Académique Documents
Professionnel Documents
Culture Documents
editors
editors
Meei-Ling Lin
(National Taiwan University, Taiwan),
Chung-Tien Chin
(Moh & Associates Inc., Taiwan),
Horn-Da Lin
(National Taiwan University of Science and Technology, Taiwan),
Yusuke Honjo
(Gifu University, Japan) &
Kok-Kwang Phoon
(National University of Singapore, Singapore)
For photocopying of material in this volume, please pay a copying fee through the Copyright
Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA. In this case permission to
photocopy is not required from the publisher.
Engineers and regulators in many countries are struggling to accommodate the complex
and multi-faceted changes occurring at the international scene. The status of local design
codes in view of globalization and their compatibility in view of evolving design
philosophies, are issues of major concern that do not admit simple solutions. A large
number of countries do not have the scale of economy, organizational structures, political
support, and perhaps financial resources to solve these complex problems on their own.
This conference intends to follow the spirit of IWS Kamakura (2002) and LSD2003 to
promote greater awareness, to facilitate debate and information exchange, and to
accelerate research and practice on important issues relating to new generation
geotechnical design codes. The bottom-line is to move geotechnical engineers forward
together as a community in response to significant changes occurring globally.
The idea behind this symposium grew out of a discussion between Chung-Tien Chin, Jie-
Ru Chen, Yusuke Honjo, and Kok-Kwang Phoon during the 16th ICSMGE in Osaka last
year. Subsequent discussion between Meei-Ling Lin and Kok-Kwang Phoon during the
GEDMR05 conference in Singapore helped set the path in commencing the organization
of this event. Given the gathering pace of geotechnical design code developments, there
is a compelling reason to consider a follow-up symposium to LSD2003. It is also timely
to discuss the possibilities of establishing a more regular series of symposiums and a joint
working group to coordinate these activities.
Thirty-five abstracts from thirteen countries were received during the initial call for paper.
Thirty-one papers were accepted for publication after review. Topics covered include
v
geohazards, geotechnical uncertainty and variability, probabilistic and reliability analysis,
design code concept and harmonization, and performance-based engineering practice. In
addition to the submitted papers, special invitations were extended for contribution as
keynote lectures, invited lectures, and Taiwan special project lectures. A total of 11
papers were obtained for these lectures. This publication contains extended summaries
of 42 papers. Complete contributions are available in the accompanying CD-ROM.
This symposium is jointly organized by the Taiwanese Geotechnical Society and TC23 of
ISSMGE. It is supported by the National Taiwan University, National Taiwan University
of Science and Technology, Taiwan Construction Research Institute, ASCE Taiwan
Chapter, JWG-DMR, ASCE Geo-Institute, TC39 of ISSMGE, and Southeast Asian
Geotechnical Society.
The publication of this proceedings will not be possible without the considerable efforts
invested by a committed editorial committee that include Jie-Ru Chen, Jian-Ye Ching,
Yo-Ming Hsieh, Chih-Ping Lin, and C.H. Wang. Papers appearing in this proceeding are
subjected to technical and editorial reviews. We are also grateful for the constant support
and timely assistance given by the technical reviewers (C. Hsien Juang, Kok-Kwang
Phoon, Robert S.R. Lo, Liming Zhang, Kenichi Horikoshi) and editorial reviewers
(Hsiang-Ju Chen, Ting-Rong Chen, Te-Wei Chen, Yu-Hua Hsieh, Wei-Nan Jian, Yueh-
Ting Lai, Jing-Hang Lin, and Mei-Ling Liu), and the secretariat (Ms Wei-Ling Lin and
Tsui-Hui Chiang). The significant assistance rendered by Prof. Der-Wen Chang in the
arrangement of travel visas for speakers is deeply appreciated. Lastly we would like to
thank Rhaimie Wahap and his team at World Scientific for working patiently with us
under a very tight publication schedule. His professional assistance is greatly appreciated.
Editors
Meei-Ling Lin
Chung-Tien Chin
Horn-Da Lin
Yusuke Honjo
Kok-Kwang Phoon
vi
Organising committee
Advisory committee
Editorial committee
vn
TABLE OF CONTENTS
Preface v
Organization vii
Keynote Lectures 1
Limit states design based codes for geotechnical aspects of foundations in Canada 3
D. E. Becker
T. L. L. On-
Invited Lectures 11
A preliminary study on load and resistance factors for foundation piles in Taiwan 19
H. D. Lin
Evaluating probability of seismic landslide based on the Chi Chi's events, Taiwan 21
M. L. Lin, C. J. Chung, M. H. Ho
The study and revision of the probabilistic seismic hazard map and dam safety
code of Taiwan 31
C. T. Cheng, S. J. Chiou, C. T. Li, P. S. Lin, Y. B. Tsai
Effect of lateral cyclic load on axial capacity of pile group in loose sand 43
S. Basak
The assessment and prediction of landslides and debris flows in Ta-Chia river
after Taiwan Chi-Chi earthquake 63
C. T. Cheng, Y. L. Chang, S. J. Chiu, Y. S. Lin, C. Y. Ku, S. M. Shu,
J. C. Chern, S. H. Yu, S. D. Yang, C. F. Wang, C. H. Chiao, L. T. Hwang
xi
Session VI: Geotechnical Uncertainties and Variabilities 85
Soil parameters used in the new design code of port facilities in Japan 97
L. P. Shi, Y. S. Hsieh
EXTENDED ABSTRACT
The geotechnical engineering profession in Canada, and elsewhere throughout the world, is in the
process of incorporating limit states design into codes of practice for geotechnical design aspects of
foundation engineering. Primary benefits of the use of limit states design are that it provides a
consistent design approach between structural and geotechnical engineers, as well as providing a
rational and consistent framework for design and risk management of design uncertainty. This paper
describes the needs and objectives for limit states design in Canada, and its development in codes;
identifies and describes the primary Canadian Codes; discusses the role of the Canadian Foundation
Engineering Manual and other authoritative references related to these Codes; discusses some of the
experiences and challenges encountered in practice during implementation and application of limit
states design; and outlines ongoing and proposed code development work, and associated future
directions and research needs. The importance of understanding fundamental principles, effective
communications between structural and geotechnical engineers, education and training is emphasized.
All of these components will be required for successful implementation and acceptance of limit states
design for geotechnical aspects of foundation engineering.
Limit states design, based on a factored strength approach similar to that of the European practice,
for geotechnical aspects of foundations was first introduced into Canadian engineering practice in the
early 1980s. However, this initial introduction did not get off to a good start because factored strength
concepts were not well accepted by geotechnical engineers; it also generated a fair amount of
confusion and controversy because the promised economy of design was not achieved. Canadian
geotechnical practitioners felt that it was not logical or rational for strength parameters to be reduced
(factored) to reflect weaker "artificial" soils and then use them directly in the same equations for
calculating design resistances. In the early 1990s, an overall factored resistance approach, based on a
Load and Resistance Factor Design (LRFD) format, was proposed for limit states design based codes.
Subsequently, a LRFD format for foundations became a mandatory requirement in the 2000 edition of
the Canadian Highway Bridge Design Code (CHBDC) and in the 2005 edition of the National
Building Code of Canada (NBCC). Nevertheless, confusion continues to exist concerning the
objectives of limit states design as engineering practitioners in Canada struggle to undergo the
transition from traditional working (allowable) stress design to design based on limit states (LRFD)
concepts.
The primary structural codes in Canada are the NBCC, the CHBDC and the Canadian Offshore
Structures Code. These codes involve the interaction of structural and geotechnical engineers; they
generally apply to the design and construction of foundations, retaining walls and other buried
structures. There is no national code document for aspects in which geotechnical engineers do not
normally interact with structural engineers. The current geotechnical state-of-practice in Canada does
not use limit states design concepts to design slopes, earth embankments, dams and other earth
structures. The code requirements are normally written as performance requirements and are based on
scientific or technical principles. The codes avoid standardizing certain methods or procedures of
design and construction. For example, the NBCC (2005) is published in an Objective-Based Code
format where each code requirement is linked to the four basic objectives of Safety, Health,
Accessibility (in particular for persons with disabilities), and Fire and Structural Protection of
Buildings. Although some countries are striving to establish Performance-Based Codes, the NBCC
code developers are of the opinion that current building science knowledge is inadequate to write a
"true" (as per their perspective) Performance-Based Code, and that the measures to verify performance
4
TAIPE12006 International Symposium on New Generation Design Codes
for Geotechnical Engineering Practice
Nov. 2-3, 2006, Taipei, Taiwan
SUMMARY
Risk assessment in rock engineering is based on the formal identification of uncertainties and
on their assessment and possible modification in the context of risk analysis and
management. The best way to include uncertainty in the engineering design process is
through the use of the basic structure of decision making under uncertainty which progresses
from information collection, to deterministic and probabilistic modeling to end up with risk
assessment and related decisions. These decisions, i.e. risk management range from
accepting the risk as is to modifying it.
Before applying this decision making process to rock engineering, it is necessary to be clear
as to what criteria engineering structures have to fulfill: safety, susceptibility, economics and
aesthetics and, particularly, to identify the relevant sources of uncertainty. In rock
engineering, the most important sources are inherent spatial variability,
measurement/estimation errors and model uncertainties.
In information collection, one needs to determine the relevant parameters and associated
uncertainties (distributions) through appropriate sampling procedures. Specifically, potential
biases have to be avoided and corrected for. Also, one needs to relate the sample to the
sample population and, most importantly, to the target population, the latter usually requiring
judgement. The result of information collection are state-of-nature models, which express
the natural variability. Stochastic fracture pattern models are examples.
In the deterministic modeling, phase one relates parameters to outcomes, i.e. predicted
performance. The performance can be related to stability, deformation, flow or economic
aspects (or combinations). In rock engineering, such performance is related to the typical
problems of slope stability, foundation performance, flow and tunneling. An important
aspect of the deterministic phase is the concluding sensitivity analysis, which is used to
identify the parameters having the greatest effect on the results. Usually only these
parameters will be varied in the probabilistic phase.
Probabilistic modeling is entirely analogous to the deterministic one but now the relevant
parameters and their uncertainties (distributions) are propagated through the model. Hence,
the state of nature models mentioned earlier provide the required input. An important issue
specifically related to rock engineering is the treatment of fracture persistence, i.e. the fact
that fractures and intact rock are interspersed; which has a significant effect on rock mass
performance. The probabilistic approach allows one to rationally solve "the persistence
problem". Probabilistic models are also well suited to deal with uncertainties affecting
economics such as the cost and time to build a tunnel.
Determining and using uncertainties in predictions have a long tradition in rock engineering.
Hence, quite a few procedures and models are available. It is, however, most important to
put all this in the context of the decision making structure as was done here.
6
TAIPEI2006 International Symposium on New Generation Design Codes
for Geotechnical Engineering Practice
Nov. 2-3, 2006, Taipei, Taiwan
SUMMARY
Introduction
There have been movements in Japan to develop a serious of comprehensive structural design codes
which can harmonize all the major Japanese structural design codes. This movement is much
motivated by the rapid development and popularisation of international and regional structural design
codes such as IS02394 and Structural Eurocodes, as well as of the performance based design concept
especially after the conclusion of WTO/TBT agreement in 1995. In proposing such efforts, it is much
contemplated to propose a concept that can harmonize all the major Japanese structural design codes
that have been developed rather separated way due to many historical reasons. The performance based
design (PBD) (or the performance based specification (PBS)) and the limit state design (LSD) are the
two concepts we introduced to achieve this aim. One of the final aims of this activity is to propose a
new framework of structural design codes for harmonizing structural codes in regional and
international levels. Two of such efforts, namely development of 'Principles for Foundation Designs
Grounded on a Performance-based Design Concept' (nick name 'Geo-code 21') by JGS (Japanese
Geotechnical Society) and 'code PLATFORM ver.l' by JSCE (Japan Society of Civil Engineers) are
presented in this paper.
The relationships among WTO/TBT, PBD/PBS and LSD in the current design framework are
illustrated in Figure. 1. It is our belief that the specifications of performance of the structures
would be described based on the concept of PBD/PBS, whereas the verification of design would
be based on LSD/RBD for all the major design codes in the world.
In order to cope with the situations explained in the previous section, movements to establish a
series of comprehensive design codes have been started in Japan. One of the initial works of this kind
of movements started in 1997 at JGS (Japanese
Geotechnical Society) as drafting of 'Geocode WTO/TBT
21'*, a proto type comprehensive foundation PBD
(Performance Ni
design code that can harmonize all the major specifications Based Design) >
foundation design codes in Japan. The by
performance
comprehensive design codes stand at the top
hierarchy level in all the structural design codes
RBD/
in Japan to give concepts, framework and Respect
LSD/ ; other
International
terminologies for structural design codes as Standards LRFD ' design
(IS02394etc.) methods
indicated in Figure. 2. It is not intended to be
legally enforced but as agreements among the
professions (more specifically, the code writers)
to draft structural codes based on the rules,
terminologies and concepts established by the Figure 1 WTO/TBT agreement, PBD and LSD/RBD
comprehensive codes. Therefore, it is thought
that it is most appropriate for professional societies such as JSCE (Japan Society of Civil Engineers)
and JGS (Japanese Geotechnical Society) to publish such codes.
* 'Geocode 21' is a nick name of this design code. This name has been used from the beginning of the
project. The final official name of the code is Principles for Foundation Designs Grounded on a
Performance-based Design Concept and the official number of the code is JGS-4001-2004.
• Unification of terminologies.
• Methods and formats to introduce the safety margin to various limit states in design.
• Standardize characteristic value determination in geotechnical design.
• Standardize information flow (i.e. documents preparation) among owner, designer,
constructor, geotechnical investigator and others.
• The limit state design (LSD) concept is introduced for design verification.
For all the major design codes in Japan, it is principal that the design changes from the next
day a revised code is enforced for the category of structures under the control of that code because
of the legal background. It is too strong constraint for a code to introduce new concepts. For this
reason, it is our experience that all the new concepts introduced to the codes are creepingly
deformed, stripped of its essential contents in the process of drafting, and finally enforced with no
substances.
It is not expected that Geo-code 21 is to be used in the actual design from the day it is issued;
it is rather pursuing an ideal code which all the code should finally merge into it in the near future.
It is expected that various foundation design codes in Japan to accept the concepts and the formats
etc. proposed in this code, and finally mildly harmonized to this code in a certain time interval.
Final Remarks
Some of the activities on harmonizing Japanese major civil engineering structural design codes
are introduced in this paper. The authors are hoping this kind of activities are extended to Asian
region so that we can cooperate together to develop our own regional codes system to promote
construction industries within this region by unifying the market, and strengthen the
competitiveness of our construction industry to the outside.
TAIPEI2006 International Symposium on New Generation Design Codes
for Geotechnical Engineering Practice
Nov. 2-3, 2006, Taipei, Taiwan
EXTENDED SUMMARY
Eurocode 7 for geotechnical design is one of the set of Eurocodes for structural design using
different materials that are about to be implemented in Europe. The Eurocodes are all based on the
same limit state design method, set out in Eurocode EN 1990, with partial factors applied to
characteristic parameter values. In this paper, the development of Eurocode 7 from the initial work in
1981 to prepare a model limit state code for geotechnical design, through the preparation of the pre-
standard, ENV version of Eurocode 7, to the publication in 2004 of the of the full European standard,
EN 1997-1, Eurocode 7 Geotechnical Design - Part 1: General rules, is outlined. The issues that
arose in developing Eurocode 7 as a code that was consistent with EN 1990, took account of the
special features of soil and geotechnical design, and was acceptable to the European geotechnical
community were:
• The scope of Eurocode 7
• The definition of the characteristic value of a geotechnical parameter
• The value of the partial factor on permanent loads
• The application of partial factors to material parameters or resistances
• The treatment of water pressures and forces
• The accommodation of national design practices.
The nature of these issues and how they were overcome is discussed in this paper.
Regarding the scope of Eurocode 7, it was accepted by CEN TC 250, the management committee
for the Eurocodes, that the requirements for ground investigations and determining parameters from
field and laboratory tests are part of the design process and should be included within the scope of
Eurocode 7.
The definition of the characteristic value of manufactured structural materials as the 5% fractile of
an unlimited series of test results is shown to be not appropriate for geotechnical design. The
principal reason for this is because the geotechnical parameter controlling, for example, a failure in
the ground, is the mean strength over the failure surface, not the strength of an individual test
element. Hence it is the 5% fractile of the mean strength along the failure surface that is required, not
the 5% fractile of the test results. Another reason is because, in geotechnical design, only a limited
number of test results are normally available and hence statistics need to be used with caution.
Eurocode 7 states that the characteristic value "shall be selected as a cautious estimate of the value
affecting the occurrence of the limit state". This definition is an important innovation in Eurocode 7
and some guidance on the selection of the characteristic value is provided in the paper.
Since the Eurocodes are for structural design, the partial factor chosen for permanent loads in EN
1990 was 1.35. This value caused a problem for Eurocode 7 because in geotechnical designs, for
example in slope stability analyses, the permanent actions due to the weight of soil are not normally
factored. If they are factored, then illogical situations can arise; for example, in the case of a circular
vertical failure surface below horizontal ground, if the unfavourable soil weight, treated as a
permanent load, is factored by 1.35 while the favourable soil weight is not factored, then analysis of
this situation can predict failure of the horizontal ground when it is not loaded. This is not logical and
therefore the Eurocode 7 drafting panel successfully resolved this issue by getting TC 250 to accept a
partial factor of unity for permanent actions in geotechnical design when factors greater than unity
are applied to the soil strength.
In the ENV version of Eurocode 7, partial material factors are applied to the soil strength
parameters c', tan<j>' and c„, in the same way as partial material factors are applied to the strength
parameters in the other Eurocodes. The partial factors for geotechnical ultimate limit states adopted
in the ENV version are the sets of partial factors referred to as Cases A, B and C. However, many
geotechnical engineers in Europe were not happy with these three Cases, with partial factors applied
to the soil strength parameters. The inclusion of partial factors applied to soil resistances was sought,
0.5 -
0 -I 1 , 1
20 25 30 35 40
Friction angle <])' (degrees)
Figure 3: Design widths for the square pad foundation
National Annexes with values for the Nationally Determined parameters (NDPs), e.g. partial
factor values, are due to be completed by November 2006 so that each member state can implement
Eurocode 7 as a national code and usher in the Eurocode era. The advantages of Eurocode 7 are that,
in providing the limit state principles for geotechnical design and not being prescriptive, it
harmonises geotechnical design with structural design, it takes account of the special features of soil
and geotechnical design, and it allows national design practices to be accommodated. Hence
Eurocode 7 is applicable, not only in Europe, but also worldwide.
10
Invited Lectures
TAIPEI2006 International Symposium on New Generation Design Codes
for Geotechnical Engineering Practice
Nov. 2-3, 2006, Taipei, Taiwan
Y. Honjo
Gifu University, Gifu, Japan
A. Iizuka
Kobe University, Hyogo, Japan
SUMMARY
More than 10 years have passed since performance-based design was introduced into the
construction industry along with a set of expectations in Japan. Nowadays, Japanese engineers are
very familiar with the term 'performance-based design'. Designers and contractors expected more
flexible application of new design and construction methods for public works. The public sector
expected more cost-effective construction by setting the required performance of individual
structures based on importance. Despite these expectations, however, the application of
performance-based design in public works is still limited.
At present, some of the large Japanese projects are based on design-build contracts. Design-
builders must take more risks to achieve the required performance. Risk sharing related to
geotechnical conditions is always a difficult issue between the client and the design-builder.
Insurance systems that cover such risks are not fully developed due to difficulty in determining
the insurance premium. It seems that there is still a number of unsolved issues to make much use
of the advantages of performance-based designs in public works.
A significant portion of the present paper is based on the activities of the JSCE (Japan Society
of Civil Engineers) Committee on 'Performance-based design of soil structures' chaired by Prof.
Atsushi Iizuka of Kobe University. The committee was organized in 2004 to discuss and clarify
the applicability of performance-based design to soil structures such as embankments, reclaimed
islands, and dams. Investigation of European systems to support performance-based design was
conducted as one of the activities.
This paper describes
1) The current situation of the Japanese construction industry;
2) New contracting and bidding systems employed in Japan that are oriented more toward
performance and quality without compromising cost; and
3) Results of investigation on performance-based design in European countries.
Following trends and findings are included through the study described in this paper.
1) The Japanese domestic construction market has become increasingly competitive. Severe price
wars have led to client concerns over maintaining the quality of public works, which has led to
the application of new contracting and awarding systems.
2) Some of the top contractors have expanded their markets to overseas. The increase in overseas
projects has resulted in designers, consultants, and contractors shifting toward performance-
based design in order to avoid or reduce potential risks in the target projects.
3) New contracting systems such as design-build system, value engineering system, technical
proposal integrated evaluation system, performance requirement ordering system, and PFI
(Private Finance Initiative) system has been applied in Japan. These systems enhance the
implementation of better performance in public works through the private sector's proposals.
Performance oriented
Code development
Bidding/Awarding systems Verification Systems of
(Technical Proposal Integrated Design/Construction
Evaluation System, etc)
method
v
Insurance Systems
M
Contracting Systems
(PI insurance etc.) (Design-Build, CM, PFI etc.)
14
TAIPEI2006 International Symposium on New Generation Design Codes
for Geotechnical Engineering Practice
Nov. 2-3, 2006, Taipei, Taiwan
Hongwei Huang
Department of Geotechnical Engineering, Tongji University, Shanghai, P.R.China
Qunfang Hu
Department of Geotechnical Engineering, Tongji University, Shanghai, P.R.China
Yuanyuan Yang
COWI Consulting Ltd. Co., Shanghai, P.R.China
SUMMARY
In general, the lining structure of shield tunnel is composed of segments and joints. According to
the difference of joints among segments, the assembly of lining structure is defined as segment
seams in sequence and segment seams in stagger. The service life of shield tunnel is usually
designed as 100 years, the lining structure is below the ground level with carrying the load of soil,
water and vehicle in its life cycle. So the risk condition of the lining structure will have great
effects on the safety during tunnel construction and operation, and risk analysis is much
significant.
In this paper, risk accidents during lining structure construction are analyzed, and various
potential risk factors are identified, and risk analysis model is established based on the faulty tree
theory. Considering the functional requirement and mechanical behaviour of lining structure, the
possible damage types mainly include:
(1) ULS: circumferential or longitudinal damage due to insufficient strength of bolts or
segments, circumferential or longitudinal joint expansion, segment crack, etc.
(2) SLS: Lining structure damages due to too large diameter deformation, longitudinal
differential settlement, durability failure, operation factors, etc.
Combining the two damage types as above and considering the uncertainty of geotechnical
engineering, the main risk factors that affect the shield tunnel lining structure are as follows in the
view of engineering risk identification:
(1) Internal factors mainly are from the structure itself, including the design model of lining
structure, material, prefabrication and curing method, assembly tolerance of segments, bolt
strength, joint tolerance, circumferential and longitudinal variation of segments and joints, which
will cause risk accidents directly.
(2) External factors mainly are the loads and environmental action on the lining structure. Due
to the constraints during geological investigation, the investigation results are usually not the same
with actual condition. Furthermore, large variation appears in the random distribution of soil, the
physical and mechanical parameter, and underground water. Therefore, the uncertain factors such
as site construction loads and environmental medium (e.g.Cr'.SOj ,Mg2 ,HC0 3 ,C0 2 ) and motors
may also cause risk accidents of lining structure.
Based on the analysis of the risk factors and accidents for shield tunnel lining structure, the
fault tree theory could be used to establish the structural risk analysis model to quantify the risk
evaluation.
Considering the variation of stratum and lining structure, the risk occurrence probability for the
events in the model is calculated by the means of engineering reliability theory and finite element
method analysis. And then the above method is used to analyze the lining structure risk in
16
TAIPEI2006 International Symposium on New Generation Design Codes
for Geotechnical Engineering Practice
Nov. 2-3, 2006, Taipei, Taiwan
SUMMARY
Seismically induced slope failures have normally been evaluated based on equilibrium of forces acting on
a potentially sliding soil mass. This force approach can evaluate the safety factor against the slope
failure or displacement not exceeding about 1 m normally by Newmark method but cannot predict slide
deformations, once large failure occurs. There exists no simple method available for performance-based
design in which different levels of slope performance including ultimate flow-type failure can be
evaluated at this moment.
In this research, an energy approach is proposed for performance-based design for predicting
residual displacement including long run-out distance in seismically induced slope failure. Shake table
tests of dry sand slopes are carried out to examine associated energy balance by comparing test results
with a Newmark-type rigid block model. The energy approach is then applied to a hypothetical slope
under extremely large earthquake energy to show its potential for evaluating different levels of slope
performance including ultimate collapse. Major findings obtained in this research are:
1) In shake table tests of dry sand slopes with different slope inclinations and different input frequencies,
earthquake energy used for slope failure could be successfully measured, quantifying the energy
balance involved in the failure of the model slope.
2) The model tests yielded a unique relationship for each slope inclination between the energy EEQ and
residual slope displacement Sr, which is independent of input frequency, indicating that energy can
serve as a better parameter for evaluating residual slope displacement than acceleration.
3) The above mentioned EEg versus 5r relationship shows clear thresholds of EEQ below that 5r =0,
which is again independent of input frequency. This implies that not only residual displacement but
also initiation of slope failure is determined not by acceleration but by energy.
4) Comparison of the test results with energy balance for a Newmark-type rigid block indicated that the
rigid block model can almost perfectly emulate continuously deforming sand slope provided that an
appropriate friction coefficient in the rigid block model can be estimated.
5) In order to back-calculate the friction coefficient, the energy-based method was applied to a recent
case history during the 2004 Niigata Chuetsu earthquake. The equivalent friction angle 12.6° was
back-calculated, much lower than the inclination of the slip plane 20°, indicating that the failed soil
mass accelerated and piled up on the opposite side of the valley.
6) A case study on a simple slope demonstrates that even under an extremely large earthquake, a
flow-type slope failure of a long run-out distance cannot occur unless equivalent friction angle
approaches to slope angle. Also indicated is that even the strongest earthquake energy is actually
far smaller than potential energy in long run-out distance failures.
7) Thus, the case study indicates that the energy-based method can serve as a powerful tool to predict
different levels of slope performance corresponding to serviceability, reparability and ultimate
collapse with long run-out distance.
SUMMARY
This paper describes the current pile design code of Taiwan, presents a rational approach for the
determination of the resistance and load factors for foundation piles subjected to axial load, and then
compares the results with some LRFD design codes. Results show that the resistance and load factors
may be significantly affected by the target reliability, the ratio of live load to dead load, the pile
failure criterion, and the design method. Therefore, these influencing factors should be carefully
considered when developing the load and resistance factors for an LRFD code. This paper also
presents a preliminary set of resistance and load factors for bored piles and driven piles in Taiwan.
2 -
Reliability Index=3.0 - Davisson
1.8 - + + Resistance Factor Fuller&Hoy
— • — • Dead Load Factor Terzaghi
# # Live Load Factor - Chin
1.6 -
1.4 -
I-
! -
0.8 -
0.6 - tsgnk?^,-.--^:-
0.4 -
0.2 -
i ' i ' i ' i ' r T
0 0.5 1 1.5 2 2.5 3 3.5
Live Load / Dead Load
Figure 1. Load and resistance factors of driven pile (design method I).
Conclusions
The load and resistance factor design method (the LRFD method) that is based on reliability analysis
offers a very rational framework to account for all aspects of the uncertainty related to pile
performance. The reliability approach also offers more flexibility for future adjustment of the safety
factors. Studies conducted by the author are in strong support of the above statements. As a result, a
set of partial factors for bored piles and driven piles in Taiwan is suggested. The suggested partial
factors seem reasonable and are consistent with available LRFD codes. However, they should be
used with caution because further study using a broader data base is warranted.
20
TAIPE12006 International Symposium on New Generation Design Codes
for Geotechnical Engineering Practice
Nov. 2-3, 2006, Taipei, Taiwan
C. J. Chung
Geological Survey of Canada, Ottawa, Canada
M. H. Ho
National Taiwan University, Taipei, Taiwan
SUMMARY
Locating in the circum-pacific seismic zone with frequent earthquake activities, Taiwan is
susceptible to hazard induced by earthquakes. In September 21 st , 1999, the Chi-Chi earthquake
with a magnitude of 7.3 struck central Taiwan, and caused severe damages to human lives and
properties. Among all the damages, more than 20000 cases of landslides were identified with a
total area of more than 8600 hectares by Council of Agriculture (2000). In order to mitigate the
hazard and plan for the land use, it is essential to evaluate probability and develop prediction map
for potential landslide induced by the earthquake. The evaluation of the probability of future
landslide can be performed by applying the conditional probability concept to the documented
events from the Chi-Chi earthquake in this case.
Conclusions
Based on the previous discussions, a prediction model is constructed using likelihood ratio method
following the theory of conditional probability. From the results of the analysis, the geological
formation, slope angle, and slope aspect factors appeared to be effective causal factors and the
prediction model established accordingly yielded satisfactory results, and the uncertainty of the
prediction results can be estimated.
22
TAIPEI2006 International Symposium on New Generation Design Codes
for Geotechnical Engineering Practice
Nov. 2-3, 2006, Taipei, Taiwan
SUMMARY
In foundation design, the serviceability limit state (SLS) often is the governing criterion,
particularly for the cases of large-diameter piles and shallow foundations. In addition, it is
widely accepted that foundation movements are difficult to predict accurately. This paper
discusses the application of a probabilistic hyperbolic model for performing reliability-based
design checks at the serviceability limit state. The important nonlinear feature of the load-
displacement curve is captured by a two-parameter hyperbolic curve-fitting equation:
Q = y
Qm a + by
in which Q = applied load, Q m = interpreted capacity, "a" and " b " = curve-fitting parameters, and
y = pile butt displacement. The uncertainty in the entire load-displacement curve is represented
by a relatively simple bivariate random vector containing the hyperbolic parameters as its
components. It is important to highlight that these hyperbolic parameters typically exhibit a
strong negative correlation. A translation-based Hermite expansion can be used to capture this
correlation aspect correctly in a fairly general way:
a = a 1 0 H 0 ( X 1 ) + a 1 1 H 1 (X 1 ) + a 1 2 H 2 ( X 1 ) + a 1 3H 3 (X 1 ) + a 1 4 H 4 ( X 1 ) + a 1 5 H 5 ( X 1 ) + ---
b = a20H0(X2) + a21H1(X2) + a22H2(X2) + a23H3(X2) + a24H4(X2) + a 2 5 H 5 ( X 2 ) + - -
in which Xi and X 2 are standard normal variables and the Hermite polynomials Hj(.) are given by:
H„(X) = 1
H,(X) = X
H k + 1 (X) = X H k ( X ) - k H k _ , ( X )
The correlation between " a " and " b " (p s , b ) is related to the correlation between Xi and X 2 (pxix2):
^ k ! a l k a 2 k p X ix2
: k=l
Pa,b
f>!a?Jjrk!aL
0 10 20 30 40 50 60 70 80 0 10 20 30 40 SO 60 70 80
Displacement (mm) Displacement (mm)
0 10 20 30 40 50 60 70 80 0 10 20 30 40 50 60 70 80
Displacement (mm) Displacement (mm)
The common assumption of statistical independence can circumvent the additional complexity
associated with a translation model, because the bivariate probability distribution reduces to two
significantly simpler uni-variate distributions in this special case. However, the scatter in the
measured load-displacement curves cannot be properly reproduced by simulation under this
assumption as shown in Figure Id. On the other hand, Figure lb looks more realistic because the
proper negative correlation is included in the bivariate probability model for "a" and "b".
The SLS is defined as that in which the vertical or lateral displacement (y) is equal to the
allowable limit (ya) imposed by the structure. The foundation is considered unserviceable if the
displacement is greater than the allowable limit. Conversely, the foundation is considered
satisfactory if the displacement is less than the allowable limit. These three situations can be
described concisely by a performance function:
P = y - y a = y(Q)-y«
P = Q . - Q = Q.(y.)-Q
Figure 2 illustrates the uncertainties associated with these performance functions. In Figure 2a,
the applied load Q is assumed to be deterministic to simplify the visualization. It is clear that the
displacement follows a distribution even if Q is deterministic because the load-displacement
24
(a) /! (b)
1.6-1 Random displacement caused by uncertainty in load- 1.6 - /[ Random allowable
displacement curve M\ . load caused by — —
/ \ uncertainty in load-
1.2 1.2 - f Jf*** displacement curve
0.8 -
1 ^f^*^ Random applied load
0.4 -
1 ^
J > Deterministic allowable
1 displacement
00 -
20 40 60 20 40 60 80
y (mm) y (mm)
curve y(Q) is uncertain. The allowable displacement may follow a distribution as well. In Figure
2b, the allowable displacement is assumed to be deterministic. In this alternate version of the
performance function, the allowable load Qa follows a distribution even if y„ is deterministic
because of the uncertainty in the load-displacement curve. The effect of a random load Q and the
possibility of upper tail values falling on the nonlinear portion of the load-displacement curve are
illustrated in this figure.
The probability of failure (pf) at the serviceability limit state can be computed easily using the
first-order reliability method (FORM) once the probabilistic hyperbolic model is established:
mQQ Ya , 1 Q '
p f =Prob = Prob
a + bya "QmVm
a + bya F S Q ;
The interpreted capacity Qm, allowable displacement ya, and applied load Q are assumed to be
lognormally distributed with a coefficient of variation (COV) = 0.5, 0.6, and 0.2, respectively.
The hyperbolic parameters are negatively correlated with an equivalent normal correlation of-0.8
and are lognormally distributed with mean and COV of "a" = 5.15 mm and 0.6 and mean and
COV of "b" = 0.62 and 0.26. The mean of y, is assumed to be 25 mm. The other parameters are
defined as: mQm, mQ = mean of Qm and Q, FS = mQm/mQ, and Qm*, Q* = unit-mean lognormal
random variable with COV = 0.5, 0.2. Note that FS refers to the ultimate limit state (ULS) global
factor of safety commonly adopted in prevailing practice.
For a typical FS = 3, the SLS reliability index for this problem is 2.21, corresponding to a
probability of failure of 0.0134. The probabilistic hyperbolic model can be simplified to a linear
model by setting b = 0 (physically, this is equivalent to an infinitely large asymptotic limit). For
the above example, the FORM solution under this linear assumption is p = 2.76 (pf = 0.00288).
Note that the solution is significantly unconservative — the probability of failure is about 4.7
times smaller than that computed for the correct nonlinear hyperbolic model. Hence, it is crucial
to model the nonlinearity in the load-displacement curve in reliability analysis. The reason is that
the applied load follows a probability distribution and it is possible for the load-displacement
behavior to be nonlinear at the upper tail of the distribution.
The allowable load can be evaluated from the interpreted capacity using a SLS model factor
(Ms) as follows:
25
Qa=-Ir-Qm=MsQm
a + bya
If M s follows a lognormal distribution, then a closed-form solution exists for the reliability index
(P):
m
l n fmMs Qm"|| 1 + COV^ ~
m
[I Q J l J ^ C O V ^ J l + COV^j
V^ll1 + C O V M S J(l + COV^, j(l + COV* )J
This procedure is similar to the bias/model factor approach used in ULS reliability-based design.
The catch is that the distribution of M s has to be evaluated for each allowable displacement. In
contrast, the foundation capacity is a single number (not a curve) and only one model factor
distribution is needed. Another practical disadvantage of the SLS model factor approach is that
M s cannot be easily evaluated from a load test database if y, follows a distribution.
Reliability analysis for the more general case of a random allowable displacement can be
easily carried out using an implementation of the first-order reliability method in EXCEL. It can
be seen in Figure 3 that a target reliability index = 2.6 is not achievable at a mean factor of safety
= 3 for a mean allowable displacement of 25 mm. However, it is achievable for mean factors of
safety larger than about 4. Note that the 50-year return period load Q50 « 1.5 mQ for a lognormal
distribution with COV = 0.2. Hence, a mean factor of safety of 4 is equivalent to a nominal
factor of safety = mQm/Q50 = 2.7. A nominal factor of safety of 3 is probably closer to prevailing
practice than a mean factor of safety of 3. It is also debatable if the allowable displacement
prescribed in practice is a mean value or some lower bound value. For a lognormal distribution
with a COV = 0.6, a mean allowable displacement of 50 mm produces a mean less one standard
deviation value of 25 mm. Using this interpretation, a target reliability index of 2.6 is achievable
at a mean factor of safety of about 3 or a nominal factor of safety of about 2. Overall, the EPRI
target reliability index for SLS is consistent with foundation designs produced from our
traditional global factor of safety of between 2 and 3.
The proposed approach is shown to be more general and more convenient to use than the
bias/model factor approach because the distributions of the hyperbolic parameters do not depend
on the allowable displacement. It can be argued that a random vector containing two curve-
fitting parameters is the simplest probabilistic model for characterizing the uncertainty in a
nonlinear load-displacement curve and should be the recommended first-order approach for SLS
reliability-based design.
mean COV
- Hyperbolic parameters: a 5.15 mm 0.6
b 0.62 0.26
Capacity: Qm mom 0.5
Applied load: Q mQ 0.2
Allowable displacement: ya mya ag^^"
•*— EN1990
•—EPRI
m
ya j / \
(mm) J^[/^'^
50-V^V^^
2 5 ' ' \ s ^
1 5 ^
1 2 3 4 5
Mean factor of safety = mQm/mQ
Figure 3. Relationship between reliability index and mean factor of safety for ACIP piles.
26
TAIPEI2006 International Symposium on New Generation Design Codes
for Geotechnical Engineering Practice
Nov. 2-3, 2006, Taipei, Taiwan
Roger Frank
ENPC (National School of Bridges and Roads), Paris, France
SUMMARY
Eurocode 7 on geotechnical design consists of two Parts: General rules (Part 1, EC 7-1) and Ground
investigation and testing (Part 2, EC 7-2). Their conversion into foil European Standards (ENs) has
now been completed. EC 7-1 is a rather general document giving only the principles for geotechnical
design within the general framework of Limit State Design. These principles are relevant to the calcu-
lation of the geotechnical actions on structures and to the design of the structural elements themselves
in contact with the ground. EC 7-1 includes fundamental sections on the basis of geotechnical design
and geotechnical data, sections on the design of various types of geotechnical structures such as spread
foundations, piles, anchorages, retaining structures and embankments and sections on geotechnical
verifications such as hydraulic failure and overall stability. Detailed design rules or calculation mod-
els are only given in informative Annexes. Annex A is important as it specifies the partial factors for
verifications of the Ultimate Limit State in persistent and transient design situations as well as correla-
tion factors for the characteristic values of pile bearing capacity. However, the numerical values of the
partial or correlation factors given in Annex A are only recommended values. All the other Annexes in
which detailed design rules or calculation models are given are informative annexes.
EC 7-2 is devoted to laboratory and field testing and states the essential requirements for the
equipment and test procedures, for the reporting and the presentation of results, for their interpretation
and, finally, for the derivation of values of geotechnical parameters for design. It consists of Section 1
General, Section 2 Planning of ground investigations, Section 3 Soil and rock sampling and ground-
water measurements, Section 4 Field tests in soils and rocks, Section 5 Laboratory tests on soils and
rocks and Section 6 Ground investigation report and contains informative annexes dealing with the
most important laboratory and field tests.
The discussions about verifications of geotechnical design usually focus on approaches performed
by calculations. Nevertheless, it should be stressed that calculations are not the only means of check-
ing that the basic requirements are fulfilled. EC 7-1 also offers the adoption of prescriptive measures,
experimental models and load tests, as well as the observational method. Thanks to the Eurocodes, a
single format will be used for the mathematical analysis of the ultimate limit states throughout the
construction sector in Europe in future. Accordingly, for any section in a structure, structure-soil inter-
face or the soil, it will have to be verified that the design value of the effects of actions, Ed, never ex-
ceeds the design bearing capacity or the design resistances, Rj, i.e.:
Ed < Rd (1)
There has to be a clear-cut distinction between the effects of actions and resistances in order for the
general limit state equation to be applied. However, in geotechnical engineering, there are many cases
in which it is not possible to make a clear-cut distinction between the effects of actions and the resis-
tances. Additional problems concerning the application of equation (1) are caused by the fact that there
are two entirely different ways of introducing the partial safety factors in geotechnical engineering. On
the one hand, the design values, Ed and Rd, of the geotechnical effects of actions and resistances can be
determined by what is known as the method of factored shear parameters. In this method, the partial
factors are applied to the characteristic shear parameters, <p\ and c\. Thus the design value of the ef-
28
Session I
—«s®**—
Shian-Jin Chiou
Sinotech Eng. Consultants, INC., Taiwan
Chyi-Tyi Lee
National Central University, Taiwan
Po-Shen Lin
National Central University, Taiwan
Yi-Ben Tsai
Pacific Gas and Electric Company, USA
SUMMARY
Before Chi-Chi Taiwan earthquake, the results of Probabilistic Seismic Hazard Analysis (PSHA)
produced by different agencies and different researchers totally underestimated earthquake hazards in
Central Taiwan. Therefore, there were some impacts on PSHA from the Chi-Chi Earthquake
experience, such as: (a) Saturation of local magnitude, (b) Significant hangingwall effect,(c) The need
to emphasize the fault source, (d) The need to use the close-distance to fault-plane. After the
earthquake, the need to emphasize the fault source was recognized, and the need to adopt a proper
attenuation relationship with considering the strong-motion data recorded during the Chi-Chi
earthquake was urged for PSHA usage. In this study, we conducted a review of readily available
information on attenuation of peak ground acceleration, tectonic setting, geology, active faults, and
seismicity of Taiwan for PSHA. However, we will consider fault activity and adopt proper attenuation
relationship in this study.
Taiwan is situated on the boundary between the Eurasian Plate (EP) and the Philippine Sea Plate
(PSP) where active oblique collisions and subduction are taking place. Therefore, Taiwan's orogenic
belt has high rates of seismicity and high fault-slip rates. We adopted available information of geology
and seismology to revise the probabilistic of Taiwan. New Seismic hazard maps were done for 10%
and 2% probability of exceedance in a 50 year period individually. In order to construct the maps we
defined four major seismic sources in Taiwan which were: (a) fault source, (b) areal/regional source,
(c) subduction plate interface and (d)subduction plate intraslab. We used the mainshocks from the
earthquake catalog of 1900 to 1999 to evaluate the earthquake recurrence rate for the regional zones
and subduction-intraslab sources by Truncated-Exponential model. We also used the fault-slip rate to
estimate the earthquake recurrence rate of faults and subduction-interface sources by Characteristic-
Earthquake model. For the first time in Taiwan our revised PSHA took into consideration the fact that
subduction plate sources induce higher ground-motion levels than crustal sources, and active faults
induce the hanging-wall effect in attenuation relationships. We also used the logic tree method to deal
with the uncertainties of each parameter in the PSHA.
The two highest hazard levels in Taiwan were shown in the areas of eastern longitudinal valley and
from the western foothills to the coastal plain. These two areas are separated by the central mountain
range which has a decidedly lower hazard level. After considering the fault activity in our revised
PSHA, we found that the PGA level of near-fields in Taiwan always exceeds 0.4g in 475-year return
period. However, in previous studies the proper hazard could not be obtained because fault sources
were not considered in the PSHA, especially in long return period. This situation was very obvious in
the central part of Taiwan (Chi-Chi earthquake disaster region) and in the HsinChu-MiaLi region. It
32
TAIPEI2006 International Symposium on New Generation Design Codes
for Geotechnical Engineering Practice
Nov. 2-3, 2006, Taipei, Taiwan
SUMMARY
This paper demonstrates that simple reliability analysis can effectively provide an objective basis
for making engineering decisions. The equations for a reliability evaluation based on reliability
index as defined in Hasofer and Lind are presented. The background of the challenge in the
design of large diameter bored piles and high capacity driven H-piles are presented. The
necessary databases are established so that real and relevant statistical attributes are available for
reliability assessment. The results of such an investigation showed that:
• Current bored pile design practice for side resistance in saprolite stratum is neither excessively
conservative, nor overly-optimistic.
• The current practice for high capacity steel H-pile is rather conservative.
• Both the CAPWAP and the HKCA methods perform equally well and one cannot conclude
which method is superior.
Reliability assessment
The pertinent equations for reliability assessment are based on Li, Lee and Lo (1993). It is
recognised that engineering design are based on Rpu, ultimate capacity as predicted by a certain
design model; and Rpu in general differs from (XR, the mean resistance. To bridge the "gap"
between Rpu and \lR and to take into account the range of design models available in pile deign, we
define a bias factor, B, for a given design model as:
/ JR
where ys, y& are partial factors on the mean values of action and resistance respectively. The
partial factor on calculated pile capacity, Rpu, is given by:
fs =
XrA)
The required overall factor of safety, FOS, is simply given by: FOS = JsfR = 7SI^B7R
where p a is atmospheric pressure in consistent unit (taken as 100 kPa), N is the standard
penetration test blow count, and fsi is the upper limit of side resistance. We take fsi = 150 kPa
based on a detailed survey of the database. The Building Department will also required site and
project specific pile load test to "prove" the design as experience suggest the side resistance can
be sensitive to detailed construction method. There has been debate on whether such an approach
is too conservative. Zhang et al (2002) examined 33 load test results for a range of foundation
consitions and construction methods, and conclude that that current practice is in fact optimistic.
The authors examined pile load test results from 10 sites (Lo and Li 1999, 2000, 2002). Only
test data from granitic saprolites are included, and piles constructed with the inappropriate
technique of using a permanent undersized liner are excluded. From these data pairs we can
calculate 19 data points for bias factor of bored pile design. This data based for granitic saprolite
yields the statistical attributes for the bias factor. The outcome of the reliability analysis showed
that current practice is neither too conservative as some might claim, nor too optimistic.
Driven H-Pile
Two methods for predicting the ultimate pile capacity from dynamic measurements have been
proposed and debated. The first method is the CAPWAP analysis, and the second method is a
simplified approach proposed by HKCA (Hong Kong Construction Association). The latter is
based on Broms and Lim (1988) but with minor adjustment to suit local conditions. We define
two Bias factors, BCAP and BHK:
The authors have established a database of 29 H-piles load tested to failure from two different
sources. From this database, the statistical attributes of for both the CAPWAP and HKCA method
were calculated. The KS goodness-pf-fit test (Ang and Tang 1975) was used to test two assumed
distribution, normal and lognormal. The results of the KS test indicate that either distribution is
acceptable. Therefore, normal distribution is assumed for sake of simplicity. We still assume Vs
= 0.15. However a target P of 3 was assumed because the resistance along the complete
embedded pile length was accounted for during dynamic measurements and pile load tests. It is
then a matter of substituting the statistical attributes of BCAP and B H K and the partial factors and
required overall FOS were calculated. The outcome of the reliability calculations
unambiguously indicated that both methods are equally reliable, and that the current Hong Kong
practice for friven H-pile may be too conservative.
Conclusions
Simple reliability analysis provides an objective basis for making engineering decisions on
debatable issues. This is illustrated by examining the design and construction of bored piles and
H-piles in Hong Kong. However, like any analysis, suitable input parameters are needed. In
reliability analysis, the inputs are statistical attributes of database compatible with the problem
under investigation. This is not a new concept, but analogous to requiring representative soil
parameters for any meaningful deterministic modelling. This requires a compatible marriage
between engineering know-how and reliability analysis. Reliability analysis is not a replacement
of our deterministic deign tool or engineering know-how. Rather it is an effective supplementary
tool.
34
TAIPEI2006 International Symposium on New Generation Design Codes
for Geotechnical Engineering Practice
Nov. 2-3, 2006, Taipei, Taiwan
W.M. Cheung
Geotechnical Engineering Office, Civil Engineering and Development Department, HKSAR
Government, Hong Kong
L.S. Lui
Geotechnical Engineering Office, Civil Engineering and Development Department, HKSAR
Government, Hong Kong
SUMMARY
Setting Standards
Prior to the 1970s, slope design and construction were generally based on rules of thumb, such as 10:6
cutting with 1 to 2 m wide berms at about 7.5 m intervals giving an average slope angle of about 50°.
Following the occurrence of disastrous landslides in the 1970s, the Hong Kong Government established
the Geotechnical Control Office in 1977 (renamed as the Geotechnical Engineering Office, GEO, in
1992). An important function of the GEO is setting geotechnical standards. Since its establishment, the
GEO has produced many publications covering a wide range of geotechnical engineering topics. The
more comprehensive ones are called Manuals, Geoguides and Geospecs (Table 1). The main objective of
publishing these documents is to allow the profession to use a series of common, up-to-date and
comprehensive geotechnical standards which are appropriate to Hong Kong conditions. The documents
present recommended standard of good practice for various geotechnical activities.
Geoguides:
• Geoguide 1: Guide to Retaining Wall Design, 2nd edition
• Geoguide 2: Guide to Site Investigation
• Geoguide 3: Guide to Rock and Soil Descriptions
• Geoguide 4: Guide to Cavern Engineering
• Geoguide 5: Guide to Slope Maintenance, 3 rd edition
• Geoguide 6: Guide to Reinforced Fill Structure and Slope Design
Geospecs:
• Geospec 1: Model Specification for Prestressed Ground Anchors
• Geospec 3: Model Specification for Soil Testing
Apart from these three categories of publications, the GEO published other related geotechnical
publications under the series of GEO Publications, GEO Reports, Technical Guidance Notes and
Geological Memoirs. GEO Publications document results of comprehensive literature reviews. GEO
Reports generally present results of applied researches and studies or reviews of aspects of geotechnical
engineering. GEO Technical Guidance Note is to satisfy the need to promulgate new or revised
technical guidance from time to time where the introduction of a new standard or updating of an existing
standard is not justified. When the technical guidance in a Technical Guidance Note has been widely
accepted by the geotechnical profession as standards of good practice, the practice will be incorporated
Process of Production
The GEO benchmarks against international standards and adapts the standards for local use as
appropriate in the course of producing geotechnical guidance documents. New specifications and
guidelines are prepared as needed to suit the specific nature of the local geological condition, works
practice, and legal and environmental requirements. Stakeholders are always consulted in the setting of
geotechnical standards. For Manuals, Geoguides and Geospecs, extensive consultation with consulting
engineers, contractors, academics, professional bodies and other government departments are carried
out. All comments are duly considered to ensure that the document would be considered a consensus
document by interested parties in Hong Kong.
Conclusion
Numerous geotechnical guidance documents in the form of Manuals, Geoguides, Geospecs and other
publications and reports are available in Hong Kong. These documents aim to promote good practice in
geotechnical engineering. Some of the guidelines are adopted as the local standards by the Government
through Technical Circulars for public development projects. The same standards are generally adopted
for private buildings and civil engineering projects through the Buildings Ordinance and its related
Regulations and Practice Notes. These standards have been benchmarked against international ones and
are adapted to suit local conditions.
36
TAIPEI2006 International Symposium on New Generation Design Codes
for Geotechnical Engineering Practice
Nov. 2-3, 2006, Taipei, Taiwan
B. Schuppener
Federal Waterways Engineering and Research Institute, Karlsruhe, Germany
SUMMARY
When Eurocode 7: Geotechnical Design, Part 1: General Rules (EC 7-1) is implemented in the European
Member States, each state will need to make two important decisions concerning the design of geotechni-
cal structures. Three design approaches are described in the code and each state can select the one that best
suits its national design traditions and stipulate its use in geotechnical design. Furthermore, the Member
States must establish the values of the partial factors in accordance with national safety requirements.
Both, the choice of design approach and the selection of the partial factors, must be seen as a single unit as
they are interdependent.
The selection of the design approach and the numerical values of the partial factors in Germany was
based on the principle that the safety level of the global safety concept that has been used successfully for
decades and should be maintained as far as possible. I.e. a geotechnical design in accordance with EC 7-1
should result in more or less the same dimensions as the former global safety concept. A comparative de-
sign, in which each of the three Design Approaches in EC 7-1 is applied to a strip footing, is used in the
paper to illustrate the option that has been selected for Germany. It shows that the Design Approach DA
2*, in which the partial factors are applied at the end of the calculation when the limit state equation is
checked, not only best fits the tried and tested safety level of the former global safety concept but also re-
sults in the most economic design.
Thanks to the Eurocodes, a single format will be used for the mathematical analysis of the ultimate limit
states throughout the construction sector in Europe in future. It has to be verified that the design value of
the effects of actions, Ed, never exceeds the design bearing capacity or the design resistances, Rd, i.e.:
E d <R,,
However, in geotechnical engineering, there are many cases in which it is not possible to make a clear-cut
distinction between the effects of actions and the bearing capacity. For instance, the action of the active
earth pressure also depends on the shearing resistance or the shear strength in the failure surface of the
active sliding wedge. In other cases, the resistance of the soil depends on the magnitude of the action. For
instance, the sliding resistance is governed by the magnitude of the effect of the action due to the vertical
component of the bearing pressure resultant. Last but not least the ground bearing resistance is dependent
on the eccentricity and the inclination of the acting forces.
Additionally, there are two entirely different ways of introducing the partial safety factors in geotech-
nical engineering:
- On the one hand, the design values, Ed and Rd, of the geotechnical effects of actions and resistances can
be determined by what is known as the method of factored shear parameters:
tan<p'd = tanip'k/Y<p c'd = c'k/Yc
Ed = f (<p'd, c' d ); Rd = f (<p'd, c'd)
38
TAIPEI2006 International Symposium on New Generation Design Codes
for Geotechnical Engineering Practice
Nov. 2-3, 2006, Taipei, Taiwan
T. Nagao
National Institute for Land and Infrastructure Management, Japan
W.Lu
CHEC Guangzhou Port Construction Company, Guangzhou, China
K.Lee
Konyang University, Nonsan, Korea
H.Kim
Korea Ocean Research & Development Institute, Ansan, Korea
SUMMARY
International Organization for Standardization (ISO) has adopted the applicability of the reliability
based technical standards based on ISO 2394, which may affect to various structures including port
and harbor facilities. Eurocode discussed among European countries might most probably be adopted
as ISO standards. It surely has some negative implications for Northeast Asian countries' economies
such as Japan, Korea, and China which have completely different types of technical standards,
especially, since the ISO is compulsory for member economies of the WTO. A new round of revision
for technical standards of port & harbor facilities are being carried out in China, reliability based
design methods are the fundamental of technical standards. At the same time, the design codes based
on reliability are being prepared in Japan, and will be completed by end of the year 2006 or later. Also,
in Korea, reliability based design methods are currently hot issue among code developers, professors
and government officers to adopt as national standards. Therefore, joint studies are necessary for the
three countries in order that the technical standards of three countries are more scientific and
reasonable. Moreover a certain compatibility among the technical standards of three countries are
critical to promote the cooperation of technology and economy. To improve the effectiveness of joint
study, the representatives of three countries discussed the subjects of joint study on September 1999.
And the work of joint study was divided among three countries and the following arrangement was
approved by the representatives of three countries. Three countries carried out their roles for
international joint study of reliability for port & harbor structure from March 2003 to September 2005.
This paper summaries some results of the joint study performed during the first and second year of the
joint study among China, Japan, and Korea.
In Conclusion
1) Korea reviewed the uncertainties of soil properties for port and harbor structure which is a vital
part in determination of partial safety factor. However, quantification of their uncertainties to design
port structure is most difficult and challenging. The Korea reviewed certain soil uncertainties and
evaluation methods of uncertainties which are closely related with design variables for port and harbor
structures with following conclusions; a) Enough information is not available in the prototype cases for
the stochastic variables of relevant soil parameters to evaluate the uncertainty of the parameters. The
variability of soil properties such as the effective friction angle, undrained shear strength of the clay,
cohesion and angle of dilation are relevant parameters in determining the bearing capacity of the
2) Japan analyzed safety index using life cycle cost analysis, and estimated sensitivity index based
on FORM for port and harbor structure focusing on the external stability problems (sliding,
overturning) of caisson-type breakwaters. Japan concluded as:, a) The dominant parameters having a
large influence on the safety index are the wave force/the uplift pressure, the coefficient of friction,
and the surcharge. The tide level is not dominant with respect to the external stability of breakwaters.
Since these dominant parameters change their sensitivity factors depending on the safety indices, their
values should be carefully evaluated, b) The same design parameter takes different values of
sensitivity factors depending on performance functions and structural types. When setting the values
of partial safety factors in the Level 1 reliability-based design method, it is necessary to use proper
values of sensitivity factors in accordance with performance functions and structural types, c) The
sliding failure mode is dominant in the present design section. The failure modes are positively
correlated. The correlation becomes stronger as the sliding safety decreases. These facts indicate that
overturning failure or foundation failure unaccompanied by sliding rarely occurs, which agrees well
with the past disaster cases.
3) China has studied how to determine partial safety factors for gravity type quay wall, which usually requires
a large sample of statistical data for probabilistic code calibration, actions that are used to derive action effects
and resistance. The reliability indices for structures of port facilities implied by use of previous deterministic
design codes have been derived by regressive calculations on some structural members by the calibration
method. Due to incomplete statistical analysis of some uncertain parameters, the derived reliability indices are
not totally accurate, but can be regarded as an indicator of the real possible reliability indices. China also
determined some results of partial safety factors for different failure modes of gravity quay wall, however,
they are not general. More comprehensive code calibration study with as more samples data and
construction cases as possible should be conducted in order to derive appropriate partial safety factors for
gravity type quay wall.
40
Session II
—*s*s*—
Performance-Oriented
Geotechnical Analysis
TAIPEI2006 International Symposium on New Generation Design Codes
for Geotechnical Engineering Practice
Nov. 2-3, 2006, Taipei, Taiwan
SUMMARY
Offshore structures, namely, oil drilling platforms, jetties, Quay & Harbour structures, tension leg
platforms, etc. are supported on pile foundations. Besides usual load from super-structure (dead load,
live load, etc.), these piles are subjected to continuous lateral cyclic loading due to ocean waves. The
quasi-static nature of such loading induces progressive alteration in the bearing capacity and head
displacement of the foundation.
A comprehensive review of literature indicates that limited research works have been done in the
related areas. The relevant contributions are made by Poulos (1981), Purkayastha & Dey (1990),
Narasimha Rao & Prasad (1992), Narasimha Rao et al (1993), Gupta & Dutta (1996), Purkayastha et al
(1997) and Basak & Purkayastha (1999 & 2003). Some of the works are theoretical while the others
are experimental (laboratory and field).
As reported by Poulos (1981), basically three reasons against alteration of strength and stiffness of
pile foundation under lateral cyclic loading can be identified. These are :
• Development of excess pore water pressure generated during cyclic loading in progress.
• Gradual accumulation of irrecoverable plastic deformation of soil surrounding the pile surface.
• Rearrangement and realignment of soil particles surrounding the pile surface.
Understandably, the first two reasons are adequate for cohesive soil, whereas for cohesionless soil, the
third reason is primarily valid.
It has also been found that the alteration in pile capacity due to transient loading depends upon the
following parameters (generally known as cyclic load parameters) : number of cycles, frequency and
amplitude.
Since no standard apparatus for imparting lateral cyclic load on piles is available, a new set up is
designed and fabricated. By means of this apparatus, a symmetrical two way lateral cyclic loading can
be applied on model piles for a specified no. of cycles at a specified frequency and with a specified
amplitude with the help of motor and other mechanically and electrically controlled units. After
completion of the specified no. of cycles, the axial post-cyclic pile capacities are determined.
Experiments are carried out in loose sandy soil bed at a relative density of 35%. Stainless steel
shafts are used as piles using 2 x 2 group, the L/d ratio being 18 and the c/c distance between the piles
are 3d.
During experiment in progress, it is observed that a basin-like depression is formed around the pile
group in the vicinity of soil surface. This may be partially due to gradual shifting of soil particle away
from the pile surface as well as partially because of compaction of the sand mass surrounding the pile
group.
The alteration in the axial capacity of the piles due to cyclic loading has been expressed by a term
degradation factor which is defined as the ratio of the post-cyclic to pre-cyclic static axial capacities of
the pile group (Poulos, 1981). It is observed that some improvement in the axial capacity of the pile
group has occurred which is indicated by the values of the degradation factor to be greater than unity.
The load deflection curves are observed to be hyperbolic in nature. The ultimate capacities are
estimated by double-tangent method. It is observed that for all the tests conducted, the degradation
factor increases with no. of cycles and also there is a tendency of asymptotic stabilization. Also, the
degradation factor increases with frequency and also there is a tendency of asymptotic stabilization.
44
TAIPEI2006 International Symposium on New Generation Design Codes
for Geotechnical Engineering Practice
Nov. 2-3, 2006, Taipei, Taiwan
1 SUMMARY
Large-diameter bored piles (LDBPs) are a popular foundation type in Hong Kong. Depending on the
geology, the pile capacity may consist of shaft resistances in soil and rock socket, and toe resistance on
rock or soil. Several design methods are available for the estimation of these resistances, and one or
several resistance components may be considered in design. The design methods include those outlined
by Buildings Department (2004) in PNAP141 and O'Neill and Reese (1999) (for comparison purpose),
and those based on empirical correlations with blow counts from the standard penetration test (SPT-N),
effective overburden stress ( a , ' ) or unconfined compressive strength (UCS) of rocks. The estimated
capacity values using these methods vary widely and the corresponding reliability levels of these methods
are not clearly known to date. In this paper, the design methods for large-diameter bored piles are
reviewed. A database containing information of load tests on 60 large-diameter bored piles in Hong Kong
is developed. The response of the test piles is studied and the resistance components of the piles are then
used as a basis to evaluate the performance of the individual design equations and combinations of these
design equations. The performance of a specific method is measured by a model bias factor and its
coefficient of variation.
2 CORRELATION STUDY
Information of 60 load tests on large-diameter bored piles in Hong Kong is collected for study. A
correlation study was performed to obtain empirical relations for calculating the shaft and end bearing
resistances of piles. For the shaft resistance in the saprolites, the measured average maximum skin friction
in a soil layer is correlated with the mean standard penetration test value (N) or the mean vertical
effective stress (^) at the centre of the layer. Correlations are established for different types of soils
including fill, alluvium, decomposed volcanics, decomposed granites. For piles socketed in rock, the
correlation study was preformed to establish the relations between the mobilized shaft resistance in the
socket or the end bearing resistance and the unconfined compressive strength of rock.
The accuracy of a design equation can be measured by a model factor that is defined as the ratio of
measured resistance to calculated resistance. Using the results in the load test database, the model factors
for the O'Neill and Reese method and two empirical correlations for the calculation of shaft resistances in
various soils are studied. The degree of dispersion for these design methods is quite large. The COV
ranges from 0.48 to 1.99.
Similar to design methods for piles in soils, the accuracy of three design methods for piles in rocks is
also evaluated. The degree of dispersion of the three design methods is also quite large. The COV ranges
from 0.54 to 1.34 for both shaft resistance and end bearing. In particular the O'Neill and Reese (1999)
method for predicting the end bearing of the piles is associated with a very large model bias factor and
severely underestimates the end bearing.
Piles in saprolites
The estimated capacity based on the method involving presumed bearing values shows a large bias factor
of 11.28 with a fairly low COVR. This means that the scatter of the model bias factor is not large
compared with other calculation methods but the pile capacity is systematically underestimated. The
predicted capacity values based on the O'Neill and Reese (1999) methods and other correlations are
associated with relatively low XR but moderately high COVR values. However, the accuracy of these
methods is still comparable to the correlations derived for driven piles in sand or clay. The COVR
predicted by the Meyerhof (1976) correlation for driven piles in sand is 0.5 while the COVR of the
correlations established in this study ranges form 0.39 to 0.53.
Piles in rocks
The estimated capacity based on the O'Neill and Reese (1999) correlations shows a large bias factor of
14.97 with a fairly high COVR, which means the predictions from this method are highly scattered. The
predicted capacity values based on the correlations with UCS established in this paper have a mean model
bias factor closer to 1 and a lower COVR of 0.40 compared with that based on the PANA 141 method. The
accuracy of the two methods is still comparable to the correlations derived for the driven piles in sand.
46
TA1PE12006 International Symposium on New Generation Design Codes
for Geotechnical Engineering Practice
Nov. 2-3, 2006, Taipei, Taiwan
M. Honda
Nikken Sekkei Civil Engineering Ltd., Tokyo, Japan
K. Ogawa
Pacific Consultants Co., Ltd., Tokyo, Japan
Y. Wakatuki
SUMMARY
Introduction
There is a trend in Japan for the structural design codes to move towards the performance based
design/specification (PBD/PBS) corresponding to the conclusion of WTO/TBT agreement.
Furthermore, there is also some expectation in the background that, since PBD/PBS scheme can
evaluate the performances of structures more directly compared to the conventional descriptive
design scheme, structures can be designed more rationally.
To design geotechnical structures, it is necessary to evaluate the interaction between structures,
e.g. foundations, and soil which may exhibit very complex behaviour. It is more so in earth
structures where the mechanical characteristic of soil need to be understood more precisely.
Moreover, grounds are always heterogeneous and amount of information is mostly not sufficient
to estimate all spatial variation of them. Consequently, some uncertainties are always included in
the evaluation of mechanical behaviour of geotechnical structures even if an external load is fixed.
The traditional descriptive type design specifications have supplied a large amount of
infrastructures successfully. Although some performances which are required to satisfy objectives
are evaluated indirectly in the traditional design scheme, the reliability of the design has been
confirmed by experience, especially those which caused damages.
To introduce PBD/PBS scheme, it is anticipated to cause some problems, e.g. difficulty of
design, evaluation of reliability etc., because it is expected to introduce more sophisticated new
design methods to evaluate performances of structures directly. However, further rationalization
of structure design can be expected by introducing such methodologies, and more flexible
responses to the required performances that corresponds to peculiar conditions at site. It is also an
important task to take into account the long-term maintenance cost and optimise the life cycle cost
of a structure.
This paper discusses technical issues in introducing the PBD/PBS scheme to design codes of
earth structures. The direct aim of this study was to add two chapters to "JGS4001-2004
Principles for Foundation Design Grounded on a Performance-based Design Concept" (whose
nickname is "Geo-code21 (JGS(2004))") on embankments and cut slopes. The basic concept of
PBD/PBS based design code is well presented in "code PLATFORM (JSCE(2003))". The
guidelines like IS02394 and "Basis of Structural Design for Buildings and Public Works
(JSCE(2002))" are also taken into consideration, which recommend LSD (Limit State Design) as a
design verification method in design code drafting.
Performance requirement
According to the code PLATFORM, the performance requirements are statements expressed in
plain language describing the performance of the structure with respect to the given objectives.
Usually, serviceability to keep functions of the structure for given combinations of actions;
safety to protect lives of inside and outside of the structure from given combinations of actions;
restorability to enable the use of a structure by rehabilitation that is technically possible and
economically feasible; are required to structures.
Among the three requirements, the restorability is newly invented performance requirement in
Japan based on the experiences of Kobe earthquake of 1995. If the buildings are kept with in this
performance requirement, property value of a building is restored, which was much concern of
many apartment room owners.
Performance criterion
Performance criteria are more quantitative and engineers oriented specifications of given
performance requirements. The performance requirements are described in more general
terminologies, whereas the performance criteria are more qualitative so that they can be used for
design verification.
The difficult aspects of performance criteria for earth structures are that most of the traditional
design method only deals with the stability of earth structures, whereas some performance
requirements are given by displacement or deformation terms. In order to overcome this
difficulty, either new design method, for example FEM, need to be introduced or the traditional
design method need to be newly interpreted and approved by a deemed to satisfy solution.
It is also felt that critical displacement/deformation values for earth structures are very difficult
to determine mainly due to lack of background data; for example, what would be the maximum
differential settlement for drivers to feel uncomfortable when driving a car on a highway?
Verification
It is the stance of Geo-code 21 that the limit state design, which clearly define the limit states and
limit the occurrence of the calculate probability of the defined limit states within the allowable
amount, serves basic framework in PBD/PBS scheme.
The most of traditional design verification methods are just calculating the stability aspects of
the earth structures, and really estimate the displacements and deformations. It becomes
necessary to apply the numerical analysis technique, like FEM, to design. Due to lack in the
performance evaluation of these sophisticated methods, many aspects of these numerical based
techniques are discussed. The conclusion is yet not reached. However, it is necessary to
introduce these modern techniques to routine design of earth structures in the near future.
Final remarks
The final conclusions are not yet reached, so as the draft. However, it is felt that the design code
of the earth structures can be put in the framework of PBD/PBS scheme.
48
TAIPEI2006 International Symposium on New Generation Design Codes
for Geotechnical Engineering Practice
Nov. 2-3, 2006, Taipei, Taiwan
SUMMARY
In the modern new era, the society demands for increasing of mega facilities and structures are
threatened by several challenges, such as the natural hazard, human crisis, and engineering
difficulty. In order to solve the challenges with sufficient safety and reasonable budge, a
performance-oriented problem-solving has been immerged among the geotechnical engineering.
This article presents the risk assessment and retrofitting strategy for the steel-grillage-anchored
foundation of electricity tower on slope. The electricity power network system is an essential
industry that provides electric power for the entire country. The Taiwan Power Company is
responsible for this duty to safely deliver electricity to all necessary corners among the island of
Taiwan. Besides, to maintain the electricity power facility in a safety level to against all possible
threats from hazards is an important issue among the emergency response and maintenance works.
In order to ensure the safety of these transmission towers, the main focus of this article aims to
evaluate the earthquake and landslide risk assessments of the electricity towers on slope for the
No.l north-south direction 345kV line. Furthermore, the adequate retrofitting strategy is
established to ensure the safety in the future service life.
This article first presents a performance-oriented procedure for the risk assessment for the
general geotechnical structures, especially for the existing structures. This procedure emphasized
on four important stages (shown in Figure 1), including the initial condition assessment, the risk
assessment, the impact and retrofitting assessment, and the long term performance maintaining
strategy.
In the first stage, in order to completely understand the initial design purpose, the past
maintenance and hazard history, and the current performance of the structure, the basic
information collecting on design, construction records, and the history on the requirement and
mitigation records are fully desired. Besides, a proper investigation and monitoring approach can
be great helps to understand the current safety level of the structure itself. Furthermore, with the
help on the modern technology, a proper database system should also be established to store and
maintain the accumulated information. Based on the information from the first stage, the second
stage emphasizes on analyzing the possible threats that producing the various hazard scenarios. In
the beginning, all possible factors that may affect the safety of the structure should be analyzed.
For example, potential factors include earthquake, rainfall, geology, tomography, previous hazard
Current Condition
Assessment ,s ,
Low * *"*•»«' i^cveJHi'
Medium Low I eve 1 2 (1) * •»>J «* S
Medium High tt (sreU $>$ '' I evel 2 (16 ^-. ^ i.. '<
s
High ; S «YJ3 I *#?> W e i 5 <S?S , L evel 2(1) i U*e* 3 i >J
w J i Mtdium ,, ,
Low Medium' ow T High
hign
Risk Level
Note: Number shown in () represents the amount of tower
Figure 2: Retrofitting Plan for the steel-griliage-anchored foundations of electricity towers on slope.
In which,
Level 4: The foundations of towers classified in this level require immediate retrofitting
attention. The possible actions (with site investigation) include re-locating tower,
foundation retrofitting (e.g. strengthening or replacing the existing foundation), and
slope instability protection (e.g. retaining structure system).
Level 3: The foundations of towers classified in this level require retrofitting attention.
However, this attention can be reasonably postponed by establishing monitoring
system for emergence. The possible actions (with site investigation) include
foundation retrofitting (e.g. strengthening or replacing the existing foundation) and
slope instability protection (e.g. retaining structure system).
Level 2: The foundations of towers classified in this level require regular inspection with
moderate retrofitting program after proper site investigation.
Level 1: The foundations of towers classified in this level require only regular inspection.
50
Session III
—-mam-—
Ting-Rong Chen
National Taiwan University of Science and Technology, Taipei, Taiwan
SUMMARY
Uncertainties are abundant in civil engineering systems. Traditionally, geotechnical engineers have been
used to design approaches based on safety factors to account for uncertainties due to their simplicity and
convenience. However, safety-factor approaches are not rigorous for the purpose. Recently, reliability
analysis is popular for quantifying these uncertainties. Intuitively, reliability increases with factor of
safety, implying that there exists a monotonic increasing functional relationship between them. However,
it is not clear whether this relationship exists for any problems and if it exists, how to determine the
relationship. This research aims to answer these questions.
Let Z and 6 be uncertain variables and design variables of the target system. Also let D be the al-
lowable design region in the 6 space. Let F denotes the failure event: F= {/?[Z,G]>1), where R[Z,Q] is
called the limit-state function. This function does not necessarily define the complete collapse of the
system but the performance of the system, e.g. serviceability and ultimate capacity. Let us further define
the nominal limit-state function R„(6), a positive function of 9. An example of R„(Q) is to take R[Z,Q] but
fix Z at certain chosen nominal values, e.g. their mean values. The safety-factor approach is to enforce
the following constraint during the design process:
7j'Rn{0)<\ (1.)
where 7] = 1 is the designated safety factor. On the other hand, the reliability-based design approach is to
enforce the following constraint during the design process:
P(R[Z,0]>\\0)=$p(Z\0)-l(R(Z,0)>l)dZ<Pr (2)
where PF* is the desirable failure probability; p(ZB) is the probability density function (PDF) of Z given
6; /(.) is the indicator function, i.e. it is equal to 1 if the inside statement is true; otherwise, it is zero.
If the equivalence between the two approaches can be proved and established, it will be significant
in the following sense:
(a) One can then achieve a reliability-based design by using a safety-factor approach, which is much
simpler and more convenient than the former.
(b) Practical geotechnical engineers who are not familiar with reliability concept can easily achieve
reliability-based design by using the equivalence.
A theorem is derived to show that under certain condition, safety-factor designs and reliabil-
ity-based designs are equivalent:
Theorem: If the distribution of R[Z,Q]/R„(Q) is invariant over the entire allowable design region D, there
exists pairs of [TJ ,PF] such that the following constraints are equivalent:
T)'Rn{0)<! (3.)
P{R[Z,0]>P)<P; (4.)
P(R[Z,0]-1]'Rn(0)>O) = P^ (5.)
K^%l{G^\e^)>rf)^±l{G^>rf) (6.)
where GP=G(Z<''\ 9 W ). By changing the rf value, one can find the corresponding PFMCS* values by re-
petitively applying (6.), i.e. the entire functional relationship between TJ and PF is obtained. Note that
the same N MCS samples can be repetitively used to estimate the entire functional relationship.
When Pr' is small, Subset Simulation (SubSim) can be used to estimate the entire functional re-
lationship more efficiently than MCS:
where CPJ-IJ denotes the i-th sample obtained in thej'-th SubSim stage. By changing the rf value, one
can find the corresponding Pr estimates by repetitively applying (7.). The performance of SubSim in
estimating the rf and PF relationship is also independent of system dimension, complexity, and the
uncertainty dimension.
Recall that the theorem is based on the premise that the distribution of R[Z,B]/Rn(Q) is invariant
over D. This can be achieved if the nominal limit-state function 7?„(0) is chosen carefully. However,
finding such a perfect choice is itself a difficult problem. Although finding the perfect choice is hard,
finding a nominal function Rn(&) such that the distribution of R[Zfi]/R„(fi) is roughly invariant is a
relatively easy task. As seen in the examples in the next section, /?„(6)=R(.E(Z),6) and R„(Q)=E[R(Z,6)IQ]
are usually acceptable choices. This is because although the distribution of R[Z,Q] may change dra-
matically with 9, the distribution of R[Z,G]/R„(&) usually does not due to the cancellation effect between
R[Zfi] and J?„(8).
54
TAIPEI2006 International Symposium on New Generation Design Codes
for Geotechnical Engineering Practice
Nov. 2-3, 2006, Taipei, Taiwan
In recent years, large diameter steel pipe piles embedded deeply are often used in port
construction in Japan. This situation is greatly different from former situation. However, the
estimation procedure of the bearing capacity of a pile has seldom changed. Comparison of the
bearing capacity estimation technique of a pile internationally and examination of the bearing
capacity mechanism of an open ended pile are conducted.
Conclusion is as follows;
In this paper, the change of the pile used for the port facilities in Japan for about 40 years was
introduced. Although the bearing capacity estimating equation of the pile hardly changed in the
meantime, it is needed to be changed.
Then, feature of the bearing capacity estimation equation of a pile used in the port facility
construction standard of Japan is compared with which used in Europe and the United States. On
these standards, a big difference is in the estimation method of a toe bearing capacity. But it is
difficult to evaluate the superiority or inferiority of bearing capacity equations only comparing
the measured results of the total bearing capacity of open ended piles with calculation results.
Then, the loading test of the pile of the large diameter which embedded to large depth was done.
The toe bearing capacity of the pile was evaluated by dividing into two, the toe resistance of a real
part and the resistance by inner surface friction. The resistance of a real part is being able to
estimated by the proposals already presented. For the inner surface friction, that working in the
section of ID to 2D from the bottom can be taken into consideration. The toe bearing capacity of
an open ended pile can be evaluated in this way.
S.R. Lo
University of New South Wales, ADFA Campus, Canberra, Australia
SUMMARY
The use of probabilistic method in geotechnical engineering has been advocated by many
researchers, probably following the pioneering publications of Lumb (1966, 1967, 1968)
since the 1960s. Reliability Based Design (RBD) is a current hot-topic in geotechnics,
either as a design methodology in its own right or for underpinning Limit state Design
(LSD) codes. The variances of soil parameters are needed for any form of reliability
analysis, simplified or complicated. The determination of this simple statistical
parameter for real engineering problems, often considered as pedestrian in reliability
research, can be challenging. The author have the experience of coming across
unrealistic "reliability predictions" because of overlooking a number of essential issues (of
practical and theoretical origins) that affect the variance of soil parameters. Having
variances that are representative of the actual engineering problem is analogous to having
representative soil parameters for any meaningful deterministic modelling. This paper
consolidates these issues and use the title of "common misconceptions" in order to
stimulate discussions.
Var{x}= Var{m}+<T2T2(L)
where Var{.) denotes variance of a random variable and X is the spatially-averaged soil property
estimated by the sample mean. The first term on the right hand side of the above equation is
associated with the sampling uncertainty while the second term is associated with the spatial
variability of the soil property. The first term is affected by the number of soil measurements and
will approach a negligible value if the number of measurements are adequate large. In most
geotechnical designs, soil data is usually limited and the contribution of sampling is normally
significant.
Conversion uncertainties
The soil parameter relevant for a given design model may be calculated from a linear conversion
of the measured soil property. For example, a conversion factor of 0.7 may be applied to the
undrained cohesion of triaxial compression to "derive" the undrained cohesion relevant to a
particular design scenario. This is referred to as derived value in EC-7. Irrespective of the
reason behind the conversion, this will introduced additional variability. The methodology of
dealing with conversion uncertainties is addressed.
58
TAIPEI2006 International Symposium on New Generation Design Codes
for Geotechnical Engineering Practice
Nov. 2-3, 2006, Taipei, Taiwan
SUMMARY
Concerns often arise during braced excavations due to the resulting excavation-induced ground
movements. When buildings are in close proximity to the excavation, building damage (referred to herein
as the violation of serviceability requirements) may occur as a result of the excavation-induced ground
movements. The application of reliability analysis can provide a valuable assessment of the building
damage potential in a braced excavation.
The purpose of this paper is to establish a framework for the simplified reliability analysis of
excavation-induced building damage potential through the development of an unbiased limit state. The
Bayesian mapping functions are established empirically based on the distributions of principal strains in a
database of cases compiled and examined by Son and Cording (2003), and subsequently used to calibrate
the model uncertainty (jici and COVci) of the assumed limit state. Based on the calibration of limit state
model, an unbiased limit state (jtci = 1.00) is established.
To perform the initial reliability analysis, the cases in the database collected by Son and Cording are
split into two categories, tolerable (undamaged) and intolerable (damaged), based on the classification of
damage by Burland where any damage levels less than or equal to "Slight damage" are considered to be
undamaged.
In addition, the limit state is established with the principal strain, a value derived from angular
distortion and lateral strain, as the critical criterion. Previously, Son and Cording correlated the damage
levels with principal strain and the limiting principal strain is empirically assigned based on classification
levels. Any principal strain associated with a damage level greater than "Slight damage" is considered
intolerable, and thus, the limiting principal strain is assumed to be 1.67xl0"3. The limit state with a
loading principal strain (epL) is subsequently equated as:
In a deterministic analysis, LS < 0 would indicate that building damage will occur.
In order to evaluate the model uncertainty and establish an unbiased limit state equation, the
probability of damage for a given principal strain must be determined. This probability may be interpreted
based on the distributions of the principal strains of the groups of intolerable and tolerable cases using
Bayes' Theorem. Using the method developed by Juang et al. (2000), the following mapping function
relating EP to PD can be established:
P(s
K pl
\D)P(D)
PD = P(D | ejp = ' K ' (2)
' P{£p\D)P(D) + P(Ep\ND)P{ND)
Using the method developed by et al. (2004), the model bias factor (pcl and COVcj) is calibrated until
the probability of damage calculated from the limit state (Equation 3) matches the probability of damage
calculated from the Bayesian mapping functions. Based on the calibration with each of the 41 mapping
functions (and thus, the results of 41 sets of [icl and COVci values), the mean values and standard
deviations of JXCI and COVcl are determined for the initial limit state: fici = 0.71 with apcl = 0.03 and COVci
- 0.33 with Bcovci ~ 0.
In addition, the effect of the limiting principal strain is examined by redefining the limiting principal
strain so that any damage greater than "Very Slight" is considered intolerable (limiting principal strain is
equal to 0.75x10"3). With the new limit state, the model uncertainty is recalibrated, and the mean values
and standard deviations of /xci and COVci respectively are determined: /xcl = 1.59 with apci = 0.07 and
COVcI = 0.33 with Ocovd ~ 0. It should be emphasized that 0.71 x (1.67 x 10"3) for the initial limit state
and 1.59x(0.75xl0" 3 )forthe new limit state equate to the same value of 1.19xl0~ 3 . This implies that
the limiting principal strain of the unbiased limit state should be equal to 1.19 x 10~3.
To more accurately formulate the serviceability limit state, the limiting principal strain is recalibrated
using a trial-and-error procedure so that the mean model uncertainty is unbiased ( ficl = 1.00). From this
analysis, the limiting principal strain is determined to b e l . l 9 x l 0 ~ 3 , and the mean values and standard
deviations of picl and COVc, respectively are determined: fici = 1.00 with apcl = 0.04 and COVct = 0.33 with
ocovd ~ 0. The uncertainty of the model bias can be combined and is equated to o c / = 0.37. Therefore,
the unbiased serviceability limit state can be expressed as:
I S = A ( c , , f p I ) = C l ( 1 . 1 9 x l 0 - 3 ) - e p , i =0 (4)
where the model uncertainty, Ci, is characterized as /xcl =1.00 and cct' =0.37.
In summary, using the database collected by Son and Cording, the framework for a
probabilistic analysis of excavation-induced building damage has been developed. The Po-ep
Bayesian mapping functions have been established based on the distributions of principal strains
for the group of intolerable cases and the group of tolerable cases. The mapping functions
provide a basis for calibrating the uncertainty of the limit state model. Based on an extensive
calibration process, the limiting principal strain has been determined to be i . i 9 x i o - 3 for an
unbiased limit state with the model uncertainty, in terms of bias factor, characterized as fici - 1.00
and c c i' = 0.37. Reliability analysis of the excavation-induced building damage potential can be
analyzed using the unbiased limit state as expressed in Equation 4.
60
Session IV
— * * ® « * —
Geohazards
TAIPEI2006 International Symposium on New Generation Design Codes
for Geotechnical Engineering Practice
Nov. 2—3, 2006, Taipei, Taiwan
S.H. Yu, S.D. Yang, C.F. Wang, C.H. Chiao, L.T. Hwang
Taipower Company, Taiwan
SUMMARY
Ta-Chia river is the one of abundant water resource in central Taiwan, and there are seven branch
hydro power plants of TPC (Taipower Company). After Chi-Chi earthquake took place, the
follow-up typhoons also caused damages along Ta-Chia river. The disastrous typhoons for Ta-
Chia river were Toraji typhoon in 2001, Mindulle typhoon and Aere typhoon in 2004, and Haitang
typhoon in 2005, which triggered new landslides and the debris flows were flushed into riverbed,
the events caused the river channel silted up and the flood level raised. Those geohazards
destroyed most infrastructures and villages nearby the river, especially the hydropower facility. In
order to assess the impact of sediment yields from landslides and debris flows and to investigate
the strategies of mitigating the geohazards, quantitative assessment was conducted by using aerial
photos and satellite images obtained at 6 stages of major earthquake and typhoon events. In order
not only to estimate the volume of the sediment yields from landslides and debris flows, but also
to establish the relationships between the volumes of sediment yields, the rainfalls intensity, and
the discharge.
0 1000 2000 3000 4000 5000 6000 7O00 8000 9000 10000 11O00 12000
Distance from Kukuan Reservoir (m)
The elevation changes from Chi-Chi earthquake to Hi-Tang typhoon(1999-2005).
64
TAIPEI2006 International Symposium on New Generation Design Codes
for Geotechnical Engineering Practice
Nov. 2-3, 2006, Taipei, Taiwan
Hung-Jiun Liao
National Taiwan University of Science and Technology, Taipei, Taiwan
SUMMARY
The total length of roads with elevation above 100m in Taiwan is more than 67,000 km. Some of
them were built with high engineering standards, but a large number of them were built with low engi-
neering standards. Therefore, landslides in different failure types are not unusual along mountain roads
when the slopes are experiencing long period of rainfalls or torrential rain accompanied with typhoons.
In this study, Route T-18 in central Taiwan is chosen to demonstrate the suitability of landslide predic-
tion using Gaussian Process model. Two main questions are of concern: (a) Given the historical land-
slide data along the demonstrative mountain roads in Taiwan, where are the locations along the roads
with high landslide potential in the future? (b) Given the historical landslide and rainfall data, what are
the landslide probabilities of the slopes along the roads in a future heavy rainfall? The former mainly
concerns with the locations of future landslides, while the latter concerns with the time of landslide
occurrence in future rainfalls.
Landslides along the mileage between 21.5km and 83.5km of Route T-18 are documented. In
total, 55 failed unprotected slopes along T-18 were extracted from the road maintenance records.
Among them, 12 slopes failed during Typhoon Herb, 18 during Toraji, 9 during Nari, and 16 during
Mindulle due to heavy rainfalls. To match the number of failed slopes, 54 not-failed unprotected
slopes were chosen. Note that the not-failed slopes in the database are roughly uniformly distributed
over the chosen Route T-18 section. The data format is as follows: D = {(*,-,/,•): i = 1 109}, where
Xj£ R15 contains the values of the 15 landslide features of the i-th slope in the database; r,- = 1 if that
slope failed, otherwise U = 0; p is the total number of slopes in the database. Fifteen landslide features
are categorized into natural features and man-made features. Among them, thirteen are natural fea-
tures, and two are man-made features. The natural features cover the aspects of topography (4 features:
slope direction, slope angle, slope height, and road curvature), geologic conditions (1 feature: outcrop
strata age), bedrock structure (2 features: slope and dip direction difference and slope and dip angle
difference), weathering & fracturing (2 features: block size and rock volume percentage), vegetation
cover (2 features: area percentage of vegetative cover and thickness of canopy cover), drainage condi-
tion (1 feature: catchment area size), and seismicity (1 feature: peak ground acceleration). The man-
made features (2 features: excavation height and change of slope grade due to toe cutting) quantify the
impacts induced by road construction.
A single index is used to capture the landslide potential: P(t=lbc,D), i.e. the probability that t = 1
given x and D. The Gaussian Process analysis is implemented to estimate the landslide poten-
tial. The discriminant function analysis is also implemented to compare with the Gaussian
Process analysis. Common practice of examining the performance of the adopted model/analysis is to
quantify the so-called training errors. For fair calculation of prediction errors, the so-called leave-one-
out (LOO) prediction errors of the adopted model are adopted here. The LOO prediction error is an
unbiased estimate of prediction error on unseen slopes of the trained model. The basic idea of LOO
prediction error is to mimic the prediction process by removing one data point out of the training da-
taset and use the removed data point for prediction testing.
Table 1 shows the traditional training error rates and the LOO prediction error rates induced by
the Gaussian Process analysis and the discriminant function analysis. Note that both training errors are
quite small (one of them is zero), but these are not realistic estimates for the actual prediction errors on
Table 1 Training errors and LOO prediction errors for the landslide location analysis
Training # of False LOO Predictions LOO
Methodology
errors (out of 109) Prediction Error Rate
Discriminant Function Analysis 7.3% 13 11.9%
Gaussian Process Analysis 0% 6 5.5%
The analysis shows that slope height, catchment area, height of toe cutting, block size, and
change of slope angle are the dominant features, among them are the two man-made features. This re-
sult implies that the slope stability along mountain roads is noticeably affected by road construction.
Also note that catchment area is among the dominant features. This result is consistent with the sense
that slope stability should be sensitive to the amount of seepage and surface water.
Besides predicting potential landslide locations, it is desirable to predict "when" (i.e. during
which typhoon) the dangerous slopes will fail. In Taiwan, landslides are mostly triggered by heavy
rainfalls during typhoons. As a consequence, we propose a second stage of analysis (landslide poten-
tial is the first stage) to predict the occurrence times of landslides for the dangerous slopes. In the oc-
currence time analysis, the size of catchment area and "effective" rainfall amount are the two features
studied. The latter is treated as the triggering feature of landslides. The goal is to develop a methodol-
ogy that determines the relationship between landslide probability and the two features for a dangerous
slope given past landslide and rainfall data. This relationship is called the rainfall fragility graph.
Figure 1 shows the results of the Gaussian Process analysis and is called the rainfall fragility
graph. In the figure, the crosses "+" indicate the failed dangerous slopes in the database (43 data
points), while the circles "o" are the not-failed dangerous slopes (33 data points). Note that in the
lower-left region, i.e. small rainfall and small catchment area, most slopes did not fail, while most
slopes failed in the upper-right region, i.e. large rainfall and large catchment area. This observation
agrees with our intuition. The contour values indicate the value of the predicted landslide probability
P(t=l\x,D) by the Gaussian Process analysis based on 43 failed dangerous slopes and 33 not-failed
dangerous slopes. This rainfall fragility graph can be used to predict landslide probability of a danger-
ous slope due to future typhoon.
Figure 1 Rainfall fragility graph for dangerous slopes along Route T-18
66
TAIPEI2006 International Symposium on New Generation Design Codes
for Geotechnical Engineering Practice
Nov. 2—3, 2006, Taipei, Taiwan
SUMMARY
In this paper, a fully probabilistic method for liquefaction evaluation is first proposed, in which
both the uncertainties associated with the earthquake loading and the cyclic strength of soils are
considered. Then, a liquefaction hazard model is established for computing the total cost of a
building including the probable liquefaction loss during its service life. They are combined to
form a decision framework to determine the optimal design solution of ground improvement
method. A case of a damaged building in Wufeng, Taichung County that caused by soil
liquefaction in the Chi-Chi earthquake was chosen to demonstrate the feasibility of the proposed
methodology. The results show that the proposed methodology is workable and provides a logical
and reasonable way to design the ground improvement for preventing the liquefaction hazard.
Case Study
The proposed liquefaction hazard analysis model is similar to that suggested by Nishimura and
Shimizu (2005) with omitting the constant initial construction cost of the structure.
The case used for the study is a residential district, named the Prince's Castle, where many
buildings were damaged by soil liquefaction during the 1999 Chi-Chi earthquake. The Prince's
Castle residential district is in Wufeng, Taichung County and surrounded by the Kehniaokengchi,
which is a 5 meters wide creek. The representative geological profile of the site is shown in Table.
Based on the proposed fully probabilistic method for analyzing the liquefaction potential, the
annual exceedance probability (AEP) of liquefaction potential index PL and post-liquefaction
0.1 0.15
Replacement Ratio, r,
Figure. Optimal analysis of sand compaction pile
Conclusion
This paper proposes a fully probabilistic method for analyzing liquefaction potential of a ground and a
liquefaction hazard model for structures. They are combined to form a decision framework to determine
the optimal design solution of ground improvement method. From the case study, the feasibility of the
proposed methodology has been demonstrated.
68
TAIPEI2006 International Symposium on New Generation Design Codes
for Geotechnical Engineering Practice.
Nov. 2-3, 2006, Taipei, Taiwan
M.J. Hsieh
Harbor and Marine Technology Center, Institute of Transportation, Taichung, Taiwan
W.J. Chang
Department of Civil Engineering, National Chi Nan University, Nantou, Taiwan
P.S. Lin
Department of Civil Engineering, National Chung-Hsing University, Taichung, Taiwan
SUMMARY
REFERENCES
Lai, S.Y., Lin, P.S., Hsieh, M.J., and Jim, H.F. (2005), " Regression Model for Evaluating
Liquefaction Potential by Discriminant Analysis of the SPT N value." Canadian Geotechnical
Journal. Vol. 42, No. 3, p.856-875.
O'/.SFCSlO'/. V*0X
0.5
C(B=0 / P^0.1i
0 0.4
* ^*/ ,
'15
% 0.3 P-005
P=0JU
t °2
ju^yXy4C4.
0.1 ~W%^ 9o°
• Liquefaction
0^s^ ONonliquefactbn
Fig.l Discriminant curves with 0%£ FCs 10% Fig.3 Physical meanings of the developed models
0.6
pc=m
0 5 10 15 20 25 30 35 40 45 50
Conected Blow Count, (Ni)u
Fig. 4 The result of liquefaction potential
Fig. 2 Comparison of discriminant curves with fines evaluation by Discriminant's model at
content of 5%, 15%, 25% and 35% for C(P)=0 Taichung Harbor area
70
TAIPEI2006 International Symposium on New Generation Design Codes
for Geotechnical Engineering Practice
Nov. 2—3, 2006, Taipei, Taiwan
SUMMARY
Development of reasonable seismic verification and design of the countermeasure for existing
structures, such as buildings, bridges, dams and etc., is an important issue in Japan because the
occurrence of big earthquakes in the near future has been predicted. A type of canal, which is built on
embankment, Figure 1 shows typical transverse section of the canal, is also one of the structures
required the reasonable seismic treatment from the viewpoints as follows;
there are many canals of such type in Japan
considerable effects of the canal damage during the earthquake on human life and social
economy are concerned
expensive seismic countermeasures will be needed for almost all the canal if they are verified
with respect to the present seismic design specification, satisfying safety factor, 1.0, of land
slide analysis, thus tremendous amounts of money and time will be required for the project
Therefore, a study on the proposal of quantitative seismic performance based design for the canal as
a reasonable treatment of the issue has been conducted.
In this paper, contents of the proposed seismic performance based design code, seismic
performance, limit states, quantitative criteria for verification, and prediction methods of seismic
response of the canal embankment, are introduced.
Study on the proposal of quantitative seismic performance based design for the canal
In order to verify the canal embankment quantitatively to satisfy seismic performance of the canal,
both quantitative criteria for verification and prediction method of seismic response of the
embankment have to be stipulated along with seismic performance and its corresponding limit states.
Therefore, firstly, seismic performance, limit states and criteria for verification of the canal
embankment are newly proposed from the viewpoints of soil strain for safety and reparability of the
embankment, and subsidence of top of the embankment for serviceability of the canal. And then, the
prediction method that can reproduce well actual seismic response of embankments was proposed
from the comparison with centrifuge tests. Furthermore, because simple prediction method is valuable
for practical design, applicability of several prediction methods, which are the method using static
elastic FEM, the method based on Newmark proposed method (N.M. Newmark, 1965) and more
simple methods adopting charts or equations as well as dynamic elasto-plastic FEM analysis, are
studied.
[Design of seismic]/
[ countermeasure y
Figure 2 Procedure of seismic verification and design of the countermeasure for canal embankment
72
Session V
Engineering Practice
and Challenges
TAIPEI2006 International Symposium on New Generation Design Codes
for Geotechnical Engineering Practice
Nov. 2-3, 2006, Taipei, Taiwan
J.A. Frame
Gammon Construction, Hong Kong
D. Sein
Lambeth Associates, Hong Kong
SUMMARY
This paper presents the author's experiences of applying Observational Design (OD) approaches to
substructure design and construction in the South East Asia. To illustrate the potential benefits that can
be achieved, a case history is described for a deep basement excavation in Singapore. The design
objectives, methodology and performance data are presented.
There is an ever-growing push to gain further economies in the construction industry. Construction
projects, particularly in congested urban areas, are becoming increasingly complex but with shorter
expected construction periods. At the same time regulators and the public are increasingly intolerant to
any form of risk. The OD approach can provide a means by which construction safety and efficiency
are improved for deep excavations in the urban environment.
Modern instrumentation has become more accurate and reliable and can provide real time
automated monitoring which directly benefits the processes in OD. Instrumentation data loggers can
remotely connect to Internet based instrumentation databases, which makes data accessible and
transparent to all stakeholders in a project.
The construction industry is also undergoing change in the way contracts are procured, which has
lead to greater interest in OD. Design and build contracts and greater emphasis on partnering amongst
stakeholders has made OD more viable compared to the adversarial nature of historical construction
contracts.
The excavation at a site in Singapore Central Business District involved a 20m deep basement
structure. The base design required 5 layers of struts and imposed constraints on the phasing of the
construction of the pile caps. Objectives of the OD were to delete the lowest level of struts and avoid
the phasing constraints.
The OD design approach involved a series of sensitivity analyses covering a range of design
parameters so that the predicted behaviour could be benchmarked against the measured performance
on site. The range of design parameters covered "most probable soil parameters" and "original design
soil parameters".
A monitoring control framework was developed to track the predicted against measured wall
deflections during basement excavation. A "traffic light" system was implemented which reflected the
status of the wall deflections at different excavation depths (green as below predicted response, amber
as alert and red as action). A similar traffic light system was used in the control framework for the strut
loads.
Decisions to proceed with deletion of struts or implement contingency measures were made at
predefined stages of the excavation based on the monitoring records. A decision flowchart was
included on the construction drawings to clearly identify the key decisions.
All of the stated objectives of the OD were achieved on the project resulting in safety, cost and
programme benefits. Site safety benefits included reduced handling of heavy steel struts and
providing more working space for manoeuvring excavation equipment. Valuable information on
the importance of temperature on strut forces was also obtained. There was a daily fluctuation in
strut force of about 25% recorded.
76
TAIPEI2006 International Symposium on New Generation Design Codes
for Geotechnical Engineering Practice
Nov. 2-3, 2006, Taipei, Taiwan
SUMMARY
Piles are normally designed to resist axial compression and uplift forces for most of the building
foundations. However, for deep foundations of structures such as towers, bridge abutments,
jetties, mooring dolphins, high-rises etc, the piles are often subject to significant lateral loads in
addition to the axial loads. The induced lateral deflection and bending moment not only will affect
the design pile capacity but also the performance of the structure. The ultimate lateral resistance
of pile depends very much on the subsoil surrounding the pile. It can be evaluated using
approximate solution as proposed by Broms (1965). Extensive researches on the deformation of a
single pile subject to lateral load had also been carried out and some analytical techniques have
been developed too. In general, the analytical model can be grouped into two main categories: the
Elastic Continuum model and the Spring Idealisation soil model (Winkler soil model or p - y
curve method). Both models had been used extensively today for the prediction of pile
deformation when subject to lateral load.
A recent development on a tidal land required pile foundation not only to resist axial load but
also to resist lateral load. The site originally was submersed at most of the time and the ground
surface level was at about 0.75 mLSD (Land Survey Datum). Reclamation had been carried out by
using hydraulic sandfill. The design platform level for the proposed development is about 4
mLSD. Due o the thick soft clay, long term post construction settlement was expected. In order to
expedite the minimise the post construction consolidation settlement of the soft marine clay,
surcharge method with the installation of prefabricated vertical drains had been adopted and more
than 90% consolidation has been completed at the time of pile tests. Soil investigation carried out
at the test piles locations revealed that the subsoil composed of 7.5m thick of sand below the
design platform level. Underlying is the treated soft clay with thickness of about 18m. Liquid
Limit and Plasticity Index of the soft clay vary from 80% to 120% and 40% to 80% respectively.
A thin layer of loose silty sand was found below the treated soft clay layer follow by medium stiff
to hard soil layer.
In order to evaluate the pile performance when loaded laterally, in-situ full-scale lateral load
tests were planned and conducted on two single piles. Test pile LTP-1 is a 600 mm diameter pre-
stressed concrete spun pile with thickness of 100 mm. The concrete strength is about 78.5 MPa.
The pile was installed using hydraulic hammer from the design platform level of 4 mLSD. The
penetration depth is about 30 m, i.e pile toe level at -26 mLSD. Inclinometer of 70 mm outer
diameter was installed into the spun pile and grouted. Test pile LTP-2 is a 750mm diameter cast-
in-situ bored pile. The pile was constructed from the design platform level to 34m depth, i.e. pile
toe at about -30 mLSD. After the boring depth had reached to the design depth, a steel cage
consisted of 10 numbers of high yield strength steel reinforcement of 32mm diameter was
installed. A steel pipe of 150mm diameter was welded to the steel cage for the installation of
inclinometer in later stage. The concrete strength is 30 MPa and it was placed by tremie pipe
method. Inclinometer with outer diameter of 70mm was installed into the pre-installed 150mm
diameter steel pipe. The annular space between the steel pipe and the inclinometer was filled with
cement grout. For both test piles, about 2m excavation was carried out for the setting up of the
test. The applied load was at the level of about 2 mLSD.
Reaction piles with similar diameter to the test piles were adopted as the reaction system. The
test load apply to the test pile was using a hydraulic jack together with a load cell. The pile head
deformation was measured by dial gauges or transducers. In addition to the dial gauges or
transducers, inclinometer was installed in the test pile to measure the pile deflection at different
depths.
78
TAIPEI2006 International Symposium on New Generation Design Codes
for Geotechnical Engineering Practice
Nov. 2-3, 2006, Taipei, Taiwan
SUMMARY
With the development and utilization of the underground space, large-area frame structure with
thick raft foundation under a single or multiple high-rise buildings is widely used in China.
Settlement calculation method for large-area thick raft foundation under regular high-rise
buildings has been already proved feasible and reasonable in studies and in-situ settlement
monitoring in recent years. In fact, with the variety and complexity of high-rise buildings, large-
area frame structure with thick raft foundation under irregular multi-tall buildings is increasing,
for this foundation form, existing approach cannot accurately calculate the settlement of the whole
range of the raft. To solve the above-mentioned problem, based on the mechanical concept of
large-area thick raft foundation, a deformation analysis method is developed in this work, which
can be employed in the settlement calculation of large-area thick raft foundation with irregular
multi-tall buildings, and it has been successfully applied in the projects of China Petroleum
Mansion and Beijing HengFu Garden Mansion.
Practical application
Optimized designs using the aforementioned method for the large-area thick raft foundation of
China Petroleum Mansion, which is of four high-rise buildings of L-shape, and Beijing HengFu
Garden Mansion, which is of single high-rise building of Z-shape, have been carried out, and in-
situ settlement monitoring has been made during the construction stage. In comparison with
theoretical calculation, measured settlements agree well with numerical results. From engineering
applications, it is demonstrated that thick raft are capable to effectively spread the intensive high-
rise loading outwards to podium area, as a result to adjust the differential settlement in between.
In comparison with flexible raft/plate, rigid thick raft can significantly reduce the unacceptable
differential settlement occurred normally at the location of 1 to 2 span outside tower perimeter.
The concept of adopting thick raft to replace the conventional delay poured strip is proved to be
feasible and practical.
Conclusion
Under the interactions of superstructures, raft and foundation soil, the measured deformation of
entire raft is irregular and continuous, the rigidity of raft foundation is limited, and the raft
foundation under high-rise building is still limited rigidity. Based on the 'rectangular cutting and
superimposition' method, the settlement of irregular high-rise building can be calculated by elastic
theory. The settlement of large-area thick raft foundation under multiple irregular high-rise
buildings can be calculated based on elastic theory and superimposition method under
serviceability limit state. It is necessary to extend the thick raft of irregular high-rise building
outside one-span podium to decrease the additional stress under the foundation of high-rise
building, and to increase the foundation's stability of irregular high-rise building.
80
TAIPEI2006 International Symposium on New Generation Design Codes
for Geotechnical Engineering Practice
Nov. 2-3, 2006, Taipei, Taiwan
H. Y. Chuay
Mott MacDonald Ltd, Kaohsiung City, Taiwan
SUMMARY
Kaohsiung is the largest city in southern Taiwan as well as the economic and political centre. To
fulfil the need of fast development of the city, the construction has started since 2000 for the
Phase 1 of Kaohsiung mass rapid transit system. The system is expected to be in full operation
from 2007. There are two lines in Phase 1, Red Line in N-S direction and Orange Line in E-W
direction. A risk assessment was carried out in the design stage for the deep excavation of a
cofferdam without any lateral strutting system, 140 m in diameter and 27 m deep, in silty sand, in
order to ensure the construction safety and programme of the only interchange station in the Red
& Orange lines of Kaohsiung mass rapid transit system (KMRTS). The main risks associated
with this type of excavation are unbalanced ground water pressure and soil pressure outside the
cofferdam before the completion of the concrete structure, which are critical to the development
of the arching effect.
A risk assessment was carried out for the excavation at O5/R10 station and it identifies that
possible risks during the cofferdam excavation include: (1) excavation error, (2) over/under-
digging, (3) collapse of trench wall, (4) design load on ground level, (5) ground conditions, (6)
groundwater level, (7) direction and velocity of groundwater, (8) temperature, (9) earthquake, (10)
structure buckling, (11) field measurement, (12) water pumping and draw-down of groundwater,
(13) uplift failure and (14) leakage of diaphragm wall.
After the identification of hazards, a performance-based evaluation of the excavation is
considered. Based on the field measurements, a deflection path of O5/R10 was made and also be
compared with reference envelope determined from excavations having different ground
conditions and wall thicknesses in Taipei. It was found that the ratio of maximum lateral wall
movement (5hmax) to the excavation depth (D) varies from 0.07% to 0.27% at O5/R10 and this
ratio is much smaller for excavations in Taipei with D greater than 4m. This might be connected
with a thicker wall, different ground conditions and excavation shape at O5/R10.
The observational method was considered to be applied in the project. In the observation
method, the design is reviewed from time to time during the construction in response to the
monitored performance of the structure. It was suggested that the observation method should be
taken for a project where a precise prediction of the geotechnical behaviour is difficult.
The management of geotechnical works at O5/R10 is reviewed and it was recommended that
the advantage and flexibility associated with a turnkey contract should be properly considered in
geotechnical works. In addition, having one consultant to carry out independent checking
services might reduce the risk in geotechnical design successfully.
The efficiency of using circular cofferdam excavation is explored in this study. It was found
that the use of circular cofferdam excavation could dramatically reduce the construction cost,
even though it increased design complexity.
82
TAIPEI2006 International Symposium on New Generation Design Codes
for Geotechnical Engineering Practice
Nov. 2-3, 2006, Taipei, Taiwan
K. Aita
Sato Kogyo Co., Ltd., Tokyo, Japan
T. Fujiyama
Central Research Institute of Electric Power Industry, Abiko, Japan
M. Honda
Nikken Sekkei Civil Engineering Ltd., Tokyo, Japan
T. Kaneko
Fukken Co., Ltd., Hiroshima, Japan
A. Morikage
Chubu Chishitsu Co., Ltd., Kanazawa, Japan
A. Murakami
Okayama University, Okayama, Japan
M. Nabetani
Sato Kogyo Co., Ltd., Tokyo, Japan
M. Nozu
Fudo Construction Co., Ltd., Tokyo, Japan
SUMMARY
Technical committee on the performance-based design for geotechnical structures in Japan Society of
Civil Engineers was founded in 2004 and has actively exchanged the ideas among the administrators,
practitioners and academia. Working group 2 (WG2) of this committee has collected and investigated
many verification methods used for previous practices to establish a new methodology of verification
methods in the context of the performance-based design.
We should notice that the preciseness of calculated results depends on not only a verification
method itself but also other data qualities such as boundary conditions, initial conditions and modeled
geometrical configurations. We discuss on the potential abilities of each verification methods with the
unified framework. However, a verification method itself is only one factor to a solution. As a nature
of a geotechnical problem is that of an initial boundary value problem, we can point out three major
factors of a problem. One factor is expressed as a term " material" which includes a constitutive
model and its material parameters. Another factor is expressed as a term "ground profile" including
the constitution of soil strata and initial conditions of ground. External forces such as seismic motions,
wind forces and tidal forces are also included in ground profile. The other factor is expressed as a term
"solver" which describes how to formulate and solve a problem.
In a performance-based design procedure, choice of verification method is one of the key issues for
the reliability of an obtained result. From the view point of cognitive science, three typical behavior
can be observed as pointed out by Rasmussen (1986) shown in figure 1.; i.e., skill-based behavior
(SBB), rule-based behavior (RBB) and knowledge-based behavior (KBB). This model is
convenient for our research purpose to consider qualitative classification of verification method. By
Goal
Knowledeg-based
level
Rule-based level
Skill-based level
Figure 2. Matrix expression of three levels of modes and three factors in geotechnical design
84
Session VI
—<&&*&*—
Geotechnical Uncertainties
and Variabilities
TAIPEI2006 International Symposium on New Generation Design Codes
for Geotechnical Engineering Practice
Nov. 2-3, 2006, Taipei, Taiwan
Jianye Ching
National Taiwan University of Science and Technology, Taipei, Taiwan
Yi-Hung Hsieh
National Taiwan University of Science and Technology, Taipei, Taiwan
SUMMARY
Uncertainties are abundant in geotechnical engineering. It is sometimes the case that the amount of
uncertainties associated with geotechnical systems is so significant that the resulting failure probability
is quite large. How to reduce the uncertainties in geotechnical systems can be an important research
topic. This research focuses on the following subject: how to reduce uncertainties and update reliability
by using monitoring data.
In this research, a new method is proposed to update reliability of general systems without the
dimensionality and linearity constraints. This new method is based on Bayesian analysis and Monte
Carlo simulation (MCS), and update reliability of an instrumented system by using its monitoring data.
In fact, as long as the probability distribution of the uncertainties is given, the functional relationship
between the updated failure probability and the monitoring value can be obtained prior to the monitoring
process. This means in real applications, it is not necessary to conduct the new algorithm in an online
manner. Instead, the relationship can be calculated a priori so that the reliability update can be achieved
right away once the monitoring data is obtained.
According to the Bayes' rule, we know
s f(<p\F)P(F)
P(F\(p)=J^\) K
> (1)
JVP)
where F is the failure event; q> is the monitoring value; P{F]ip) is the failure probability given that the
actual monitoring value is f. In fact, P(F\<p) is the updated failure probability;fi<p\F)is the probability
density function (PDF) of the monitoring value conditioned on the failure event; P{F) is the failure
probability without the monitoring information; aadfifp) is the marginal PDF of the monitoring value.
For our purpose, the goal is to find P(F\<p). According to (1), iff(tp\F), P(F) and fly) are all available,
P(F]tp) can be readily obtained.
The failure probability P(F) can be estimated with MCS. Let Z denote the uncertain variables of
the target system. First, N sets of Z samples {Z*' : i = 1.. .N} are drawn from the prescribed PDF of Z.
According to the Law of Large Number, we have
1
* ;=1
where R denotes the limit-state function that defines failure event F, i.e. failure event is defined as
R[Z]>\.
Please note that in the process of MCS, samples distributed asf{(p\F) snAfij^F0) can be obtained
(F° denotes the non-failure event): Corresponding to the N sample sets {Z*' : /' = I...N) are the JV
samples of the monitoring value {#>*' : i = 1.. .JV }. Assuming that among the N samples, there are NF
failure samples, i.e. samples satisfying J?[Z"]>1, so the corresponding <p samples are distributed &sj{(p\F).
Please note that in the optimization problem, the variable is the entire g(f) function, where g(cp) can be
either ftjp\F) or/[^IF c ); L and U are the upper and lower bounds, respectively, of the monitoring value <p
(if they exist, otherwise, they are -co and oo); the quantity -lUi}og(g(<p)). g(<p)d<p is the entropy of g(<p), and
Hi, k=\..A are the first four sample moments. Furthermore, j\q>) can be estimated with the following
equation:
20
Monitoring value (m) Days
Figure 1. An infinite slope. The left-hand-side plot is the estimated P(F\<p) function. The upper-right plot
is the monitoring value, while the lower-right plot is the resulting failure probability time history.
88
TA1PEI2006 International Symposium on New Generation Design Codes
for Geotechnical Engineering Practice
Nov. 2-3, 2006, Taipei, Taiwan
Limin Zhang
The Hong Kong University of Science and Technology
SUMMARY
It is well recognized that the geotechnical properties of soil and rock show considerable spatial
characteristics in all directions, though in different degrees. In this paper the spatial variability of
a weathered rock at a site in Hong Kong is quantified using the statistical models based on random
field theory and geostatistics. The borehole information available at the site is used in the
analyses. Depth of Grade-III weathered rock at a construction site is an important indicator in
design to evaluate the preliminary founding depth of the bored piles. The theory of geostatistics is
made use of to enhance the available subsoil information. Additional data at unsampled locations
are obtained using the spatial interpolation technique, called Kriging. The spatial variability
characteristics of Grade-III rock in the horizontal direction are evaluated in terms of scale of
fluctuation, using random field theory. The advantage of getting such information is that it helps
the engineer in charge of nearby sites or sites of similar geological origin for an effective and
efficient planning of the future site investigation. An effort has also been made in this study to
evaluate the scale effect of sampling size on the autocorrelation characteristics of Grade-III
profile. The results clearly demonstrate that the scale of fluctuation increases with the sampling
domain; the highest value is obtained for the data from the whole site and the least value when the
borehole information from only one building block is considered.
In general, the Grade-III weathered rock is deemed to support the loads of the structure
effectively by producing distortions that are within permissible limits as specified in the codal
provisions. Hence, for the preliminary designs, depth of Grade-III weathered rock is an important
reference for the determination of bored pile lengths.
Similar studies have been reported for the evaluation of spatial characteristics for soil and rock
properties for various applications using either geostatistics or random field modelling by many
researchers over past 3 decades (Vanmarcke 1977, Kulatilake 1989, De Groot and Baecher 1993,
De Groot 1996, Fenton 1999, Jaksa et al. 1999, Murakami et al. 2006, Liu and Chen 2006).
The whole site is divided into 7 blocks for site exploration purpose, which are referred to as
blocks 1 to 7. Figure 1 shows the locations of borehole data, which are irregularly spaced over the
areas of each block and whole site. The spatial correlation characteristics are evaluated for three
data sets with varying sampling size, as shown in Table 1. The first data set contains the depth of
Grade-III information obtained from the boreholes within block 1. The second data set contains
the depth of Grade-III information from the boreholes within blocks 1, 2, and 3 together.
Similarly, the Grade-III data obtained from all the boreholes within the whole site comprise the
third data set.
A clear trend is observed in the depth of Grade-III profile for this site, and it is found that the
slope of the Grade-III surface is as high as 1V:3H.
E- 1000
a>
13
c
S. 950
o
o
o
V,
iS 900
Table 1. Size of exploration area and number of boreholes for individual blocks and the whole site
Size of exploration area No. of boreholes
Whole site 160.2 m x 252.4 m 174
Blocks 1,2, and 3 together 122 m x 137.8 m 87
Block 1 alone 55 m x 59.3 m 49
CONCLUSIONS
Depth of Grade-Ill weathered rock is an important indicator used in Hong Kong design practice to
evaluate the preliminary founding depth of bored piles.
To ease the evaluation of semivariogram using irregularly spaced data, the separation distance
of all pairs of data are grouped into different classes of 1 m interval, and semivariance
corresponding to each class is calculated considering all the pairs of data grouped within that
class. This approximation could result in slightly inappropriate experimental semivariogram, and
might have affected the reliability of the predicted depths of Grade-Ill at unsampled locations.
N/4 pairs of data are chosen as minimum number for evaluating the experimental
semivariogram reliably, where 'N' is the number of measured data points. The results obtained
from the study reveal that the maximum lag of N/4 for constructing the experimental
semivariogram suggested in case of regularly spaced data may not be effective when the data is
randomly spaced.
The detrending process reduces the sill and range of semivariogram. The range obtained using
the observed data (without trend removal) from the whole site is many scales higher than the
sampling size. This demonstrates that the data set is non-stationary and has to be de-trended.
The horizontal scale of fluctuation of depth of Grade-Ill surface using the linearly detrended
data from the whole site is 137 m. This value is well comparable with horizontal scales of
fluctuation of some of the geotechnical parameters available in the literature.
The horizontal scale of fluctuation of depth of Grade-Ill depends on the sample size. Higher
scale of fluctuation is obtained from the analysis when the data from the whole site is used, as
shown in Table 2.
Table 2. Horizontal autocorrelation distances and scales of fluctuation of depth of Grade-Ill
Block 1 Blocks 1,2, and 3 Whole site
Size of exploration area 55 m x 59.3 m 122 m x 137.8 m 160.2 m x 252.4 m
Correlation distance (m) 13.5 45.4 77.3
Scale of fluctuation (m) 24.0 80.5 137
90
TAIPEI2006 International Symposium on New Generation Design Codes
for Geotechnical Engineering Practice
Nov. 2-3, 2006, Taipei, Taiwan
K.K. Phoon
National University of Singapore, Singapore
M. De Wet
University of Stellenbosch, Department of Civil Engineering
J.V. Retief
University of Stellenbosch, Department of Civil Engineering
ABSTRACT
Database
A database of 172 compression pile load tests was compiled in this study. The test loads were
generally carried out in accordance with SABS 1200F (1983) reaching a maximum load of 1.5 to 3
times the design load. In both cohesive and non-cohesive materials, the available geotechnical data
was limited to: soil profiles, SPT, Dynamic Probe Super Heavy (DPSH), and CPT (very few). For
evaluation of geotechnical properties, the DPSH and CPT measurements were converted to equivalent
SPT values. The geotechnical properties required for the calculation of ultimate capacity were then
obtained from correlations with SPT N-values.
The collected data was processed and used to determine the predicted capacity and the
interpreted capacity. The predicted ultimate capacity was obtained from the basic static
formulae while the measured capacity was interpreted from load settlement curves. The model
factor for each case was calculated as the ratio of the interpreted capacity to predicted
capacity, thereby generating a model factor dataset of 172 individual values. A statistical analysis of
the model factor realisations was carried out to reduce the data to manageable form as well as to
facilitate the extrapolation of the results beyond the current database.
Graphical display
Histograms for the model factors in cohesionless materials, cohesive materials and all soil types were
constructed. Visual examination of the histograms showed similar characteristics for the three cases.
These characteristics were: most of the data points were clustered around the mean value; the
histograms were unimodal, i.e. they had one point of concentration or a single peak indicating that the
Although graphical presentation condenses a set of data for easy visual comprehension of its general
characteristics, numerical sample characteristics are required for calculations, statistical testing, and
inferring the population parameters. These are quantities used to describe the salient features of the
sample. The key statistics are the mean (jtM) and the standard deviation (<rM). These key statistics
together with the coefficient of variation (COV) are given in table 1. In addition to the measure of
centrality and dispersion, in this study the mean and standard deviation of the model factor are
considered as indicators of the accuracy and precision of the predication method. An accurate and
precise method gives HM = 1 and OM = 0 respectively, which means that for each pile case, the
predicted pile capacity equals the measured capacity (an ideal case). However, due to uncertainties the
results of an ideal case can not be attained in practice. Therefore in reality, the method is better when
fiM is close to 1 and aM is close to 0. When /JM > 1, the predicted capacity is less than the interpreted
capacity, which is conservative and safe. When HM < 1, the predicted capacity is greater than the
interpreted capacity, which is unconservative and unsafe.
The correlation between the bias factor and the various pile design parameters was accessed on the
basis of the p-value associated with a null hypothesis of zero rank correlation. The results of the
correlation analysis indicated that:
• Cohesive materials: The bias factor was not significantly correlated to pile diameter, pile length,
and undrained shear strength for both driven and bored piles. However, the model factor appears
to be correlated to the predicted capacity in driven piles but not in bored piles.
• Cohesionless materials: The bias factor was not significantly correlated to pile length. However, it
was significantly correlated to the pile base diameter for both driven and bored piles; angle of
friction in driven piles but not bored piles; the predicted capacity in both driven and bored piles.
A lognormal distribution was assumed to be the most appropriate theoretical distribution for the model
factor on the basis of theoretical and historical considerations. This assumption was verified using
goodness of fit statistical tests. When a random variable X is lognormally distributed, its natural
logarithm, ln(X) is normally distributed. Accordingly the natural logarithms of bias factors were
determined and some normality tests supported by the available software were applied to the resulting
data set. The normality tests included Kolmogorov-Smirnov (K-S) test, Lilliefors test and Shapiro-
Wilk's W test. The P-values for all the three tests are greater than 0.05 and therefore there is no
92
evidence to reject the null hypothesis of normal distribution for the logarithms of bias factors of all the
cases considered. Since the distribution of the natural logarithms of the bias factors is normal, it
follows that the distribution of the bias factors is lognormal.
Conclusions
It has been shown that the model factor varies with individual observation and therefore it is best
modelled as a random variable. In this regard the lognormal distribution was found to be the most
appropriate theoretical model for the model factor. It was also established that the bias factor was
significantly correlated to pile design parameters in non-cohesive materials while such correlation in
cohesive materials was not significant.
93
TAIPEI2006 International Symposium on New Generation Design Codes
for Geotechnical Engineering Practice
Nov. 2-3, 2006, Taipei, Taiwan
SUMMARY
We describe how to draft a formula for estimating the side resistance of a pile by reviewing a great
number of in situ pile loading tests conducted in the past. Design values of the unit side resistance by
an average SVT-N value (fj/N) provided as estimation formulae in the Specifications were determined
by selecting respective peak values of unit side resistances in each layer from the relation of the unit
side resistance, /, and displacement, S, during a test. Consequently, it has been pointed out that
estimated side resistances using fd /N tend to overestimate the resistances observed in the tests.
Furthermore, the magnitude of the safety margin excluding the safety factor in the current
Specifications is unclear.
We examine two SVT-N value normalized unit side resistances in each layer: (1) fy/N at the yield
point of a P0-S0 curve and (2)fp/N at the peak side resistances at pile tops based on the in situ pile test
results. To determine the yield point of the P0-S0 curve, a Weibull curve was used here. The resistances
observed in the test piles based on past tests have collected, in which SPP means driven steel piles and
CCP means cast-in-place bored piles. These tests all had measurement devices for dividing the
resistance into the side and base, although their loading sizes were different.
Figure 1 shows the relations of total resistance, Pa, at the pile top, base resistance, Pp, and side
resistance Pf - displacement S0 at the pile top for pile A-4, in which dotted lines mean yield points of
the P0-S0 curves and chain lines mean peak values of side resistances. This figure reveals the
following: (1) peak side resistances are mobilized at the displacement of around 2 to 5% of the pile
diameter, (2) base and total resistances increase in proportion to the increased load at the pile top with
displacement of more than 10% of the pile diameter, and (3) total resistances tend to yield at around
the points where the peak side resistances are fully mobilized.
Figure 2 show the relations of unit side resistance by normalized by SVT-N value (f/N) - pile top
displacement normalized by pile diameter (S0 /D) in each layer for pile A-4, in which yield points
(dotted lines) of P0-S0 curves and peak values of side resistances (chain lines) are also shown. In this
figure, the symbols As, Ac, Ds, and Dc mean alluvium sand, alluvium clay, diluvium sand and
diluvium clay, respectively. As shown in these figures, we can obtain the yield point (fy/D) but it is
hard to find the peak value (fp/D) of the side resistances compared to fy/D.
We examined the reliability of two SVT-N value normalized unit side resistances in each layer, (1)
fy/N at the yield point of a Po-S0 curve and (2)fp/N at the peak side resistance by using the reliability
index, (3. Statistical values offy/D and fp/D for each pile type and ground condition are shown in Table
1, in which 95% reliability lower levels of mean values (hereafter, characteristic values) from the mean
values are also provided. Figures 3 show fy /D and fp /D histograms of pile types SPP based on the
values provided in Table 1. In this figure, sample mean values of ;t^, D and x^^ with full lines and
characteristic values of fi ID and ll^,ID with dotted lines are also marked. With regard to the reduction
ratio of the characteristic to mean values, those of fp /D are greater than those of / , /D because the
number of the former is remarkably smaller than that of the latter.
In order to estimate the reliability of the resistances fy/D andfp/D, we first set up (fd/D)d = 0.667 by
using current design values fa/D = 2 and a safety factor (= 3 for the present situation) in the case of
SPP sand, for example. The safety margin can be estimated with the reliability index, P, considering
the characteristic values and variation of fy/D,fp/D and (fd/D)d-
1200 Po
1000
p 800
P
(kN) 600 / - • —
'
400
// P„
200
1/ 2.0 10.6
Figure 1. Total, base and side resistances of pile A-4 Figure 2. Unit side resistances of pile A-4
*IWN —
= 2.)1
y
•
Xfp/N
=5. i72
| MtL
6 8 10 12 14
•• •5 10 15 20 25 30
fy/N fp/N
96
TAIPEI2006 International Symposium on New Generation Design Codes
for Geotechnical Engineering Practice
Nov. 2-3, 2006, Taipei, Taiwan
ABSTRACT
This paper proposes a new practical method, which will be adopted in the annex of the Japanese
design code of port facilities, to evaluate soil parameters for performance based design. The
characteristic value of the soil parameter is the expected value in principle, but it is not the mere
average in consideration of statistical errors in association with the testing method, the
inhomogeneity of the soil itself, and the limited number of the test data. The new method uses a
correction factor corresponding to the coefficient of variation.
3 Correction factor
In order to evaluate the variation of the derived values a, it is convenient to use the standard
deviation (SD) of a/a*, where a* is the estimated parameter obtained by modeling the depth
profile of the derived values. Note here that the profile of a* must be estimated as either the
expected constant value or the regression function to minimize the errors. This is the major
premise based on Table 1, in which the characteristic value is generally defined as the expected
value of the derived values. The SD of a/a* is equivalent to the COV of a.
It has been known that the COV of the derived values obtained by soil tests carried out
carefully by a well skilled technician becomes less than 0.1 (Watabe et al., 2004). In other words,
the variation in this level is inevitable because of the ground heterogeneity and the laboratory test
errors. Ground heterogeneity, sample disturbance, inappropriate soil test, bad modeling of the
depth profile, and so on result a larger COV. In such cases, it is reasonable to conservatively
determine the characteristic value taking account of the uncertainties.
In order to calculate the characteristic value at from the estimated value a*, correction factor b\
is introduced as a function of the COV of a, then a k is defined as b\Xa*. When the soil parameter
a contributes to either resistance in the performance verification (e.g. the shear strength in the
stability analysis) or safety margin in the prediction (e.g. the consolidation yield stress pc; the
coefficient of consolidation cv in the consolidation calculation), the correction factor is defined as
bi-l-(COV/2). On the other hand, when it contributes to either sliding force in the performance
verification (e.g. the unit weight of the earth fill in the stability analysis) or safety dearth in the
prediction (the compression index Cc; coefficient of volume compressibility mv in the
consolidation calculation), the correction factor is defined as b\ = 1 +(COV/2). In these definitions,
the characteristic values correspond to either 30% or 70% fractyle value. Because a simplified
method is aimed in this proposal, the values listed in Table 1 are to be used instead of the
correction factors with detailed fractions. When the COV is larger than 0.6, it is judged that the
reliability of the soil parameter is too low for design. In this case, the test results are reexamined;
i.e. the depth profile is remodeled if necessary. In some cases, the ground investigation may be
reexecuted.
As mentioned above, some soil parameters, such as pc; cv; mv are examined on the logarithmic
axis. It is, thus, reasonable to evaluate the variation with the logarithmic normal distribution. For
the soil parameter a, the SD of (log a)/(log a*) is the COV of log a. Using this relationship, the
values listed in Table 1 can be used as the correction factor b\ on the logarithmic axis. In this
case, since ak = a* b'' is obtained from log ak = b\ log a* = log a* b\ the correction factor b\
corresponds to the exponent on the arithmetic scale.
If the number of the data is not sufficient for statistic treatment, another correction factor b2 is
introduced to correct the b\. Since about 10 or more data in the depth profile concerned can be
thought sufficient to reliably calculate the COV. In the case with less than 10 data, when the soil
parameter contributes to either resistance in the performance verification or safety margin in the
prediction, the correction factor is defined as bi = 1 - (0.5 / n). On the other hand, when it
contributes to either sliding force in the performance verification or safety dearth in the
prediction, the correction factor is defined as b2= 1 +(0.5/n). Here, b2 for the case with only one
data is set to 0.5 or 1.5, respectively, and the reliability is assumed to rapidly increase with the
data number. In this regard, however, the correction factor b\ cannot be obtained in the case of n =
1, because the COV cannot be calculated. This indicates that more than two data are required in
this proposed method. Note here that b\ = 1 and b2 = 1 can be used for soil parameters that
contribute equivalently to both action and counteraction.
98
Taiwan Special
Project Series
TAIPEI2006 International Symposium on New Generation Design Codes
for Geotechnical Engineering Practice
Nov. 2-3, 2006, Taipei, Taiwan
SUMMARY
The capacities and behavior of a cost-in-place reinforced concrete pile is largely determined by the
way it is installed. The design of the pile foundation for Taipei 101 was based on a series of full scale
pile trial installation and a series of comprehensive instrumented pile load tests with downward and
pull-out loads up to 4000 tons and 2200 tons, respectively.
The load transfer within the pile and the characteristic t-z curves for each soil strata were evaluated
and used in the pile length determination for each of the 380 piles, in accordance with the soil
stratification each pile is located and the maximum anticipated load to be acted on the pile during
service.
Base on results of pile trial installation and load tests, pile installation specifications, including
equipments to be used and installation procedures were finalized, and were strictly enforced under the
full time supervision of geotechnical engineers.
The structural design of the foundation take into consideration not only the pile-mat system, but
also the 5 levels of basements and the surrounding slurry wall retaining structures. The combined
effect on, and the behavior of the foundation system under loadings from the tower and the podium
were also investigated and evaluated.
The estimation of the foundation behavior under various load combinations were conducted using
the above mentioned structure model and the estimated distribution of sub-grade reactions under the
foundation mat. The estimated distribution of sub-grade reactions takes into consideration the
settlements of both the compression of each pile under load and the compression of the bearing
soil/rock strata under pile group loads. The high and low ground water conditions were also considered
in the foundation behavior estimates.
SUMMARY
The Taiwan High Speed Rail Project is considered one of the largest BOT (Build-Operate-Transfer)
projects in the world with an estimated construction cost of US$ 15 billions. The THSR runs through
populated west coast of the Taiwan Island, and its 345-km guideway links major cities from Taipei to
the southern city of Kaohsiung. The maximum design speed of the THSR is 350 km/hr with a capacity
of carrying 300,000 passengers per day, the Rail is expected to contribute significantly in alleviating
the overcrowded traffic of the west corridor. In this paper, an overview of the THSR Project is first
provided, and then important design requirements and considerations are discussed, which include
design requirement, seismic effect, near/crossing fault effect, ground subsidence and foundation
scouring etc. This is followed by discussions of load test programs implemented during the
installation of piles. Finally, several issues related to design and analysis of foundation systems were
discussed from a retrospective perspective. Extensive technical matters regarding bored piles have
been learned through this project.
The Taiwan Island is located on the western edge of the Pacific-rim earthquake belt and sits over
the juncture of the Eurasian and Philippine Sea Plates. Complex geological structures of the island
were formed by the active tectonic activity, which also created active faults islandwide. As a result,
seismic forces and earthquake related effects are major concerns for the design of the THSR. The
geological conditions along the route vary substantially, having mountainous terrain in the north to
thick sedimentary deposits in the south. As a result, the northern section is constructed mostly by cut-
and-fill, with bridges and tunnels, while the southern half is mostly elevated supported by large bored
piles. More than 30,000 piles were installed because of the presence of large earthquake forces, thick
alluvium soil deposit, and limited right of way. Due to significant variations in ground conditions,
different construction methods also were adopted for the installation of piles along the THSR route.
The design of the THSR structures followed the basic concept of performance-based design, in
which different performance requirements were considered for different earthquake levels. However,
the performance requirements were not implemented in an explicit form of performance matrix. Table
1 attempts to summarize the loosely specified design requirements in the context of the performance
matrix. Although the basic performance-based design was attempted in the THSR design of civil
works, some redundancy may still exist because performance requirements and verification methods
were not addressed in a systematic manner. With experiences of the THSR design, it seems important
considering the essence of performance-based design concept and making critical review on current
practice.
Given the large number of piles to be constructed, there is a great benefit to optimize the pile
design. The full-scale pile load tests were executed to determine the actual shearing resistance of local
soils. The test piles were instrumented with rebar stress transducers installed at different depths. The
results of the pile load tests in compression, tension and lateral direction were evaluated. Some
correlations between the unit skin friction and the SPT N values were established for various types of
soils.
In this series of pile load tests, the construction methods of bored pile installation were improved to
a great extent through the use of proper stabilizing agent (polymer), shorter construction time, multi-
stage toe grouting etc. Based on some of the available results, the ratio of ultimate tension capacities
to compression capacities (without end bearing) ranges from 62 to 105%, with a mean of 81±16 %.
Lateral pile load tests were conducted at locations with different soil conditions for determining
horizontal modulus of subsoil reaction and to verify the design parameters. Test results revealed that
the responses of test piles are quite similar to those predicted by the "p-y" curve method; hence this
method was adopted by some of the designers for predicting response of piles in design
With the benefit of hindsight, it can be concluded that improved foundation engineering may be
achieved with more elaborated studies of the load test data. The project also revealed the need for
sound and realistic design concepts. Performance-based design with due considerations of explicit
reliability levels should be the framework for implementation in future geotechnical seismic design.
The geotechnical community is way behind the structural community in this subject area. It calls for
serious efforts in conducting relevant investigations and developments such that a harmonized design
on both structural and geotechnical aspects may be achieved.
104
TAIPEI2006 International Symposium on New Generation Design Codes
for Geotechnical Engineering Practice
Nov. 2-3, 2006, Taipei, Taiwan
Y. S. Hsieh
RSEA Engineering Co., Taiwan
SUMMARY
The main structures of Hsuehshan Tunnel are composed of one pilot tunnel and two main tunnels
(southbound and northbound). These 12.9 km long tunnels are the longest traffic tunnel in Taiwan.
There are three ventilation stations each including twin shafts for fresh and exhaust air
interchanging respectively, their depths vary from 250 m to 500 m. The experiences from the
above tunnelling works are discussed. These tunnels go through the Hsuehshan Ranges in the
northern Taiwan. There are 6 major faults and 98 shear zones in the field of construction site. The
maximum width of these weak zones is about 80 m. The tunnels were bored by tunnel boring
machines (TBM) from the south portals due to the environment constrains of the Feitsui
Reservoir.
The difficult problems during construction mainly came from TBM excavation stuck in the
Szeleng Sandstone. Difficulties occurred in the excavation of the pilot tunnel, main tunnel and
ventilation shafts are described. The rock mass properties of the Szeleng Sandstone can be
categorized as high strength, joint well developed to highly fractured with many faults and shear
zones, and with large quantities of groundwater behind them. The distributions of the geological
structure and groundwater are so complicated that they cannot be exactly predicted before
excavation, especially for the irregular high or low angle of the shear zone that left great
quantities of groundwater behind. These interactive effects create a lot of problems in probing
from excavation faces and TBM workings. The difficulties encountered include collapses of rock
masses, obstructions of borings and TBM excavations, wears and breakages of probing bits and
TBM disc cutters, cut and indent of casings and shields, and huge inflows of pressurized water
discharge..
The strategies for solving the problems proposed by the team are described as the follows:
a. Top heading method
By using this protection method, the southbound TBM worked well in the Szeleng Sandstone
and then make a greater excavation rate after passing that difficult ground condition.
b. Bypass tunnel for rescuing stuck TBM
The bypass tunnel whose excavation area almost same as the pilot tunnel was successfully used
to solve the TBM stuck problems.
c. Prolongation of tunnelling from northern portal
The excavation length from the north portal working face is more than 1/3 both in pilot and
main tunnel. This plan really increased the excavation rate.
d. Additional tunnelling faces from ventilation shaft No. 2
The total length excavated by these additional working faces was about 5,454 m. It actually
promoted the construction rate for this Project.
e. Raise boring applied for shaft No. 1
By using this sophisticated directional drilling system, it took about 5 months to complete the
reaming work of the fresh air shafts and 2 months for exhaust air shaft.
f. Application of long distant boring
It is worth to note that long-distance probing investigations are very good for the TBM
excavation, especially in the difficult geological condition.
g. Reformation of boring machine on the pilot TBM
106
AUTHOR INDEX
Aita, K. 83 Hsiung, B. B. C. 81
Askew, I. 75 Hsiung, K. P. 49
Basak, S. 43 Hu,Q. F. 15
Becker, D. E. 3 Huang, H. W. 15
Chang, C. F. 49 Huang, X. L. 79
Chang, M. H. 49 Hwang, J. H. 67
Chang, W.J. 69 Hwang, L. T. 63
Chang, Y. L. 63 Iizuka, A. 13
Chen,C. S. 77 Ito, Y. 71
Chen, D. D. S. 101 Juang, C. H. 59
Chen, J. R. 103 Kaneko, T. 83
Chen,T. R. 53 Kasai, Y 71
Cheng, C. T. 31,63 Kato, T. 71
Chern, J. C. 63 Kikuchi, Y. 55,97
Cheung, W. M. 35 Kim, H. 39
Chiao, C. H. 63 Kobayashi, S. 83
Chin, C. T. 103 Kokusho.T. 17
Ching,J. Y 53,65,87 Ku,C. Y 63
Chiou, S. J. 31,63 Kung,G. T. C. 59
Chu,F. L. F. 45 Lai, S. Y 69
Chuay,H. Y 81 Lam, J. 33
Chung, C.J. 21 Lee, C. T. 31
Dasaka, S. M. 89 Lee, K. 39
deWet, M. 91 Li,K. S. 33,57
Di,D. H. 79 Liao,H.J. 65
Dithinde, M. 91 Lin, H. D. 19
Duann, S. W. 103 Lin, M. L. 21
Einstein, H. H. 5 Lin, P. S. 31,69
Frame, J. A. 75 Lin, Y S. 63
Frank, R. 27 Lo, S. R. 33,57
Fujiyama, T. 83 Lu, C. C. 67
Gong, J. F. 79 Lu, W. 39
Hara,T. 71 Lui, L. S. 35
Ho, M. H. 21 Matsui, K. 95
Honda, M. 47,83 Morikage, A. 83
Honjo, Y. 7,13,47 Murakami, A. 83
Horikawa, T. 71 Nabetani, M. 83
Horikoshi, K. 13 Nagao, T. 39
Hosoyamada, M. 71 Nakatani, S. 95
Hsiao, E. C. L. 59 Nozu, M. 83
Hsieh,M. J. 69 Ogawa, K. 47
Hsieh, Y H. 87 Orr, T. L. L. 9
Hsieh,Y S. 105 Otake, Y 71
107
Phoon, K. K. 23,91 Tsai, Y. B. 31
Pun, W. K. 35 Vogt,N. 37
Retief, J. V. 91 Wakatuki, Y. 47
Roth, M.J. S. 59 Wang, C. F. 63
Schuppener, B. 27,37 Wang, C. H. 49
Schuster, M. J. 59 Watabe, Y 97
Sein, D. 75 Wu, D. C. 49
Shi, L. P. 105 Yang, C. W. 67
Shirato, M. 95 Yang, S. D. 63
Shu, S.M. 63 Yang, Y. Y. 15
Su,T. C. 103 Yoon, G. 39
Suzuki, M. 95 Yu, S. H. 63
Tanaka, M. 97 Zhang, L. M. 45,89
108
Listing of ISSMGE TC23 proceedings and other related publications
Orr T. L. L. & Farrell E. R., Geotechnical Design to Eurocode 7, Springer, London, 1999,
166 pp., ISBN 1852330384.
Honjo, Y, Kusakabe, O., Matsui, K., Kouda, M. & Pokharel, G. (Editors), Proceedings,
International Workshop on Foundation Design Codes and Soil Investigation in View of
International Harmonization and Performance Based Design (IWS Kamakura 2002),
Tokyo, Japan, 10-12 April 2002. Balkema 2002, 459 pp., ISBN 90-5809-381-6.
Lin, M. L., Chin, C. T, Lin, H. D., Honjo, Y. & Phoon, K. K. (Editors), Proceedings
of the International Symposium on New Generation Design Codes — Taipei 2006 (With
CD-ROM), Taipei, Taiwan, 2-3 November 2006. World Scientific 2006, 122 pp., ISBN
981-270-382-9 (pbk).
109
Proceedings of the International Symposium on
New Generation Design Codes for
Geotechnical Engineering Practice -Taipei 2006
(with CD-ROM)
Circular patterns symbolized the ground and the water conserved in it. The harmony
between these elements represents the spirit of fullness — an ideal pursued by the
Chinese. The rectangles correspond to buildings, highlighting the goal towards more a
631 Osc
ISBN 981 -270-382-9 (pbk)
i 9 8 i - 2 o ~o—6 www.worldscientitic.com