You are on page 1of 218

INTERNATIONAL SOCIETY FOR SOILMECHANICS

AND GEOTECHNICAL ENGINEERING


TC 40.

FORENSIC GEOTECHNICAL ENGINEERING


OCTOBER 2009
INTERNATIONAL SOCIETY FOR SOIL MECHANICS AND
GEOTECHNICAL ENGINEERING

FORENSIC GEOTECHNICAL ENGINEERING

EDITED BY

Dr.V.V.S.Rao

CHAIR 2005-2009
TC40: FORENSIC GEOTECHNICAL ENGINEERING
OCTOBER 2009
FORENSIC GEOTECHNICAL ENGINEERING

CONTENTS

PREFACE
1. INTRODUCTION TO FORENSIC GEOTECHNICAL ENGINEERING V.V.S.RAO
2. WHAT IS FAILURE PETER DAY
3. CHARACTERIZATION OF DISTRESS ANIRUDHAN,I.V.
4. COLLECTION OF DATA PETER DAY
5. COMPILATION OF DATA V.V.S.RAO
6. LABORATORY TESTS ROBINSON,R.G
7. BACK ANALYSIS SIVAKUMAR BABU
8. BACK ANALYSIS OF SLOPE FAILURES POPESCU,M.E.et.al
9. BACK ANALYSIS IN FGE RICHARD N. HWANG
10 GEOTECHNICAL FAILURE:THE CAUSE JAN HELLINGS
11. INSTRUMENTATION AND MONITORING IWASAKI,Y.
12. LEGAL PROCESS AND JURISPRUDENCE SAX SAKSENA
13. POTENTIAL ROLE OF RELIABILITY PHOON,K.K,et.al.

CASE HISTORIES:
1. TURNING HINDSIGHT INTO FORESIGHT MIKE MARLEY
2. FAILURE IN CONSTRUCTION OF AN UNDERGROUND STATION RICHARD HWANG
3. FORENSIC ENGINEERING FOR UNDERGROUND CONSTRUCTION BROWN,E.T.
4. IMPORTANCE OF UNDERSTANDING LANDFORMS THOMPSON,R.P.
5. MONITORING IN FORENSIC GEOTECHNICAL ENGINEERING IWASAKI,Y.
6. EXPERTS’ DILEMMA DIRK LUGAR
7 R.E.WALL DISTRESS SIVAKUMAR BABU
8. DISTRESS TO REINFORCED EARTH WALL EMBANKMENT SITHARAM,T.G, etal
9. ANALYSIS OF SOIL NAILED STRUCTURES RAJAGOPAL,K.,etal.
10. FAILURE ANALYSIS OF DAMAGED RE WALL MURTHY,B.R.S.
11. APPROACH EMBANKMENT ON SOFT GROUND MADHAV,M.R.,et al.
12 COLLAPSE OF SOILNAILED WALL SITARAM,T.G.
13. THE CASE OF TILTED CHIMNEY RAO,V.V.S.
14. INFLUENCE OF VIBRATIONS SANTOSH RAO, N.

ANNEXURE.
PREFACE

Forensic Geotechnical Engineering (FGE) deals with the procedure to be followed while analyzing
a distress / failure in a structure which is attributed to geotechnical origin, not only from technical,
but also from legal and contractual viewpoints. In cases of remedied installations the analysis and
evaluation of adopted remedial measures may be subjected to legal scrutiny with regard to their
effectiveness and economy. Geotechnical based distress in structures due to natural hazards
including seismic damages also come under this purview. The commonly adopted standard
procedures of testing, analysis, design, and construction may not be adequate for forensic analysis.
The test parameters and design assumptions must simulate the actual conditions encountered at
site. Thus, in the forensic investigations, every micro aspect of the design, construction and
maintenance actions are studied in detail to analyze what, when, how, and why something went
wrong and more importantly, who is responsible for it. This procedure not only assists in
litigations, but also helps in improving the standards of geotechnical aspects of a project.

In order to develop this subject, Prof. Pedro Seco e.Pinto, President, ISSMGE established during
February 2006 a Technical Committee (TC40) on Forensic Geotechnical Engineering. The
composition of the committee is given in the annexure. The committee proposed to prepare a book
on FGE as a guide to geotechnical professionals.

The present book highlights the aspects of a forensic investigation illustrated by a number of case
studies. This is a first attempt and more detailed one has to follow.

While compiling the chapters, the papers presented in the workshop on FGE conducted in
Bangalore on 12th Sept.2009 have been included. This w/s was organized by the Indian Institute of
Science in collaboration with TC 40 and Karnataka Chapter of IGS.
Chapter 1.

INTRODUCTION TO FORENSIC GEOTECHNICAL ENGINEERING

Dr.V.V.S.Rao
Nagadi Consultants Pvt. Ltd., New Delhi, India,
dr_rao@nagadi.co.in

INTRODUCTION

Forensic analysis in geotechnical engineering involves scientific and legalistic investigations and deductions to
detect the causes as well as the process of distress in a structure, which are attributed to geotechnical origin.
Cases of remedied installations where the analysis and evaluation of adopted remedial measures with regard to
their effectiveness and economy may be subjected to judicial scrutiny also fall under this purview. The normally
adopted standard procedures of testing, analysis, design and construction are not adequate for the forensic
analysis in majority of cases. The test parameters and design assumptions will have to be representative of the
actual conditions encountered at site. The forensic geotechnical engineer (who is different than the expert
witness) should be able to justify the selection of these parameters in a court of law. Hence he has to be not only
thorough in his field of specialization, but should also be familiar with legal procedures. This paper presents
principles of planning and executing a forensic investigation

SCOPE

While investigating any distress, the engineer should meticulously follow a well planned programme. The scope
of work would broadly be under the following heads:
a. Compulsory tasks
i. survey and documentation of the distress
ii. scrutiny of all design documents including design criteria chosen
iii. review results of original geotechnical investigations, their analysis, and selection of
design parameters.
iv. study the field reports of construction
v. interview persons involved in planning, design, construction and
performance monitoring, etc.
b. Optional tasks:
i. perform additional investigations
ii. develop and conduct special tests
iii. non-destructive testing of structural element
c. Analyze all data and evaluate
i. the distress history
ii. causes of distress
iii. identify the shortcomings in the original investigation and analysis
d. Report
i. authority and scope
ii. history
iii. summary of original documents
iv. data collected
v. interviews
vi. meteorological information
vii. earthquake
viii. investigations performed , their methodology and their results
ix. analysis
x. conclusions.
TYPES OF DISTRESS

The visual distress which commonly occurs in normal structures and the most probable causes are listed in the
following table

Structure Visual distress Causes

Buildings and bridges Cracks Structural


Geotechnical: settlements
heaving
uplift
vibrations
Tilts Differential settlements
Earthquakes
Collapse Excessive loading
Soil erosion
Repetitive loading, fatigue

Retaining structures Lateral movement Inadequate base resistance


Tilting Differential settlement between toe
and heel
Excessive surcharge on the backfill
Excessive water pressure due to
poor drainage
Slopes Excessive settlements Improper compaction
Slope failure Settlement of virgin strata
Longitudinal cracks Erosion due to water, rainfall
Inadequate drainage

Apart from the above causes, most important causes would be inadequate and/or inappropriate soil
investigation, selection of design parameters and use of inappropriate theories

DIAGNOSTIC TESTS.

After identifying the cause of distress, the following questions arise :


a. Has the distress fully occurred and if not, how much more can be expected? Quantify.
b. What were the precise causes for distress?
c. Whether the soil underwent same stress-deformation history as was anticipated? If not, what was
the actual history at site?
d. The effect and efficacy of remedial measures on the soil+structure behaviour.

To answer these questions, detailed tests have to be conducted both in the field and in the Laboratory. The
choice of tests will normally be from among the following tests depending upon the problem.

A. Field tests
i. Borehole investigations including SPTs to a depth deeper than the influence zone
ii. Cone penetration tests
iii. Load tests
iv. Special tests like, pressuremeter tests, vane shear tests, seismic or dynamic tests
B. Laboratory tests
i. Triaxial shear tests; to simulate the actual field conditions, these tests should be done on stress
increment basis on partially saturated sample. The effect of fluctuation in degree of saturation
on the deformation behaviour of the soil should also be investigated. In case of clays, the field
stress history is also to be considered
ii. Repeated cyclic shear tests, in cases like water towers, bridges,etc.
iii. Large deformation tests, to assess the residual strength and magnitude of final deformations
in cases of slopes,etc..
iv. Compaction tests with different compaction energies
v. Permeability tests

In case of residual and wind deposited soils, the effect of change of soil structure due to loads, both static and
moving ones

In all cases it is advisable to conduct regular borehole investigations and evaluate the sub-soil profile.

ANALYSIS

After collecting all data detailed analysis can be done to evaluate the design parameters. Use of empirical
relationships should be avoided, unless their validity in the particular site is established. The analysis should be
based on:

- limit conditions
- partial factors of safety
- equilibrium state vis-à-vis flow state
- liquefaction potential
- critical void ratio in compacted fills

With these design parameters the load – deformation history of the soil+structure combine can be reconstructed.
This process will lead to identification of the causes of distress. A suitable and economically viable remedy can
then follow.

LEGAL ISSUES

In the entire process of investigation, the forensic engineer should be careful to ensure that all the experimental
and analytical procedures as well as the selected parameters for tests and analysis fully conform to the field
conditions. The report should be comprehensive and intelligible to a legal person also. It is advisable to avoid
“hi-tech” terminology and strong verbs like-should, must, etc. As far as possible, it is better to avoid too many
details in the main text. At the same time, the report should have sufficient details for the client to give a
comprehensive brief to the executing agency.

One should realize that the association of the engineer with the project is based on the principle of-“contract of
skill”. Hence the consultant should ensure competent and reliable advice to the client. It is also imperative for
the consultant to explicitly detail the risks that might be involved or expected in using the conclusions and
recommendations.

The consultant should be aware of the importance and implications of the following guidelines (ref: Guidelines
for the provision of Geotechnical Information in Construction Contracts- The Institution of Engineers,
Australia, 1987.):

Facts: These have to be true and should not be erroneous.


- exploration locations
- samples and cores available for inspection
- lithological descriptions of soils and rocks
- measured water tables
- test results

Interpretations: the skill of the engineer is judged here


- borelogs
- inferred stratigraphy between boreholes
- properties of various layers for use in the analysis
- seismic interpretations yielding velocity and layer depths

Opinion: may or may not be disputable


- assumptions
- judgement based on facts and interpretations

Negligence: obviously, very serious


- performance of investigations
- description and analysis of information
- communication
- accuracy, in general

Overall, it is emphasized that application of standard of skill and care is expected from a professional,
irrespective of quantum of remuneration. However, the liability of the consultant is also limited as the owner
pays for the “skill and service” and not for “insurance”.

REFERENCES

Leonards, Gerald A., “Investigation of failures”, Journal of Geotechnical Engineering Division, ASCE, GT 2,
Feb. 1982.
Task Committee on Guidelines for Failure Investigation, “Guidelines for Failure Investigation”, ASCE,1989.
Green, D.C., “Principles for providing Geotechnical Data in Construction Contracts”, Conference on
Dams, Queenstown, Tasmania, 1988, (also in Ancold Bulletin No. 81.)
Robert W Day, “Forensic Geotechnical and Foundation Engineering” Mc Graw Hill, 1998.
Chapter 2

WHAT IS FAILURE
Peter Day
Jones & Wagener Consulting Engineers,
Livonia, South Africa, 2128
day@jaws.co.za

1. BACKGROUND

Forensic geotechnical engineering involves the application of scientific methods and


engineering principles in the investigation of failures of geotechnical origin, not only
from a technical view point but also with the possibility of legal proceedings in
mind(Rao,2005)

Forensic investigations differ from conventional geotechnical investigations in that


they are retrospective. They seek to explain what has happened rather than to predict
future performance. A further distinguishing factor is that, following a failure, there
is an urgency to clean up the site and rebuild or repair the works. This limits the time
available for investigation and makes it essential that all relevant data is recorded
before the evidence is removed. There is also the added difficult that the ground
conditions may have been altered by the failure and testing of the ground in areas that
have not failed is not always representative.

2. WHAT CONSTITUTES FAILURE ?

2.1 General

When a structure collapses, there can be little doubt that a failure has occurred.
However, legal disputes often arise in cases whee the distress is considerable more
subtle. Leonards(1982), defines failure as an “unacceptable difference between
expected and observed performance”1. This performance may either involve the
stability of the structure, its appearance or its ability to fulfil its intended function in
either the short or long term.
The distress of geotechnical work can be classified in a number of ways ranging from
the type of structure involved, the nature of the distress, the consequences of failure
and many others. However, from a legal perspective, it is preferable to have a
classification that corresponds with the performance requirements laid down by
international standards or codes of practice. Classification of distress in this manner
will assist in determining the acceptability or otherwise of the observed performance
of the structure.

Many modern codes of practice are based on limit states design principles. These
codes clearly define the standards of performance required for various design
situations or limit states. Eurocode En 19902 defines the two main limit states as the
ultimate and serviceability limit states.

1 Leonards,G.A. (1982). “Investigation of Failures.” Journal of the Geotechnical


Engineering Division, ASCE, 108 (GT2):187-246
In broad terms , the ultimate limit state deals with the stability of the structure(or of its
component parts) whereas the serviceability limit states must be satisfied. This firmly
establishes the principle that unsatisfactorily performance of a structure in terms of
serviceability is equally as much a failure as its collapse.

The criteria applied in adjudicating compliance with these limit states differs as
described below.

2.2 Instability

According to EN 1990, the ultimate limit state concerns the safety of people and/or
the safety of the structure. Ultimate limit state failures are generally easy to recognise
as they involve visible collapse or instability of a part or the whole of the structure3.
4
For geotechnical engineering, EN1997-1 (Geotechnical Design) requires
consideration of the following ultimate limit states:

• loss of equilibrium of the structure or the ground, considered as a rigid


body, in which the strengths of structural material and the ground are
insignificant in providing refinance (EQU);

• internal failure or excessive deformation of the structure or structural


elements. Including e.g. footing, piles or basement walls, in which the
strength of structural materials is significant in providing resistance
(STR);

• failure or excessive deformation of the ground, in which the strength of


soil or rock is significant in providing resistance (GEO)

• loss of equilibrium of the structure or the ground due to uplift by water


pressure (buoyancy) or other vertical actions (UPL)

• hydraulic heave , internal erosion and piping in the ground caused by


hydraulic gradients (HYD)

Although the design requirement to be satisfied in the case of each particular ultimate
limit state differs, the basic requirement is that the design action effect (typically the
effect of loads or deformations ) should not exceed the design resistance (typically its
strength).

In the case of abnormal events (such as accidental impact, fire, explosions,


earthquakes, and the consequence of human error), there are additional requirements
for structural integrity and robustness in terms of which the damage to the structure
should not be disproportionate to the original cause. The structure should be capable
of withstanding local damage this causing widespread collapse.
2.3 Serviceability

EN 1990 describes serviceability limit states as those that concern the functioning of
the structure under normal use, the comfort of people and the appearance of the
works.

Verification of the serviceability limits states requires consideration of :

• deformations that affect the appearance the works, the comfort of the users and the
functioning of the structure.

• Vibrations that cause discomfort to people limit the functional effectiveness of the
structure,

• damage that is likely to adversely affect the appearance, durability or functioning of


the structure.

In the case of the serviceability limit state, the criteria to be applied is that the design
action effect must be less that the limiting design value.

It is often considerably more difficult to adjudicate whether a serviceability limit state


has been transgressed than is the case with the ultimate limit state as the assessment of
the serviceability limit state requires the establishment of the performance criteria (or
limiting design values). Although no such criteria are given in the Eurocodes., they
may be specified in the National Annexes or in other national standards. An example
of this is the categorisation of damage to single storey masonry strucutres5 or the
performance requirements and categorisation of damage given in the Australian code
of practice for foundations for residential buildings6.

3. CAUSES OF FAILURE

During a forensic investigation, it is important not to anticipate the outcome of the


investigation by pre-judging the cause of distress or failure. However, the
investigator should be aware of the most common causes of geotechnical failures to
ensure that the data collected is sufficiently comprehensive to enable the problem to
be analysed from all angles.

A study of several published case histories suggests that the most common causes
failure are as given in Table 1. It is often found that failures involve two or more of
the elements given in Table 1 acting in combination. In geotechnical failures, water
or water pressure is one of the most frequent contributing factors.
Table 1 : Most common causes of geotechnical failures

Cause Description

Inadequate geotechnical Budgetary or programme restraints can result in insufficient


investigation investigation being undertaken to adequately model of the
conditions on the site. Alternatively, even the most
comprehensive investigations may fail to reveal critical
conditions that affect the geotechnial behaviour of profile.

Incorrect parameters This can occur for many reasons, including :


- poor sampling and testing procedures
- selection of inappropriate parameters for particular desig
situation (e.g. mean values, lower characteristic values or
upper characteristic value)
- underestimation of variability of soil properties

Inappropriate analysis Failure to recognise the critical failure mechanism , e.g.


model drained v. undrained failure of slopes or foundations,
internal stability v. external stability of reenforced fills.

Underestimation of actions Either the magnitude, distribution or combination of actions


(forces or displacements) incorrectly assessed, particular
load case or combination not considered, use of structure
changed over lifetime.

Unexpected groundwater Changes in ground water levels can increase the loading on
regimes or changes in the structure and decease the shearing resistance of the soil.
moisture content Seepage forces can also have an adverse effect on stability.
Changes in the moisture content of partially saturated soils
can cause softening, heave or collapse settlements.

Sub-standard workmanship Required construction procedures (including sequence and


or materials timing) not followed, specification requirements not met,
inappropriate construction techniques employed, material
properties not in accordance with design assumptions.

Abnormal events not Extreme meteorological events (including temperature,


catered for in design precipitation or wind), accidental impact, errors in
construction or use of structure.

3. CLASSIFICATION OF DISTRESS

4.1 General

Classification of distress should not be confused with the identification of its likely
causes. It is simply a method of classifying the damage to the works prior to any
investigation into its origin. It is simply a case of what has happened to the structure
and not why it has happened.

When viewed in this manner, distress can be classified in terms of two criteria,
namely the manifestation of the distress and its severity.

4.2 Distress of a structure or geotechnical works can manifest itself in two ways, namely
loss of stability and loss of serviceability. These are may be broadly linked to the
ultimate and serviceability limit state requirements of most codes of practice.

Ultimate limit states conditions include the manifestation of distress given in Table 2.

Manifestation Geotechnical examples

Failure of excessive movement of the Bearing capacity failure, slope instability


ground

Failure or excessive movement of the Failure of a retaining wall in bending or


structure or structural component shear, structural failure o fa pile shaft

Loss of equilibrium of the structure or Overturning of a rigid structures or bodily


ground considered as a rigid body translation of a reinforced soil mass

Loss of equilibrium of a structure due to Floatation of a tanked basement structure or


uplift by water pressure buried tank

Hydraulic heave, internal or piping in the Base instability or upward movement inside
ground due to hydraulic gradients supported excavations or caissons,
headward erosion of dam embankments

Serviceability limit state conditions include the manifestations of distress given in Table 3.

Table 3 : Manifestations of distress - serviceability limit state

Manifestation Geotechnical examples

Deformations that affect the appearance of Total and differential settlement of


the works or their functionality foundations causing cracking, jamming of
doors etc.

Vibrations of transient movements that Dynamic response of machine foundations


cause discomfort to people or the function or supporting structures
of structure

Damage that affects the appearance, Water ingress into structures, excessive
durability or functioning of the structure seepage from dams, cracks in piles or
retaining structures that permit corrosion of
reinforcement, etc.
4.3 Severity of Distress

The severity of distress may be classified according to the degree to which the
functionality and stability of the structure is impaired and the ease with which it can be
repaired as indicated in Table 4.

Table 4 : Severity of Distress

Severity Degree of Impairment Remedial work required

Negligible Some visual impairment but Cosmetic only


structure/works fully
functional

Slight Structure/works still Minor non-structural repairs


serviceable but users and redecoration
inconvenienced.
Appearance affected.

Moderate Structures still deemed safe Structural repairs required


but use of structures/works
restricted. Damage clearly
visible.

Severe Structures still standing but Significant structural repairs


no longer serviceable. of partial reconstruction
Access restricted required.

Very severe Structure in state of partial Demolish and rebuild


or complete collapse

References:
1 Leonards,G.A. (1982). “Investigation of Failures.” Journal of the Geotechnical
Engineering Division, ASCE, 108 (GT2):187-246
2 EN 1990:2002. Eurocode - Basis of structural design, European Standard.
European Committee for Standardisation, Brussels.
3 In this context, the term structure is taken in its broadest sense and includes
constructed works such as slopes and embankments, or the natural ground in the case
of landslides.
4 EN 1997 -1:2004. Eurocode 7 : Geotechnical Design - Part 1 : General Rules,
European Standard. European Committee for Standardisation, Brussels.
5 SAICE ? Joint Structural Division (1995) Code of Practice for Foundations and
Superstructures for Single storey Residential Buildings of Masonry Construction.
1st Edition.
6 AS 2870-1996: Residential Slabs and Footings Construction. Standards Australia,
Chapter 3
CHARACTERIZATION OF DISTRESS AND SOME CASES

Anirudhan I.V.
Geotechnical Solutions
19 Usha Street, Dr. Seethapathy Nagar, Velachery, Chennai – 600 042
anirudhen@eth.net

ABSTRACT

Types of distress found in structures due to different geotechnical reasons are briefly outlined in this
paper. The patterns of distress generally found in near failure cases or at a stage where a remediation
is required are typical and these patterns can be used to arrive at the possible remedial measures.
Legal disputes arising out of the distress cannot be simply resolved by these patterns and in such
cases, the patterns are used for formulate further investigation of the problem and then fixing the
responsibility. This step of the forensic geotechnical engineering is not dealt with in this paper.
Three cases of distress identification during construction are discussed in the second part of the
paper. Deficiencies in the design, construction methods and the deficiencies in the geotechnical
investigation are the causes for these distresses. The attempt to identify the causes for distress and
then to recommend remedial measures in these cases may not be very scientific and legally
acceptable as the problems were not looked into from a legal point of view.

1 INTRODUCTION
Any geotechnically induced distress in a structure is a result of deformation or displacement
of the soil over which the structure is supported. The relative stiffness of the structure vis-à-vis the
soil supporting it plays a major role in most of the distress. Unfortunately the relative stiffness itself
is constantly changing due to several factors, which includes the influence of the structure itself.
Deformations and displacements due to swelling and shrinkage point to a continuing, but erratic
change in the relative stiffness in the soil by the influence of environmental factors. Changes in the
otherwise harmonious relative stiffness because of the changes in in-situ stress conditions due to
excavations, dewatering, blasting, etc., also cause distress. Earthquake induced distresses are the
results of such changed stress environments, but with a distinct feature of stress reversals at very
rapid successions.
Most of the structures have complex dimensions and properties that cannot get along with
more complex behaviour of soil due to the changes imposed by the structure. The limits put forth on
such several changes within the soil as well as in the structure to some extend prevent the distress.
Several studies on these limits revealed serious limitations on generalising them and only resulted in
more and more limiting parameters (Boone, S.J., 2001).
While the analysis of distress is complex and requires in depth case to case study, patterns of
distress can point to an initial assessment of the possible cause that initiated the distress.
Characterising the distress with respect to such different causes may be the first step towards such
detailed analysis.

1
In forensic geotechnical engineering, emphasis is given on the study of causes and remedial
measures after a structure is undergone undue distress and deformation significantly reducing the
utility of the structure. Often, the symptoms and initial stages of distress are ignored and patched up
leading to a major distress making it more complex for further remediation. Most of the classical
examples of back analysis cited by Madhav (2003) are well planned failures helping a step to step
back analysis leading to constructive conclusions. The slope failures, testing of foundations to large
deformations (footing load tests, pile load tests, retaining wall load tests, etc.), piping failures, etc.
have less complex modes or the modes can be simplified significantly.

The symptoms that caution the designer can be identified during several stages of
construction itself helping the designer to take corrective steps to prevent a major distress after the
completion of the structure. The defect can be in the basic design, in the investigation of sub-soil
conditions that can be revealed during a deep excavation, and also in the construction of foundation
including the defect in the workmanship and wrong methods. The back analysis in such case is not
very complex in terms of geotechnical engineering. Such cases are complex mostly because of
concealment of various factors relating the construction procedures and the symptom of distress.

This paper attempts to briefly put the reasons for distress and then profiles possible distresses
in various geotechnical problems cases. Since the distress is often exhibited in the form of cracks,
some of the common features of these crack patterns in relation with the reason for such distress are
identified and explained. Starting from lightly loaded conventional buildings and multi-storeyed
buildings built on conventional shallow foundations and special foundation, distress patterns for
structures like retaining walls, deep excavations, bridge supports, high embankments, roads, etc. are
outlined. Earthquake induced distresses are not reviewed here.

The second part of the paper discusses three cases where the major distress is prevented
identifying (i) a wrong construction procedure. (ii) a design defect, and (iii) a wrong assessment of
the soil condition.

2 CAUSES OF DISTRESS

It is well understood that a construction, which does not undergo displacement, settlement, or
deformation is impossible. The practice is to adopt a design that limits these to allowable levels.

‘Limits’ to foundation movements / rotations / displacements, etc. were developed from


various studies undertaken all over the world. These limits, as on today, have wide ranges and the
debate is still on about the most significant parameter that would explain the cause of distress in
structures. Skempton and MacDonald (Burland 1977) relate distress to angular distortion β. This
implies that the damage results from shear distortion within the building, which is not necessarily
the case. These studies separate load bearing brick walls from framed structures. Some of the studies
introduced deflection ratio Δ/l into the realm of criteria for damage (Burland 1997). Burland in his
earlier studies (Burland and Wroth, 1974) related the building damage to limiting tensile strain εlim
along with L/H and E/G, the ratio between Young’s modulus and shear modulus, as the deciding
parameters. The typical damage criteria with respect to individual crack width have been in use for a
long (Burland & Wroth, 1974). Burland had however cautioned about the use of crack width alone
as the criteria of damage. Studies by Boscardin and Cording (1989) later introduced limiting lateral
strain εh also into the damage criteria.

2
Boone (2001) emphasised that simplified

3
procedures using angular distortion, deflection

Horizontal strain εX10


ratio, lateral strain and limiting tensile strain
would not be sufficient in the damage related
studies of more complex structures we have
today. Figures 1 & 2 reproduced from his case
studies, will suggest that all these formulations
do not really predict damage by distress and in
Angular distortion, βX10
3
turn it suggests that there is a risk of damage
even after painstakingly taking care of these
Figure 1: Angular distortion, horizontal strain and
possible hazards. damage category and reported damages
(Boscardin and Cording, 1989)
‘Strain superposition method’ suggested
by Boone (2001) assumes influence of the
building stiffness on the final ground profile
rather than accepting the general assumption of
structure deforming to match the ground

Δ/l (%)
movement. The later may be conveniently
accepted, though bit conservative, in the case of
one to four storied structures usually found in
urban areas. He analysed many cases and found
that the said procedure agrees well in the case of
these relatively simple structures. However exact
Horizontal strain εX10
3
ground movement cannot be predicted in a
simple way. Figure 2: Horizontal strain and deflection ratio and
damage category and reported damages
3 DISTRESS PATTERN AND MODE OF (Burland 1997)
DEFORMATION
Distress in a structure is often related to the
foundation displacements due to various reasons. A
distress is ‘sighted’ only when it exceeds a certain limit
that can cause ‘cracks’ in the elements of the structure.
It is understood that all the structures undergo
deformations for various reasons and the ‘undue’
deformation not confirming to safety standards are
usually referred as ‘distress.’ Most of the studies
discussed above consider three modes of deformation,
bending, shear and extension, for describing the crack
pattern. These modes are illustrated in Figure 3, (Boone,
2001). The crack patters are typical. However, what one
sees in a structure with undue distress are many cracks
without a definite pattern that would fit into any of these
three. Thus what one sees in a distressed structure are
manifestations of two or more modes of deformations.
Just observing the cracks and even measuring different Figure 3: Separation of deformation modes
(Boone, 2001)

3
parameters like angular strain, deflection ratio, lateral strain, etc. will not explain the root cause for
the distress. Behaviour of some uniformly loaded structures such as storage tanks can be explained
by standard deformation theories (Marr, Ramos and Lambe, 1982).

4 DISTRESS IN RELATION WITH MODE OF ‘FAILURE’ AND CAUSE FOR


DEFORMATION

4.1 Bearing Capacity Failure:


Bearing capacity failure often occurs under undrained condition especially in the case of
clayey soil at foundation. Punching through very loose fine sand is also possible when the
foundation is over loaded almost instantaneously. The displacement resulting from a bearing
capacity failure is often very excessive and the resulting distress is failure in most cases. There is an
element of rotation displacement in such failures. Slope failure is one example. Bearing capacity
failure will cause excessive differential movements that can cause shear. The result is vertical cracks
or rotation failure. Simultaneous and uniform bearing capacity failure of all foundations of a
structure is just not possible.

4.2 Compression Settlements - During Construction:


Loose sand deposits as founding soil can cause large settlements during construction. Large
differences in the loading of different components of a structure can lead to undue differential
settlements that are sudden and prohibitive. The resulting distress is also like bearing capacity
failure because of large shear in the structure. Results almost vertical shear cracks and failure.

4.3 Consolidation Settlements – After Construction:


On the other hand, long term consolidation settlement in excess of permissible limits can
cause differential settlements, during which the structure is allowed to adjust itself depending up on
its capability to do so. Large storage tanks allowing to settle gradually under consolidation
settlements, but avoiding a bearing capacity failure by not loading suddenly, are common. Large
consolidation settlements can cause differential settlements in a structure with non-uniform loading
and with the subsoil having varying thickness of the consolidating layer. The result is often diagonal
cracks. Non-uniform foundation sizes, non-uniform column loading and relative positions of the
foundations are the cause for large differential settlements. Prolonged settlements under secondary
consolidation of grounds improved by pre-loading causing distress to structures with non-uniform
loading are also reported.

4.4 Delayed Compression – Collapsible soils:


Ramanathan Ayyar & Jaya (2003) describe delayed failure in laterite as one of the reasons
for mass collapse of wells in Kerala during 2001. Large cavities in lithomarge clay below laterite
crust due to continuous water seepage through this soft soil is common. Collapse of lithomarge and
the top crust is possible due to sudden rise in ground water level. Figure 5 shows such cavity of
roughly 10.0m x 12m at a depth of about 2.50m below the surface found in construction site in
North Kerala. The soil around the cavity was very soft and wet, whereas the crust, the soil above the
cavity was very hard laterite that required mechanical excavation. The reason for such large cavity
was then deduced as the loss of lithomarge (fine soil mss) through seepage into large excavations for

4
quarrying the laterites stones very close to the
site. As seen form the picture it is very difficult
to assess such situation in a large site where the
exploratory boreholes are made at a normal close
spacing 25m to 30m and the presence of such
cavity just below a heavily loaded column will
be a disaster.
Similar collapsible formations are
reported in many parts of western Tamilnadu
also. Some sand deposits in Rajastan are also
reported to be collapsible. Distress in these cases
may be leading to failure. Figure 4: large natural cavities found in laterite
formations, Pattuvam, Kannur, Kerala
4.5 Squeezing of Soft Soil – Tooth Paste Phenomenon:
This ground movement is different from settlement or punching. Lateral displacement of very
soft soil sandwiched between relatively strong soil / structure above and below is possible when the
incremental stress in this soft soil is more. The displacement in this case also is often ‘sudden’ and
the result is lateral displacement of the foundation apart from vertical movements. Tilting can be an
associated feature.
Poulos (2003) describes a case of lateral flow of very soft soil due to adjacent excavation that
caused undue lateral forces resulting failure in the deep foundation supporting a huge structure.
Though this case is not strictly a case of squeezing of soft soil, sudden relief of confining pressure
giving rise to increased K could be the reason for this flow.

4.6 Rotation and Sliding:


Rotation is related to bearing capacity failure and slope stability. Sliding of retaining
structures can be independent of rotation, while sliding in a slope is also a rotation with very large
curvature even though some models assumed a wedge failure. Wedge failure shall be termed as
sliding. The distress noticed under these cases is in the form of relative ground movement, failure of
retaining structures and embankments. In most of the cases, the distress is sudden. Many rotation
failures of abutment piers manifested by bearing capacity failure because of overweighing back fill
are reported.

4.7 Hydraulic Fractures such as Piping:


Continuous removal of fines from soil
due to seepage or piping causes deformation in
earthen structures such as reservoir bunds, earth
dams, etc. The distress is noticed in the form of
unequal settlements and local slope failures.
These indications are warning of a major failure
such as slope failure and sliding. A case of
failure of a wide RCC channel due to piping is
Figure 5: Failure of a canal (and adjacent large sump)
presented in Figure 5.
because of erosion of soil initiated by piping due to seepage
of water stagnated on the side of the channel, Ludhiyana

5
4.8 Heaving:
Heaving is associated with deep excavations in soft clay or very stiff soil or excessive filling
over a desiccated crust of firm soil of small thickness over a thick layer of soft clay. Heaving of
excavation bottoms in the case of soft clay formations is common because of reduced effective
stresses. Bulging of excavation sides is also common in the case of very soft deposits. Lateral earth
pressure exceeding the effective vertical stress is common in such cases.
Heaving of ground due to lateral and upward movement of very soft deposits below relatively
thin desiccated crust or better soil occurs when heavy area loads are placed. Heave occurs away
from the loaded area and often develops gradually. Indications are displacements and resulting
distress in small structures in the heaved area.
Deep excavations into very hard over-consolidated clay formations or sedimentary rock
formations like shale and mudstone can result large up-heaving of excavation bottom. This is apart
from possible swelling under saturation if the soil in-situ is partially saturated. Shrinkage cracks
developed within the natural soil immediately below the excavation bottom and sides because of
increased σ3 and subsequent saturation can cause considerable reduction in bearing capacity and
increase in compressibility. Small footings placed over large common excavation area can be
affected by this softening.
Even very shallow excavations in
expansive soils results significant heaving of
the excavation bottom by the migration of
moisture under changed stress conditions.
The heaving is aggravated by the presence of
ground water within the capillary range.
Lowering of ground water table alone is not
reducing the evil of heaving in such cases as
the process of heaving is not merely related
to the presence of ground water table, but
significantly related to moisture locked in
the soil itself. Examples of disintegration of
highly weathered shale upon shallow Figure 6: Disintegration of weathered shale upon one cycle of
excavations are plenty (Figure 6). drying (after excavation) and wetting (during construction)
Oragadam, Chennai
4.9 Subsidence:
General ground subsidence can happen due to ground water lowering in relatively loose and
soft deposits. These movements are typically gradual and slow giving enough warning to possible
damage to structures in the affected area. The net result is however a more dense sub-soil, and this
does not help in avoiding distress in the existing structures because of possible differential
movements Mair (2003).
Subsidence due to ground movements during tunnelling and mining are often sudden because
of the involvement of pore pressure. Such ground movements usually result in loosened soil.
Similarly the ground movements due to reduction in density due to particle loss along seepage
during heavy ground water lowering by dewatering are also sudden. Caving in of several wells in
Kerala during 2001 after a relatively intense rains is an example (Ramantha Ayyar & Jaya, 2003).

6
4.10 Swelling and Shrinkage:
Foundation displacements and slope
failures are common where swelling soil is
encountered. Moderately swelling soil
experiences large volume reductions by loss
of moisture. Displacements due to swelling
and shrinkage will be highly varying
between different parts of a building
because of varying moisture build up or
loss below different foundations. Usually
footings in the corners are affected more
because of relatively small loading and
more exposure to environmental changes.
Moisture variations within the building area
in normal conditions will be small leading
to small movements for foundations in the Figure 7: Differential Heave Pattern of The Building at Anna
interior developing large differential Nagar, Chennai (Ramaswamy and Narasimhan, 1978)
movements between exterior and interior
foundations.
The differential heaving associated with the swelling of soil has resulted distress in many
structures Ramaswamy and Narasimhan (1978) had investigated a single storey building at Anna
Nagar which has undergone severe cracking. The differential heave pattern of the building is shown
in Figure 7 as indicated by the plinth line at the time of inspection.
Progression of moisture towards interior of buildings where swelling soil is used in
backfilling the excavations and plinth can cause swelling of soil in the interior after a prolonged
period. Interestingly the cracks usually associated with swelling of foundation soil is vertical and
horizontal. Horizontal cracks between roof slab and load bearing wall on one side results when the
cross wall foundation is subjected to swelling displacement. Heaving of plinth beams supported on
swelling soil can cause hogging in the walls allowing to develop tensile cracks at the top portion of
the walls. Outward movement of walls are also noticed due to rotation of plinth beams along
horizontal axis.
Displacement due to shrinkage is also gradual similar to that due to consolidation settlement,
but can be very erratic. Alternate shrinkage and swelling displacements cause distress in the
structure over a period of time and this later manifest into a major crack or deformation. By the time
this severe crack is noticed in the structure, many minor unnoticed cracks due to stress reversals
could have been inflicted weak zones in the structure. Shrinkage of topsoil under drought or ground
water depletion causes sinking of floors and development of additional stresses in plinth beams
designed as ground beams. Foundations initially resting over medium stiff clays of moderate shear
strength can displace due to large volume reduction under loss of moisture. Cases of failures are
reported due to shrinkage of foundation soil under boiler foundations under severe moisture loss due
to induced heat. These are occurring after a period and can be differentiated from other types of
distress.

7
5 DISTRESS PATTERNS

5.1 Lightly Loaded Buildings on Shallow Foundation


5.1.1 Due to Differential Loading:
The most common distress found in lightly loaded
buildings on shallow foundation is vertical and diagonal
cracks below the window sills. This, usually very
prominent and eye catching distress, is not strictly due to
any geotechnical issue. The imbalance in the load
distribution towards foundation due to large openings
provided for the windows and the resulting tensile stresses
in the generally brittle wall is the reason for such distress. Figure 8: Distress in walls due to
This distress happens within few months from the differential loading
construction (Figure 8).
5.1.2 Due to Differential Settlements:
If the diagonal cracks are towards the corner or to the
edges of the walls, such distress could be associated with
differential settlement within or between the foundations.
Usually such distress appears in one or two years after the
construction and involves the settlement of underlying
clayey stratum. The floors are generally intact. The floor
portion close to the walls may try to move along with the
foundation causing distress little away from the intersection Figure 9: Distress in walls due to
of the floor with the wall. This happens when the settlement differential settlements (consolidation)
is relatively large.
More settlement for heavier interior
column footings is a clear indication of
consolidation of underlying clay stratum. The
apron provided around the outer walls is usually
detached and cracked when the foundation
settlement is excessive. In the case of severe
SHRINKING SOIL SHRINKING SOIL
settlements, several horizontal cracks associated
with the main diagonal cracks will be sighted
(Figure 9). Figure 10: Distress in walls due to differential
settlements (shrinkage in founding soil)
5.1.3 Due to Shrinkage in Soil:
Differential settlements occurring after a
long period without any change in the loading of
the structure shall be attributed to volume
changes in founding soil due to desiccation or
shrinkage. The foundations below the exterior
walls and columns are more affected because of
more exposure to environmental changes. Figure 11: Distress due to differential settlements
(shrinkage in founding soil)

8
Shrinkage of founding soil aggravated by the water consuming trees and bushes rather than
physical interference by the tree roots is also common. There is no definite pattern for such distress,
but can be identified from the period at which it occurs. The floors are affected badly in such cases
mainly because of the detachment from supporting soil that is dried up. The plinth protection around
the outer walls shows severe distress by way of detaching from the walls and settles more than the
walls (Figures 10 and 11).
5.1.4 Due to Swelling of Founding Soil:
Distress due to swelling / heaving of founding
soil is more haphazard in comparison to the distress
resulting from increased differential settlements.
Generally wide cracks develop between the interface of
the columns and walls or between the roof slab and the
interior walls. Shallow foundations supporting the outer SWELLING SOIL

columns and walls are generally affected and the Figure 12: Distress due to differential
displacements by swelling of founding soil
problem is aggravated by the fact that these columns and
walls are with less loads in comparison with the interior
ones. Hogging of the foundation portion below large
window openings will result in the case of strip footing
placed over expansive soil. This distress is similar to the
one attributed to imbalance loading over the wall below
the window sill. However, one of the major differences
BOWING
between these two distresses is sort of a cyclic nature of WALL
the distress due to swelling. Another difference is the
BRICK WORK
appearance of distress above the window opening or at
DETACHING THE
the interface of the lintel beam and the wall (Figure 12). DOOR FRAME

Lateral separation between window / door frames from


the walls is a clear indication of differential heaving in Figure 13: Distress due to differential
the continuous footing (Figure 13). displacements by swelling of founding soil

Complete shear of the walls displacing it from its alignment is an indication of swelling of
founding soil. Severe uplift pressure caused by the heaving soil can cause shear of infill walls. The
distress is spread to upper floors when the structure is with load bearing walls and more distress will
be noticed near the window openings. Detachment of interior cross walls from the outer ones is also
very common in such case. Such distresses make the load bearing walls considerably weak and the
in-fill walls loosing their confinement.

5.2 Lightly Loaded Buildings Supported on Short Piles / Under-reamed Piles


The major cause for distress in such structures is improper selection of the founding level for
the supporting system. Even though the active zone with respect to severe moisture changes can
easily be identified from local data, severe draughts and measures like rain water harvesting, etc. can
trigger large changes in the active zone. Many case of shearing of wall and column interfaces are
reported due to uplift of the supporting system because of changed environments. An extended
draught period will result loss of moisture below the active zone causing tremendous swell pressure
on the supporting system during next wet season. The floors and plinth walls properly treated for
swelling pressures from the soil from shallow depths may remain intact in such cases triggering

9
distress in the interfaces. Horizontal cracks in upper portions of the infill walls, separation of infill
walls from the roof, etc. are other distresses noted in such cases.

5.3 Structures Supported on Pile Foundation


Distress due to settlement of pile foundations is not generally expected unless very large pile
groups are involved the supporting system. Large differential settlements between medium and large
pile groups are reported because of underlying weaker layers. Excessive and continued settlements
are also reported because of inadequate bearing stratum. Severe distress in the form of shear cracks
on the beams and the columns supported on piles is reported where one structure on pile foundation
abutting another structure on spread foundation is constructed. Consolidation of upper soft clay
layers under the weight of the structure with spread footing transfers considerable amount of load
through negative drag on the piles of the adjoining structure.
Failure of end bearing pile foundations under a load much less than the design load was
reported from a site where the filling required for reclamation was the order of four metres. The
piles failed in shear just under the tremendous negative drag caused by the consolidation of roughly
6.0m soft clay under the gravely soil fill of 4.0m. The failure occurred just in three to four months

5.4 Bridge Supports


The most common distress found in bridge structures is the level difference between
approach length and the deck portion. This is definitely due to settlement of the approach portion.
Generally there is a thick embankment or fill below the approaches and a well laid fill or
embankment is not expected to undergo compression. Consolidation settlement of the founding soil
is the major cause. The level difference or opening up of the deck joints over a pier suggests
differential settlement between the piers or tilting of piers. Joints opening up between the deck slabs
can also indicate uplift of piers founded on very hard clayey soil or shale formation when the river
swell after a long draught period. Lifting of the first deck slab supported on the abutment pier is an
indication of the rotation of the abutment pier under the embankment load.

6 CASE STUDIES OF DISTRESS DURING CONSTRUCTION

A Geotechnical Engineer will


1. ‘Foresee’ several scenarios
2. ‘Assume’ most reliable design parameters including the loading (with the help of an
investigation)
3. ‘Anticipate’ deviations from the assumptions
4. ‘Analyse’ such possible deviations
5. ‘Set the limits’
6. ‘Estimate’ the stresses and strains for all the possible scenarios, and
7. ‘Reworks’ if the stresses and strains exceed the limit
before finalising a geotechnical design. However, often the inadequate data on the structure itself,
continuous changes in the overall planning of the structure by the architects and owners, etc. compel
the geotechnical engineer to finalise a design well before the finalisation of the final project the
owner decides to execute. Very rarely the geotechnical design is re-looked or re-analysed to fulfil all

10
the requirements of the finalised project. At the same time there are instances where the geotechnical
engineer failed to foresee the problems of a particular construction procedure that lead to
unsatisfactory performance of the foundation. There are several cases in which the geotechnical
investigation was inadequate of inaccurate that led to failures of near failures. The three cases
described below try to illustrate some of these lapses.

6.1 The Case of ‘Lifted’ Piles


6.1.1 Problem
In 1994, in Chennai, about 1100 driven cast-in-situ piles were constructed to support a
multistoried building with two basements. The construction of piles was done from (-) 4.0m level,
whereas the basements were to be as deep as (-)9.5m and (-)11.5m. 600mm diameter and 550mm
diameter piles extending to (-)19.0m were driven. Since the piles were to be cut-off below the
basement levels, the concreting depth was limited to (-)8.0m, while the remaining length up to the
working level was filled with sand.
When the excavation started after the construction of piles, the concrete top levels were
found to be at much higher level than planned. The owner’s engineer thought that the piles were
somehow lifted and there could be gap between the pile tip and the soil. Some piles were then
stripped up to the cut-off levels and found reduction in pile diameter over a large length (Figure 14)
and also discrepancies in the levels of reinforcement. The reinforcement top levels were however
lower than the expected ones in most of the cases of discrepancy, while the cases of ‘cage lifting’
were also significant (Anirudhan, 1997).
6.1.2 The Distress
The major distress apprehended here was the loss of
contact between pile tip and the supporting soil. The
discrepancy in reinforcement top level brought the fear of
cavities in the pile concrete at deeper levels.
6.1.3 Perusal of Construction Method
The foundation consultants had cautioned about
‘necking’ (reduction in pile diameter) in pile concrete over a
portion between (-)6.5m and (-)10.0m where soft clay existed.
Ground water table was at (-)4.50m during construction and the
excess pore pressure developed in this soft clay could exert
tremendous pressure on the green concrete forcing it to squeeze.
Ideal option for preventing this was to provide concrete up to the Figure 14: Necking in driven cast-in-
working level irrespective of deep cut-off levels. situ piles, Chennai

However, the concrete top level during construction was decided as roughly 2.0m above the
cut-off level and filling of sand in the rest of the length was adopted to save cost. This remedial
measure proved ineffective because of arching of sand within the casing pipe completely relieving
the load over the green concrete that was supposed to have been provided with an overburden.
Similarly the excess hydrostatic pressure developed in the very dense clayey sand (residual
type) at the pile tip during driving could be very significant, while the hydrostatic pressure inside the
casing is zero. The minimum height of concrete column within the casing pipe was hence suggested

11
as 11m before allowing even a small lift of the casing pipe so that the weight of green concrete is
adequate to counter the hydrostatic pressure developed at the pile tip.
6.1.4 Actual types of distress
Now the types of distress in question were modified to the following
Reduction in pile diameter due to necking
Possible reduction in pile diameter below the soft clay levels also
Deterioration of concrete at pile tip due to excess pore pressure
Vertical displacement of reinforcement cage
6.1.5 Issues to be Investigated
1. Whether the necking was limited to the soft clay between (-)6.50m and (-)10.0m alone?.
Even though there was no weak layers below (-)10.0m, the soil layers up to (-)16.0m were
not very dense or stiff.
2. Whether the pile really lifted from the base?
3. Weather the reinforcement was intact in its position or was there a lift of the
reinforcement on account of the necking?
4. Whether the piles were deficient in their structural capacity due to reduction in cross
section?
5. Whether the piles had deficiency in end bearing resistance due to bad concrete at pile tip?
6.1.6 Investigation of Distress
A systematic investigation of these distresses was executed in the site.
The extend of necking in pile shaft: This task was achieved by systematic volume
measurements with the aid of reasonably thorough pile driving records and, concreting &
reinforcement details available for each pile executed. The top level of pile concrete was measured
using precision level. Diameters at every 300 to 500mm length interval were then measured to a
level at which the pile diameter was equal to the design diameter after the occurrence of first
necking. The total volume of the concrete within the necking portion was found to be equivalent to
the theoretical volume of the pile up to the theoretical concrete top level. Such observations were
made for almost all the piles suggesting that there was no reduction in pile diameter (necking) below
the first level of necking found at about (-)10.0m.
Lifting of pile from the base: The above measurements also expelled the fear of lifting of pile
from the base
Reinforcement displacement: The fear was that when the necking in concrete took place
resulting an upward movement of the green concrete, this concrete might have also lifted the
reinforcement cage. The length of reinforcement cage above the level up to which the necking took
place is only one sixth of the total reinforcement length. The resistance offered between the
remaining length of the reinforcement and the concrete around this length is much more than the
lifting force that can be offered by the concrete moved upward while necking took place.
Physical verification of the top level of the reinforcements for each pile were made.
Interestingly the observations revealed that the top levels of the reinforcements for almost all the

12
piles in a group remain same. There were groups of 6 piles to 18 piles. Very rarely there was relative
difference within a group. There was a drawback of not recording the actual ground levels at each
pile location that complicated this review process. An average ground level of (-)4.50m was
mentioned in the records of all the piles which could not have been the case in such a large piling
area. The length required for reinforcement cage in each pile was estimated based on this average
ground level. This caused small differences between the levels of reinforcement in piles of different
groups.
About 1% piles found suffering from relative displacement of reinforcement by 50mm to
350mm, in which most of the cases were less than the lengths as per record. A slip at lap joints when
the laps were towards the bottom of the cage was suspected to be the reason for this phenomenon
and this was verified from some records. Only 5 piles recorded lifting of reinforcement. One pile
recorded a cage lift of 560mm.
Compression capacity of piles with respect to bearing: Direct measurement of static vertical
compression capacity by means of maintained load tests was resorted to. Eleven piles out of about
1000 piles were subjected to load tests apart from two initial load tests carried out from a higher
level. Eight piles recorded settlements within the permissible limits and two piles recorded slightly
excessive settlements. One pile failed at very small load and this pile had the history of cage lifting
by 560mm. However, the range of settlements recorded in some of the load tests were towards the
maximum limits and there were deficiencies when compared with the results from initial load tests.
The piles were driven to a very hard set of 3 to 5mm for ten blows of 4 tonnes hammer falling from
1.20m height at the time of construction and such large settlements were not expected under static
load.
The effect of excavation for basement: The relief of roughly 6.0m thick soil for the
construction of basement will cause a ‘heave’ in the soil below the excavation level since these
layers are relatively stiff and dense. This heave might cause lifting of piles through the friction
between the heaving soil and the pile. This might also cause a relative movement between the pile
tip (flat shoe in this case) and the soil at this level. However, such relative displacements shall be
fairly uniform and the load test results should also have resulted a uniform pattern. Large differences
in the behavior of piles pointed to different reason/s.
The excess pore pressure at the time of pile driving: The completely weathered rock present
at the pile bearing level has reasonable amount of clay suggesting low permeability. Large amount
of excess pore pressure could develop during such hard driving of pile. Anticipating that the entire
excess pressure could not have dissipated before completing the concreting and lifting of the casing
pipe, the piling instructions included a direction not to lift the casing pipe before placing minimum
length of 11m concrete column within the casing. This concrete column provided a pressure of 22
t/m2 against excess pore pressure of about 8 t/m2 plus the static water pressure of 14.5m at the pile
tip. It was reported that some of the piling rig did not have the capacity to lift the casing pipe
carrying a large column of concrete. There was also significant resistance from the soil around the
casing. Therefore, the piling agency resorted to an initial lifting of 500mm after filling only about
three metre concrete in the casing pipe. The concrete exposed to the existing excess pore pressure
should have suffered loss of cement particles resulting a weak concrete at the pile tip. Then the real
worry was to identify such piles since the piling records did not mention the initial lifting of casing
pipes.

13
Investigation by pile integrity test, PDA: A large number of piles were subjected to low
energy PDA to confirm the pile diameter, length of pile and the quality of concrete. These factors
were determined with some accuracy, but the results were not conclusive on two aspects. One is the
quality of concrete at pile tip and the other was inconsistencies in shear wave velocities. In almost
all the cases intact pile tips were recorded, which was proved wrong by other means later.
Investigation by high energy hammer tests: It was decided to conduct some high energy tests
by ‘driving’ the piles using a 4 tonne hammer. A suitable frame with leader was devised and initially
around 100 piles were driven. While about 70% piles did not penetrate more than 4mm under three
or four hammer blows, 30% piles penetrated more than 10mm in one or two blows. This was
attributed to poor tip conditions caused by premature lifting of the casing pipe since the piles were
initially driven to very hard set. Most of these piles, however, offered resistance after this ‘initial’
displacement. These 30% piles would have settled more under the working load while the pile are
expected to settle only 6 to 8mm under the working load.
These observations were not in comparison with the PDA test results, particularly in the case
of quality of pile tip. Therefore, PDA tests were discontinued. However, driving the piles using high
energy drop hammer was resorted for identifying the piles with bad tip conditions and also for
driving it further to achieve adequate seating of piles with relatively large displacement during initial
blows. Except for 5% piles, the piles could be driven to a hard set with maximum movement of 10
to 12mm.
Piles with more set: The 5% piles with more set had to be underrated from its safe capacity
and some piles were to be rejected altogether because of continued displacement. About 1% of the
total piles were rejected and new piles were driven to compensate the deficiency.
6.1.7 Conclusions
Concreting the piles up to about 3.0m above the cut-off level and filling the remaining length
up to working level using sand did not work against necking of concrete within soft clay potion.
This was identified to be attributed to arching of sand within the casing pipe completely relieving
the intended load over the green concrete. However, in this particular case, the cut-off level was
very close or below the portion affected by necking. Foundation consultants had to learn this lesson.
The piling agency failed to comply with the requirement of minimum length of concrete
column within the casing for countering the excess pore pressure developed during driving. This
resulted bad concrete at the tip of a significant number of piles
The piling agency also failed to take ground level measurements at each pile location and
instead resorted to approximate measurements leading to misjudgment of reinforcement lengths.
This resulted difficulty in identifying piles with real problem of reinforcement cage lifting.
Placing un-welded laps on main bars towards the tip of the reinforcement cage had the risk
of slipping at the lap by self weight of remaining cage above the lap.
These lapses resulted huge investment on rectification measures like re-driving of all the
piles and, underrating of few piles and rejecting some piles apart from heavy loss of time.

6.2 The Case of Settling Bungalows


6.2.1 The Problem

14
A vast residential project site was developed from a lowland by filling good quality murrum
(residual gravely soil) and then by preloading, supplemented with band drains for increased rate of
consolidation settlement. The effect of preloading was confirmed by time-settlement measurements
using plate settlement markers. The observed settlements were 780mm to 850mm, close to 90% of
the total estimated settlement of 900 to 950mm. The time settlement curves also suggested 90 %
consolidation under the prescribed pre-loading period of 60 days. Construction of two bungalows
was taken up immediately and a systematic measurement of settlement under every stage of
construction was carried out.
Two bungalows were then constructed and a detailed monitoring of the settlement was made.
One bungalow settled by about 145mm and the other one settled by 80mm, whereas the expected
settlement of the structure after ground improvement was less than 40mm. Fortunately, the
settlements were uniform because continuous RCC strip footing foundation was provided for these
buildings, whereas the initial design was based on RR masonry strip footing. Even though the
structure did not experience any distress because of uniform settlement, the excessive settlement far
beyond the expected settlement, was considered as a ‘failure’ in the design of ground improvement
programme.
6.2.2 Distress and Owner’s Worry
Even though the settlements were uniform, there were apprehensions about settlements
larger than acceptable limits. The differential settlements at every stage were negligible because of
fairly uniform construction schedule adopted for these two buildings. The foundation consultant and
the ground improvement agency were called upon to explain the large settlements irrespective of the
commitment that the settlement of finished structure not to exceed 40mm.
6.2.3 The Soil Profile
The detailed investigation prior to the foundation design revealed presence of 1.20m thick
residual soil fill followed by soft to very soft marine clay of 4.50m to 7.00m thick below which
relatively stiff residual clay existed. Weathered rock stratum followed. Compressibility of plastic
clay was high and long term settlement due to consolidation of these layers under the weight of
existing fill and another 0.80m fill proposed was estimated.
Shallow foundations resting in the fill and deep foundation resting in weathered rock
suffered doubts of long term performance. The distress in infrastructure like roads, sewage lines,
water pipe lines, etc. were anticipated and ground improvement using pre-loading was considered
more appropriate.
6.2.4 Original Design
The amount of preload was decided based on a pre-determined finished ground level of
0.80m above the existing ground level. There already existed an original fill of 1.20m thick that has
not undergone adequate compaction. A fill of about 3.80m above the existing level imposing a load
of about 7.0 t/m2 was the recommended pre-load as the average load intensity from the proposed
construction including the weight of the 0.80m fill required to reach the finished ground level was
close to 6.2 t/m2 (roughly 90% of the pre-load). Vertical band rains at an interval of 0.90m were
introduced for the depth of soft clay for accelerating the consolidation process. It was expected that
about 2.20m thick preload fill (equal to a load of 4.2 t/mm2) could be shifted to other locations

15
leaving the FGL as required. Figure 15 illustrates the preload fill and the expected settlement under
the preload. Figure 16 illustrates the design principle of pre-load for ground improvement.

Area for building


Other areas

Preload for removal

Finished ground
level

Original ground
level

Soft clay
compressed by Soft clay compressed by 800 to
400 to 500mm 1000mm under pre-load
under pre-load Pre-consolidation improved by 2.5 t/m2

Figure 15: Pre-load fill and the settlement


6.2.5 Changes in the Design
The owner and architect meanwhile
decided to raise the ground level by another PRESENT PRE-CONSOLIDATION-
(LOADING HISTORY)
0.80m for better appearance and also to LOADING

avoid any possibility of flooding of the LOAD FROM PRE-LOAD +


PLASTIC YIELD

FILL UP TO FINISHED GL
premises. But this was informed to the
consultant and the ground improvement TOTAL
SETTLEMENT
agency only after the removal of the pre- NET
SETTLEMENT
load and the strat of foundation excavation..
SETTLEMENT

The consultant advised further pre-loading REMOVAL OF PRE-LOAD


to compensate the deficiency arose because EQUAL TO LOAD FROM STRUCTURE

of revision in the finished ground level. PROCESS OF CONSOLIDATION


However, the owner decided to compact the (EXPELLING WATER FROM VOIDS)
soil at founding level and carried out plate
Figure 16: Design principle of the preload
load test on the compacted soil. Satisfied
with the small settlements measured in the
plate load test, the owner proceeded without
further pre-loading.
The plate load test using a 450mm x 450mm plate naturally showed less settlement since the
test was done on the compacted fill. Based on the results the construction proceeded, but after
agreeing to provide continuous strip raft as foundation (Figure 17).

16
The construction of both the
buildings was fairly uniform and took SETTLEMENT VILLA # 42
about 6 months. Regular measurements of NO OF DAYS
settlement at different corners of the 0 20 40 60 80 100 120 140 160 180 200

buildings were made. The band drains 0

functioned very effectively during the 20

40
construction and the settlements were CUM SETTLEMENT MM
First floor roof REAR
completed FRONT
very rapid. The settlements measured 60

after completing the construction were 80

100
145mm and 80mm for the two buildings. Roof tiling completed

120
The Figure 18 illustrates the settlement of
140
one villa. Settlement stabilised within
160
fifteen days from the completion of the
180
construction. Uniform construction
200
schedule helped in resulting a uniform
settlement without any significant
Figure 18: Settlement observation of Villa # 42
differential settlement.
6.2.7 Analysis
Very careful study of the load settlement curves will show that the rate of settlement
significantly increased after 20 to 30mm settlement suggesting the major settlement is resulting form
virgin compression and not from re-compression expected from pre-loaded soil. This means that
there is a deficiency in the pre-load intensity. In this case, the pre-load fill that could be removed
after the preloading period was limited to 1.40m because of the upward revision of finished ground
level. Thus the effective preloading suffered a deficiency of about 1.5 t/m2 equal to 0.80m thick fill
(Figure19) that caused this additional virgin compression.
However, it is very clear from the settlement observation in the front and rear of the building,
the settlements were very uniform. The main reason for such uniform settlement was very uniform
vertical progress of the construction possible for a load bearing wall construction. The initial design
was to have independent footings with RCC frame and infill walls that would have resulted very
non-uniform construction pattern. Having the vertical drains in position, the settlements could have
been very fast (also as seen from the time settlement curve in Figure 18) resulting large differential
settlements. Another reason for such uniform settlement is that the inverted ‘T’ RCC strip footing
with much higher stiffness contributed to a more uniform load distribution.
A detailed settlement analysis
revealed that the probable settlement due to EXTRA LOAD FROM THE
REVISED FGL
this extra loading is 80 to 105mm. More
settlement for one villa could have been the
SETTLEMENT
EXPECTED

ACTUAL
17 SETTLEMENT
result of compression of initial portion of the
preload fill that remained below the
founding level.
6.2.8 Further Constructions
However, such large settlements
could not be allowed because of the fact that
the RCC strip raft foundation was relatively
expensive and a uniform construction pattern
could not be enforced in the case of RCC
frame construction.
Even though the owner decided not to listen to the arguments, adopted required additional
pre-loading for the remaining sections of the project. The settlements observed during preloading
increased to 900 to 1000mm and the completed bungalows recorded settlements less than 30mm.
6.2.9 Conclusions
Ground improvement programmes like pre-loading with settlement accelerators require good
planning with respect to site development. All the possible changes in the development programme
shall be thoroughly investigated and the improvement shall take care of the worst case. Co-
ordination between the foundation consultant, the architect, the structural engineer and the owner is
a must in such programmes.
A thorough and systematic monitoring of improvement in terms of settlements versus the
load and time is needed.
Adoption of expensive RCC strip raft foundation for the area with under-improvement
helped in preventing large distress in the structure. However, adoption of similar measures for all the
villas could have affected the economy of the project.

6.3 The Case of Rotating Abutment Pier


6.3.1 Problem
Two abutment piers and three intermediate piers for a rail over bridge was constructed on
shallow footings placed at about 2.60m below the natural ground level. The construction was
planned without hindering the rail traffic and hence launching of main girders through rails
supported on approach embankments was in the offing. The general layout of the piers and the
embankment is shown in Figure 20.
Failed pier

Embankment fill

Embankment fill

Piers

18 piers and intermediate piers


Figure 20 Layout of the abutment
The embankment construction progressed from both the sides and when the embankment
near one of the abutment piers reached almost maximum height, the embankment fill failed. The top
of the abutment pier moved towards the embankment fill while the embankment cross section close
to the pier slipped by more than 1.50m (Figure 21). The embankment fill did not reach the abutment
rear face when failure occurred.
6.3.2 The Distress
Fill sandy clay, cu = 0.35
It was a clear case of base failure kg/cm2

of almost vertical embankment section


facing the abutment pier, without a Fill clayey sand, φ = 28º
support by the pier. A wedge shaped gap Sandy clay, cu = 0.70 kg/cm2
was provided between the abutment pier
and the embankment for the provision of Sandy clay, cu = 0.30kg/cm2

conventional filter layer. Rotation of the sandy clay, cu = 0.28 kg/cm2


abutment pier about its base and towards Soft clay, cu = 0.21 kg/cm2
the embankment clearly suggested the Sandy clay, cu = 0.65 kg/cm2
involvement of soil beneath the pier
foundation in the embankment base
Figure 21 Slip of the embankment
failure. The major distress was failure of
the abutment foundation because of its
rotation.
6.3.3 The Investigation
A possible slip circle was constructed based on the shape and position of the slip line. The
slip circle passed very close to the abutment pier foundation and the diameter of the slip circle was
roughly estimated as 36 metres. The design of pier foundation was done on the basis of net safe
bearing capacity equal to 18 t/m2 arrived at on the assumption that stiff sandy clay with good shear
strength present 2.60m below the ground level is continuing towards depth.
Further investigation through four exploratory boreholes and five dynamic cone penetration
tests revealed presence of soft clay and soft sandy clay with undrained shear strength in the range of
0.20kg/cm2 to 0.28 kg/cm2 between 3.0m and 8.0m below the ground level. The thickness of
relatively good bearing stratum between the founding level and the weak layer was less than 0.50m.
The soil profile and the failure imagined are illustrated in Figure 20. The detailed geotechnical
investigation data from one borehole is presented in Figure 21.

SPT / VST
Depth below
GL (metres)
Soil Profile

Test Depth

Field Description Blow Counts


m

15 30 45 60 N**

0.5 Greyish dry sandy clay


1.0 0.75 2 1 0 0 1
Brownish grey soft clay with sand pockets
1.5 2
1.50 Su=0.44kg/cm
2.0 Light grey soft clay with fine sand
2.25 Sunk @63.5kg 0
3.0
3.00 Sunk @63.5kg 0
Greyish soft sandy clay/clayey sand 2
4.0 3.75 Su=0.24kg/cm 0

4.50 Sunk @63.5kg 0


5.0 19
5.1
2
Dark grey soft clay 5.50 Su=0.31kg/cm 0
6.0
6.2 2
6.25 Su=0.38kg/cm 0
Figure 22 Soil profile 15m away form the failed pier foundation
6.3.4 The Analysis
Standard slip circle analysis resulted a minimum factor of safety of 1.29 based on the soil
data revealed from the investigation. Relatively high shear strength of 0.70kg/cm2 for the sandy clay
available up to 3.0m from original ground level is providing considerable resistance to failure. This
called for a re-look into the back analysis. It was then observed that the possible slip circle passed
through the interface between the backfill of pier foundation excavation and the original stiff sandy
clay. The large excavation made for the construction of the abutment pier left a weak zone having
very poor backfill in place of the stiff clay. A slip circle analysis was then carried out assuming only
about one sixth of the shear strength of the stiff sandy clay along the foundation excavation line, and
a factor of safety 0.83 was obtained.
The soil layers below the abutment foundation have very high compressibility and any load
over the abutment pier beyond the self weight of the pier should have resulted excessive settlement
and a bearing capacity failure.

20
6.3.5 Conclusions
The foundation provided for the abutment pier was inadequate in view of soft clay layers
immediately below the founding level. Very large load from 8.0m high embankment triggered the
failure. Major load over the abutment pier from the main girders, road formation and the road traffic
was due. Even if the failure of abutment was not happened during the embankment construction, a
major failure when the abutment was loaded fully was sure to happen.
Changes in the shear strength parameters because of possible construction activities like
foundation excavation and backfilling of such excavations are very relevant in similar cases.
This case illustrates clear failure from the part of the designer and the owner who went ahead
with the design and construction without ascertaining the soil conditions below the embankment and
the piers.

REFERENCES
Allen Marr, W. Ramos J.A, Lambe T.W. (1982), ‘Criteria for Settlement of Tanks’, Journal of the
Geotechnical Engineering Division, Proc. ASCE, Vol 108, No: GT8, Aug 1982, 1017-1039
Anirudhan I.V. (1997), ‘Driven cast-in-situ piles – Execution and performance’, Proc. Indian
Geotechnical Conference, IGC 1997, Vadodara, pp 233-236
Anirudhan I.V. (2005), ‘Types of distress in Geotechnical Structures’, Proc. Indian Geotechnical
Conference IGC 2005, Ahmadabad, pp 165-168
Boone S.J. (2001), ‘Assessing Construction and Settlement-induced Building Damage: A Return to
Fundamental Principles’, Proceedings Underground Constructions, Institution of Mining and
Metallurgy, London, 559-570
Boscardin M.D. & Cording E.J. (1989), ‘Building Response to Excavation Induced Settlement’,
Journal of Geotechnical Engineering, ASCE, 115(1), pp1-21
Burland J.B. (1997), Assessment of Risk of Damage to Buildings due to Tunnelling and
Excavation’, Earthquake Geotechnical Engineering, Ishihara (ed), Balkema, Rotterdam, 1189-1201
Burland J.B. Wroth C.P. (1974), ‘Settlement of Buildings and Associated Damages’, State of Art
Report, Proc. Conference of Settlement of Structures, Cambridge pp 611-654
Burland J.B., Broms B.B, DE Mello V.F.B. (1977), ‘Behaviour of Foundations and Structures’,
State of Art Report, Proc. 9th ICSMFE, Tokyo, Vol 2 495-546
Madhav M.R.(2003), ‘Modelling Methods in Geotechnical Forensic engineering’, Proc. of a
Workshop by Committee on Professional Practice of Indian Geotechnical Society, Chennai, Feb28 –
March 1, 2003, pp 75-81
Mair R.J. (2001), ‘Research on Tunnelling Induced Ground Movements and Their Effects on
Buildings – Lessons from the Jubilee Line Extension’, Proc of the Intnl Conference on Response of
Buildings to Excavation Induced Ground Movements, Imperial College, London, UK, July 17-18,
pp 3-26
Poulos H.G. (2003), ‘Á Framework for Forensic Foundation Engineering’, Proc. of Workshop on
Forensic Geotechnical Engineering, Committee on Professional Practice of IGS, Feb 28-Mar 1,
2003, pp 7-13

21
Radhakrishnan R and Anirudhan I.V. (2003), ‘Ground improvement with pre-fabricated V drains
and pre-load – A case study, Proc. Symposium on Advances in Geotechnical Engineering, SAGE
2003, IIT Kanpur, pp 426-430
Ramanatha Ayyar T.S. & Jaya V (2003), ‘Geotechnical Aspects of Mass Collapse of Shallow Wells
in Kerala During 2001’, Proc. of workshop on Forensic Geotechnical Engineering, Committee on
Professional Practice of IGS, Feb 28-Mar 1, 2003, pp 33-36
Ramaswamy, S.V. and Narasimhan, S.L. (1978), “Behaviour of Buildings on Expansive Soils –
Some Case Histories”, Jnl. Institution of Engineers (India), Vol. 58, Pt. CI 4, pp 141 – 46.

22
Chapter 4

FORENSIC GEOTECHNICAL ENGINEERING INVESTIGATIONS:


DATA COLLECTION

Peter Day
Jones & Wagener Consulting Engineers, Rivonia, South Africa, 2128
day@jaws.co.za

ABSTRACT: This paper describes the initial stage of a forensic geotechnical investigation,
namely the gathering of data on the site, the works and the failure for further detailed analysis. It
has been written as part of the effort by the ISSMGE Technical Committee TC40 to prepare a
handbook on forensic geotechnical engineering. The paper provides guidance to the investigator on
the objectives of the investigation, the nature of the data required, sources of information
to be considered and the recording and storage of data.

1. BACKGROUND

Forensic geotechnicalengineering deals with the investigation of failures of geotechnical origin,


not only from a technical viewpoint but also with the possibility of legal proceedings in mind (Rao,
2005).

Forensic investigations differ from conventional geotechnical investigations in that they are
retrospective. They seek to explain what has happened rather than to predict future
performance. A further distinguishing factor is that, following a failure, there is an urgency to clean
up the site and rebuild or repair the works. This limits the time available for investigation and makes
it essential that all relevant data is recorded before the evidence is removed.
The ideal outcome of the data collection stage of a forensic investigation would be to
have a body of information that is (a) as complete as reasonable possible, (b) accepted by all
parties as an accurate record of the facts and events and (c) is stored in an accessible and
readily understood way. This paper describes procedures aimed at achieving this outcome.

2. SCOPE OF INVESTIGATION

Although some geotechnical failures such as landslides occur in the absence of any human
intervention, most geotechnical failures involve both the ground (soil, rock and groundwater)
and the works (some man made structure or intervention). The works may be a structure that
imposes loads on the ground, the alteration of the surface geometry (cuts or fills), alteration
of drainage patterns or the creation of underground openings. Thus, the forensic geotechnical
investigation must include a study of the event or failure which gave rise to the
investigation, the site on which the failure occurred and the nature of the works. Only after
these three aspects have been investigated and recorded can post-failure diagnostic testing and
back-analyses commence.
2.1 The Failure
In any forensic investigation, it is essential that the circumstances and events surrounding
the failure are investigated and recorded as soon as possible, before any evidence is removed.
Details of this stage of the investigation will vary from case to case. Nevertheless, there are
common aspects that apply to all failure investigations. These include the conditions that
prevailed immediately prior to the failure, the sequence of events, and the condition of the works and
surrounding areas following the failure.

2.1.1 Circumstances prior to failure


Before one can establish the cause of the failure, it is essential to investigate and record the
condition of the works immediately prior to the failure. Typical factors to be recorded include:
¤ the stage of completion of the works at the time (see sub-section 2.3.3 below),
¤ the occurrence of accidental actions (impact, explosion, earthquake, flooding or water leakage,
etc) or abnormal loading,
¤ abnormal meteorological conditions (wind, snow, rainfall, temperature, etc),
¤ the results of any monitoring (pore pressures, deformations, settlements, anchor loads, etc),
and
¤ any early warnings of incipient failure (cracking of ground or structure, falls of ground,
changes in anchor loads, etc).
2.1.2 Sequence of events
Obtaining an accurate record of the sequence of events that lead to the failure will assist
greatly in determining the failure mechanism and, in many cases identifying the trigger. The
information to be recorded will vary from site to site. However, an attempt should be made to
obtain as much information as possible from the time when the first signs of distress were noted.
An attempt should be made to draw up a time line from which the sequence of events and the
speed of progression can be ascertained.
2.1.3 Resulting distress
The two sub-sections above deal with the “before” and “during” situations. This sub-section
deals with the “after” or post-failure state of the works and surrounding areas. An accurate
description of the distress caused by the failure may be pivotal in determining the value of any
claim for compensation or damages which may follow.
Unlike the “before” and “during” situations which must be investigated by reference to historical
records, the “after” situation can be directly observed, photographed and recorded. Note that
the recording of the condition of the works after the occurrence as described in this paper is
distinct from the detailed diagnostic tests that may be required to provide parameters for
back-analysis of the failure.
Typical information to be recorded may include:
¤ the extent and severity of distress,
¤ the magnitude of deformations and trajectory of movement,
¤ any signs of where rupture may have occurred
(slickensided shear zones, yielded construction elements, etc),
¤ indications of abnormal surface or subsurface water conditions (seepage, high water marks, etc),
¤ condition of any exposed rupture surfaces whether on the structure or in the ground,
¤ any deviations from expected ground conditions (paleo channels, intrusives, faults,
adverse jointing, seepage, etc), and
¤ damage or changes in surrounding areas (physical damage, settlement, lateral movement,
drop in groundwater levels, rupture of services, etc).
2.1.4 Sources of Information
When determining condition of the works prior to the failure, reference should be made to any
available reports (geotechnical, structural, etc), to construction records, progress payment
certificates, photographs, as-built drawings and any information that can be provided by site
staff.Depending on the circumstances, it may be necessary to obtain additional information
from outside the site such as water meter readings, flood levels, metrological data, etc.
Recent aerial photos or satellite images may also be of assistance.
Determining the sequence of events normally relies on eye-witness accounts and photographs
taken immediately prior to, during and immediately after the event. Care should be taken when
interpreting eye-witness accounts as vested interests may be involved. For this reason, it is
preferable to obtain information from as many site personnel as possible and preferably
not only from personnel employed by either the engineer, the owner or the contractor. Any
discrepancies in the information provided by witnesses should be revisited and clarified at the
time. Written records should be kept of eye-witness interviews and these should preferably be
signed by both the investigator and the witness. Where appropriate, verbatim transcripts or
recordings of eye-witness evidence should be kept.
Information of the condition of the works after the occurrence is generally obtained by
direct observation. This will typically include photography, sketches, written descriptions, post-
failure survey drawings, etc.
It is preferable that the post-failure information is jointly recorded by the parties involved
(owner, engineer, contractor, insurer etc) to minimise disputes at a later stage.
2.1.5 Potential failure mechanisms
The scientific method relies on a process of postulation and verification. Many scientific
endeavours have floundered as a result of failing to consider alternative postulates and gathering
only information which supports a particular point of view. In any scientific investigation, it is as
important to record both supporting data and data which is inconsistent with various postulates.
The purpose of this first phase of the investigation is to gather data. Back analyses and the
identification of the most likely failure mechanism(s) will follow at a later stage. In fact,
forming a fixed opinion on the cause or mechanism of failure may result in certain essential
information relating to possible alternative causes being overlooked. Nevertheless, it is
recommended that investigators should seek to identify all the potential triggers, sequences
of events and failure mechanisms. Simple logic is likely to eliminate many of these from the start.
An attempt should then be made to obtain the data which will enable the likelihood of any
plausible failure mechanism to be assessed at a later stage.

2.2 The Site


In order to carry out a competent analysis into the causes of the failure, information is required on
the site on which the works were undertaken. Much of this information should already be
available in the form of existing reports and other documents.
2.2.2 Essential Data
The essential data required will vary from site to site. In most instances, it will include:
¤ the location and extent of the site,
¤ surface topography and alterations thereto,
¤ surrounding services and development
¤ site description including vegetation, drainage, climate, previous land use, existing
development, etc,
¤ geological setting, site geology, regional geological structures (faults, folding, etc) and
seismicity,
¤ site stratigraphy including the identification of typical soil profiles for various areas of the site,
¤ detailed soil and rock profiles from boreholes, pits or other exposures at particular locations on the
site,
¤ information on groundwater including depth of water table(s), seasonal fluctuations, gradients,
flow characteristics, etc, and
¤ results of field and laboratory tests.

2.2.2 Sources of information


The main source of geotechnical information on the site should be the geotechnical and
geological reports prepared for the project. The absence of such reports may, in itself, be a
contributor to the failure.
The next most likely source of geotechnical information on the site is the construction records.
Much valuable information may be gleaned from site instructions, recorded founding depths, tunnel
face maps and other similar information.
On some projects, borehole core or samples from the original investigation may still be
available. This creates the possibility of re-inspecting the core or re-testing of samples.

2.3 The Works


In considering the works, there are three main aspects which require consideration. These are
(a) the works as designed, (b) the works as constructed and (c) if appropriate, the state of
completion of the works at the time of failure.
2.3.1 Works as Designed
Information on the works as designed is generally contained in the construction
drawings and project specifications. These documents specify the work to be carried out by
the contractor and provide a basis from which any deviations may be assessed.
Of equal importance are the calculations on which the design of the works was based. A
review of the design calculations will enable the investigator to ascertain whether the
conditions on site are consistent with the those assumed by the designer during the design
process.
2.3.2 Works as Constructed
The works as constructed may differ from the works as described on the construction
drawings and project specifications for a number of reasons such as on-site design
modifications, substandard materials or workmanship, geometric deviations (both construction
tolerance and setting out errors) and concessions granted to remedy non-compliance.
Identification of differences between the “as designed” and “as constructed” works will generally
require physical measurement, survey, inspection and testing on site. Supplementary
information can be obtained from site documents such as non-conformance reports, design
modification reports, site instructions, minutes of site meetings, etc.
2.3.3 State of completion
A significant proportion of all geotechnical failures occur during construction. Where this is the
case, it is essential to determine:
¤ the state of completion of the works at the time,
¤ applied loading (both self weight and imposed loads),

¤ changes in loading or construction activities


immediately prior to failure and
¤ strength of materials that show time dependency at the time of failure (eg clays
undergoing consolidation, freshly cast concrete or grout, etc).

3 RECORDING OF DATA

3.1 Attention to Detail


The forensic investigator should strive to ensure that all information obtained during the
course of the investigation is complete in every respect. Not only does this
require that the scope of the investigation is adequate, that all plausible failure mechanisms are
investigated and that all relevant documentation is considered, it also requires meticulous
attention to detail.
In this regard, it is recommended that written records be kept of all discussions and inspections
and that these records be dated and signed. All photographs should be uniquely referenced and
the date, time, location and orientation of the photographs should be recorded. Any samples
taken should be photographed in situ (prior to sampling), provided with a unique sample
number and records kept of the tests undertaken and the results obtained.

3.2 Agreement between Parties


Forensic geotechnical investigations may either be carried out by an independent investigator
or by different investigators employed by the various parties (employer, contractor, engineer,
insurer, etc).
Wherever possible, all parties should be afforded the opportunity to witness critical stages of
the investigation such as exhuming of foundations, taking of samples, removing collapsed
structures, etc and observations made during such crucial stages of the investigation should be
shared with all parties. This is to avoid the situation where crucial evidence is ruled to be
inadmissible during the legal process as the accuracy of the observations, location of the
samples or photographs and other such details can only be vouched for by one party.
In investigations of this nature, it is preferable that as much agreement as possible is
obtained during the investigation stage as it is likely that much of the evidence will have been
removed when the time arrives for resolving disputes or legal argument,.

3.3 Reporting and data storage


It goes without saying that all data obtained during the course of the investigation must
be adequately documented and stored in a well referenced and easily retrievable format. Even
the simplest of failure investigations can drag on for a number of years. Given the mobility in
the job market, it is likely that the data will be analyzed by a different team of engineers to
those responsible for its collection.
In the case of small investigations, it is preferable that a report be compiled which details the
extent and findings of the investigation and appends any relevant supporting documents.
In the case of larger investigations, the report is more likely to be a summarised version of
events and an index of the various supporting documents. On major projects, the number of
documents may run into the hundreds and possibly thousands. Under these circumstances,
electronic storage of documents should be considered due to the ease of reproduction and portability
of electronic records.

4 CONCLUDING REMARKS

The gathering of data for a forensic geotechnical investigation should be conducted with
an open mind. Collection of selective data intended to support a particular hypothesis may
be counterproductive in that the data so collected will be inadequate to test the veracity of
alternative theories. It is as important to collect supporting evidence as it is to note
evidence that is inconsistent with postulated failure mechanisms.
The investigator should be mindful of the fact that the data gathered may be subject to scrutiny in
subsequent legal proceedings. Obvious bias in the collection and reporting of data will discredit the
findings of the investigation.
Where the investigator has access to all parties (e.g. in the case of a joint appointment),
consultations should be held with all parties involved. More often than not, follow-up consultations
will be required as information obtained from one party may be queried or refuted by
another. Wherever possible, an attempt should be made to obtain agreement between the parties
on important issues at the time as this reduces the amount evidence to be lead in any future legal
proceedings.

4 REFERENCES

Rao, V.V.S. (2005) –TC40 Terms of Reference. Report submitted to ISSMGE, December 2005.
Chapter 5

COMPILATION OF DATA
Dr.V.V.S.Rao,
Nagadi Consultants Pvt. Ltd., New Delhi, INDIA
dr_rao@nagadi.co.in

INTRODUCTION:
Although some geotechnical failures such as landslides occur in the absence of any human
intervention, most geotechnical problems involve both the ground (soil, rock, and ground water)
and the works ( man made structure or intervention). The works may be a structure that imposes
loads on the ground, the alteration of the surface geometry (cuts or fills), alteration of drainage
patterns or the creation of underground openings. For a forensic analysis of such problems the
first step is to compile all available data about the project. This compilation begins with the
detailed description of the distress or malfunction of the structure which has been attributed to
geotechnical causes. The distress or malfunction can be a complete failure, or excessive
deformations, or unacceptable responses to vibrations, or excessive seepage in case of water
retaining structures.

TYPES OF DISTRESS:
a. Complete failure
In the case of a complete failure the stress conditions in the supporting or participating soil have
reached their ultimate resistance level and the structure has collapsed. Typical examples are
foundation failures, collapse of retaining structures, land slides, and slope failures in cuts and
embankments. Generally, water seeping through the slopes is the culprit. In case of natural
slopes like in hills, the water seeping through the porous seams and rock joints play havoc.
Piping action due to poor compaction in some pockets of an earthen dam body can even lead to
its sudden collapse.
b. Excessive deformations
A soil-structure system is always designed to tolerate certain magnitude of deformations. These
deformations include both total deformations occurring in the system as a whole and differential
deformations between elements of the structural system. We are mainly concerned with the
deformations caused by the supporting soil medium. These deformations are combinations of
immediate ones which occur within a short time after the forces are transferred to the soil and the
ones which occur slowly over a period of time after the commissioning of the project. The soil
strata which has cohesionless soil within the influence zone undergoes mainly immediate
settlements while in the cohesive strata the deformations are time dependent. In both cases the
major causation factor is the water. This water may be from the ground water table fluctuation
or/and ingression from external sources. Normally, immediate settlements which can be
estimated and provided for in the designs are acceptable. Differential settlements may lead to a
condition when the structure can not be utilized fully for the designed purpose.

c. Excessive vibrations
In case of industrial as well as laboratory structures having testing facilities with sensitive
instrumentation, vibrations of individual foundations due to forced excitation on them and also
due to vibrations transmitted through the soil may not be acceptable. Similarly, in case of
structures built in seismic sensitive zones, the effect of earthquakes can lead to damage to
foundations including excessive tilt/collapse of the structure due to liquefaction. In all these
cases, the response of the soil-structure should be within tolerable limits.

d. Seepage problems.
In case of water retaining structures like dams and water/sewage treatment plants at ground level,
the water seeping through the soil strata may cause erosion due to piping, flow of soil, or
excessive seepage in spite of foundation treatments. The seepage problems are particularly
common during rains.

CAUSES OF DISTRESS
During the data collection stage it will not be possible to identify the primary cause of distress.
However, as the data collected should be comprehensive and sufficient to analyze the problem
from all angles. A study of several published case histories suggest the primary causes can be
grouped under five headings:
a. Underestimation of forces
A design engineer normally assumes certain magnitudes of different types of forces that may act
on the soil-structure system, depending upon relevant standard codes and also upon his
experience. As “Engineering Judgment” is involved, there can be difference of opinions between
designers at this stage. It is advisable to highlight the forces considered in the design.
b. Meteorological data
Factors like high flood level, maximum intensity of rains/snow, direction and velocity of winds,
and variations in daily as well as seasonal temperatures, etc. are used while selection of design
forces. Proper selection of their magnitudes is important for design.

C .Inadequacy of geotechnical investigations


Trying to economize on investigations leads to lack of sufficient data for scientific evaluation of
subsoil properties.
d. Design soil parameters
This is one of the critical stages in design. Improper values may bring disaster.
e. Improper design criteria
At this stage important decisions regarding tolerable deformations and allowable stresses on the
materials to be used are made. As described above, the designers’ engineering judgment plays a
very important role. The “efficiency” of the structure depends on this stage.
f. Inappropriate/inadequate design
The design theories to be adopted depend upon the type of sub-soil strata and substructures.
Adopting design methods irrelevant to the type and nature of the subsoil will lead to inadequacy
of the system response. Soil-structure interaction is another design aspect which needs to be
considered in depth. Short-cut methods like empirical designs should be strictly avoided.
g. Improper construction methodology
Construction methods and equipment needed for the same depend directly upon the subsoil
conditions. Contractors having suitably skilled labor should have been selected for the
construction.

DATA TO BE COLLECTED
A. Predesign stage
This is mainly desk work. The history of the project, from the concept stage up to the selection of
the site form the first part of the study. The factors considered while selecting the site as also the
persons involved in the process should be identified. In the second part, the data collected for the
design including their sources should be compiled. The data to be compiled include:
a. Detailed topographical survey,
b. Metrological including hydrological data,
c. Results of preliminary soil survey data from trial pits , shallow boreholes, etc.,
d. Sources of energy and water, and,
e. Details about equipment and machinery to be installed.

B. Design stage
As this is the crucial stage, all data, even though on the outset looks superfluous, should be
collected. The important ones are:
a. Data regarding the structural aspects including the details of the machinery, if any,
b. Detailed report of subsoil investigations:
1. Locations of points of investigation w.r.t. structures
2. Type of exploration like, boreholes, penetration tests, load tests, etc,
3. Depth to which explorations are done,
4. Results of field and laboratory tests: their adequacy and accuracy w.r.t. the
importance of the structures,
5. Interpretation and analysis of all results, theories used for analysis,
6. Selection of strength as well as consolidation parameters of soil strata
c. Designs including assumptions regarding loads, tolerances in deformations, and strengths
of all building materials used, along with calculations.
d. Drawings, whether all information needed for execution, including the precautions to be
taken, etc. are included,
e. Construction sequence to be followed.
f. Proof checking of all designs
g. Approval of the project authority

C. Construction stage.
a. Selection of suitably qualified and experienced executing agency
b. Contract agreement
c. Project organization, field and office. The responsibilities and liabilities of all involved
persons should be clearly identified.
d. Monitoring the construction, verification of soil conditions
e. Works as designed vis-à-vis as constructed:
Information on the works as designed is generally contained in the construction drawings
and project specifications. These documents specify the work to be carried out by the
contractor and provide a basis from which any deviations may be assessed. The works as
constructed may differ from the works as described on the construction drawings and
project specifications for a number of reasons such as on-site design modifications,
substandard materials or workmanship, geometric deviations (both construction tolerance
and setting out errors) and concessions granted to remedy non- compliance.

f. Quality control during construction:


Details regarding the personnel who did the quality control tests and the methods adopted
need to be documented. Detailed documentation should have been maintained.
g. Metrological conditions during the constructionperiod
h. Details regarding the instrumentation used for maintaining the accuracy of construction
and also for monitoring the behavior of the structure along with the readings and their
interpretation should have been maintained.
i. Construction sequence and progress reports: detailed documentation is necessary
j. Completion report:
The design engineers should conduct detailed and thorough checking of the construction
to verify whether all the design details have been properly adhered to in the construction.
k. Photos and video recordings.

D. Post construction stage:


a. Trial runs (first loadings), instrumentation, observations and conclusions
b. Final approvals
c. First observation of distress/ malfunction
d. Observation of progress of distress, instrumentation and records
e. Emergency remedial measures taken.
CHAPTER 6

LABORATORY TESTS IN FORENSIC INVESTIGATIONS

Robinson R. G.
Associate Professor, Dept. of Civil Engineering, IIT Madras, Chennai-600 036
robinson@iitm.ac.in

ABSTRACT: Forensic geotechnical engineering involves systematic scientific investigations


to detect the causes of failure or distress in a structure. Scientific method, involving the
postulation of probable hypothesis and proving the hypothesis, is generally adopted in
forensic investigations. In order to prove the hypothesis very often field and laboratory
testing is needed. Conventional laboratory testing are also often very important. In this paper
some of the aspects to be considered while analysing the experimental data obtained from
laboratory studies are outlined.

1. INTRODUCTION

Field testing and laboratory testing form one of the major components of forensic
investigations. The objective of the testing program is to determine the in-situ properties of
the soil and the cause for distress. The field tests can be divided into two categories: non-
destructive testing and destructive testing. Nondestructive testing such as geophysical surveys
can be used to evaluate the ground conditions without causing damages to the site. The other
category of non-destructive testing involves the determination of elevation of the ground and
structure that will be used to estimate any settlement or heave.

Destructive testing involves removing a limited section of the building or undergoing


subsurface explorations. Details of destructive testing are given in Day (1998). The
geotechnical forensic engineer has various choices among the following:

• Borehole investigations including SPT


• Vane shear
• Pressuremeter test
• Flat dilatometer
• Cone penetration tests like static cone, dynamic cone, piezocone, seismic cone,
nuclear density cone, Video cone, etc.
• Plate load tests
• Insitu density measurements, etc.

It is very important to note that the engineering properties derived from the field tests
depends strongly on the type of test and method of analysis. Therefore, they should be used to
test the hypothesis with caution. Also the following aspects should be considered during
interpretations.

• Variations due to seasonal change


• Variation in ground water table
• Variation due to the new stress path the soil has undergone due the construction of the
structure in question
• Variation due to any environmental change like chemical contamination, temperature
change, etc.
2. LABORATORY TESTS

The common laboratory tests in geotechnical investigations are:

(i) Classification Tests such as Specific gravity, Atterberg limits, Grain size distribution, etc.
(ii) Strength Tests such as Direct shear, Triaxial shear, plane strain and simple shear
(iii) Consolidation tests such as anisotropic consolidation, isotropic consolidation and can be
either stress controlled or strain controlled.
(iv) Permeability tests
(v) Compaction tests, etc.

The in-situ properties may differ from properties that were used in the design. It is
essential to evaluate whether the difference is caused due to the construction of the structure
in question or due to other factors. Some aspects to be considered while interpreting the
results are discussed below.

2.1 Classification tests

It is very important to evaluate the pre-test conditions adopted while doing the
classification tests. For example the classification test results strongly depend on the degree
and method of drying. It is the usual practice to oven dry the soil sample before conducting
classification tests, though the code of practice suggests appropriate drying techniques.
Liquid limit, plastic limit and shrinkage limit of some soils are severely affected by drying as
can be seen in Table. 1. The grain size distribution is also significantly affected. Therefore, if
any correlations were used in the design based on the index properties, the pre-test condition
should be properly analysed. In addition, the soil classification also may get changed. For
example, as per Indian Standard Classification system, the air-dried sample of Parur clay, in
Table 1, falls as CH in the plasticity chart. However, the oven dried sample at 105 oC is
classified as MH.

Table 1. Index properties of Parur Clay (data from Pandian et al. 1993)

Condition wL wP PI Grain size Distribution


(%) (%) (%) Clay size (%) Silt size (%) Sand size (%)
Natural 106 47 59 51 42 7
Partially air dried 91 40 51 32 48 20
o
Air-dried (25-30 C) 85 34 51 21 52 27
o
Dried at 60 C 70 32 38 16 43 41
o
Dried at 105 C 60 32 28 15 39 46

The values of liquid limit also depend on the type of apparatus used for the
evaluation. Sridharan and Prakash (2000) observed that the percussion method gives higher
liquid limit values for the montmorillonitic soils than the cone method and that the cone
method gives higher liquid limit values than the percussion method for kaolinitic soils. It may
be noted that the plasticity chart is based on the liquid limit value obtained from
Casagrande’s apparatus. One may end up with different classification depending on the type
of test adopted.

2.2 Compaction Tests

Very similar to the index properties, compaction characteristics also are affected by
the pre-test conditions. If the compaction tests were conducted in the laboratory on oven-
dried soil samples, the results cannot always be expected in the field that were initially wet
but allowed to dry up to the optimum moisture content. Wesley (1973) reports such examples
on allophone clays. It is also to be noted that over-compaction reduces the strength.

Fig. 1 Standard compaction curves for allophone clay (Wesley 1973)

2.3 Strength Tests

Soils in general are not elastic materials and their behaviour in the field depends on
many factors including the magnitude of the imposed stress changes; the way in which they
change; drainage conditions and the previous history of loading, etc. When a load is applied
or removed from a mass of soil in the ground by a foundation or excavation, respectively,
each element of soil experiences changes in its state of stress. A stress path gives a
continuous representation of the relationship between the components of stress at a given
point as they change (Lambe 1964, Lambe and Whitman 1969, Lambe and Marr 1979). Use
of a stress path provides a geotechnical engineer with an easy recognisable pattern which
assists him in identifying the mechanism of soil behaviour. It also provides a means of
selecting and specifying the sequences of stresses to be applied to a sample in a test for a
particular purpose. In geotechnical engineering practice, if the complete stress path of the
problem is understood one is well along the way towards the solution of that problem (Holtz
and Kovacs 1981).

It is essential to consider the drainage conditions, rate of loading, stress


controlled or strain controlled tests, degree of saturation, etc. It is also well established
in the literature that the type of test also has influence on the shear strength values
(Leonards 1982) and the tests appropriate to the situation is to be selected. In practice,
the soil is subjected to stress increments. But the standard tests are performed in a
strain controlled apparatus. The effect of the type of loading is illustrated in the
following fig.

The other important factor that affects the engineering properties is the sampling
disturbance. Method of boring and sampling and their influence on the parameters are also to
be analysed.

2.4 Volume change and Permeability

The volume change behaviour of the soil depends on the type of clay mineral present
in the soil and the pore medium. It is well known that the swell shrink behaviour of soils is
due to the presence of the clay mineral montmorillonite. However, the non-swelling soils
under normal conditions can exhibit heave when exposed to chemical solutions. For example,
Rao and Rao (1994) reported a case study where the kaolinitic soils, commonly classified as
non-swelling, showed excessive have and damage to structures due to caustic soda leakages
on the ground.
The consolidation and permeability characteristics also depend to a large extent on the
pore medium chemistry. Therefore, the possibility of volume change due to chemical
alteration due to contamination of chemicals cannot be ruled out and due consideration be
given in laboratory evaluations in such situations.

3. EFFECT OF WATER TABLE FLUCTUATION

A single story building, founded on a silty stratum experienced severe distress with
many cracks on the walls (Fig. 2a). Investigations indicated that the foundations are adequate
in terms of bearing capacity, considering the worst case of water table on the surface.
Consolidation tests indicated that the soil is overconsolidated and the settlement is within the
permissible limits. It was observed that the water table fluctuates between the ground level
during wet season (Fig. 2b) to a depth of 3 m during summer. This was confirmed by
observing the water table levels in the nearby wells. The effect of water table fluctuation on
the distress needs proper evaluation through experimental study.

Fig. 2a Crack in the building Fig. 2b View of the test pit

4. EXPANSIVE SOILS

4.1 Identification:

Index properties and differential free swell tests are commonly used to identify
expansive soils. It is often observed that the sites with soil that were classified as soils of low
potential for swelling cause severe distress (Fig. 3). The basic properties of the soil are shown
in Table. 2. From the table, the soil may be classified as Soil of low to marginal potential for
swelling. However, the damage to the floors is severe. Probably, the sample needs to be
examined by subjecting it to repeated cycles of wetting and drying.
Fig. 3 Distress due to swelling soil

Table 2. Index properties of soil

Property Value
Atterberg Limits
Liquid limit (%) 39
Plastic limit (%) 20
Plasticity Index (%) 19
Shrinkage limit (%) 10

Grain Size Analysis


Sand fraction (%) 40%
Silt size (%) 35%
Clay size (%) 25%

Compaction Characteristics
Maximum dry unit weight (kN/m3) 17.0
Optimum moisture content (%) 21

Swelling Characteristics
Differential Free Swell (%) 20
Swelling Protential (%) 0.6

Potential for swelling Low to marginal

4.2 Swelling Pressure

An important variable required in the prediction of heave in swelling soils is the


swelling pressure, which is the pressure required to hold the soil at constant volume when
water is added. Knowledge of swelling pressure is essential for the design of a variety of
geotechnical structures on expansive soils. The swelling pressure is evaluated in the
laboratory by a number of testing methods which include oedometer testing of samples,
suction measurements, triaxial methods, etc.

Out of all the methods, laboratory oedometer testing method is extensively used to
determine the swelling pressure due to its simplicity and operational ease. Brackley (1973)
lists three different oedometer methods for the determination of swelling pressure as
explained below.

Method A - The sample is inundated and allowed to swell vertically at a small seating
pressure until primary swell is completed. The sample is then loaded in intervals similar to
the procedure of conventional consolidation testing until the specimen reaches its initial
thickness. The pressure required to bring back the sample to its initial thickness is regarded as
the swelling pressure. This method is also often termed as Swell-consolidation method.
Method B - Three identical samples are loaded with different pressures near the expected
swelling pressure and submerged in water. The vertical movements were plotted against the
applied pressure and the pressure corresponding to zero volume change is taken as swelling
pressure. While only one sample is enough to determine the swelling pressure in method A,
at least three identical samples are needed in method B. This method is also often called as
Different pressure method.
Method C – In this method, also called as Constant volume method, a specimen is
maintained at constant height by adjusting the vertical pressure after the specimen is
inundated in free water. The pressure required to maintain constant volume is the swelling
pressure.
Each of the methods is equally sensible, but gives entirely different swelling pressure
values for the same placement conditions of the soil. A number of investigators have
attempted to study the cause for the variation of the swelling pressure values by these
methods. Johnson and Snethen (1978) compared the swelling pressure values by different
oedometer methods and found that the magnitude of swelling pressure depends on the degree
of confinement. Ali and Elturabi (1984) conducted methods A and C for the measurement of
swelling pressure of expansive soils. Results obtained show that method A gives higher
swelling pressure values than method C. Sridharan et al. (1986) compared the results from the
three oedometer methods (Methods A, B and C) to determine the swelling pressure and
concluded that method A gives an upper bound value, method B gives the least value and
method C gives intermediate values. They also found no definite relation between the three
methods.

Soundara and Robinson (2009) also observed that the swelling pressure depend on the
test method, as can be seen in Table. 2. The reason for this difference was attributed to the
structure change that occurs during the tests. Therefore, it is essential to identify which
method that was used to evaluate the swelling pressure during an investigation.

Table 3 - Comparison of Swelling Pressure values by different methods


(Soundara and Robinson, 2009)

Sample Swelling pressure (kN/m2) by Method


A B C
A1 325 160 185
A2 270 150 175
B1 240 160 175
B2 210 115 140
B3 60 - 50
5. INTERFACIAL FRICTION ANGLE

The interface friction between soils and construction materials is a very important
input parameter required for the design of a variety of geotechnical structure. Studies
indicated that the mode of shear significantly affect the magnitude of interface friction
(Subba Rao et al. 1996). In the direct shear mode, if the test material is placed above the solid
material, the limiting maximum value of interface friction is the critical state angle of internal
friction of the soil. However, if the test material is placed below the soil, the limiting
maximum value is the peak angle of internal friction of the sand.

6. SUMMARY

Laboratory testing and field studies are important components of forensic investigations.
Many of the parameters obtained in the laboratory depend on the pre-test conditions and also
method of testing. The influence of pre-test conditions and the type of tests adopted needs to
be carefully analysed in the forensic investigations.

7. REFERENCES

Ali, E.F.M., and Elturabi, M.A.D. (1984). “Comparison of two methods for the measurement
of swelling pressure.” Proc. of 5th Int. Conf. on Expansive soils, Adelaide, Australia,
84/3, 72-74.
Brackley, J.J.A. (1973). “Swell pressure and free swell in compacted clay.” Proc. of 3rd Int.
Conf. on Expansive soils, Haifa, 1, 169-176.
Chandrasekaran, V. S. (2003). Centrifuge Modelling: A useful technique for forensic
geotechnical Engineering. Proc. Workshop on Forensic geotechnical Engineering,
Chennai, pp. 57-60.
Day, R. W. (1998). Forensic Geotechnical and Foundation Engineering. McGraww-Hill.
Holtz, R. D. and Kovacs, W. D. (1981). An Introduction to Geotechnical Engineering,
Prentice-Hall, 733 pp.
Johnson, L.D., and Snethen, D.R. (1978). “Prediction of potential heave of swelling soils”.
Geotech. Test. J. ASTM, 1, 117-124.
Lambe, T. W. (1964). Stress path method. Jl. Soil Mech. and Found. Div., ASCE, 93(SM6),
309-331.
Lambe, T. W. and Whitman, R. V. (1969). Soil Mechanics, John Wiley and Sons, New York.
Lambe, T. W. and Marr, W. A. (1979). Stress Path Method: Second Edition. Jl. Geotech.
Engg. Div., ASCE, 105(GT6), 727-738.
Leonards, G. A. (1982). Investigation of failures. Jl. Of Geotechnical Engg. Div., ASCE, Vol.
108, GT2, 187-246.
Pandian, N. S., Nagaraj, T. S. and Babu, G. L. S. (1993). Tropical Clays. I: Index Properties
and Microstructural Aspects. Jl. of Geotech. Engg. ASCE, Vol. 119, No. 5, 826-839.
Rao, S. M. and Rao, K. S. S. (1994). Ground heave from caustic soda leakages- a case study.
Soils and Foundations, Vol. 34, 13-18.
Sridharan, A., Rao, S.A. and Sivapullaiah, V. (1986). “Swelling pressure of clays”. Geotech.
Test.J., ASTM, 9(1), 24-33.
Sridharan, A. and Prakash, K. (2000). Percussion and Cone Methods of Determining the
liquid Limit of Soils. Geotechnical Testing Journal, Vol. 23, No. 2, pp. 242–250.
Soundara, B. and Robinson, R. G. (2009). Influence of test method on swelling pressure of
compacted clay. International Journal of Geotechnical Engineering, Vol. 3, No. 3, pp.
439-444.
Subba Rao, K. S., Allam, M. M. and Robinson, R. G. (1996). A note on the choice of
interfacial friction angle. Geotechnical Engineering, ICE, London, 119(2), 123-128.
Wesley, L. D. (1973). Some basic engineering properties of halloysite and allophone clays in
Java, Indonesia. Geotechnique, Vol. 23, No. 4, 471-494.
CHAPTER 7

BACK ANALYSES IN GEOTECHNICAL ENGINEERING

G L Sivakumar Babu and Vikas Pratap Singh


Department of civil Engineering
Indian Institute of Science
Bangalore 560012

1. Introduction

Back-analysis is an approach commonly used in geotechnical engineering to estimate


operable material parameters in situ (Deschamps and Yankey 2006). Further, back
analyses are required to provide technical evidences to prove or to disprove the
hypotheses made on the cause of failures and to establish scenarios of failure (Hwang
2008). The approach of back analyses is popular because there are significant limitations
in the use of laboratory and in-situ test results to accurately characterize a soil profile.
Back analyses have been commonly used to study the causes of failures in geotechnical
engineering applications such as slope stability, landslides, earth retaining structures,
dams, highways and foundations. Back analyses are being used extensively in the
geotechnical engineering practice, several studies including Leroueil and Tavenas (1981),
Leonards (1982), Stark and Eid (1998), and Tang et al. (1999), describe various
applications and the limitations of back-analyses.

2. Basic Considerations

In accordance with Hwang (2008), following points shall be considered while performing
back analyses for the given problem.

1. Back analyses should be performed for ‘as-built” conditions because many of the
assumptions made in design are either non-existing or different from reality.

2. Reconnaissance of the site must be done preferably jointly with all the parties
involved so as to sort out differences in opinions, if any. Site observations may be
captured in the form of photographs/videos that plays a vital role in decision making
and provides support to the conclusions drawn.

3. To ensure that the results of the analyses are reliable, the data available must be
carefully verified. This includes an appropriate appraisal of local geology of the site
of interest to help in understanding actual ground conditions and related historical
events.
4. Design drawings and calculations, if available, must be checked to ensure that the
works have been executed in accordance with appropriate design.

5. Depending upon the complexity of the problem, analyses can be performed by using
one or more of the following: (a) rules of thumb that includes indices such as stability
number and overload factor, (b) empirical relationships, (c) closed from solutions, (d)
simple numerical models, and (e) sophisticated numerical models.

6. While using numerical methods (generally using commercial software packages), an


understanding of the suitability, capability and the limitations of the method for the
particular problem must be developed and the judicious interpretation of the output
shall be made by the expert analyst. The software package adopted for analyses must
have sufficient technical backup.

7. Depending on the algorithm and the numerical scheme adopted, different software
packages may lead to drastically different results. Therefore, it is necessary to
conduct a few test runs so as to calibrate the parameters to be adopted by comparing
the results of analyses with observations or with known solutions.

8. Following important source of errors in the numerical analyses shall be given due
importance while drawings conclusions and final recommendations based on the
interpretations from the computed output: (a) implications of modeling of a 3D
system as a 2D model, (b) modeling time dependency of soil behaviour (e.g., rate of
dissipation of pore water pressures), and (c) modeling of the nonlinearity of the soil
behaviour (i.e. choice of an appropriate constitutive soil model).

3. Scope

Follwing are some the areas that can be identified to have scope for back analyses in
geotechnical engineering applications:

1. Back analyses may be used to study the settlement response of different types of
foundations and structures, classification of cracking damage, backfill settlement, to
identify causes of settlements such as limestone cavities or sinkholes, consolidation of
soft and /or organic soils, underground mines and tunnels, extraction of ground water
or oil, landfills and decomposition of organic matter.

2. Back analyses may used to study failures/behaviour of structures founded on


expansive soil by analyzing factors such as lateral and vertical movements of
expansive soils, special considerations in design of foundations and pavements on
expansive soils and treatment measures.

3. Back analyses may be used to study potential causes of lateral movement in


applications such as rock falls, surficial slope failures, landslides, slope softening and
creep, and dam failures.
4. Back analyses may be used to earthquakes induced phenomenons such as surface
faulting and ground rupture, liquefaction, slope movement and settlement, and
foundation behaviour.

5. Back analyses may be used to study bearing capacity failure of buildings, roads,
retaining walls and historical structures.

6. Back analyses may be used to study problems in structures such as slopes and
pavement due to the ground water and presence of moisture.

4. Examples of Potential Errors

Deschamps and Yankey (2006) provided a few project examples to illustrate the
challenge and potential errors that can be present in back-analysis for material strengths,
rupture (slip) surface, pore pressures, and three-dimensional or “End Effects”. A brief
overview of the examples related to material strength and 3D effects from the study of
Deschamps and Yankey (2006) is presented below.

4.1. Material strengths

This example was drawn from a case history related to the Grandview Lake Dam located
in Bartholomew County, central Indiana to illustrate the dependence that the back-
calculated strength along a weak seam has on assumptions of strength in other zones. It
was desired to perform back-analysis for estimating the strength along the planer slip
surface. An assumed rupture surface was developed from inclinometer measurements.
The dam was constructed primarily of glacial till and residual soils weathered from
claystones. The first challenge was to select the operable strength of the dam materials.
There was no distinct zonation of materials in the dam, and therefore, no basis for
subdividing the dam into discrete materials.

To select the operable strength of the dam materials, results of the consolidated
undrained (CU) triaxial tests were used. Four different characterizations of embankment
strength were considered for back analyses purposes to characterize weak seam strength.
A summary of the back-calculated friction angles in the weak seam is shown in Table 1.

The strength along the weak seam was characterized as having a zero cohesive
intercept because it was rationalized that this material was at or near its residual strength
because the deformations along the very thin seam were significant, at least several
inches. Moreover, there is little tendency for volume change at residual conditions such
that shear-induced pore pressure changes were considered negligible

Table 1. Back-calculated strength (Deschamps and Yankey 2006)

Embankment strength Back calculated friction angle (degree)


Lower bound 22-24
Upper bound (high friction angle) 16
Upper bound (high cohesion) 11
Average 18

Table 1 illustrates the range in back-calculated friction angles is from 11°


to 24°, with the average strength providing 18°. Although this range can be viewed as
extreme, it is apparent that even if a narrower range of strengths were used to characterize
the dam, the back-calculated strength would still vary over an appreciable range. A
conservative (low) estimate of embankment strength leads to a relatively high
interpretation of strength along the weak seam. Note also the significant difference in
back-calculated strengths for the two upper bound cases, wherein the case with primarily
cohesive strength leads to a much lower back-calculated friction angle for the geometry
considered because of the higher shear strength of the compacted materials.

4.2. Three-dimensional or “end effects”

This example was drawn from a case history related to the ‘Lock and Dam 10’ on the
Kentucky River to demonstrate the importance of understanding of three-dimensional
effects when back-calculating strengths. ‘Lock and Dam 10’ is a relatively small concrete
gravity dam owned by the Commonwealth of Kentucky and built circa 1905. The dam is
a spillway over its complete length of 240 ft; it has a height of 34 ft; a width of 32 ft; and
is made up of ten monoliths 24 ft long. Follwing are key features of the study:

Problem: Stability analyses are conventionally performed on idealized two-dimensional


cross sections, which are based on plane strain conditions. At ‘Lock and Dam 10’, coring
through the dam indicated that the construction joints between concrete monoliths were
essentially rubble, and could not be relied upon as shear connections between monoliths.
Although it was considered imprudent to rely on the shear resistance between monoliths
as a design consideration, it was recognized that some resistance was likely to be present.

Back Analyses: An attempt was made to estimate the magnitude of this resistance in order
to understand the inconsistency between required strength and interpreted strength.
Accordingly, numerical modeling with the program FLAC was used to model the entire
dam as a beam. The assumption was made that only a nominal frictional resistance (35°)
was available between monoliths (no tensile or cohesive strength) and that the ends were
fixed at the ends. Figure 1 illustrates the idealized model of the dam taken in plan view.
The dam is attached to an abutment and training wall on the left, and the lock river wall
on the right, both assumed to be stable. The distributed load on the beam was
progressively increased to represent increasing the net hydrostatic pressures from higher
pools.
Figure 1. Plan view of Kentucky River Dam No. 10, modeled as a beam (Deschamps and
Yankey 2006)
Interpretation: The modeling effort produced a surprising result in which a zone within
the dam formed a compressive arch that developed significant flexural resistance. Based
on this analysis, the compressive arch that develops has sufficient capacity to carry the
complete hydraulic load acting on the dam during the maximum design flood,
independent of any frictional resistance at the base, and with only frictional resistance
between monoliths. This example clearly illustrates how difficult it would be to back-
calculate strengths if there is a significant, but uncertain, three-dimensional influence.
Although, the three-dimensional effects are extreme in the present case, influences of 5 to
30% are generally believed to be expected.

5. Concluding Remarks

Back analyses provide a means to analyze the failures that frequently occurs in various
technical fields including geotechnical engineering. It is of prime importance that back
analyses must represent the in situ conditions to the extent possible. Choice of back
analysis methodology must be based on the technical inputs and data available from the
failure site, detailed laboratory investigations, and other factors such as complexity of
problem, availability of experts, and the cost of analyses. Back analyses also form the key
step in the field of forensic studies, and provides basis for the techno-legal aspects in
geotechnical engineering.

6. References

Deschamps, R. and Yankey, G. (2006). Limitations in the back-analysis of strength from


failures. ASCE Journal of Geotechnical and Geoenvironmental Engineering, 132(4),
532-536.
Hwang, R. N. (2008). Back Analyses in Forensic Geotechnical Engineering.
ISSMGE/TC40 - Forensic Geotechnical Engineering. (Draft: 2008/10/20), 1-8.
Leonards, G. A. (1982). Investigation of failures. ASCE Journal of Geotechnical
Engineering Division, 108(2), 185-246.
Leroueil, S., and Tavenas, F. (1981). Pitfalls of back-analysis. Proceedings of the 10th
International Conference on Soil Mechanics and Foundation Engineering, Balkema,
Rotterdam, 1, 185–190.
Stark, T. D. and Eid, H. T. (1998). Performance of three-dimensional slope stability
methods in practice. ASCE Journal of Geotechnical and Geoenvironmental
Engineering, 124(11), 1049-1060.
Tang, W. H., Stark, T. D., and Angulo, M. (1999). Reliability in back analysis of slope
failures. Soils and Foundations, 39(5), 73–80.
Chapter 8

Back Analysis of Slope Failures to Design Landslide Stabilizing Piles


M. E. Popescu, Ph.D., P.E., Eur.Ing.
Parsons Brinckerhoff Americas Inc. / Illinois Institute of Technology, Chicago, Illinois, U.S.A.
V.R. Schaefer, Ph.D., P.E.
Iowa State University, Ames, Iowa, U.S.A.

ABSTRACT: It is generally accepted that shear strength parameters obtained by back analysis of slope
failures ensure more reliability than those obtained by laboratory or in-situ testing when used to design reme-
dial measures. In many cases, back analysis is an effective tool, and sometimes the only tool, for investigating
the strength features of a soil deposit. The fundamental problem involved is always one of data quality and
consequently the back analysis approach must be applied with care and the results interpreted with caution.
Procedures to determine the magnitude of both shear strength parameters (c' and φ') or the relationship be-
tween them by considering the position of the actual slip surface within the failed slope are discussed Using
the concept of limit equilibrium the effect of any remedial measure (drainage, modification of slope geome-
try, restraining structures) can easily be evaluated by considering the intercepts of the c'-tan φ' lines for the
failed slope (c0', tan φ0') and for the same slope after installing some remedial works (c'nec, tan φ'nec), respec-
tively. This procedure is illustrated to design piles to stabilize landslides taking into account both driving and
resisting force acting on each pile in a row as a function of the non-dimensional pile interval ratio B/D. The
accurate estimation of the lateral force on pile is an important parameter for the stability analysis because its
effects on both the pile-and slope stability are conflicting. That is, safe assumptions for the stability of slope
are unsafe assumptions for the pile stability, and vice-versa. Consequently in order to obtain an economic and
safe design it is necessary to avoid excessive safety factors.

1 INTRODUCTION 1996). The cost of non-structural remedial measures


is considerably lower when compared with the cost
Correction of an existing landslide or the prevention
of structural solutions.
of a pending landslide is a function of a reduction in
Terzaghi (1950) stated that, “if a slope has started
the driving forces or an increase in the available re- to move, the means for stopping movement must be
sisting forces. Any remedial measure used must in- adapted to the processes which started the slide”.
volve one or both of the above parameters. The For example, if erosion is a causal process of the
IUGS Working Group on Landslides (Popescu, slide, an efficient remediation technique would in-
2001) has prepared a short checklist of landslide re- volve armoring the slope against erosion, or remov-
medial measures arranged in four practical groups, ing the source of erosion. An erosive spring can be
namely: modification of slope geometry, drainage, made non-erosive by either blanketing with filter
retaining structures and internal slope reinforcement, materials or drying up the spring with horizontal
as shown in Table 1. As many of the geological fea- drains, etc.
tures, such as sheared discontinuities, are not well Morgenstern (1992) followed this theme when he
known in advance, it is better to put remedial meas- noted that post-failure analyses can be used to pro-
ures in hand on a “design as you go basis”. That is vide a consistent explanation for landslide causal
the design has to be flexible enough for changes dur- events. The back-analyses can then be used as a ba-
ing or subsequent construction of remedial works. sis for design of the stabilizing measures if engineer-
Although slope instability processes are generally ing works are required. This approach has the added
seen to be “engineering problems” requiring “engi- appeal that the remedial design is normalized in
neering solutions” involving correction by the use of terms of the post-failure analytical model.
structural techniques, non-structural solutions in- Most landslides must usually be dealt with sooner
cluding classical methods such as drainage and mod- or later. How they are handled depends on the proc-
ification of slope geometry, as well as some novel esses that prepared and precipitated the movement,
methods such as lime/cement stabilization, grouting the landslide type, the kinds of materials involved,
or soil nailing, are increasingly being used (Popescu, the size and location of the landslide, the place or
components affected by or the situation created as a
result of the landslide, available resources, etc. The
technical solution must be in harmony with the natu- Table 1. A brief list of landslide remedial measures
ral system, otherwise the remedial work will be ei-
ther short lived or excessively expensive. In fact, 1. MODIFICATION OF SLOPE GEOMETRY
landslides are so varied in type and size, and in most 1.1. Removing material from the area driving the land-
instances, so dependent upon special local circum- slide (with possible substitution by
stances, that for a given landslide problem there is lightweight fill)
more than one method of prevention or correction 1.2. Adding material to the area maintaining stability
that can be successfully applied. The success of each (counterweight berm or fill)
measure depends, to a large extent, on the degree to 1.3. Reducing general slope angle
which the specific soil and groundwater conditions 2. DRAINAGE
are prudently recognized in an investigation and in- 2.1. Surface drains to divert water from flowing
corporated in design. onto the slide area (collecting ditches and pipes)
2.2. Shallow or deep trench drains filled with free-
In this paper a methodology involving back anal-
draining geomaterials (coarse granular fills
ysis of the slope failure and the use of piles to reme- and geosynthetics)
diate the landslide are presented. 2.3. Buttress counterforts of coarse-grained
materials (hydrological effect)
2.4. Vertical (small diameter) boreholes with
2 BACK ANALYSIS OF FAILED SLOPES TO pumping or self draining
DESIGN REMDIAL MEASURES 2.5. Vertical (large diameter) wells with
gravity draining
2.1 Failure envelope parameters 2.6. Subhorizontal or subvertical boreholes
2.7. Drainage tunnels, galleries or adits
A slope failure can reasonably be considered as a
2.8. Vacuum dewatering
full scale shear test capable to give a measure of the 2.9. Drainage by siphoning
strength mobilized at failure along the slip surface. 2.10. Electroosmotic dewatering
The back calculated shear strength parameters, 2.11. Vegetation planting (hydrological effect)
which are intended to be closely matched with the 3. RETAINING STRUCTURES
observed real-life performance of the slope, can then 3.1. Gravity retaining walls
be used in further limit equilibrium analyses to de- 3.2. Crib-block walls
sign remedial works. The limit equilibrium methods 3.3. Gabion walls
forming the framework of slope stability/instability 3.4. Passive piles, piers and caissons
analysis generally accept the Mohr-Coulomb failure 3.5. Cast-in situ reinforced concrete walls
criterion: 3.6. Reinforced earth retaining structures with strip/ sheet
- polymer/metallic reinforcement elements
τf = c'+ σ' tan φ' (1) 3.7. Buttress counterforts of coarse-grained material (me-
chanical effect)
3.8. Retention nets for rock slope faces
where τf and σ' are the shear stress and effective 3.9. Rockfall attenuation or stopping systems (rocktrap
normal stress respectively on the failure surface and ditches, benches, fences and walls)
c' and φ' are parameters assumed approximately 3.10. Protective rock/concrete blocks against erosion
constant for a particular soil. 4. INTERNAL SLOPE REINFORCEMENT
A significant limitation in the use of this criterion 4.1. Rock bolts
is that the constant of proportionality is not really a 4.2. Micropiles
constant when wide range of stress is under consid- 4.3. Soil nailing
eration. There is now considerable experimental evi- 4.4. Anchors (prestressed or not)
dence to show that the Mohr failure envelope exhib- 4.5. Grouting
its significant curvature for many different types of 4.6. Stone or lime/cement columns
soil and compacted rockfill. Therefore, if the as- 4.7. Heat treatment
4.8. Freezing
sumption of a linear failure envelope is adopted, it is 4.9. Electroosmotic anchors
important to know what range of stress is appropri- 4.10. Vegetation planting (root strength
ate to a particular slope instability problem. To mechanical effect)
avoid this difficulty a curved failure envelope can be
approximated by the following power law equation:

τf = A (σ')b (2) 2.2 Procedures for back analysis of slope failures


Shear strength parameters obtained by back anal-
which was initially suggested by De Mello (1977)
ysis ensure more reliability than those obtained by
for compacted rockfills and subsequently found ap-
laboratory or in-situ testing when used to design re-
propriate for soils (Atkinson and Farrar, 1985).
medial measures. In many cases, back analysis is an lated shear strength parameters based on the same
effective tool, and sometimes the only tool, for in- requirements is illustrated in Fig.1c.
vestigating the strength features of a soil deposit
(Duncan, 1999). However one has to be aware of the
many pitfalls of the back analysis approach that in-
volves a number of basic assumptions regarding soil
homogeneity, slope and slip surface geometry and 1
pore pressure conditions along the failure surface 0
Trial Sliding
(e.g. Leroueil & Tavenas 1981). A position of total 1′
Surfaces, Failure
confidence in all these assumptions is rarely if ever Surface,
achieved. (a
While the topographical profile can generally be
determined with enough accuracy, the slip surface is tan φ′
almost always known in only few points and inter-
polations with a considerable degree of subjectivity Envelope
are necessary. Errors in the position of the slip sur-
face result in errors in back calculated shear strength Region II
parameters. If the slip surface used in back analysis
is deeper than the actual one, c' is overestimated and Contact Points
φ' is underestimated and vice-versa.
The data concerning the pore pressure on the slip
surface are generally few and imprecise. More ex- Region I
actly, the pore pressure at failure is almost always c′
unknown. If the assumed pore pressures are higher 1.0 1 0 1′
than the actual ones, the shear strength is overesti-
mated. As a consequence, a conservative assessment (b)
of the shear strength is obtainable only by underes-
timating the pore pressures. F
Procedures to determine the magnitude of both TSS 1′ TSS 1
shear strength parameters or the relationship be-
tween them by considering the position of the actual FS
slip surface within a slope are discussed by Popescu
and Yamagami (1994). The two unknowns - i.e. the
shear strength parameters c' and φ' - can be simulta- c′
neously determined from the following two re- Range of shear
quirements: strength
tan φ′
(a) F = 1 for the given failure surface. That means (c
the back calculated strength parameters have to
satisfy the c'-tan φ' limit equilibrium relation- Figure 1. Shear strength back analysis methods.
ship;
(b) F = minimum for the given failure surface and The procedures discussed above to back calculate
the linear strength envelope parameters, c' and φ' in
the slope under consideration. That means the
equation (1) can be equally applied to back calculate
factors of safety for slip surfaces slightly inside the nonlinear strength envelope parameters, A and b
and slightly outside the actual slip surface in equation (2) (Popescu et al., 1995).
should be greater than one (Fig.1a). The fundamental problem involved is always one
of data quality and consequently the back analysis
Based on the above mentioned requirements, Sai- approach must be applied with care and the results
to (1980) developed a semi-graphical procedure us- interpreted with caution. Back analysis is of use only
ing trial and error to determine unique values of c' if the soil conditions at failure are unaffected by the
and tan φ' by back analysis (Fig.1b). An envelope of failure. For example back calculated parameters for
the limit equilibrium lines c' - tan φ', corresponding a first-time slide in a stiff overconsolidated clay
to different trial sliding surfaces, is drawn and the could not be used to predict subsequent stability of
unique values c' and tan φ' are found as the coordi- the sliding mass, since the shear strength parameters
nates of the contact point held in common by the en- will have been reduced to their residual values by
velope and the limit equilibrium line corresponding the failure. In such cases an assumption of c' = 0 and
to the actual failure surface. A more systematic pro- the use of a residual friction angle, φr is warranted
cedure to find the very narrow range of back calcu- (Bromhead 1992). If the three-dimensional geomet-
rical effects are important for the failed slope under tan φ′
consideration and a two-dimensional back analysis
is performed, the back calculated shear strength will LIMIT EQUILIBRIUM RELATIONSHIP
be too high and thus unsafe. FOR THE FAILED SLOPE: Fs=1 or En=0
tan φ′0

2.3 Design of remedial measures based on back LIMIT EQUILIBRIUM RELATIONSHIP


analysis results tan φ′nec
FOR THE STABILIZED SLOPE

In order to avoid the questionable problem of the


representativeness of the back calculated unique set
of shear strength parameters a method for designing
remedial works based on the limit equilibrium rela- c′
tionship c' - φ' rather than a unique set of shear c′nec c′0
strength parameters can be used (Popescu, 1991). (a)
The method principle is shown in Fig. 2. It is tan φ′ Unload
Load
considered that a slope failure provides a single
piece of information which results in a linear limit
equilibrium relationship between shear strength pa-
rameters. That piece of information is that the factor
INITIAL GEOMETRY
of safety is equal to unity (F=1) or the horizontal
force at the slope toe is equal to zero (E=0) for the UNLOADING ACTIVE
conditions prevailing at failure. Each of the two LOADING PASSIVE PARTS
conditions (F=1 or E=0) results in the same relation-
ship c'-tan φ' which for any practical purpose might
be considered linear.
The linear relationship c'-tan φ' can be obtained c′
using standard computer software for slope stability
limit equilibrium analysis by manipulations of trial (b)
values of c' and tan φ' and corresponding factor of tan φ′
WT 1

safety value. It is simple to show that in an analysis WT 2

using arbitrary φ' alone (c'=0) to yield a non-unity


factor of safety, Fφ*, the intercept of the c'-tan φ' line BEFORE DRAINAGE (WT 2)
(corresponding to F=1) on the tan φ' axis results as:
tan φ0' = tan φ' / Fφ* (3)
AFTER DRAINAGE (WT 2)
Similarly the intercept of the c'-tan φ' line (corre-
sponding to F=1) on the c' axis can be found assum-
ing φ'=0 and an arbitrary c' value which yield to a
non-unity factor of safety, Fc*: c′
c0'=c' / Fc* (4)
(c)
Using the concept of limit equilibrium linear rela- PILE
tionship c'-tan φ', the effect of any remedial measure ROW
d B d D
(drainage, modification of slope geometry, restrain-
ing structures) can easily be evaluated by consider- tan φ′
ing the intercepts of the c'-tan φ' lines for the failed
slope (c0', tan φ0') and for the same slope after in- NO PILE

stalling some remedial works (c'nec, tan φ'nec), respec- PILE INTERVAL (B/D)1

tively (Figure 2). The safety factor of the stabilized PILE INTERVAL (B/D)2<(B/D)1
slope is:
⎛ c' tan φ0 ' ⎞
F = min⎜⎜ Fc = 0 , Fφ = ⎟ (5)
⎝ c' nec tan φ' nec ⎟⎠ c′
Errors included in back calculation of a given (d)
slope failure will be offset by applying the same re-
sults, in the form of c' - tan φ' relationship, to the de- Figure 2. Limit equilibrium relationship and design
sign of remedial measures. of slope remedial measures.
The above outlined procedure was used to design makes all details and difficulties simple by a sound
piles to stabilize landslides (Popescu, 1991) taking and profound understanding.
into account both driving and resisting force. The
principle of the proposed approach is illustrated in
Figure 3 which gives the driving and resisting force 3 APPLICATION
acting on each pile in a row as a function of the non-
3.1 Site Conditions
dimensional pile interval ratio B/D. The driving
force, FD, is the total horizontal force exerted by the The described methodology is applied to a landslide
sliding mass corresponding to a prescribed increase in Ohio in the United States. The site is located
in the safety factor along the given failure surface. along the Ohio River in south-central Ohio. A re-
The resisting force, FR, is the lateral force corre- placement bridge was proposed at the site and site
sponding to soil yield, adjacent to piles, in the preparations reactivated an ancient slide. The cross
hatched area shown in Figure 3. FD increases with section is shown in Figure 4. The slope consists of
the pile interval while FR decreases with the same shale bedrock overlain by shale weathered to a re-
interval. The intersection point of the two curves sidual clay. Overlying the residual clay is alluvial
which represent the two forces gives the pile interval silts and clays. Construction activities at the site led
ratio satisfying the equality between driving and re- to the reactivation of an ancient slide. The slip plane
sisting force. discerned from surface scarps and inclinometer data
UPHILL is shown in Figure 4. It can be seen that the failure
surface is planar in nature and occurs just above the
shale bedrock in the weathered residual clay.
To accommodate the new bridge a fill was pro-
posed on the existing slope, which was now moving
PILE
ROW and would have exacerbated the instability. Hence
d B d D
the use of piles to stabilize the slope was proposed.
DOWNHILL
62

60
FD – Driving Force Existing Fill
New Fill
FR – Resisting Force 58
Height

F D, F R 56
Alluvial Silts Ohio River
54 & Clays
Shale
52
FR FD
Failure Surface
50
Residual Clay
48

46
B/D 0 10 20 30 40 50 60
Length (ft)
narrowly largely
spaced piles spaced piles Figure 4. Cross section of slope.

3.2 Back analysis


Figure 3. Driving vs. resisting force for stabilizing piles
Back analyses were conducted of the slope failure
The accurate estimation of the lateral force on using limit equilibrium techniques as described pre-
pile is an important parameter for the stability analy- viously (Popescu, 1991). The back analyzed rela-
sis because its effects on both the pile-and slope sta- tionship between friction angle and cohesion for the
bility are conflicting. That is, safe assumptions for residual clay and the failure surface are shown in
the stability of slope are unsafe assumptions for the Figure 5. The resulting strength parameters vary
pile stability, and vice-versa. Consequently in order depending upon the water level in the Ohio River.
to obtain an economic and safe design it is necessary The relationship between the friction angle and the
to avoid excessive safety factors. cohesion are shown in Figure 5 for cases of no pile
The problem is clearly three-dimensional and and pile B/D ratios of 0.75 and 0.5. The back ana-
some simplification must be accepted in order to de- lyzed friction angle of about 13º for the no pile case
velop a two-dimensional analysis method based on compares favorably with residual shear test results.
the principles outlined above. However the only The numerical results for the B/D ratios in Figure
simplicity to be accepted and trusted is the simplic- 5 were obtained using the methodology proposed by
ity that lies beyond the problem complexity and Liang (2002) and coded into an Excel spreadsheet.
lected to provide a margin of safety for the drilled
15 shafts.

12
Friction Angle, deg

4 CONCLUSIONS

9 This paper has outlined an approach to back analyz-


No ing the strength parameters in a slope failure and de-
termining the force required to stabilize a slope us-
6 ing piles considering the back analysis results. The
B/D = 0.75 use of the technique has been demonstrated through
application to a case history.
3

B/D = 0.50
0
5 REFERENCES
0.5
0 1
Cohesion, ksf Atkinson, J.H., Farrar, D.M. 1985. Stress path tests to measure
soils strength parameters for shallow landslips. Proc.11th Int.
Figure 5. Back analyzed relationship between fric- Conf. Soil Mech. Foundation Eng., San Francisco, 2: 983-
tion angle and cohesion. 986.
Bromhead, E.N. 1992. Slope Stability. 2nd Edition, Blackie
Academic & Professional, London, 411 pp.
3.3 Driving and resisting forces De Mello, V.F.B. (1977). Reflections on design decisions of
The driving forces were determined using limit practical significance to embankment dams. Géotechnique,
equilibrium analyses utilizing the program XSTABL 27(3):281-354.
(Interactive Software Designs, Inc. 1994) and Duncan, J.M. 1999. The use of back analysis to reduce slope
spreadsheet analyses. The driving forces are shown failure risk. J. Boston Soc. Civil Eng. 14:1:75-91.
in Figure 6 for various B/D ratios. The resisting Interactive Software Designs, Inc. 1994. XSTABL An Inte-
forces were determined using the Ito and Matsui grated Slope Stability Analysis Program for Personal Com-
(1975) method as outlined by Popescu (1995). The puters. Reference Manual, Version 5, Moscow, ID.
resisting forces are shown in Figure 6 for various Ito, T. and Matsui, T. 1975. Methods to estimate landslide
B/D ratios. forces acting on stabilizing piles. Soils and Foundations,
15:4:43-59.
6000
Liang, R.Y. 2002. Drilled shaft foundations for noise barrier
walls and slope stabilization. University of Akron, Akron,
5000
Ohio, Report prepared for the Ohio DOT and FHWA.
Leroueil, S. and Tavenas, F. 1981. Pitfalls of back-analyses.
Force, Driving & Resisting, kips

4000
Proc. 10th Int. Conf. Soil Mech. Found. Eng. 1:185-190.
Resisting Force Popescu, M.E. 1991. Landslide control by means of a row of
piles. Keynote paper. Proc. Int. Conf. on Slope Stability En-
3000
Driving Force gineering, Isle of Wight, Thomas Telford, 389-394.
Popescu, M.E. 1995. Keynote Lecture: Back analysis of slope
2000
failures to design stabilizing piles. 2nd Turkish Symposium on
Landslides, 15-26 October, Adapazari, Turkey.
1000
Popescu, M.E. 1996. From Landslide Causes to Landslide Re-
mediation, Special Lecture. Proc. 7th Int. Symp. on Land-
0
0.30 0.40 0.50 0.60 0.70 0.80
slides, Trondheim, 1:75-96.
B/D Ratio Popescu M.E. 2001. A Suggested Method for Reporting Land-
slide Remedial Measures. IAEG Bulletin, 60, 1:69-74.
Figure 6. Driving and resisting forces as a function of Popescu M.E., Yamagami T. 1994. Back analysis of slope fail-
B/D ratio. ures - a possibility or a challenge? Proc.7th Congress Int. As-
soc. Eng. Geology, Lisbon, (6), p.4737-4744.
Popescu M.E., Yamagami T., Stefanescu S. 1995. Nonlinear
From the results in Figure 6 it can be seen that the
strength envelope parameters from slope failures. Proc.11th
resisting force and driving force cross at a B/D ratio ECSMFE, Copenhagen,. (1), p. 211-216.
slightly larger than 0.5 with a required resisting Morgenstern, N.R. 1992. Keynote Paper: The role of analysis
force of about 1800 kips. A shear force of this mag- in the evaluation of slope stability. Proc. 6th International
nitude could be obtained using six-foot diameter Symposium on Landslides, Christchurch, 3:1615-1629.
shafts; however, eight-foot diameter shafts were se-
Saito M. 1980. Reverse calculation method to obtain c and φ
on a slip surface. Proc. Int. Symp. Landslides, New Delhi,
1:281-284.
Terzaghi, K. 1950. Mechanisms of Landslides, Geological So-
ciety of America, Berkley, 83-123.
Chapter 9

BACK ANALYSES IN FORENSIC GEOTECHNICAL ENGINEERING

Dr. Richard N. Hwang


richard.hwang@maaconsultants.com

Abstract: Back analyses are required to provide technical evidences to prove or to disprove the
hypotheses made on the causes of failures and to establish scenarios of the failures. Analyses must be
performed by experienced engineers who are familiar with the analytical tools to be adopted for
analyses. The analytical tools adopted must be suitable for the cases to be investigated and the
constitutive laws to be used in the analyses must be representative of the materials to be simulated. It
is important to realize the fact that there are limitations associated with analyses and the results
obtained must be interpreted by experienced engineers who have sufficient practical experience.

Keywords: forensic, geotechnical, failure, analyses

1 INTRODUCTION

Failures seldom occur for a single reason and for cases in which litigation is involved, the causes
of failures are inevitably difficult to ascertain. It is necessary to make various assumptions regarding
why and how failures happen and perform analyses to prove or disprove these assumptions.
Concession among the parties involved is often required in reaching conclusions. Sometimes, it has to
be left to moderators to make final judgments.
Discussed herein are the general principles of back analyses. A case history is presented to
illustrate the complexity of back analyses and the difficulties associated with the interpretation of the
results obtained.

2 COMMON CAUSES OF FAILURES

Defective design of temporary works will likely lead to failures during construction and such
failures usually occur rather suddenly. The collapse of Nicoll Highway during the construction of the
Circle Line of the Singapore Mass Rapid Transit System in April 2004 is a good example. The
retaining system of a cut-and-cover tunnel collapsed suddenly while the excavation was about to reach
the formation level and a section of the 6-lane expressway fell into the sinkhole. Failure started as the
waling on the northern wall buckled at 9am on 20 April 2004 and by 3pm all the struts in a 100m
section totally failed. The incident led to the death of 3 construction workers and 1 supervisory staff.
The expressway was closed for seven and a half months. A Committee of Inquiry was immediately
appointed by the Ministry of Manpower to investigate the causes of failure. It was concluded that the
failure was mainly caused by the inadequate design of the retaining system (COI, 2005; Moh and
Hwang, 2007; Yong and Lee, 2007). Effective stress parameters were adopted for marine clay which is
essentially an undrained material, resulting in much overestimates of soil strengths and underestimates
of wall deflections and bending moments. However, the failure was triggered by buckling of the
stiffeners of walings and propagated to all other members of the retaining system.
Defective design of permanent works, on the other hand, will lead to problems which may last for
a very long time. The leaning Tower of Pisa is a notable example. The construction of the tower
started in 1173. The incline of the tower was already apparent when the construction was halted in
1178. The tower was only 10.6m tall and consisted of three stories then. The tower, 58m in height,
was completed in 1319 and the bell-chamber was not finally added till 1372. Geographical surveys
show that the tower is founded on loose sediments accumulated in a river course. The tilt was
obviously caused by differential settlements. There have been several attempts made for preventing
the tilting from worsening but resulted in only adverse consequences. The incline reached 5.5 degrees
and the government officially closed the tower in 1990. The tilt was finally halted by extracting soil
beneath the foundation on the north side and the tower was re-opened at the end of 2001. It is
necessary to study the history of the tower and the measures taken in all these attempts to establish the
entire scenario.

1
Failures are not always related to geotechnical problems. If there are no apparent ground
movements, failures are most likely due to defective design of structures. A 15-year old building, Lian
Yak Building (commonly referred to as “Hotel New World”), in Singapore collapsed in 1986 and
claimed 33 lives. An intensive investigation was conducted by the government and failure of the
foundation was one of the possible causes studied. Of 33 columns in the building, the foundations of 5
of them were selected for detailed examination and serious defects were observed in piles and in pile
caps (Hulme, et al., 1993). However, it was found that the basement walls and slabs were virtually
undamaged and there was no evidence of differential settlements. Checking of design calculations
revealed that the dead loads of the structures were either grossly underestimated, as in the case of
brickwalls, or completely missed, as in some case of self-weight of slabs. It was thus concluded that
the failure was due to under-design of the structures and was not related to geotechnical problems.
Many failures could be attributed to conditions, such as adverse soil conditions, abnormal changes
of groundwater level, heavy storms, or unexpected loading, etc., which were not accounted for in
designs. Some of these conditions are unforeseeable but most of them are merely unforeseen because
of the limited data available. Groundwater rushed into an arrival shaft as a portal was made on
diaphragm wall for receiving the shield machine during the construction of the Taipei Metro. As a
result, a section of twin tunnels was seriously damaged and 39 rings in the up-track and 34 rings in the
down-track tunnels had to be replaced (Ju, Duann and Tsai, 1998). An PVC pipe was found at the
invert of the up-track tunnel after the shaft was drained and the shield machine exposed. It could be
onetime used for pumping water from the underlying water-bearing gravelly layer for irrigation or fish
farming. The soils surrounding the tunnel had been solidified by jet grouting previously. A large piece
of timber was found next to the pvc pipe and it is suspected that the treated ground was much disturbed,
resulting in cracks which became water paths, as the shield machine forced its way out. Groundwater
was then able to rush from the water bearing gravelly layer into the shaft from the abandoned pvc pipe.
Drift woods were frequently encountered during constructions in the Taipei Basin, however, there is no
way to ascertain their locations beforehand. It was also a common practice to pumping groundwater
for irrigation and fish farming in the area, however, pumping wells are equally difficult to locate.
Failures due to defective constructions are not uncommon. Of the five disastrous events in metro
constructions in Asia Pacific in the period of 2001 to 2007, one was caused by defective ground
freezing work, one was caused by defective diaphragm wall and two were caused by defective ground
treatment by jet grouting (Moh and Hwang, 2007). In all these 4 cases, water was the major source of
the problem. In fact, water was responsible for a majority of failures in underground works carried out
in soft ground.

3 BACK ANALYSES

A timeline showing all the events chronologically will be very helpful in understanding what has
happened. It will even be better if all major events can be presented in a Gantt chart with their
durations clearly identified.
Analyses should be performed for “as-built” conditions because many of the assumptions made in
design are either non-existing or different from reality. Therefore, reconnaissance of the site is a must
and may have to be conducted for more than once. It is preferably to be conducted jointly with all the
parties involved in the case so differences in opinions can be sorted out at the site. Photos and videos
will be very useful in helping one’s memory and will provide vital supports to the conclusions to be
made.
In many cases, failures involve soil-structural interaction and advices from structural engineers are
desirable. For cases involving failure of structures, analyses should be performed jointly with
structural engineers.

3.1 Data Collection and Verification

A checklist should be prepared to document the data available in hand and the data which are still
missing. The data needed will depend on the modes of failures. Although the cause of failure may
appear to be obvious right at the beginning, it is still necessary to investigate all the potential modes of
failure before they can be eliminated.
To ensure that the results of analyses are reliable, the data available must be carefully verified.
For example, ground conditions may be mis-interpreted, instrument readings may be erroneous,
construction activities may be wrongly logged, etc. Some of the data may be misleading without being
realized, and some of data may be contradictory and have to be sorted out. It is natural for patties

2
involved to hind facts, intentionally or unintentionally, which are not in their favor, therefore, data must
be critically reviewed and judgments must be applied whenever data appear to be dubious.
An appropriate appraisal of local geology sometimes is helpful in understanding ground conditions
and historical events related to the site of interest should be investigated.

3.2 Checking Design and Calculations

If design drawings are available, it is necessary to check whether or not the works were
constructed as designed. Design calculations, if available, should be firstly checked to see if they are
appropriate. However, inconformity with specifications and/or codes of practice will not necessarily
lead to failures and the redundancy demanded by specifications and codes may not be required at the
time of failure.
It should be noted that ground conditions might have been altered once a failure occurs and it is
necessary to figure out what they were prior to the event. This sometimes may not always be possible
and guesswork may sometimes be required.

3.3 Analytical Tools

Depending on the complexity of the problem, analyses can be performed by using:


a) rules of thumb using indices such as stability number, overload factor, etc.
b) empirical relationship
c) closed form solutions
d) simple numerical models
e) sophisticated numerical models
Simple models are available for many failure modes and many of them can be solved by hand
calculations. Complicated problems can be solved by numerical analyses. In any case, the tools
adopted for analyses must be suitable for the type of problem to be solved and must have sufficient
technical backup.
The analytical tools adopted for analyses must have been validated in accordance with stringent
quality assurance program and well documented. For this reason commercial software packages are
preferred to in-house programs because the former are usually well tested and improved based on the
feedback from users.
Numerical methods have their limitations and the results can not be trusted blindly. Depending on
the algorithm and the numerical scheme adopted, different software packages many lead to drastically
different results. Therefore, a few test runs are necessary to calibrate the parameters to be adopted by
comparing the results of analyses with observations or with known solutions.

3.4 Limitations

With the rapid advancement of computer technology, finite element method and finite difference
method have become important tools for design and they are very useful tools for back analyses as well.
They have become so user-friendly that even fresh graduates from colleges can perform the analyses
with little guidance. This, however, leads to the danger of mis-handling of problems and mis-
interpretation of the results of analyses. After all, geotechnical engineering is an art rather than science.
The results of analyses provide important evidences for judgments to be based on and experience
should prevail at the end.
It is important to realize the limitations associated with analyses so the results can be correctly
interpreted. First of all, although tools are available, three-dimensional analyses are extremely labor
intensive and time consuming. Therefore, unless the cases are critical enough and resources are
available, one-dimensional and two-dimensional analyses are performed and the results are usually
appropriate for practical purposes. However, simulation of a 3D system by a 2D model inevitably
alters the nature of the problem and introduces errors in the results of analyses.
Secondly, soils are usually classified as either drained materials, i.e., pure sand, or undrained
materials, i.e., pure clay, in analyses. However, in reality, most of soils are neither pure sand nor pure
clay and are mixtures of sand and clay. Their behavior will depend on how fast porewater pressures
dissipate. In other words, problems are time-dependent. The rate of dissipation of porewater pressure
will not only depend on the rate of loading but will also depend on the permeability of soil and the

3
length of drainage path. The reliability of the results of analyses will depend on how well this time-
dependency of soil behavior is handled.
Thirdly, it is a well-known fact that soil behavior is highly non-linear and there are many soil
models to handle the nonlinearity. The results of analyses may vary considerably if different
constitutive laws, which describe the stress-strain relationship of soils, are selected.

4 CASE STUDY

The collapse of Nicoll Highway, refer to Fig. 1, is an ideal case for illustrating the points
mentioned above. The site is located in a piece of land reclaimed in the 80’s. As shown in Fig. 2, the
subsoils at this site contain mainly two thick layers of marine deposits (namely, the upper marine clay
and the lower marine clay) and are underlain by the Old Alluvium which is a competent base stratum.
Figure 3 is a plot of the results of a cone penetration test carried out in the vicinity of the site. The
excavation was supposed to be carried out to a depth of 33.5m and diaphragm walls with a thickness of
800mm (locally, 1000mm) were used.
N

Golden Mile
Golden Mile Complex
Tower

Area
affected

0m 200m
Fig Reference: COI, 2005
. 3

Fig. 1 Collapse of Nicoll Highway

M2 M3
Waling buckled at
9am, 20 April, 2004

I-65
S331

S333
S330

S332

S334

S335
S328

S329

S336

S337

S338
S327

S339

S340
S326
S325

66kV Cable Crossing


I-104

資料來源: COI, 2005


Fig. 3
Fig. 2 Retaining system in Section M3

4
I-104 I-65
South RL 102.90
North
105 105
3/26 Fill 1 -1.00 3/26
100 282mm 2 -4.00 202mm 100
Estuarine -7.50
95 3 95
Upper 4 -11.00
Marine

Elevation, m
90 5 -14.50 90
Clay
F2 upper
6 -17.50
85 7 -20.50 85
8 -23.80
80 Lower 80
9 -26.80
Marine
75 Clay 10 -29.80 75
-33.50
70 Estuarine 70
JGP
65 F2 lower 65
4/20
60 175mm 60
4/20 Old Alluvium
55 441mm 55
50 Diaphragm Walls 50
800/1000mm
Reference: COI, 2005

Fig. 3 Ground conditions and excavation scheme in Section M3

4.1 The Incident

Excavation for constructing the cut-and-cover tunnels was carried out by using the bottom-up
method of construction. The collapse occurred in Section M3 on 20 April 2004 while the 10th dig was
completed and excavation reached a depth of 30.5m four days earlier. There were two inclinometers,
i.e., I-65 and I-104, available for monitoring wall deflections. Figure 4 shows the wall deflection paths,
which are the plots of maximum wall deflections versus depth of excavation in a log-log scale, for
these two inclinometers (Hwang, Moh and Wong, 2007). Wall deflections on the two sides of the
excavation were about the same till 9 March, 2004, when excavation reached a depth of 25m, and
deflections of 198mm and 215mm were recorded by Inclinometers I-65 and I-104, respectively.
Subsequently, there was a period in which I-104 was not read because it was damaged. When
monitoring resumed on 26 March, the deflection of the southern wall was found to have increased by
67mm to 282mm while the readings for Inclinometer I65 on the north were fairly steady in this period.
The readings for Inclinometer I-104 kept on increasing while those for I-65 remained to be steady
subsequently, presumably, because of the asymmetry of ground conditions. In fact, I-65 appeared to
move outward by 27mm and the maximum deflection reduced from 202mm on 26 March to 175mm on
20 April, as depicted in Figs. 2 and 4. On the other hand, Inclinometer I-104 moved inward by 90mm
to 441mm in the 3-day period from 17 April to 20 April.

Tip Resistance, qp Shaft Friction, fs


MPa MPa
0 5 10 0.00 0.05 0.10
0

10
Depth, m

20

30

40

50

Fig. 4 Results of a cone penetration test

Failure started as the waling at the northern end of Strut S338 at the 9th level, refer to Fig. 2,
buckled at 9am on 20 April and failure propagated to other struts. By 3pm of the day, all the struts for
a 100m section totally failed. As Nicoll Highway sank, gas, water and electricity cables ruptured,

5
causing power to go out for about 15,000 people and 700 businesses in the Marina and Suntec City area.
Tremors were felt at Golden Mile Complex. Tenants and residents in the building were also evacuated.

Maximum Wall Deflection, mm


1 10 100 1000 10000
1

I-65
Depth of Excavation, m

I-104

10

100

Fig. 5 Deflection paths for diaphragm walls

4.2 The Design

The contract was awarded as a design-and-build contract and temporary works were designed by
the contractor. The computer program PLAXIS, which was first developed in 1987 at the Technical
University of Delft and made commercially available by PLAXIS bv of Delft, Netherlands, was
adopted in the design of retaining system. PLAXIS allows users to select different material types
(drained, undrained and non-porous) and different material modes (Mohr-Coulomb, soft soil, etc.)
Mohr-Coulomb Model was adopted for defining failure of soils for the case of interest. When using
the Mohr-Coulomb model, it is possible to input either effective stress parameters (c’ and φ’) or the
undrained strength parameters (c’ = cu and φ’ = 0). The use of a Mohr-Coulomb soil model with
effective strength parameter has been referred to as Method A in the report by the Committee of
Inquiry (COI, 2005). Method B refers to the use of Mohr Coulomb soil model with undrained strength
parameters in combination with undrained material type. Method A was adopted in the analyses and
effective strength parameters, c’ and φ’, instead of the undrained strength parameters, c’ = cu and φ’ = 0,
were used for marine clays.
This, as depicted in Fig. 6, drastically over-estimated the undrained strength of clays and analyses
indicated that wall deflections and the associated bending moments were grossly under-estimated by a
factor of, roughly, 2. However, except Levels 6 and 9, the strut loads computed by using Model A
were larger than those computed by using Model B. The strut loads computed by using Model B were
107% of those computed by using Model A at Level 6 and 110% at Level 9.

Kf

2cu

effective Total stress path


φ' stress path

Confining stress p'. p

(a) Model A

6
q

Kf

2cu
Total stress path
effective
φ'
stress path
Confining stress p'. p

(b) Model B

Fig. 6 Comparison of undrained strength of clay in Models A and B

4.3 Back Analyses

Back analyses were carried out by reducing the stiffness of soils and JGP slabs to match the
performance of the wall when the reading of I-65 reached 159mm, exceeding the alert level of 105mm,
as the excavation reached a depth of 18m in the 6th stage of excavation on February 23, 2004. The
results of analyses predicted a maximum wall deflection of 253mm in the subsequent stages of
excavation. On April 1, the reading of I-104 reached 302.9mm, back analyses were again carried out
by further reducing the stiffness of materials and the maximum deflection was revised to 359mm.
Model A was continuously adopted in these back analyses.

4.3 Forensic Studies

Subsequent to occurrence of failure, the Singapore Government immediately formed an


independent Committee of Inquiry (COI), headed by a Senior District Judge, to look into the incident.
After thorough investigations, in which 173 witnesses were interviewed and 20 experts offered their
professional opinions, an Interim Report was released on 13 September 2004 and a very comprehensive
Final Report was made available to the public on 13 May 2005 (COI 2005).
The Committee identified critical design and construction errors, particularly the design of
stiffeners on the walings at the connections between the diaphragm walls and the struts, that led to the
failure of the earth retaining system. The Committee also found deficiencies in the project
management that perpetuated and aggravated the design errors, including inadequate instrumentation
and monitoring of works, improper management of instrumentation data, and lack of competency of
persons carrying out specialized work.
At the request of the Land Transport Authority, the Engineering Advisory Panel (EAP) conducted
an extensive study to review the case and the design of other parts of the Circle Line. Three-
dimensional analyses were carried out using the commercial software ABAQUS to analyze the
performance of the retaining system. Each of the first nine layers of excavation was modeled as a
uniform lift with all the elements removed at the same time. The 10th excavation step was simulated in
three sub-steps, starting with the removal of the elements at the eastern end of the excavation and
proceeding westwards (Yong and Lee, 2007) .

4.4 Causes of Failure

As mentioned in Section 4.2, wrong strength parameters were adopted for marine clay, leading to
under-design of wall deflections and bending moments. Although back analyses were performed
during the course of construction, they were performed by engineers who did not have experience on
the program PLAXIS and the results were not reviewed by experienced engineers. In this regard,
however, regarding soil parameters, the experts did differ in opinions in the Inquiry. It is amusing to
note the statement made by the representative of the Contractor in the Inquiry that:
(Quoted) “….. despite more than six months of intensive work by the six teams of experts who
have been reviewing the collapse at M3, the experts still can not reach any agreement on the
correct input parameters to be adopted in a back analysis. Very significant difference still remain,

7
particularly respect to the parameters to be adopted for JGP. It is also noted that because of the
stiffness characteristics of the ground, diaphragm wall and JGP are all highly non linear, it is
virtually impossible to obtain agreement between the monitored strut loads and wall displacements
throughout all the stages of excavation sequence using a single set of linear elastic stiffness value,
as adopted in Plaxis analyses.” (Unquoted)
Although these statements may be exaggerating, they nevertheless illustrate the difficulty associated
with analyses.

Plate stiffener

Strut diaphragm
wall
Load

(a) Original design

C-channel

Strut diaphragm
Load wall

waling

(b) Revised design

Fig. 7 Stiffeners of walings

Model A was adopted by the contractor in analyses for all the cut-and-cover sections in this
contract. Although considerable efforts have been made to deal with excessive wall deflections, the
excavation in other sections was able to be completed. This indicates that the under-design of
diaphragm walls is not solely responsible for the incident. The failure must be due to a combination of
adverse factors.
As depicted in Fig. 3, the soft deposits on the two sides of the excavation differed in thickness
while analyses were performed for only the southern half of the mesh on the assumption that the results
would be conservative. This assumption may not be valid reality because the deflections of the
southern wall tended to be larger due to the imbalance of earthpressures on the two sides of the
excavation. The 3D finite element analyses conducted by EAP also indicated the significance of the
curvature of the alignment. The wall panels tended to split at the joints as the wall moved inward as
the excavation proceeded. This is particularly true for the south wall where wall deflections tend to
widen the gaps.
The failure in fact was triggered by the buckling of walings. As mentioned in Section 4.2, except
Levels 6 and 9, the strut loads computed by using Model A were larger than those computed by using
Model B. Even for Levels 6 and 9, the strut loads computed by using Model B were 7% to 10% larger.
Therefore the buckling was not a result of the use of wrong material model for clay.
The Inquiry revealed that the loads at some of the connections between struts and walings were
under-estimated either because a wrong type of joint was installed or because slays were omitted.
Secondly, around February 2004, several instances of stiffener plate buckling as well as the buckling of
a waling were reported at the Nicoll Highway Station, the contractor proposed replacing the stiffener
plates by C-channels in an attempt to improve the performance of the connection. The replacement of

8
double stiffener plates with C- channels, refer to Fig. 7, provided only minor improvement in terms of
axial load bearing capacity for the waling connections, but this came at the expense of ductility. The
change rendered it more susceptible to the brittle “sway” failure mode. This is clearly seen in results of
the post collapse finite element analyses and physical tests. The poor detailing is exacerbated by the
discontinuity of walings, as pointed out in the EAP study, which is particularly important at the curved
south wall where deflection due to excavation caused separation of the joint between diaphragm wall
panels.
The instrumentation and monitoring system was not properly executed and some of the crucial
readings were found erroneous and misleading. This made it impossible to correlate the strut loads
with the performance of the retaining system.
In addition to all the above-mentioned shortcomings in the design, the performance of the two
grouted slabs was also questioned and investigated in the Inquiry and in the EAP study. The removal
of sacrificial JGP layer above the formation level was not followed with timely installation and
preloading of struts to compensate for the loss of reaction load sustained in the JGP layer. As a result,
it caused significant lateral movement to the temporary walls and increase in load to the struts above.
This triggered the local failure of connections at the joint between struts and walings at critical
locations and contributed to the chain of collapse of diaphragm walls.

5 CONCLUSIONS

Failures frequently occur and back analyses enable their causes to be identified. The experience
learned can then be passed on from generation to generation. For the results of analyses to be reliable,
(1) the analytical tool used must be suitable for the type of problem to be analyzed
(2) the analyses must be conducted by experienced engineers who have sufficient background on the
algorithm of analyses
(3) the input parameters must be representative of the materials to be studied
(4) the construction sequence must be reliable
(5) instrument readings must be faithful
(6) the results must be interpreted by experienced engineers who have sufficient practical experience
on the performance of soils and structures.
It is however important to realize that there are limitations associated with analyses and the results
obtained only play a role of providing supporting evidences and engineering judgments should always
prevail.

REFERENCES

COI 2005. Final Report of the Committee of Inquiry into the Incident of the MRT Circle Line Worksite that Led
to the Collapse of Nicoll Highway on 20 April 2004, Presented by Committee of Inquiry to Minister for
Manpower on 10 May 2005, Singapore
Hulme, T. W., Parmar, H. S., Hou, K. H. and Sripathy, P. (1993) The collapse of the Hotel New World, Singapore,
a technical inquiry, The Structural Engineer, v71, no. 6/16, March, pp. 91~98
Hwang, R. N. and Moh, Z. C. Wong, K. S. 2007. Reference envelopes for deflections of diaphragm walls in
Singapore Marine Clay, Proc., 16th Southeast Asian Geotechnical Conference, Kuala Lumpur, Malaysia, 8 ~
11 May
Ju, D. H., Duann, S. W. and Tsai, H. H. (1998) Ground freezing for restoration of damaged tunnel, Proc., 13th
Southeast Asian Geotechnical Conf., November 16~20, Taipei, Taiwan, pp. 615~620
Moh, Z. C. and Hwang, R. N. (2007) Lessons learned from recent MRT construction failures in Asia Pacific,
Opening Keynote Address, Proc., 16th Southeast Asian Geotechnical Conf., 8~11 May, pp.3~20, also,
Geotechnical Engineering, Special Issue, v38, no. 3, December, Bangkok, Thailand, pp. 121~137
Yong, K. Y. and Lee, S. L. (2007) Collapse of Nicoll Highway – A global failure at the curved section of a cut-
and-cover tunnel construction, Chin Fun Kee Lecture, Proc., 16th Southeast Asian Geotechnical Conf., 8~11
May, pp.23~36

9
Chapter 10

GEOTECHNICAL FAILURE: THE CAUSE IS NOT ALWAYS


OBVIOUS AND MAY BE COMPLEX
Jan E Hellings
MBA, PhD, DIC, MSc, CEng, FICE,
Dr Jan Hellings & Associates, Headley, UK, RG19 8AB
jan@janhellings.com
ABSTRACT: This article considers the failure of four projects, three in the UK and one in the Netherlands. In
many cases, the failure is not straightforward and may be due to a number of causes rather than to a single
cause.

The tunnel was constructed using the “NATM”


1. INTRODUCTION (New Austrian Tunnelling Method) form of
construction, or spayed concrete lining (SCL). The
Designers and constructors clearly do not anticipate safety of this approach depends upon close
that failure of their project will occur; the principle monitoring as the work progresses; it was the first
risks and concerns of any job are usually identified time it had been used in London clay. In view of
and effort is channelled into controlling, this, an extensive full scale trial was undertaken at
monitoring and providing for behaviour scenarios the site prior to initiating the actual works. The
arising out of these. But, perhaps because of too method involves the use of SCL, applied in stages
focused a view, less critically perceived features in such a manner that the ‘stand up’ time of
can be overlooked. In concentrating on elements – exposed clay is not compromised.
efficiently and often very cleverly dealing with
such elements – the overall ‘scheme’ view can be The collapse is well documented in the UK Health
forgotten. The situation is often exacerbated by and Safety Executive publication (HSE 2000).
interfaces, whether it be by environment, by
discipline, by techniques, by participants, by The HSE labelled this failure as having all the
cultures or by trading practices. (Fernie 2007). hallmarks of an “organisational accident”. The
This paper will describe four projects where failure report states that “hazards were not identified by all
has occurred, in some cases spectacularly and at the parties and risks were not controlled, during
great cost. The projects are: collapse of rail tunnels construction, through the “defensive” systems (i.e.
at London’s Heathrow Airport in 1994, the collapse preventative management systems) used by the
of a section waste water tunnel in Kingston Upon parties”.
Hull, UK in 1999, a failure during construction of
the Tramtunnel in The Hague, The Netherlands in The investigations’ findings were as follows:
1996, and finally the failure of foundations to a
house in central London. 2.1 Contractual Arrangements and Culture
• A lack of awareness of risks: the risk
2. HEATHROW AIRPORT TUNNELS evaluation at the outset was not sufficiently
COLLAPSE comprehensive.
• The quality of workmanship was according to
One of the worst civil engineering disasters in the the “self certification” procedure which
United Kingdom in the last quarter of a century followed the competitive tendering
occurred during the night of the 20-21st October arrangement. This, in the author’s view, is not
1994 when tunnels in the course of construction as robust as an independent certification
beneath Heathrow Airport’s central terminal area procedure.
collapsed. They continued to collapse over the • The separation of permanent and temporary
following days. The public and those engaged in NATM works design: this is often a problem
the construction work were exposed to grave risk in construction generally.
of injury. Although there were no injuries, many • Separation of compensation grouting and
people were put at risk and the consequential costs tunnelling monitoring processes.
was significant. A number of the lessons arising
from this collapse can be applied to engineering 2.2 Design
projects generally.
• There was a lack of appreciation of the a few metres of the shaft which at that time was
differences between hard rock and soft clay about 200m behind the face.
behaviour.
• The design was not considered sufficiently The tunnel ring comprised 6 concrete segments.
robust. The rings were assembled with temporary bolts
• The invert of the tunnel was of a form which which were removed once there was access beyond
made construction tolerances more critical. the rear of the TBM train. The removal of these
• The monitoring regime was unsatisfactory. bolts is common in mainland Europe; the
• The ground conditions were as expected. alternative would have been expensive stainless
steel bolts and the bolts contribution to the rings
2.3 Construction Quality strength and stiffness was ignored in the permanent
• The tunnel wall profiles were not correct. design.
• There was defective invert construction.
The geology at the collapse location was different
• There was defective joint construction.
from that elsewhere along the route; in this location
• The invert was “over flat”.
there was a deep infilled former channel of the
River Hull the tunnel arisings here being in
2.4 Construction Management alluvium, whereas elsewhere the tunnel is in glacial
• There was insufficient specialised staffing. deposits. The first sign of the impending collapse
• There was poor communication between was a small leak not considered significant at the
parties involved. time between segments at the knee joints. The
• There was poor sequence of tunnel progress of the collapse, at least in its early stages,
construction. was seen by a number of witnesses. Sand, silt and
• There was bad timing of invert repairs. clear water was observed bubbling through the
• There was no integration in planning joint. The invert segments began to twist and at
construction activities. this time leaks from the shoulder joints worsened.
• There was a lack of awareness of This was approximately 6 hours after the leak had
instrumentation data warning of impending first been noticed.
failure.
As the situation worsened significant volumes of
It can therefore be seen from the Heathrow sand had flowed into the tunnel, enough to derail a
experience that failures are likely to arise from train which remained in the collapsed zone until
multiple causes. Designs need to take account of recovery some months later. Segments were
the ease of construction. Monitoring should be spalling, and the extent of damaged segments were
continuous and autoprocessed and preferably spreading away from the shaft. Just before the
available on the web. Supervision must be tunnel was abandoned, spalled concrete was
informed of the design objectives and there needs observed to be dropping from the crown segments.
to be integration and communication of the team. Movement continued at the surface until a crater
60m in diameter and 2.5m deep had been formed,
The cost of recovery of the project is estimated to the centre of which was around 3m east of the
have been of the order of 250 million U.S dollars; shaft.
there was no loss of life but a successful HSE
prosecution. Fortunately the area around the shaft was being
used as a surface car park.

3. HULL WASTE WATER FLOW A relatively unusual feature of the collapse was
TRANSFER TUNNEL COLLAPSE that there was no immediately obvious cause or
contributing factors.
In November 1999, tunnel construction was
progressing normally with 80% of the roof Despite a very thorough investigation, the team did
successfully completed (Grose and Benton 2005). not find an incontrovertible single root cause for
About two weeks earlier the tunnel boring machine the collapse. Although in some ways the situation
(TBM) heading westwards had passed through the was not the desired outcome, pragmatic balance
previously constructed 7.5m diameter access shaft had to be struck between the cost of investigations
having stopped in the shaft for essential desire to provide answers.
maintenance. The centre of the collapse was within
It was established that the two primary factors sandboil not to occur therefore, the percolating
leading to the collapse were: water should be discharged in a controlled manner
(i) The presence of substantial volumes of fine using a drainage system in the soil above the jet
sand, under considerable water pressure, adjacent grout later, without the risk of erosion. A covering
to the tunnel; and layer of sand of certain thickness is therefore
(ii) A leak through the lining large enough to necessary and was absent to a sufficient extent in
allow sand to wash through. the case of the Tramtunnel.

All the credible causation factors were investigated Compressed air was introduced in the tunnel while
to the extent that their likelihood of occurrence imperfections in the jet grout layer were repaired
could be judged, and in this way steps could be by additional grouting.
taken to ensure safe completion and commissioning
of the tunnel.
5. LONDON HOUSE FOUNDATION
FAILURE
4. THE HAGUE TRAMTUNNEL
The house is a two storey building constructed over
The Tramtunnel was constructed using the ‘cut and a hundred years ago. The house was originally
cover’ system; the walls of the tunnel were constructed with a shallow strip footing on London
constructed variously of diaphragm and sheetpile clay. The present owner purchased the property in
walls. After construction of the roof slab, 1994 and in 1995 noticed cracking appearing in the
excavation took place beneath. An arched jet grout walls of the property. There are six very large lime
layer was installed between the toes of the trees in close proximity to the house.
diaphragm and sheetpile walls which was intended
to fulfil three functions: In 1997 an investigation was undertaken and it was
• Lateral support for the walls. decided that the property needed underpinning as
• Water resistant layer to minimise water the roots of the lime trees were drying out the
seepage into the building pit. underlying London clay and leading to subsidence.
• Resist upward water pressure Continuous mass concrete footings were designed
to be constructed to 3m below ground level. A
The jet grout-arch consists of interlocking jet grout contractor was appointed and the work completed
columns with varying heights of 1.5m at the crown in 1998. However in 2003 further cracking started
to 2.5m at the side of the arch. The design relied to appear in the walls of the property, despite the
on a completely watertight arch. underpinning. In recent years the cracking has
become considerably more serious and at one stage
In 1998, during the last stage of excavation, a there were concerns for the structural stability of
serious sandboil occurred. As a result 60m 3 of soil the walls of the house.
flowed into the tunnel, leaving a hole in the street
next to the tunnel. As this was the first part of the It was thought that the roots of the trees had
tunnel to be constructed, all parties involved lost surprisingly extended below the underside of the
confidence in this construction method. 3m deep footings and were therefore drying out the
London clay at some considerable depth. However
Research into the collapse (Van Toll and Sellmeijer a further investigation revealed that the footings
2003) found that there had in the years before the had not in fact been constructed to 3m as they
construction of the Tramtunnel been several near should have but in some cases were around 1.5m or
catastrophic events involving these jet grouts less. Clearly the contractor had not undertaken the
screens in Cairo, Berlin and Bilbao. work in the manner that he should have.

The research focused on two aspects: the reliability


of jet grout screens and the consequences of CONCLUSIONS
imperfections in jet grout screens.
The four case studies have all involved failures
It was realised that all injection layers will show related to the geotechnical situation of a variety of
some imperfections. This would allow ground structures. In general it can be seen that the
water from beneath the gel layer to percolate failures were rather surprising in the manner that
through the gel layer along the ‘imperfection’; for a they occurred; often several factors combined in
order to make the failure occur. It is not always
obvious what has led to failure. Experience tells us
that often it is at interfaces and in areas where risk
is perceived to be low that problems occur. In
forensic investigation of failure in civil engineering
it is almost a reverse of “Occam’s razor” that needs
to be applied (Fernie 2007); Occam’s razor is a
principle that states that the explanation of any
phenomenon should make as few assumptions as
possible eliminating those that make no difference
in the observed predictions of a hypothesis or
theory. Clearly all possible risks should be
investigated thoroughly before and during any
construction.

REFERENCES

Fernie, R (2007) – Personal communication with


the author

Grose, W.J. and Benton, L (2005) – Proceedings of


the Institution of Civil Engineers, Geotechnical
Engineering 158, October, issue, GE4, 179-185

HSE (2000) – The Collapse of NATM Tunnels at


Heathrow Airport

Toll, A.S. and Sellmeijer, J.B. (2003) – Souterrain


The Hague: Imperfections in Jet Grout Layers,
Reclaiming the Underground Space, Saveur (ed.),
381-386
Chapter 11

Instrumentation and Monitoring for Forensic Geotechnical Engineering

Y.Iwasaki
Geo-Research Institute, Osaka, Japan

ABSTRACT

Instrumentation plays an important role in forensic geotechnical engineering. In soil mechanics, there are at least several reasons or
mechanisms to explain a problematic phenomenon. The scientific facts that were recorded by instrumentation during the construction
and/or post construction give a direction to select one of several possible hypotheses.
Two case histories are presented to show the effects of instrumentation upon the forensic investigation of the failure of geotechnical
constructions. One is Teton Dam, Idaho in the U.S.A. that was earthen dam and failed during the first filling stage in 1976. Some
conclusions were given by Independent Panel to review the failure; however, Teton Dam has been providing endless long debate
because of not enough monitored data to rely on.
Another one is Nicoll Highway Collapse that was caused by failure of retaining wall during subway construction in Singapore in
2004. In this second case, monitored records of the deflection of retaining wall gave very important information. From the seventh
excavation step, deflection of wall at south side became much larger than at north side. One sided deformation was due to the inclined
dense gravel layer towards to south below the soft layer and deeper soft soil that differed from the initial design of horizontal layer
condition. After the failure, it was confirmed thicker soft soil layer in the south compared to north side by boring study. If the
retaining wall had failed without knowing different performance between these walls, it might be much works needed to reach the
right conclusion.
Instruments and sensors in geotechnical engineering are briefly reviewed. In recent decades, new technologies were introduced to
sensor and instrumentation in geotechnical engineering. Among these innovative instruments, two techniques are introduced. Carrier
Phase Tracking GPS for measuring displacement between reference GPS receiver and target GPS receiver with a high accuracy of
displacement of 2-5mm. BOTDR (Brillouin Optical Fiber Reflectometer) is another cutting edge technology that provides strain and
temperature. Technical basic knowledge of these instruments is explained to understand the principle that is useful not only to follow
but also to take advantage of the next generation of instrumentation becoming to prevail at present and in near future.

Keywords: forensic, Teton dam, Nicoll Highway Collapse, Carrier Phase Tracking GPS, BOTDR

1 INTRODUCTION 2.1 Teton Dam Failure


In the preface to second edition of the textbook on Soil The Teton Dam was a federally built earthen dam on the Teton
mechanics by Terzaghi and Peck in 1967, Prof. Ralph B. Peck River in southeastern Idaho, USA which when filling for the
added a new chapter on Performance Observation. This chapter first time suffered a catastrophic failure on June 5, 1976. The
12 was intended to aid the engineer in the use of observational collapse of the dam resulted in the deaths of 11 people and
method which is at the very heart of successful application of 13,000 head of cattle. The dam cost about USD $100 million to
soil mechanics. Among several good reasons to adapt build, and the federal government paid over $300 million in
observation, Dr. Peck indicates field observations for providing claims related to the dam failure. Total damage estimates have
evidence in lawsuits. The text book indicates that lawsuits ranged up to $2 billion. The dam was never rebuilt. (Wikipedia)
frequently arise from conflicts between the owner and Not enough instrumentation was installed to monitor pore
contractor because of the completed structures or between the pressure in the dam.
contractor and neighbour because of damage to the latter’s
property.
Firstly, two cases are presented where instrumentation was key
factor in providing fact process of failure. Secondly,
instrumentations and sensors are reviewed including some of
the cutting edge technology that provides innovative
instrumentation in geotechnical engineering at present and in
near future.

2 LESSONS IN THE PAST


In recent cases, big disasters in geotechnical engineering show
the need of the instrumentation and monitoring. Two examples
are shown in the following section. One is Teton Dam that
failed on June 5, 1976 and another is Nicoll Highway Collapse
on April 21, 2004.

Photo-1 Teton Dam, Idaho, June 5, 1976


from <http://en.wikipedia.org/wiki/Teton_Dam>
Hydraulic fracture in the core was considered as most likely the The Panel referred on monitoring as
key mechanism estimated by desk work with laboratory aids. “There were not enough instruments in the dam to provide
Independent Panel to Review Cause of Teton Dam Failure adequate information about changing conditions of the
(1976) reached a conclusion as follows, embankment and abutments.”
The Panel had quickly identified piping as the most probable The debate on the cause of the failure still continues until
cause of the failure, and then focused its efforts on determining recently (Solava and Delatte, 2003) and at present. If pore
how the piping started. Two mechanisms were possible. The pressure monitored data had been provided, we could have
first was the flow of water under highly erodible and reached the valuable understanding.
unprotected fill, through joints in unsealed rock beneath the
grout cap, and development of an erosion tunnel. The second 2.2 Collapse of Nicoll Highway
was “cracking caused by differential strains or hydraulic


fracturing of the core material.” The Panel was unable to The Collapse of Nicoll Highway The Committee of
determine whether one or the other mechanism occurred, or a Inquiry(2005) that took place on 21 April 2004 was caused by
combination. an accident of subway construction in Singapore.
If pore pressure sensors had been installed in the dam section, The Failed section in the subway construction near Nicoll
process of increasing pore pressures during filling water might Highway was a cut and cover excavation shown in Fig.1.
have provided vital information of the failure process.

Fig.1 Geotechnical condition for design of the failed section

(Number with underline is displacement of JGP slab)


Fig.2 Deflection of retaining wall of the failed section
The excavation is planned in a very deep soft marine clay layer
and some special design of JGP (Jet Grouting Pile) was The original design condition was assumed as the same soil
considered. The slab near the bottom of the excavation was to layers at both sides. Boring study after the failure have revealed
take strong horizontal load from the soft clay layer. that the dense sand gravel layer below the soft marine clay is
Among various instrumentations, the most important a
inclined tow rd south and the soft marine layer is deposited
information was the inclination of the temporal retaining wall. deeper than the original design as shown in Fig. 4. If the site
9/1 10/1 11/1 12/1 1/1 2/1 3/1 4/1 5/1 engineer could evaluate the different response of the walls of
GL=102.9m 2nd south and north, he might proceed to perform boring study and
100 1st strut
3rd
could prevent the failure. The site engineer just agreed with the
90
4th 5th proposal of modifying design parameter by the contractor and
H eight(m )

6th allowed to increase the allowable design level without any


Strut and Excavation Stages 7th8th
geotechnical consideration as shown in Fig.3.
The
80
9th
upper JGP
JGP slab is rather stiff elements compared to the soft soil at
70
lower JGP the site. The horizontal compression displacement allowable in
60 the slab might be controlled by failure strain level under
compression of the order less than possibly a few percents.
500
Since the horizontal length of the JGP slab is about 20m, 20 cm
W all Deflection(mm)

400
Maximum Deflection of Retaining Walls of the compression at 8th strut stage (Fig.2) corresponds to 1%
South.I104 of compression strain, which was the design allowalble strain.
300
North.I65 The instrumentation and monitored data really had provided not
200 Design Level only before the failure but also after the failure to give
constrains of the possible mechanism discussed in forensic
100
engineering process.
0
9/1 10/1 11/1 12/1 1/1 2/1 3/1 4/1 5/1
2003 Time 2004
3 INSTRUMENTATION FOR MONITORING
Fig.3 Maximum deflection of walls of the south and north side
Recent technologies of monitoring displacements or strains
Though under emergency, the measurement was not made daily. have much advanced within the past decade. Among these
However, the obtained data gives clear tendency of the several technologies, only two methods of GPS (Global
deformation as shown in fig.2. Two things should be noted. Positioning System) and BOTDR (Brillouin Optical-fiber Time
One is the deflection of retaining wall. The other is the Domain Refletometer) are introduced here.
horizontal compression of JGP slab. It should be noticed not However, conventional methods are still important and used
only the increament of the maximum deflection itself but also as daily practice.
the difference between south and north walls. As shown in Fig.2 Due to the limitation of the number of paper, the author will
and 3, the maximum deflections at sixth and seventh excavation review only a trend of instrumentation for monitoring in
level show the same order of 160mm and 200mm for north and geotechnical engineering in some aspect.
south sides. Reader should refer an introductory book as well as
However, after the seventh excavation, the maximum deflection containing professional level by John Dunnicliff (1993).
of the south side wall increased with the excavation depth. At
3.1 Mechanical Sensor
the same time, the deflection of the north side was kept constant
about 20cm and even decreasing. However, this difference of Photo.2 shows a simple and yet accurate monitoring device
deflection needs some reasons of either weaker soil condition or to measure gap movement for foundation of Hudson-Athens
smaller stiffness of retaining wall in south side. Lighthouse (U.S. National Register of Historic Places) in the
Hudson River in the state of New York in the United States.

Earthpresure Earthpresure
by soft clay by soft clay Photo.2 Mechanical type of gap sensor
southside northside
(http://en.wikipedia.org/wiki/Strain_gauge)

The two halves of the device are rigidly attached to the


Fig.4 Deflection of walls at the final step and the confirmed foundation wall on opposite sides of the crack. The red
geotechnical condition by boring after the failure reference lines are on the transparent half and the grid is on the
opaque white half. Both vertical and horizontal movement can
be monitored over time. In this picture, the crack can be seen to becomes available in the market.
have widened by approximately 0.3mm (and no vertical
movement) since the gauge was installed. 4 RECENT DEVELOPMENT OF SENSORS

3.2 Strain Gage Sensor


In the past decade, sensor technology had entered another era of
Electric sensor is also common for automatic monitoring. innovative principles like MEMS (Micro Electro Mechanical
Photo.3 is strain gage based displacement sensor. The range of Systems).
measurement is 5 to 10 mm with accuracy of 0.5-1%/Full Scale. In the chapter, among these cutting edge technology, GPS and
fiber optic sensors are introduced.

4.1 Carrier-phase tracking of GPS

Carrier-phase tracking signal of GPS has resulted in a


revolution in land surveying. A line of sight along the ground is
no longer necessary for precise positioning. Positions can be
measured up to about 30 km from reference point without
intermediate points. This use of GPS requires specially
Photo-3 Strain gage based displacement sensor (Tokyo Sokki) equipped carrier tracking receivers.
Displacements of multiple points may be monitored by Carrier
Strain gage for steel plate was also developed based upon Phase Tracking GPS method that provides relative
vibrating wire sensor as shown Photo-4. displacements of target monitoring points to the reference point.

Photo.4 VW spot-welded strain Gauge


(Durham Geo Slope indicator Co.)

Vibrating wire transducer measures change of natural


frequency of a wire installed in a very small pipe on a steel plate.
It is designed to set the sensor by spot-welding to steel surface. (Difference in travel distance=n19+δ (cm))
The accuracy of the vibrating wire sensor is about 0.05% of
full scale of 2500µstrain. Fig.5 Basic Concept of the Carrier Phase Tracking
modified from (http://www.shamen-net.com/word/img/)

3.3 Displacement, Load cell and Pore pressure sensor


Conventional strain gauge and vibrating wire are basic
transducer for not only strain but also other application of
displacement, load, and pore pressure sensors.
Other types of transducers like LVDT (Linear Voltage
Differential Transducer) are also used for various types of
sensors. The accuracy of LVDT sensor is around 0.1-.0.5%/F.S.

3.4 Inclinometer
Inclinometer sensor system that had been developed in 1970s
was based upon force balance sensor. The force balance
principle that is to measure how much force needed to get back
the original zero position of the displacement sensor have
resulted in the wider range to measure more than 30 degrees Photo 5 GPS receiver
with the accuracy of 0.01-0.02%/FS. (http://www.furuno.co.jp/product/gps/terrain/dana2000.html)
Recently, Micro-Electro-Mechanical Systems (MEMS) have
realized much cost effective with the same performance As shown in Fig.5, distance between the monitoring station and
GPS is divided into two segments of equi-distance and the space conditions like humidity, temperature, and air pressure.
difference. To avoid such errors, it is better to set monitoring position with
The difference in travel distance is obtained as to detect the time relative position within some limited range of several hundred
difference between the receiving times of the same phase of the meters.
signal carrier signals. The L1 and/or L2 carrier signals are used The most time consuming process is to estimate the wave
in carrier phase surveying. The wave length used for the system number n. Satellite is moving over the receivers from L1 to L2
is about 19cm of the L1 carrier band. along the known orbit. Using four GPS satellite signals, the
The difference in travel distance = n19 + δcm) phase difference as δ is obtained for every satellite. As shown in
<where n is an integer number and called integer bias. Fig.6, point P is the crossing point between ray path from
With accuracy of 1% of wave length in detecting the leading satellite to the monitoring receiver and the equi-distance circle
edge, the error might be as low as 2 millimeters. (Circle 1 in Fig.6). As the satellite moves from L1 to L2, the
crossing point P1 moves to P2. Since the radius of the circle is
very long compared with base line, the arc from P1 and P2 to
the reference receiver (RR) is considered as a straight line. The
locus of the points P1 and P2 forms an elliptical orbit.
The wave number n is obtained as to give the minimum error to
show the orbit as elliptical shape in 3D space.
Vertical displacement(mm)
40
30 forced displacement
20
10
0
-10
5/10 5/15 5/20 5/25 5/30 6/4

Fig.8 Response of GPS in vertical component

Fig.8 shows response of the vertical component against forced


displacement that is shown in red bold line.
It is found that the response of vertical component is not sharp
Fig.6 Determining wave number of integer bias for Carrier- to small displacement but becomes very clear when the input
phase tracking of GPS method displacement is larger than the 5mm of the level of standard
GPS Monitoring at G-3, Mikage
20 20
deviation.
UD
10 10
UDav=0.017mm
0 0 4.2 Optical Fiber Sensors
SD=5.02mm
-10 -10

-20 2008/2/1-3/26 -20


(%) Fiber optics, though used extensively in the modern world, is a
Displacement(mm)

2/1 2/8 2/15 2/22 2/29 3/7 3/14 3/21 3/28 0.0 10.0 20.0 30.0 fairly simple and old technology. Guiding of light by refraction,
10 10 NSav=0.43mm
NS SD=2.06mm
the principle that makes fiber optics possible, was first
5 5
demonstrated in Paris in the early 1840s. Prof. Jun-ichi
0 0
Nishizawa, a Japanese scientist at Tohoku University, was the
-5 -5
(%) first to propose the use of optical fiber for communications in
-10 2008/2/1-3/26 -10
2/1 2/8 2/15 2/22 2/29 3/7 3/14 3/21 3/28 0.0 10.0 20.0 30.0 1963. Nishizawa invented other technologies that contributed to
10 10 EWav=0.31mm
EW the development of optical fiber communications as well.
5 5 SD=1.86mm

0 0

-5 -5
(%)
-10 2008/2/1-3/26 -10
2/1 2/8 2/15 2/22 2/29 3/7 3/14 3/21 3/28 0.0 10.0 20.0 30.0
Number in total 1252

Fig.7 Scattering of monitored position by GPS carrier phase


tracking (Iwasaki, 2008)

Table-1 Errors of displacement obtained by GPS

Component Standard Deviation


(mm)
UD 5.02
NS 2.06 http:/ / en.wikipedia.org/ wiki/ Optical_fiber
EW 1.86
Photo.6 Optical fiber
Errors are caused by several factors. One of the common factors
is the variation of wave velocity near the surface among the
monitoring points. Wave velocity may be affected by such
The RI of 1.0 means light travel in the medium with the speed
in vacuum air. The RI greater than 1.0 means light travel in the
medium with lower speed than in vacuum air. To confine the
optical signal in the core, the refractive index of the core must
be greater than that of the cladding. If the wave length of the
light is longer than compared to diameter of the core cable, the
light is confined within the core and can travel very long
distance with little scattering. The boundary between the core
and cladding may either be abrupt change of refractive index, in
step-index fiber, or gradual, in graded-index fiber. Single mode
optical cable(SM) is designed as abrupt change to show little
loss with capability of travelling long distance of 10km.
Fig.9 Structure of optical fiber

4.3 BOTDR (Brillouin Optical-fiber Time Domain


Refletometer)
When a sharp and strong optical pulse is sent into an optical
fiber cable, some weak optical signal was recognized as
returned to the entered gate. The signals are originated from the
scattering of the light at the propagating front. Three different
kind of scattering are identified as Rayleigh, Raman, and
Brillouin. The scattered signals of Brillouin are found to show
special characteristics. The frequency of the scattered pulse is
shifted about 10GHz from that of the input pulse and changes
Figure 10. Refractive Index proportional to strain and temperature in the fiber as shown in
(A and B: refraction, C: full reflection) Fig.11

Optical fiber has been developed for sending images of 


medical use and for optical signal as communication cable as
shown in Photo. 9.
Fig.9 shows a typical structure of the optical fiber that
composes from core and clad.
An optical fiber consists of fiber of quartz or plastics in a
cylindrical shape that works as waveguide for transmitting light
along its axis, by the process of total internal reflection. The
fiber consists of a core surrounded by a cladding layer.
The refractive index (or index of refraction, see Fig.10) of a
medium is a measure of how much the speed of light (or other
waves such as sound waves) is reduced in the medium
compared to a reference medium. In treating light, the reference
medium is usually vacuum air.
<http://park.itc.u-tokyo.ac.jp/hotalab/research/bocda-r>
Fig.11 Brillouin shift vs. Strains and Temperature

Fig.12 Basic Concept on BOTDR 


As shown in Fig.12, when a signal is sent to an optical fiber, An optical fiber was fixed along a plastic pipe. The fiber
backward travelling scattered light continuously arrives at the cable was given some pretension strain. The pipe strain gauge
front gate. The point where the scattered signal originated is of optical fiber gives continuous change of the strain compared
estimated by the delayed arrival time from the time of sending to conventional pipe strain measurement that gives only discrete
pulse. points. As shown in Fig.13, the strain shows continuous change
Since the relationship between the strain and Brillouin from compression to extension that corresponds to bending
frequency shift is known, the analyzed frequency shift is easily deformation at the depth of 10 to 11m. This depth was
transformed to strains or temperature. When temperature effect considered as a sliding surface.
should be known independently, another optical fiber cable with
free strain may be installed. The accuracy of the measurement Pile 126: BOTDR Strain Measurement
4
depends upon accuracy to obtain the number of the frequency of between 12/08/05 and 8/11/05 ∆T [°C]

Brillouin shift. At present, the measurement is usually given for 2


1m segment of the cable. The error range of strain is about 10-
20µstrain and that of temperature is 0.5-1 degree in Celsius. 0

-2
4.4 BOTDR as Pipe Strain Gauge

E leva tio n (m )
-4 Exc.
Nakano et al. (2003) reported a case history of the Lvl.
application of the BOTDR to identify sliding surface by -6
inserting a pipe equipped with optical fiber in a slope as shown
-8
in Fig.13.
-10
Outerside
-12 of Excavation
Nearside
-14 of Excavation
Axial strain
-16
-0.04 -0.03 -0.02 -0.01 0 0.01 0.02
Strain (%)
(a) (b)

((a) strains at pipe edges) ((b) strain in a loosed tube cable)


Mohamad et al.( 2009)
Fig.15 Measured strains along edges and in a loosed tube

+
6

4
1st Prop
2

Fig.13 Monitoring slope instability by vertical pipe with 0 2nd Prop


optic cable 3rd Prop
-2 (26 Oct)
Elevation (m)

-4
Exc. Base
Lvl. Slab Floor
-6

-8
BOTDR
-10 8 Nov 05
-12 Incl 3 Nov 05
Incl 21 Nov 05
-14 Incl Averaged
12 Nov 05
-16BOTDR vs. conventional inclinometer
-2 0 2 4 6 8 10 12 14
Displacement (mm)

Fig.16 Comparison of Lateral Displacement


(Mohamad (2007, 2009))
Fig.14 Lateral deformation estimated from equivalent
curvature with no axial strain (Mohamad (2007, 2009))
4.5 BOTDR as Inclinometerequiped with Secant Pile Wall where most of the phenomena takes place in dark underground,
should use and take advantages of installation and monitored
data to reach the most likely mechanism among several
Mohamad et al.(2007, 2009) show some BOTDR application
possibilities.
to monitor axial strain of bored piles and lateral deflection of
retaining wall of secant pile wall of a series of intersecting
reinforced concrete piles. 6 REFERENCES
The diameter of the secant wall was 450mm and a pair of
optical fiber cable was installed along the reinforced steel was The Committee of Inquiry, 2005,”Report on the Incident at the
installed along steel cage giving a pretension of 2000µstrain. MRT Circleline Worksite that led to the Collapse of the
Another single optic fiber loose tubed cable was equipped along Nicoll Highway on 20 April 2004,” Ministry of Manpower,
the monitored pile. Fig.14 shows how to evaluate lateral Government of Singapore
deformation from the edge strain along a pile. Since the edge Dunnicliff J. 1988. “Geotechnical Instrumentation for
strains are induced by mix modes of axial compression and
bending deformation. To obtain bending strains, axial
Monitoring Field Performance ,” John Wiley & Sons,
New York (USA)
compression should be separated. These axial components
should be subtracted from the apparent edge strains as shown in Iwasaki, Y. 2008, “Basic Study of Carrier-phase tracking of
Fig.14. GPS method at Mikage site,” GRI report
Temperature is another factor that affects upon the apparent
strain values. Independent Panel to Review Cause of Teton Dam Failure,
Fig.15 shows an example of the measured strains of BOTDR 1976, “Report to the U.S. Department of the Interior and
for retaining wall after excavation. The excavated level was State of Idaho on Failure of Teton Dam. Idaho Falls,” Idaho.
about -4.5m. Fig.15 (a) shows two edge strains of outside and December 1976
nearside of the pipe as well as axial strain of the averaged Nakano, M., Yamazaki, H., and Okuno, M., 2003, “The
values. earthquake disaster prevention monitoring system using
Fig.15 (b) shows rather constant temperature in the soil up to optical fiber technology,” JSCE Journal of Earthquake
a level of -6.5m and gradually decreases to -13 degree at the top Engineering, Vol.27, pp.1-4
of the pipe. Mohamad, H., Bennett, P.J., Soga, K., Klar, A. & Pellow,
The air temperature was lower than the in the ground, A.2007.” Distributed Optical Fibre Strain Sensing in a
temperature decreases from near the ground surface to the top of Secant Piled Wall,” Proceedings of the Seventh
the pile. International Symposium on Field Measurements in
Fig.16 shows comparison between a lateral deformation Geomechanics (FMGM2007), ASCE Geotechnical Special
curve obtained from BOTDR and two curves by conventional Publication, No. 175
inclinometer measurements. It is noticed that the averaged curve Mohamad, H., Soga,K., & Bennett, P. J.,2009, “Fibre optic
of the two deflection curves from the conventional inclinometer installation techniques for pile instrumentation,” Proc.,17th
corresponds well to the one from BOTDR. ICSMGE(in print)
Compare to the conventional inclinometer of manual Solava and Delatte, 2003, “Lessons from the Failure of the
handling, inclinometer system by BOTDR provides automatic Teton Dam,” Proceedings of the 3rd ASCE Forensics
and realtime evaluation of deformation of retaining wall. Congress, October 19 - 21, 2003, San Diego, California
Inclinometer measures inclination angle relative to gravity.
BOTDR measures a pair of the edge strains under bending
mode. It should be notice that the distance between two edges
must be wide enough to keep acceptable accuracy of bending
deformation.
At present, the facility to send and receive the signal as well
as analyzing the BOTDR is rather expensive, In the near future,
the cost effective system is expected available and become
useful especially for big and deep excavation works.

5 CONCLUSION
In the first part, two comparative case studies are overviewed.
Teton Dam that was failed by piping. No further detailed
mechanism is clarified due to not enough instrumentation.
In the collapse of Nicoll Highway, though the data were not
collected as it should be the available data of inclination of the
subway construction nearby the Highway show clear fact of one
sided deformation of the wall from which the geotechnical
condition for the design was anticipated different from the
design condition. If the construction had been modified, the
collapse might well be avoided.
After the failure, the results of instrumentation contributed to
identify the process of to the failure from the geotechnical point
of view.
In the second part, review of sensors is made including some
edge cutting technologies of two sensors of Carrier-phase
tracking of GPS method and BOTDR.
These sensing technologies in addition with other sensor like
MEM (Micro-machine Electronic Machine) sensor are still on a
way of development.
Forensic engineers, especially in the field of geotechnical field
Chapter 12
LEGAL PROCESS AND JURISPRUDENCE

Dhirendra S. Saxena (Sax), P.E., F NAFE


Senior Principal and Chief Consultant, ASC geosciences, inc., Lakeland, Florida, USA

9.1 INTRODUCTION

For geotechnical engineers it is only a myth to believe that practicing perfect engineering, or
conforming to normal standards of care, will provide immunity from civil liability. Unfortunately,
when problems or failures occur, all parties including engineers get named in the lawsuit
regardless of their innocence. In USA, attorneys sue everyone involved with a damaged project
to secure compensation or claim to their liability insurance limit. A practicing geotechnical
engineer cannot provide services without the fear of a lawsuit. Strategies for limiting liability
range from assessing risk to securing professional liability insurance and including limitation of
liability classes in consulting contracts between engineer and the client. Although experts
retained by opposing parties generally disagree on issues resulting from differences in
professional judgment, they are invaluable to the jurisprudence system in America.

A new discipline known as forensic geotechnical engineering (FGE) has been created to deal
with investigation of soil-interaction related failures of engineered facilities or structures.
Services of forensic geotechnical engineers, experienced in the jurisprudence system, are
generally commissioned to investigate such failures. They also prepare reports, and
sometimes provide expert witness testimony, in an attempt to assist their client in
understanding the cause and liability for the failure. Because of the adversarial nature of most
failure investigations many forensic engineers are generally under pressure to give an opinion
that benefits their client’s position. Accordingly, the technical facts are not presented, yielding
to this kind of pressure sometime leads to inappropriate occurrences, practicing outside one’s
competence, manipulation of facts, crafted testimony, inadequate or defective investigations,
misrepresentation of standards of practice, or other ethical violations.

A forensic geotechnical engineer looks at a pile of rubble as a problem that needs to be solved.
Each broken piece in the pile tells a story about the forces that acted upon it and somewhere
in the twisted mass of intertwined building components is the first facture surface that resulted
in the progressive collapse of the structure. At the end of the failure analysis process, which
may include laboratory testing of critical items and hours of analysis in the office, a clear
picture of what happened begins to develop, and the forensic engineer must then be able to
explain the mode of failure and the forces, action, or process that caused the failure. The
explanation may be a simple letter, a detailed report, or expert testimony in the courtroom
(Bell, 2007).

“What happened?”, and “Why did it happen?” are usually the first two questions asked of the
forensic engineer. Of course, these are often followed by, “How can it be fixed?”, and all too
often, “Who’s fault is it?”, and “Who is going to pay?” (Bell, 2007)

When something breaks, falls down, or otherwise fails to perform as it was intended, our
society demands to know why.

This inevitably becomes the job of the forensic engineer, who must be equally comfortable
in the field, the office, and the courtroom while providing an objective analysis of the failure,
whether the facts are helpful or harmful to the client’s interests.

FGE prepares civil engineers to read, think, speak, and analyze like a lawyer. In addition, it
familiarizes him with jurisprudence system so that he is much able to understand and deal with
legal issues since he has to work closely with statutes and regulations may become involved in
litigation, or who may serve as an expert witness.

9.1.1 Who Wants to Know What

A forensic geotechnical engineer needs to answer questions, such as:

• The building owner who wants to know why the building fell down and who is
going to pay for it.

• The insurance adjuster who needs to determine if the cause of the collapse
is covered under the building owner’s insurance policy, and if it was the
result of a defect that someone else might be liable for.

• The contractor and subcontractors, who built the building and their liability
carriers, want to know if the defect is construction-related and if so, which
trade or trades are responsible.

• The design professionals and their liability carriers want t o know if the
defect is design-related.

Each of these entities may retain their own forensic engineers to assist them with understanding how
they may or may not be responsible.

There will be attorneys for each entity who will evaluate and argue how the law should
be applied to the insurance policies and contracts. The forensic engineer’s opinion of
what happened may be explored during a deposition in which the attorneys for all sides have
the opportunity to probe the engineer as to the process used and the basis for the
opinion. If the parties cannot reach a settlement, the forensic engineer may be called to
provide expert testimony in the courtroom, with the goal of helping the judge and/or jury
understand the facts and how they relate to the particular incident.

Opposing attorneys normally attempt to discredit the engineer through cross-examination with
the hopes of revealing some flaw in the forensic engineer’s thought process.

The forensic engineer must be able to apply the art and science of engineering during the
entire process, from the extreme conditions encountered in the field, through the intense
pressure experienced in the courtroom, and never lose sight that seeking the truth is all
that matters.

By understanding what went wrong, the forensic engineer is able to add to the body of
knowledge that hopefully will prevent a similar occurrence from happening again.

9.2 TECHNICAL ISSUES

Within the framework of dispute relating to construction, particularly within the jurisprudence
system, forensic engineering may include investigation of the physical causes of accidents and
other sources of claims and litigation, preparation of engineering reports, testimony at hearings

2
and trials in administrative or judicial proceedings, and the rendition of advisory opinions to
assist in the resolution of disputes affecting life or property.

Within the construction industry, disputes usually result from unfulfilled expectations of the
project developers and purchasers, from errors or omissions of the design team, or from
construction or material defects. Non-technical issues, principally time and cost, are frequently
considerations and unanticipated influences such as weather and unforeseen conditions that
may negatively influence the performance of construction.

In the “perfectly” constructed project, the expectations of the owner/developer are completely
understood by the design team and are converted to a clear, comprehensive set of drawings
and specifications. After execution of an equitable contract outlining the terms and conditions,
schedule, and compensation arrangements, the competent construction team executes the
construction according to the design within budget and on schedule. And finally, the
expectations of the owner/developer/occupant are met. Unfortunately, in our imperfect world,
few projects are perfect. Expectations are not met, designs are inadequate, construction is
improper, materials fail, and nature exerts its influence. Whether before, or after calling the
lawyers, a forensic engineer is usually needed to provide the technical expertise to investigate
and answer the questions.

The process used by the forensic engineer is an adaptation of the scientific method, “the
process by which scientists attempt to construct an accurate (that is, reliable, consistent and
non-arbitrary) representation of the world.” The scientific method has four steps:

1. Observation and description of a phenomenon (an occurrence, observation, event,


or trend) or group of phenomena i.e. identify the problem.
2. Formulation of a hypothesis (a theory or statement regarding cause and effect) to
explain the phenomena, i.e. decide on the procedure.
3. Use of the hypothesis to predict quantitatively the results of new observations.
4. Performance of experimental tests of the predictions with properly performed
experiments, i.e. collect the data and perform the analysis.

It is the fourth step that often is overlooked. Because the experimental tests may lend either to
confirmation of the hypothesis or to the ruling out of the hypothesis, the scientific method
requires that a hypothesis be ruled out or modified if its predictions are clearly and repeatedly
incompatible with experimental tests.

Further, no matter how elegant or desirable a hypothesis or theory is its predictions must agree
with experimental results if it is a valid description of nature.

The scientific method attempts to minimize the influence of the forensic engineer’s bias or
prejudice on the outcome of an experiment. That is, when testing a theory, the engineer may
have a preference for one outcome or another, and it is important that this preference not
influence the results or their interpretation. Another common mistake is to ignore or rule out
data (measurements or information) which do not support the engineer’s theory. Ideally, the
engineer is open to the possibility that the theory is correct or incorrect. Sometimes, however,
the engineer may have a strong belief that the hypothesis is true (or false), or feels internal or
external pressure to get a specific result. In that case, there may be a psychological tendency
to find “something wrong” with data which do not support the engineer’s expectations, while
data which do agree with those expectations may not be checked as carefully.

9.3 ETHICAL ISSUES

3
Ethics is defined as a discipline dealing with good and bad, right and wrong, moral duty and
obligation.

Why does a group need a code of ethics? One does not have to look any further than the
daily newspaper to find that unethical behavior exists in the highest levels of government,
in the clergy and in virtually every segment of society. Social scientists have studied and
found that in any group a certain percentage will deviate from the norm. Therefore,
there is the need to codify and regulate or influence the behavior of various professions and
groups. But, when we speak of a code of ethics applicable to engineers do we mean
merely the business practice with respect to another professional or a client, or are there
higher goals and standards to which a professional engineer should aspire?

The NSPE code of ethics states that in the fulfillment of their professional duties engineers
shall:

“1. Hold paramount the safety, health and welfare of the public in the performance of
their professional duties; 2. Perform services only in the areas of their competence;
3. Issue public statements only in an objective and truthful manner; 4. Act in
professional matters for each employer or client as faithful agents or trustees; 5. Avoid
deceptive acts in the solicitation of professional employment.”

Upon close examination of the Canons, it is seen that none are in conflict with the practice of
forensic engineering. The forensic engineer should adhere to the SPE Cannons not only as a
member of NSPE but also as an engineer (Dixon, 1992).

Engineers are recognized as the most ethical professionals above doctors, accountants, and
lawyers in a survey of 200 CEO’s of national companies.

While ethical core values for engineers are generally identified as integrity, honesty,
faithfulness, charity, responsibility, and self discipline the classical seven deadly sins are
described as pride, greed, anger, envy, gluttony, lechery, and sloth.

It is clear that there are problems in ensuring the integrity of expert testimony. There are no
minimum qualifications for becoming an expert witness, nor any means of regulating them.
Code of ethics are voluntary and not often enforced (Grover et al., 2003).

In summary, the middle of the road rule for an ethical engineer is to be honest, tell the truth
no matter how bad it may seem, and let your conscience be your guide (Baker, 1997).

9.4 LEGAL ISSUES

The legal considerations of these forensic geotechnical engineering services illustrate the
reality that the engineering investigation of a failure incident is a fact-finding mission that
results in uncovering the probable causes of that failure. It concentrates on the identification
of hidden clues. The procedures adopted for the analysis, testing, opinions, and written reports
should be able to satisfy even legal scrutiny of their validity.

9.4.1 Role of Expert Witness

Experts are vitally important to universal jurisprudence. Expert witness can assist retained
counsel in fact gathering, request for documents, and evaluation of evidence; fashioning precise
questions for opposing experts and through candid discussions of strengths and weaknesses of
the case; retained counsel to properly qualify and question an expert witness to maximize his

4
impact on the jury; and expansion of the traditional testifying roles in the litigation process.

An expert in forensic engineering is not automatically dubbed an “expert witness”. An


individual is not granted the privileges of an expert witness until a legal forum confers that
recognition upon a properly qualified professional.

Experts must possess specialized knowledge of science, profession, business or occupation


that is beyond the experience of the lay juror and can assist the fact finder (Grover, 2003).
Additionally, expert witnesses in the field of engineering are governed by their own code of
ethics.

Well over 90% of civil cases settle prior to expert witnesses being called for a trial. As such
experts can expect that the majority of testimony that they give will be given at deposition.
Accordingly, the experts need to excel during their depositions as they are key element of the
discovery process.

If the expert witness is a litigation consultant, he/she is not subject to discovery by the opposing
counsel. If disclosed as an expert, then he/she is subject to discovery by the opposing counsel.

Engineers should perform services only in areas of their competence and they should
undertake assignments when qualified by education and/or experience in the specific fields
involved. Direct examination is an expert’s opportunity to persuade the jury that his/her opinion
should be believed. The goal during direct testimony is to persuade the jury to find one’s
opinion creditable. To connect with the jury an expert needs to understand what jurors want
and employ the best ways to communicate with them.

Experienced expert prepares a well written, persuasive expert based upon reliable accepted
methodology. Expert witness can express his/her opinion in an unambiguous and legally
sufficient manner. Expert witness can dramatically improve their performance by being good
teachers, both to the lawyers who hire them, and to jurors and other fact- finders. They must
keep it simple with illustrations, explanation and non-technical language.

The four key elements of a successful expert witness assignment are (SEAK, 2006):

9.4.1.1 Part I: Prevention

The most valuable experts isolate themselves from attacks and deny opposing counsel
ammunition to attack their credentials and credibility. It helps to identify a detailed checklist of
potential areas of attack that experts may be subject to regarding their credentials and
credibility including: every word on their CVs, past testimony, their image, controversial or
political associations, missing credentials, fee schedules, fee agreements, marketing
materials, web page, speeches, work on past case, apparent and actual conflicts of interest,
non-related litigation, hobbies, professional complaints or discipline, presentations and writings.

The best experts form and express airtight opinions and opinions that hold up under the most
rigorous scrutiny and cross-examination. There are many ways in which opposing counsel is
able to poke holes in an expert’s opinion. The expert must provide specific actions to bullet-
proof his opinion including, proper case and client selection, avoiding time crunches, using
careful and confident language, not overstating or understand facts or opinions, consistency,
dealing with the opinions of other experts, knowing exactly what needs to be proved, testing
alternative theories, taking careful and precise measurements, being well-trained and well-
versed in any computer program used, verifying computer results, leaving no stone
unturned, taking photographs, verifying factual assumption, gaining as much first hand

5
knowledge as possible, thoroughly researching the issues at hand, obtaining and carefully
reviewing all relevant documents, not sharing draft reports with counsel, and avoiding “junk
science”.

9.4.1.2 Part II: Preparation

Peak performance requires proper disciplined preparation done correctly prior to deposition
and trial. Well-prepared experts are able to deliver confident testimony, deal with cross-
examination far more effectively, and are in much better position to articulate and defend their
opinions. It should include advanced techniques that can and should be used to prepare for
depositions, direct examination and cross-examination.
9.4.1.3 Part III: Performance

The best experts recognize that most cases are won and lost in the discovery phase and that
the expert’s deposition is a crucial, often outcome determinative, component of the case. In
order to excel at the highest level during a deposition, experts need to be able to recognize
and defeat opposing counsel’s deposition tactics and recognize how these tactics differ from
those used during trial.

The best experts deliver powerful and understandable direct testimony. They also learn to
explain and demonstrate numerous advanced techniques for delivering captivating, memorable
and persuasive direct expert testimony. These advance techniques include, showing – not
telling, getting to the point up front and explaining later, being well-prepared and well-
organized, making the complex simple, entertaining, being likeable, highlighting your most
relevant qualifications, working on a smooth flow and style, getting out of the jury box early and
often, using visual aids that work, aggressively self-editing, employing powerful, memorable
analogies, showing your human side and bonding with the jury, using precise language,
using confident language, employing short preview and review summaries, using numbered
lists, citing references, speaking conversationally, conforming your testimony to the theme of
the case, and reading and reacting to the jury.

9.4.1.4 Part IV: Practice

The best experts skillfully answer cross-examination questions and explain how the question
could have been avoided, how they could have and should have prepared to answer the
question, identifying the tactic that counsel is using and delivering a response that defeats the
tactic and/or allows the expert to go on the offensive.

9.4.2 Legal Considerations

In order to fully evaluate the forensic expert’s role in the legal process, the expert should
understand the definitions and concepts of a dispute resolution.

The professional services provided in the investigative process and the duties expected of the
expert witness are generally the same no matter what method of formal dispute resolution is
selected by the client. Following is an o v ervi ew o f the v arious methods of dispute
resolution that may be encountered (Task Committee on Guidelines for Failure Investigation,
1989).

9.4.2.1 Civil Litigation

This process occurs in a court of justice having jurisdiction in the dispute. The court is
presided over by a judge who decides on the applications of the law and procedure. These

6
decisions usually are determined by a jury of lay-persons or by the judge in the absence of a
jury. The decision of this court and jury may be subjected to appeals and retrials.

9.4.2.2 Arbitration

Instead of a judge and/or jury, an arbitrator or a panel of arbitrators are used in a private
proceeding. Unlike litigation, arbitration eliminates or significantly relaxes the discovery
process and is usually limited to the production of documents to an extent determined by the
arbitrators.

9.4.2.3 Administrative Hearings or Special Investigations

Hearings are held by governmental authorities to resolve disputes or to gather information of


public opinion prior to making an administrative decision, or for inquiries into the integrity of the
structure or the cause of a failure of an engineering facility.

9.4.2.4 Private Litigation

This form of dispute resolution requires both parties to agree to resolve their differences in a
private court that is created, and paid for, by both parties. A mutually respected individual,
such as a retired jurist, is chosen as a judge, and civil litigation rules are followed.

9.4.2.5 Mediation

This method of “non binding” dispute resolution is used to resolve certain construction-oriented
disputes. A mediator acceptable to both sides is selected and probes the disputants to
determine their position relative to “minimum” settlement. He or she then strives to attain a
middle ground that both can accept comfortably.

9.5 OVERVIEW

A well organized, methodical approach to failure investigation is needed because of the


complexities of modern engineered facilities and the vast variety of possible failure causes.
The investigative process should n o t employ a rigid “cookbook” approach, but there are
certain steps that are common to an effective failure investigation.

An emphasis on an open-minded acceptance of all pertinent data, while recognizing the


relevant information, is the fundamental basis of a failure investigation. As various hypotheses
develop and change, continuous review of previously rejected data is required.

There is seldom a single cause of a failure but rather a complex interaction of components and
forces. As a result, the outcome of a failure investigation seldom leads to absolutely
irrefutable results but rather to a most probably cause of failure. While the collection of data
and facts surrounding the failure are specific, expert opinions arising from these data may
differ widely.

The most accepted failure investigation findings will be the one employing a qualified
investigation team that presents the most plausible failure scenario, based on well documented
support data. (Task Committee on Guidelines for Failure Investigation, 1989).

A close coordination between the forensic geotechnical engineer, his client, and client’s

7
attorney is important in order to understand the facts as they relate to a particular incident.
This teamwork normally lends credence to their case and is instrumental in making it a winning
combination.

Two case histories that include all elements of forensic geotechnical engineering as well as
legal issues of jurisprudence system are included with this chapter. They include situations
where forensic geotechnical engineering was effectively utilized to identify, investigate,
remediate, litigate, and resolute legal issues.

9.6 REFERENCES

ASC geosciences, inc. (2002) – Report of Field Investigation, Data Evaluation, and Engineering
Consultation Services, ASC geosciences, inc., Report No. 02L2016.

Baker, P.E., John A. (1997) – Professional Engineering Ethics versus Expert Witness
Requirements, Los Angeles Section ASCE, Forensic Engineering Technical Group.

Bell, P.E., John T. (2007) – What is Forensic Engineering?, Florida Engineering Society
Journal, May 2007, pp.8-9.

Dixon, E. Joyce (1992) – The NSPE Code of Ethics Applied to Forensic Engineering, Journal
of the National Academy of Forensic Engineers, Vol. IX No. 1.

Grover, J.D., P.E., J.L. (2003) Ethical Considerations for Expert Witnesses in Forensic
Engineering, Ethical Dilemmas of Technical Forensic Practice, pp. 441-52.

Saxena, D.S., P.E. (2005) – "Forensic (Geo-technical and Foundation) Engineering Case
History", National Academy of Forensic Engineers (NAFE) Seminar, Chicago, Illinois, 10 July
2005.

Saxena, D.S., P.E. (2007) – "Case Studies in Forensic Geotechnical and Foundation
Engineering", 6th International Conference on Case Histories in Geotechnical Engineering,
Arlington, VA, Paper No. OSP-2.

SEAK inc. (2006) – Advanced Testifying Skills for Experts – The Master’s Program, pp. 4-11.
Task Committee on Guidelines for Failure Investigation (1989) – Guidelines for Failure
Investigation, American Society of Civil Engineer (ASCE).

B. PART 2

9.7 Appendices

9.7.1 Case History One

Project specifics and forensic facts summary;

• The backyard slope of a one-story dwelling subsided abruptly, failed, and

8
slid into the lake along with the rip-rap from lake edge. It also extended
into the neighboring property on the south side.

• The lakefront backyard of a residence was damaged and stability of the


structure was threatened, and removal/replacement of the completed
structure was considered a viable yet costly option.

• The owner filed a claim against the developer/builder for negligence and for
not informing them of a potentially unstable pre-existing condition.

• The owner secured the services of a Forensic Geotechnical Engineer


(FGE) to investigate, remediate, and assist in litigation as well as serve as an
expert witness during the resolution of dispute and legal issues.

• The developer/builder offered a band- aid solution that the owner then
rejected upon advice of the FGE.

Figures 1-4 illustrate the salient features of this case history.

Fig. 1 Close-up view of the subsided backyard

9
Fig. 2 View of subsided slope of adjoining house

Fig. 3 View of sodded and finished backyard

10
Fig. 4 Reoccurrence of slope failure in finished
backyard two weeks after completion

9.7.1.1 Forensic Field Exploration and Subsurface Condition Evaluation


Subsurface stratigraphy beneath the site, as illustrated in Figure 7, consisted of very soft,
loose low strength subsoils to 15.0 ft depth. These low strength subsoils consisting of highly
plastic clays (mc 113%, LL-107, PI-89) which were unstable under the weight of the additional
fill material.

Slope stability analyses were performed for the original subsurface profile as well as the
contractor/developer proposed restructured slope that yielded Factors of Safety (F.S.) of 0.87
and 0.93, respectively, for the two conditions.

9.7.1.2 Observations and Findings

The expert’s investigation at the project development site revealed that;

• Slope failure was triggered by the presence of unconsolidated sediment


layers under the site fill.

• Unstable sediments to continue to consolidate resulting in continued


movement of the back of the house and the backyard.

• A retaining wall should have been constructed prior to the house being
built to contain and stabilize the soils beneath the house foundation.

11
Fig. 7 Subsurface stratigraphy identifying delineation of compressible phosphatic clay

Upon the advice of their legal counsel the owner put the developer on notice, and elected to
proceed with the proper fix pending resolution of litigation and the claim.

9.7.1.3 Remediation

Based on findings and observations from the FGE investigation, a two part remediation
program was recommended and consisted of:

1. Installation of helical anchor piers under the exterior wall footing, along the back of
the house, for the underpinning stabilization.

2. Installation of a helical anchor bulkhead to stabilize the backyard along the lake.

3. Based upon field monitoring of pier installation operations, anchored timber


bulkhead, post stabilization walk through, and review/evaluation of the field
stabilization data, it was concluded that the methods and procedures utilized and
installed by the specialty contractor were effective, satisfactory, and acceptable.
Slope stability analyses of this remediated structure indicated a Factor of Safety
(F.S.) of 1.57.

Figures 6-9 illustrate various sequences of the underpinning and bulkhead stabilization program.

12
Fig. 6 View of support cap and bracket assembly over
helical support anchor under the footing

Fig. 7 Close-up of geotextile lay down and wrap up


at helical anchor tie back on timber bulkhead

13
Fig. 8 Additional fill being dozed into the anchored geotextile

Fig. 9 View of completed and restored backyard

9.7.1.4 Concluding Remarks

1. Following the close coordination between the forensic geotechnical engineer, the
owner, and the remediation contractor and based upon results of remediation
monitoring it was determined that the repair of the residence structure had been
achieved satisfactorily.

2. The final repair resulted in restoration of the residential structure and backyard to its
originally planned and constructed stage, as illustrated in Figure 9.

3. As a result of the monitored and satisfactory remediation program, the consultant


recommended acceptance of the restored structure.

4. The property owner retained an attorney who sent a notice of claim, and remediation

14
cost summary, to the developer/builder. Following litigation, arbitration the property
owner, through their legal counsel, was able to recover these remediation costs
along with attorney fees.

9.7.2 Case History Two

This case history identifies a request received from a building owner for a forensic engineering
review to investigate a situation where a severe site specific soil-structure deficiency occurred
and caused post construction damage to an office building in west central Florida. Evaluations
were also made to see if repairing or replacement of the structure was, in fact, necessary.

9.7.2.1 Project Description

Construction of an office building was completed in June 2006 as per design plans, permitting,
and pre-construction technical support from the civil engineer. Foundation support included
installation of 12 inch diameter timber piling, driven to 24 tons capacity (as per the pile driving
records prepared by a geotechnical engineering representative at the site) along the exterior
perimeter grade beam supporting the load bearing walls. The interior slab was a 4 inch thick
fiber mesh concrete supported on compacted soil for the entire building footprint.

Soon after occupying the building, the owner(s) observed certain impacts on their property.
Concerns identified by the owner(s) had included:

• noticeable settlement and deflection of the interior soil supported floor


slab.

• visible cracks throughout the interior of the building.

• differential settlement of building foundations (exterior piles and interior


floor slab) resulting in misalignment of the doors.

• possible damage to the below grade utility lines.

• substantial deflection distress and development of cracks throughout the


interior and foundation floor slab within the office building as a
consequence of differential settlement due to the consolidation of
existing peat material within the building footprint area. Additionally, total
and differential settlement of the foundation support system and resulting
distress appeared to be of a continuing and progressive nature.

Figures 1 – 8 illustrate the salient features of the case history and owner’s justifiable concern.

15
Fig. 1 Two close up views of sinking floor at conference room door

Fig. 2 Close up view of interior partition wall separation from sinking floor slab

Fig. 4 Close up view of wall crack showing new crack and additional movement

16
Fig. 8 Separation and sinking of slab along building exterior

A 2-stage program was carried out by the forensic geotechnical consultant in conjunction with
filing of a claim by the building owner.

9.7.2.2 Stage 1 Investigation

Stage 1 consisted of a review of the field inspection of the structure, review of the project
foundation drawings, and pre-design geotechnical exploration report.

The findings, comments, and conclusions derived from the Stage 1 investigation at the project
site revealed that:

• there were wide spread separations between the walls and the floors,
and between partition walls and ceilings. The separations appeared to
be greatest toward the center of the building.

• a floor elevation survey of the building floor slab conducted in March


2007 indicated partial settlement of 7.5 cms (3 inches) at the center of
the building, as illustrated in Fig. 9.

• subsoil test borings performed by the pre-design geotechnical engineer


were 12.0 m (40 ft) deep and consisted of organic compressive soils.
Pressure treated timber piles with an embedment depth of 20 to 25 ft (for
a 12 inch diameter) were utilized and a net allowable capacity of 12 tons
was recommended.

• during construction pile lengths of 16.0 m to 18.0 m (50 to 55 ft) were


used. Furthermore, 30.0 cm (12 inch) square prestressed precast
concrete (PCC) piles were substituted in lieu of pressure treated timber
piles. In addition, the contractor elected to use an unusually high energy
hammer in lieu of the recommended hammer of low to moderate energy
for the timber pile.

17
• All three of the borings had significant weight-of-rod (WOR) and weight-
of-hammer (WOH) zones throughout the in filled soils. In those cases
where underlying sandy soils were encountered, it appeared that the
deep sands were in a more stable condition. This profile indicated an
ongoing raveling associated with the relic sinkhole, or an indication that
the in filled organic soils were weak and very loose.

• the interior floor slab was only soil supported and not structural (with
interior grade beam supported).

• proper pile capacity determination, its installation, and structural grade


beam supported floor slab was not performed.

9.7.2.3 Stage 2 Investigation

In an effort to properly address, investigate, and evaluate the alleged subsoil deficiency a
detailed subsoil exploration program was undertaken by advancing test borings to depths
ranging from 40.0 m (130 ft) to 60.0 m (190 ft). Test boring stratigraphy showing internal
erosion and solution features is illustrated in Fig. 5. It consisted of elements listed below.

• the site was underlain by very poor soil conditions not capable of
providing support for any type of construction that assumes bearing to
be developed from the underlying soils.

• upper fill soils were mixed with debris underlain by predominately


organic soils with organic content ranging from 5% (at 80 ft depth) to
18.5% (at 60 ft depth) and 85% (at 135 ft depth).

• no limestone was encountered to the termination depth of 43.0 m (135 ft)


to 60.0 m (190 ft). Project site stratigraphy showing internal erosion and
solution features is illustrated in Fig. 5.

18
Fig. 5 Project site stratigraphy showing internal erosion and solution features

• damage to the building was the result of factors including consolidation


of significant factors including consolidation of significant depth of very
loose organic soils (located within a large relic sinkhole) along with the
consolidation of buried debris located within or just below the develop fill
soils and loading street zone.

• the old aerial photographs from 1941 through 2005 showed the wetland
feature and confirmed that the building was constructed in a wetland
drainage basin area over deposits of very deep, very loose organic soils.

9.7.2.4 Concluding Remarks

• the owner reported that the damage had been ongoing since the building
was constructed, which was consistent with a concrete slab supported by
very loose, compressible, debris laden organic soils. The pile supported
perimeter walls supported by the timber pile showed very little damage,
indicating that no significant differential movement of the perimeter wall
had yet occurred.

• it was determined that deleterious soil condition was known to exist prior
to construction of the building and is the reason that a pile foundation
was recommended by the project geotechnical engineer. However, the
foundation design only accounted for support of the perimeter walls and
did not address the need for support of the slab. It was concluded that
this design flaw/oversight was the primary reason that the building had
sustained substantial interior damage.

• it was also concluded that no feasible and effective options were

19
available to remedy the relic sinkhole condition and that the wetland site
over deposit of very deep organic soil was not a suitable site for this
building construction.

• additionally, as a direct result of structural and geotechnical engineer’s


severe design deficiency, as well as contractor’s failure to exercise, due
care the building had sustained substantial irreversible damage.

• the property owner’s claim to the developer/builder was successful and


the owner was able to recover a major portion of the damage cost
resulting from design oversight and relic sinkhole related damage.

20
Chapter 13.

POTENTIAL ROLE OF RELIABILITY IN FORENSIC GEOTECHNICAL ENGINEERING

K K Phoon
Department of Civil Engineering, National University of Singapore, Singapore 117576
E-mail cvepkk@nus.edu.sg

G L Sivakumar Babu
Department of Civil Engineering Indian Institute of Science, Bangalore, India, 560012
E-mail: gls@civil.iisc.ernet.in

M Uzielli
Georisk Engineering S.r.l., Florence, Italy
E-mail : muz@georisk.eu

ABSTRACT

This chapter discusses some possible roles for reliability and risk in forensic geotechnical engineering. A
preliminary statistical framework is presented to quantify the difference between expected and observed
performance in the presence of unavoidable and potentially significant geotechnical variabilities. The probable
ranges of geotechnical variabilities are reviewed. Other potentially useful results in the recent reliability and
risk literature are highlighted. The intention of this chapter is to stimulate further discussions and research in
this important but somewhat overlooked area.

1
1. INTRODUCTION
Forensic geotechnical engineering (FGE) is a relatively new field. ISSMGE TC40 was formed only in 2005
under Dr VVS Rao as chair. Subcommittee 6 was convened under Dr KK Phoon to explore the potential role of
reliability in FGE. A cursory Google search using keywords “reliability, forensic, and geotechnical” shows that
this concept is entirely new. This chapter seeks to explore if reliability/risk concepts are potentially useful to
forensic geotechnical engineering. This study is very preliminary given the dearth of previous literature. Brown
(2006) noted that: “The concept of reliability as ‘the likelihood that a system will perform in an acceptable
manner’ (Bea, 2006) is important in forensic geotechnical engineering.” No details were discussed.

Engineering with natural materials such as soils and rocks is difficult and challenging owing to the variability
and the uncertainties associated with engineering decisions. In fact, one key aspect that distinguishes
geotechnical engineering from structural engineering is the large variability (natural/intrinsic variability, testing
errors, and transformation uncertainties introduced when measured parameters are converted to engineering
parameters) related to naturally occurring geomaterials. In the case of quality of concrete, variability (in terms
of coefficient of variation or COV) can be classified as follows (Rétháti 1988 citing 1965 specification of the
American Concrete Institute):

COV < 10% excellent


COV = 10 – 15% good
COV = 15 – 20% satisfactory
COV > 20% bad

The COVs related to the natural variability of geomaterials as well as those pertaining to measurements errors
and transformation uncertainties can be much larger and do not fall within a narrow range. Phoon & Kulhawy
(1999a; 1999b) compiled an extensive database of soil statistics for calibration of simplified reliability-based
foundation design equations. Details are given in Section 3. One key observation that emerged from their
calibration studies is that geotechnical variability of common design parameters can be broadly grouped using
the following qualitative categories:

Geotechnical parameter Property variability COV (%)


Undrained shear strength Low 10 – 30
Medium 30 – 50
High 50 – 70
Effective stress friction angle Low 5 – 10
Medium 10 – 15
High 15 – 20
Horizontal stress coefficient Low 30 - 50
Medium 50 – 70
High 70 - 90

These observations are not entirely surprising given that the volume of geo-materials investigated by direct or
indirect means is extremely small in comparison to the volume of interest. Chiles and Delfiner (1999) cited
volume fractions investigated at Brent field, North sea, to be 10-9 for cores and cuttings and 10-6 for logging.
The immediate impact of these observations is that Leonards (1982) definition of “failure” - unacceptable
difference between expected and observed performance – cannot be evaluated in a meaningful way using
deterministic methods. To elaborate, “expected performance” must vary given the backdrop of potentially
significant geotechnical variability. We can argue that this variability is “inherent”, in the sense of being
irreducible, in geotechnical engineering based on the prevailing standards of best practice. Some reduction in
some of the components of variability is always possible by carrying out more tests, but it is generally not
realistic to expect orders of magnitudes reduction. In view of this widely acknowledged and widely established
reality in geotechnical engineering, it is more realistic and perhaps more credible to quantify “unacceptable
difference” in a statistical sense. In broad terms, forensic engineering is related to the investigation of failures
with the view of rendering an opinion regarding responsibility. Hence, an objective statistical measure of
“unacceptable difference” (specifically, a difference not explainable by underlying variability) should provide
useful additional information in the formulation of such an opinion. It should be emphasized, however, that the
statistical criteria tentatively proposed in the following sections should never replace – nor downplay the
importance of - preliminary qualitative assessments of the quality of design based on geotechnical knowledge

2
and experience. Rather, they should serve as a complementary tool that could improve the degree of objectivity
in forensic geotechnical assessments.

2. STATISTICAL FRAMEWORK
2.1 Rejection criteria for factor of safety
A preliminary statistical framework can be constructed by using the widely used global factor of safety (FS) as
an indicator of performance. It is plausible to imagine FS as a lognormal random variable with mean = μ,
standard deviation = σ and coefficient of variation = σ/μ = θ. Silva et al. (2008) provided some empirical
evidence to support this lognormal assumption for slope problems. The lognormal distribution satisfies the most
basic constraint that most load and strength parameters are non-negative. It is implicit in this model that one
speaks of the “likelihood” of failure, rather than failure in our usual absolute deterministic sense. Fig. 1
illustrates a lognormal distribution for μ = 3 and θ = 0.3. It can be shown that ln(FS) is normally distributed
with mean (λ) and standard deviation (ξ) given by:

λ = ln(μ) – 0.5ξ2 and ξ2 = ln(1+θ2). (1)

The equivalent normal distribution for Fig. 1 is shown in Fig. 2 with λ = 1.056 and ξ = 0.29. Note that the
standard deviation of ln(FS) (ξ) is approximately equal to the COV of FS (θ) up to θ of about 0.5.

0.6

Fail
Probability density

0.4

0.2

0
0 2 4 6
FS

Fig.1 Log-normal distribution of FS

3
1.5

Fail
Probability density

1.0

0.5

0.0
-1 0 1 2 3
ln(FS)

Fig. 2 Equivalent normal distribution

The observed performance, FS , can be viewed as a sample mean (of sample size 1) from a population of
factors of safety such as that shown in Fig. 1. It is obvious that a low value of FS near to 1 is not impossible,
although it is “unlikely” if the population follows a lognormal distribution with μ = 3 and θ = 0.3. A naïve
interpretation is that “failure” (as in FS < 1) is always possible and geotechnical variability is responsible,
rather than human errors. Sowers (1993) noted that the majority of foundation failures were due to human
shortcomings. This naïve interpretation essentially misses the key principle of hypothesis testing described
below.

Our proposed approach is to assume that: (a) “expected performance” can be described by the reliability index
and (b) a conventional hypothesis test can be performed to ascertain if the target reliability index had been
achieved in the original design based on the observed sample mean, FS . To do this, it is necessary to define
“expected performance”. If FS follows a log-normal population, the reliability index is given by:

E[ln(FS)] λ
β= = (2)
σ[ln(FS)] ξ

Numerous reliability calibration studies (e.g., Phoon et al. 1995) have shown that existing foundations are
typically designed to achieve a target reliability index (β) of about 3 (corresponding to a probability of failure of
about one in a thousand). Target reliability indices for other types of geotechnical design are not as well
established presently.

In Fig. 2, the reliability index is β = 3.64 or probability of failure = Φ-1(-β) = 0.000136, where Φ is the standard
normal cumulative distribution function. In EXCEL, Φ(.) and Φ-1(.) are computed using the NORMSDIST(.)
and NORMSINV(.) functions, respectively. Based on the definition of the reliability index and β = 3, the
following null and alternate hypotheses on the population mean [actually, the mean of ln(FS)] can be formulated
as:

H0: λ = 3ξ
H1: λ < 3ξ

4
Assuming that ξ is known and a sample size of 1, the null hypothesis is rejected at the customary 5% level of
significance if:

( ) =
( )
ln FS − λ ln FS − 3ξ
< Φ −1 (0.05) = −1.645 (3)
ξ ξ

[
FS < exp(1.355ξ) = exp 1.355 ln 1 + θ 2 ( )] (4)

The above rejection criterion provides an example of a simple numerical yardstick to evaluate “unacceptable
difference between expected and observed performance” in the presence of potentially significant geotechnical
variability. A rejection means that the observed factor of safety does not support the claim that β = 3 at the 5%
level of significance. Different values of the threshold factor of safety can be computed for different significance
levels. It is also possible for a rejection to arise because the underlying “true” geotechnical variability was
grossly over-estimated. A “do not reject” scenario means that the observed factor of safety is not unreasonably
“low” and failure may be caused by geologic “surprises”, limitations in the existing factor of safety, critical
failure mechanism not identified, etc. Again, it is possible for “do not reject” to arise because we have grossly
under-estimated the underlying geotechnical variability.

The above framework is by no means perfect and comprehensive, given the diversity and complexities of actual
failures. However, it does provide an example of an objective framework to perform an initial evaluation of the
observed factor of safety, particularly to eliminate the more obvious claim that it is an “unfortunate” realization
caused by geotechnical variability.

Fig. 3 presents the critical values of factor of safety below which the difference between expected and observed
performance is “unacceptable”. For example, if the expected performance corresponds to β = 3 and the
underlying variability corresponds to θ = 0.3, an observed factor of safety of 1.4 is too “low” and not
explainable as a random outcome from a population of FS with β = 3.0 and θ = 0.3. It will be useful to interpret
actual case studies using this chart to evaluate its usefulness in forensic geotechnical engineering and to fine-
tune it if necessary. In the general case where we observe n factors of safety, the rejection criterion for β will
become:

(
FS < exp βξ − 1.645 ξ n )
[ ( )]
(5)
= exp ln(1 + θ 2 ) β − 1.645 n

The rejection curves for β = 3.0 corresponding to various sample sizes (n) are shown in Fig. 4. If this statistical
framework proves to be useful, more sophisticated rejection criteria can be developed based on a sample
estimate of ξ (rather than the population version used in the above equations). It suffices to note here that they
do not follow the t-distribution in standard statistical texts.

5
3.0
2.8 β = 3.5

2.6

Critical factor of safety 2.4


2.2 β = 3.0
2.0
1.8
1.6 β = 2.5

1.4
1.2
1.0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
Coefficient of variation
Fig. 3 Rejection criteria for different expected performance levels - a possible role for risk in FGE based
on one observed factor of safety

3.0
2.8
n=5
2.6
Critical factor of safety

2.4
2.2 n=2

2.0
1.8
n=1
1.6
1.4
1.2
1.0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
Coefficient of variation
Fig. 4 Rejection criteria for different sample sizes of observed factors of safety based an expected
performance of β = 3.

2.2 Example
An example of the use of Fig. 3 can be illustrated with reference to the results presented in Duncan (2000)
wherein a case study of underwater slope failure was reported. The failure took place entirely within San
Francisco Bay mud, a normally consolidated, slightly organic clayey silt or silty clay of marine origin. Previous
experience in the area indicated that 1(H):1(V) slopes with a factor of safety of 1.25 were satisfactory. To
reduce the quantity of excavation, slopes of 0.875(H): 1(V) having a factor of safety of 1.17 were excavated
which failed subsequently. A risk-based back-analysis indicated a probability of failure of 18% which is

6
unacceptable though the factor of safety is 1.17. Duncan (2000) mentioned that the coefficient of variation of
shear strength parameters was high and hence the probability of failure was high. This case study can be plotted
in Fig. 3 as an open circle. It is clear that an observed FS = 1.17 cannot support the claim that β = 2.5 for any
COV larger than about 0.2. Following the same argument, we can conclude that β = 3 is not supported for any
COV larger than about 0.1. Because COV of 0.1 is the lower bound for most geotechnical problems, it is
reasonable to say that β = 3 was not achieved in the original design with a fair degree of confidence.

3. GEOTECHNICAL UNCERTAINTIES
The evaluation of geomaterial (soil and rock) parameters is one of the key design aspects that distinguish
geotechnical from structural engineering. The basic premise underlying Section 2 is that the factor of safety
(FS) is a lognormal random variable. It is possible to infer the underlying coefficients of variation of FS from
the empirical evidence presented in Silva et al. (2008). This approach of lumping all sources of geotechnical
uncertainties into a COV of FS is quite crude, but it does allow practitioners to apply the statistical framework in
Section 2 rapidly to test the claim that failure is an unfortunate “roll of the dice” caused by geotechnical
uncertainties. Some important COV thresholds for FS are given in Section 3.1.

It is preferable to distinguish between two main sources of geotechnical uncertainties explicitly. The first arises
from the evaluation of design soil properties, such as undrained shear strength and effective stress friction angle.
This source of geotechnical uncertainty is complex and depends on inherent soil variability, degree of
equipment and procedural control maintained during site investigation, and precision of the correlation model
used to relate field measurement with design soil property. A total variability analysis that lumps all these
components produces only site-specific statistics which should not be generalized to other sites. Realistic
statistical estimates of the variability of design soil properties have been established by Phoon & Kulhawy
(1999a, 1999b) and presented in Section 3.2. The second source arises from geotechnical calculation models.
Although many geotechnical calculation models are “simple”, reasonable predictions of fairly complex soil-
structure interaction behavior still can be achieved through empirical calibrations. Model factors, defined as the
ratio of the measured response to the calculated response, usually are used to correct for simplifications in the
calculation models. Recent literature includes estimation of model statistics for the calibration of deep
foundation resistance factors for AASHTO [American Association of State Highway and Transportation
Officials] (Paikowsky 2002). None of these studies addresses the applicability of model statistics beyond the
conditions implied in the database. This question mirrors the same concern expressed previously on the possible
site-specific nature of soil variabilities. More rigorous model statistics for both ultimate and serviceability limit
states are presented in Section 3.3.

It is crucial to highlight that the COV of FS is indeed exceedingly crude, because it lumps parametric and model
uncertainties into a single number.

3.1 Coefficient of variation of factor of safety


Silva et al. (2008) proposed several relationships between the annual probability of failure and the factor of
safety based on 75 projects (zoned and homogeneous earth dams, tailings dams, natural and cut slopes, and
some earth retaining structures) and expert judgment. As shown in Fig. 5, there are 4 categories of earth
structures. They are defined by Silva et al. (2008) as:
ƒ Category I—facilities designed, built, and operated with state-of-the-practice engineering. Generally these
facilities have high failure consequences;
ƒ Category II—facilities designed, built, and operated using standard engineering practice. Many ordinary
facilities fall into this category;
ƒ Category III—facilities without site-specific design and substandard construction or operation. Temporary
facilities and those with low failure consequences often fall into this category;
ƒ Category IV—facilities with little or no engineering.
Silva et al. (2008) compared the empirical data in Fig. 5 with some theoretical curves, but did not provide any
mathematical details. Their theoretical curves can be easily reproduced using the following procedure:
1. Assume that FS is lognormally distributed with parameters λ and ξ.
2. The horizontal axis of Fig. 5 is the mean factor of safety, μ.
3. If the COV of FS (θ) is sufficiently small, λ ≈ ln(μ) and ξ ≈ θ.
4. The vertical axis of Fig. 5 is the probability of failure, given by:

7
⎛0−λ⎞ ⎡ − ln (FS) ⎤
p f = Pr ob(FS < 1) = Pr ob[ln(FS) < 0] = Φ⎜⎜ ⎟⎟ = Φ ⎢
⎝ ξ ⎠ ⎣ θ ⎥⎦

Fig. 5 Annual probability of failure versus factor of safety for earth structures (Silva et al. 2008)

The theoretical lognormal curves from Silva et al. (2008) are reproduced in Fig. 6 using the above procedure
with θ = 0.072 (Category I), 0.109 (Category II), 0.174 (Category III), 0.316 (Category IV). Based on Silva et
al. (2008), it would appear that the COV of FS for standard earth structures is between 0.1 and 0.2. These COV
thresholds are important for FGE and they are summarized in Table 1. Lognormal probability curves for a more
complete and more systematic range of COVs are shown in Fig. 7.

8
1.E-08

1.E-07
Annual probability of failure

1.E-06

1.E-05

1.E-04

1.E-03
I (COV = 0.072)
1.E-02 II (COV = 0.109)

1.E-01 III (COV = 0.174)

IV (COV = 0.316)
1.E+00
0.5 1.5 2.5 3.5 4.5
Mean factor of safety
Fig. 6 Lognormal probability curves back-calculated from Silva et al. (2008)

1.E-08

1.E-07
Annual probability of failure

1.E-06

1.E-05

1.E-04 Coefficient of
variation (COV)

1.E-03 0.05
0.10
1.E-02 0.20
0.30
1.E-01 0.40
0.50
1.E+00
0.5 1.5 2.5 3.5 4.5
Mean factor of safety
Fig. 7 Lognormal probability curves for higher COVs of FS

9
Table 1. COV thresholds for factors of safety corresponding to different categories of earth structures
(back-calculated from Silva et al., 2008)

Categories of earth structures COV(FS)


I 0.072
II 0.109
III 0.174
IV 0.316

3.2 Parametric uncertainties


There are three primary sources of geotechnical uncertainties as illustrated in Fig. 8: (a) inherent variability, (b)
measurement error, and (c) transformation uncertainty (Phoon and Kulhawy 1999a). A comprehensive effort
was undertaken to provide realistic geomaterial statistics of sufficient generality to underpin current and future
developments of practical reliability-based design codes (Phoon and Kulhawy, 1999a, 1999b; Kulhawy et al.,
2000). The results of this compilation effort are also useful for FGE and are summarized below. Others are
reported by Jones et al. (2002) and Uzielli et al. (2006).

3.2.1 Inherent or natural variability


Inherent variability results primarily from the “real” heterogeneity which is inherent to geomaterials. Such
heterogeneity stems from natural geologic and physical/chemical/biological processes that produced and
continually modify the soil/rock mass in-situ. Uzielli et al. (2006) provided a state-of-the-art review of methods
for quantifying inherent variability in geotechnical engineering analyses and design. Approximate guidelines for
inherent soil variability are given in Table 2. A more detailed presentation containing the mean, standard
deviation (S.D.), and range of the COV are shown in Table 3 with the total number of data groups per test (m).
The statistics for rock are given for comparison.

Fig. 8 Sources of uncertainties contributing to overall uncertainty in design soil parameter.

Table 2. Approximate guidelines for ranges of mean values and COVs of inherent soil variability
(Phoon and Kulhawy 1999a).

Test type Parametera Soil type Mean COV(%)


2
Lab strength su(UC) Clay 10-400 kN/m 20-55

10
2
su(UU) Clay 10-350 kN/m 10-30
2
su(CIUC) Clay 150-700 kN/m 20-40
φ
o
Clay & sand 20-40 5-15
2
CPT qT Clay 0.5-2.5 MN/m < 20
2
qc Clay 0.5-2.0 MN/m 20-40
2
Sand 0.5-30.0 MN/m 20-60
2
VST su(VST) Clay 5-400 kN/m 10-40
SPT N Clay & sand 10-70 blows/ft 25-50
2
DMT A reading Clay 100-450 kN/m 10-35
2
Sand 60-1300 kN/m 20-50
2
B reading Clay 500-880 kN/m 10-35
2
Sand 350-2400 kN/m 20-50
ID Sand 1-8 20-60
KD Sand 2-30 20-60
2
ED Sand 10-50 MN/m 15-65
2
PMT pL Clay 400-2800 kN/m 10-35
2
Sand 1600-3500 kN/m 20-50
2
EPMT Sand 5-15 MN/m 15-65

Lab index wn Clay & silt 13-100 % 8-30


wL Clay & silt 30-90 % 6-30
wP Clay & silt 15-25 % 6-30
PI Clay & silt 10-40 % b
LI Clay & silt 10 % b
γ, γd Clay & silt 13-20 kN/m
3
< 10
c
Dr Sand 30-70 % 10-40
d
50-70
a- su = undrained shear strength; UC = unconfined compression test; UU = unconsolidated-undrained
triaxial compression test; CIUC = consolidated isotropic undrained triaxial compression test; φ =
effective stress friction angle; qT = corrected cone tip resistance; qc = cone tip resistance; VST = vane
shear test; N = standard penetration test blow count; A & B readings, ID, KD, & ED = dilatometer A
& B readings, material index, horizontal stress index, & modulus; pL & EPMT = pressuremeter limit
stress & modulus; wn = natural water content; wL = liquid limit; wP = plastic limit; PI = plasticity
index; LI = liquidity index; γ & γd = total & dry unit weights; Dr = relative density
b- COV = (3-12%) / mean
c- total variability for direct method of determination
d- total variability for indirect determination using SPT values

Table 3. Inherent soil variability for soil and rock (Kulhawy et al. 2000).

Test Parameter Material Coefficient of variation (%)


type type m mean S.D. range
Index γ, γd fine-grained 14 7.8 5.8 2-20
wn fine-grained 40 18.1 7.9 7-46
wP fine-grained 23 15.7 6.0 6-34
wL fine-grained 38 18.1 7.1 7-39
PI - all data fine-grained 33 29.5 10.8 9-57
- ≤ 20% fine-grained 13 35.0 11.4 16-57
- > 20% fine-grained 20 26.0 9.0 9-40
γ, γd rock 42 0.9 0.7 0.1-3
n rock 25 25.9 19.4 3-71
Strength φ, tan φ sand, clay 48 13.9 10.4 4-50

11
sand 32 9.0 3.0 4-15
clay 16 23.5 13.0 10-50
su clay 100 31.5 14.2 6-80
qu rock 184 14.2 11.7 0.3-61
qt-brazilian rock 74 16.6 10.4 2-58
Stiffness Et-50 rock 32 30.7 15.0 7-63
CPT qc sand, clay 65 36.6 15.5 10-81
sand 54 38.2 16.3 10-81
clay 11 28.4 6.8 16-40
qT clay 9 7.9 4.9 2-17
VST su(VST) clay 26 25.3 6.5 13-36
SPT N sand, clay 23 38.0 10.8 19-62
DMT A,B sand, clay 56 27.9 11.9 12-59
sand 30 34.8 11.3 13-59
clay 26 19.9 6.2 12-38
ID - all data sand 30 40.7 21.6 8-130
- w/o outliers 29 37.7 14.2 8-66
KD - all data sand 31 41.2 19.2 15-99
- w/o outliers 29 37.6 13.5 15-67
ED - all data sand 31 42.7 19.6 7-92
- w/o outliers 30 41.1 17.6 7-69

3.2.2 Measurement error

Inherent variability is caused primarily by the natural geologic processes that are involved in soil formation.
Measurement error, on the other hand, arises from equipment, procedural/ operator, and random testing effects.
A summary of total measurement error is given in Table 4 for laboratory tests and in Table 5 for field tests.

Table 4. Summary of total measurement error of some laboratory tests (Phoon & Kulhawy 1999a).

Parametera Soil No. data No. tests/group Property value Property COV (%)
type group (unitsb)
Mean Range Mean Range Mean Range
su(TC) Clay, silt 11 - 13 7-407 125 8-38 19
su(DS) Clay, silt 2 13-17 15 108-130 119 19-20 20
su(LV) Clay 15 - - 4-123 29 5-37 13
φ(TC) Clay, silt 4 9-13 10 2-27 o 19.1o 7-56 24
φ(DS) Clay, silt 5 9-13 11 24-40o 33.3o 3-29 13
o
Sand 2 26 26 30-35 32.7o 13-14 14
tan φ(TC) Sand, silt 6 - - - - 2-22 8
tan φ (DS) Clay 2 - - - - 6-22 14
wn Fine-grained 3 82-88 85 16-21 18 6-12 8
wL Fine-grained 26 41-89 64 17-113 36 3-11 7
wP Fine-grained 26 41-89 62 12-35 21 7-18 10
PI Fine-grained 10 41-89 61 4-44 23 5-51 24
γ Fine-grained 3 82-88 85 16-17 17.0 1-2 1

a - su = undrained shear strength; φ = effective stress friction angle; TC = triaxial compression test;
UC = unconfined compression test; DS = direct shear test; LV = laboratory vane shear test; wn = natural
water content; wL = liquid limit; wP = plastic limit; PI = plasticity index; γ = total unit weight
b - units of su = kN/m2; units of wn, wL, wP, and PI = %; units of γ = kN/m3

12
Table 5. Summary of measurement error of common in-situ tests (Kulhawy & Trautmann 1996).

Test Coefficient of variation, COV (%)


Equipment Procedure Random Totala Rangeb
Standard penetration test (SPT) 5c - 75d 5c - 75d 12 - 15 14c - 100d 15 - 45
Mechanical cone penetration test
5 10e -15f 10e -15f 15e -22f 15 - 25
(MCPT)
e f e f
Electric cone penetration test (ECPT) 3 5 5 -10 7 - 12 5 - 15
Vane shear test (VST) 5 8 10 14 10 - 20
Dilatometer test (DMT) 5 5 8 11 5 - 15
Pressuremeter test, pre-bored (PMT) 5 12 10 16 10 - 20g
Self-boring pressuremeter test
8 15 8 19 15 - 25g
(SBPMT)
2 2 2 0.5
a - COV(Total) = [COV(Equipment) + COV(Procedure) + COV(Random) ]
b - Because of limited data and judgment involved in estimating COVs, ranges represent probable
magnitudes of field test measurement error
c, d - Best to worst case scenarios, respectively, for SPT
e, f - Tip and side resistances, respectively, for CPT
g - It is likely that results may differ for po, pf, and pL, but the data are insufficient to clarify this issue

3.2.3 Transformation uncertainty


The third component of uncertainty is introduced when field or laboratory measurements are transformed into
design parameters using empirical or other correlation models (e.g., correlating the standard penetration test N
value with the undrained shear strength) as shown in Fig. 9. Obviously, the relative contribution of these
components to the overall uncertainty in the design parameter depends on the site conditions, degree of
equipment and procedural control, and quality of the correlation model. Therefore, geomaterial statistics that
are determined from total variability analyses only can be applied to the specific set of circumstances (site
conditions, measurement techniques, correlation models) for which the design parameters were derived. In
other words, the COV of geomaterials cannot be viewed as an intrinsic statistical property.
For each combination of soil type, measurement technique, and correlation model, the uncertainty in the
design soil property is evaluated systematically by combining the appropriate component uncertainties using a
simple second-moment probabilistic approach:

2 2 2
⎛ ∂T ⎞ 2 ⎛ ∂T ⎞ 2 ⎛ ∂T ⎞ 2
s ξ2d ≈ ⎜ ⎟ s w + ⎜ ⎟ se + ⎜ ⎟ sε (6)
⎝ ∂w ⎠ ⎝ ∂e ⎠ ⎝ ∂ε ⎠

in which ξd = T(ξm, ε), T(⋅) = correlation function between test measurement (ξm) and design parameter (ξd), ε =
transformation uncertainty, w = inherent variability, e = measurement error, and s2 = variance.

13
Figure 9. Transformation uncertainty (Phoon & Kulhawy 1999b).

The above equation refers to the variance of the design parameter at a point in the soil mass. For foundation
design, it is not uncommon to evaluate the spatial average of the design parameter over some depth interval,
rather than using the value of the design parameter at a point. The spatial average of ξd is defined as:

1 zb
ξa =
La ∫z t
ξ d (z) dz (7)

in which ξa = spatial average, zt and zb = top and bottom coordinates of a depth interval, respectively, and La =
zb - zt = averaging length.
If t and ∂T/∂w are constants, it can be shown that the variance of the spatial average (sξa2) is given by (Phoon
& Kulhawy 1999b):

2 2 2
⎛ ∂T ⎞ 2 ⎛ ∂T ⎞ 2 ⎛ ∂T ⎞ 2
s ξ2a ≈ ⎜ ⎟ Γ (L a )s w + ⎜ ⎟ s e + ⎜ ⎟ s ε
2
(8)
⎝ ∂w ⎠ ⎝ ∂e ⎠ ⎝ ∂ε ⎠

in which Γ2(⋅) = variance reduction function, which depends on the length of the averaging interval, La. The
following approximate variance reduction function is proposed for practical application (Vanmarcke 1983):

Γ2(La) = 1 for La ≤ δv
Γ2(La) = δv/La for La > δv

in which δv = vertical scale of fluctuation. It can be seen that the variance reduction function decreases as the
length of the averaging interval increases. Therefore, the effect of averaging is to reduce the uncertainty
associated with inherent variability (sw2).
The scale of fluctuation quantifies the spatial extension in which a property of interest can be considered
significantly autocorrelated. Within separation distances smaller than the scale of fluctuation, the deviations
from the trend function are expected to show relatively strong correlation. When the separation distance
between two sample points exceeds the scale of fluctuation, it can be assumed that little correlation exists
between the fluctuations in the measurements. From a geological and geomorphological perspective, it is
intuitive that soil formation and modification processes, as well as factors contributing to the definition of the

14
in-situ state (e.g. stress) would result in a greater heterogeneity of soil properties in the vertical direction and,
hence, in a weaker spatial correlation. Hence, the scale of fluctuation of a given soil property in the vertical
direction is generally much smaller than in the horizontal direction.
The scale of fluctuation is not an inherent property of a soil parameter. Numerical values of the scale of
fluctuation depend at least on: (a) the spatial direction [e.g. horizontal, vertical]; (b) the measurement interval in
the source data; (c) the type of trend which is removed during decomposition; (d) the method of estimation of
the scale of fluctuation from residuals; and (e) modelling options from the specific estimation method. Uzielli et
al. (2006) discussed these issues in detail.
Vanmarcke (1977) proposed the following approximate relationship for evaluating the scale of fluctuation:

2
δ≈ Δ (9)
π
where
nc
1
Δ=
nc
∑Δ
i =1
i (10)

is the average distance between the intersections of the fluctuating component and the trend of a given profile
(see Fig.10).

Figure 10. Simplified estimation of the scale of fluctuation as proposed by Vanmarcke (1983)

Other methods to estimate the scale of fluctuation have been implemented in the geotechnical literature. Uzielli
et al. (2006) provided an overview of such methods. Uzielli (2004) compared the estimates of the scale of
fluctuation of profiles of normalized cone tip resistance using different estimation methods. None of the models
were shown to rank consistently above or below the others, though however, in some cases the scatter among

15
estimates from different methods is significant. Ranges of v for a number of geotechnical parameters are given
in Table 6.

Table 6. Literature values for the horizontal ( h) and vertical ( v) scale of fluctuation of geotechnical
parameters (from Uzielli et al. 2006)
Property* Soil type Testing method** h (m) v (m)
su clay lab. testing - 0.8-8.6
su clay VST 46.0-60.0 2.0-6.2
qc sand, clay CPT 3.0-80.0 0.1-3.0
qc offshore soils CPT 14-38 0.3-0.4
1/qc alluvial deposits CPT - 0.1-2.6
qt clay CPTU 23.0-66.0 0.2-0.5
qc1N cohesive-behaviour soils CPT - 0.1-0.6
qc1N intermediate-behaviour soils CPT - 0.3-1.0
qc1N cohesionless-behaviour soils CPT - 0.4-1.1
fs sand CPT - 1.3
fs deltaic soils CPT - 0.3-0.4
FR cohesive-behaviour soils CPT - 0.1-0.5
FR intermediate-behaviour soils CPT - 0.1-0.6
FR cohesionless-behaviour soils CPT - 0.2-0.6
Ic cohesive-behaviour soils CPT - 0.2-0.5
Ic intermediate-behaviour soils CPT - 0.6
Ic cohesionless-behaviour soils CPT - 0.3-1.2
N sand SPT - 2.4
w clay, loam lab. testing 170.0 1.6-12.7
wL clay, loam lab. testing - 1.6-8.7
´ clay lab. testing - 1.6
clay, loam lab. testing - 2.4-7.9
e organic silty clay lab. testing - 3.0
´p organic silty clay lab. testing 180.0 0.6
KS dry sand fill PLT 0.3 -
ln(DR) sand SPT 67.0 3.7
n sand - 3.3 6.5

* su=undrained shear strength; qc=cone tip resistance; qt=corrected cone tip resistance;
qc1N=dimensionless, stress-normalised cone tip resistance; fs=sleeve friction; FR=stress-
normalised friction ratio; Ic=CPT soil behaviour classification index; N=SPT blow count;
w=water content; wL=liquid limit; ´=submerged unit weight; =unit weight; e=void ratio;
´p=preconsolidation pressure; KS=subgrade modulus; DR=relative density; n=porosity
** VST=vane shear testing; CPT=cone penetration testing; CPTU=piezocone testing;
SPT=standard penetration testing; PLT=plate load testing

Useful guidelines on typical “total” coefficients of variation of many common design soil strength properties
have been summarized by Phoon and Kulhawy (1999b) and are given in Table 7 for reference. Once again,
tabulated values should not be applied uncritically, i.e. if it has not been assessed that the variability related to
site conditions, testing methods and transformation models are significantly similar to those used to obtain
specific literature values.

Table 7. Approximate guidelines for design soil parameter variability (Phoon & Kulhawy (1999b).

Design b Point Spatial avg. Correlation


a Test Soil type c
parameter COV (%) COV (%) Equationf
su(UC) Direct (lab) Clay 20-55 10-40 -
su(UU) Direct (lab) Clay 10-35 7-25 -

16
su(CIUC) Direct (lab) Clay 20-45 10-30 -
su(field) VST Clay 15-50 15-50 14
e e
su(UU) qT Clay 30-40 30-35 18
e e
su(CIUC) qT Clay 35-50 35-40 18
su(UU) N Clay 40-60 40-55 23
d
su KD Clay 30-55 30-55 29
e
su(field) PI Clay 30-55 - 32
φ Direct (lab) Clay, sand 7-20 6-20 -
φ (TC)
e e
qT Sand 10-15 10 38
φcv PI Clay 15-20
e
15-20
e
43
Ko Direct (SBPMT) Clay 20-45 15-45 -
Ko Direct (SBPMT) Sand 25-55 20-55 -
e e
Ko KD Clay 35-50 35-50 49
e
Ko N Clay 40-75 - 54
EPMT Direct (PMT) Sand 20-70 15-70 -
ED Direct (DMT) Sand 15-70 10-70 -
EPMT N Clay 85-95 85-95 61
ED N Silt 40-60 35-55 64
a - su = undrained shear strength; UU = unconsolidated-undrained triaxial compression test;
UC = unconfined compression test; CIUC = consolidated isotropic undrained triaxial
compression test; su(field) = corrected su from vane shear test; φ = effective stress friction angle;
TC = triaxial compression; φcv = constant volume φ; Ko = in-situ horizontal stress coefficient;
EPMT = pressuremeter modulus; ED = dilatometer modulus
b - VST = vane shear test; qT = corrected cone tip resistance; N = standard penetration test blow
count; KD = dilatometer horizontal stress index; PI = plasticity index
c - averaging over 5 meters
d - mixture of su from UU, UC, and VST
e - COV is a function of the mean; refer to COV equations in Phoon & Kulhawy (1999b) for details
f - Quality of correlation affects the COV of design parameters. The correlation models used are
referenced by the equation numbers in Phoon & Kulhawy (1999b)

3.3 Model uncertainties


3.3.1. Ultimate limit state
Robust model statistics can only be evaluated using: (1) realistically large scale prototype tests, (2) a sufficiently
large and representative database, and (3) reasonably high quality testing where extraneous uncertainties are
well controlled. It is common to correct for simplifications in the calculation model using the following
multiplicative form (as exemplified by laterally loaded drilled shafts):

Hm = M·Hu (11)

in which Hm = “measured” lateral capacity (more precisely, capacity interpreted from load test), Hu = ultimate
lateral capacity computed using limit equilibrium analysis, and M = model factor, typically assumed to be an
independent log-normal random variable. It is well known that many different models exist for the computation
of Hu.
The distributions of the model factors for Hm determined using 2 different criteria (HL or Hh) and Hu
computed from 4 different lateral soil stress models are shown in Fig. 11. Note that M < 1 implies that the
calculated capacity is larger than the measured capacity, which is unconservative. If Hm is defined as the
hyperbolic capacity (Hh), M < 1 is most likely unsafe as well since there is no reserved capacity beyond Hh and
it is mobilized at very large displacements.

Capacity model Lateral or moment limit (HL) Hyperbolic capacity (Hh)


(soil type)

17
Reese (1958) (clay) 0.4 0.4

Relative Frequency
Relative Frequency
Mean = 0.92 Mean = 1.42
S.D. = 0.27 S.D. = 0.41
0.3 COV = 0.29 0.3 COV = 0.29
n= 72 n= 74
0.2 pAD = 0.633 0.2 p AD = 0.186

0.1 0.1

0 0
0.0 0.8 1.6 2.4 3.2 0.4 1.2 2.0 2.8 3.6
HL/Hu(Reese) H h/Hu(Reese)
Broms (1964a) (clay) 0.4 0.4
Mean = 1.49 Mean = 2.28
Relative Frequency

Relative Frequency
S.D. = 0.57 S.D. = 0.85
0.3 COV = 0.38 0.3 COV = 0.37
n= 72 n= 74
pAD = 0.122 pAD = 0.149
0.2 0.2

0.1 0.1

0 0
0.0 0.8 1.6 2.4 3.2 0.4 1.2 2.0 2.8 3.6
HL/Hu(Broms) Hh/Hu(Broms)
Randolph & Houlsby 0.4 0.4
Relative Frequency

(1984) (clay) Mean = 0.85 Relative Frequency Mean = 1.32


S.D. = 0.24 S.D. = 0.38
0.3 COV = 0.28 0.3 COV = 0.29
n= 72 n= 74
0.2 pAD = 0.555 0.2 pAD = 0.270

0.1 0.1

0 0
0.0 0.8 1.6 2.4 3.2 0.4 1.2 2.0 2.8 3.6
H L/H u(Randolph & Houlsby) H h/H u(Randolph & Houlsby)
Broms (1964b) (sand) 0.4 0.4
Relative Frequency

Relative Frequency

Mean = 0.88 Mean = 1.30


0.3 S.D. = 0.36 0.3 S.D. = 0.49
COV = 0.41 COV = 0.38
0.2 n= 75 0.2 n= 77
pAD = 0.736 pAD = 0.141

0.1 0.1

0 0
0.0 0.8 1.6 2.4 3.2 0.4 1.2 2.0 2.8 3.6
H L/H u(simplified Broms) Hh/Hu(simplified Broms)

Figure 11. Distribution of model factors (Phoon & Kulhawy 2005).

3.3.2. Serviceability limit state


Phoon et al. (2006) reported a probabilistic characterization of load-displacement curves using an augered cast-
in-place (ACIP) pile load test database. The normalized hyperbolic curve considered in their study is expressed
as:

18
Q y
= (12)
Q STC a + by

in which Q = applied load, QSTC = failure load interpreted using the slope tangent method, “a” and “b” = curve-
fitting parameters, and y = pile butt displacement. Note that the curve-fitting parameters are physically
meaningful – the reciprocals of “a” and “b” are equal to the initial slope and asymptotic value, respectively.

Each continuous load-displacement curve can be reduced to two curve-fitting parameters (plotted as a single
point in Fig. 12). The scatter in the load-displacement curves is captured by the scatter between “a” and “b”. If
the values of “a” are plotted as a histogram, a non-uniform distribution will be obtained (Fig. 13). For example,
values close to 5 mm are more frequently encountered. The standard approach is to fit one of the many classical
distributions (e.g. lognormal distribution) to the histogram. The crucial point to be emphasized here is that such
an approach implicitly assumes that “a” and “b” are statistically independent random variables. However, Fig.
12 clearly shows that the variation of “a” (variation along y-axis) and the variation of “b” (variation along x-
axis) are coupled. In other words, it is incorrect to assume that “a” can vary independently of “b”. The correct
probabilistic model in this case is not independent random variables but a bivariate random vector.
Computational details on the construction of this bivariate random vector are given elsewhere (Phoon and
Kulhawy, 2008).

Augered cast-in-place pile Spread foundation (uplift)


20 (compression) 20
clay
a parameter (mm)
a parameter (mm)

16 16 sand
12 12

8 8

4 4

0 0
0.2 0.4 0.6 0.8 1.0 1.2 0.2 0.4 0.6 0.8 1.0 1.2
b parameter b parameter

No. of tests = 40 No. of tests = 85


a Mean = 5.15 mm a Mean = 7.13 mm
Standard deviation = 3.07 mm Standard deviation = 4.66 mm
Coefficient of variation = 0.60 Coefficient of variation = 0.65
b Mean = 0.62 b Mean = 0.75
Standard deviation = 0.16 Standard deviation = 0.14
Coefficient of variation = 0.26 Coefficient of variation = 0.18

Correlation = -0.67 Correlation = -0.24

19
Drilled shaft (uplift) Pressure-injected footing
5 5 (uplift)
clay

a parameter (mm)
a parameter (mm)

4 sand 4

3 3

2 2

1 1

0 0
0.4 0.6 0.8 1.0 1.2 1.4 0.4 0.6 0.8 1.0 1.2 1.4
b parameter b parameter

No. of tests = 48 No. of tests = 25


a Mean = 1.34 mm a Mean = 1.38 mm
Standard deviation = 0.73 mm Standard deviation = 0.95 mm
Coefficient of variation = 0.54 Coefficient of variation = 0.68
b Mean = 0.89 b Mean = 0.77
Standard deviation = 0.063 Standard deviation = 0.21
Coefficient of variation = 0.07 Coefficient of variation = 0.27

Correlation = -0.59 Correlation = -0.73

Figure 12. Correlation between hyperbolic parameters (Phoon et al. 2006; 2007)

0.3 Mean = 5.15 mm


0.3 Mean = 0.62
Relative Frequency
Relative Frequency

S.D. = 3.07 mm S.D. = 0.16


COV = 0.60 COV = 0.26
0.2 n= 40 0.2 n= 40
pAD = 0.183 pAD = 0.887

0.1 0.1

0 0
0 5 10 15 20 0.0 0.5 1.0 1.5 2.0
a parameter (mm) b parameter
Figure 13. Marginal distributions of hyperbolic parameters (Phoon et al. 2006).

4. OTHER POTENTIAL ROLES OF RISK-BASED ANALYSIS


One of the important requirements in forensic geotechnical engineering is the identification of failure
mechanisms. A possible role for risk-based analysis can be discerned from the simulation studies detailed
below.

Popescu et. al (1997) studied the effects of spatial variability of soil properties on soil liquefaction for a
saturated soil deposit subjected to seismic excitation. They compared standard results derived from
deterministic inputs and probabilistic results derived from stochastic finite element analyses. They concluded
that both the pattern and the amount of dynamically induced pore water pressure buildup are strongly influenced
by the spatial variability of soil parameters. For the same average values of soil parameters, more pore water

20
pressure build up is predicted by the stochastic model than by the deterministic model, which is attributed to a
water injection phenomenon triggered by the presence of loose pockets in the spatially variable soil deposit.

Griffiths et. al (2002) demonstrated that failure mechanisms of a strip footing are considerably different in the
case of uniform soils and spatially variable soils. The spatial correlation structure of the foundation soil in terms
of autocorrelation structure has an important effect on the failure mechanism. More extensive studies are
presented by Kim (2005).

Goldsworthy (2007) presented a fairly comprehensive risk-based study of the effect of site investigations on
foundation failures. One conclusion is that an optimum site investigation minimizing the risk of failure can be
planned based on the variability and autocorrelation structure of the foundation soil.

The above studies and other similar efforts demonstrate that failures can be studied in a more realistic way using
spatially variable soils, rather than the traditional uniform or layered soil profiles. Notwithstanding this, a
significant number of failures does arise from human failures and a lack of proper quality assurance and control
plans. Bea (2006) suggested that the failure development process can be categorized into three phases viz,
initiating, contributing and propagating causes and a proper risk management strategy is necessary using
observational methods.

A risk-based approach to forensic geotechnical engineering may include elements such as:

i. Detailed soil investigations in the area in the form of vertical and horizontal soil profiles to formulate
plausible hypotheses concerning prevalent/relevant failure mechanisms. The number and spatial location of
boreholes and spacing should be adequate to provide a proper estimate of the mean, variance and
autocorrelation properties of in-situ soils.

ii. Analysis of the actual failure mechanism in relation to results from probabilistic studies.

iii. Reanalysis of different loading and the associated variability.

iv. Re-examination of quality assurance plans which include specifications on quality control of materials and
construction, construction sequences proposed and adopted, observations of performance depending on the
nature of project such as deformations, pore pressures etc. which are expected in important projects; and

v. Reliability-based computational back-analysis.

5. CONCLUDING REMARKS
This chapter seeks to explore if reliability/risk concepts are potentially useful to forensic geotechnical
engineering. This study is very preliminary given the dearth of previous literature. One key aspect that
distinguishes geotechnical engineering from structural engineering is the natural variability of geo-materials.
Within this context, Leonards (1982) definition of “failure” – “unacceptable difference between expected and
observed performance” – cannot be evaluated in a meaningful way using deterministic methods. In broad terms,
forensic engineering is related to the investigation of failures with the view of rendering an opinion regarding
responsibility. Hence, a statistical measure of “unacceptable difference” (specifically, a difference not
explainable by underlying variability) should provide useful additional information in the formulation of such an
opinion. A preliminary statistical framework is presented to quantify the difference between expected and
observed performance in the presence of unavoidable and potentially significant geotechnical variabilities. It
should be emphasized that the quality of the statistical analysis can only be as precise, accurate and meaningful
as the engineer’s characterization of uncertainties. Other potentially useful results in the recent reliability and
risk literature are highlighted. The intention of this chapter is to stimulate further discussions and research in
this important but somewhat overlooked area.

REFERENCES
Bea, R. (2006). Reliability and human factors in geotechnical engineering. Journal of Geotechnical and
Geoenvironmental Engineering, ASCE, 132(5): 631-643.

21
Brown E. T. (2006). Forensic engineering for underground construction. Proceedings of the ISRM International
Symposium 2006 and the 4th Asian Rock Mechanics Symposium, Singapore 8 - 10 November 2006, Chapter
1, 1-16.

Chilès, J-P. and Delfiner, P. (1999). Geostatistics – Modeling Spatial Uncertainty. John Wiley and Sons, New
York.

Duncan, J.M. (2000). Factors of safety and reliability in geotechnical engineering, Journal of Geotechnical and
Geoenvironmental Engineering, ASCE, Vol.126, No.4, pp. 307-316.

Goldsworthy, J. S. (2007). Quantifying the risk of geotechnical site investigations. PhD Thesis, University of
Adelaide.

Griffiths, D V, Fenton, G. A and Manoharan, N. (2002). Bearing capacity of rough rigid strip footing on
cohesive soil: probabilistic study. Journal of Geotechnical and Geoenvironmental Engineering, Vol. 128(9),
73-755.

Jones, A.L., Kramer, S.L. & Arduino, P. (2002). Estimation of uncertainty in geotechnical properties for
performance-based earthquake engineering. PEER Report 2002/16, Pacific Earthquake Engineering
Research Center, University of California, Berkeley.

Kim, H. (2005). Spatial variability in soils: stiffness and strength, PhD Thesis, Georgia Institute of Technology,
2005.

Kulhawy, F.H. & Trautmann, C.H. (1996). Estimation of in-situ test uncertainty. Uncertainty in the Geologic
Environment - From Theory to Practice (GSP 58), ASCE, New York, 269 – 286.

Kulhawy, F.H., Phoon, K.K. & Prakoso, W.A. (2000). Uncertainty in the basic properties of natural
geomaterials. Proc. 1st International Conference on Geotechnical Engineering Education and Training,
Sinaia, Romania, 297-302.

Leonards, G. A. (1982). Investigation of failures. Journal of the Geotechnical Engineering Division, ASCE,
108(GT2): 187-246.

Paikowsky, S.G. (2002). Load and resistance factor design (LRFD) for deep foundations. Proc. International
Workshop on Foundation Design Codes and Soil Investigation in view of International Harmonization and
Performance Based Design, Tokyo, Japan, 59 – 94. Balkema, Netherlands.

Phoon, K.K, Kulhawy, F. H & Grigoriu, M. D. (1995). RBD of Foundations for Transmission Line Structures.
Report TR-105000, Electric Power Research Institute (EPRI), Palo Alto.

Phoon, K. K. & Kulhawy, F. H. (1999a). Characterization of geotechnical variability. Canadian Geotechnical


Journal, 36(4):612-624.

Phoon, K. K. & Kulhawy, F. H. (1999b). Evaluation of geotechnical property variability. Canadian


Geotechnical Journal, 36(4):625-639.

Phoon K.K. and Kulhawy, F.H. (2005). Characterization of model uncertainties for laterally loaded rigid drilled
shafts. Geotechnique, 55(1), 45-54.

Phoon, K. K. & Kulhawy, F. H. (2008), Serviceability limit state reliability-based design, Chapter 9, Reliability-
Based Design in Geotechnical Engineering: Computations and Applications, Taylor & Francis, April 2008,
344-383.

Phoon, K.K., Chen, J.-R. & Kulhawy, F.H. (2006). Characterization of model uncertainties for augered cast-in-
place (ACIP) piles under axial compression. Foundation Analysis & Design: Innovative Methods (GSP 153),
ASCE, Reston, 82-89

22
Phoon, K. K., Chen, J. R. & Kulhawy, F. H. (2007), Probabilistic hyperbolic models for foundation uplift
movement, Probabilistic Applications in Geotechnical Engineering (GSP 170), ASCE, Reston, CDROM.

Popescu, R. Prevost, J. H and Deodatis, G. (1997). Effects of spatial variability on soil liquefaction: some design
recommendations. Geotechnique, 47(5):1019-1036.

Rétháti, L. (1988). Probabilistic solutions in geotechnics. Elsevier, New York.

Silva, F., Lambe, T. W. and Marr, W. A. (2008), Probability and risk of slope failure. Journal of Geotechnical
and Geoenvironmental Engineering, ASCE, 134(12): 1691-1699.

Sowers, G. F. (1993). Human factors in civil and geotechnical engineering failures. Journal of Geotechnical
Engineering, ASCE, 119(2): 238-256.

Uzielli, M. (2004). Variability of stress-normalized CPT measurements and application to seismic liquefaction
initiation assessment, University of Florence (Italy). Thesis downloadable at: http://www.georisk.eu.

Uzielli, M., Lacasse, S., Nadim, F. & Phoon, K.K. (2006). Soil variability analysis for geotechnical practice.
Proc., Second International Workshop on Characterisation and Engineering Properties of Natural Soils,
Singapore. Balkema, Netherlands, 1653 – 1752.

Vanmarcke, E.H. (1983). Random Field: Analysis and Synthesis, MIT Press, Cambridge, Massachusetts.

23
-1-
Case Studies 1
TURNING HINDSIGHT INTO FORESIGHT

(LEADERSHIP LESSONS FROM A LANDSLIDE)

Mike Marley
BE, MEngSC, FIEAust, RPEQ, ACIS
Principal – Golder Associates Pty Ltd
611 Coronation Drive, Toowong, Qld
mmarley@golder.com.au

INTRODUCTION

Tom Peters (one of the great business thinkers of our time) proclaims in his “Leadership Essentials” that “great leaders are
great story tellers”. He states” “We need stories, Riveting Tales that fire the imagination of ---- as-yet-reluctant heroes-in-
waiting”.

This is the story of a forensic engineering study to determine the cause of a major landslide in P.N.G., of the combination
of good luck and good management that led to a successful outcome, and of the lessons the author took from the study
which have informed his views on project management during the years since the study.

Drawing on Peters’ “Leadership Essentials” (the ability to inspire, liberate, achieve) some lessons in leadership are
highlighted, and the importance of populating the leadership literature with similar stories of real projects, as a means of
recruiting and nurturing the next generation of inspirational leaders, is emphasised.

THE STORY

Background

The Ok Tedi gold and copper mine was developed in the remote Star Mountains in the Western Province of P.N.G., close
to the Irian Jaya border (Figure 1). A tailings storage dam was included in the mine development plan, and construction of
this dam was on the critical path for the mine construction programme to ensure that key milestones could be met for the
project’s financial model.

OK
OO KTK
OT
TKOEK O
DIKOK OK
I OK OK TTE
T
T E
ED
ED
DD
DIIII
M
MN
M
MI NE
IN ES
IE SSIIT TE
IT
IM E
IMI
E MIMIMIN
NEESS T
IIM M IM N E SIIIT
S TE
EE
ii
Ta
TT
abau
T Ta
Tb
a a
T
Ta llll T
iiiT
a aMaoT
T auanb
b
bu
b
b
b tu
u
ub
u b
b
b
Hb
biia
iiiilllllgen
TT
Ta a abb
buu
ub bM
b M
MMo
M o
o
o
oou
o un
ununtu
n H
tt n
H
u u
n aa g
ng
un
u n
een
uu
nntttttt H
n Ha
HHaHa
H
Haage
aa ge
ge gge
egg
eggen
ee n
n
n
n
n
T
TA
T
TA
AB
ABU
U
BUUB
BB
U BU BU
BUBU IBIBss
I ee
BrIBiiiIB
bi lll IBIL
T
T AB BBU
IIIIL
LBU BUBIBW
W e
WeW
W
We eBI
set rW
n
rn
nest
L
L oun est
estM e unuu
ost
Mo s
nn
La en
tee
urrru
n
n
La a L
ntttt L L
n aLaL
aL aL
aL a ae
e
W We tteW eM
Mst
o eu
o
o stn un
sutee
nrrrn
n
n ttt
P ro inrP
vvP cree PP H
Ha
H
oH Ha
viia H a
agege gge
egg egee gen
n
n
PPrP
Pr orP
rro v incc
riin PrePr r ro
o
o
o v nH
iin
n
n
n
n ca
c
c
c
ca
e
ee ge
e n
n
n
idi i di NNN
ddd

T
iA
T AABUrru
B U
BmBU
U BU IBIBIBIBIBIBIBIL
NBiNU
B U B U
B UBI B L
Luu
ii

i di didiiiiiNi
edd
nNi
gNe u
rriN m
uiNi
m iNiNin
n
n
ngg
g
gee
errrrru
u
u
umm
m
eee

n g m
m Da
DD rDu
aa a
DaDa
DaDa Darrru
Da uurtM orreessb y
Po rrttM
Mo orresb byy
T
T
OkkkT

TT
eTee W
We
W WeWW e estestestestestestsPo
tte
sP ee
ePrrrrrro
n
Pn
no
n Poorrrrtrtrtrtttt M
o M
M
M
Moo
o
oorrrrrresb esb
esb e
e esb
esb s
esb
s
IIIIIG BIBIBIBIB LIB IBL G
aEE
E bay
y
y
O

B GE IB
La aL
L a IBL
a aIG
G
L sb
a
L b
L
a e
y
y
ye
O

Ok
OOkOkO
kOkO
kO
kO
kk PAPU AN
NNE
EW
EWG
W GU
GUIIN
U NE
IIN E
EA
AA
PrPrPrPrPPr
r Pr
KI
KIIK IKI
IKN KIKIKIKIUN
UNUN UN G A
A
IU G
KIA UN
r Pr ooovviiinnncceee PU
UNro
o
oN
UvN
v nUN
iiin
n GA
ccee G A
A
e
iv e
0 50
NiN0iNiNiNiNiNiNiNin
100
ididididididDa
idDa Da Da R nn
n
ng
n ge
g
geerrrrrru
u
um
u
umm
m
m
m
e L Da Da Da Darrrrrru
Da u
u
Laaakkkeee u
u killometres
LL ake ake M
akeuarke
raayke ake ake ake ake M Mu urrrra yay ay y P Pdo
akilometres
y PoPo Po orrtrtrtrtt M
M
Moo
orr
esbesb esbeesb esb esbsby y
y
TTeTee M an d a n
MMaan nd daa l a l
kIBIBIBIBIBIBIG
M a M an
ckIB
BIB G
GE E
E
a

Ma Ma M a nnd ddaa
kMA a
kAnn
ndd
da
kaaA
a

i
ua

Ok OO kO kO O
kIIIO O
k O kK tr
u

M B
ppu

SAIAIAIAIAIA
ap

t AI AIIAM AM AMAM AMAMAM


A M
PB
B
B AK
B B AK B
PAK
A
APU K
KK
a

S PA PA PA APA
PA PAPU U PUPU
Pa

PA PA PU P
puapua pua pua pua puaW est er n P Prrro
P oovvviin innccceee
P

W est er n
P

a PUPUAA
A
AN
A N
N
NE
N
N E
E
EW
EWWG
W G
GU
GUU
U
UIIIIIN
N
N
NEE
EA
EAA
A
A
Papua
Pr
P PrPrPapua
rProoPv
vapua
iin
nW
Wce
c ee ststsW
Wee
terW nePr
rn ste
Pst sro
rPW ts
osts
a
pte
aa
v
v
ap
er
pie
a
ien
n
aeree
ac
c rn
e n
KIKIKIKIKIKIKIKIKIUN
UNUN
UN UN
UN UN
UNUNG
GA
G A
A
A
A
r
S
S t urrtt Is
Stttu
urrtt IIs
lan d
nd
lan
Isllla e
d Wa WW a Waa Wa WW a Wa W a as s
ssa ap
a
a pp
p
pee
e
eaa
e a
iv W 0a ssu a 50 100
inea

W aa suuaa
R F lyW
nesiiaa

uii

StStSS S t StStu uurrrrrrtttttt IIIs


Is
sllllla
a
a
an
n
nnd
d
dd R
neessiia

St t t St u
uu s a
anndd ivey w a i Is la n d
LLake rrat ike ake MM rararaa ay y Kiw
Gu

a ke
'Aa ke saIs
ske
Isllla ke ake akee wu
audrd a y
akilometres
y iwd aaiiiiIIs lan
IIslla nd
Gu

A b e laan d d
aa d
wG

a
a D'''A llb
lb ee rttiiisi la nn dd r
L e Wa Wa WW W
W
a asu
W asu a
asu asu su asu su a
s u aa
a
e
on

TAPIIILLA
New

ia iinn ILA
a Neew

a A
do

n
do
nd

u
l a KiKiKiKiKiKiKiw aiii
uaaN

l Le
S
S eLwL
eeeww
wa
r im ad d
daaa
abu w
wa a
Seew w errriimimaab bu
In

IIs
s
sllla
an
annd
d
d u
IIn

Ik
M kIs KiK iw
wa ai i Is l
laan
nd d
MaMD'a
D M a
D'D 'Da
n
n
nd M
d
da
'AAa
AM
a
a lllb
bbae
n
n
nrrd
e rd
d
tttaa
is
iis
is
isc IiIs sllla a
an
nnd
d
d i
Suiii
pu
apu
Paap

i tdr D
DDAAARRRU U Paraam
U am aa
ma
t ss G u D
D'D' A A
Alllbb
beerrrtttisiis
s IsIIsllla a
annnd d
SAIAIAIAIAIALe LLe L
eLeAMe
Lew ww
Aa a
ad
d
dMa
aa
P

AI AM BB AK AK K
P

IIAM AM AM M
A M B
B B
IIs AK
laB n d
nAK dA K
S G IIsslllaan d
pua pua puapuapua pua G
a ne ne G nn
eee
P
Pr nnee
rP
Papua
PLALA
L
ne
ne
rP rPrno
Aapua
do do do
oPv
vw
winW
apua
e inc
W ee ststsW
ce Wee
terWrneTA
Tst
nTAPre
AA
TTTA
A
T
P Ps
st
A
TA
rPIPPt
Ps
Iro
o tsv
IIILALtA
L
v iLer
L
A
in
neA
c
cree
rn
e n
dd
ooN dodd
do oo e
In
I aa
N Sew
Sew Se
SewSSeS
eeSe w
Sewweerrriim
e maab
m b
bu u
u
Wa
W Wa SSW S WSWa
SSs s
saa
ap
p
ip
iie
e
eaa
a
-2- r u ss lan d
u IIIslla
Darrru nd
lan d

Figure 1
A fast track approach was adopted, requiring investigation, and design to proceed concurrently with construction. This
approach increases the overall risk associated with the work, and requires the design programme to be sufficiently flexible
to accommodate design changes where investigations identify conditions significantly different from assumptions adopted
in the preliminary design.

The site is located in one of the most remote and challenging environments in the world: the terrain is rugged; the rainfall
extreme (up to 11m per annum); and the jungle is almost impenetrable.

Lesson 1: Understand and respect the environment in which you are operating

History

Preparation for construction of the tailings dam commenced in May 1983.

Excavation for the dam footprint commenced in early November 1983.

In mid December 1983, (when approximately 250,000m³ of excavation had been completed) a landslide occurred in the
eastern side of the valley. Approximately 3.4 million m³ was involved in the slide mass. This was followed in early
January 1984 by a much larger slide (approximately 35 million m³) covering an area approximately 800m wide and 1.2km
long) (Figure 2).

Extent of December 1983 Sl

Extent of January 1984 Slid

Figure 2

An investigation of the slides was initiated (specifically to asses whether the footprint excavation was the primary cause,
and whether there were grounds for an insurance claim under the contractors insurance).

Initial Assessment

Initial assessment revealed:

– significant variability in the magnitude of translational movement within the slide mass:

– ratio of excavation volume to slide mass volume for first slide of approximately 1:15;

– ratio of excavation volume to slide mass volume for second slide of approximately 1:154;
-3-
– the disposition of the slide mass with respect to excavation was particularly asymmetrical;

– slide had apparently occurred on a very low angle failure plane (Figure 3).

Figure 3
There was considerable doubt as to whether the excavation was the sole (or even a significant) contributor to the slide.
Other potential causes canvassed at the time include earthquake and/or an intense rainfall event.

Clearly assessment of the cause would not be straight forward.

Assembly of a team with a wide range of geo-scientific expertise was the essential first step.

Lesson 2: Complex tasks require high performance teams

The Investigation

Geologically, the Star Mountains are part of the deformation zone associated with the active boundary between the Pacific
an Indo-Australian tectonic plates (Figure 4). At this boundary the sub-oceanic crust of the Pacific Plate plunges beneath
the more stable continental crust of the Indo-Australian plate causing uplift (Figure 5). There is considerable seismic and
volcanic activity associated with this plate sub-duction and the associated rapid uplift has resulted in thick layers of
limestone and mudstone, formed from marine deposits, being exposed in the rugged heights of the Star Mountains. A
simplified cross-section through the PNG shows the impact of this tectonic activity (Figure 5), including the high central
mountain spine and zones of low amplitude folding to the north and south of the spine. This reminder of the mechanics of
continental creation proved to be the first vital clue in understanding the cause of the slide.

Figure 4
-4-

Figure 5
Initial helicopter reconnaissance surveys to assess the magnitude and extent of the damage caused by the slide also
highlighted an unusual feature of the Ok Ma valley (viz. that the valley was broadly leaf-shaped with a narrow, meandering
entrance gorge upstream of the dam site, and another similar exit gorge downstream of the site). This feature was repeated
regularly throughout the surrounding area, and this repetition of topographic form proved to be the second important clue in
understanding the slide.

Detailed study of the aerial photography of the site and its surrounds revealed a number of regional characteristics, in
particular several sets of linear features crossing the folded topography surrounding the site (Figure 6). These lineaments
(identified under the stereoscope as distinct partings in the tree cover) are indicators of geological structures such as faults,
fractures and shears in the rocks underlying the area.

Figure 6
The orientation of the predominant sets of defects was noted.

Lesson 3: Take time to look at the big picture – get above the trees so you can see the form of the forest.

Review of the considerable volume of existing investigation data and records showed:
-5-
– no significant earthquake or rainfall events that were likely to have initiated the slides,

– the stratigraphic sequence at the site (Figure 7) with the upper levels of the sequence removed by erosion exposing the
Warre limestone as a relatively thin cap overlying approximately 1000m of dark fine grained mudstone (Pnyang
formation) in which the slide occurred.

– failure could not be readily explained based on measured strength in mudstone.

Figure 7
Field investigations concentrated on detailed geological mapping of the site, and a drilling and instrumentation installation
programme. A number of inclinometers and piezometers were installed throughout the slide mass to record any on-going
movement of the slide mass, as well as details of fluctuations in groundwater levels which have a major influence on slope
stability.

Airphoto interpretation and field mapping yielded valuable understanding of the regional geomorphology. The leaf-shaped
valleys were found to be the results of transverse erosion through the anticlinal folds formed by the compression and uplift
caused by regional tectonic forces.

A mechanism was deduced for the valley formation processes as shown in Figure 8:

– Streams originating in overlying formations (since removed)erode down through the upper moderately resistant
limestone cap.

– As uplift and erosion continue, the streams enter the underlying less resistant mudstone.

– Progressive erosion of the mudstone and removal of the toe of the valley slope leads to undercutting and breaking up
of the limestone cap.

– Evolution of the slope by repetitive slides in the mudstone creates weak zones approximately parallel to the bedding
with very much lower strength than the surrounding materials.
-6-

Figure 8

The establishment of the mechanism of formation of the characteristic leaf shaped valley (Figure 9) was the next critical
factor in understanding the cause of the slide.
-7-

Figure 9
With a widely scattered team, daily debriefings were held to enable the whole team to stay in touch with progress of the
various facets of investigation work, each proceeding at its own characteristic pace.

Lesson 4: Good communication is critical to the effective functioning of high performance teams.

Geological mapping of the immediate area surrounding the slide revealed a closely spaced grid of near vertical joints in the
mudstone (Figure 10) and the orientations of these defects were found to align closely with the large-scale lineations
observed on the aerial photography.

Figure 10
The orientations of these lineations at both macro and micro scale aligned closely with the theoretically calculated
directions of failure planes which would be induced in the mudstone by the tectonic compression forces causing the uplift
and folding at the site. Laboratory scale triaxial compression tests on mudstone specimens confirmed this.

All of the major defect sets identified at the site were now able to be explained. Figure 11 shows the defects:

– bedding planes originally horizontal, but now distorted by folding,

– axial defects parallel to the fold axis (major tensile zone),

– conjugate joint sets induced by tectonic compression.


-8-

Figure 11
Figure 12 shows a diagrammetric representation of a typical wedge-shaped block of mudstone which would be produced as
a result of the intersection of the various defects sets at the site.

Figure 12
Instrumentation records showed that movement was continuing in the slide mass and that the movement appeared to be
occurring on multiple planes stepping up in elevation progressively in the downstream direction.

Measured values of shear strength on samples recovered from the vicinity of the shear planes identified in the inclinometers
confirmed that actual strengths on these planes (formed initially in the valley forming process) was about one third of the
average strength of intact mudstone. These measurements confirmed values of shear strength estimated from back analysis
of the stability of the slope to be necessary for slides to be mobilised on planes at the locations indicated by inclinometers.
They also confirmed that use of the average values of mudstone strength (as had apparently been done by the original
designers) would seriously under-estimate the likelihood of instability.

Lesson 6: Averages are statistical constructs. Understand the numbers that went into their calculation.

Piecing all of the data together allowed a model for the slide mechanism to be developed involving uniform planar
movement of a relatively few large wedge-shaped blocks of mudstone. Figure 13 shows a schematic representation of the
site and the three dimensional spatial relationships between the slide blocks and the terrain surface. Initial failure of a
relatively small block at the upstream end following removal of support at the toe of the slope by the footprint excavation
in turn removed support from a succession of further blocks (with different failure plane elevations) resulting in progressive
enlargement of the failure downstream.
-9-

Figure 13
This model actually recreated the form of the landslide remarkably closely.

Lesson 7: Sketch what you think is happening. It allows a better understanding of complex relationship.

LESSONS

The lessons highlighted during this investigation while significant in their own way, and useful in informing the author’s
approach to project management since, have greater real value than simply providing guidance for the future solving of
similar technical problems.

While technical papers are an essential tool in the advancement of the science that underpins our professional practice,
stories of the execution of major iconic projects are considered to be as important (or more important) in the advancement
of the art of management and leadership in our profession.

Stories can identify heroes who have executed iconic projects, point to these pioneers, and provide a real-life point of
reference for the next generation of aspiring leaders as a guide to how it is done.

According to Peters, the Essentials of Leadership are to :

• INSPIRE (enthusiastic followers);

• LIBERATE (team members’ imagination and passion);

• ACHIEVE (outstanding results through others).

In these terms the more profound learnings from the study can be set out as follows:

• INSPIRE

− Leadership is not a matter of doing excellence------ it is a matter of inspiring excellence in others.

− Recognise your dependence on the effective functioning of a high performance team.

− Honour those team members whose perspective may appear unconventional or off-beat.

(In this study, listening to an initially improbable notion that plate tectonics was a root cause of the slide helped inspire a
cascade of lateral thinking breakthroughs).

• LIBERATE

− Create opportunities: encourage the team to apply their talents and grasp these opportunities;

− “Forget command and control; Forget knowing one’s place; Forget hierarchy”
- 10 -

(In this study, liberating the team from the constraints of adhering solely to conventional quantitative engineering analysis
was a key to the successful modelling of the slide mechanism).

• ACHIEVE

− Honour those who try to achieve something rather than those who try to stop something wrong being done;

− Achieving outstanding results requires passion, persistence and imagination

(In this study persistence in pursuing the enigma of failure on planes apparently too shallow to allow movement assuming
average material strengths, allowed the achievement of a quantitatively supportable slide mechanism).

THE IMPORTANCE OF STORYTELLING

As Peters attests: “A scintillating story makes an abstract strategy real.”

On completion of a WOW project (according to Peters; one that matters; one that makes a difference; one that transforms
the organisation; one that it is worth bragging about) write up the story for inclusion in the library of leadership literature.

And then, recount the story (often) to help inspire and liberate aspiring leaders to achieve their potential.

REFERENCES

1) Hollingswoth D’Appolonia, July 1984 “Ok Tedi Project, Ok Ma Tailings Dam Site, Evaluation of
Landslide”. Report to Ok Tedi Mining Ltd.

2) Griffiths, J.N. Hutchinson, D. Brunsden, D.J. Petley and P.G. Fookes; “The Reactivation of a Landslide
During the OK Ma Tailing Dam, Papua New Guinea”, J.S. Quaterly Journal of Engineering Geology and
Hydrogeology Vol. 37, 2004 pp 173-186.

3) Tom Peters “Tom Peters Essentials, Leadership”, DK Publishing, 2005.

U:\MIKE MARLEY\TURNING HINDSIGHT INTO FORESHIGHT-REV0.DOC


Case studies 2

Forensic Studies for Failure in Construction of An Underground Station of


the Kaohsiung MRT System

Richard N. Hwang
11F, 3, Dunhwa S. Road, Sec. 1, Taipei, Taiwan 10557
wspc@wspc.com
Kenji Ishihara
Chou University, Japan
other@tp.ox.uk
Wei F. Lee
Taiwan Construction Research Institute, Taipei, Taiwan
other@tp.ox.uk

Abstract: Several buildings collapsed as a result of leakage on diaphragm wall during the excavation for
constructing O1 Station of the Kaohsiung MRT System. Resistivity image profiling was carried out to check
the quality of diaphragm walls and the effectiveness of ground treatment using CCP. In addition, undisturbed
soil samples were taken by using GP-75S sampler and laboratory tests were conducted for determining the
characteristics of soils. This paper presents the findings of these studies.
1 THE PROJECT was successfully completed and sub-structure erected in June
2006.
The construction for Kaohsiung MRT System (KMRTS) was
commenced in October 2001. The system comprises 2 lines, i.e., 3 GROUND CONDITIONS
the Red Line in the NS direction and the Orange Line in the EW
direction with a total of 37 stations and 3 depots and is expected Fig. 4 shows the soil profile obtained at Borehole WB-11. Al-
to be open for revenue services at the end of year 2007. though the thicknesses of sublayers vary from place to place, the
The station of interest, i.e., O1 Station (Sizihwan Station), is sequence shown in the figure is quite typical. Soil properties
a terminal station at the western end of the Orange Line. It is adopted in designs are listed in Table 1.
located on the north bank of the Kaohsiung Harbour as depicted
in Fig. 1. Fig. 2 shows the layout of O1 Station which is 287m
in length and 16m in width and is connected to a section of tun-
nel at its eastern end.

2 THE EXCAVATION

This 2-level station with side platforms is buried at a depth of


5.12 below ground surface. Excavation for the station was car-
ried out to a depth of 20m by using the cut-and-cover method of
construction. Diaphragm walls of 800mm in thickness were in-
Ai R

stalled to a depth of 39m in Zones A and B and walls of


iver

1000mm in thickness were installed to a depth of 37.5m in


Zones C and D. The pit was propped by six levels of struts as
depicted in Fig. 3.
On 9 August 2004, a sinkhole was formed behind Panel Kao
hsiu
S60M on the south side when excavation already reached a n gH
depth of 15m and 4 buildings (3-story) collapsed due to leakage arb
o ur
of the diaphragm wall. Excavation was halted and measures
were taken to improve the watertightness of diaphragm walls.
Excavation in Zone B was resumed in July 2005 but a sinkhole
Fig. 1 Location of O1 Station, Kaohsiung MRT System
was formed shortly at the same location for the second time.
Additional ground improvement work was carried out and exca-
vation resumed at the end of November 2005. The excavation

1
Table 1 Basic soil properties adopted in design of diaphragm determined in major projects to check whether groundwater is
walls aggressive. Table 2 shows the results of chemical tests on soils
and Table 3 shows the results of tests on groundwater in bore-
Unit weight C’ Φ’ Su holes in the section of route between O1 and O2 Stations. The
KN/m3 kPa degrees kPa aggressiveness of chemical attack is classified as “severe” in ac-
SF 20.1 0 28 cordance with Table 4 which ….(ref. )
CL 18.9 0 27 18
SM 20.2 0 32

N
CL 19.5 0 29 85
CL 19.5 0 28 113 Leakage

Table 2 Results of chemical tests on soils


Entrance
B
Hole Depth, m PH Organic Cl- SO4-2 SO3-2
(%) (mg/kg) (mg/kg) Entrance
A
OA-1 20 7.7 7.38 0.053 2176 5.1 4 buildings (3-story)
collapsed
OA-3 13 9.5 3.96 0.038 99.6 11.9
OA-4 15 8.3 5.29 0.11 324.7 7.5
OS-1 16 8.3 1.8 0.049 341.4 159 Cut-and-cover
OS-2 16 8.3 1.8 0.049 323.3 2.6 O1 Station Tunnel

Zone A Zone B Zone C Zone D

Table 3 Results of chemical tests on groundwater Fig. 2 Layout of O1 Station, Kaohsiung MRT System

Hole Depth PH Cl- SO4-2 SO3-2 Ca+2 Ca+2


m (mg/l) (mg/l) (mg/l) (mg/l) (mg/l)
OA-1 15 7.9 180 112 <2.5 59.7 80.3 4 THE EVENT OF 9 AUGUST 2004
OA-3 15 7.2 5030 680 <2.5 113 2480
OS-1 17 6.8 15800 2450 <2.5 984 5530 Water spurted from the bottom of the excavation in front of
OS-2 15 7.7 17300 2450 <2.5 1030 4780 Panel S60M, refer to Fig. 2 for location, on the southern wall at
13:20 on 9 August 2004 as excavation reached a depth of 15m.
A sinkhole of 3m maximum depth was formed behind the dia-
Table 4 Classification of aggressiveness of groundwater phragm wall and the area affected was around 500m2. Four 4-
(ref?) story buildings collapsed in less than an hour and were demol-
Cl- (mg/l) SO4-2 /SO3-2 (mg/l) ished overnight. Several low-rise shops were severely damaged
mild 0 ~ 1,000 0 ~ 150 and were demolished sometime later.
moderate 1,000 ~ 5,000 150 ~ 1,500
severe > 5,000 1,500 ~ 10,000 15.78m
very severe > 10,000 Ground Level 0
1

The subsoils on west coast of Kaohsiung City are young sedi- 2


ments with high water contents and low plasticity. Such soils GL-5.12m
can easily be softened once disturbed, or liquefied when sub- 3

jected to steep hydraulic gradients and, therefore, failures of 4

trenches were quite common during installation of diaphragm 5

walls. In many cases, mini-piles and/or micro-piles were in-


6
stalled to prevent trench from collapsing. Even so, necking still
frequently occurred and reduced the sectional areas of dia- GL-19.85m
phragm walls. One row of CCP piles was indeed used along the
perimeter of the area to be excavated for maintaining the stabil-
ity of trenches prior to the installation of the diaphragm walls. Diaphragm Wall
In some cases, the soil pockets were continuous from back to
back of the walls and became water paths. Ground treatment had Zones A & B
t = 800mm
to be applied behind defective diaphragm walls to stop the in-
L = 39m
gress of groundwater into pits. Zones C & D
The site was a salt pan a century ago. It was later used for t = 1000mm
fish farming for decades. Groundwater table was much lowered L = 37.5m

as fish farmers relied groundwater as supply because surface


water was contaminated. Ingression of sea water increased the
high chloride content in groundwater. It has been suspected that Fig. 3 Excavation scheme and retaining system
the quality of diaphragm walls may deteriorate as a result of
chloride attack and, therefore, chemical contents are routinely

2
JSG piles
1200mm diameter at 1m spacing
Bored piles 1,000m2.Subsoils at this site, being closer to the sea, are even
800mm diameter at 1m spacing
GL-12m to GL-29m GL-1m to GL-30m worse than those at O2 Station so the incident was not a sur-
prise.

6 FORENSIC STUDIES
S58M S60M
Subsequent to the event of 9 August 2004, the contractor en-
1 2 3 4 5
gaged Kaohsiung Professional Civil Engineers Association,
Kaohsiung Professional Geotechnical Engineers Association and
JSG Piles JSG piles Kaohsiung Architects Association on 16 August to form a joint
1200mm diameter at 1m spacing
(#1, 3 and 5 were in place
GL-12m to GL-29m
committee to investigate the causes of failures. After the event
#4 was being installed)
of 7 July 2005, Taiwan Construction Research Institute (TCRI)
was engaged by the contractor in December 2005 to further in-
vestigate the causes of failures.
Plan

6.1 Investigation by the Professional Associations


CL JSG Piles
2<N<3 The committee studied the field records of diaphragm wall
installation and checked the structural design of diaphragm
SM GL –12m
walls. Upon the recommendation of the committee, additional
GL –15m
10<N<28 GL –15m
soil exploration was carried out to determine ground conditions
and tests were carried out to determine the quality of diaphragm
walls. The design was found to be adequate and the failure was
CL/ML Defective between
GL-15.95m to GL-16.55 attributed to defects in wall panels.
20<N<27 GL –29m
Coring was carried out on Panels S59F, S60M and S61F in
Diaphragm Wall
CL t = 800mm September 2005 to check the integrity of the walls. Panel S60M
to GL-39m JSG Piles
appeared to be in good quality within the depth of coring of 16m.
Segregation of concrete was detected in S59F at depths of
15.95m to 16.55m. Coring was abandoned at Panel S61M as
rebars were encountered at a depth of 6m. Laboratory tests
Section were carried out on the cores obtained from Panels S59F and
S60M and an average unconfined compressive strength of 58.7
Fig. 4 The event of 7 July 2005 MPa (with a standard deviation of 83.8 MPa) was obtained on
the 11 specimens from Panel S59F and an average strength of
713.1 MPa was obtained on the 9 specimens from Panel S60M,
One row of 11 bored piles was installed behind Panels S58M while the design strength of concrete is ? MPa.
to S60M. Pumping tests were performed in November 2004 to Chemical tests on the cores indicate chloride contents vary-
see if there were other defective panels. A total of 3,285 cubic ing from 0.016 to 0.022 kg/m3, less than the tolerance of 0.30
meters of water was drawn from 6 wells and water levels inside specified in CNS 3090 for ready-mixed concrete.
the excavation closely monitored at 60 wells. The groundwater The thicknesses of Panels S60M and S61F and the locations
table inside the excavation dropped; on an average of 2.2m as a of rebars were confirmed to be adequate by using radar. How-
result of pumping. The recovery of water levels was monitored ever, Panel S59F was found to be defective.
for more than 10 days, however, the desired purpose was not
achieved as the locations of leakages could not be identified 6.2 Investigation by Taiwan Construction Research Institute
(Ho, et al., 2007).
In addition to routine tests, such as CPT and SPT, TCRI
5 THE EVENT OF 7 JULY 2005 carried out resistivity image profiling (RIP) of diaphragm walls
and laboratory tests on soil samples taken by using GP-75S
To be on the safe side, one row of JSG piles was added along sampler which is a tool specifically developed for undisturbed
the entire perimeter of the station. Furthermore, the joints be- sampling of difficult soils.
tween JSG piles were treated by using CCP piles. Pumping tests Cross-hole resistivity image profiling was carried out at
were again performed subsequently to confirm the effectiveness Panel S72MF. Basically, electrical current is applied to the
of these measures. The results were not satisfactory as the rate ground through two current electrodes and a electrical field is
of recovery of groundwater inside the excavation was only generated. The resulting voltage difference between two poten-
slightly smaller than that obtained previously (Ho, et al., tial electrodes is measured to determine the resistance of the
2007). It was decided to add more JSG piles at the back of ground. By moving the current electrodes and the potential
Panel S58M. Three new piles were installed without problem. electrodes, ground resistances can be measured for different
As No. 4 pile, refer to Fig. 9, was installed on 7 July 2005, situations. With the modern computer technology, the data ob-
groundwater brought a large quantity of soil into the pit. A tained can be analyzed to obtain the resistivities of underground
nearby hospital was endangered and the patients in the hospital materials at various locations and the results can be can be pre-
were urgently evacuated as a precautionary measure. It how- sented in a graphic form.
ever survived with only minor damage. The sinkhole was about Table 4 shows the typical values of resistivities of different
1m in depth and settlement spread over an area of about types of soils and rocks. As can be noted, the resistivities for
concrete vary from 10,000 ~ 40,000 ohm-m, while those for

3
saturated sands, silts and clay vary from 15 to 1,000 ohm-m. Crosshole RIP Crosshole RIP

The resistivity of sea water is as low as 1 to 5 ohm-m. It is


Y=1.5
therefore possible to identify defects, if any, in diaphragm walls Y=3.5 Y=3.0
Y=2.5
Y=2.0 Y=1.0 X

accordingly.
X=3.5
Pole-pole array was adopted and 3 holes were sunk on each X=3.0
X=2.5
side of the panel as shown in Fig. 5.Sensors were installed at 1m X=2.0
X=1.5 S72
intervals to a depth of 40m below ground surface. (more de- X=1.0

scription on operation) Y
0


Station 挖
Excavation區

Table 4 Resistivities of different material


JSG

Material Resistance, ρ ( Ω -m) Fig. 6 3D presentation of Resistivity Image Profileing

Fresh water 10,000 Crosshole RIP


Diaphragm
Sea water 1~5 Wall
Coarse sand, gravel (dry) 20,000 ~ 80,000
Coarse sand, gravel (saturated) 1,000 ~ 5,000 Crosshole RIP
JSG 12m
Sand (dry) 5,000 ~ 20,000
Sand (saturated) 200 ~ 1,000
Silt (dry) 400 ~ 2,000
Silt (saturated) 30 ~ 200 17m 40m

Clay (saturated) 15 ~ 30
Sandstone 100 ~ 8,000
Granite, gneiss 7,000 ~ 15,000
Concrete, gabbro 10,000 ~ 40,000
Quartzite 5,000 ~ 10,000
Fig. 7 Side view of Resistivity Image Profiling
One row of JSG piles were previously installed at the back of
the panel as shown in Figs. 6 and 7. To further improve the wa-
tertightness of the wall system, CCP piles (not shown) were in- GL

stalled at the joints to fill up the gaps, if any, between JSG piles.
RIF was carried out between depth of 11m and 40m before the Low resistivity
indicating soil
installation of CCP and again after the installation of CCP to see pocket
the differences made. GL
Fig. 8 shows the results obtained for a longitudinal section
(XZ plan, at Y = ) of the wall panel and soil pockets are identi- JSG Gr out
fiable.
GL

Low resistivity
BH-1 BH-2 BH-3 indicating soil
1.8m 2.2m
pocket with sea
water
1.35m 1.2m 1.4m
GL

S72MF S73M Resistivity (ohm-m)


0 0 3 180 10,000

Y 1.7m 1.9m 1.95m


Fig. 8 Interpretation of Results of Resistivity Image Profiling
1.57m 1.8m

BH-4 B1 BH-5 BH-6

Fig. 5 Resistivity image profiling at Panel 72

4
Before CCP Treatment After CCP Treatment

New CCP
Pr evi ous JGP Tr eat ment
Tr eat ment

X=1.0 X=1.5 X=2.0 X=2.5 X=3.0 X=3.5

Fig. 10 Results of Resistivity Image Profiling in YZ sections


Resistivity (ohm-m)
0 0 3 180 10,000

Fig. 12 3D presentation of results of Resistivity Image Profil-


ing

Fig. 13 compares the results of RIP before and after CCP


treatment.

Y=1.0 Y=1.5 Y=2.0 Y=2.5 Y=3.0 Y=3.5

Fig. 11 Results of Resistivity Image Profiling in XZ sections


ACKNOWLEDGMENTS

The authors are grateful to Da-Cin/Shimizu Joint Venture for


the authorization to publish the data presented herein.

REFERENCES

Ho, S. K., Chou, C. C., Chen, D., Chung, L-J, Liao, Z-B, Chen,
Y-Y (2007), The pumping tests at KMRT O1 Station, Proc.,
2007 Cross-Strait Symposium on Geotechnical Engineering,
April 15~17 (in Chinese)

5
CASE HISTORY 4

IMPORTANCE OF UNDERSTANDING LANDFORMS WHEN


DEVELOPING ON LANDSLIPPED TERRAIN

R.P. Thimpson
Coffey Geotechnics Ltd

INTRODUCTION
The landscape in northern England, U.K, includes many areas where there are relict
landslides which are a reflection of former glacial or periglacial conditions. Valley
landslides were typically oversteepened by meltwaters and the rapid dropdown of
glacial lakes following the breaching of ice or moraine dams. As the climate became
warmer, stability of the hill slopes returned but was often only marginal.

Increased demand for building land has led to areas of former landslipping being
considered for development. However, the first stage is to recognise and understand
that the landforms are actually relict landslides and that the margins of safety against
renewed movement are modest. Before even embarking on an intrusive ground
investigation, it is important to review both the geology and the landforms so that the
possible, if not probable factors dictating the stability can be established. These
include the depth to the plane of sliding, the shear strength operating on this plane,
and the groundwater conditions. The sensitivity of the slope to renewed changes
needs to be understood so that the appropriate range of hypotheses that could again
trigger failure is identified.

Two cases are described below in which development of adjacent areas of hillside was
planned. Each had somewhat different landforms and therefore required separate
failure hypotheses to ensure properly targeted ground investigations were conducted
and suitable geotechnical works designed and constructed.

Site A
The first site was on east-facing hillside and was planned for multi-storey industrial
development, which was to form a further part of an already developed location
(ref1). No alternative sites were available.

Published information (ref2) on the geology of the area indicated that it comprised
inter-layered shaley mudstones and sandstones belonging to the Carboniferous period.
A particularly distinctive marine fossil band, Gastrioceras cumbriense, was noted as
being present in the relevant stratiagraphy. The site itself lay about 8km south of the
ice sheet at its maximum glaciation. It is thought that a glacial lake had been formed
which subsequently drained rapidly and caused landslipping of the valley side. The
movement would have been on the shaley mudstone layers. As the water levels fell
the reprofiled valley side adopted a stable condition again, albeit marginal.

Concerning the topography of the area, the lowest part of the valley side had slopes of
about 10 degrees which gradually rose to about 15 degrees. However the site for the
development itself was backed by a mound with slopes of about 40 degrees which in
turn was followed by a stretch of level ground before the main hillside rose at an
angle of about 60 degrees. The stretch of level ground and the mound were
tentatively classified as comparable to graben and horst structures. The graben was
backed by faults with the main slope and with the back face of the horst while the
ground downslope of the front face if the horst was also landslipped terrain, see fig. 1.
The development was to be sited in the last area, downslope of the horst. The
proposed construction included a semi-basement which entailed excavation works that
had a serious potentonal to create renewed landslip movement. It was therefore
essential to gain the fullest understanding of the ground conditions and test the
hypothesis that the location was a series of relict landslides.

The initial step was to propose a ground model in which the depth to the plane of
former sliding was comparable to the width of the graben i.e. the level stretch of
ground (ref 3). A second step was to then propose that the mound or horst had
undergone a planar slide on gently dipping plane consistent with the general dip of the
strata. In addition it was recognised that it would be benefit if the marine fossil band
could be detected. Once the hypothetical ground model for the area upslope of the
site to be developed was established, the next stage would be to focus on the site itself
so that it could be safely excavated and that suitable foundations could be designed.

Boreholes were drilled firstly in the area of the supposed graben. The core revealed
steeply inclined mudstone layers intermixed with more broken rock, consistent with
slope movement. At a depth of nearly 20m, similar to the width of the graben, intact
shaley mudstone rock was encountered which thereby demonstrated that the base of
the graben had been reached and allowed location of the plane of former sliding to the
established. Sandstone was encountered after a further 20m of drilling.

The second part to be investigated was the horst block. This was found to comprise
essentially intact layers of gently dipping shaley mudstone. Marginally above the
plane of sliding, as determined from the investigation in the area of the graben block,
a marine fossil band was recovered which was identified as being the Gastruiceras
cynvruese band. Having found the fossil band, its plan position and elevation was
measured at this location. By relying on the published value for the geological dip of
the strata a search was made for possible evidence of the fossil band in the main intact
hillside. Fortunately there was a deeply incised narrow gorge with fresh exposures of
rock and the fossil band was able to be located. Its exact plan position and elevation
in this second location was measured so that an accurate appreciation of the dip of the
undisturbed strata could be determined.

The third part that was targeted for the investigation was the site selected for the
development and the area immediately downslope. With the knowledge that the fossil
band was likely to be present and that it lay marginally above the plane of sliding , the
drilling was conducted with great care. On the site itself, landslide debris was initially
encountered and close to the base of this fossil band was again met and was directly
underlain by mudstone with one thin layer, exhibiting a polished surfacing consistent
with material that had been subject to former sliding. This was carefully sampled for
subsequent laboratory testing. It was determined that the residual shear was between
11 degrees and 13 degrees, the slight variation being a reflection of testing under a
range of effective stress conditions (ref 4). Immediately downslope the interface
between the landslipped debris and the intact mudstone was at a slightly steeper
grade. This raised particular cause for concern regarding its stability, noting that the
toe of this slope had been cut away with only a low height retaining wall being in
place at this location.

By this stage the surface topography was known and the original hypothetical ground
model was proven with the location and strength of material relating to the plane of
sliding established. The final main factor was information the ground water
conditions. There were two possible sources. Firstly the underlying sandstone rock,
which would be relatively permeable, could collect rainfall and run-off from the main
hillside and feed this from below into the overlying mudstone, in effect giving rise to
an unward flow and critical porewater pressures at the plane of sliding. Alternatively,
the ground surface of the graben block was seen to become flooded after heavy rain.
The water could then, possibly , drain down through the broken rock fragments
forming the block until the water met the relatively low permeability intact mudstone
at a depth of the plane of sliding. The water would then flow along this plane in
preference to downwards through the intact mudstone. Targeted instrumentation and
careful monitoring established that this later hypothesis was most likely to be correct.
Furthermore it revealed that the downslope of the mound, on the site for the
development itself, the phreatic surface could lie very close to the current ground
surface.

Stability analyses of the existing slope gave varying factors of safety with the area
downslope of the proposed development site having only marginal stability, of the
order of 1.1. The next stage was to consider the implications of excavating the site for
the development. Again such works were found to give rise to unacceptably low
margins of safety in respect of the stability of the mound which could potentially
move on its former plane of sliding. Finally the weight of the proposed development
would render unstable the downslope section of the slope.

The design to address the various geotechnical problems was divided into three parts.
On the uphill side of the proposed excavation it was decided to install a contigious
bored pile wall with temporary ground anchors. For the permanent condition the piles
would be propped by the base slab of the development. This part was designed to
ensure that movement of the uphill side did not occur under any of the foreseeable
temporary or permanent conditions. The next part concerned the foundations of the
building itself. It was important that not only was adequate bearing capacity achieved
,but that the building weight did not impose load on the downslope area. Accordingly
bored pile foundations were selected with the piles being founded below the plane of
sliding. To increase stability of the downslope area itself, trench drains were designed
to lower the ground water levels. By this means the factor of safety against instability
was raised to at least 1.3 even under the most severe conditions.

Conclusions for Site A

i) The initial ground model and associated failure hypothesis were rigorously
researched and were based on a comparatively complex series of landslipped
blocks.
ii) The location of the plane of sliding was provisionally estimated together with
the groundwater conditions and formed the basis of a tentative model.
iii) Ground investigations and detailed forensic geotechnical examination of the
data obtained confirmed the validity of the original model and failure
hypothesis.
iv) With this information to hand, the appropriate geotechnical design involving
an anchored bored pile wall, bored pile foundations and drainage trenches
could be determined and implemented.
v) The case history A has shown that a comparatively complex landslip area can
be effectively and safety developed provided the mechanism of the landslip is
fully understood.
Site B

The second site included a section of the access road to site A and was also on a hillside that
included landforms consistent with relics landsliding. Similar geological conditions to site A
existed with the solid strata comprising mudstone ad sandstone.

The general topography of the area included a gentle convex toe section of the slope which
then rose at a gradient of about 10 degrees, increasing to about 15 degrees before reaching a
level section or plateau area at about 100m distant from the tone, see fig. 2. Beyond this the
ground rose relatively steeply.

Some excavation works in the toe area of the slope had been carried out already to allow
constructon of the access road to be started. The contractor seemed to be peridiocally finding
that after excavating a section on one day to a specified profile he was returning the next day
only to find that he had to repeat the works to achieve the required cut profile. The problem
was noted as being more marked in wet weather.

Examination of the overall slope area revealed that fissures in the ground were appearing
about 100m upslope of the toe at the junction between the top of the 15 degree slope and the
plateau area. There was little doubt that the excavation works for the access road had
triggered movement of a relict landslide. In addition it was the lower stage being disturbed
by the construction works. However the detailed mechanism controlling the behaviour of the
slope was not known. Two possible hypotheses were proposed. The first model involved a
relatively shallow failure surface at about 3m depth with surface run-off ponding on the
plateau area and then infiltrating the ground to give rise to periodic increases in the pore
pressure on the failure surface. The second model included a failure surface at 6m to 7m
depth with the pore pressures being controlled by variations in groundwater flow rates and
therefore rainfall.

Boreholes were drilled on the landslipped area in order to determine the actual ground
conditions with depth and piezometers were installed so that the groundwater conditions
could be recorded. The piezometers were also designed to act as basis slip indicators. In
addition movement surface monitoring pins were placed spanning the zone at the edge of the
landslipped surface so that magnitude and direction of movement between the landslip and
adjacent stable ground could be established.

Remedial solutions to stablise the movement were planned to be centred around the
installation of drains to reduce the porewater pressures on the sliding plane. However until
the ciritical plane and associated water pressures were known, it was not possible to proceed
with the design of a solution let alone its implementation. One aspect that became apparent
from the borehole investigation was that a geological fault diagonally traversed the centre of
the landslip area so that mudstone rock was displaced downwards over the lower section of
the landslip. At the fault interface, therefore, more permeable sandstone rock was juxtaposed
against less permeable mudstone rock.

The boreholes also revealed that non-solid deposits, being drift or landslipped material,
extended to about 6m to 7m over most of the landslipped area increasing to about 10m depth
near the head of the slip. However there was no horizon that could be conclusively
determined as being the slip plane, other than possibly the interface between non-solid and
the solid deposits.
A rockfill berm was placed at the toe of the slope to provide a modest improvement in the
short-term stability of the slope. It was not until about nine months afterwards that there was
a period of particularly heavy rainfall. The piezometers revealed that there was a marked
increase in pore pressure at the interface of the non-solid and solid deposits. Futhermore the
slip indicators in the piezometer tubes confirmed that movement was taking place on this
horizon. The original ground model and its associated failure hypothesis were duly modified
to embrace the information that had been established regarding the location of the movement
plan, the presence of the faulting and that this fault dictated the critical porewater pressures.

From this information, a 7m deep drainage trench was designed and constructed running
diagonally across the slope and positioned immediately downslope of the line of the fault.
This ensured that under very wet conditions the groundwater was drained away from the fault
zone before it had the opportunity of reaching and disturbing the stability of the toe of the
slope. With the toe stablised, the upper pat of the slope was effectively buttressed so that it
likewise was rendered stable.

Conclusions for Site B

i) The intial ground model and associated failure hypotheses were reasonably based
on the site being a straightforward two-stage relict landslip.
ii) The details relating the plane of sliding and critical porewater pressures were not
known but two tentative models were proposed.
iii) Ground investigations revealed that the original ground model required
modification in order to properly understand the behaviour of the landslip.
iv) With this information to hand, appropriate drainage works to address the critical
porewater pressures could be determined.
v) The tentative drainage solutions based on either of the original modesls would
have been seriously flawed and would not have provided stability to the landslip.
vi) The case history B has shown that an apparently relatively straightforward
landslip may, in fact, be complex and that rigorous forensic geotechnical
engineering is essential to gain a correct understanding.

Overall Conclusions
1. The examples for the two sites, which concerned relict landslides, demonstrate
that it is essential from the outset to develop well thought out and researched
failure hypotheses together with ground models and that this should be done
prior to intrusive ground investigations being carried out.
2. On the basis of the hypotheses, the ground investigations can be target to give
the optimum results. It is essential to review and modify, if necessary, the
initially chosen hypothesis in the light of the findings from the investigations.
Provided this is achieved, the geotechnical aspects of a development can be
designed and implemented with known levels of safety.
3. One of the sites, Site A, was complex but the rigorous forensic geotechnical
examination of the findings from the borings and field exposures allowed the
original ground model and associated failure hypothesis to be confirmed.
4. The other site, site B, appeared to be straightforward two-stage landslide.
However the two originally proposed possible failure hypotheses were found
not to match the actual ground conditions as revealed by intrusive
investigations. It demonstrated the requirement to carefully examine the
information obtained and then to modify the original hypothesis accordingly.
5. From the examples it is evident that if proper failure hypotheses and targeted
ground investigations are not conducted, the subsequent development could
lead to renewed slope movement. This is exemplified, in particular, by the
initial works that were carried out at site B.

References
1. Leach B.A. and Thomson R.P. (1987), The influence of underdrainage of the stability
of relic landslides. 9th ECSMFE Dublin Vol.1pp 447-450, Balkema
2. The Geology of the Country around Huddersfield and Halifax (1930), Explanation of
Eological Sheet 77, Memoirs of the Geologifcal Survey of England & Wales. DSIR
His Majesty’s Stationary Ofrice London.
3. Holmes A. (1965), Principles of Physical Geology. Second Edition pp 1046-1051,
Nelson.
4. Thompson R.P (1991) Stabilisation of a landslide on Etruia Marl.International
Conference on Slope Stability, Isle of Wight UK. Pp 403-408. The Institution of
Civil Engineers. Thomas Telford.
Case Studies 5

Monitoring in Forensic Geotechnical Engineering


- Dam Construction –
-
Yoshi Iwasaki, Ph.D., Dr of Eng, P.E.
Managing Director
Geo-Research Institute, Osaka

1 Observational Procedure
The text book of “Soil Mechanics in Engineering Practice, Second Edition” by K.
Terzaghi and R.Peck(1967) added Chapter 12 Performance Observations to the first
edition. It
was the first to point the enrolment of the monitoring soil and structural performance
during construction work as to provide evidence in lawsuits.
It states that a fair decision is to be expected only if the causes and the nature of the
mishap are known. If the contractor or owner can prove that he has anticipated the
undesirable condition, has observed its progress during construction, and has done
everything possible to avoid it, he is in a much more favorable position than if the
condition has taken him by surprise. The element of surprise not only injures his
professional reputation, but it may also injure his financial standing.
When the third edition by the authors of K.Terzaghi, R.Peck, and G.Mesri was published,
the Chapter 12 in the second edition was deleted and refers observational procedure
only in a single paragraph of “10.5 Observations during Construction,” where no legal
aspects was mentioned for usage of monitoring.

2 Enrolment of observation in forensic geotechnical


In the third edition of “Soil Mechanics in Engineering Practice”, it is stated that “in
earth-dam and foundation engineering the permanent structures are designed before the
construction operations start, and the consequences of unanticipated sources of trouble
do not appear until the structure is in an advanced state of construction or is in service.”
This statement sounds observational method is not useful for earth dam and foundation
engineering.
The author will review several case histories that are constructed with or without
monitoring including forensic problems and try to consider any meaning of monitoring
in dam construction.
In very early stage of dam construction, St.Francis Dam, Los Angeles County,
California, U.S.A. was constructed as a concrete gravity dam in March 1926. During the

1/7
initial filling, cracks in the dam began to appear and was tried to seal the cracks. During
the final stage of the second filling in March 1928, new leakage was found and the dam
was failed. The St.Francis Dam is considered the greatest American civil engineering
failure of the twentieth century. Approximately 450 mostly sleeping people died by
flood caused by the failure of the dam. When the dam was constructed, dam engineering
was just to start to develop based upon these failures and lessons learnt.
Another example is Vajont dam, Italy that was a concrete arch dam constructed in 1962
and failed in 1963.

2.1 Vajont Dam


Vajont Dam is a dam completed in 1961 under Monte Toc, 100km north of Venice, Italy.
It was one of the highest concrete dams in the world measuring 262m in height with
27m in base thickness and 3.4m in top thickness.
Shifting of rock was noticed during the first filling in February 1960 before final
completion of the dam of September 1960. By March 1960, the level of the reservoir
reached 130 above the level of the river, when the first small slide was observed.

Table-1 The Sequence of Filling/Draw Down leading to Failure


water movement
date mass movement/land slide
level rate
1960/2
1960/3 130m
The First Filling
1960/10 170m 3.5cm/day potential huge slide area recognized
Stage 1960/11/4 180m 0.7millon slid
185m 8cm/day
Draw Down
135m 0.1cm/day

1961/10 135m
The Filling 1962/2 185m
Second 1962/11 235m 1.2cm/day
Stage 1962/11 1cm/day
Draw Down
1663/2 185m 0cm/day

1663/3 185m
1993/5 231m 0.3cm/day
1993/6 237m 0.4cm/day
Filling
The Third 1993/7 240m 0.5cm/day
1993/8 240m 0.8cm/day
Stage 1993/9 245m 3.5cm/day
1993/10 245m
Draw Down 1993/10/9 235m 20cm/day
1993/10/9 22:38 heavy rain 260million m3 slid

2/7
In October 1960, the level reached 170m, a rapid increase in the rate of displacement to
approximately 3.5cm/day was observed. At this time, a huge joint of 2km in length was
found to open up in an area of slope 1700m long and 1000m wide, suggesting very large
slide was mobilized. On November 4, 1960, there took place a land slide of 700,000
cubic meters at the level of 180m.
The water behind the dam was lowered to 135m and the movement was reduced to
1mm/day. The safety of the slope of the dam was discussed and calculations ordered by
management showed that catastrophic failure was unlikely and the reservoir allowed to
refill water under controlled monitoring.
Thus the reservoir was filled and slowly emptied three times. On October 9, 1963 at
10:38 pm, the combination of third drawing down of the reservoir and heavy rains
triggered a big landslide of about 260million cubic meters of forest, earth, and rock,
which fell into the reservoir. The resulting splashed water caused some 50million cube
meters of water to push up the opposite bank and destroyed the Village Casso, 260m
above the lake level before overtopping the concrete dam of 250m in height. The water,
estimated about 30million m3 then fell more than 500m onto the downstream villages of
Longarone, Pirago, Villanova, Rivalta, and Fae destroying them and killed some 2500
persons.
The court condemned eight persons in guilty not to take any action of evacuation of the
inhabitants.
In the case of Vajon dam, monitoring was being carried out, however, it was very
difficult to predict the failure of the slope.

2.2 Teton Dam


Teton dam constructed and failed in1976, Idaho, U.S.A. is another case that has no
monitoring except measurement of deformation.

Teton dam, Idaho, the major feature of the multipurpose Teton Basin Project of the
United States Bureau of Reclamation, failed on June 5, 1976 while the reservoir level
was 1m below the spillway sill. The dam was a center core zoned earth fill structure,
with a height of 93m above the river bed and 123.4m above the lowest foundation.
The instrumentation of Teton dam was not prepared because performance records were
considered to be available for dams “constructed of similar material and on similar
foundations” and only settlement monuments were provided.

2.2 Takase Dam

3/7
Takase dam is a central core earth-rock dam, 176m in height, 362m in crest length built
in 1978 by Tokyo Electric Power Company Inc. as a part of Takase River Hydraulic
Power Development Project developed in Takase Basin, central Japan.

Fig.1 Pore pressure measurements at Takase dam during initial filling (1)

4/7
Fig.2 Pore pressure measurements at Takase dam during initial filling (2)

Fig.3 Change of equi-pore water pressure line during the first filling
(Dec.,’78, Feb.,79, Apr.,79, and July, 79)

Instrumentation of Takase dam was arranged to provide pore pressure, earth pressure,
displacement in dam section, and surface targets for measurement of surface
movements.
In placing the pore pressure gauges for control of the dam construction work, it was
decided to concentrate mainly on the core and transition zones, where comparatively

5/7
high pore pressure was expected to develop, and also to a central extent on the inner
shell zone, which has a relatively high fine grain content.
Pore pressure gauges for dam security control during and filling of the reservoir were
located on the core contact area and within the foundation bedrock, and also in positions
that enable tracking of the downstream phreatic line.
Fig.1 and 2 show changes of pore water pressure during the initial filling water in the
reservoir. The filling started in December, 1978 and reached nearly completion in July,
1979. Each figure shows the measured pressure at three different heights within the
embankment of EL=+1107m, EL=+1165m, and EL=1135m. Fig.3 shows change of
equi-contour line during the filling (Dec,’78(left top)-Feb., 79(left bottom, Apr.,79(right
top), and(July,79(right bottom )). It is seen that the distribution of the pore pressure in
the field is sound condition from which no leakage is expected.

2.3 Dams that were saved from failure based upon monitoring
ASCE Task committee on instrumentation and monitoring dam performance
summarized various case histories that showed saved dams from catastrophic failures
based upon instrumentation and monitoring.
The following three examples are from the case studies shown in the guidelines for
Instrumentation and Measurements for Monitoring Dam Performance.
Findlay(1988) reports a dam for a hydroelectric project in 1986 included the installation
of instrumentation and monitoring system. Pneumatic piezometers were installed to
monitor hydrostatic pressure during and following the initial filling. It was found the
reading of two sensors increased equivalent with rising water elevation. Piping might
have occurred if the head condition were continued. The reservoir was immediately
dewatered. The pore water gages responded with the same amount of the dewatered
level. The two piezpmeters were installed near a contraction joint. Modification was
made for inadequate sealing and a safe condition was achieved.
Only after one year of a dam of USBR(Fontenelle Dam) in 1965, piping of the
embankment and surface erosion caused a partial failure. Damage to the 36.6m high,
1626m long rolled-earth embankment was repaired and the dam put back in service. In
1982, instrumentation and monitoring indicated the presence of new seepage. A cut-off
wall of about 60cm thick and 48m deep along the whole embankment of about 1.6km
was installed to stop the leakage. (ENR,Nov.26, 1987)
Navajo Dam of U.S. Bureau Reclamation was detected seepage problem during the
initial filling in 1963. Seepage reading of the dam had been increased until the
installation of a 100cm thick, 120m deep, 135m long cutoff wall.(Civil Engineering,

6/7
March 1987)
It is clear that the observational procedure has been applied successfully in dam
constructions based upon monitoring of the dam behavior during the first filing as the
final stage of construction or even later phase under service stage.

3 Conclusions
If you search for “Teton Dam,” through internet Home Page, you will find hot
discussions still going on the true reason that made the dam failed.
If the performance was monitored, the discussion might be more fruitful and practical
based upon real data.
Failure of Teton Dam clearly shows the measurement of power pressure within dam
section is necessary and if it had been measured, the failure of the dam itself had been
prevented. Vajon dam shows inadequate theory and practice of the slope failure in
geotechnical engineering.
The statement in the textbook of “Soil Mechanics in Engineering Practice” the
consequences of unanticipated sources of trouble do not appear until the structure is in
an advanced state of construction or is in service” should be replaced by “the
construction of a dam continues until the initial filling water and should be confirmed as
assumption by design or ready to be modified based upon the monitoring results during
the filling water in the dam.

References:
W.L.Chadwick(1977), Case Study of Teton Dam and Its Failure, Proceedings of the
Ninth International Conference on Soil Mechanics and Foundation Engineering, Case
Study Volume, Tokyo, Japanese Society of Soil Mechanics and Foundation Engineering,
pp431-455
R.Takai, T.Iwakata, and Y.Miyata (1977), Results of Soil Tests and Measurements
during and after Construction of the Takase Dam, Proceedings of the Ninth International
Conference on Soil Mechanics and Foundation Engineering, Case Study Volume, Tokyo,
Japanese Society of Soil Mechanics and Foundation Engineering, pp497-554
ASCE Task Committee(2000)”Guidelines for Instrumentation and Measurements for
Monitoring Dam Performance”
Findlay, R.Craig (1998), “Hydrostatic Pressure at a Soil-Structure Interface,”
Proceedings of the Second International Conference on Case Histories in Geotchnical
Engineering, University of Missouri-Rolla, St.Louis

7/7
Case Study 7

FORENSIC ANALYSIS ON RE WALL DISTRESS: A CASE STUDY

Sivakumar Babu, G. L.1, Murthy, B. R. S.2 and Singh, V. P.3

1. Associate Professor, Dept. of Civil Engg., IISc, Bangalore, India, 560 012, gls@civil.iisc.ernet.in
2. Retd. Professor, IISc, Bangalore, India, 560 012, murthybrs@gmail.com
3. Research Associate, Dept. of Civil Engg., IISc, Bangalore, India, 560 012, vikasps@civil.iisc.ernet.in

ABSTRACT

The article presents a forensic analysis of the observed distress of an RE wall. The
investigation involved field observations, extensive laboratory testing of backfill soil properties and
reinforcement material (i.e. geogrids), analysis of failure cause and justification for the soil nailing
technique as a rehabilitation measure. The article also highlights soil nailing technique as a viable
alternative for rehabilitation of failed reinforced earth (RE) structures.

1. INTRODUCTION

Forensic geotechnical engineering (FGE) may be defined as the discipline that prepares
geotechnical engineers to investigate failures of structures such as embankments, pavements,
foundations and earth retaining systems, in purview of professional ethics and legal bindings. FGE
being a relatively new terminology in geotechnical engineering profession, it is necessary to create
general awareness and understanding about the subject among geotechnical engineers, clients, and
contractors.

In this article, a case study of RE walls constructed for a flyover is presented wherein a
detailed investigation was sought by the client with regard to the examination of the distress noted
during the construction. The methodology adopted to carry out detailed investigation involved
following sequence of steps.

1. Survey and characterization of the distress i.e. site visit and observations.
2. Diagnostic tests i.e. collection of material (backfill and geogrid) samples for laboratory testing to
evaluate the behavior of soil and the strength characteristic of geogrids used.
3. Development of most probable failure hypothesis and cross check with original design.
4. Back analysis of RE walls stability using conventional methods, and rigorous numerical analysis
using computational code.
5. Suggestion and justification for the rehabilitation measure in the form of driven soil nails.

2. SURVEY AND CHARACTERIZATION OF DISTRESS

Site visits were made to (a) identify and study locations of distress, (b) collect the samples of
backfill and geogrids for laboratory testing, and (c) procure a preliminary report on the possible
cause of distress from the client/contractors. For the purpose of identification, the two ends of
flyover were named as the Siliguri end and the Dalkola end. The construction work of the RE wall
started in June 2006 and the Siliguri end of the flyover was completed in April 2007 and the Dalkola
end by August 2007. During June –October 2007, the area experienced severe rains which were of
unprecedented magnitude causing distress in the RE wall that was being constructed. On the Siliguri
end, distresses in RE wall were noted in the following forms: i) damage to reinforced earthwork ii)
damage near abutment leading to misalignment of panels, and iii) sand flow was noticed and one of
the panels got dislocated in the region near the design height of 8 to 8.8 m. On the Dalkola end, the
client informed that the sand removal led to overstressing of the connection, resulting in the
breakage of grid and panels. A few photographs of the distress locations in the RE walls taken
during the site visits are shown in Figures 1(a)-(c).

Figure 1(a): Damage near flyover abutment

Figure 1(b): Bulging of facing panel


Figure 1(c): Damage near top facing panel and the approach slab

3. DIAGNOSTIC TESTS ON MATERIAL SAMPLES

3.1 Test Results of Backfill Soil Samples

Backfill soil samples collected during site visits were tested in the laboratory and used to
obtain shear strength parameters in both the undisturbed and saturated conditions. Table 1 shows the
summary of the shear strength parameters obtained from laboratory tests on backfill soil samples.

Table 1: Shear strength parameters of backfill

Bulk density Peak friction angle Residual friction


Soil state 3 Cohesion (kPa)
(kN/m ) (degrees) angle (degrees)
In situ
17.50 40.0 35.1 0
condition
In situ
16.08 39.1 33.2 0
condition
Saturated
17.50 30.54 29.8 0
condition

It is noted that the shear strength parameters for the in situ condition are higher than the
design value of 32o. Tests on saturated sand indicate that there is a marginal decrease in the friction
angle. The permeability coefficient of the backfill corresponding to the bulk density of 17.50 kN/m3
is in the range of 0.089 mm/sec. It is likely that the rate of infiltration during the intense rainfall was
higher than this value.
3.2 Strength and Permittivity Characteristics of Geogrids Samples

The permittivity of the geogrid (geotextile) used is 1.85 sec-1 and the thickness of the geogrid
is 1.68 mm which corresponds to permeability coefficient of 3.1 mm/sec. For proper filter behavior,
as per NCMA (1996) guidelines, the geogrid permeability coefficient needs to be not less than 10
times higher than the permeability coefficient of the backfill. In this case, the geogrid is highly
permeable and the factor of safety with respect to filter criteria is in the range of 3.48 (3.1/10 x
0.089) and hence, the geogrid used was satisfactory and a separate drainage medium behind the
facing panels was not essential. Further, the design strength and other properties of geogrid samples
were assessed by another agency and the result supplied by them showed that the used geogrids
satisfied their technical specifications.

4. DEVELOPMENT OF MOST PROBABLE FAILURE HYPOTHESIS

Rainfall leads to saturation of the backfill and/or pore water pressure development which
needs to be addressed properly. The effect of rainfall on reinforced wall can be understood from
design charts for reinforced soil slopes for dry and saturated conditions developed by Jewell (1990).
Figures 2(a) and (b) show the minimum required force coefficient (kreq) for reinforced soil structures
with slope angles varying from 0 to 90 degrees. The state of pore water pressure is expressed in
terms of pore pressure coefficient ru = u/γz, where u is the pore pressure, γ is backfill bulk density
and z is the depth at the point of considerations of pore water pressure (u) and vertical stress (γz).
The pore pressure coefficient varies from 0 to 0.5 and dry/fully drained backfill has ru = 0, whereas
presence of saturated water in the backfill is indicated by ru = 0.5.

It is expected that minimum required force expressed in terms of (kreq) in the reinforcement
for equilibrium is lesser for fully drained backfill. From Figures 2(a) and (b), it can be noted that for
fully drained backfill, for the slope angle 900 representing wall, for design friction angle of 320, the
kreq value is 0.31. This value increases to values in the range of 0.60 when pore pressure in the
backfill exists. The pore pressure generated in the backfill in a momentary condition as a result of
heavy downpour and is not considered in design. Thus, when rain of unpredicted magnitude occurs
and if the infiltration rate is higher than rate of discharge, seepage pressures build up and
momentarily induce additional pressure on the retaining system. The local bulging observed in a few
locations in the RE walls of the flyover is attributed to the above phenomenon.

Further, for a completely covered surface in the plan area of the wall, percolation of water is
not allowed. For this condition, using design friction angle of 32o, the minimum required force
coefficient (kreq) for reinforced soil structure of 90o (vertical wall) is 0.31 and the corresponding
force required to be supplied by the reinforcement at a depth of 10.40m is (kreq)γH is
0.31x17.5x10.4 = 56.42 kN/m. It is noted that the geogrids provided at this depth have higher
strength than required value of 56.42 kN/m. However, under the influence of pore pressures, the
minimum required force coefficient (kreq) is in the range of 0.64 and hence, the force momentarily can
increase to a maximum value of 116 kN/m, applying considerable force on the reinforcement
and panels which could have contributed to the distress noted.

It is also noted that for the fourth facing panel from the top, two types of geogrids (SS120
and SS150) which have different stress-deformation characteristics were used. The tensile force
required for stability is marginally in the same range whereas elongations required to obtain the
same force in each geogrid layer are different. This could have contributed to marginal drift in the
facing panels under the influence of earth pressures.

Figure 2: Jewell (1990) design charts for reinforced slopes (a) ru = 0; (b) ru = 0.5.

5. BACK ANALYSES OF RE WALLS STABILITY

The back analyses to examine the stability of the geogrid reinforced retaining walls was
carried out using conventional method (FHWA 2001) and by performing rigorous finite element
analysis using a computational code (Plaxis 2006). Various sections of the RE walls with different
design height of wall (e.g., 10.40 m, 9.60 m and 8.80 m) were analysed separately. In the present
paper, results of the analyses corresponding to the maximum design height (i.e. 10.40 m) are
presented and discussed. Table 2 shows the material properties and other parameters adopted for the
study.

Table 2: Material properties considered for analysis

Parameter Value
Design height of the wall, H (m) 10.40
Unit weight of wallfill, γ (kN/m3) 18
Angle of internal friction of wallfill, φ (degree) 32
Surcharge load, ws (kPa) 22
Length of geogrids member, L (m) 7.80 (7.20)
Vertical spacing of geogrids members, SV (m) 0.80 (0.365)
Geogrid-soil interaction coefficient, μ 0.85
Design strength of geogrid members As per the data supplied by the client
Note: Figures in bracket are for the bottom most layer of geogrid.
It is to be noted that the properties of foundation and backfill are assumed to be same as that
of wallfill. Tables 4 and 5 shows the typical results. Figures 1 and 2 shows the finite element models
of the RE wall of 10.40 m design height with and without use of soil nailing as remedial measure
respectively.

5.1 Results of Conventional Analysis

RE wall is found safe against overturning failure with minimum eccentricity of 0.85 mm
which less than the permissible vale of 1.30 mm (i.e. L/6). Also, wall is safe against sliding failure
having a minimum factor of safety = 2.50 which is more than the minimum recommended value (i.e.
1.50). Further, no evidence of bearing capacity failure were reported or found during site visit. Thus,
it can be concluded that the RE wall is externally stable. Table 3 shows the result of the internal
stability analyses. According to FHWA (2001), minimum recommended values for factor of safety
against pullout FSP and tensile failure FST of the reinforcement are 1.50 and 1.30, respectively. It
can be seen (indicated in bold) from Table 3 that the FSP values for the top four geogrid layers and
FST values for the bottom four geogrid layers are less than the minimum recommended values.

Table 3: Results of conventional analyses for section with design height 10.40 m

Design tensile
Geogrid layer Depth Factor of safety
strength of Factor safety against
number ( from z (m) against tensile failure,
geogrid, Pd pullout, FSP
top ) FST
(kN/m)
1 0.44 29.70 0.81 4.36
2 1.24 29.70 0.99 2.85
3 2.04 39.60 1.19 2.79
4 2.84 39.60 1.41 2.17
5 3.64 59.50 1.64 2.62
6 4.44 59.50 1.91 2.17
7 5.24 59.50 2.19 1.82
8 6.04 74.40 2.51 1.93
9 6.84 74.40 2.85 1.66
10 7.64 74.40 3.24 1.43
11 8.44 74.40 3.68 1.24
12 9.24 74.40 4.16 1.09
13 10.04 74.40 4.35 0.90

5.2 Results of Finite Element Analysis

As mentioned earlier, the section of RE wall with design 10.40 m was simulated using finite
element computational code Plaxis (2006). Mohr-Coulomb model was used to represent soil
behaviour and geogrid layers were simulated using ‘Geogrid’ structural elements (yellow color)
available in the code. Details on simulation procedure can be found in the user manual of the
computational code. Figure 3 shows a typical example of numerically simulated RE wall.
Figure 3: Typical example of numerically simulated RE wall.

The results of the finite element analysis are shown in Table 4. From Table 4, it can be seen
that the bottom geogrid layers are not safe against tensile failure. Further, the maximum
displacement of the wall is observed as 26.53 mm.

Table 4: Results of finite element analysis for section with design height 10.40 m

Axial force
Geogrid Design strength Horizontal
Depth z developed in
layer of geogrid, PD displacement FST
(m) each member
(from top) (kN/m) (mm)
(kN/m)
1 0.44 29.70 16.34 26.53 1.82
2 1.24 29.70 16.56 26.44 1.79
3 2.04 39.60 12.21 26.42 3.24
4 2.84 39.60 16.52 26.39 2.40
5 3.64 59.50 19.20 26.36 3.10
6 4.44 59.50 18.99 26.32 3.13
7 5.24 59.50 22.36 26.27 2.66
8 6.04 74.40 32.05 26.22 2.32
9 6.84 74.40 37.06 26.16 2.01
10 7.64 74.40 46.30 26.10 1.61
11 8.44 74.40 60.73 26.04 1.23
12 9.24 74.40 78.99 25.98 0.94
13 10.04 74.40 133.59 25.94 0.56
5.3 General Observation from Back Analyses

From Tables 3 and 4, it is evident that in the bottom layers of geogrids, the axial force
developed is more than design tensile strength of geogrids; at depths beyond 6m as the factor of
safety is close to 2 or less. The height of the wall above the ground level is in the range of 8m.
Hence, in this range it is preferable to increase the factor of safety with respect to tension, and, it is
recommended that soil nailing could be employed.

6. SOIL NAILING AS REHABILATION MEASURE

FHWA (2003) may be referred for detailed information on the analysis, design and
construction of soil nail walls. Figure 4 shows the numerically simulated model of RE wall with soil
nails. Table 5 shows the results of finite element analyses of the RE wall, additionally reinforced
with 25 mm diameter horizontally driven nails of length 7.80 m spaced at 0.80 m vertically between
geogrid layers and 1.00 m horizontally along the length of the wall. From Table 5, it is evident that
improvement in the values of factor of safety against tensile failure FST is observed. For the bottom
most geogrid layer FST value has increased from 0.56 (for geogrids alone) to about 1.11 (for
geogrids along with soil nails) which almost two times improvement. It is also observed that
maximum axial force developed in the bottom most soil nail is about 58.08 kN which yields a factor
of safety value against nail tensile failure FST equal to 3.53. Hence, use of soil nailing significantly
improves the stability of RE wall and is therefore, suggested as the possible remedial measure for
short term and long term stability of RE wall. Soil nails may be provided where the factor of safety
with respect to tensile forces is less than 2 and it can be noted from Table 4 that soil nailing is
required at the top as well as bottom. It may also be noted that the wall height above the ground
level is 8m. Hence, the results presented in Table 5 needs to be considered for 8 m depth only.

Figure 4: Numerically simulated RE wall with soil nails (black elements indicate soil nails).
Table 5: Results of finite element analysis of RE Walls with soil nails*

Axial force
Geogrid Design strength Horizontal
Depth z developed in
layer of geogrid, PD displacement FST
(m) each member
(from top) (kN/m) (mm)
(kN/m)
1 0.44 29.70 8.57 13.07 3.46
2 1.24 29.70 9.12 12.86 3.25
3 2.04 39.60 6.38 12.75 6.20
4 2.84 39.60 7.07 12.69 5.60
5 3.64 59.50 7.50 12.65 7.93
6 4.44 59.50 6.00 12.63 9.91
7 5.24 59.50 6.72 12.61 8.85
8 6.04 74.40 10.42 12.58 7.14
9 6.84 74.40 13.87 12.55 5.36
10 7.64 74.40 18.96 12.53 3.92
11 8.44 74.40 25.15 12.50 2.95
12 9.24 74.40 35.04 12.48 2.12
13 10.04 74.40 67.12 12.47 1.11
* - 25 mm diameter driven nails of length 7.80 m spaced at 0.80 m vertically between
geogrid layers and 1.00 m horizontally.

6.1 Advantages of Using Soil Nails as Rehabilitation Measure

As indicated from the results of the numerical analyses of RE wall with soil nails (see Table
5), the contribution of soil nails in improving the stability of the RE is found to significant.
Introduction of soil nails helped in the aspects:

1. Redistribution of axial forces among soil nails and geogrid layers and thus improving the tensile
capacity of the original reinforcement (i.e. geogrids).
2. Reduction in net force at facing connection thereby improving connection stability.
3. Reduction in the lateral displacement of the RE wall to almost 50%.
4. Improvement in the overall stability of the RE wall against any unforeseen event such as
unprecedented heavy rainfall.

7. FINAL RECOMMENDATIONS AND PRECAUTIONARY MEASURES

The analysis shows that the unprecedented rains during construction (temporary
phenomenon) resulted in very high lateral pressures contributing to minor movement of panels in the
area. The geotextile provided for filtration was adequate. The backfill properties measured from
undisturbed samples were satisfactory. However, it is essential that proper drainage measures should
be implemented so that the backfill is not saturated during the remaining part of construction.
Soil nailing to improve the stability of RE wall is recommended as the rehabilitation
measure. Driven soil nails of 20 mm diameter and length 5.6 m are suggested near the Siliguri end
abutment. In general, nail lengths of 0.7H may be provided at locations of distress. The height shall
be reckoned as the height above the ground level. Two nails per facing panel may be provided when
the height exceeds 6 m. It is to be noted that the above recommendation about the provision of soil
nails is based on the similar analysis performed for other sections of RE wall with smaller heights,
keeping in view the cost of rehabilitation process, and ease of application.

7.1 Suggested Precautionary Measures

1. It is essential that the nailing process should not leave the facing ugly.
2. As noted during the field observations, gap was not maintained between the top of the facing
panel and the approach slab, creating additional loads on the facing panels. This contributed to
movement of the facing panel. It is suggested that gap may be created such that the stress
transfer is avoided.
3. It is reiterated that normal conditions considered in design shall be ensured during execution.
4. Unexpected loads in the form of seepage/pore pressures, loading on the facing from approach
slab shall be avoided.
5. It is advisable that during impending monsoon, the embankment area be covered properly and
handle local problems if any from outside only.

8. CONCLUDING REMARKS

The article presents a case study of failed RE wall, wherein, a through investigation about
the cause of distress observed during the construction of RE walls was sought by the client.
Although, this study do not addresses the legal issues related to the investigation, it attempts to
describe the general procedure that may be followed in the forensic geotechnical engineering.

9. REFERECNES

FHWA. (2001). Mechanically Stabilized Earth Walls and Reinforced Soil Slopes: Design and
Construction Guidelines. Report FHWA-NH1-00-043, U. S. Department of Transportation,
Federal Highway Administration, Washington D. C.
FHWA. (2003). Geotechnical Engineering Circular No. 7 - Soil Nail Walls. Report FHWA0-IF-03-
017, U. S. Department of Transportation, Federal Highway Administration, Washington D. C.
Jewell, R. A. (1990). Revised design charts for steep reinforced slopes. Proceedings of the
Symposium on Reinforced Embankments: Theory and Practice, Thomas Telford, 1-30.
NCMA. (1996). Design Manual for Segmental Retaining Walls. National Concrete Masonry
Association, 2nd edn, NCMA, Herndon, VA.
Plaxis (2006). Reference Manual. Delft University of Technology & PLAXIS B.V., The
Netherlands.
Case Study 8

DISTRESS TO REINFORCED EARTH WALL EMBANKMENT: A CASE STUDY

Prof. T.G. SITHARAM


Professor and Chairman, PMG,

Prof. K.S. SUBBA RAO


Professor (retd) Department of Civil Engineering
Indian Institute ofScience
Bangalore – 560012

SUMMARY

In this case study, the distress to the Road over bridge (ROB) with a 6 lane highway
(about 30m wide) constructed using reinforced earth wall technique with a maximum height
of about 11m is discussed. A view of constructed RE wall is shown in Fig1. The longitudinal
cracks as seen on the ROB-II are shown in Figure 2. The typical cross-sectional details of
ROB are given in Fig 3. After the completion of the ROB-II, it was opened for traffic on
15.07.2007. During inspection on 20th September 2007, some longitudinal cracks were noted
on the pavement parallel to the reinforced earth wall face (As in Fig 2) on both Right
Carriage Way (RCW) and Left Carriage way (LCW) of the road between chainages 195.0m
and 390.0m (see Fig 4). On both sides of the road cracks have appeared just at the end of the
GI mesh reinforcements of the RE wall which is at a distance of 5.0 to 8.5 m from the
Reinforced earth wall panels. The maximum width of crack was mostly 1 to 3mm and at one
place it was 9mm. Immediately, the longitudinal cracks were filled up and rolled. At that
time, some cavities have also been observed as shown in fig 4. The asphalt filling up which
had been done on the observed cracks showed partial success in that the crack on the left
carriage way from chainage 195 to 295 did not reappear. But the one on the right carriage
way from 295 to 390 did reappear. In addition, on the 19th of October, 2007 new hair line
cracks were observed on the left carriage way and the right carriage way as indicated in the
Figure 2. These hair line cracks were seen to be confined to the top bituminous concrete
layer, however extending upto Ch 405 on LCW and Ch +205 on RCW.
The observation on the verticality of the end panels were also recorded from
26.09.2007 to 15.10.2007. The recorded movements during the above mentioned period
remained fairly stable and showed no evidence of perceptible distress propagation. The
movements mostly are in the range of 5 to 30 mm which are generally expected in RE walls
for mobilization of shear forces on the soil-reinforcement interfaces.
In order to get some additional information on the compacted state of the backfill,
geotechnical investigations at three more locations in the ROB section of the carriage ways
were suggested and carried out.
Fig 1. View of RE wall which is constructed for ROB

Fig 2. Longitudinal cracks on the carriage way

Fig 3. Typical cross section of the ROB Fig 4. Cavity observed on RCW
Fig 5. Longitudinal cracks as seen on both RCW and LCW of ROB –II
Most of the boreholes in the reinforced earth wall embankment section show a good
penetration resistance with depth but only in one o the borehole (which is close to the
longitudinal crack on the right carriage way) shows some evidence of loose pockets and with
very low penetration resistance (at a depth of 4.5m from the road surface) and the presence of
boulders (between depths of 5 to 8m from the road surface). The results of ground penetrating
radar survey indicate the presence of loose pockets and cavities in the top layers of
compacted fill.
Design of RE wall was correct and the Factor of Safety (FS) for the global stability is
high, which rules out the global stability concern (as expressed by the designer and the
owner) for the ROB-II section with distress. GPR survey has also not indicated any formation
of slip circle type of failure passing through the embankment section. The design calculations
for the retained earth wall consisted of internal stability checks (reinforcement rupture and
reinforcement pullout) external stability checks (for overturning and sliding) and they are
found to be in order. It is unlikely that the longitudinal cracks that appeared might have been
caused due to these internal and external stability issues. Some settlement might have been
caused due to poor foundation (as observed in BH-10 with values of N=10 corresponding to a
depth of 8.5m from the road level). However, this may not be very significant as we did not
observe any level difference across the crack.
Selection of borrow materials and Compaction control was found to be the cause of
distress and cracks. Most of the cracks are confined up to the bottom of the granular sub base
layer. Granular sub base layer seems to be most affected layer. From the detailed
investigations (forensic), the cause of longitudinal cracks are attributed to:
a) From the median area or the voids that might have been present in the top layers of
pavement (BC, DBM and BM), rain water got seeped into the granular sub base layer and
further onto subgrade which might have caused differential settlement of soil layers at the
junction of reinforcement section and un-reinforced section of the embankment and resulted
in the form of longitudinal cracks on the pavement. This is evident from the differential
settlement in various soil layers as observed in GPR survey.
b) These longitudinal cracks are mainly confined in the top 1.5m from the road
surface. However, there are indications that these cracks might have extended up to 4 to 5m
into the compacted soil (as indicated in one of the boreholes) during the heavy rainy days and
subsequent seepage at some locations, where it had small boulders causing a high
permeability zone. At some locations, densities of the backfill material obtained in the
embankment section did not meet the desired values.
c) Selection of granular sub base material and improper drainage through the granular
sub base layer.
Cement grouting is recommended for the compacted backfill and has been done all
along the longitudinal cracked zones at regular intervals of 3m c/c. c. The median portion was
recommended to be made impermeable by providing LDPE layer along with PCC of about 50
mm thick all along the length of the median to prevent water entering the pavement.
Premixed seal coat comprising of a fine aggregate premixed with bituminous binder was also
recommended for sealing the voids in a bituminous surface laid to the specified levels, grade
and camber for the full width of the carriage way.
Case Study 9

EXPERIMENTAL AND NUMERICAL TECHNIQUES FOR


ANALYSIS OF SOIL NAILED STRUCTURES

K. Rajagopal1 and T. Rambabu2


1
Professor, Dept. of Civil Engineering, IIT Madras, Chennai 600036, e-mail: gopalkr@iitm.ac.in
2 T. Rambabu, M/s Utraco-Ryobi Ground Engineering Pvt. Ltd., Chennai-87, e-mail:
utracoryobi@yahoo.com

ABSTRACT

This paper discusses the comparison in the response of soil nailed retaining walls and slopes
obtained by different methods i.e. experimental, limit-equilibrium and finite element analysis. The
paper describes some laboratory tests in which 1.75 m high soil nailed retaining walls were
subjected to failure by applying uniform surcharge pressure. Their response was studied by three
different methods and comparisons were made. Later, the paper describes the stabilization of deep
excavations at a construction site in Chennai using pre-stressed nailing technique. The construction
site is surrounded by fully-developed residential areas. Any slope failure at the site will result in loss
of foundation support to the adjacent structures which may lead to many legal problems. After
carefully considering various methods for stabilizing the excavated faces, it was decided to use pre-
stressed grouted nails for stabilizing the slope faces. This paper discusses the results from different
parametric studies and the results obtained. The implications of different possible alternatives are
discussed in the paper.

1. INTRODUCTION

The use of soil nailing is a new and innovative technique that can be used even in very
difficult site conditions. It involves in driving in steel rods into the soil beyond the rupture plane to
bind the active soil mass to the interior mass. This technique falls under the category of earth
reinforcement and consists of passive bars, named nails, which are generally used to retain
excavations, and to stabilize unstable slopes. This technique has been widely used in different types
of projects, e.g. soil slopes (Nicholson and Boley 1985, Bruce and Jewell 1987), retaining structures
(Stocker et al. 1979, Shen et al. 1981, Guillox and Schlosser 1982), tunneling and other civil and
industrial projects. Babu and Singh (2009) have described some latest techniques for the design of
nailed soil walls and their verification by finite element based numerical methods. Gosavi et al.
(2009) have described some analysis method for nailed soil cuts and their verification by laboratory
model tests. The design and construction aspects of soil nailing for stabilizing of slopes and
excavations was presented by Elias and Juran (1991).
In this paper, the response of the nailed soil walls is studied through different methods, viz.
limit equilibrium method based on bi-linear wedge analysis, laboratory model tests and finite
element analyses. The results are analysed for comparing the accuracy of different methods and the
implications for the field performance. Later, one case study of a real field case is described in the
paper.

3
2. BI-LINEAR WEDGE ANALYSIS METHOD

This limit equilibrium analysis method assumes the failure surface to consist of two surfaces
intersecting within the soil mass. The reinforcement elements are assumed to develop only tensile
forces and the effect of the bending stiffness is ignored in the analysis. The length of the
reinforcement element beyond the rupture wedge is defined as the embedment length and the pullout
resistance is calculated based on this length of reinforcement only. These tensile forces are divided
into the components parallel and perpendicular to the assumed rupture plane. The normal force and
tangential component in each reinforcement element crossing the potential surface are added to the
resisting forces mobilised in the soil when determining the factor of safety of the entire mass. The
factor of safety of the soil mass is computed as the ratio between the available and the required
forces.
The above calculations are performed repeatedly for different failure surfaces using a computer
programme developed in C++ language. The computer program is interactive and reads the input
data through the console and the results are saved in a file. The output from the program consists of
the factors of safety at each and every rupture plane considered in the analysis. The program also
gives the minimum factor of safety for the given configuration.

3. EXPERIMENTAL PROGRAMME
All the retaining wall tests were performed in a specially constructed retaining wall test facility
of 2000 mm long, 740 mm wide and 2000 mm height. This facility consists of a mass concrete base
slab of 100 mm thickness and two longitudinal side walls that are anchored into the base slab. The
side walls of the facility were constructed with hollow concrete blocks of size 400 mm × 200 mm ×
200 mm. These blocks were anchored into the foundation soil through tie rods of 12.5 mm extending
through the concrete floor. During the tests, steel channels (ISMC 200) were fixed across the two
side walls to create a self-reacting rigid frame consisting of the base slab, side walls and the steel
channels as illustrated in Figure 1.

The test set up is constructed 1.0 m below ground level and 1.0 m above the ground level for ease of
handling large quantities of soil during the tests and also from the stability consideration of the two
side walls. A free space of 500 mm was left in front of these walls for construction of the test wall
and for access to provide necessary instrumentation.
The backfill of the model retaining walls was subjected to uniform surcharge pressure by
inflating an air bag placed between the backfill surface and the cross-channels connected between
the two side walls. The top of the air bag was covered with a non-woven geotextile and a 25 mm
thick plywood plank was placed between the geotextile and the steel channels to distribute the load
uniformly on the soil. This air bag was specially made of double faced neoprene coated nylon
reinforced fabric.
An air compressor capable of supplying air at a maximum pressure of 1000 kPa was run
continuously during the test to supply pressurised air. The air is routed through a non-return valve.
The regulator was adjusted to control the pressure of the air in the air bag to apply the pressure in
small increments.

44 FGE 2009, Bangalore


The lateral displacements of the wall facing were measured through LVDTs that were connected
to display unit to directly read the displacements. The vertical and lateral soil pressures were
measured using strain gauge type pressure cells of 300 kPa capacity that have an accuracy of 0.1
kPa. The loads developed in the soil nails at the wall facing were measured through pre-calibrated
load rings fabricated within the geotechnical engineering laboratory of IIT Madras. These load rings
were made of 45 mm wide steel pipes (85 mm internal diameter and 2.5 mm wall thickness) fixed
with resistance type strain gauges on diametrically opposite sides. Strain gauges were also fixed at
four locations on the soil nails to measure the strains along the length of the nails. These strain
gauges were protected from damage during the installation by placing them in side a thin plastic
membrane wound around the soil nail.
3.1 Construction of the Model Retaining Walls

The soil used in the present investigation was a dry river sand with coefficient of uniformity (Cu) of
1.96, coefficient of curvature (Cc) of 1.05 and effective size of particle (D10) of 0.25 mm. The soil
can be classified as poorly graded sand with the letter symbol SP according to the Indian soil
classification system. The minimum and maximum dry unit weights of the soil are 14.97 and 18.5
kN/m3. The direct shear tests have given the peak friction angle of the soil as 41°.

reaction frame
wooden plank and
nonwoven geotextiles
0.2 m inflatable air pressure bag facing panel

LVDT

2.0 m load ring


sand
G.L

pressure
cells

strain
gauge

2.0 m

Figure 1. Set up for model retaining wall tests

The model walls were then built in the retaining wall test facility with the same configuration to
evaluate the accuracy of the limit equilibrium method of analysis of this type of earth structures. All

45 FGE 2009, Bangalore


the tests were performed with a model wall height of 1800 mm. This has left a clear gap of 200 mm
to accommodate the air bag between the surface of the backfill and the cross channels to apply the
uniform surcharge pressure.
The front facing was made of individual facing blocks (8 rows and 3 columns). The soil nails used in
these test walls consisted of 12.5 mm diameter plain mild steel rods. The length of the soil nails was
varied from 600 mm to 1600 mm. The number of soil nails also was varied from 9 to 24.

All the tests were conducted at a relative density of the backfill soil of 70%, which corresponds to a
peak friction angle of 41°. The backfill sand was placed in the tank in layers of 225 mm thick. For,
each lift, the amount of soil needed to produce the desired dry density (at relative density of 70%)
was weighed out and placed loosely in the tank. Each layer was then compacted manually using a
drop weight of 65 N falling on a wooden plank of plan size 250 mm × 600 mm from a height of 350
mm. This procedure of soil compaction was arrived at by trials to achieve a uniform soil density
throughout the test tank. The compacted density of the sand was monitored by collecting samples in
steel cups of 90 cm3 volume placed at different depths in the sand bed. The difference in the density
of the soil at different locations was found to be less than 1%.

The front facing units were temporarily supported laterally using cement blocks. After the backfill
soil was placed to the full height of the model wall, the bricks in each layer were successively
removed after the nails in that layer were installed. The nails were installed by driving. A template
was used to guide the inclined nails at 10° below the horizontal. After the nails were driven in, the
facing panels were connected to the nails through pre-calibrated load rings as shown in Figure 1.
This method of construction of the test walls corresponds to the top-down construction technique
adopted for slope stabilisation works in the field.

3.2 Test Procedure


The constructed walls were subjected to failure by applying uniform surcharge pressure on the
backfill. The surcharge pressure was applied in small increments until the failure of the wall by
controlling the flow of air through the non-return valve. The applied surcharge pressure was
monitored through the vertical pressures read from the pressure cell fixed at the back end of the wall.
Each increment of surcharge was kept constant until the deformations under that particular load
increment have reached a steady state.

As the surcharge pressure was increased, the lateral deformations and the load in the load rings were
observed to increase. As the failure pressure was approached, the load in the load ring connected to
the upper row of the soil nails became constant and had suddenly decreased at some surcharge
indicating the pullout failure of the nails at that elevation. The lateral earth pressures measured
behind the facing panels also had shown a similar trend. At that stage, the walls were deemed to
have failed due to the pullout of the soil nails. The top row of nails became so loose that they could
be pulled out by hand. The test was stopped at that stage and the entire test set up was dismantled.
Some of the initial tests were repeated again with the same nail configuration to ascertain the
repeatability of the test results. Very small differences were obtained in the results obtained from
such tests. The tests were repeated with different nail configurations to examine the influence of the
length, inclination and spacing of nails.

46 FGE 2009, Bangalore


4. FINITE ELEMENT ANALYSIS OF TEST WALLS
All the finite element analyses in this research work were carried out using the general purpose
finite element program GEOFEM developed at IIT Madras, Rajagopal (1995). The behaviour of
retaining walls supported by soil nails was analysed by two dimensional finite elemental analysis
using plane strain idealisation by smearing the effect of soil nails over 1 m width perpendicular
to the plane of the analysis. The influence of some of the important parameters which influence
the behaviour of these walls was studied through parametric studies. In order to compare the
finite element results with those from experimental work, the exact dimensions used in the
laboratory experiments were employed in these analyses.
The finite element mesh used for the analysis of nailed walls on sand consisted of eight-node
quadrilateral elements to represent the soil and the front facing units, 3-node bar elements to
represent the soil nails and six-node joint elements of zero thickness to represent the interfaces
between the front facing and soil and between the nails and soil. The elastic-plastic behaviour
and the failure of the soil were simulated by using Mohr-Coulomb yield criterion.
The data from the pullout tests was used to estimate the interface friction angle between the soil
nails and the soil. The interface elements consist of two stiffness values, one in the shear
direction and the other in the normal direction. The shear stiffness was given a large value to
maintain the compatibility between the nail and soil before the failure and was set to a small
after the failure to allow for relative deformation. The stiffness in the normal direction was set to
a large value to maintain the contact in the normal direction at all stages of the analysis. The
interface element is not allowed to develop tensile stresses in the normal direction.
The construction of the wall was simulated by placing the soil in the mesh and applying the self
weight as gravity forces with a Ko of 0.50. During this stage, the facing panels were laterally
supported by external props. The resulting deformations and strains were set to zero at the end of
this stage. During the next stage of analysis, the front props were removed and uniform pressure
was applied on the backfill surface in small increments to simulate the pressure loading in the
laboratory tests. The analysis was iterated to reduce the out-of-balance forces at each load step to
less than 0.5%. The failure of the wall was defined at the stage at which the out-of-balance force
increases to a high value and divergence takes place in the solution. The finite element analyses
were performed with nail lengths of 0.6, 0.8 and 1.2 m, and with nail inclinations of 0°, 10° and
20°.

5. COMPARISON OF DIFFERENT RESULTS

Typical lateral deformations along the height of the wall at different surcharge pressures are
shown in Figure 2 for 1.2 m long horizontal nail. The forces measured in the load rings were
found to increase with increase in surcharge pressures, Figure 3. At the ultimate surcharge
pressures the top row of facing panels deformed excessively accompanied by the sudden drop of
force in the load ring. At that stage, the top row of nails became so loose that they could be
pulled out by hand easily. This behaviour shows that the top row of soil nails had failed in
pullout. The lateral pressures measured behind the facing units also showed a linear increase
with surcharge pressure until failure and showed a sudden decrease at failure. In general, the
walls with inclined nails failed at higher surcharge pressures because of the higher pullout
capacity of these nails. In all the walls, the lateral wall displacements have decreased towards the

47 FGE 2009, Bangalore


toe of the walls. This is because the length of embedded portion (where the nail resistance is
developed) beyond the rupture plane increases towards the toe of the wall. As the deformations
are lower towards the toe of the wall, the relative deformations between the soil and nail are also
lower and hence the mobilised force in the nail is lower in the bottom row of nails as shown by
the results in Figure 3.

2000

1500 applied pressure (kPa)


elevation (mm)

10
1000 20

30

500 40

50

60
0

0 5 10 15 20
lateral displacement (mm)
Figure 2 Lateral displacements at different surcharge pressures (1.2 m long horizontal nail)

2000 elevation (mm)

1687.5

1500 1237.5
force in load rings (N)

1012.5
562.5
1000 112.5

500

0 20 40 60 80
surcharge pressure (kPa)

Figure 3. Load ring forces in soil nailed wall (1.2 m horizontal nail)
The measured lateral earth pressures behind the facings panels are shown in Figure 4. It may be
observed that the lateral earth pressures have also increased with the applied surcharge pressure until

48 FGE 2009, Bangalore


the collapse state of the model walls. When the failure pressure was approached, the facing panels
had excessive lateral outward deformations leading to a decrease in the lateral earth pressures in the
top most pressure cell. The reduction of earth pressures was taken as one indication of the failure of
the walls.

The comparison of the internal rupture plane measured in the laboratory tests and those predicted by
the finite element analysis and the bilinear wedge mechanism is shown in Figure 4. The points with
the peak strain in the nails were joined together to obtain the possible rupture plane in both FEA and
experiments while the surface with the least factor of safety is taken as the rupture plane in the case
of limit equilibrium analysis. It could be seen that the rupture plane predicted by the bilinear wedge
method is widely different from the measured surface. This may be another reason for the low
ultimate pressures estimated by the limit equilibrium method.

The responses predicted by the finite element analyses with 0.6 m long nails placed at three different
inclinations are shown in Figure 5. These responses closely matched those measured in the
laboratory tests. It could be observed that the deformations decrease (at the same surcharge pressure)
with increase in the nail inclination as also noted from the experimental results. This result clearly
shows that the finite element model is able to replicate the improvement in the performance of nailed
soil retaining walls with the inclination of the nails.

The comparison between the measured ultimate surcharge pressures on the soil nailed model walls
and the corresponding pressures predicted by the limit equilibrium and the finite element methods is
presented in Table 1. It can be observed that the pressures estimated from the limit equilibrium
method are lower than the measured values in all the cases while the values predicted by the finite
element analyses are reasonably close to the measured values. One of the reasons for the discrepancy
between the predicted and the estimated results could be the effect of sidewall friction.
2.0

1.5
elevation (m)

1.0

fem
0.5
exp

design method
0.0
0.50
0.0 1.0
points of maximum strain locations over the length of the nail (L=1.2 m)
Figure 4. Internal rupture surfaces from different methods

49 FGE 2009, Bangalore


Table 1. Comparison of ultimate surcharge pressures on retaining wall

Configuration of soil Limit Laboratory FE analysis (kPa)


nails Equilibrium model tests (kPa)
analysis (kPa)
1.6 m (horizontal) 100 130 125
1.2 m (horizontal) 50 60 55
1.2 m (10° inclination) 65 80 85
0.8 m (horizontal) 35 45 40
0.8 m (10° inclination) 45 60 55
0.6 m (horizontal) 25 35 30
0.6 m (10° inclination) 35 45 45

80

60
pressure (kPa)

40

20o inclination
20
10o inclination
horizontal
0
10 30
0 20 40
displacement (mm)
Figure 5. FE predicted retaining wall behaviour with different nail inclinations

6. CASE STUDY OF A PRE-STRESSED NAILED EXCAVATION

At one construction site in North Chennai, it is required to make deep excavations for construction
of two basement floors of a shopping complex. The proposed construction site is surrounded by one
to two storey residential and commercial buildings. Due to heavy rain fall in the area, the surface
rain water collects into the construction area. Hence, dewatering system had to be installed to keep
the construction area dry.

50 FGE 2009, Bangalore


In general, the soil profile at the site consists of four distinct layers. The top one meter or so from the
ground level is loose silty reddish brown fine to medium sand. This upper soil layer is followed by a
three meter thick medium dense to dense brown silty sand of fine to medium grade with average
SPT values of 20. This second layer is underlain by a firm to medium dense brown silty sand of 3 m
thickness with average SPT values of around 10. The soil layer below consists of silty sand of
approximate thickness 6 m thick extending up to 13 m. This layer is a medium dense to dense and is
brownish in colour with an average SPT of 30. This layer is followed by dense to very dense, hard
silty sand with average SPT values in excess of 50.

The ground water table at the site is around 3 to 4 m from the ground level as per the local
information. However due to effective dewatering at the site, the water table is maintained as a depth
lower than 8 to 9m below the ground level i.e. at one to two meters below the excavated ground
surface. Assumed profiles and proposed slope is indicated in Figure 6.

Figure 6. Schematic sketch of the proposed excavation and ground support

The factor of safety of the slope without any stabilization comes to 0.63 which is unacceptable. The
factor of safety of the slope was increased to 1.20 by providing pre-stressed soil nails as illustrated
above. This factor of safety was adequate as the excavated slopes are only of temporary nature. The
details of the soil nails provided at the site are shown in Table 2. The factors of safety obtained from
finite element analyses are slightly larger than those obtained from these analyses.

Table 2. Details of the pre-stressed soil nails at different depths


Row name Location of Force in Free length Required Total Provided
soil nails kN/m (m) fixed length required length (m)
(m) length (m)
A -1.0m 10 3.0 1.5 4.5 6.0
B -2.5m 20 2.0 3.0 5.0 6.0
C -4.0m 25 1.5 3.75 5.25 6.0
D -5.5m 30 1.0 4.5 5.5 6.0
E -7.0m 40 0.0 4.2 4.2 6.0

51 FGE 2009, Bangalore


The slope stability analysis was performed using a limit equilibrium based slope stability software
“DC slope” from DC Foundations, Germany. A uniform surcharge of 20 kPa was assumed at the
ground level to take account of the loads from moving machinery etc. The results from these
analyses were independently verified from finite element analysis using the program PLAXIS-2d.
Typical result from one analysis of limit equilibrium analysis is shown in Figure 7.

Figure 7. Typical result from program DC slope

The surface was treated by spraying 50 mm thick concrete to prevent surface erosion. Some typical
pictures from the site are shown below to illustrate the nail pre-tensioning and the beam
arrangement.

Figure 8. General view of the construction site

52 FGE 2009, Bangalore


Figure 9. Pre-tensioning of the soil nail after the grouting

7. CONCLUSIONS
Any design of soil structure is complicated due to several unknown parameters. Hence empirical
methods are employed to come out with safe designs. Invariably, these are based on earlier
experience or on some limit equilibrium analysis. The current study has shown that the stability of
the nailed wall or excavation is very much influenced by the length and inclination of the soil nails
and pre-tension effects.. Some of the major conclusions that can be drawn from this study are as
follows:

• The design based on the bilinear wedge failure mechanism is always on the conservative side.
• The length and inclination of the soil nail plays significant role in the overall behaviour of nailed
soil wall.
• The nail should project sufficiently beyond the rupture plane to develop significant pullout
capacity and also to intercept the potential rupture planes.
• The internal rupture plane in the soil nailed walls can be idealised as a bilinear surface and can
be obtained by drawing a vertical line at 0.3H behind the facing and an inclined line through the
toe at (45+φ/2)°.
• The influence of the nail inclination is very well represented by the finite element models.

ACKNOWLEDGEMENTS
The financial support for this work from the Ministry of Human Resources and Development,
Government of India, New Delhi is gratefully acknowledged. Some part of the work reported in this
paper was performed by Mr. G.V. Ramesh as part of his thesis work for Master of Science degree by
research at IIT Madras.

RFERENCES
Babu, G.L.S. and Singh, Vikas Pratap (2009) Appraisal of Soil Nailing Design, Indian Geotechnical
Journal, Vol. 39, No. 1, 81-95.
Bruce, D.A. and R.A. Jewell (1987) Soil nailing: application and practice – part 2. Ground
Engineering, The journal of British Geotechnical society, Vol. 20, No.1, 21- 28.

53 FGE 2009, Bangalore


Elias, V. and Juran, I. (1991) Soil Nailing for Stabilisation of Highway Slopes and Excavations,
FHWA-RD-89-198, Federal Highway Administration, McLean, Virginia, USA.
Gosavi, M., Swami Saran and Mittal, S. (2009) Model Tests for validation of analysis of nailed cuts,
Indian Geotechnical Journal, Vol. 39, No. 1, 96-115.
Guilloox, A. and G.Notte (1982) Experiences on a Retaining structures by Nailing. Proceedings of
8th European Conference on soil Mechanics and Foundation Engineering, Helsinki.
Nicholson, P.J. and D.L. Boley (1985) Soil nailing supports excavation. Journal of Geotechnincal
Engineering, ASCE, 111, No. 4. 45-47.
Rajagopal, K. (1998) Users Manual for the Finite Element Program GEOFEM, Geotechnical
Engineering Division, Department of Civil Engineering, IIT Madras.
Shen, C.K., L.R. Herrman, K.M. Romstad, S.Bang, Y.S. Kim and J.S. De Natale (1981) An In Situ
Earth reinforcement Lateral support system. Department of Civil Engineering, Univ. of
California, Davis, Report No. 81-03, U.S. National Science Foundation Report No. NSF/CEE-
81059, March, 130 pp.
Stocker, M.F. and G.Riedinger (1990) The Bearing Behaviour of Nailed Retaining Structures,
Design and performance of Earth retaining Structures. ASCE Geotechnical Special publications,
No.25, New York, 612-628.

54 FGE 2009, Bangalore


Case Study 10

THE FAILURE ANALYSIS AND REHABILITATION OF


DAMAGED RE WALL

Murthy, B. R. S.
Retd. Professor, IISc, Bangalore, India, 560 012, murthybrs@gmail.com

1. BACK GROUND

A Geo-grid Reinforced earth retaining wall for a fly-over along the National High
way No 4 was constructed a few years back. The wall height is varying from 2 to 13
meters. At a location of about 7 meter height of the wall there was snapping of
connection between facing element and the geo-grid reinforcement. The authorities
requested the author to visit the site and suggest a suitable permanent solution for
rehabilitation of the failed zone. Also it was requested to find out the reason for failure.
The site was inspected along with all concerned. There was not much information
available with the project authorities regarding the quality control or the designs not even
a sketch of the actual implementation.

2. SITE CONDITION

The failure had taken place over a length of about 10 meters. The failure was
essentially in the form of collapse of the panels due to snapping of the connecting plastic
cleats/clamps, perhaps initially of the upper layers. The number of geo-grid
reinforcement layers provided was found to be reasonable. There was variation in the
grade of geo-grid increasing with depth. The soil appeared to be in good condition as
required from an RE wall point of view. In fact the failure plane had not gone beyond a
meter from the wall face which was the boundary of the drainage zone of granular
material. The soil behind the failed zone was standing near vertical indicating the
sufficiency of the reinforcement layers and good quality of compaction. It was reported
that the PHI value for the soil was found to be more than 34 degrees.

In such a situation it was difficult to evaluate the reasons for failure. A few
theories were proposed by engineers and local people for the cause of failure. One
particular theory was that a heavily loaded truck rammed against the crash barrier. Really
no such sign was visible. All the theories were discounted and it was indicative that most
of the reasons were only to divert the attention and hence search started for a realistic
one.

3. INVESTIGATION

A closer examination of the detailing and the site condition of failure indicated
that the failure has initiated from snapping of upper layers of geo-grid/ connections. This
was indicative from large elongations of the upper layers. Further the crash barrier and

11
the friction slab are rigid and continuous where as the pavement beyond the friction slab
is flexible. The integrity between them cannot be established by a thin bitumen pavement
on the top. The wheel loads will transmit the load and vibrations differently in the two
systems. In rigid pavement portion the vibrations are transmitted down into the granular
medium whereas the same is damped/ dissipated in the flexible pavement. The
transmitted vibration will tend to compact the granular soil resulting in the loss of contact
of the friction slab and soil. This will in turn makes the friction slab to transmit the
reaction due to the wheel load to the facing element. As the facing element assembly is
not designed to transmit the axial loads the reaction will tend the assembly of facing
element to BUCKLE causing the snapping of the cleats/ joints between the facing and the
geo-grid. This is the primary reason for the failure. Figures A and B indicate the
schematic representation of the wall and details of the analysis of failure.

4. REHABILITATION SCHEME

To reconstruct the wall, the following scheme was proposed and implemented and the
same has been presented in the drawings. The essential steps were as follows.

1. Support the failed surface with sand bags and provide support to the exposed
crash barrier using wall form jacks.
2. Providing a suitable staging, remove the sand bags to a depth of 1.65 meters. Trim
the failure plane to remove loose materials.
3. Drive 6 meter long 20 TOR bars at a spacing of 550 mm c/c (three rows) in both
directions.
4. Fasten 50x50x2.8 mm MS grid reinforcement vertically along a defined line
which can facilitate fixing of the permanent RE wall panels and horizontally
along the lowest layer of nails.
5. Fill the void space beyond the MS grid with 50 mm aggregates as densely as
possible.
6. Provide 50mm thick shot-crete on the weld mesh as per relevant IS code.
7. Repeat the steps 2, 3, 4, 5 and 6 to reach the full depth.
8. Install the RCC pre-cast facing panel for one row and fasten the same suitably
with the projected nails.
9. Back fill the gap between the shot-creted surface and the Pre-cast panel with self
compacting coarse sand.
10. Repeat the steps 8 and 9 to reach the top level.

12 FGE 2009, Bangalore


CRASH BARRIER WITH
FLEXIBLE PAVEMENT
FRICTION SLAB (RIGID)
BITUMIN WEARING SURFACE

GEOGRID LAYERS
RCC FACING

GRANNULAR DRAINAGE MEDIUM about 1 meter thick

FIGURE A

13 FGE 2009, Bangalore


FLEXIBLE PAVEMENT
CRASH BARRIER WITH
FRICTION SLAB (RIGID)
BITUMIN WEARING SURFACE

REACTIONS

GEOGRID LAYERS
RCC FACING

GRANNULAR DRAINAGE MEDIUM about 1 meter thick

FIGURE B

14 FGE 2009, Bangalore


ROAD SURFACE

SAND BAGS TO A SLOPE OF 1:1

WALL FORM JACKS TO SUPPORT


FAILED SOIL SURFACE THE CRASH BARRIER

Figure1. The failed reach is protected using sand bags to a slope


of 1:1 and supporting the crash barrier with wall form pipe jacks

15 FGE 2009, Bangalore


20 TOR BARS OF LENGTH 0.7H DRIVEN
AT SPACING OF 550 C/C IN BOTH THE
DIRECTIONS

Figure 2. The sand bags are removed to a depth of 1.65 meters and three rows of nails are
driven in to the back fill material.

16 FGE 2009, Bangalore


50x50x2.8 mm MS WELD MESH FASTENED
TO THE NAILS AND VOID SPACE BEHIND
FILLED CAREFULLY WITH AGGREGATES

50 mm THICK SHOTCRETE
AS PER IS 9812

Figure 3. MS weld mesh of size 50x50x2.8 mm is fastened to the nails and at the
top of the sand bags. The gap between the soil and the weld mesh is carefully packed
with aggregates of suitable size. 50mm thick shot-crete is provided on the weld mesh.
Repeat the procedure of removal of sand bags, driving of nails, fastening of weld
mesh, backfilling with aggregate and shot-creting to reach the full depth of the wall as
indicated in Figures 4 and 5.

17 FGE 2009, Bangalore


Figure 4

18 FGE 2009, Bangalore


Figure 5

19 FGE 2009, Bangalore


PRECAST RCC FACING PANELS
ASSEMBLED FROM BOTTOM

THE FACING PANEL IS CONNECTED


TO THE NAIL SYSTEM AND THE GAP
IS BACK FILLED WITH SAND

Figure 6 The pre-cast RCC facing panels as per the original designs shall be
assembled along the original alignment starting from the ground level. The panels
are connected to the nails using suitable MS cleats and fastening system. The gap
between the shot-crete surface and the RCC pre-cast panel shall be backfilled with
good self compacting material like sand. The above sequence shall be followed till
the full height is covered.

20 FGE 2009, Bangalore


INTERACTION BETWEEN APPROACH EMBANKMENT ON SOFT GROUND
AND ROAD OVER BRIDGE FOUNDATIONS – A CASE STUDY

M.R.Madhav1, B.Vidyaranya2 & G.B. Rajendra Prasad3


1
Professor Emeritus, 2Research Scholar & 3Asst. Executive Engineer
1
JNTU College of Engineering, 2Osmania University, Hyderabad & 3R&B Amalapuram Division
madhavmr@gmail.com & vidyaranya_b@yahoo.co.in

ABSTRACT
The paper presents an interesting instance of approach embankment construction affecting the already
constructed Road Over Bridge (ROB) and its foundations. The ground consists of very thick layer of very soft to
soft soil extending to a depth of nearly 15 to 16 m underlain by dense sands. As such, the bridge was constructed
on well foundations sunk into strong and stiff underlying strata. In spite of this fact, the bridge suffered
extensive damage due to the construction of nearly 4.5 m high embankment. The lateral squeezing of the soil is
believed to be the cause of the failure/damage to bridge foundations. The case history highlights the need for an
integrated approach for the design and construction of a bridge and its approaches.

1. INTRODUCTION lined with concrete panels to support the surrounding


soil. Even the panels moved in to the canal by about
The design of structures on or over soft soils needs half a meter.
special care and concern. The interaction between
structures adjacent to each other is an important The plan of ROB is shown in Fig. 1. The ROB
aspect that should not be ignored. The paper presents consists of a single span of 16.0 m long RCC ‘T’
a case study of a Road over Bridge which was beam slab. The abutments were founded on well
constructed prior to the construction of the approach foundations two under each. The wells, 4.50 m in
embankments. The embankment was constructed diameter, were plugged at depth between 18.30 m to
subsequently. However, the effect of possible 18.60 m below the ground level. Wing and return
squeezing of soft soil in the longitudinal direction was walls (Fig. 2) were made in Mass CC M-20 grade up
not considered leading to a major geotechnical to ground level and Mass CC M-30 grade above
problem. The paper presents the details of the bridge, ground level with 40 mm grade hard granite stone
damage caused to it and the results of soil aggregate.
investigations at the site.

2. DESCRIPTION OF THE SITE

The site under investigation is on a bypass road


constructed to facilitate free movement of traffic over
a railway line and avoid the use of a level crossing.
The proposed road connecting two towns passes over
an irrigation canal and a railway track approximately
400 meters apart. The irrigation canal runs parallel to
the railway track. An embankment of height 6.5 m
was to be constructed to connect the ROB and the
bridge crossing the canal. The site is surrounded by Figure 1 Plan of the Road over Bridge
paddy fields all-round. The ground water table was at
a depth of about 1.5 m below the existing ground level
at the time of investigation but is known to rise close
to GL following monsoon rains.

A road runs parallel to the canal on the side away


from the ROB. The bridge structure over the canal
was founded on a raft 15 m wide. The bridge settled
by about 0.5 to 1.0 m and the piers tilted leading to
severe cracking of the deck, piers and the abutments,
the bottom of the canal heaved due to the construction
of the approach embankment to the bridge on the Fig 2 Section of Return Wall
canal. The sides of canal supporting the road were

6
Foundation provided up to 4.0m below ground level block. The crack had widened to a width of about 25
consisted of 2.0 m sand filling well compacted in mm since then. The top of the abutment moved by
layers of 300 mm and RCC raft foundation of 0.50 m about 15 to 20 mm towards the backfill. Earthwork on
followed by 1.50 m sub-structure with 1:1 slope on the rear side of the abutment No. 2 was carried out to
one side and straight face on other side. Figs 1 a height of 4.4 m from the well cap level. Horizontal
through 4 show the plan, sectional details of the cracks 25 mm wide were noticed at 3.60 m 4.80 m
abutments and the return walls in detail. and 6.0 m from the top of the slab on wing walls
No.4. Figs 5 through 10 shows the tilt generated in the
wing walls due to movements caused by embankment
construction and Table 1 summarize the movements
reported.

Figure 5 Sectional Elevation of Abutment with wing


walls

Figure 3 Section of Abutment

Figure 6 Sectional Elevation of Wing Wall 1

Figure 4 Section of Wing Wall

The embankment was constructed using locally


available soil. A layer of geogrid was laid at the
bottom of the fill. The soil used for the embankment
appeared to be heavy clay, possibly with slightly
higher percentage of fine sand and silt. Sandy soils
were used for the central core while clay was used for
the remaining portion of the backfill.

The wing walls got settled and heavy cracks were


generated. The construction of the bridge was
completed in Jan’ 01 but the construction of the
embankment was delayed till year 2005. The work
was recommenced with work on Abutment No.1 side
to a height 6 m from the top of the well cap and to 1.2 Figure 7 Sectional Elevation of Wing Wall 2
m from the bottom of the bed block. A crack was
observed at a level of 1.20 m below the top of bed

7 FGE 2009, Bangalore


Table 1 Details of Movements of Abutments

Horizontal
Settlement Lean
Loca Movements
at top Toward
tion Well Bed
Fill
Cap Block
W1 270 340 120 20
Figure 8 Sectional elevation of Abutment with Wing Butting
Walls W2 400 290 100 against
abutment
W3 Nil 75 75 40
No
W4 120 20 Nil
formation

3. SITE INVESTIGATIONS

The site was investigated by several agencies prior to


the construction of the bridge. The soft clay layer was
reported to extend up to a depth of 18.5 m. The water
table was close to GL. The site conditions needed
fresh investigations to evaluate the soil properties and
to examine the reasons for the movements. Fresh
boreholes were advanced to a depth a 25- 30 m at
three locations close to the ROB structure.

The soil profile indicated black soft clay up to 15 to


18 m depth. Medium to fine sand underlies the soft
clay. SPT ‘N’ values were negligibly low and some
times equal to zero. The variations of the SPT values
with depth are shown in Fig. 12. SPT N values were
slightly high (4 to 5) near the top (2-3 m) possibly due
to desiccation but reduce to 2 to 3 until a depth of 15
Figure 9 Sectional Elevation of Wing Wall 3 in BH.1, 16 m in BH.2 and 18.0 m in BH.3. N values
increased to about 20 to 27 at 18 to 20 m depth range.
N values in the lower sand stratum were greater than
40.

The liquid limit values ranged from 72 to 85 while the


plasticity index values were of order 45 to 55. The
natural moisture content increases with depth over the
top 4.0 to 5.0 m and lie in the range of 75 to 90. The
liquidity index values are greater than 1.0 at few
locations. The shrinkage limit values measured over
selected depths are 8.0 for depths less than 11.0 m and
about 10 at a depth of 16.0 m. Low shrinkage limit
and high plasticity index attest to the fact that the soil
is of expansive type and the predominant clay mineral
could be montmorillonite. The soils at the site are
within the flood plains of the river Godavari. The
river over a major part of its length traverses through
the Deccan Plateau wherein basaltic formations exist.
These rocks normally weather in to expansive clay,
locally called Black Cotton clay.

4. IN SITU TESTING
Standard Penetration Tests were conducted in the
three boreholes. Results from other agencies indicate
Figure 10 Sectional Elevation of Wing Wall 4 N values of 2 to 3 up to a depth of 12.5 m and
increase to about 10 to 12 at 17 to 18 m. The N values

8 FGE 2009, Bangalore


in the sand stratum were in excess of 30. Another but reduce to 2 to 3 until a depth of 15 in BH.1, 16 m
investigation conducted in 2001 also indicated N in BH.2 and 18.0 m in BH.3. N values increase to
values of 45 to 60 in the depth range of 19 to 22. about 20 to 27 at 18 to 20 m depth range. N values in
the lower sand stratum were greater than 40. Fig. 12
The investigations carried out by JNTU corroborate shows the variation of the SPT values conducted by
the above studies. SPT N values were slightly high (4 various consulting agencies.
to 5) near the top (2-3 m) possibly due to desiccation

JNTU Professional Agency 1 Professional Agency 2


N N N
0 10 20 30 40 50 60 0 10 20 30 40 50 60 0 20 40 60
0 0 0

5 5 5

10 10
10

Depth, Z
D epth, Z
Depth, z

15 15
15

20 20
20
25 25
25
30 30
Series1 Series2
30 .
BH - 1 BH - 2 BH - 3

Fig. 12 Comparison of SPT N Values from Different Agencies

5. RESULTS AND DISCUSSION Water Content


0 20 40 60 80 100
0
The laboratory tests were conducted on the
undisturbed soil samples to determine the Atterberg
limits, the natural moisture content and the 5
undrained strength of the soil at various depths. Fig
13 shows the variation of the moisture content with
depth for the three boreholes investigated. The 10

water content is close to 60% in the top 3 to 4 m


depth and increase to 80% in the soft clay layer. 15
The values reduce to about 20 to 25 at depth 20 m
Depth

within the sandy soils underlying the soft clay. Fig.


14 shows the variation of the Liquidity index with 20

depth for the three boreholes investigated.


25

30

35 BH-1 BH-2 BH-3

Figure 13 Variation of Natural Moisture Content


with Depth.

9 FGE 2009, Bangalore


-0.5 0.5 LI 1.5 2.5
0

10
z

20

30 BH-1 BH-2 BH-3

Figure 14 Variation of Liquidity Index with Depth

The displaced soil induced lateral pressures on the


surrounding structure as a result of which the wing
walls settled and cracked. The damage to the ROB
structure is severe due to large lateral pressures
generated by the embankment fill.

6. ANALYSIS

The undrained strength of the in situ soft soil is


around 10 to 15 kPa. The ultimate bearing capacity
of the embankment would be of the order of 55 to
65 kPa. An embankment of 4.5 m height generated
a vertical stress of 83 kPa (=4.5x18.5 kN/m3). The
construction of the embankment induced plastic
flow and failure of the soft soil. Since the soil is
highly plastic clay and has low bearing capacity,
the soil was squeezed laterally due to embankment
loading. The lateral movements were less across
the embankment possibly because of the berms and
basal reinforcement provided. However, there was
no such restraint in the longitudinal direction
towards the wing walls and the abutments. Hence
the substructure moved significantly.

7. CONCLUSIONS

The case study illustrates an interesting interaction


between an already constructed bridge abutments
and embankment constructed subsequently by
another agency with no communication between
the two. The stability of the embankment was
ensured in the transverse direction by the provision
of berms and basal geogrid reinforcement. The
interaction between the embankment and the
abutment in the longitudinal direction was not
considered resulting in large unacceptable
movements of the bridge structure rendering it
dysfunctional.

10 FGE 2009, Bangalore


Case Study 12
AN INVESTIGATION INTO THE COLLAPSE
OF A PORTION OF SOIL NAILED WALL

Prof. T.G. Sitharam,


Professor of Geotechnical Engineering, and
Chairman, Centre for infrastructure, Sustainable Transportation and Urban Planning
Indian Institute of Science, Bangalore – 560012.
Email: sitharam@civil.iisc.ernet.in

SUMMARY
The excavation for a site, which has three basements, the soil nailing has been
adopted for retaining the earth so as to have a vertical cut temporarily until a permanent
concrete retaining wall is constructed. On March 12th, west side of the soil nailed wall was
collapsed (see Fig 2) for a length of 20m and road on the adjacent property was also
damaged. A temporary restoration scheme (See Fig 4) consisting of earthen embankment fill
and sand bags fill was already adopted as an immediate measure. Geotechnical report clearly
indicates that the top soil (is about 0.5m) and lateritic red soil (up to a depth of 2.75m to
3.5m) from the ground level. Very close to the failure area, the lateritic red soil exists up to
3m depth below which disintegrated rock is present. Soil gradation also clearly indicates
presence of uniform sand (less than 2mm particle size), silt and clay sizes non presence of
any gravels or boulders in the area. N values are invariably greater than 15 in insitu condition.
Driven soil nailed walls have been constructed as a temporary measure all along the
boundary to retain the earth for a depth of about 10m. 20mm dia TOR steel nails of length 5m
long at spacing of 450 mm c/c were driven in to the ground as shown in cross section. 50mm
thick shotcrete of 1:2:2 grout with 50x50x2.6mm wire mesh was adopted (see Fig 2 for the
final finished wall surface). Nailing was done in stages of 1.5m deep at every stage. This was
reported to be repeated up to a depth of hard rock. Typical cross section of the soil nailed
wall is shown below in fig 1.
The soil nailing has been reported to be completed on 14th October 2008. On March
12th, west side of the soil nailed wall was collapsed for a length of 20m and road on the
adjacent property was also damaged. The typical diagram clearly indicates that the failure
(yellow colored area) is only in the top 3m of the soil mass which corroborates well will the
geotechnical borehole data. The fig 2 clearly indicates there is no failure below 3m from the
ground level (Ground RL is at 99.85 m). The failure surface does not pass through the
disintegrated rock mass. Figure 3 shows the collapsed area in plan.

39
Fig 1. Cross Section of the soil nailed wall and Plan view

 
 
 
Fig 2. Typical Cross-section of the failed slope

Fig 3. View of the soil nailed wall after collapse (left


fig) and restored slope with sand bags (right photo)

40 FGE 2009, Bangalore


Fig 4. Temporary scheme adopted

Considering the adopted temporary restoration scheme with embankment (see


photograph in Figure 3 and Fig 4), the non-availability of working space on the west side
road, and noise/vibration due to piling, it was decided to work within the property of the
client from inside their land initially to restore the slope.

The following scheme (as shown in Figure 5) is suggested as a remedial measure for
permanent restoration of the failed section of the wall (about 20m long). The scheme consists
of construction of a retaining wall at the toe of the temporary embankment, grouting of the
pervious embankment to take the applied pressures from the bottom approach ramp and
grouting of the sand bags/filled up soil in the failed slope section. As a precautionary
measure, even the temporary embankment and the adjacent road re-constructed using sand
bags has been grouted by drilling holes using clean cement grout, so that it can take the load
of the approach ramp uniformly.

41 FGE 2009, Bangalore


The whole area was reported with oozing water on the face of the wall (See photo in
Fig 2). This might be due to the puncturing of the live manhole at several points due to soil
nailing. Further, the failure is attributed to not providing the deisgned length of the nails in the
execution. The actual length of nails provided were shown through analysis shorter than what
has been proposed to have a factor of safety less than 1. The present paper highlight the scheme
of restoration suggested by the consultant rather than the cause of failure. Consultant was
involved only after the failure in the month of April 2009. This presentation describes the scheme
of restoration. The scheme involves a retainng wall away from the boundary (inside the clients
area) along with a backfill and a retaing wall at the toe of temporary embankment coupled with
constructing approach ramp in stages. As this scheme did not cause any disturbance to
the neighbouring areas, the scheme was accepted and executed. The
disadvantage of the scheme is that the client lost a small area of about 6 to 8 car parks.

Fig 4. The final view of the restored retaining wall with the approach slab

42 FGE 2009, Bangalore


CASE STUDY 13

THE CASE OF A TILTED CHIMNEY – A FORENSIC ANALYSIS


RAO.V.V.S.
Nagadi Consultants Pvt. Ltd., New Delhi.

THE PROBLEM:

A R.C. Chimney of 45m.height having a diameter of 5m. at GL was constructed in a sugar factory in the
year 2001. By about June 2002 the chimney was observed to have tilted by 450mm. at the top. To
correct the tilt, sand bags were piled up on the diametrically opposite side of the direction of tilt. The
chimney regained the original plumb, but it developed wide vertical cracks at three locations on the
chimney throughout its height. To develop an appropriate repair solution, detailed investigation was
conducted to determine the cause of tilt.

DATA COLLECTION:

From the design as well as “as constructed” drawings, it was observed that the chimney had been
constructed as per designs. Detailed borehole investigations had been conducted and proper design
parameters had been used. However, there was no borehole data at the location of the chimney. Hence
bore hole investigations were performed around the chimney. The sub-soil profiles were uniform. The
profile is shown in fig. 1.

Fig.1.SUB - SOIL PROFILE


BH 3
E.G.L
-0.0
FILL
FILLED UP STRATA (GREY SANDY SILT/SAND WITH
-1.0 -5 BRICK BATS AND DEBRIS)
……… GREY CLAYEY (4-12%) SANDY (9-22%) SILT (65-82%)
-2.0
………
-4 WITH GRAVEL (1-10%)
………
-3.0
………
………
-4.0 -10
………
………
-5.0
DEPTH IN METRES

………
-16
………
-6.0
………
………
-7.0 -22
………
………
-8.0
………
-20
………
-9.0
……… GREY SAND(90-99%) WITH TRACES OF SILT (1-10%)
………
-10.0 -26
………
………
-11.0
………
-28
………
-12.0
………
………
-13.0
………
………
-14.0
………
-37
………
-15.0

SOIL PROFILE

LEGEND : 5 Observed 'N' Values Water Table

E.G.L.: Existing Ground Level

The foundation raft was founded at a depth of -3m. The soil below consists of medium dense sand.
During the site visit, it was found that a single storey shed had been built very near to the chimney on the
tilted side. Upon enquiry, it was told that the shed was constructed after the factory went into operation,
without consulting the design or geotechnical consultants. The crack was observed a few weeks later. The
cross-section of the construction was as shown in fig.2.

Fig.2.

In plan, the section at GL was as shown in fig.3.


VIEW OF A CRACK VIEW OF CHIMNEY AND SHED

CONCLUSIONS:

The additional loads acting on diametrically opposite ends of the foundation raft resulted in inducing
splitting action of the chimney resulting in cracks.
CASE STUDY 14

INFLUENCE OF VIBRATIONS ON INSTALLATION OF


BORED PILES

Dr. N. Santosh Rao


Nagadi Consultants Pvt. Ltd., New Delhi

1. INTRODUCTION

This is a case study of a forensic geotechnical investigation to assess the possible


cause/s which prevented the satisfactory construction of large diameter bored piles using the
bailer technique. The investigation involved first identifying the possible causes, carrying
out field and laboratory tests where necessary to assess the same, analysis of the test results
and finally arriving at the conclusion.

2. PROJECT DETAILS

The project in this case study is an elevated corridor through the Kishanganj town,
Bihar, India, located parallel to the existing road as also the railway line abutting the road.
The elevated corridor had been proposed to be constructed in reinforced cement concrete
with the deck consisting of box girders supported by two column piers resting on 1m
diameter 16m long bored piles.

3. PROBLEM DEFINED

The piling contractor had been unable to ensure satisfactory construction of the large
diameter bored piles using the bailer technique, primarily because of frequent collapse of the
sides of the bores made for the purpose of construction of the bored piles. As per the
conditions of the contract, bentonite slurry was not to be used in the pile bores and use of
liner/casing had also not been envisaged on account of the nature of the subsoil as indicated
by the report of the geotechnical investigations carried out earlier for this project.

4. POSSIBLE CAUSES

An evaluation of the above problem indicated that one or both of the following could
be the possible causes :
1. Subsoil being unsuitable for construction of the bored piles under the constraints
put up by the contract.
2. Vibrations transmitted from the trains moving on the nearby railway tracks and
vehicle movement on the adjacent road.
5. INVESTIGATION OF POSSIBLE CAUSES

a. Subsoil Conditions

To check whether the subsoil conditions at the site of the proposed elevated corridor
are similar to those considered in the design and included in the tender documents, a few
boreholes had been carried out at select locations along the length of the elevated corridor. A
comparison of the typical subsoil profile indicated by these boreholes with the typical subsoil
profile indicated by the geotechnical investigation report included in the tender documents is
given below:

Geotechnical Investigation Report New Boreholes


Depth
(m) Soil Range of N- Soil Range of N-
Description values Description values
Fine Sand with
0–5 10 – 15 10 – 14
Silt
Silty Fine Sand Medium to
5 – 10 / Sandy Silt 15 – 20 12 – 20
Coarse Sand
10 – 20 20 – 25 Coarse Sand 25 – 35
Coarse Sand
20 – 25 - - > 50
with Gravel

The above comparison clearly shows that the actual subsoil conditions differ
significantly from that indicated by the geotechnical investigation report. While an
unsupported pile bore may remain intact in a silty sand / sandy silt soil the same will
invariably tend to collapse in a medium to coarse sandy soil. Therefore, while based on the
results of the geotechnical investigation report, an unsupported pile bore can be considered
feasible, the same will be considered difficult to infeasible when the actual subsoil conditions
as indicated by the new boreholes are taken into account.

b. Vibrations from Rail and Road Traffic

To assess the level of vibrations caused by the rail and road traffic in close proximity
to the site of the proposed elevated corridor, vibration measurements had been carried out
using geophones (i.e. velocity pick-ups with a sensitivity of 28.3 mv/mm/s) connected either
to a storage oscilloscope or to a computer based data acquisition system. The locations of the
vibration measurement points have been indicated in the schematic layout plan of the area
given in fig.1.

The recorded vibrations from different types of trains as also from trucks plying on
the road have been presented in figs. 2 to 6. The absorption coefficient for the subsoil at the
site has been determined from the vibrations recorded during the passage of a Rajdhani
express train by locating the geophones in a line perpendicular to the railway tracks on the
opposite side of the road under the proposed elevated corridor.

From the recorded vibrations and the absorption coefficient determined, the values of
the acceleration ratio at various distances from the tracks will be :
Distance (m) : 6 10 14 18 22

Acceleration Ratio : 0.03 0.013 0.009 0.005 0.002

As per Tscheboterioff (1965), dynamic forces caused by pile driving operations


having high accelerations in the range of 0.01g to 0.001g have been shown to have caused
significant soil settlements to cause distress to the nearby structures and substructures. As the
levels of vibrations caused by the rail traffic are in the above range, the vibrations caused by
the rail traffic can be considered be a major contributory factor in the collapse of the pile
bores.

6. CONCLUSIONS

The investigation of the possible cause/s indicates that both the considered possible
causes have contributed to the frequent collapse of the pile bores in that:
1. The subsoil profile given in the tender documents as indicated by the geotechnical
investigation report is wrong.
2. The actual subsoil conditions indicate that the subsoil at the site is unsuitable for
use of the bailer technique.
3. The agency that carried out the initial geotechnical investigation is the main
culprit in having provided wrong data about the subsoil conditions leading to the
inappropriate choice of method of construction of the bored piles.
4. The engineers of the highways authority should have properly verified the
geotechnical investigation report before going in for tendering of the work and
insisting on bored piles with restrictions on method adopted for the same.
5. Vibrations from the rail and road traffic are significant enough to have contributed
to the frequent collapse of the pile bores.
6. Use of full length liner would be the only method by which bored piles could be
constructed in such subsoil conditions where significant vibrations are generated
by rail and road traffic.
Fig.1 : Schematic Layout Plan

Fig.2 : Express Train on Track 3 in E-W direction


Fig.3 : Rajdhani Express Train on Track 3

Fig.4 : Truck Plying on the Road


Fig.5 : Passenger Train on Track 1

Fig.6 : Goods Train on Track 3


ANNEXURE
COMPOSITION OF TC 40: FORENSIC GEOTECHNICAL ENGINEERING

Chairman:Dr.V.V.S.Rao, India dr_rao@nagadi.co.in


Secretary: Prof. G.L. Sivakumar Babu, India, gls@civil.iisc.ernet.in
Core Members:
1. Prof. Yoshi Iwasaki, Japan, yoshi-iw@geor.or.jp
2. Dr. Richard Hwang, Moh & Associates, Taiwan, richard.hwang@maa.com.tw
3. Mr. David Starr, Golder Associates, Australia, david_starr@bigpond.com
4. Prof. K.K. Phoon, Singapore, cvepkk@nus.edu.sg
5. Mr. P.W. Day, South Africa, day@jaws.co.za
6. Prof. J. Mesci, Hungary, mecsi@pmmk.pte.hu
7. Dr. Jan Hellings, U.K., jan@janhellings.com
8. Mr. D.S. Saxena, USA. dsaxena@ascworld.net
9. Prof. Mihail e Popescu mihail.e.popescu@gmail.com

General Members:
1. Prof. L.G.De Mello, Brazil lgdemello@usp.br
2. Dr. Dianging Le, China dianging@whu.edu.cn
3.Prof. Zeliko Arbanas, Croatia zeliko.arbanas@gh.hr
4.Mr. Jack Pappin, Hong Kong jackpappin@arup.com
5. Prof. Lingwei Kong, China lwkong@dell.whrsm.ac.cn
6. Mr. W.K.Pun, Hongkong wkpun@ceddigov.hk
7. Dr. R.Szepeshazi, Hungary szepesr@sze.hu
8. Prof. S. Hamsin, Kazakstan
9. Dr.W.Cichy, Poland wcic@pg.gda.pl
10. Prof. R. Gilbert, USA bob.gilbert@mail.utexas.edu
11. Dr. Chan S. F. Malaysia sfchan@pc.jaring.mv
Corresponding Members:
1. Mr. Marco Uzielli Italy, marco.uzielly@ngi.no
2. Mr. Gil Yoon, South Korea glyoon@kordi.re.kr
3. Mr. Masahiro Shirato, Japan, shirato@pwri.go.jp
4. Mr. Makoto Suzuki, Japan,
makoto.suzuki@shimz.co.jp
5. Mr. Limin Zhang, Hong Kong, cezhangi@ust.hk
6. Mr. Chris Basile P.E. (USA) christopherbasile@yahoo.com
7. Mr. Grey Stephan P.E. (USA) gstephan@ascworld.net
8. Dr.Phili Pells pells@psmtoo.com.au
9. Mr. Michael Marley, mmarley@golder.com.au
10. Strath Clarke
11. Prof. Mihail Popescu , mepopescu@usa.com
THE SUBCOMMITTEES
SC 1: Characterization of distress. Convener: Mr. P.W. Day
SC2: Diagnostic Tests. Convener: Mr. David Starr
SC 3: Back Analysis. Convener: Dr. Hwang
SC 4: Instrumentation. Convener: Prof. Yoshi Iwasaki
SC 5: Development of failure hypothesis: Convener: Dr. J.Hellings
SC 6: Reliability Checks. Convener: Prof. K.K.Phoon
SC 7: Legal Issues: Convener: Mr. D.S.Saxena
SC 1: CHARACTERIZATION OF DISTRESS.
Convener: Mr. Peter W. Day day@jaws.co.za
Members:
1. Prof. Luiz de Mello lgdmello@usp.br
2. Jack Pappin jackpappin@arup.com
3. Prof. Bob Gilbert bob_gilbert@mail.utexas.edu
4. Dr. Jan Hellings jan@janhellings.com
5. Nick Shirlaw nshirlaw@golder.com.sg
SC 2: DIAGNOSTIC TESTS.
Convener: Mr.David Starr david_starr@bigpond.com
Members:
1. Dr Philip Pells, Pells Sullivan & Meynink (PSM) pells@psmtoo.com.au
2. Mr Michael Marley, Golder Associates mmarley@golder.com.au
3. Strath Clarke (he works for PSM )
SC-3: BACK ANALYSIS.
Convener: Dr. Mihail Popescu
mihail.e.popescu@gmail.com
Co-Convener:Dr.RichardHwang. richard.hwang@maa.com.tw
Members:
1.Dr.StephenButtling stephen.buttling@hyderconsulting.com
2.Mr.NiclauseShirlaw nshirlaw@golder.com.sg
3.Dr.TingWenHui tingwh@pop.jaring.my
4.Prof.FusaoOka foka@mbox.kudpc.kyoto-u.ac.jp
5.Dr. Chung Tien Chin ct.chin@maaconsultants.com
SC 4: INSTRUMENTATION
Convener: Dr. Yoshinara Iwasaki yoshi-iw@geor.or.jp
Members:
1.Mr. Craig A. Davis (USA) craig.dvis@ladwp.com
2. A. WADA (Singapore) : wada@agacorp.com.sg

SC 5: DEVELOPMENT OF FAILURE HYPOTHESIS.


Convener: Dr. Jan Hellings. jan@janhellings.com
Members:
1. Bill Grose. Ove Arup & Partners (UK).
bill.grose@arup.com
2. Rab Fernie. Cementation Foundations Skanska Ltd
rab.fernie@skanska.co.uk
3. Kjell Karlsrud. NGI (Norway). kjell.karlsrud@ngi.no
4. Dirk Luger. GeoDelft (The Netherlands).
dirk.luger@geodelft.nl
5. Mahmoud Mahmoud. GES Geotech (Canada).
mahmoud@gesgeotech.com
6. Darren Page. OTB (UK). geoscience@hotmail.com
7. Roger Thomson. EDGE Consultants UK Ltd.
rpt@edgeconsultants.co.uk
SC6: RELIABILITY CHECKS.
Convener: Prof. Kok-Kwang Phoon (Singapore,) –
cvepkk@nus.edu.sg
Members:
1. Prof. Sivakumar Babu (India) - gls@civil.iisc.ernet.in
2. Mr. Robert Gilbert (USA) - bob_gilbert@mail.utexas.edu
3. Mr. Dian-qing Li (China) - dianqing@whu.edu.cn
4. Mr. Marco Uzielli (Italy) - Marco.Uzielli@ngi.no
5. Mr. Gil Yoon (South Korea) - glyoon@kordi.re.kr
6. Mr. Masahiro Shirato (Japan) - shirato@pwri.go.jp
7. Mr. Makoto Suzuki (Japan) - makoto.suzuki@shimz.co.jp
8. Mr. Jack Pappin (Hong Kong)
9. Mr. Limin Zhang (Hong Kong) - cezhangl@ust.hk

SC 7- LEGAL ISSUES INVOLVING JURIPRUDENCE SYSTEM


Convener: Mr. Dhirendra S Saxena P.E. (USA) dsaxena@ascworld.net
Members:
1. Mr. Christopher C Basile P.E. (USA) christopherbasile@yahoo.com
2. Mr. Greg A Stephan P.E. (USA) gstephan@ascworld.net

You might also like