the displacement due to
the slippage on the joint and the shear deformation of the intact rock. One has to consider this fact when
investigating the shear behavior of a rock joint by experiment. To measure the slip of the joint more
accurately, pairs of monitoring points should be located close to the joint surfaces, instead of just measuring
the relative displacement (u) between the top and bottom of the sample. Normally only one shear stress, one
normal stress and one slip value is measured at one time during the shearing test. However, the analyses in
this paper show that different portions of the joint could have different stress and slip values at one time. Thus
the measured stresses and slip actually represent aggregated values and not point-specific values, and the
resultant slip-stress curve actually depends on the experimental set-up and the way loading is applied. To
minimize this source of inaccuracy, one may consider measuring the stresses and slips at different locations
along the joint. The mesh used by FROCK is much simpler because only the boundaries of the sample are
modeled, while PLAXIS 2D models the entire domain. However, PLAXIS 2D is more versatile as different
material types are allowed while FROCK deals with a single linear elastic material only, even though the
discontinuities can possess non-linear slip-stress relations. Based on this study, it is recommended that the
PLAXIS interface element should be modelled by directly specifying the material properties, rather than by
specifying an Interface Factor. This recommendation can be implemented in many practical applications. We
have also proposed potential improvements of PLAXIS.
ACKNOWLEDGEMENTS
This paper is published with the permissions of the Head of the Geotechnical Engineering Office and the
Director of Civil Engineering and Development, The Hong Kong SAR Government.
REFERENCES
Brebbia, C.A., Telles, J.C.F. & Wrobel, L.C. 1983. Boundary Element Techniques. Springer-Verlag, NY.
Chan, H.C.M. 1986. Automatic two-dimensional multi-fracture propagation modelling of brittle solids with
particular application to rock. MIT Sc.D. thesis. (unpublished).
Chan, H.C.M. 1990. The Principle of Asymptotic Proportionality. Journal of Applied Mechanics (ASME
Transactions), Vol 57, March.
Chan, H.C.M. 1993. Fracture Mechanics Analysis of the North West Fault Block of the Prudhoe Bay Field.
International Journal of Rock Mechanics and Mining Sciences, 30(2).
Chan, H.C.M., Li, V. & Einstein, H.H. 1990. A Hybridized Displacement Discontinuity and Indirect
Boundary Element Method to Model Fracture Propagation. International Journal of Fracture, Vol. 44.
Crouch, S.L. & Starfield, A.M 1983. Boundary Element Method in Solid Mechanics. Allen & Unwin.
Hoek, E. 2015. Practical Rock Engineering. Available at http://www.rocscience.com.
PLAXIS 2012. PLAXIS Finite Element Code for Soil and Rock Analyses: Reference Manual 2D-Version 12.
Delft, The Netherlands.
Sternberg, E. 1954. On Saint-Venants principle. Quarterly Appl. Math., Vol 11, pp. 393-402.
Vsrhelyi B. & Bobet, A. 2000. Modeling of Crack Coalescence in Uniaxial Compression. Rock Mechanics
and Rock Engineering 33(2): 119-139.
168
Evaluation of the Consolidation Behavior of a Settling Ground at

Reclaimed Ground of a Container Port in South China
S.L. Chiu, T.T. Fong & S.H. Yung
AECOM Asia Co. Ltd., Hong Kong
ABSTRACT
In reclamation projects, it is often that consolidation of marine deposits would be occurring
throughout a long time such that the primary settlement is still ongoing during operation stage
of the facility on the reclaimed land. For port work facilities, uneven settlement of the
reclaimed ground would cause different settlement of port work structures placed on it, which
affects the performance of the port work facilities.
In this paper, site investigation and settlement analysis as well as projection of amount of
ongoing settlement in a portion of a container port in South China is presented. At this portion
of the Port, uneven settlement of the ground was observed and the performance of gantry crane
was affected. The settlement analysis was conducted based on existing settlement monitoring,
additional site investigation (including laboratory tests and field tests) and pore water pressure
monitoring. The current degree of consolidation was back calculated based on the available
information and further residual settlement was estimated. Based on the estimation results,
different options of remedial measures to cater for the residual settlement are suggested.
1 INTRODUCTION
1.1 Background
At a portion of a container port in South China which the construction was completed in 2010 and has
operated for a few years, uneven settlement was reported within two areas of approximately 55m x 25m and
54m x 21m. In order to maintain good serviceability and performance of the port work facilities, the Authors
were commissioned by the port owner (client) to evaluate and provide preliminary advice on the settlement
issue.
Based on the information provided by the client and site inspection, the differential settlement within the
settlement area was approximately 600mm.
1.2 Site history
Ground improvement was carried out in the site during the construction of the port. As indicated in the asbuilt drawings provided by the client, the area that uneven settlement occurred included a mud pit, where
ground improvement works - dynamic compaction was carried out after the installation of prefabricated band
drains (PVD) and surcharging. The relative location of the mud pit and the uneven settlement area are shown
in Figure 1.
It was reported that the clay/silt in the mud pit was very weak, and mud wave of approximately 1m in
height and 100m in width had occurred during the sand filling in the course of site formation. This
information also suggested that the disturbance of clay/silt and low stiffness and strength in the mud pit could
be a source of the problem causing uneven settlements between this area and the adjacent areas.
169
Uneven
settlement area
Road Hai 8
Mud pit
Road Hai 7
Road Gang 24
Figure 1: Location of mud pit and uneven settlement area in the port
2 SETTLEMENT MONTORING RESULTS

2.1 Monitored settlement
Ground level survey carried out in this study was compared with the levels shown on the as-built site plan.
The results indicated that level difference within the concerned area ranges from 420mm to 1000mm,
indicating an average differential settlement of 600mm observed. Figure 2 shows the settlement contours
generated based on the surveyed ground level differences aforementioned.
Boundary of the mud pit

A
BH1
BH2
~850 mm settlement
BH6
~1000 mm settlement
BH3
BH4
Road
Gang
24
BH5
Figure 2: Location of mud pit and uneven settlement area in the port
170
3 ADDITIONAL GROUND INVESTIGATION AND THE RESULTS INTERPRETATION

3.1 Additional ground investigation
In order to investigate the geological profile of the problematic areas, additional ground investigation
including 6 exploration boreholes (locations as shown in Figure 2) was carried out. Field tests and laboratory
tests on undisturbed samples collected were also conducted to determine soil parameters and estimate sub-soil
conditions of the study areas. Based on the additional ground investigation results, the geological profile of
the site is given in Table 1 below. The clay/silt layers were considered to be the layers attributable to the
settlement of the ground. A typical geological profile (Section A-A), based on the GI results, is shown in
Figure 3. The location of Section A-A is shown in Figure 2.
Table 1: Summary of geological layers of the area with uneven settlement
Soil layer
Depthm
Avg. thk. (m)
Rock Fill
0.0 - 8.5
7.1
Clay/siltupper
5.8 - 19.6
10.7
Sand
17.1 21.8
3.0
Clay/siltlower
20.0 26.0
4.4
Coarse sand / gravel
23.9 - 28.4
1.9
Highly decomposed granite (HDG)
Below 25.8
--
BH2
BH6
BH1
BH3
BH5 BH4
Rock Fill
Clay/silt (upper)
Sand pocket
Sand
Clay/silt (lower)
Coarse sand / gravel
HDG
Figure 3: Geological section A-A
3.2 Observations and interpretations from field tests and laboratory tests
Field tests including vane shear tests, Cone Penetration Tests (CPT), piezometers and laboratory tests
including unit weight measurements, triaxial tests (Consolidated Undrained, CU and Unconsolidated
Undrained, UU), Oedometer tests, Atterberg limits tests were carried out. The data points of plasticity index
and liquid limit of the clay/silt are shown in the plasticity chart in Figure 4. The results indicate that the soft
layer is generally silt (or clayey silt) of intermediate plasticity. The unit weight and consolidation parameters
of the soft layer as determined from the results of laboratory testing are also summarized in Table 2.
171
Figure 4: Plasticity index and liquid limit of the clay/silt layer
Soil layer
Table 2: Consolidation parameters of the soft layer

Initial void
Compression
Saturated unit
3
ratio, e0
index, Cc
weight, (kN/m )
Coefficient of vertical
consolidation, cv (m2/yr)
Mud ()
16
1.69
0.508
1.081
Clay ()
17.4
1.23
0.313
2.701
Silty clay ()
18.7
0.886
0.188
8.799
The undrained shear strength (Su) results of the clay/silt layer are also plotted against depth in Figure 5.
The data points of vane shear test results are also plotted against its estimated preconsolidation stress (p)
when consolidation is completed. It is taken as the effective vertical stress (vf) based on the estimated final
soil profile and loading when no excess pore pressure remains. The relationship of Su and vf in Figure 5 is
compared with the published data in the literatures for vane shear test data from different sites and its
relationship with effective vertical stress (v). e.g., Figure 58 of FHWA-IF-02-034 (2002).
The data of undrained shear strength (Su) (uncorrected by Bjerrums factor) of this study generally ranges
from 0.1 to 0.4 pa, and the estimated preconsolidation stress (p) when consolidation completed are generally
at 2 pa, which lies slightly above the average range of the data points in Figure 58 of FHWA-IF-02-034; and
the ratio of Su/vf of the site is lower than the fitted line in the Figure 58 of FHWA-IF-02-034. It suggests that
the effective vertical stress (v) of the test locations are lower than its expected final effective vertical stress
(vf) when all excess pore pressure is dissipated at the end of primary consolidation.
Figure 5: Undrained shear strength of the clay/silt layer and its relationship with final preconsolidation stress when
consolidation is completed, the base plot of the rightmost is extracted from Sabatini et.al. (2002), FHWA-IF-02-034
Evaluation of Soil and Rock Properties, where pa is atmospheric pressure (~100kPa)
172
The relationship between undrained shear strength (Su) and effective vertical stress (v) have been studied
by different researchers:
In Figure 58 of FHWA-IF-02-034 by Sabatini et.al. (2002), most data points are located at around Su/v =
0.25.
Mesri (1975) proposed the relationship of the ratio of Su/v = 0.22.
Skempton (1957) proposed the relationship of the Su/v ratio with plasticity index (Ip) which is Su/v =
0.11 + 0.37Ip. For the case of this study, the plasticity index of the clay/silt layer generally ranges from 12% to
18%, the Su/v ratio would be approximately 0.166 for the mid value 15% taken.
Bjerrum (1972) also proposed a relationship of Su/v ratio with plasticity index. The proposed curve and
the data that the curve is based on are extracted and copied in Figure 6. For this site with Ip generally ranges
from 12% to 18%, the Su/v ratio would be approximately 0.18.
Leroueli el. al. (1983) also proposed a relationship of Su/v ratio with plasticity index. The proposed curve
and the data that the curve based on are extracted and copied in Figure 7. For this site with Ip generally ranges
from 12% to 18%, the corresponding Su/v ratio would be around 0.20 to 0.24.
Figure 6: Relationship of Su/v and plasticity index Ip by

Bjerrum (1972) (figure extracted from Soil mechanics in
engineering practice by Terzaghi et. al., 1996)
Figure 7: Relationship of Su/v and plasticity index Ip by

Leroueli et. al. (1983) (figure extracted from In situ
testing in geomechanics by Schnaid, 2009)
From the above literatures, it is noted that the values of Su/v ratio are 0.25, 0.2, 0.18, 0.166 and 0.22,
which are around 0.2. In this study, a ratio of Su/v for 0.2 has been assumed in back calculation of v from
the measured undrained shear strength (Su). Figure 8 compares the Su of vane shear tests and UU tests with the
adopted empirical relationship of 0.2 vf that is based on final soil profile and loadings.
The following observations can be made from the plot of Su from vane shear test and UU triaxial test:
1. At borehole BH3, BH6 and bottom of BH1, BH2, the difference between Su and 0.2v is closer,
suggesting that its clay/silt layer might have achieved a higher degree of consolidation.
2. At top of BH1 and BH2, the difference between Su and 0.2v is still large, indicating that at the time
of present study its clay/silt layer was still at low degree of consolidation such that excess pore water
pressure was still to be dissipated.
Based on the assumed relationship of Su = 0.2v, the effective vertical stress v is back calculated.
Together with the fill thickness and final soil profile, the degree of consolidation (U) is estimated and
summarized in Table 3.
The results show that the degree of consolidation at BH4 is the lowest, and the degrees of consolidation at
top of BH1 and BH2 are also relatively low. It matches with the observation that the observed settlement near
these boreholes is larger (lower settlement rate at later part of consolidation). The degree of consolidation at
BH6 is relatively high, and the observed settlement during operation near BH6 is less (higher settlement rate
at early part of consolidation).
It is also observed that UU triaxial test tend to give lower Su values than vane shear tests. As triaxial UU
test samples are subjected to sample disturbance and the samples are unconsolidated before triaxial shearing,
while the vane shear tests are in-situ tests conducted on site, it is considered that the results from vane shear
tests would have higher reliability than triaxial UU tests.
173
Figure 8: Comparison of undrained shear strength (Su) with 0.2 vf
Table 3: Estimated degree of consolidation at different boreholes at different depths
Borehole
Depth
(m)
Su
(kPa)
BH1
9.1
11.3
12
13.5
15
16.5
10.5
12.8
15
8.4
11.8
15
17.9
8.8
11.3
14.4
17.4
9.2
12
14.8
9.5
12.4
15.4
11.5
17.5
40.6
26.3
36.2
38.5
7.9
36.8
69.8
10
14.1
26.6
38.8
6.9
9.7
11.6
14.5
9.6
13.3
15.1
14.8
30.1
29.3
BH2
BH3
BH4
BH5
BH6
Fill
thk.
(m)
5.8
5.8
5.8
5.8
5.8
5.8
6.3
6.3
6.3
6.8
6.8
6.8
6.8
7.1
7.1
7.1
7.1
8.1
8.1
8.1
8.5
8.5
8.5
vf'
(kPa)
v' by
Su(kPa)
v0'
(kPa)
v'
(kPa)
190.54
204.16
208.49
217.78
227.06
236.35
202.68
216.92
230.53
179.22
200.27
220.07
238.03
178.78
194.25
213.44
232.01
188.69
206.02
223.35
195.18
213.14
231.71
57.50
87.50
203.00
131.50
181.00
192.50
39.50
184.00
349.00
50.00
70.50
133.00
194.00
34.50
48.50
58.00
72.50
48.00
66.50
75.50
74.00
150.50
146.50
20.43
34.05
38.38
47.66
56.95
66.23
26.00
40.24
53.85
9.90
30.95
50.76
68.71
10.52
26.00
45.19
63.76
6.81
24.14
41.47
6.19
24.14
42.71
37.07
53.46
164.62
83.84
124.05
126.27
13.50
143.77
295.15
40.10
39.55
82.24
125.29
23.98
22.50
12.81
8.74
41.19
42.36
34.03
67.81
126.36
103.79
174
Degree of
consolidation,
U
21.8%
31.4%
96.8%
49.3%
72.9%
74.2%
7.6%
81.4%
100.0%
23.7%
23.4%
48.6%
74.0%
14.3%
13.4%
7.6%
5.2%
22.6%
23.3%
18.7%
35.9%
66.9%
54.9%
4 REVIEW OF DESIGN
4.1 As-built records of works carried out in the mud pit area
The as-built records of the area that the mud-pit area with uneven settlement occurred were reviewed.
Prefabricated vertical drains (PVD) were installed in the area and followed by application of a surcharge
preloading. After that, dynamic compaction was carried out with an impact-energy per blow (WH) of 3000 kJ.
Based on FHWA-SA-95-037 Dynamic compaction, the depth of improvement can be estimated by the
equation D = n (WH) 0.5 (WH in ton-m). The value of n can be referred to Table 7 of the document. For
previous soil deposits, the recommended n value is 0.5. By n = 0.5 and WH = 300 ton-m, the improvement
depth can be calculated to be 8.7m.
From the borehole logs, the fill thickness is 6m to 8m. Hence the improvement depth of dynamic
compaction should be sufficient to improve the fill layer. Also, as the improvement depth of dynamic
compaction only reached the top surface of the clay/silt layer, it is unlikely that the dynamic compaction has
caused significant disturbance to the clay/silt layer. Excess pore pressure during transient loading would not
persist to permanent condition.
4.2 Review on the lower clay/silt layer
As observed from the geological section shown in Figure 3, thickness of the lower clay/silt layer is generally
1/2 to 1/3 of the upper clay/silt layer. Also, the lower clay/silt layer is mainly composed of silty clay, while the
compression index (Cc) of silty clay (0.188) is lower than Cc of mud (0.313) which is the main component of
upper clay/silt layer. Therefore, for the same increase of effective vertical stress, the total primary
consolidation settlement of lower clay/silt layer would be approximately 1/5 to 1/8 of the upper clay/silt layer.
It may suggest that the settlement of area is mainly caused by the upper clay/silt layer.
5 SETTLEMENT EVALUATION
5.1 Evaluation of the current site situation
Referring to as-built records of the project, the time from project completion to time of the present study is 7.3
years. The measured settlement after the project completion ranged from 500 to 1000mm. The back-calculated
degree of consolidation of clay/silt layer indicated that the consolidation of clay/silt layer is higher at the
bottom of clay/silt layer than at the top, it might be probably due to:
1. Mud wave occurred within the area which caused disturbance to the top part of soft clay/silt layer,
resulting in damaging of band drains near the top of layer. The function of the drains was affected.
2. Mud wave may have caused a smear zone of low permeability of clay/silt in contact with band drains.
3. Damage to PVD and reduction of permeability lead to lower process of consolidation at the time of
surcharge removal. The soil had not been properly over-consolidated under the design load.
4. Excess pore pressure had not yet been fully dissipated and settlement occurred during operation.
5.2 Estimation of current degree of consolidation and residual settlement
As the average fill thickness is 7.1m, and water depth is 4.4m, with 60kPa design loading, the increase of final
effective vertical stress is 175.9 kPa. With average increase of effective vertical stress (v) of 78.8kPa
occurred as back calculated from Su/v = 0.2, the overall degree of consolidation, U = 78.8 / 175.9 = 44.8%.
As the overall settlement of the area is 500 to 1000mm, taking mid value 750mm, the final settlement is
1674mm. Hence the residual settlement is 924mm.
As the time of consolidation is 7.3 years, with an average clay/silt layer thickness of 11.1m, by equations
from Terzaghis 1-D consolidation theory (as it is observed that the consolidation rate is very slow and
performance of PVD within this area is questionable, the back analysis is conducted by vertical drainage), the
overall coefficient of consolidation is back calculated to be 0.67 m2/yr. The overall cv value based on actual
site performance is smaller than the laboratory test value, suggesting that disturbance to the clay/silt layer
caused reduction of permeability or increased the clay/silt layer thickness.
175
Also, based on compression index from laboratory tests and the above loadings, the calculated total
primary settlement is 1421mm, which is of similar magnitude to the estimation from overall site performance.
5.3 Pore water pressure measurements
Pore pressure measurements records from piezometers were also reviewed. The monitoring period was
approximately one month. By comparing the pore pressure measurements with the hydrostatic water pressure
based on the observed water level, the measured pore pressure is slightly above the hydrostatic line and
quickly dropped to below the hydrostatic line within the 1 month monitoring period (see Figure 9). The rate of
dissipation of pore pressure in the piezometer is much quicker than observed from surface settlement. It is
considered that the measured pore pressure could be generated during installation of instruments and quickly
dissipated over the monitoring period. Hence the measured pore pressure may have not reflected the actual
pore pressure in the soil stratum.
5.4 Limitations of the evaluation method
The method of evaluation in this study is mainly based on the empirical correlation of Su/v ratio. As the soil
properties of different sites are different, the Su/v ratio can be varied among different sites. If the site
specific Su/v ratio is different from that adopted based on other studies, the estimated degree of
consolidation would also vary. However, it is essential that the settlement evaluation based on the empirical
correlation had to be verified by the observed site settlement. The settlement evaluation in the present study is
generally in good agreement with the site observations and measurements of ground settlement.
Figure 9: Measured pore water pressure from piezometers installed during additional ground investigation
6 REPAIRING WORKS
6.1 Site constrains to repairing works
Different repairing works were considered for the project. Additional PVD plus surcharge can increase the
settlement rate, but require heavy works and long construction time, and penetrating the fill layer would be
difficult. Installing soil grout columns can increase the stiffness of the soil to reduce settlement, but would
also face construction difficulties for penetrating fill layer and require relatively high construction cost.
176
Resurfacing is considered to be a more preferred solution as there is no load bearing capacity problem, and the
works would require less time and cost.
6.2 Considerations of resurfacing of formation level
Given the site constraints aforementioned, two options were proposed for the resurfacing solution:
1. Resurface the settled ground to be flush with the adjacent areas.
2. Resurface the settled ground to a certain level higher than adjacent areas for future settlement.
As the differential settlement requirement for the gantry crane is 1 : 200, with the settlement area of width
60m, option 1 requires resurfacing for every 150mm settlement, and option 2 requires resurfacing for every
300mm settlement.
Also, additional load would lead to an additional loading of approximately 10 kPa, causing approximately
100mm additional settlement. Hence the total future settlement would be approximately 1000mm.
For option 1, it is estimated that the time of future resurfacing would be at 2.5, 6, 10, 16, 25, 40 years. For
option 2, the estimated time of future resurfacing would be at 6, 16, 40 years. The residual settlement-time
curve together with the resurfacing time intervals are shown in Figure 10.
Option 1
Option 2
(time t starts from first resurfacing repairing conducted shortly after the study )
Figure 10: Settlement time curve and resurfacing time intervals
7 CONCLUSIONS
This study has made use of the existing monitoring records, additional ground investigation to assess the
settlement issue of a site of containing port. The settlement evaluation in the present study is generally in
good agreement with the site observations and measurements of ground settlement. Based on the results of
present study, repairing works were proposed in order to maintain the site in use as it has been designed for as
well as the serviceability of the gantry crane structure for the port operations.
REFERENCES
Bjerrum, L. 1972. Embankments on soft ground. Proc. ASCE Speciality Conf. on Performance of Earth and
Earth-Supported Structures, Purdue University, USA, 2:154.
Bjerrum, L. 1973. Problems of soil mechanics and construction on soft clays. Proc of 8th Int. Conf. Soil Mech.
Found. Engng. Moscow, 3: 111-159.
Leroueil, S., Tavenas, F., Samson, L. & Morin, P. 1983. Preconsolidation pressure of Champlain clays Part
II: laboratory determination. Canadian Geotechnical Journal, 20(4): 803816.
Lukas, R.G. 1995. FHWA-SA-95-037 Dynamic compaction. U.S. Department of transportation.
Mesri, G. 1975. Discussion on new design procedure for stability of soft clays. ASCE Journal of the
Geotechnical Engineering Division, 101(4): 409-412.
177
Sabatini, P.J., Bachus, R.C., Mayne P.W., Schneider, J.A. & Zettler, T.E. 2002. FHWA-IF-02-034 Evaluation
of soil and rock properties. U.S. Department of transportation.
Schnaid, F. 2009. In situ testing in Geomechanics, the main tests: 275-276. Taylor & Francis Group.
Skempton, A.W. 1957. Discussion of Planning and design of New Hong Kong Airport. Proc. Instn Civ.
Engrs, 7: 305307.
Terzaghi, K., Peck, R.B. & Mesri, G. 1996. Soil mechanics in engineering practice. Wiley-Interscience
Publication.
178
With unparalleled underground expertise

and worldwide experience, aeCom is always
a trustworthy partner to deliver innovative
solutions to overcome complex geology
challenges faced by our clients.
nearly 100,000 employees serving clients

in more than 150 countries, we are a leader
in all the key markets we serve, including
transportation, facilities, environmental,
energy, water and government.
aeCom, recently joined by Urs, is a premier,

fully integrated support services firm. With
www.aecom.com
Worlds largest underground terminal aerial view of the

West Kowloon Terminus under the Hong Kong Section of the
Express Rail Link project
The HKIE Geotechnical Division 35th Annual Seminar, 2015

Soil-Structure Interaction: From Modelling to Observations
Designing more
innovatively
The HKIE
Geotechni
cal
Division
35th
Annual
Seminar,
2015
SoilStructu
re
Interact
ion:
From
Modelli
ng to
Observa
tions

As 2015 Opt

Uploaded by

Copyright:

Available Formats

As 2015 Opt

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

As 2015 Opt

Uploaded by

Copyright:

Available Formats

With unparalleled underground expertise

and worldwide experience, aeCom is always

nearly 100,000 employees serving clients

aeCom, recently joined by Urs, is a premier,

Worlds largest underground terminal aerial view of the

The HKIE Geotechnical Division 35th Annual Seminar, 2015

Lambeth brings together the diverse experience of operations, planning,

Proceedings of the 35th Annual Seminar

The HKIE Geotechnical Division Annual Seminar 2015

Any opinions, findings, conclusions or recommendations expressed in this material do

The HKIE Geotechnical Division Annual Seminar 2015

The HKIE Geotechnical Division Annual Seminar 2015

The HKIE Geotechnical Division Annual Seminar 2015

Soil-structure Interaction in Tall Building Foundation Design

The Role of Pile Structural Capacity in Pile Lateral Load Effect on

Lateral Load Tests on Four Large Diameter Piles in Hong Kong

Subgrade Reaction vs Elastic Solutions Theoretical Considerations

Piled Raft Foundation Design for a Supertall Tower

Does 3D Finite Element Analysis Allow Better Design of Piles?

Application of Modern Analytical Tools to a Four Decade Old

Geotechnical Challenges in Construction of an Underground

Recent Experiences of Numerical Prediction & Assessment

The HKIE Geotechnical Division Annual Seminar 2015

Outward Wall Movements Induced by Jet Grouting

Effects of Soil-structure Interactions on the Design of ELS Systems

Numerical Modelling of Flexible Barriers Subject to Landslide 115 - 122

Effects of Positioning of Debris-resisting Baffles on Impediment of

Study of Interaction between Landslide Debris and Debris-resisting 133 - 139

Performance Monitoring of a 50 m High 75 Cut Slope Reinforced 141 - 149

Considerations for Consolidated Undrained Triaxial Testing of 151 - 160

Interface Modelling - A Comparison Study Using Finite Element

Evaluation of the Consolidation Behavior of a Settling Ground at

The HKIE Geotechnical Division Annual Seminar 2015

Soil-structure Interaction in Tall Building Foundation Design

The HKIE Geotechnical Division Annual Seminar 2015

Raft or mat foundations;

The HKIE Geotechnical Division Annual Seminar 2015

The HKIE Geotechnical Division Annual Seminar 2015

Rs* = design structural strength = s. Rus,

where Rgs* = design geotechnical shaft capacity,

where max = maximum computed settlement of foundation,

The HKIE Geotechnical Division Annual Seminar 2015

all = allowable angular rotation.

Limiting Tolerable Settlement

Based on 52 cases of deep foundations.

Based on 52 cases of deep foundations.

Based on 57 cases of deep foundations.

Observed Intolerable Angular

Based on 57 cases of deep foundations.

6.4 Dynamic loading

The HKIE Geotechnical Division Annual Seminar 2015

6.5 Earthquake loading

The geological complexity of the site;

The HKIE Geotechnical Division Annual Seminar 2015

Geotechnical reduction factor, g, applied to

Unfactored geotechnical strength parameters

Unfactored geotechnical strength parameters

7 DESIGN ISSUES RELATED TO SOIL-STRUCTURE INTERACTION