Innovations in Geotechnical Design: 2021 Australian Geomechanics Society Victorian Symposium
Innovations in Geotechnical Design: 2021 Australian Geomechanics Society Victorian Symposium
VICTORIA CHAPTER
PREFACE
The Victorian chapter of the Australian Geomechanics Society (AGS) is pleased to announce a one-day
symposium titled “Innovations in Geotechnical Design” which is to be held on 21 October 2021. The event
platform changed to an online-only event due to development of new restriction due to COVID-19 pandemic.
This event will bring together geotechnical and other civil engineering professionals to share and discuss
their knowledge and experiences related to geotechnical design and the latest technology and innovative
methods.
Victoria’s construction industry is still up and running, recovering quite well from the hiccups of the pandemic
and several lock-downs. To catch up with the fast pace of global advancement in technology, incorporating
more and more of the emerging and innovative methods in geotechnical design and construction seems to
be the way forward, facing the challenges in the built environment. This symposium will present overviews
of the state-of-the-art practices innovations in design, including new research, case studies, advanced
technology and simulations related to various geo-structure systems, as well as reliability, safety, and
observational design implemented in Victorian projects.
The symposium will bring together professional engineers, researchers, specialist contractors, regulators,
educators and students to share and discuss their experiences on the above topics. This will be a great
networking opportunity post-pandemic.
GOLD
Global Synthetics Pty Ltd
Insitu Geotech Services (IGS)
All right reserved. No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form without the permission of the Australian Geomechanics Society.
© 2021 Australian Geomechanics Society.
SESSION 2 27
INSTRUMENTATION AND
MONITORING IN DESIGN
Ground truth, control and design of driven piles: 29
implementing old ways with a new twist
Julian Seidel (Foundation QA)
Go to table of contents
SPONSORS
PLATINUM SPONSOR
GOLD SPONSORS
TECH SERV
EO I
CE
G
INSITU
S
re d u
cin
ge in
ty
g
ote ta
c h nic al u n cer
Go to table of contents
General Services
Chadwick Geotechnics is a leading supplier of services to the
Geotechnical, Civil and Environmental disciplines across Australia, its
Territories, and throughout the Asia Pacific region.
Discipline of Civil, Surveying and Environmental Engineering, College of Engineering, Science and Environment, The University of
Newcastle, Callaghan NSW 2308, Australia; PH: +61 2 4921 5118; email: Jinsong.huang@newcastle.edu.au
ABSTRACT
Due to the intrinsic inhomogeneous nature of soils and rocks, the minimal site investigations, and the need to extrapolate
available information over a large domain, geotechnical designs have inevitable uncertainties. To be conservative,
geotechnical engineers traditionally use a safety factor to account for uncertainties. A more rigorous way of considering
uncertainties is to use probabilistic methods. To promote the use of probabilistic methods in geotechnical engineering, this
paper tries to address the following commonly encountered questions: 1) Why do we need to use probabilistic methods?
2) How can we use probabilistic methods if we don’t have enough test data? 3) How much field/test data do we need? 4)
How can we use multiple sources of information? 5) How can we use monitoring data to predict future performance?
Keywords: probabilistic methods, geotechnical engineering, limited data, Bayesian methods, prediction
2.1 Limited site investigation data Considering two examples of drained slope stability, they
both have the same geometry as shown in Figure 1. The
Soils and rocks in their natural state are among the most characteristic values of the soil properties in the two
variable of all engineering materials, and geotechnical examples are shown in Eq. (1) and (2) respectively.
engineers must often “make do” with materials that
present themselves at a particular site. In a perfect world
with no economic constraints, we would drill numerous
boreholes and take multiple samples back to the
laboratory for measurement of standard soil properties
such as permeability, compressibility, and shear strength.
Armed with all this in-formation, we could then perform
our design of a seepage problem, foundation, or slope
and be very confident of our predictions. In reality we must
usually deal with very limited site investigation data, and
the traditional approach for dealing with this uncertainty in
geotechnical design has been through the use of
characteristic values of the soil properties coupled with a
Figure 1. Slope profile
generous factor of safety.
Go to table of contents
2
Use of probabilistic methods in geotechnical engineering J. Huang
Probability of where
0.27% 2.27% N () represents the Normal distribution, and
failure
µ0 and σ 0 are the mean and standard deviation of µ
Consequence (mean of mean pile capacity, standard deviation of mean
10 100 10 100
(million dollar) pile capacity). The parameters µ0 and σ 0 were assumed
to be associated with prediction methods (a static or
Risk
1.027 1.27 0.727 2.77 dynamic method).
(million dollar)
Go to table of contents
3
Use of probabilistic methods in geotechnical engineering J. Huang
P (y | µ ) f ′(µ )
f ′′ ( µ | y ) =
P (y) (5)
(a)
nσ 2 y + σ 2 µ0
2
µ − 0 2
nσ 0 + σ 2
∝ exp −
σ 2σ 2
2 20 2
nσ 0 + σ
(b)
.
where y is the average value of y Figure 2. True in-situ distribution of undrained shear
strength
A detailed derivation of Eq. (5) can be found in Huang et
al. (2016). Before we do the excavation or test, we actually don’t
know the undrained shear strength distribution in Figure
The posterior mean and standard deviation of µ are 2(b). We can look in the literature for the statistics of the
undrained shear strength of the soil in this particular site.
nσ 02 y + σ 2 µ0
µ1 = (6) And based on the statistics, we can use random field
nσ 02 + σ 2 theory to guess the possible spatial distributions of the
undrained shear strengths and perform a set of Monte
and Carlo simulations as shown in Figure 3. We generate a
set of realisations of the slope, some of them will fail (e.g.,
Figure 3 a), some will not (e.g., Figure 3 b). If you divide
σ 02σ 2
σ1 = (7) the total number of failures by the total number of
nσ 02 + σ 2 simulations, we can estimate the probability of failure,
which is 17% in this case.
Let us assume the prior mean and standard deviation of
µ are µ0 =1.3 and σ 0 = 0.5 and suppose five load tests For the same slope, if we perform a cone penetration test
have been conducted. The average capacity of tested as indicated by the red box in Figure 4, we then know the
piles is y = 0.8 . According to Eqs. (6) and (7), the undrained shear strength of this column. Now we have
statistics of the undrained shear strengths plus some test
posterior mean and standard deviation of µ are data. Based on the statistics and data, we can perform a
set of conditional Monte Carlo simulations (e.g., Yang et
nσ 02 y + σ 2 µ0 al. 2019). Conditional means that the values of the known
µ1 = undrained shear strengths will not change from simulation
nσ 02 + σ 2 to simulation, but fixed. From a set of conditional Monte
5 × 0.52 × 0.8 + 0.22 ×1.3 Carlo simulations, we can estimate the probability of
= failure, which is 6% in this case.
5 × 0.52 + 0.22
= 0.82 If we perform a second CPT, we will know two columns of
undrained shear strengths as shown in Figure 5. We can
and perform another set of conditional Monte Carlo
simulations, and estimate the probability of failure, which
σ 0σ
σ1 = is 0.7%. If we perform a third CPT, the estimated
nσ 02 + σ 2 probability of failure is essentially close to zero as shown
in Figure 6. If we keep doing test, eventually we will know
0.5 × 0.2 the undrained shear strength distribution everywhere.
=
5 × 0.52 + 0.22 And it turns out that the slope is actually safe. we can see
based on three CPTs, we can reach a very reliable
= 0.09
prediction of the slope stability. Please note that this
The uncertainty of the mean capacity (µ ) has been conclusion only holds for this particular slope. For a
reduced significantly from σ 0 = 0.5 to σ 1 = 0.09 by the different slope, the slope may fail, and we may need
different number of CPTs to find it out. But the same
five load tests. It is interesting to note that even when we
methodology can be used to study how much tests do we
have only one test, we can still use Eqs. (6) and (7).
need to draw a reliable prediction.
Go to table of contents
4
Use of probabilistic methods in geotechnical engineering J. Huang
.
. Conditional Monte Carlo simulations
(a) .
.
. Monte Carlo simulations
.
(b)
Go to table of contents
5
Use of probabilistic methods in geotechnical engineering J. Huang
Table 2: Risk assessment of slope failure the soil profile based on conditional random field and the
CPT test only. It can be seen from Figure 8 that at the
Approach Input data place where we have CPT test, the mean cone tip
resistance is the same as we tested by CPT and the
standard deviation of cone tip resistance is low. However,
Deterministic Characteristic values at the places where we don’t have any test, the mean
cone tip resistance is similar to and informed only by the
CTP test results, and the standard deviation of cone tip
Mean, standard resistance is large. This means that we have less
Unconditional
deviation, spatial confidence on the profile of cone tip resistance based on
probabilistic analysis
correlation structure one CPT only.
CPT-6A
Mean, standard MASW-1
0 200
Conditional probabilistic deviation, spatial
Depth (m)
5
analysis correlation structure, test
VS (m/s)
150
data 10
100
15 qt (MPa)
0 2.5 5
0 10 20 30 40 50 60 70
Machine learning Test data Distance (m)
Depth (m)
5 4
qt (MPa)
Site characterisation is a fundamental step of collecting 10
3
qt (MPa)
0.08
situ geotechnical investigation including several 10
0.06
boreholes and laboratory testing at the intended location 15 0.04
of the future structure. Each type of investigation explores 0 10 20 30 40 50 60 70
a specific volume of the subsoil and has different degrees Distance (m)
of uncertainty. Geophysics provides a wide variety of tools
which can help to identify the subsoil stratigraphy. Data Figure 8. Soil profile based on one CPT test, a) mean
are obtained on a two- (or three-) dimensional section of cone tip resistance, b) standard deviation of cone tip
the ground, often with low resolution. On the other hand, resistance.
in situ tests are performed along a one-dimensional line
with depth and laboratory tests performed at point If we use the CPT and MASW test results together, we
locations within the ground. These tests provide direct can have a higher confidence on the cone tip resistance
measurements of physical and mechanical properties, but profile. Figure 9 shows the soil profile based on Bayesian
cover only a small volume of the subsoil. In current updating and both CPT and MASW test results. It can be
engineering practice, the integration of geophysical data seen from Figure 9 that at the place where we have CPT
with geotechnical data is done manually, often by visual test, the mean cone tip resistance is still the same as we
inspection based largely on engineering judgement and tested by CPT and the standard deviation of cone tip
experience, which does not only introduce additional resistance is low. However, at the places where we don’t
human error, but also causes loss of information. For have CPT test, the mean cone tip resistance is informed
example, the geophysical data is used to find the by the MASW results, and the standard deviation of cone
desirable locations of geotechnical tests, but often tip resistance is reduced.
ignored in deriving geotechnical properties.
Go to table of contents
6
Use of probabilistic methods in geotechnical engineering J. Huang
6 HOW DO WE USE MONITORING DATA TO consolidation periods for the test embankment built at
PREDICT FUTURE PERFORMANCE? Ballina, New South Wales, Australia. The results show
that surface settlement can be well predicted using 116
In design stage, we predict the safety level of the days of observed settlements, while the pore pressure
structures based in lab and field test results. After can be predicted using 292 days of pore pressure
construction, the behaviour of the structure most of the measurements. The predictions are shown to converge to
time is different from our prediction. For important projects the field measurements, regardless of some assumptions
such as large dams and bridges, we usually install some about the measurement errors. It is also demonstrated
sensors or devices to monitor the behaviour of the that incorporating more monitoring data into the Bayesian
structure. In this case, we need to use the monitoring data updating process enables more accurate predictions.
to reassess the safety level of the structure. This is called
back analysis. 7 CONCLUSIONS
Another example where back analysis becomes Deterministic factor of safety approach has been used in
important is embankment built on soft soils. Due to geotechnical engineering for many decades. Although
variations in a soft soil profile and its properties, the probabilistic approaches can provide more information for
actually settlement behaviour is usually different from the risk assessment, there are still quite some resistances to
performance anticipated in design. If the decision on the the use of probabilistic approaches. The most common
settlement behaviour (i.e., quicker or slower than excuse for not using probabilistic approach is lack of data.
expected) can be made sooner, then a smaller amount of This paper has shown that Bayesian approaches can be
surcharge is required and road construction can start used even when we have only one test result. Bayesian
earlier, which leads to significant financial benefits. approaches are also useful for combining different types
of test results and back analysis based on monitoring
Table 3 lists some of the most commonly used back data. It is anticipated that probabilistic approaches will
analysis methods. The simplest back analysis is manual become more and more popular in geotechnical
calibration where engineers try to minimise the different engineering to supplement, not replace the factor of
between observation D and model output g (θ ) . Least safety approach.
square method is probably the most widely used method
8 ACKNOWLEDGEMENTS
for back analysis because it always provides a solution no
matter how complicated the problem is. Least square is
The research supports by the Australian Government
actually a special case of the Maximum Likelihood
through the Australian Research Council's Discovery
method where measurement error is normally distributed
Projects funding scheme (project DP190101592) and
with zero mean. In contrast to the Maximum Likelihood
Linkage funding scheme (project LP200100367) are
method, Maximum A Posterior method not only consider
acknowledged.
the likelihood, but also the prior distribution of the
parameters. Bayesian back analysis is the most generic
REFERENCES
type of back analysis. Instead of solving for the set of
parameters that maximum posterior probability density as Huang, et al. (2016). "Updating reliability of single piles and pile
in the Maximum A Posterior method, Bayesian updating groups by load tests." Computers and Geotechnics 73:
samples the whole posterior probability density function. 221-230.
The most commonly used method for this purpose is the Huang, et al. (2018). "Probabilistic characterization of two-
Markov chain Monte Carlo method. dimensional soil profile by integrating cone penetration test
(CPT) with multi-channel analysis of surface wave
Table 3: Commonly used methods for back analysis (MASW) data." Canadian Geotechnical Journal 55(8):
1168-1181.
Method Formula Kay (1978). "Safety Factor Evaluation for Single Piles in Sand."
Journal of the Geotechnical Engineering Division-Asce
Manual calibration Min D − g (θ ) 104(1): 148-149.
Kelly, et al. (2015). "Bayesian updating for one-dimensional
consolidation measurements." Canadian Geotechnical
Min ( D − g (θ ) )
2 Journal 52(9): 1318-1330.
Least square
Michalowski (2002). "Stability Charts for Uniform Slopes."
Journal of Geotechnical and Geoenvironmental
D − g (θ ) − µe Engineering 128(4): 351-355.
Maximum Likelihood Max L
σe
Yang, et al. (2019). "Importance of soil property sampling
location in slope stability assessment." Canadian
D − g (θ ) − µe Geotechnical Journal 56(3): 335-346.
Maximum A Posterior Max L P (θ )
σe
Zheng, et al. (2018). "Embankment prediction using testing data
and monitored behaviour: A Bayesian updating approach."
D − g (θ ) − µe Computers and Geotechnics 93: 150-162.
Bayesian updating P (θ D ) ∝ L P (θ )
σe
Go to table of contents
Reliability-based geotechnical design in an Australian context
1
Principal Geotechnical Engineer, Golder Associates Pty Ltd, Melbourne, PO Box 6079, Hawthorn East, VIC 3121; PH (+61) 3 8862 3500;
FAX (+61) 3 8862 3501; email: alochaden@golder.com.au
2
Principal, National Geotechnical Consultants, PO Box 485, The Gap, QLD 4061; PH (+61) 7 3300 0088; email:
stephen.buttling@ngconsult.com.au
ABSTRACT
The design philosophy adopted by the international geo-engineering community over the last few decades has largely
evolved from working stress design (WSD), in which a single or lumped factor of safety (FoS) is adopted, to a load and
resistance factor design (LRFD) approach. In LRFD, partial factors are applied to actions (i.e. loads), soil parameters and
/ or resistances. These partial factors vary in magnitude depending on the relative uncertainty of each parameter to which
they are applied. The approach adopted in Australia has differed, with both WSD and LRFD approaches being widely used.
In the authors’ experience, there is limited awareness in the Australian geo-engineering community of the relationship
between the concept of reliability and the partial factors adopted in LRFD, and therefore of the potential benefits of
undertaking reliability-based design (RBD). The outcome of this is that RBD, in which the uncertainty of the variables which
may affect the design is individually assessed, is rarely undertaken. This paper discusses the concept of RBD and its place
within the framework of Australian Standards and presents practical means of adopting RBD with accompanying examples
from the literature. The intention of the paper is to encourage practitioners to consider uncertainty in geotechnical design
more rigorously, whilst acknowledging the importance of maintaining engineering judgement in design.
This paper sets out the current state of geotechnical Thirdly, whilst most international design standards have
design in Australia, describes the relationship between moved from WSD to LRFD, Australia has effectively
LRFD and RBD, and presents ways in which RBD can maintained both approaches. For example, pile
be adopted in practice. foundations are typically designed using a limit state
approach to AS2159, whereas slopes are typically
designed using a lumped FoS approach.
Go to table of contents
8
Reliability-based geotechnical design in an Australian context A. L. E. Lochaden and S. Buttling
To further complicate matters, the analysis tools One of the benefits of LRFD over WSD is that the
commonly adopted in geotechnical design have changed engineering professional is required to apply partial
over the last number of years such that parts of the factors to individual components of load and resistance,
national design standards have lost some of their thereby allowing the higher risk of variation in (say) live
relevance. For example, embedded retaining walls are loads compared to (say) the self-weight of concrete to be
now commonly designed using tools such as WALLAP considered in the design. This is not possible in WSD, as
and PLAXIS 2D. However, both AS4678 (Standards a single lumped FoS is applied.
Australia, 2002) and AS5100.3 (Standards Australia,
2017a) provide insufficient guidance on how such soil- LRFD has been subject to criticism by Australian
structure interaction analyses should be undertaken. practitioners, for example Day (2001), Pells (2011), Day
Although beyond the scope of this paper, the reader is et al. (2007), and Wong et al. (2007). It is acknowledged
directed to Haberfield (2017), for example. that many of the arguments made in these references
are valid, but note that a discussion of these arguments
3 THE RELATIONSHIP BETWEEN RBD AND lies outside the scope of this paper. It is noted that
LRFD despite the criticisms set out in these references, Wong
et al. (2007) state that “…despite these difficulties, there
3.1 Introduction is definitely a place for limit state design in geotechnical
engineering” and “the use of limit state design…has
As stated in Section 2, design must balance cost with the benefits as long as the underlying design principles and
safety / serviceability of a structure. Regardless of the soil-structure interaction effects are properly understood
various design methodologies which may be adopted, or and communicated”.
whether the design is a structural or a geotechnical one,
an appropriate design must establish that the design 3.3 Reliability based design
resistance is not less than the design actions (applied
loads). Reliability theory was introduced initially as a design
concept for structural engineering, and subsequently for
The degree of safety of a structure has traditionally been geotechnical engineering, primarily to counter this lack of
assessed in a deterministic manner by engineering robust consideration of the potential variability. The use
professionals in terms of a calculated FoS. The of reliability theory in the field of geotechnical
engineering professional undertakes this assessment engineering is not new – it was the subject of the 17th
using either what he / she considers to be appropriate Terzaghi Lecture in 1981 (Whitman, 1984).
values for applied loads and resistances to take account
of their potential variability, or what the relevant design The requirement for this more robust consideration is
standard mandates. This is illustrated in Figure 1, in discussed below with reference to Figure 2. This has
which the margin of safety is defined as the difference previously been presented in the literature (including
between the resistance and the loads. However, this Becker, 1996a). Loads and resistances are variables
methodology neglects to robustly consider the potential and are not defined by a single value as considered in
variability of both the applied loads and the resistances. the deterministic approach described in Section 3.1.
Sources of geotechnical uncertainty include spatial
variation, measurement error, transformation error (for
Probability of occurrence
1.0
example, the calculation of undrained shear strength of
a cohesive material from Cone Penetration Test data),
and model calculation uncertainty (i.e. the degree to
which the analytical model represents the true behaviour
Margin of safety = R - S
of the system).
Go to table of contents
9
Reliability-based geotechnical design in an Australian context A. L. E. Lochaden and S. Buttling
distribution curve is relatively broad). This could be • The ellipses are centred about characteristic values
the case where little geotechnical investigation has of φ’ and δ.
been undertaken at a site and where there is • The “critical combination” is the most probable
relatively high variability in the geotechnical data. combination of φ’ and δ which would result in failure,
The area of intersection of the load and resistance and is the shortest distance on the plot from the
curves (the combined green and red areas) is characteristic values of φ’ and δ to the failure
relatively large and therefore so is the probability of surface.
failure. • β is defined as the quotient of the shortest distance
• The higher probability of failure of Case (b) from the centre of the ellipse to the failure surface
compared to Case (a) is not captured by the (R in Figure 3) and one standard deviation (r in
traditional FoS approach, as the mean margin of Figure 3).
safety (and therefore FoS) for both Cases (a) and
(b) is the same. 45 One standard-deviation dispersion ellipse
FoS = 1.75
Distribution of load
= 1.00 r
30
(i.e. limit
state surface) R
25 Safety
20
Critical combination
The intention of RBD is to provide adequate confidence 0
that the probability of failure (Pf) is sufficiently low. RBD 0 5 10 15 20 25 30
allows a more consistent assessment of reliability to be Soil-wall interface friction angle, δ (°)
calculated compared to either WSD or LRFD. Although
the theory of reliability analysis lie beyond the scope of Figure 3. Definition of β (after Low, 2005)
this paper, reliability analysis requires assumptions to be
made in relation to the frequency distributions of the More than two variables can be considered in reliability
loads and resistances (for example the normal analyses, the first step being the replacement of the
distributions shown in Figure 2), and therefore the Pf ellipses by ellipsoids in three dimensions. However, to
calculated should not be considered to be the “true” allow graphical representation in two-dimensional space,
probability of failure, but a reasonable estimate. the example above has been limited to two variables
Furthermore, as with any design, reliability analysis only.
requires the engineering professional to use appropriate
means of calculation, interpretation of geotechnical data, If for some design scenario the calculated β value was
etc.. too low, the engineering professional could increase the
strength and size of structural members while leaving the
As an alternative to Pf, the likelihood of failure may be uncertainty changed. This would result in an increase in
expressed in terms of the reliability index (β), which is a the distance from the characteristic points to the failure
description of the safety of the structure normalised by surface (R) and leave r unchanged, thereby increasing
its uncertainty. A higher β value indicates a more robust the calculated β value. An alternative solution would be
design. The target β for a specific structure should be to leave the strength and size of structural members
based on a number of considerations, but primarily the unchanged and decrease the uncertainty, for example by
consequences of failure (for example, risk to life and undertaking more and / or higher quality geotechnical
economic consequences). The target β for Australian investigations. This would cause a reduction in the radius
structures for various considerations is set out in AS5104 of the one standard-deviation dispersion ellipse (r), also
(Standards Australia, 2017b), and which includes an resulting in an increase in the calculated β value.
appendix focussed on geotechnical engineering.
The partial factors presented in national design
The concept of β is explained in Low (2005) with standards and which are required to be adopted in LRDF
reference to Figure 3 and the First Order Reliability may have been (but are not always) calculated based on
Method (FORM, also known as the Hasofer-Lind reliability analysis, as discussed in Section 3.2. The
method, Hasofer and Lind, 1974), demonstrated using benefit of the presentation of partial factors within
the example of the rotational failure of a gravity wall, as national design standards is that the engineering
follows: professional is not required to undertake reliability
• The ellipses in Figure 3 represent combinations of analyses for a design, and can instead use the provided
the effective angle of friction (φ’) of the retained partial factors to achieve the required level of reliability.
material and of the soil-wall interface friction angle However, in deriving partial factors to be applied for all
(δ) which have the same probability of occurring potential scenarios to achieve adequate reliability,
concurrently. reasonable ranges of uncertainty must be considered.
This means that in some design situations, the partial
factors presented in national design standards will result
Go to table of contents
10
Reliability-based geotechnical design in an Australian context A. L. E. Lochaden and S. Buttling
in a β for a structure which is greater than that required, quotient of the standard deviation and mean of the
i.e. the resulting design will be overly conservative. A sample, and is a description of its relative dispersion.
site-specific RBD therefore presents the engineering
professional with the potential opportunity to adopt lower Baecher and Christian (2003) present a summary of
partial factors, if: various numerical methodologies which can be adopted
• The engineering professional has the expertise to for the practical adoption of RBD, some of which are
undertake such an analysis, summarised below.
• Sufficient geotechnical investigation data are • First Order Second Moment (FOSM) method.
available to undertake a valid statistical assessment • Point Estimate method.
of the data, and • FORM.
• The relevant national design standard allows it. • Monte Carlo simulation.
Fenton et al. (2016) present a general methodology for Low and Tang (1997) present a more sophisticated
the adoption of RBD which is common to many reliability analysis of a shallow footing. As the
geotechnical design scenarios. This methodology methodology is similar to that set out in Low (2005) and
includes the following (albeit high-level) steps: discussed in Section 4.5, this is not discussed further
• Identify the limit state under consideration, for herein.
example bearing capacity of a shallow footing.
• Describe the relevant parameters (for example 4.4 Design example – pile group
effective cohesion, c’ and φ’) statistically by using
appropriate frequency distribution curves. Randolph and Buttling (2022) demonstrate how the
• Use a numerical model to assess for many Monte Carlo method, scripted using Python, can be
scenarios if the limit state is exceeded. applied to undertake a probabilistic analysis of a pile
• Assess the probability of failure based on the group analysed using the spreadsheet program PIGLET.
number of cases which result in the limit state being In the example presented by Randolph and Buttling
exceeded. (2022), a number of inputs to the pile group analysis are
treated as random variables, including the shear
Of the above steps, a typical geo-engineering modulus (both axial and lateral) of the soil in which the
professional will likely be least familiar with the piles are constructed, the axial capacity of the piles, and
assessment of frequency distribution curves and with the the normalised lateral pile head displacement at which
identification of a suitable numerical model to assess the the secant stiffness is 50% of the initial tangent value. As
limit states for the purposes of a reliability analysis. the analysis undertaken was a test case (and not a
design to be constructed), the range of the random
An assessment of an appropriate frequency distribution variables used as inputs to the probabilistic analysis was
requires a working knowledge of statistics. Ideally, the expressed through the use of published COV data,
geo-engineering professional would have a thorough rather than a statistical assessment of site-specific
understanding of the site conditions from an appropriate geotechnical investigation data.
site investigation, with many data points to allow an
appropriate frequency distribution to be identified (e.g. The Python subroutine was used to carry out 100,000
normal, log-normal, etc.). However, this is not always analyses for 38 load cases, which took approximately
practical and so the geo-engineering professional may 150 minutes on a standard computer. It was found that
use information presented in the literature in relation to the pile group lateral displacement limit of 50 mm
the mean and coefficient of variation of various soil (considered as a ULS case) was exceeded by 0.33% of
properties (refer Uzielli et al., 2006 for example). It is the simulations, giving a β value of 2.72. This compares
noted that the coefficient of variation (COV) is the with a value of 3.1 recommended in AS5104 (Standards
Australia, 2017b) for Class 2 structures with a high cost
of safety measures. A typical bridge such as a highway
Go to table of contents
11
Reliability-based geotechnical design in an Australian context A. L. E. Lochaden and S. Buttling
Go to table of contents
12
Reliability-based geotechnical design in an Australian context A. L. E. Lochaden and S. Buttling
• The improvements in recent years in the way in Fenton, G.A. and Griffiths, D.V. (2003). Bearing-capacity
which electronic site investigation data are gathered prediction of spatially random c - φ soils. Canadian
and manipulated has allowed significant efficiencies Geotechnical Journal, 40, pp. 54-65.
to be gained in the management of geotechnical Fenton, G.A., Naghini, F., Dundas, D., Bathurst, R.J. and
data. The time and budget gained through these Griffiths, D.V. (2016). “Reliability-based
improved efficiencies needs to be re-invested in geotechnical design in 2014 Canadian Highway
more sophisticated interpretation methodologies to Bridge Design Code”. Canadian Geotechnical
allow RBD to be undertaken. Journal, 53, pp. 236-251.
• Greater engagement with RBD by the Australian Green, R. (1989). Limit states design: some thoughts.
geo-engineering community. Proceedings of the Symposium on Limit States
• Increased co-operation between industry and Design in Foundation Engineering. Canadian
academia in relation to how RBD may be adopted in Geotechnical Society – Southern Ontario Section,
practice, e.g. the Recent Trends in Geotechnical Toronto, May 26-27, pp. 91-116.
and Geo-Environmental Engineering and Education Haberfield, C.M. (2017). Practical applications of soil
(RTGEE) workshops. structure interaction analysis. Gregory
• Development of an overarching Australian Standard Tschebotarioff Lecture 2017, ISSMGE. 19th
for geotechnical design, similar to that which exists International Conference on Soil Mechanics and
in Europe and the USA, which addresses the Geotechnical Engineering, Seoul, pp. 81-100.
concept of RBD. Hasofer, A.M. and Lind, N. (1974). Exact and invariant
second-moment code format. J. Eng. Mech.,
Subject to the above, the authors consider that RBD has 100(1), pp. 111-121.
the potential to be a highly useful design tool for the geo- Low, B.K. (2005). Reliability-based design applied to
engineering professional. Notwithstanding this, the retaining walls. Geotechnique, vol. 55 no. 1 pp. 63-
authors consider that, as described by Becker (1996a): 75.
• Whilst probability and statistics are useful tools if Low, B.K. and Tang, W.H. (1997). Automated reliability
properly applied, they must not become a substitute based design of footing foundations. Proceedings
for trying to understand the behaviour of of the 7th international conference on structural
geotechnical materials. safety and reliability. ICOSSAR ’97, Kyoto, Japan,
• Geotechnical design must not become overly vol. 3 pp. 1837-1840.
distracted by levels of safety and their quantification, Pells, P.J.N. (2011). “Against limit state design in rock”.
but instead focus on the understanding of the basic Tunnels & Tunnelling International, February
failure mechanisms and fundamental material edition, pp. 34-38.
behaviour. Randolph, M.F. and Buttling, S. (2022). “An investigation
of probabilistic analysis in relation to the design of
REFERENCES pile groups”. Proceedings of the 20th International
Conference on Soil Mechanics and Geotechnical
AASHTO (2017). “AASHTO LRFD bridge design Engineering, Sydney, Australia (awaiting
specifications”. Washington DC, USA. publication).
Baecher, G.B. Christian, J.T. (2003). “Reliability and Schweckendiek, T. Courage, W.M.G., and van Gelder,
Statistics in Geotechnical Engineering”. P.H.A.J.M. “Reliability of sheet pile walls and the
Becker, D.E. (1996a). “18th Canadian Geotechnical influence of corrosion – structural reliability analysis
Colloquium: limit state design for foundations. Part with finite elements”. Proceedings of the European
I. An overview of the foundation design process”. Safety and Reliability Conference (ESREL 2007),
Canadian Geotechnical Journal, 33, pp. 956-983. Stavanger, Norway, 25-27 June, pp. 1791-1799.
Becker, D.E. (1996b). “18th Canadian Geotechnical Standards Australia (2002). “Earth-retaining structures,
Colloquium: limit state design for foundations. Part AS 4678:2002”. Sydney, NSW.
II. Development for the National Building Code of Standards Australia (2009). “Piling – design and
Canada”. Canadian Geotechnical Journal, 33, pp. installation, AS 2159”. Sydney, NSW.
984-1007. Standards Australia (2011). “Residential slabs and
British Standards Institution (BSI) (1986). “Code of footings, AS 2870”. Sydney, NSW.
Practice for Foundations”. Standards Australia (2017a). “Bridge design, part 3:
British Standards Institution (BSI) (2004). “Eurocode 7: foundation and soil-supporting structures, AS
Geotechnical design”. 5100.3”. Sydney, NSW.
Canadian Standards Association (2014). “Canadian Standards Australia (2017b). “General principles on
Highway Bridge Design Code”. CAN/CSA-S6-06, reliability for structures, AS 5104”. Sydney, NSW.
Mississauga, Ontario. Uzielli, M., Lacasse, S., Nadim, F. and Phoon, K.K.
Cherubini, C. (1990). “A closed-form probabilistic (2006). “Soil variability analysis for geotechnical
solution for evaluating the bearing capacity of practice”. Characteristic and engineering properties
shallow foundations”. Canadian Geotechnical of natural soils, vol. 3, pp. 1653-1752.
Journal, vol. 27, pp. 526-529. Whitman, R.V. (1984). “Evaluating calculated risk in
Day, R.A. (2001). “Factored material properties and limit geotechnical engineering”. J. Geotech. Engrg.,
state loads – unlikely extreme or impossible 110, pp. 143-188.
pretence”. Geotechnical Engineering, vol. 149, no. Wong, P.K., Day, R.A. and Poulos, H.G. (2007) “Fifteen
4, pp. 209-210. years of geotechnical limit state design in Australia.
Day, R.A., Wong, P.K. and Poulos, H.G. (2007) “Fifteen Part II – foundations”. Proceedings of 10th
years of geotechnical limit state design in Australia. Australia New Zealand Conference on
Part I – soil retaining structures”. Proceedings of Geomechanics, Brisbane.
10th Australia New Zealand Conference on
Geomechanics, Brisbane, pp. 596-601.
Go to table of contents
Application of soil nail wall to roadway widening using GFRP rebars as per
Australian design guidelines
J. Lee1, PhD, PE, CPEng, FIEAust, E. Lee2, ME, MIEAust, M. Leong3, ME, M. Wei4, PhD, CPEng, FIEAust
1
Executive Director, GS E&C Australia, (formerly AECOM, Melbourne), Level 3, 100 Walker Street, North Sydney, NSW 2060; PH (02)
9135-2961; email: jeawoo.lee@gsenc.com
2
Senior Geotechnical Engineer, AECOM, Level 10, 727 Collins Street, Docklands, VIC 3008; PH (03) 9653-8705; email:
erin.lee@aecom.com
3
Design Manager, CPB Contractors, Level 6, 567 Collins Street, Melbourne, VIC 3008; PH (03) 9228-7000; email:
melvyn.Leong@cpbcon.com.au
4
Technical Principal, EIC Activities, Level 6, 567 Collins Street, Melbourne, VIC 3008; PH (03) 9228-7700; email:
michael.wei@eicactiv.com
ABSTRACT
This paper presents the application of soil nail wall technology to roadway widening. An arterial road upgrade project in
Melbourne consists of widening the freeway in the northern part of Melbourne, including massive cuts into a hillside on the
southern side of the existing freeway. Cuts up to 12 meters necessitated the use of retaining walls at 1H:10V batter to stay
within the right of way. A soil nail retaining wall was adopted to facilitate reduced excavation, less impact on the existing
slope, and improved construction speed with a top-down process. The soil nail retaining wall is 520 m long with a maximum
height of approximately 12 meters, including the undercut for pavement and drainage. The ground comprises clay fill
overlying residual clay and subsequent weathering profiles of basalt from Newer Volcanics group, partly overlying Brighton
Group sediments. Eleven boreholes were drilled sufficiently below the bottom of the wall. Laboratory tests were conducted
to estimate the soil and rock strength, including triaxial compression tests with pore pressure measurement to determine
effective strength parameters for Brighton group clayey soil. The design employed Glass Fibre Reinforced Polymer (GFRP)
bars to enhance the work efficiency by removing encapsulation of steel bars, the durability of which was reviewed for the
100 year design life with the proven data provided by the manufacturer. The soil nail retaining wall was designed as per
AS5100.3 and AS4678 guidelines selectively depending on the importance of the wall with reference to VicRoads
Specification Section 683 and FHWA-NHI-14-007.
Keywords: retaining wall, GFRP permanent nail, roadway widening, Australian design guidelines for soil nail walls
Go to table of contents
Application of soil nail wall to roadway widening using GFRP rebars as per Australian design guidelines J. Lee, E. Lee, M. Leong and M. Wei
Go to table of contents
Application of soil nail wall to roadway widening using GFRP rebars as per Australian design guidelines J. Lee, E. Lee, M. Leong and M. Wei
In addition, the traffic barriers adjoining the bottom of the 3.1 Design soil/rock parameters and bond
soil nail wall shall be designed to act as independent strength
systems or to prove no detriment to the soil nail wall in the
event of any collision with the traffic barrier where the Geotechnical material parameters adopted for design
barriers and the wall are integrated. To this end, the have been derived based on the available geotechnical
impact load is considered to be taken by the traffic barrier information, including both in-situ and laboratory tests,
designed as an independent system to the soil nail wall, and practically accepted correlations in literature (e.g.
hence no damage to the integrity of the wall or load Burt 2005), as summarised in Table 1. The highly
transfer to the wall. This is further discussed in Section weathered (Hw) to moderately weathered (Mw) basalt
3.3.1. strength parameters are determined based on the
empirical relations between the RMR class and Mohr-
2.4 Other Wall Features Coulomb failure criterion strength parameters and Mohr-
Coulomb strength parameters equivalent to Hoek and
The retaining wall requires a balustrade or other barrier to Brown strength for rock mass.
alleviate the risk of a person falling from the top of it as The soil nail bond strength was determined from the
depicted in Figure 4. The extension at the top of the soil lesser of empirical bond stress presented in literature
nail wall, which forms a safety barrier upstand, is such as CIRIA report C637 (CIRIA 2005) and effective
assumed to be constructed at the same time as the upper- stress based estimation employing the effective strength
most nails. This barrier upstand is subjected to wind design parameters in Table 1.
loading and handrail loading from AS5100 Cl 12.5 (a-c).
Go to table of contents
16
Application of soil nail wall to roadway widening using GFRP rebars as per Australian design guidelines J. Lee, E. Lee, M. Leong and M. Wei
The adhesion between the nail grout and Hw basalt for method as per AS 5100.3-2017, achieving a minimum
the rotary-drilled hole was determined with reference to global FoS of 1.8 in long-term
empirical literature charts such as those in FHWA Condition. The important structures sections include:
(FHWA 2003). The nail verification pull-out tests
conducted during the installation of nails have verified • located within 30 m of the existing road bridge
the design ultimate bond stress to be adequate. • located within 30 m of the existing transmission tower.
Seismic stability has been analysed based on the
3.2 Soil Nail Wall Design as per AS4678 and working stress method, achieving a minimum global FoS
AS5100 for Soil Nail Retaining Walls of 1.2. The stability analysis has taken into account
possible construction staging with 0.5 m excavation
The soil nail wall design has been completed below each row of nails prior to installing the subsequent
predominantly relying on the stability of soil nail walls, row of nails and 1.5 m over-excavation below the
which has been analysed using the 2D limit equilibrium pavement level at the end of the excavation included.
software Slide2 by Rocscience and the 2D finite element The short term analysis accounting for ‘during-
software PLAXIS 2020 with the strength reduction construction cases’, including those subject to the build-
method employed. up of temporary water pressure and seismic loading, has
been performed employing undrained soil design
Static stability design has primarily adopted the limit parameters, which have been switched to effective
state method with material strength partial factors and stress design parameters for the long term stability
load factors adopted in accordance with AS 4678-2002 analysis.
as summarised in Table 2 and Table 3, achieving a
minimum global “factor of safety” (“FoS”) of 1.0 in both As shown in Figure 5, the long term analysis has been
short-term (i.e. during construction) and long-term static carried out excluding the bottom row of soil/rock nails
conditions. behind the road protection barrier that is considered
sacrificial for collision so that the wall stability is
Table 2: Partial Factors as per AS4678 independent of the bottom row soil/rock nails. As a
Partial Factor Items Existing Fill In-situ Material result, the minimum global “FoS” has been calculated to
Effective friction 0.9 0.85 be between 1.1 and 1.4 for the sections designed as per
angle, ɸɸ AS4678 and 1.9 for those as per AS5100 in long term
Effective cohesion, 0.75 0.7 conditions.
ɸc
Undrained shear 0.5 0.5 On the other hand, the strength reduction method in the
strength, ɸuc finite element models has generated the minimum global
FoS slightly higher than that from the conventional limit
Table 3: Load Factors as per AS4678 equilibrium method. The working stress method analysis
Partial Factor Items Factor adopted against all the sections designed as per AS4678 or
Dead load behind wall 1.25 AS5100 has indicated the minimum global FoS greater
Dead load ahead of wall 0.8 than 1.5 specified as the design criteria in FHWA-NHI-
Traffic load behind wall 1.5 14-007 and CIRIA report C637.
The following design reduction factors have been The nails have been laid out at spacings of 1.5 m in soils
considered for the global stability analysis in as per and 1.8 m in Hw to Mw basalt with lengths up to 0.9 H at
AS4678-2002 or other standards as stated below: maximum. The hole diameter has been set to 125 mm
• Tendon structural capacity reduction factor = 0.4 as in soils and 105 mm in Hw to Mw basalt.
per VicRoads Section 683 (06.b.vii);
• Pull-out (bond) resistance reduction factor for rock
nails = ɸk x ɸb = 0.56;
where:
o Importance category reduction factor, ɸk = 0.8
(AS4678-2002 Table B1);
o Minimum material reduction factor ɸb = 0.7 (AS4678-
2002 Table B2) for the bond between rock and grout;
o Pull-out resistance reduction factor for soil nail = ɸn x
ɸb =0.63;
where:
o Structure classification design factor, ɸn = 0.9 for
structure classification C (AS4678-2002 Table 5.2); Figure 5. Soil Nail Wall Stability Analysis with the
and bottom nail excluded
o Minimum material reduction factor, ɸb = 0.7 (AS4678-
2002 Table B2) for the bond between soil and grout; 3.3 Deformation Analysis Considering
Interaction with Adjacent Structures
The bond capacity within the first 1 m length of nail has
been ignored in the design as per VicRoads Section 683. The Finite Element analysis undertaken has also
assessed the deformation of the soil nail walls, including
For the design sections in proximity to important the influence of collision loading and its impact on the
structures, the design has adopted the working stress existing structures in proximity. The maximum lateral
wall deflection has been estimated to be less than 0.3 %
of the wall height, which results in insignificant settlement
Go to table of contents
17
Application of soil nail wall to roadway widening using GFRP rebars as per Australian design guidelines J. Lee, E. Lee, M. Leong and M. Wei
Go to table of contents
18
Application of soil nail wall to roadway widening using GFRP rebars as per Australian design guidelines J. Lee, E. Lee, M. Leong and M. Wei
Go to table of contents
Risk Prediction Model for Formation of Underground Cavities and Sinkholes due to
Defective Sewer Pipes
S.R. Indiketiya1, P. Jegatheesan2 and R. Kuwano3
1
Former PhD Student, Swinburne University of Technology, Hawthorn, Melbourne, Vic 3122, Australia; email: samanthi.uom@gmail.com
2
Former Senior Lecturer, Swinburne University of Technology, Hawthorn, Melbourne, Vic 3122, Australia; email:
piratheepan.jegatheesan@yahoo.com
3
The University of Tokyo, Meguro-ku, Tokyo 153-8505, Japan; email: kuwano@iis.u-tokyo.ac.jp
ABSTRACT
Sinkhole formation due to internal erosion around defective sewer pipes is identified as a serious threat in urban
infrastructure system. Post-repair and rehabilitation after pipe failure are not effective as emergency pipe repairs are very
costly and pipe failure leads to various public and environmental consequences. Only a few studies have been conducted
on the prediction of the risk of ground erosion around pipe defects. Therefore, the main objective of this article is to propose
a model which can predict the risk of formation of sinkholes around sewer pipelines based on the weighted factors method
when a pipe defect is certain. The proposed methodology relies on different factors which contribute to void development
and severity of the consequences. The Risk of Erosion (ROE) combines the effect of Likelihood of Erosion void formation
(LOE) and Consequence of Erosion cavity formation and ground failure (COE). The LOE rating is related to many
parameters, including soil properties, hydraulic conditions, and pipe defect characteristics, while the COE rating is related
to the environmental, economic, and social consequences of pipe failure. Therefore, this model, which can predict the risk
of developing a sinkhole close to an existing pipe defect, will enable sewer assets maintenance teams to evaluate each
pipe and prioritize the maintenance and rehabilitation work based on the risk to each pipe.
1 INTRODUCTION rehabilitation projects and for sewer and storm water pipe
designers to design high stability backfill design
Wastewater and stormwater collection systems are considering the erosion resistance of the backfill.
critical components of urban infrastructure systems in any
country. Failure of deteriorated buried sewer pipes can 2 METHODOLOGY
occur in two ways. One process is when the soil backfill
in the vicinity of a pipe defect is gradually eroded, the pipe 2.1 Risk assessment method
will bend and collapse due to loss of confinement from the
ground (Balkaya et al. 2012). The second process occurs Most of the risk assessments conducted for sewer failures
when the cracks are formed at either sides or the crown as outlined in available literature have considered some
of the pipe. Voids can develop above the pipe while the form of likelihood or probability of failure, consequence of
pipe is still well supported from the bottom. In this case, failure and risk of failure. Emilio (2015) adopted a similar
erosion may propagate towards the ground surface method to predict the risk of failure of sewer pipes.
causing a sinkhole which eventually breaks most of the Therefore, modified version of the same approach is used
other buried service lines (Sato and Kuwano 2015). in this study to predict the risk of erosion void formation
(ROE) which combines the likelihood of erosion void
Many studies have been conducted on sewer formation (LOE) and consequence of erosion void
deterioration and risk prediction models for pipe failures formation (COE) in sewer pipes. The risk assessment
(Yan and Vairavamoorthy 2003; Baah et al. 2015; Emilio involved identifying influencing parameters for LOE and
2015). To the author’s best knowledge, only one risk COE then quantifying the LOE and COE by assigning
prediction model is available in the literature (Kaddoura critical scores based on the literature and by allocating
and Zayed 2017) which proposed a model that can predict factor weights based on experts’ opinion. After both
the risk of void erosion present outside sewer pipelines. ratings were determined, they were multiplied together to
This is based on weighted factors method using Fuzzy find the risk score for ROE as shown in Equation 1.
theory. However, this model has several limitations. It
does not account few a few other critical factors which ROE = LOE * COE (1)
control the process of cavity formation such as the relative
density or compaction of the ground and the influence of Following an extensive literature review, seven most
rainfall. Moreover, the erosion susceptibility of clay has influencing parameters which control the erosion initiation
been considered as higher than fine sand and silt which and progression through pipe defects were identified for
is contradicting with available literature (Rogers 1986, the LOE function as shown in Equation 2. The contribution
Indiketiya 2019). The model only predicts the probability from each factor to LOE, λi was estimated based on the
of erosion void presence outside sewer pipes and the risk expert's belief where, each 0 ≤ λi ≤ 1 and Σλi = 1.
due to consequences of erosion void formation is not
accounted for the overall risk. LOE = λ1.Soil type score +λ2.Relative density of
backfill score + λ3.Depth of sewer pipe score + (2)
Therefore, it will be useful to develop an efficient model λ4.Location of GWT score + λ5.Pipe defect size
score + λ6.Frequency and magnitude of sewer
which can predict the overall risk of developing erosion
exfiltration score + λ7.Frequency and magnitude
voids in defective sewers which can be useful to sewer of rainfall score
asserts management teams for allocating priorities in
Go to table of contents
Risk Prediction Model for Formation of Underground Cavities and Sinkholes due to Defective Sewer Pipes S. R. Indiketiya, P. Jegatheesan and R. Kuwano
The consequences of the erosion void formation and 2.2.3 The depth of the sewer
sinkhole formation are very similar to the consequences
of the pipe failures. Therefore, four parameters which Pipes which are closer to ground surface are often
have been commonly adopted in previous risk prediction disturbed and damaged due to heavy traffic, construction
models for pipe deterioration (Baah et al. 2015; Emilio works and maintenance of other underground services
2015) were selected for COE of this model. Selected four (Davies et al. 2001). In addition, O’Reilly et al. (1989)
factors to represent environmental, economic, and social found out that the defect rate decreases with the
impacts of the pipe location are depicted in Equation 3. increasing of the pipeline depth. The chance of erosion
Where, µi was estimated based on the expert's belief initiation and propagation also could increase at a lower
while satisfying 0≤µi ≤1 and Σµi = 1. depth due to lower confining pressure and faster rain
infiltration through the opening. Kaddoura and Zayed
COE = µ1.Environmental-pipe diameter score + (2017), also proposed a depth categorisation for sewer
µ2.Economic-commercial zone score) + (3) pipes in his void erosion prediction model. Based on the
µ3.Social-critical infrastructure score) + µ4.Social- above-mentioned discussion and studies, pipes are
traffic volume score)
scored based on it’s depth in LOE function as displayed
in the Table 3.
2.2 Assigning critically scores for LOE
Table 3. Scoring for depth of the pipeline
The condition of each pipe is evaluated for the LOE based The depth of the sewer (m) Attribute Score
on seven factors and each parameter was assessed and >4 Excellent 0
scored between 0 and 4 as given in Table 1 to Table 7. 2-4 Good 1
Therefore, the overall risk of LOE of pipe will also be 1.2-2 Fair 2
scored out of 4 as 0 ≤ λi ≤ 1 and Σλi = 1. as depicted in 0.6-1.2 Poor 3
Equation 2. The scoring criteria for each of the seven 0-0.6 Worst 4
factors are summarized through section 2.2.1 to 2.2.7.
2.2.4 Location of the GWT
2.2.1 Soil Characteristics If the GWL exists above the sewer pipeline or above any
structural defect, there is a higher possibility of infiltration
The soil type, its particle-size distribution and plasticity are and soil migrating to the sewer pipeline (Davies et al.
identified as key parameters which controls the initiation 2001). Indiketiya (2019) also discussed propagation of
of erosion through pipe cracks (Rogers 1986; Sato and erosion voids considering the geometry of the void and
Kuwano 2015; Indiketiya 2019). Considering all those the relative location of the GWT. Similar studies have
data, soil types were arranged into a hierarchy based on categorised the influence of GWT into two as pipe located
the erosion resistance through underground openings above and below the GWT (Rogers 1986; Davies et al.
and tabulated as shown in Table 1. 2001). However, the most important fact is the location of
the GWT relative to the pipe defect. As crack or the
Table 1. Scoring for soil type opening in a pipe can be at the bottom or the crown,
Soil type choosing the depth of pipe can be misleading for large
Attribute Score
(USCS specification) diameter pipes. Therefore, in this model, sewer depth was
GW, GC, CH Excellent 0 categorised and scored into three sections as shown in
CL-CH, SC, GM Good 1 Table 4.
SW, CL, GP Fair 2
SM, CL-ML, MH Poor 3 Table 4. Criticality scoring for the GWT
SM, ML, SP, Dispersive Clay Worst 4 Groundwater table Attribute Score
Well below the pipeline Excellent 0
Closer to the pipeline Poor 3
2.2.2 Relative density of the backfill Above the pipeline Worst 4
Go to table of contents
Risk Prediction Model for Formation of Underground Cavities and Sinkholes due to Defective Sewer Pipes S. R. Indiketiya, P. Jegatheesan and R. Kuwano
Table 5. Criticality scoring for the pipe defect The parameters include pipe diameter, distance from
B/Dmax B/D85 Attribute Score commercial zones, distance from critical infrastructure,
<0.75 <1 Excellent 0 and vehicular traffic volume as parameters for quantifying
0.75-1 1-2 Fair 2 the environmental, economic, and social consequences.
1-2 2-3 Poor 3 The scoring criteria for each of the four factors above are
>2 >3 Worst 4 summarized below in Section 2.3.1 to 2.3.4.
Go to table of contents
Risk Prediction Model for Formation of Underground Cavities and Sinkholes due to Defective Sewer Pipes S. R. Indiketiya, P. Jegatheesan and R. Kuwano
2.3.3 The distance from critical infrastructure As this theme is still quite new and thorough knowledge
about the pipe deterioration and erosion process is
Critical infrastructure enables the provision of essential required to fill up the questionnaire, 15 researchers were
services such as food, water, health, energy, selected from various countries referring to their
communications, transportation, emergency services and publications which indicates their expertise on this study
banking. Sinkholes, pipe breaks and the subsequent area. All the information required for survey was provided
repairs may influence a city’s ability to respond to and experts were requested to assign λi and µi values
emergencies. Emilio (2015) also counted schools as corresponding to parameters from their expertise so that
critical infrastructure in his model, as schools are large ∑(λi) = 1 for LOE and ∑(µi) = 1 for COE separately.
public gathering points that may be negatively affected.
The same scoring breakdown used by Emilio (2015) is 2.5 Data analysis and determination of weighted
applied in this COE model as shown in Table 10 since influence factors
pipe failures have very similar consequences to sinkhole
events. The proposed λi and µi values by experts were scattered
in a wide range. Therefore, the weighted average (Mean)
Table 10. Scoring for the distance from critical and the mode of each data set for λi and µi are calculated
infrastructure and presented in Table 12 and Table 13 respectively. The
Distance from critical Attribute Score mean and the mode of all seven λi values are close and
infrastructure (m) consistent. Therefore, final λ values were designed clearly
>2500 Excellent 0 as closely following the mean and the mode values while
2499 - 2000 Very Good 1 satisfying the condition of ∑(λi) =1.
1999 - 1500 Good 2
1499 - 1000 Fair 3
However, all four µi, parameters are having higher
999-500 Poor 4
standard deviations while mean and the mode values are
<500 Worst 5
also slightly different. Therefore, considering the mean,
mode and the distribution of the normal frequency,
2.3.4 The traffic volume/road type of the
judgment on final value was decided as presented in
location
Table 13. Emilio (2015) had the same parameters in the
COE model and the judgment was µ1=0.3, µ2=0.3, µ3=0.2
The type of the road or the volume of traffic where the
and µ4=0.2. These have been decided manually based on
sinkhole appears are crucial which affect the
judgment where more emphasis was applied to social
consequences of the event. However, as the traffic
concerns because they directly affect the public.
volume highly depends on the location, the city and the
country, it is difficult to normalise the volume of traffic as
Therefore, the final equations for LOE and COE with the
an indicator for general practice. Therefore, road type
proposed influencing factors can be written as shown
would be beneficial as it is a relative gauge which can be
below in Equations 4 and 5.
easily replaced for different countries. There is a variety
of classifications for road types, and no single variable is
Table 12 The Mean, mode and proposed λi values
available to completely describe a class of roads. The
λ
road classification implemented throughout Victoria Parameter
Mean Mode Proposed
according to (Austroads 2005) are used here. According λ1 0.19 0.2 0.2
to this hierarchy, “M” routes deserve a higher critical score λ2 0.11 0.1 0.1
as it carries the highest traffic volume and has greatest λ3 0.09 0.1 0.1
connectivity within the road network. The scoring system λ4 0.18 0.2 0.2
proposed in this study is given in Table 11. Footpaths are λ5 0.21 0.2 0.2
also added with the lowest critical score as some sewer λ6 0.12 0.1 0.1
pipes are buried in footpaths which are not included in the λ7 0.10 0.1 0.1
national road numbering system. ∑( λ i) 1.0
Table 11. Criticality scoring for the road type Table 13 The Mean, mode and proposed µi values
Distance from critical µ
Attribute Score Parameter
infrastructure (m) Mean Mode Proposed
Footpaths Very Good 1 µ1 0.26 0.3 0.3
C Good 2 µ2 0.25 0.2 0.2
B Fair 3 µ3 0.25 0.3 0.25
A Poor 4 µ4 0.25 0.2 0.25
M Worst 5 ∑( µ i) 1.0
2.4 Data acquisition for weighted influence factors LOE = 0.2*(Soil type score) +0.1*(Relative density
of backfill score) + 0.1*(Depth of sewer score) +
Two of the recommended methods for estimating 0.2*(Location of GWT score) + 0.2*(Effect of pipe
(4)
defect size score) + 0.1*(Frequency and magnitude
influence factors (λ and µ values) in Equation 2 and 3 is
of sewer exfiltration score) + 0.1* (Frequency and
to get the expert's judgment (Joseph et al. 2010) or magnitude of rainfall score)
iterative computation through field data (Emilio 2015;
Kaddoura and Zayed 2017). As collecting actual field data COE = 0.3 * (Environmental-pipe diameter
related to defective sewer pipes and sinkhole events are score) + 0.2 * (Economic-commercial zone (5)
extremely difficult, multiple experts are selected by score) + 0.25 * (Social-critical infrastructure
considering the experience and confidence on the topic. score) + 0.25 * (Social-traffic volume score)
Go to table of contents
Risk Prediction Model for Formation of Underground Cavities and Sinkholes due to Defective Sewer Pipes S. R. Indiketiya, P. Jegatheesan and R. Kuwano
Since the LOE is scored out of 4 and COE is scored out and the location of the GWT have the highest relative
of 5, this model considers pipes with LOE of 3 to 4 as the weights.
higher vulnerability for ground subsidence with pipe Four parameters which evaluates the consequences
failure and in need of urgent attention. Similarly, COE is of erosion void formation were considered as the pipe
scored out of 5 and pipes with COE score of 4 to 5 are diameter, the distance from commercial zone, the
considered as critical pipes which causes many social, distance from critical infrastructure and the road type
economic and environmental consequences. Various
of the location
studies have graphically illustrated the risk matrix system
(Baah et al. 2015) which combines the probability and • Criticality scores for LOE function was scored out of 4
consequences for an event. Referring those studies, risk and COE function was scored out of 5. Therefore, the
level for a pipe to develop an erosion cavity was overall risk of erosion, ROE is predicted rating out of
categorised through the risk matrix based on the score as 20. Then the level of risk for a defective pipe to
low, moderate, high and very high as illustrated with a develop an erosion cavity is categorised according to
colour code in Table 14. a risk matrix based on the ROE score: low (0 to 4),
moderate (5 to 9), high (10 to 14) and very high (15 to
3 VALIDATION 20).
• Unfortunately, due to lack of published data, and the
The model can be validated one of these methods: (1) legal requirements of different institutes and countries,
comparison of the predicted risk with that of previous
it proved impossible to access real data to validate the
models, (2) validation based on real field data or (3)
model. However, this model development contributes
implementation of the model in a case study. Detailed
reports of investigations of sinkhole accidents are significantly to existing knowledge. Therefore, it can
extremely rare, as the relevant authority’s primary goal is be validated and improved with collaboration from
to restore the infrastructure as quickly as possible to industry partners in a future study.
minimise the public, economic and social inconvenience. • Ultimately, this allows defective pipes to be ranked
General information about the events is usually published based on the risk of erosion void formation and
in media as news reports. All the attempts to access some consequences which assist to organise a
field data to validate the model from local and maintenance and rehabilitation schedule by allocating
international authorities were unsuccessful as the present the priority.
research is a university-based study without any industrial
collaborations.
5 ACKNOWLEDGEMENTS
The ROE model presented in this chapter was developed
following a thorough study of previous research related to Authors would like to acknowledge:
erosion through pipe defects, risk prediction models for • Swinburne University of Technology in supporting this
pipe failures and the experimental investigations available research by offering Swinburne University
in literature. The output of COE is closely aligned with Postgraduate Research Award.
Emilio 2015). Nevertheless, the LOE function needs to be • All the research experts who contributed for the
validated before implementing it in industrial applications. survey to determine the weightage factors for the
model.
4 CONCLUSIONS • Chadwick Geotechnics in providing the financial
support for publishing the article.
Based on the study, the following conclusions were
driven. REFERENCES
• Seven key parameters which affect the likelihood of
erosion void formation are identified. Of those Austroads. 2005. Towards a Nationally Consistent
parameters, the soil type, the size of the pipe defect Approach to Route Numbering. Available from
Go to table of contents
24
Risk Prediction Model for Formation of Underground Cavities and Sinkholes due to Defective Sewer Pipes S. R. Indiketiya, P. Jegatheesan and R. Kuwano
Go to table of contents
SESSION 2
INSTRUMENTATION
AND MONITORING
IN DESIGN
Keynote Address
Foundation QA Pty Ltd, Managing Director, POBox 4280, Croydon Hills, Victoria, Australia, 3136; email: julian@foundationqa.com
1
Golder Associates Pty Ltd, Foundation/Pile Testing Department, Building 7, Botanicca Corporate Park, 570 – 588 Swan Street,
2
ABSTRACT
Piling design and verification is a fraught and risky business. The spread of pile capacity estimates submitted to conference
predictions exercises is often staggering and sobering. This underlines why design of driven piles does not stop at the
design engineer’s desk but continues through construction, and relies on the valuable information provided by the
installation process. Each installation blow is a test - a test of the ground response to hammer input delivered into the pile.
Traditionally, pile capacity has been interpreted from this input-response relationship through various and many pile driving
formulae. Five decades ago, measurement systems were first used to measure and interpret the stress waves in piles
generated from the hammer inputs and reflected from the ground response to infer capacity in a more sophisticated and
reliable way using wave mechanics principles. Today, PDA testing and wave matching are routinely accepted practice.
However, each PDA test has direct relevance only to the individual pile which is tested. This paper will argue that our
fundamental task as designers and supervisors is to establish ground truth, by synthesizing the results of PDA tests into a
locally-evidenced and locally-targeted dynamic formula. Therefore, only dynamic formulae, properly modified and
correlated, must be the vehicle for delivering local ground truth and ultimately being the basis for sign-off. On a foundation-
wide basis, the role of PDA tests is critical but subservient, and principally to provide the evidence on which a correlated
dynamic formula is developed. Consequent implications for the foundation sign-off process, and for a proposed new
approach to establishing capacity reduction factors for driven piles will also be discussed.
Keywords: pile driving formulas, pile acceptance, PDA testing, capacity reduction factors, wave equation analysis
Go to table of contents
27
Ground truth, control and design of driven piles: implementing old ways with a new twist J. P. Seidel and D. Denes
Fifty years ago, electronic systems were first used to subsequent argumentation regarding the role of dynamic
measure and interpret the stress waves in piles generated pile formulae.
from the hammer inputs and reflected from the ground
response to infer capacity in a more sophisticated and In the analysis phase of our pile design case, activities
reliable way using wave mechanics principles (Rausche include:
et al., 1972; Goble et al., 1975). Today, PDA testing and • Desktop studies
wave matching are routinely accepted practice • Site investigation
(Hannigan, et al., 2016). But prior to 1985, pile driving • Insitu testing
acceptance was based only on static load testing and pile • Laboratory testing
driving formulae.
The synthesis phase activities include:
Our fundamental task as designers and supervisors is to • Development of simplified site stratigraphy
ensure that capacity and integrity of each pile installed • Assigning characteristic properties to the layers
meets the demands of the structure that it supports. To • Selecting and applying a pile design method
do this we need to establish a ‘ground truth’, albeit a
• Considering the structural loads to establish a
locally-evidenced and locally-targeted ground truth
design pile size and length
capable of application to all piles in a foundation system.
Dynamic formulae, properly modified and correlated,
Let’s suppose that the pile is a floating pile in a deep clay
must be the vehicle for delivering local ground truth and
layer, and that the design method adopted is the
ultimately being the principal basis for sign-off.
𝛼𝛼 method, in which the pile adhesion, 𝜏𝜏 at any depth is
computed based on the local undrained shear strength,𝑐𝑐�
It is important to remember that each PDA test has direct
and an adhesion factor, 𝛼𝛼
relevance only to the individual pile which is tested. On a
foundation-wide basis, the role of PDA tests is critical but 𝜏𝜏 � 𝛼𝛼𝛼𝛼� (1)
subservient, and principally to provide the evidence of
ground truth on which a correlated dynamic formula is Figure 2, reproduced after Coduto (1994), is typical of
founded. design charts which relate 𝛼𝛼 to 𝑐𝑐� . The Coduto curve is
similar to the design chart which was included in the
Consequent implications for the foundation sign-off Appendix to AS2159 (1978).
process, and for a new approach to establishing capacity
reduction factors for driven piles will also be discussed. Such design curves are often used by designers in the
belief that they are based on some universal truth,
2 ANALYSIS AND SYNTHESIS unaware of the significant scatter which lies behind this
design relationship.
The engineering design process can be generalized to
comprise two phases Figure 3, includes the 124 test results from which the
Coduto relationship was developed. Based on the
1. An analysis phase which includes collecting reference tests alone, it can be seen that if a designer
evidence, organization and processing of data uses this design curve, they could easily overestimate or
2. A synthesis phase in which the data collected and underestimate adhesion by a factor of 2.
analysed is transformed into a connected design
In this section, we will consider the design for a project to The development of this design curve is the embodiment
be supported on driven piles, and the process of of the synthesis process, in which the engineer
designing a single pile to support a column load. necessarily has to reduce the diverse test results into a
design algorithm, despite the up to four-fold range of the
The purpose of this section is only to provide an analogy data on which the algorithm is based.
to a process that is well understood as a basis for
Go to table of contents
28
Ground truth, control and design of driven piles: implementing old ways with a new twist J. P. Seidel and D. Denes
Improved predictions of 𝛼𝛼 are possible if some of the Figure 6. Simulated adhesion relationship for specific
influencing factors are taken into account. Thus Nowacki geology and pile type
et al. (1992) proposed the design chart shown in Figure
5. More sophisticated interpretations of data sets can 2.2 Key findings
lead to more reliable assessments with reduced
uncertainty. To summarize, the key findings (kf) from this review are :
Putting aside whether the 𝛼𝛼 method is the most suitable kf1. Geotechnical design relationships are generally
design method, it is clear that the variability of results will based on data sets with wide scatter
reduce as the number of variables decreases. By kf2. Geotechnical engineers using these
reducing the scope of a relationship to a specific site, or a relationships must understand and account for
small region within the same geological setting, the the inherent uncertainty in their design (with
possibility of developing a more reliable relationship appropriate reduction factors)
increases. Further, if that data is collected for a single pile kf3. Empirical relationships are by nature not
universal, but are specific to the particular
Go to table of contents
29
Ground truth, control and design of driven piles: implementing old ways with a new twist J. P. Seidel and D. Denes
Because of the traditional use of the Hiley Formula in This design line shown in Eq (4) and Figure 7 is equivalent
Australia, only the Fragaszy results in relation to this to the 𝛼𝛼 design line in Figure 3, and demonstrates the
formula will be discussed here. same key findings kf1, kf2 and kf3 (see Section 2.2).
The Hiley formula is based on a simple theoretical model Although on average the predictions have low bias (the
of pile driving characterized as the inelastic collision with average ratio of Hiley to static capacity is 0.985), the
coefficient of restitution, 𝑛𝑛 of 2 masses (the hammer, 𝑊𝑊� scatter is very wide which is reflected by the coefficient of
which drops from a height ℎ with drop efficiency 𝑒𝑒� and variation of 72.6%.
the pile and helmet, 𝑊𝑊� ). The pile experiences a
permanent set, 𝑠𝑠 , and three components of transient Given that the basic premise of monitoring the installation
movement, 𝐶𝐶. Equation (2) predicts the pile capacity, 𝑅𝑅� . of piles is to reduce the uncertainties inherent in pile
design, such a high coefficient of variation undermines the
𝑒𝑒� 𝑊𝑊� ℎ 𝑊𝑊� � 𝑛𝑛� 𝑊𝑊� very premise on which pile monitoring is based.
𝑅𝑅� � . (2)
𝑠𝑠 � �𝐶𝐶� � 𝐶𝐶� � 𝐶𝐶� � 𝑊𝑊� � 𝑊𝑊�
The basis of this equation is to estimate the striking
hammer energy, determine the energy transferred to the
pile, and then to equate that to the work expended in the
transient and permanent movements of the pile. With the
exception of the Gates Formula (Gates, 1957), pile driving
formulae are based on some simplified energy balance
equation.
1
And in practice are generally assumed rather than measured
Go to table of contents
30
Ground truth, control and design of driven piles: implementing old ways with a new twist J. P. Seidel and D. Denes
Fragaszy et al. do not further analyse the data of Figure 7 Ramey and Johnson (1978) report a data set of 153 piles
in order to determine whether the spread is affected by comprising steel-H, steel pipe, concrete and timber piles.
pile type, hammer type or soil type. Five dynamic formulae, including Hiley Formula, are
compared with static tests. All things being equal, one
The difficulties of comparing end of drive installation data would expect the Hiley/Static capacity ratio to be relatively
(on which driving formulae are based) with static load highest for the piles installed in cohesionless soils (which
tests undertaken some days or weeks later have been are generally not known for setup), and lowest for the
discussed elsewhere (Seidel, 2015a). The phenomenon piles exclusively in cohesive soils (for which end of drive
of pile setup is well documented and is particularly capacity may be significantly less than the capacity
associated with post-installation pile capacity increases in measured later during static load testing).
cohesive soils (Skov and Denver, 1998; Lee et al., 2010).
Figure 8. Hiley capacity vs Static capacity after Ramey and Johnson (1978)
Figure 8 shows that the spread of data is similar to the This resistance comprises two parts – a static component,
Fragaszy data set. However, the data broadly supports 𝑅𝑅� , and a transient dynamic component 𝑅𝑅� . thus
the expected differentiation of test results by soil type,
𝑅𝑅� � 𝑅𝑅� � 𝑅𝑅� (5)
with the spread of cohesive soil data generally being
below the non-cohesive data, and showing the influence As discussed in Seidel (2018a), the relative proportion of
of pile set-up. Key findings kf 4 and kf 5 are demonstrated static and dynamic components of total driving resistance
in Figure 8. vary with the ease of driving. For hard driving, with low
set, the total resistance is dominated by the static
As noted in regard to the design parameter, 𝛼𝛼 insightful component. As driving becomes progressively easier
consideration of the influencing factors (in that case (increasing set), the proportion of dynamic resistance
overconsolidation ratio and plasticity index), can reduce increases as indicated schematically in Figure 9.
uncertainty.
Referring to Equation (3), the correction function, 𝜒𝜒2 will
In order to further reduce the uncertainty in the application be relatively small at low sets, and progressively increase
of dynamic formulae, it is important to understand that as pile sets increase. 𝜒𝜒 can either be evaluated by wave
dynamic formulae do not predict static pile capacity but equation analysis prior to piling, or evaluated empirically
rather predict the total resistance of the soil to driving, 𝑅𝑅� . from dynamic pile testing, particularly at the
commencement of a project.
2
𝜒𝜒 is referred to as Dynamic Reduction Function (DRF) in Seidel (2018)
Go to table of contents
31
Ground truth, control and design of driven piles: implementing old ways with a new twist J. P. Seidel and D. Denes
Go to table of contents
32
Ground truth, control and design of driven piles: implementing old ways with a new twist J. P. Seidel and D. Denes
these results into a unifying relationship from which a pile 3.1 Pile acceptance based on energy, set and
acceptance criterion can be developed. Locally temporary compression
correlated driving formulae provide the ground truth, and
are fundamental to providing the basis for acceptance of Despite the certainty provided to the evaluation of 𝜒𝜒 ,
any untested pile. there still remain challenges in evaluation of static
capacity, 𝑅𝑅� , in Equation (3) which is reproduced here for
3 PILE ACCEPTANCE convenience.
𝐸𝐸𝐸𝐸𝐸𝐸
The current practices in Australia for driven pile 𝑅𝑅� = (3)
acceptance vary considerably. Generally, piles are driven 𝜒𝜒. (𝑠𝑠 + 𝐶𝐶 2)
to a capacity rather than a penetration criterion. What
might be considered a normal practice is that a There are a range of techniques for measurement of pile
percentage of piles are PDA tested and matched with set and temporary compression. The possibilities, their
wave equation analysis, and this percentage defines the benefits and limitations are discussed in Denes et al.
adopted capacity reduction factor, 𝜙𝜙� in accordance with (2021).
the provisions of Section 4 of AS2159 (2009).
However, the most significant challenge to
AS2159 (2009) introduced the concepts of a basic implementation is 𝐸𝐸𝐸𝐸𝐸𝐸, the energy transferred to the pile.
geotechnical reduction factor, 𝜙𝜙�� , an intrinsic test factor, Although some modern hammers measure and report
𝜙𝜙�� , and a testing benefit factor, 𝐾𝐾, which allowed the kinetic impact energy, 𝐾𝐾𝐾𝐾, these represent only a small
capacity reduction factor, 𝜙𝜙� to vary depending on the percentage of the existing hammer fleet. Even for these
percentage of piles tested, 𝑝𝑝.3. For a typical project with hammers, the challenge still remains to evaluate the loss
low perceived risk category and low redundancy 𝜙𝜙�� = of energy, 𝐾𝐾𝐾𝐾𝐾𝐾𝐾𝐾𝐾𝐾𝐾, through the helmet, particularly for
concrete piles, for which the pile cushion can be highly
0.56, the computed capacity reduction factors are as
variable in thickness and stiffness, especially when the
indicated in Table 1:
cushion properties change during driving.
Table 1: AS2159 (2009) Effect of increasing Flynn and McCabe (2016) compare 𝐸𝐸𝐸𝐸𝐸𝐸 and 𝐾𝐾𝐾𝐾
percentage of PDA tests measurements for hydraulic hammers on 5 projects, and
𝑝𝑝 (%) 3 5 10 25 demonstrate the uncertainties with inferring 𝐸𝐸𝐸𝐸𝐸𝐸 from
𝜙𝜙� (−) 0.69 0.72 0.76 0.80 hammer energy measurements, as shown in Figure 13.
The Flynn and McCabe data indicates that the percentage
Table 1 suggests that by the very act of testing, a reduced of hammer energy delivered to the pile typically varied
risk is imparted to the pile foundation system. The between 75% and 100%, but in isolated instances was as
benefits of testing are manifest, and higher percentages low as 50%. This variability is just the transfer variability.
of testing will increase the likelihood of detection of any The variability of the hammer stroke and kinetic energy is
piles with insufficient capacity. However, there is an not even included in this assessment.
implied assumption in AS2159 (2009) that the installation
of the 75% to 97%4 of piles not tested is informed by the For uninstrumented hammers, transferred energy can
results of the tests. This is generally not true, or is not only be estimated on the basis of estimated ram drop
implemented in a rigorous way. height, assumed hammer drop efficiency, and an
assumed loss of energy in transferring from the hammer
One common practice is to treat the untested piles as to the pile. As just noted, this can be particularly
unrelated to the tested piles. The tested piles must challenging for concrete piles.
demonstrate a capacity in excess of the contractual
requirement. The untested piles must pass a predefined Seidel (2018b) presents the energy measurements for a
acceptance criterion in the specification, such as the project in which 155 PDA tests were undertaken on
traditional Hiley Formula. In this approach, the valuable prestressed concrete piles over the extent of a large
learnings of the PDA tests do not inform acceptance of bridge project. The project is distinguished by the
the untested piles. requirement for all piling and all PDA tests to be
undertaken with a drop height of 0.5m. Despite this
The second common practice is to adopt a maximum set control on drop height, delivered energies averaged
criterion which corresponds to the set measured for a 69.2kJ (88% efficiency), but with a range of 37.5 to 91.0kJ
tested pile with the target capacity, or inferred if the tested (48% to 116%5 efficiency). The coefficient of variation of
capacity exceeded the target. This implicitly assumes EMX was 15.4%. The three projects that the paper details
that delivered energy efficiency is constant and that the demonstrate energy ranges of 36% and 40% for the steel
set-capacity relationship is fixed with no variation. We will pile projects, and 77% for the concrete pile project.
see in the next sections that neither of these assumptions
is correct. Li et. al., (2022) report in regard to another case study
“The driving hammer energy efficiency recorded
It will be shown (see Figure 14) that transferred energy variations of up to 40%, casting doubt on the use of
cannot be assumed to be constant. Figures 10 to 12 achieved set as a means of inferring pile capacities”.
demonstrate that the set-capacity relationship is not
constant. Any acceptance criterion must take into The histogram in Figure 14 presents the sequential EMX
account the demonstrated variability of both the energy values for 51 of the 155 PDA tests.
delivered and of the set-capacity relationship.
3 5
Up to a limit of 𝑝𝑝 = 25(%) of piles PDA tested Efficiency greater than 100% is physically impossible, indicating that the
4 true drop height was significantly in excess of 0.5m
i.e. (100 – 𝑝𝑝)%
Go to table of contents
33
Ground truth, control and design of driven piles: implementing old ways with a new twist J. P. Seidel and D. Denes
The only possibility for individual pile energy implied geotechnical reduction factors based on a
measurements6 is from high frequency remote non- transferred energy uncertainty of 5%7.
contact displacement measurements. Peak pile
velocities can be computed from differentiating the It is noted that the 𝜙𝜙� value of 0.70 is derived exclusively
displacement-time signal. It is possible to identify low from a statistical assessment of the pile testing results
energy blows from low velocity measurements. Although and hammer performance and is independent of any
EMX is related to the square of peak velocity VMX, the assessment of the basic geotechnical reduction factor,
relationship is affected by variations in cushion stiffness, 𝜙𝜙�� . This is a logical approach, because assessment of
so that no unique and reliable energy can be computed in the installation process should be independent of any of
most cases. the risk factors considered in developing project average
Figure 15 is a representation of Figure 12, with upper and risk rating, 𝐴𝐴𝐴𝐴𝐴𝐴 and 𝜙𝜙�� in Table 4.3.2(C) of AS2159
lower 95% confidence limits shown. Also shown are the (2009).
6 7
Other than PDA testing or attachment of an accelerometer to each pile, The uncertainty in transferred energy evaluation, either from
both of which have practical issues instrumented hammer or pile velocity measurements should be
determined from each case.
Go to table of contents
34
Ground truth, control and design of driven piles: implementing old ways with a new twist J. P. Seidel and D. Denes
Figure 15. Confidence limits and inferred geotechnical reduction factors for Figure 11 data
3.2 Pile acceptance based on energy and set Three key points are shown on the graph at blow counts
of 67, 200 and 870 blows/m (15mm, 5mm and 1.15mm
Pile acceptance methods based on energy and set alone set). What this graph shows is that for any given energy,
are fundamentally premised on use of bearing graphs the capacity at 15mm set will be 50% less than the
correlated to field test results. Bearing graphs represent capacity at 5mm set, and the capacity at 1.15mm set will
the relationship between capacity and pile set, which is of be 50% more than the capacity at 5mm set. This
course dependent on hammer energy. Seidel (2015b) relationship is unique to this data set, but equivalent
and Denes et al. (2021) discuss the importance of relationships will hold for any site.
Energy-Capacity-Movement (ECM) relationships.
Go to table of contents
35
Ground truth, control and design of driven piles: implementing old ways with a new twist J. P. Seidel and D. Denes
To better understand why a non-dimensional equation Figure 18. Gates blow-count capacity modifier
performs well, consider Equation (6) for kinetic energy 𝐾𝐾𝐾𝐾
of the pile driving hammer with mass 𝑀𝑀� , which strikes
with velocity 𝑣𝑣��� , The purpose of the parametric study was to investigate
the factors affecting normalized capacity with a view to
�
𝐾𝐾𝐾𝐾 = 12 𝑀𝑀� 𝑣𝑣��� (6) developing reliable predictions of the relationship
between normalized capacity9 and pile set. Only
from which it follows that moderate success was achieved, and a reliable prediction
√𝐾𝐾𝐾𝐾 ∝ 𝑣𝑣��� (7) method is yet to be developed.
Similarly, the energy transferred to the pile, 𝐸𝐸𝐸𝐸𝐸𝐸 is Figure 19 shows the results for 4 of the cases
computed from pile-top PDA measurements of force, 𝐹𝐹(𝑡𝑡) investigated. These cases include concrete and steel
and velocity, 𝑣𝑣(𝑡𝑡) as piles, and span ratios of (a) ram weights of 10.4, (b) pile
impedances of 3.5; (c) damping factors of 3.4 and (d)
𝐸𝐸𝐸𝐸𝐸𝐸 = � 𝐹𝐹. 𝑣𝑣. 𝑑𝑑𝑑𝑑 (8) quakes of 2.1. Despite these large variations in hammer
One dimensional wave mechanics provides the following and pile parameters, the indicated normalized capacity-
relationship which holds between F and v at the set relationships have a limited range. Furthermore, in
commencement of the dynamic event when no reflections practice, the range of this relationship appears to be more
from the soil have reached the pile head. limited than indicated by Figure 19, because the
parametric study included hammer/pile combinations
𝐹𝐹 = 𝑍𝑍. 𝑣𝑣 (9) which could be considered extreme and outside the
bounds of practical experience.
where 𝑍𝑍 = 𝐸𝐸𝐸𝐸𝐸𝐸 is the pile impedance, a function of the
pile modulus, 𝐸𝐸, pile cross-section, 𝐴𝐴 and pile wavespeed, Regardless of whether it will be possible to reliably predict
𝑐𝑐. these relationships in advance with a simple algorithm,
predictions for any project can be made in advance using
From Equations (8) and (9) it follows that there is an wave equation analysis.
(imperfect) relationship between √𝐸𝐸𝐸𝐸𝐸𝐸 and pile-top
velocity, 𝑣𝑣, and pile-top force, 𝐹𝐹.
Seidel (2018c) presented results from a parametric study Figure 19. Wave Equation study of Normalized
of 194 Wave Equation simulations involving 20 different Capacity – Set relationship
piling hammers ranging from 7 to 168 tonne ram weight;
25 concrete or steel pile sections varying from 0.2m
Furthermore, any predicted relationship should be
square concrete to 3m diameter and 80mm wall thickness
calibrated against the results of testing once piling
steel; cushion stiffnesses, short and long piles, and
commences. It is recommended that for the initial pile
varying resistance quakes, damping factors and
installations, PDA testing and wave matching analyses be
resistance distributions. The study therefore ranged
undertaken at a range of pile sets – for instance at 10mm,
across an extreme range of piling scenarios.
5mm and 2mm set per blow.
8
For example FHWA-Gates, modified FHWA-Gates or WSDOT formulas
9
Capacity normalized by the impact force, 𝑅𝑅� 𝐹𝐹���
Go to table of contents
36
Ground truth, control and design of driven piles: implementing old ways with a new twist J. P. Seidel and D. Denes
Figure 20 is an example drawn from a project covering practical, however, peak impact velocity, 𝑣𝑣��� , can be
multiple dispersed bridge sites. The data includes both used as a proxy for 𝐹𝐹���
end of drive and restrike tests, but is constrained to a using Equation (11).
single pile geometry (1200mm diameter steel pipe piles)
and hammer system. 𝐹𝐹��� = 𝑍𝑍. 𝑣𝑣��� (11)
Figure 20 also shows the upper and lower 95% It is possible to measure 𝑣𝑣��� with high frequency remote
confidence limits, and the corresponding geotechnical pile measurement devices (see Denes et al., 2021).
reduction factors, 𝜙𝜙� , which vary depending on pile set
and the spread of data. For typical acceptance criteria The basis for the popularity of the Gates formula can be
with sets below 5mm /blow, 𝜙𝜙� is in excess of 0.7 and as seen in the similarity of the capacity-set relationships
much as 0.74. shown in Figures 18, 19 and 20. The empirical success
of the Gates formula reinforces the essential validity of a
Of course, the relationship is premised on measurement force-based, but locally correlated acceptance criterion.
of impact force, 𝐹𝐹��� , for every (untested) pile. This is not
Go to table of contents
37
empirical pile acceptance relationship, the reliability of the capacity. Transp. Res. Record No. 1105: Struct.
energy, force and movement measurements and the Foundations, Transp. Res. Board, pp.32-40.
number of tests on which the relationships are based. Lee, W., Kim, D., Salgado, R. and Zaheer, M., 2010. Setup of
driven piles in layered soil. Soils and foundations, 50(5),
These approaches do not rely on any evaluation of the
pp.585-598.
basic geotechnical reduction factor defined in AS2159 Li, I., Vinod. M. and Hsi, J. 2022. Case Study: Lessons learned
(2009), which is in itself a logical outcome. from large scale pile driving in Waikeria, New Zealand.
Proc. 20th International Conference on Soil Mechanics and
5 ACKNOWLEDGEMENTS Geotechnical Engineering, Sydney 2022 (submitted for
review).
The authors acknowledge the important contributions of Long, J.H., 2009. Comparison of five different methods for
their ex-colleagues at FSG Geotechnics and determining pile bearing capacities. Wisconsin Highway
Research Program.
Foundations, in particular Pearse Casey, Ben Kroenert,
Olson, R.E. and Flaate, K.S., 1967. Pile-driving formulas for
Nick Medley, Alireza Foroughi and Hossein Ahmadi, to friction piles in sand. Journal of the Soil Mechanics and
the development and refinement of the novel pile Foundations Division, 93(6), pp.279-296.
acceptance methods presented here. Oxford University Press. 2021 Available at:
https://www.lexico.com/definition/wake
Paikowsky, S., and LaBelle, V. 1994. Examination of the Energy
Approach for Capacity Evaluation of Driven Piles.
6 REFERENCES Proceedings of the International Conference on Design
and Construction of Deep Foundations, December 6–8,
Orlando, FL. FHWA, Vol. II, pp. 1133–1149
Allin, R., Likins, G. and Honeycutt, J., 2015. Pile driving formulas
Ramey, G.E. and Johnson Jr, R.C., 1979. Relative accuracy and
revisited. In IFCEE 2015 (pp. 1052-1063)
modification of some dynamic pile capacity prediction
AS2159 (1978) SAA Piling Code. Rules for the design and
equations. Ground Engineering, 12(6).
installation of piling. Standards Australia
Rausche, F., Liang, L., Allin, R. and Rancman, D., 2004.
AS 2159-2009 (2009). Piling – Design and Installation.
Applications and correlations of the wave equation
Standards Australia
analysis program GRLWEAP. In Proceedings, VII
Bond, A.J., Schuppener, B., Scarpelli, G., Orr, T.L., Dimova, S.,
Conference on the Application of Stress Wave Theory to
Nikolova, B. and Pinto, A.V., 2013, June. Eurocode 7:
Piles (pp. 107-123).
geotechnical design worked examples. In Workshop
Rausche, F., Moses, F. and Goble, G.G., 1972. Soil resistance
“Eurocode (Vol. 7).
predictions from pile dynamics. Journal of the soil
Broms, B. and Choo, L.P., 1988. A simple pile driving formula
mechanics and foundations division, 98(9), pp.917-937.
based on stress-wave measurements. In Proc. of the 3rd
Seidel, J.P., 2015a. Overview of the Role of Testing and
Intl. Conf. on the Application of Stresswave Theory to Piles
Monitoring in the Verification of Driven Pile Foundations. In
(pp. 591-600).
Proceedings on the 12th Australia New Zealand
Cambridge University Press 2021. Available at
Conference on Geomechanics (ANZ2015).
https://dictionary.cambridge.org/dictionary/english
Seidel, J.P., 2015b. Enhanced Use of Dynamic Pile Testing in
Chellis, R.D. (1961) ‘Pile Foundations’. 2nd (ed). New York:
Foundation Engineering. In Proceedings on the 12th
McGraw-Hill Book Company, Inc.
Australia New Zealand Conference on Geomechanics
Cherubini, C. and Vessia, G., 2007. Reliability approach for the
(ANZ2015).
side resistance of piles by means of the total stress
Seidel, J.P., 2018a. The normalized bearing graph and dynamic
analysis (α Method). Canadian Geotechnical Journal,
reduction function concepts in pile acceptance. DFI-EFFC
44(11), pp.1378-1390.
International Conf. on Deep Foundations and Ground
Coduto, D.P. 1994. Foundation design, principles and practices.
Improvement, Rome, Italy 6-8 June, 2018.
Prentice Hall Inc., Englewood Cliffs, N.Y.
Seidel, J.P., 2018b. The importance of energy evaluation on an
Denes, D., Foroughi, A. and Seidel, J.P. 2021. Pile Testing
individual pile basis. DFI-EFFC International Conf. on
Verification – an Alternative Approach. Australian
Deep Foundations and Ground Improvement, Rome, Italy
Geomechanics Society Victoria Seminar, Oct. 2021.
6-8 June, 2018.
Fellenius, B.H., 2013. Capacity and Load Movement of a CFA
Seidel, J.P., 2018c. Mobilization of Pile Capacity
Pile: A Prediction Event. In Foundation Engineering in the
and Pile Acceptance. Invited presentation to Australian
Face of Uncertainty: Honoring Fred H. Kulhawy (pp. 707-
Geomechanics Society, Victorian Chapter, 8 August, 2018.
719).
Skov, R. and Denver, H., 1988. Time-dependence of bearing
Flynn, K. and McCabe, B.A., 2016, August. Energy transfer ratio
capacity of piles, Proc. 3rd International Conference on
for hydraulic pile driving hammers. In Proc., Civil
Application of Stress-Wave Theory to Piles, Canada, 1–10.
Engineering Research in Ireland Conf.
Fragaszy, R.J., Argo, D. and Higgins, J.D., 1989. Comparison of
formula predictions with pile load tests. Transportation
Research Record, 1219, pp.1-12.
Gates, M., 1957. Empirical Formula for Predicting Pile Bearing
Capacity, Vol 27, No. 3, March : 65-66
Goble, G.G., Likins Jr, G. and Rausche, F., 1975. Bearing
capacity of piles from dynamic measurements (No. OHIO-
DOT-05-75 Final Rpt.).
Groom, D. 2019. An Assessment of Dynamic Pile Driving
Formulae for use by Military Engineers
Hannigan, P.J., Rausche, F., Likins, G.E., Robinson, B.R. and
Becker, M.L., 2016. Geotechnical Engineering Circular No.
12–Volume II Design and Construction of Driven Pile
Foundations (No. FHWA-NHI-16-010).
Hiley, A., 1930. Pile-driving calculations. The Struct. Eng, 8.
Kulhawy, F.H. and Phoon, K.K., 2002. Observations on
geotechnical reliability-based design development in North
America. In Foundation Engineering in the Face of
Uncertainty: Honouring Fred H. Kulhawy (pp. 142-159).
ASCE.
Lawton, E.C., Fragaszy, R.J., Higgins, J.D., Kilian, A.P. and
Peters, A.J., 1986. Review of methods for estimating pile
Go to table of contents
The Application of 3D Finite Element Method in the Design of Large Piled
Foundation System - Case Study: Melbourne Cement Facility
ABSTRACT
This paper provides an overview of the foundation design and analysis process carried out for the Melbourne Cement
Facility silo located in Port Melbourne, Victoria, Australia. The proposed silo is a cylindrical multi-compartment cement
storage facility supported on a 2.6m thick concrete ring beam with an external diameter of 38.5m. The ring beam is
supported by a piled foundation system comprising 155 CFA piles in an annular pile layout. The site is underlain by
Quaternary Sediments of Yarra Delta which are further underlain by Werribee Formation. This paper describes a detailed
soil-structure interaction analysis performed using the finite element program PLAXIS 3D, which was used to assess the
foundation performance with particular attention to global and differential settlement of the pile group. The study evaluated
the complex load sharing between the piles and the ring beam, and the differences in load mobilisation between piles
within the group. The results of this study highlight the capability of 3D FEM analysis for obtaining an optimised foundation
design solution and understanding and addressing various technical challenges associated with silo foundation systems
of this type.
Keywords: piled raft, piled foundations, foundation design, finite elements, storage facilities, settlement analysis
Go to table of contents
The Application of 3D Finite Element Method in the Design of Large Piled K. Ranjbar Pouya, B. Collingwood and A. Judi
Foundation System - Case Study: Melbourne Cement Facility
Figure 1. Site location and the proposed location of the new silo - source: Google Earth
The silo structure is supported on a 2.6m thick annular assessment of the contribution of the piles and ring beam
ring beam which is shown in Figure 2. Loads coming to the foundation performance will be discussed in
from the superstructure will be transferred from the subsequent sections. The overall dead load and live load
cylindrical silo walls to the ring beam and the piles of the silo structure is in the order of 900 MN in total. The
constructed underneath. The foundation system is load critical load cases that were adopted in foundation
designed to perform as a piled raft as the ring beam has modelling and design are summarised in Table 1.
a large dimension (and contact area) and can contribute
to the load-bearing of the system. A quantitative
Serviceability limit state load cases SLS-1 and SLS-2 foundation system which will be discussed in more detail
were used for settlement assessment. It was assumed in the subsequent sections.
that all silo compartments are fully loaded for SLS-1. It is
understood that some compartments of the silo can be 3 GROUND CONDITIONS
either fully loaded or completely empty at times.
Therefore, SLS-2 was adopted to account for the worst- According to the Geological Survey of Victoria 1: 31, 680,
case scenario in terms of differential settlements when the construction site is located in an area of recent
half of the silo is empty (DL only) and the other half is Quaternary sedimentation which forms part of the well-
fully loaded (DL+LL). Ultimate state limit load cases ULS- known Yarra delta. Yarra delta sediments at the site are
1 and ULS-2 were used to check the generated actions present beneath a fill layer, and comprise Port
in piles against the structural capacity. ULS-1 was also Melbourne Sand (PMS), Coode Island Silt (CIS),
used to check the overall geotechnical stability of the Fishermen’s Bend Silt (FBS) and Moray Street Gravel
(MSG). The Yarra Delta sediments are further underlain
Go to table of contents
The Application of 3D Finite Element Method in the Design of Large Piled K. Ranjbar Pouya, B. Collingwood and A. Judi
Foundation System - Case Study: Melbourne Cement Facility
by Werribee Formation and Silurian age Siltstone of the information from both borehole and CPTs. The ground
Melbourne Formation. The groundwater level has been conditions were found to be reasonably consistent with
reported consistently at a depth range of 2.2m to 2.7m our general expectation of the Yarra Delta profile as
below ground surface, consistent with the adjacent river described in many existing references, for example,
water levels. Ground models were developed based on Ervin (1992). A summary of the adopted ground model is
the available site investigation data, which includes presented in Table 2.
4 FOUNDATION DESIGN The foundation system has the ring beam supported on
155 No. 900mm diameter reinforced concrete CFA piles
The large depth to siltstone rock meant that the piled with pile toes founded in the Moray Street Gravel (MSG)
foundations needed to be founded in the overlying soil unit. The layout for the CFA piles includes five rows of
profile at depths that could be readily achieved by an piles from inside to outside of the ring beam. The pile
available piling plant. The size and significance of the silo spacing varies between rows, but averages about 2.1m.
structure, and the significant loading it imposes on the
founding soils, meant that both geotechnical strength Available site investigation data suggested that the MSG
and serviceability requirements had to be carefully is comprised of interbedded sand and clay layers. To
evaluated. account for the lithological variation of interbedded
granular and cohesive sublayers within the MSG unit and
According to the adopted serviceability design criterion, the possibility of encountering clay lenses at the toe of
calculated foundation settlements under the some of the piles, a series of sensitivity analyses were
serviceability load combinations (SLS-1 and SLS-2) performed to evaluate the foundation performance with
should not exceed the allowable settlements. Allowable an appropriate range of ultimate base resistance values.
settlements for silo were specified as below:
• Total settlement is not to exceed 300mm. Appropriate pile design parameters for shaft friction and
• Differential settlement under Load Case SLS-2 is base resistance were determined in consultation with
not to exceed 60mm from one side of the silo to the Wagstaff Piling, based on previous experience and
other. available load test results from previous tests in similar
ground profiles, ensuring that the selected values were
Design for geotechnical strength was based on the limit reasonably expected to be achievable at the site for CFA
state design principles documented in the Australian piles installed with the available equipment. Table 3
Piling Code (AS 2159-2009). This requires piles to be presents the unit stresses that were adopted in our
designed in such a way that the design geotechnical modelling.
strength (which is the ultimate geotechnical strength
multiplied by an appropriate geotechnical strength Table 3: Summary of pile resistance parameters
reduction factor) will be not less than the design action Depth (m) Ultimate
Ultimate
effect. However, it has been well documented that Shaft
Unit Base Stress
applying such geotechnical criteria to each individual pile Friction
From To (kPa)
within a pile group can result in considerable (kPa)
conservatism in the design (e.g., Poulos, 2017). This is Fill 0 3.5 0 -
acknowledged in AS 2159-2009 in clause 3.2.2, which PMS 3.5 6.5 20 -
allows individual piles within the group to be overloaded CIS 6.5 8 20 -
for the ULS load case provided that the ultimate FBS 8 17 50 to 80 -
geotechnical design strength of the pile group satisfies
MSG 34 40 100 to 150 3000-5000*
the requirements of the code. To this end, the ULS-1 load
* Sensitivity checks were carried out for the range of base
divided by an appropriate Geotechnical Strength stresses presented which cover the expected range of
Reduction Factor (φg), was modelled with the outcomes in practice.
acceptance criteria being that the pile group must
maintain overall stability. This is indirectly a check on the Using the estimated unit stresses, the ultimate
overall factor of safety for the foundation system design geotechnical strengths of piles considered in the
and stability under this load case demonstrates that the modelling process ranged between 9.2MN and 11.3MN,
requirements of AS2159-2009 are met with respect to depending on the various founding assumptions.
strength.
Go to table of contents 39
The Application of 3D Finite Element Method in the Design of Large Piled K. Ranjbar Pouya, B. Collingwood and A. Judi
Foundation System - Case Study: Melbourne Cement Facility
Simulation of piles in a sufficiently accurate and realistic A linear elastic perfectly plastic model with Mohr-
manner is a key aspect of a piled foundation analysis. Coulomb failure criteria was adopted for all soil layers.
There are two distinct methods in PLAXIS for modelling Effective stress analyses were carried out using drained
piles in a 3D analysis. One involves modelling piles as material properties to evaluate the long-term settlement
volume elements with a user-defined interface. Although behaviour of the foundation system. Long term
considered to be the more robust approach, the consolidation over the design life of the structure was
application of volume elements can be extremely also considered, by estimating the secondary
demanding in terms of processing time for a large compression of soils over the 50-year design life of the
foundations system with a considerable number of piles. structure. The adopted geotechnical modelling
parameters are summarised in Table 2. To ensure
The alternative option, which is adopted in this study, is loading was accurately distributed to the foundation
the application of embedded beam elements in which the system, loads were converted to equivalent pressures on
piles interact with the surrounding soil by means of top of the ring beam. Figure 3 shows the model geometry
special embedded interface elements. Lee et al. (2010) and pile arrangements.
6 RESULTS AND DISCUSSION be because a significant proportion of the SLS loads are
being resisted by shaft friction over the length of the
Table 4 summarises the estimated total and differential piles, and shed to the surrounding ground. As a
settlements for the foundation system under SLS load consequence, the load is distributed to a large soil mass
cases, for each of the adopted founding levels and end and small differences in founding level and end bearing
bearing assumptions. It can be inferred from the data in do not have a significant impact on settlements. Figure 4
Table 4 that the settlement performance of the presents the predicted absolute and differential
foundation system is not greatly sensitive to the range of settlement contours for load case SLS-1 and SLS-2 for
assumptions that were used. We consider this is likely to founding condition case 4 specified in Table 4.
Go to table of contents
The Application of 3D Finite Element Method in the Design of Large Piled K. Ranjbar Pouya, B. Collingwood and A. Judi
Foundation System - Case Study: Melbourne Cement Facility
Figure 4. Absolute and differential settlement contours for load cases SLS-1 and SLS-2
The load sharing behaviour between the piles and the unnecessary over-conservatism in the design by
ring beam is summarised in Table 5, as the ratio of the designing all piles to the highest individual pile load.
sum of the pile head loads over the total load being
applied on the foundation system. This is sometimes
referred to as the combined piled raft foundation
coefficient (Katzenbach et al. 2005). Table 5 indicates
that the superstructure loads are mostly carried by the
piles, with a bearing contribution of about 80%, and the
remaining 20% of the loads are transferred to the subsoil
directly through the ring beam contact area.
Go to table of contents 41
42
The Application of 3D Finite Element Method in the Design of Large Piled K. Ranjbar Pouya, B. Collingwood and A. Judi
Foundation System - Case Study: Melbourne Cement Facility
design, our approach was to average the individual pile to develop an optimised design solution for the piling
loads across a selection of adjacent piles for each row of scheme. The nature of the load transfer between the
piles. Table 6 presents the average maximum axial pile piles and the ring beam was able to be assessed, as well
head loads across inside, middle and outer rows of piles as the distribution of loads to individual piles and inner,
under SLS-1 and ULS-1 load cases. It is noteworthy that middle and outer pile rows. The analysis indicated that
the outside row piles attract significantly higher loads the distribution of vertical loads was highly non-uniform
compared to the middle row and inner row piles. among piles in different rows, which can be explained by
the relative displacements experienced by piles in each
Table 6: Summary of calculated pile head loads row relative to the surrounding soil. The study illustrated
the importance of considering the foundation system as
Case Maximum axial pile head load (kN) a group, in order to avoid over-conservatism in the pile
Outside Row = 9413 design.
SLS - 1 Middle Row = 3159
Inside Row = 5765 8 ACKNOWLEDGEMENTS
Outside Row = 10050
The authors wish to acknowledge the contribution and
ULS - 1 Middle Row = 5747
assistance of Melbourne Cement Facility and Fitzgerald
Inside Row = 6591
Construction (Australia) in the execution of the study and
for allowing the publication of this technical paper.
To better understand the load sharing mechanism
between piles in different rows, it is necessary to REFERENCES
consider the differential movements between these piles
and the surrounding soil. Figure 6 shows a cross-section AS 2159-2009 Piling - Design and installation, Standards
of the vertical settlement contour plot through the Australia.
midsection of the silo from the PLAXIS analysis. Burland, J. B., Broms, B. B., and De Mello, V. F. B. (1977).
“Behaviour of foundations and structures.” Proc. 9th int.
This figure needs to be interpreted in conjunction with the Conf. on Soil Mech. & Found. Eng., 2, 495-546.
tabulated values in Table 6. As shown in Figure 6, the Davis, E. H., and Poulos, H. G. (1972). The analysis of piled raft
soil mass enclosed inside the ring of piles settles much systems. Aust. Geomech. J 2, 21-27.
Ervin, M.C., (1992), Engineering properties of Quaternary age
more compared to the soil mass outside the footprint of sediments of the Yarra Delta, Seminar, Engineering
the silo. Due to the relative rigidity of the ring beam, the Geology of Melbourne, 245–260.
absolute settlements of all piles are very similar. Katzenbach, R., Arslan, U., Moorman, C., and Reul, O. (1998).
Therefore, the outer row piles settle more relative to the “Piled raft foundation: interaction between piles and raft.”
surrounding soil mass, which results in mobilisation of Darmstadt Geotechnics (Darmstadt University of
higher shaft and end bearing resistances in these piles. Technology), 4, 279-296.
In contrast, the soil mass inside the silo footprint Katzenbach, R., Schmitt, A., and Turek, J. (2005). “Assessing
Settlement of High-Rise Structures by 3D Simulations.”
undergoes a much larger settlement and the
Computer-Aided Civil and Infrastructure Engineering, 20,
displacement of the inner row piles relative to the soil is 221-229.
correspondingly lower; hence, less resistance is Lee, S. W., Cheang, W. W. L., Swolfs, W. M., and Brinkgreve,
mobilised in the inner row piles. The middle row piles R. B. J. (2010). “Modelling of piled rafts with different pile
mobilise the least resistance, as the soil mass around models.” Numerical Methods in Geotech. Eng. Benz &
these piles is being loaded by adjacent piles in both the Nordal (eds), Taylor & Francis Group, 637-642.
inner and outer rows; hence, middle row piles experience Mandolini, A., Russo, G., and Viggiani, C. (2005). “Pile
the least relative pile/soil displacement. foundations: experimental investigations, analysis, and
design.” Proc. 16th Int. Conf. on Soil Mech. & Geotech.l
Eng., 12–16 Sep, Osaka, Japan, 177-213.
Poulos, H. G. (1991). “In computer methods and advances in
geomechanics (eds Beer et al.), 183-191. Rotterdam:
Balkema.
Poulos, H. G. (1994). “An approximate numerical analysis of
pile-raft interaction.” Int. J. for Numerical and Analytical
Methods in Geomechanics, 18, 73-92.
Poulos, H. G. (2017). “Tall Building Foundation Design.” CRC
Press, Taylor & Francis Group.
Randolph, M. F. (1983). “Design of piled foundations.” Proc. Int.
Symp. on Recent Development in Lab. & Field Tests &
Analysis of Geotech Problems, Bangkok, 525-537.
Randolph, M. F. (1994). “Design methods for pile groups and
piled rafts: state-of-the-art report.” Proc. 13th Int. Conf.
Soil Mech. Found. Eng., New Delhi 5, 61-82.
Russo, G., and Viggiani, C. (1998). “Factors controlling soil-
Figure 6. Vertical settlement contour plot structure interaction for piled rafts.” Intern. Conf. on Soil-
Structure Interact. in Urban Civil Eng., Ed. Katzenbach,
7 CONCLUSION R., and Arslan, U., Darmstadt, 79-102.
Smith, I. M., and Wang, A. (1998). “Analysis of piled rafts.” Int.
This paper presented the analysis undertaken as part of J. Numer. Anal. Meth. Geomech., 22, 777-790.
the design of the foundation system for a large cement Ta, L. D., and Small, J. C. (1996). “Analysis of piled raft systems
in layered soils.” Int. J. for Numerical and Analytical
silo storage facility. A satisfactory outcome was achieved
Methods in Geomech., 20, 57-72.
in terms of both geotechnical strength and serviceability Viggiani, C. (1998). “Pile groups and piled rafts behaviour.”
performance. The study showed that 3D FEM analysis Proc. 3rd Int. Geot. Seminar on Deep Foundations on
can be employed to understand the group performance Bored and Auger Piles, Ghent, 77-94.
of the foundation system and the distribution of load to Viggiani, C., Mandolini, A., Russo, G. (2012). “Piles and Pile
foundation elements. The results were used successfully Foundations.” Spon Press, Taylor & Francis Group.
Go to table of contents
Pile Testing Verification – an Alternative Approach
ABSTRACT
Dynamic pile testing is undertaken for a number of reasons including: 1. To confirm that the pile meets serviceability and
geotechnical capacity requirements; 2. To assess pile integrity, either during installation (driven piles) or after construction
(cast-insitu piles) and 3. To verify that the piling hammer delivers the energy required to satisfy the design criteria and that
stresses during testing are kept within acceptable limits. In addition, testing allows us to establish and calibrate acceptance
criteria - relationships such as resistance vs set curves, and/or correlated pile driving formulas. These relationships are
premised on the interrelationship of capacity (C), transferred energy (E) and pile movement (M) which is represented
primarily by pile set. These ECM relationships allow capacity to be inferred from measurement of transferred energy and
pile movement and are used to infer the capacity of untested piles. However, for a variety of reasons, transferred energy
can vary significantly between piles which, being undetected, undermines the reliability of ECM relationships. An alternative
approach to using ECM relationships is proposed based on pile set and pile force (F). We demonstrate through parametric
studies and review of project data, that these FCM relationships are reliable alternatives which bypass the problems with
variable energy transfer. Of course, impact force will also vary with hammer performance, but impact force can generally
be accurately determined from the measured impact velocity as a proxy. Pile velocity can be measured by attaching a
single accelerometer to the pile, or by using a high frequency displacement monitoring device. FCM-based acceptance
criteria have the significant advantage that both the necessary force (F) and displacement (M) inputs can be verified by
simple measurements on all untested piles.
Keywords: Pile Testing, Pile Verification, Dynamic Formulae, PDA, Case Study
Go to table of contents
Pile Testing Verification – an Alternative Approach D. Denes, A. Foroughi and J. Seidel
The focus of this paper is on driven displacement piles bearing relationship. The bearing graph may be further
which allow a wider range of testing opportunities, calibrated based on observed field results taken during
including high strain dynamic testing (commonly referred end of initial drive or restrike testing. At the end of the test
to as PDA testing). PDA testing has been used in piling process, the aim is to use the pile test results to
Australia and New Zealand since the early 1980’s. PDA produce a relationship that reliably predicts resistance of
testing has substantial cost and time advantages relative the untested piles to the tested piles.
to static testing, and this allows more testing to be
conducted for the same testing budget. PDA testing The result of Wave Equation analysis is a relationship
therefore allows statistically significant, and widely between pile capacity and pile set, which is energy
distributed testing to be conducted – a significant benefit dependent – i.e. a form of ECM relationship. This is
compared to traditional static testing. traditionally presented as one or more curves relating
capacity to blow count (blows/metre) called a bearing
Typically, less than 15% of piles on a project are tested, graph, as shown in Figure 1.
leaving the remaining 85% of piles untested. The
verification of the untested piles can be done in several The problem with this traditional form, is that at very low
ways. For example, by simply comparing installation data pile sets (high blow counts), blow counts become
of untested piles to installation data of tested piles and excessively high and the relationship below 1.0mm set
ensuring that the untested piles have been driven to (1000 blows/m) is difficult to evaluate. Therefore, in the
similar or harder conditions. In practice, dynamic following, the bearing graphs shown will relate capacity to
formulae, such as the Hiley formula (Hiley, 1930) are set, as shown in Figure 2 which represents the same data
often used for this verification. Hiley formula is shown in Figure 1.
recognised in AS2159-2009 and is also in the Auckland
Structural Group Piling Specification (2002). However, 3.1 Set and Hammer Energy Measurements
academic debates from as early as 1941 (American
Society of Civil Engineers, 1941) and to this date (Allin, In order to use the bearing graph for pile capacity
2015) continue to discuss the inaccuracies of the use of verification, measurements of the pile set and hammer
dynamic formulae. Allin recommends using wave energy are required. Different methods can be used to
equation analyses instead of generic dynamic formulae measure pile set, with various degrees of accuracy. The
for the assessment of untested piles. Seidel (2015) traditional method used to measure set involves a person
proposes the use of site-specific dynamic formulae, which (a) standing beside a pile and marking a piece of paper
are to be calibrated based on test data, which has been attached to the pile
further investigated with success (Damen & Denes 2017,
Denes & Kroenert 2019). Nevertheless, the use of
dynamic formulae in projects may lead to mis-guided
results and inaccuracies and should be used with a
correct understanding and informed approach.
Go to table of contents
Pile Testing Verification – an Alternative Approach D. Denes, A. Foroughi and J. Seidel
subsequent to the final 10 hammer blows; (d) onboard Using Seidel’s observation, and assuming a 5mm set per
piling hammer instrumentation (such as the iPilier system blow, if we apply a +/- 33% variation to the 44 kNm energy
by Junttan or NDT lazer & computer system); (e) counting example in Figure 2, the resistance acting on the pile can
the number of hammer strikes for the pile to move a range between 2200 kN and 3050 kN which is an
certain distance (blow count); (f) non-contact high unacceptably large range.
frequency displacement monitoring devices, such as the
Pile Driving Monitor (PDM) which has sub-0.1mm 4 AN ALTERNATIVE APPROACH - FORCE
accuracy.
As noted, an alternative approach to using the standard
It can be unreliable to rely on options (a) to (d), especially ECM relationship has been developed using the pile set
when pile set is small. Ideally, more focus towards set and the force applied to the top of the pile. This FCM
measurement using digital equipment should be approach was first discussed and presented by Seidel
embraced, as anything that involves manual pile marking (2018).
and/or counting is likely to introduce an element of error.
In simple terms, the general principle that applies to piling
Set is usually measured over a sequence of 10 blows. It is similar to driving a nail into timber. For a nail to drive
is important that these blows be consistent so that the into timber, the force applied must overcome the force
average set is meaningful. Any ‘warm up’ blows to reach resisting the nail to move. If the nail does not move under
the specified drop height should be discarded from the force of the hammer, one can assume that the force
consideration. In summary, set is a fundamental resisting the nail to movement is equal to or greater than
measurement used in all ECM relationships, so it is the force from the hammer. On the other hand, if the nail
important that set measurement is reliable. moves, then resisting force is expected to be less than the
force applied by the hammer. This is often seen with pile
It is noted that temporary compression measurement is testing. If the pile movement is zero under a hammer
used in the Hiley Formula. This can only be measured strike, one can infer that the resistance acting on the pile
using methods (a) and (f) above. Method (a) is not is equal to or greater than the applied force.
recommended for safety reasons, as discussed.
Based on the principles of one-dimensional wave theory
Hammer energy for an untested pile can be estimated pile force and velocity are related by pile impedance for
from the potential energy of the ram which is a function of as long as there is only a single wave in the pile. Under
the weight and drop height. The drop height of a hammer those conditions, the following holds true.
is usually recorded for the full sequence of pile driving
(referred to as pile driving record) which can be used to F = vZ (1)
estimate the energy applied to the pile. The actual
delivered energy to the top of the pile is often less than where F is the pile-top force, v is the pile-top velocity and
the potential energy of the ram due to frictional losses in Z is a constant of proportionality called the pile
the hammer, energy losses in the hammer and pile impedance, and
cushion, energy losses during impact, and restrictions on
energy transfer that are functions of the pile material, Z = EA/c (2)
cross-section, length and driving resistance. There have
been several papers that document and emphasize the where E is the pile modulus, A is the cross-sectional area,
high variability in energy transfer from impact hammers and c is the pile wave-speed.
and caution for the sensitivity of driving formulae and
wave equation analyses to this parameter (Allin (2015), During impact, and generally up to the time of maximum
Seidel (2015b), Flynn & McCabe (2016)). Given the impact force and velocity, this proportionality between F
sensitivity of driving formulae to this parameter, a and v can be reliably assumed. Therefore, knowing Z, the
calibration using either measurements of actual energy in peak pile-top impact velocity can be used as a proxy for
the test piles or an energy efficiency factor based on measurement of peak impact force.
previous performance of the hammer is usually
recommended for estimation of hammer energy. it is important to distinguish between capacity and
However, the performance of a hammer, including the resistance in relation to pile driving. The resistance to
efficiency, can vary from pile to pile and over time. penetration that the pile experiences when struck by the
Various factors such as hammer alignment, pile-hammer hammer is called the driving resistance. Driving
contact surface and hammer maintenance frequency can resistance comprises a dynamic component due to the
also affect the energy transferred to the pile which in turn velocity of the pile movement, and a static component,
affect the accuracy of the bearing graph correlation. which is the (static) capacity of the pile at the time of
Further to this, there can be errors in identifying the actual impact, and would be the expected capacity if it were
hammer drop if it is undertaken visually when the hammer possible to undertake an instantaneous static load test.
is in operation.
The parametric study and case studies described relate
Seidel (2015) has reported in one comprehensive study to the relationship between pile capacity, C, and peak
for a major project that the measured energy observed impact force, F. The capacity/force (C/F) ratio is
from PDA testing on the same pile type (and hammer) dependent on hammer, soil and pile characteristics, and
varied by +/- 33% about the median energy. The first is also shown to be a function of pile set, being maximum
author’s experience has often seen variation in energy at low set, and progressively decreasing as pile set
ranging between +/- 15 %. This demonstrates the increases (and dynamic resistance represents a
uncertainty associated with estimation of the delivered progressively larger proportion of the driving resistance).
energy to untested piles.
Figures 1 and 2 presented the results of a GRLWEAP
analysis for a 9t hammer hitting a 400mm square concrete
Go to table of contents
Pile Testing Verification – an Alternative Approach D. Denes, A. Foroughi and J. Seidel
pile from drop heights of 0.4m, 0.6m and 0.8m. The three In each case, the CAPWAP results were normalized using
curves in each Figure relate capacity to blow count or set the PDA-measured peak impact force and the data
for each of the three drops/ energies. The parametric presented as data points on a C/F ratio vs set plot.
study takes this one step further by normalizing the
capacities by peak impact force. The FCM relationship has also been computed using
GRLWEAP based on the site hammer, pile and soil
The same data presented in Figures 1 and 2 is shown in characteristics using representative resistance
Figure 3 in normalized form – i.e. the FCM relationship - distribution, damping and quake values determined in the
C/F ratio vs set. What is striking is that all three curves CAPWAP analyses.
reduce to a single and unique relationship between C/F
ratio and set, regardless of drop height. 5.1 Case Study One – Open Ended Steel Tube
Piles
5 CASE STUDY CORRELATIONS Figure 4 compares the C/F ratio vs set data from PDA
tests and the FCM relationship from GRLWEAP analysis.
Two case studies will be presented for historical piling At pile sets less than 1mm per blow, the inferred capacity
projects. PDA testing equipment from Pile Dynamics is approximately equal (96%) to the peak impact force. At
(PDA-PAX and 8G systems) were used to collect the pile sets of 5mm, the achieved capacity is only 70% of the
force and velocity data. The data was analysed using peak impact force; at 10mm the ratio reduces to 55% and
CAPWAP to determine pile capacity. at 15mm pile set the ratio drops further to 44%.
It is clear that there is very good correlation between the 2. The FCM approach offers a piling control and
site data and the theoretical prediction. There are two acceptance method which can accommodate
important findings relating to this case study. large variations in hammer efficiency without
impacting reliability. The capacity of all non-PDA
1. If the GRLWEAP parameters can be reasonably tested piles, can be estimated with greatly
estimated in advance, then the FCM relationship improved accuracy provided that the set and pile
can be predicted, and initial target sets can be velocity can be measured accurately.
established in advance of piling.
Go to table of contents
Pile Testing Verification – an Alternative Approach D. Denes, A. Foroughi and J. Seidel
5.2 Case Study Two – Precast Concrete practicalities of driving precast piles where different drop
height may need to be applied due to varying soil/ground
The project used for case study two is the construction of conditions, controlling driving stresses (especially tension
13 bridge pier foundations for a river crossing. At this stresses) and/or different load requirements.
project the soil stratigraphy consisted of very stiff to hard
sandy clay layers. All piles were 400mm square simply- Figure 5 presents the correlation between the C/F ratio
reinforced precast concrete with total lengths ranging and pile set for PDA/ CAPWAP data compared to the
between 20m to 41m and penetration ranging between GRLWEAP models with 3 different drop heights. Similar to
18m to 36m. The end of drive sets were between 1mm Case 1, the correlation between the C/F ratio and pile set
and 13mm using 300mm to 500mm drop height. The piles can be clearly seen from these results. It is also evident
were driven with a 9t hammer. All piles were restrike that GRLWEAP models can predict the correlation with
tested between 1 to 24 days after installation using the reasonable accuracy if the necessary parameters for the
same hammer with drop heights between 400mm and GRLWEAP models are available.
800mm. The restrike sets ranged between 0.3mm and
9mm and estimated CAPWAP resistances varied Nevertheless, there is some increase in scatter evident
between 1100kN to 4500kN. Timber ply was used as a which may result from minor PDA data issues
pile cushion for all tests. (proportionality), accuracy of the manual set
measurement especially at small sets or variable pile
For the GRLWEAP models, three different drop heights cushion stiffness.
were used for assessment. The reason is related to the
The general form of the FCM relationship is similar for 1B. Undertake PDA testing and CAPWAP analysis on the
both Case 1 and Case 2, albeit with different specific first pile. Analysis to be undertaken at various different
values, and again, predictable by theoretical analysis pile sets (e.g., 12mm, 6mm and 2mm).
using relevant analysis parameters. 2. Use data from either step 1A, 1B or both, to establish a
site correlated FCM relationship.
For Case 2, at pile sets less than 1mm per blow, the 3. Measure pile velocity and pile set for untested piles.
inferred capacity is slightly less (92%) of the peak impact 4. Estimate pile force using the measured velocity and
force. At sets of 5mm, the achieved capacity is only 63% calculated pile impedance
of the peak impact force; at 10mm the ratio reduces to 5. Use the site correlated FCM curve to find the C/F ratio
42% and at 15mm pile set the ratio drops further to 29%. for the measured set and then calculate resistance using
the force estimated in step 4.
The results also confirm that the use of different drop
heights has little effect on the correlation between C/F Ongoing pile testing with CAPWAP analysis should also
ratio and measured set. be undertaken on a percentage of piles to check the
relationship. Minor adjustments/refinement may be
6 RECOMMENDED FCM RELATIONSHIP required as more data is collected.
PROCEDURES FOR A PROJECT
7 FURTHER DISCUSSION
The FCM relationship can be developed using the
following method: 7.1 Measuring Velocity
1A. Undertake a GRLWEAP analysis with estimated driving There are only two known methods currently undertaken
parameters from site investigation using proposed pile to measure the velocity during pile driving. The first is by
and driving hammer details. using accelerometers recording the acceleration of the
pile which can then be used to obtain the velocity.
Go to table of contents
Pile Testing Verification – an Alternative Approach D. Denes, A. Foroughi and J. Seidel
Examples of equipment used are accelerometers from the both the pile set and velocity. Pile set is an important
PDA system (PDI / Allnamics). aspect of this force base verification and an accurate
measurement of set should always be a strict
The second is by measuring displacement and requirement.
differentiating with respect to time to calculate velocity.
An example is the Pile Driving Monitor (PDM). PDM is a The use of this force relationship for projects should follow
device that records the displacement of the pile with high the same process used for standard bearing graph
frequency and can therefore accurately measure pile set relationships. That is to undertake a GRLWEAP analysis
and pile velocity, and hence, can be further used to prior to piling works to establish an initial correlation.
estimate pile resistance for all untested piles from a FCM Then once the first pile(s) has(have) been PDA tested,
relationship created from tested piles or GRLWEAP refinement of the relationship is required for application to
modelling. the remaining piles. Ongoing testing should also be
undertaken to identify the need for further refinements as
7.2 Set Limitation the project progresses.
Go to table of contents
SESSION 3
OPTIMIZATION OF RISK
AND SAFETY IN DESIGN
Keynote address
ABSTRACT
Project risk is commonly defined as an uncertain event or condition that, if it occurs, has a positive or negative effect on a
project’s objectives. As the construction industry continues to deliver larger and more complex projects, improving how
we manage project risk is a key priority for clients, contractors and designers. Within this digital age, innovation and change
in our industry will play a key part in how we manage risks and deliver projects. This paper considers how we can define
‘innovation’ and how innovation can be promoted within organisations and on projects. The classification of something as
innovative can vary widely depending on its novelty and context. The journey from creativity (an idea) to a productive
solution (innovation) is difficult and frequently does not materialise. Three examples of innovation are presented to highlight
the range of productive solutions that can be considered as innovation within their context.
1. The development and benefits of public Geotechnical Databases; novel within the context of Victoria
2. Incremental innovation in the use of InSAR satellite monitoring.
3. Radical/modular innovation using automated processed to import piling construction records back into project
models to validate QA and geotechnical ground models.
Go to table of contents
51
Using geotechnical innovation to reduce project risk C. Lyons, C. Cairns
Figure 1. Innovation life cycle showing with succession planning within a market for continued high performance
(Mazouz, et al., 2019).
Within this digital age, the opportunities for innovation to Innovation can be classified as the adoption of existing
revolutionise how we operate are emerging into all facets technology or system and applying them within a local
of our industry. Opportunities include; market. Within the Melbourne market, there is a gap in
the collation, management and delivery of geotechnical
• Automated processes ranging from simply eliminating data that supports industry and maximises the return on
repetitive activities, to scripting complex models to public investment. Public geotechnical borehole
parametric design process. databases, and associated geoscience datasets acquired
• The evolution of Digital Twins connected to the as part of geotechnical investigations, are a relatively
common structure in other States within Australia and
Internet of Things (and conversely their Evil Twin,
globally, although the structure and sophistication of the
bloated with excess data) (Hannel, 2019).
databases vary. The Geological Survey of Victoria (GSV)
• Machine Learning and Artificial Intelligence are is working with government partners to scope the
changing our use of data and design processes. development of a Victorian geotechnical database. The
• Robotics and drone automation as our construction aim of the database is to augment existing free public
sites become smarter, with autonomous trains, resources delivered by the GSV such as GeoVic, the
vehicles, and drone ‘swarms’ online mapping application that enables users to build
• Virtual reality and augmented reality their own maps, using an array of datasets, perform
• 3D printing searches and access data, the GSV catalogue, and the
Victorian Drill Core Library located in Werribee.
• Big Data, Blockchain and Cloud technologies
supported by 5G and Wi-Fi 6 networks
2.1 Benefits of public geotechnical databases
For this paper, three examples of innovation have been
selected to show different ranges of activities which could The following are the key benefits generally identified with
be defined as innovative. A key feature of all three are public geotechnical data bases, many of which directly
the degree of novelty, the kind of innovation (process, assit reducing project risk:
product, service) and the generation of productive
outcomes. • The greater the availability of geotechnical
information to inform design and construction, the
1. The development and benefits of public Geotechnical more efficient and less wasteful design and
Databases. construction can be. This logic would be shared by
most geotechnical practitioners.
2. Incremental innovation approaches for the use of • Increased information and knowledge of wider
InSAR satellite monitoring. geotechnical conditions mitigates the risk of project
delays, cost over runs and environmental impacts.
3. Radical innovation in adopting construction monitoring • Databases present geotechnical logs and other
data into digital models, akin to integrating the Internet information in a single accessible location and
of Things into construction. The Smart Piles platform thereby reducing data management inefficiency,
presented is modular application of the broader optimise site investigation planning and support
concept, which is globally novel from a design and QA scoping, early-stage design and feasibility stage
perspective geotechnical risk assessments.
Go to table of contents
52
Using geotechnical innovation to reduce project risk C. Lyons, C. Cairns
• Reduce the cost of projects both by reduced databases. The NZGD has over 130,000 data points
investigations and less conservative designs with and is used by over 6,000 users.
more informed geological (3D) modelling. • The Queensland Geotechnical Database (QGD) was
• Provide a cost-effective way of compiling and sharing launched in 2017 with a view to consolidate primarily
information, opening opportunities for cumulative tax and toll payer subsidised exploratory
crowd sourcing from smaller, private projects, which investigations. The database holds over 1,600
reduces inefficiencies in retrieving achieved files and geotechnical investigation records.
relying on historic knowledge of projects when • The NSW government is currently working on the
planning investigations. Government Geotechnical Repot Database Project
• The data lives on in the public domain following the (GGRD) to collate geotechnical investigation records.
end of the project and/or agency. • Progressive European nations have recognised the
• Facilitate the research into geotechnical properties of value and importance of geoscience (geotechnical
materials, geological profiling, and education. investigation) data and knowledge (urban geology
• The re-purposing of geotechnical investigation data models) in infrastructure planning and development
from projects for other areas of public good for sustainable cities. Sub-Urban group compiles data
geoscience, examples in greater Melbourne include and models from multiple European Geological
improved understanding of sea level rise and fall and Surveys, Cities and Researchers (including
volcanic eruption frequency. international partners) http://sub-
urban.squarespace.com/
It is estimated the total site investigation saving for
residential property from the Christchurch Geotechnical 2.3 Development of a Victorian geotechnical
Database is in the order of $50-$100 million NZD (Scott, database
et al., 2015). This excludes savings in infrastructure and
commercial development project and qualitative benefits There have been recent changes to Victoria’s public
such as the improved confidence in project risk and more construction guidance. When procuring services for
efficient delivery. geotechnical investigations, or Works or Services that
may require geotechnical investigations, Agencies must
2.2 Geotechnical database examples ensure that their contracts provide for the ownership and
custody of geoscience data collected for the project to be
A of review of relevant geotechnical data bases is transferred to the State of Victoria, where:
summarised, demonstrating the recognised benefit to the
construction industry: • ‘geoscience data’ includes geological, geotechnical
and environmental information, reports, maps,
• In the 1970’s the Urban Geotechnical Automated images, recordings, survey results and drill core, drill
Information System (UGAIS) was developed in cutting and associated materials embodied in any
Canada, funding the creation of geotechnical form; and
databases for 27 cities involving records from over • ‘geoscience data collected for the project’ includes
110,000 boreholes. While several of the data bases geoscience data generated, placed, stored,
are no longer in use, many still are. The Ontario processed, retrieved, printed, accessed, or produced
databases includes data from over 90,000 boreholes using data supplied by the Principal, for the purpose
and in 2015 was accessed on average almost twelve of the contract.
times every workday (Thompson, 2016).
• The British National Geotechnical Properties Within Victoria, the GSV is in the process of scoping the
Database was launched by the British Geological compilation and public delivery of all geotechnical
Survey (BGS) in 1992. The database includes over investigation physical (i.e. drill core) and digital
100,000 boreholes and the web viewer portal is geoscience data acquired as part of geotechnical
accessed over 60,000 times a month (Thompson, investigations undertaken as part of State of Victoria
2016). infrastructure projects.
• The Perth, Australia Central Business District (CBD)
The GSV are working through agreements with current
database was created in 2004 with 649 boreholes major projects and stakeholders on the transfer and
from the 1970’s. release of geotechnical data. The Suburban Rail
• The Canterbury Geotechnical Database (CGD) was Authority Loop Authority (SRLA) is coordinating with GSV
created to generate efficiencies in sharing to provide data from 560+ geotechnical investigations
geotechnical information following the earthquakes of already completed for Stage One of the project.
2010 and 2011 around Christchurch. In addition to Discussions are ongoing for the provision of construction
hosting over 26,000 borehole and CPT records, it also stage face mapping and geological records to supplement
includes regional assessments and maps of the project investigation data.
liquefaction susceptibility. The CDG assisted with the
design and construction of residential, commercial 3 DEVELOPMENT OF INSAR MONITORING
and infrastructure rebuilds, as well as city planning
Interferometric Synthetic Aperture Radar (InSAR) is a
and enabling the insurance industry to understand
radical, innovative satellite-based ground movement
and manage development risks. In 2016 the New monitoring system that is increasing become commonly
Zealand Ministry of Business and Innovation launched place on major infrastructure projects and forensic
the New Zealand Geotechnical Database (NZGD), studies. It significantly enhances and complements
which combined the CGD with other regional traditional survey monitoring approaches.
Go to table of contents
53
Using geotechnical innovation to reduce project risk C. Lyons, C. Cairns
Satellite based InSAR has been applied across a range required to assess movements during construction.
of major infrastructure, mining, forensic and geohazard Additionally, data requires processing, which is
assessment projects. Within the sphere of metro rail typically undertaken in batches of 3-to-6-month
tunnel projects, InSAR came to prominence in Europe intervals.
slightly over 10 years ago on multiple projects (Crossrail • The location of monitoring points cannot generally be
in London, Paris Metro Line 4, Barcelona Metro, Warsaw
controlled. The processing requires consistent
Metro). Within the Victorian context it was adopted on the
backscattering to develop a reliable data point for
Metro Tunnel Project in Melbourne and has been
procured by the Suburban Rail Loop project to baseline processing (such as kerbs, outcrops, or structures)
and understand existing historic ground movements. and may not be reliable over vegetated areas. On
Crossrail the X-band data generated one monitoring
InSAR measures ground movements by comparing two point approximately every 40m2 (Garcia, et al., 2016);
or more Synthetic Aperture Radar (SAR) images of an and
area to identify surface movements over time (refer Figure • “Unwrapping” errors if rapid ground movements occur
1). InSAR enables ground movements to be monitored between satellite passes. Where large settlement (i.e.
over spatially extensive areas to a high level of accuracy <10mm) occurs between satellite passes the
(typically 1-3mm). Radically innovative technology comparison of SAR images can assess the data as
considering the data comes from satellite hundreds of heave instead of settlement. Unwrapping errors in
kilometres above the earth. Benefits of InSAR include:
data can be easily re-calibrated with manual
settlement monitoring if required.
• Extensive spatial coverage which is suitable for
assessing regional ground movement trends. While InSAR can’t replace automated systems, it can
• Can assess seasonal shrink/swell effects, thermal successfully complement them, providing historical
effects (e.g., bridges) existing consolidation and/or ground movements, assessing regional
groundwater related settlement. settlement/heave, reduce the extent of conventional
• Satellites have been collecting and storing data over monitoring and provide cost effective long-term
most urban centres in Australia for more than seven monitoring. On the Crossrail project the procurement and
years, enabling historic data to be processed from data management of InSAR was in the order of a
time periods prior to project commencement and magnitude cheaper than maintenance and data
management of automated data monitoring (Gonzalez, et
assist with establishing baseline ground movements;
al., 2017).
and
• Monitoring can cover sensitive urban areas and InSAR’s massive spatial coverage, high accuracy, regular
structures, mitigating the need for access (e.g., rail readings, and ability to model historical movements make
corridors) and approval (e.g., heritage structures). it a radically innovative approach to surveying.
Limitations with InSAR include:
Figure 2. Two SAR images of the same area are acquired at separate time intervals and the phase shift in the reflected
wavelength is used to assess the differential movement (Geoscience Australia, 2021)
Go to table of contents
54
Using geotechnical innovation to reduce project risk C. Lyons, C. Cairns
Figure 4. InSAR reflectors installations being picked up clearly within a grassed area (before and after images courtesy
of Sixense Oceania).
3.1.2 Commercialisation of space such as Capella Space are rapidly launching additional
satellites and within three years, by using readings from
The miniaturisation of satellites combined with the multiple satellites the monitoring frequency may reduce to
increasing availability and affordability of commercial daily, or even multiple times per day. Subject to the
space payloads has significantly reduced the cost of InSAR image overlay and batching process becoming
launching satellites. Subsequently the number of more automated, this increased frequency has the
commercial satellites orbiting the earth is quickly potential to significantly alter how InSAR monitoring
increasing, enabling incremental innovation in the use of integrates with traditional monitoring approaches. This
InSAR monitoring. higher frequency of monitoring has the potential to move
InSAR through a subsequent cycle of innovative growth.
Within the context of InSAR, the increased number of
satellites relates to an increased frequency of readings. Within a broader industry context, the commercialisation
Currently InSAR reading using a single satellite are of space with increased frequency and higher resolution
typically recorded on a two-week frequency (8-14 days), of satellite imagery may have large impacts in fields such
which is one of limitations compared with traditional as coastal engineering, hydrology and tracking
survey approaches. However, commercial organisation
Go to table of contents
55
Using geotechnical innovation to reduce project risk C. Lyons, C. Cairns
construction progress, particularly when combined with to provide insights into construction productivity and
advanced in artificial intelligence and data-driven models. quantities. Smart Piles takes the same data and uses
automated processes to provide:
4 CONSTRUCTION DATA TO DIGIAL DATA
1. Automated quality control reviews.
While advances are being made in developing existing 2. Verification and development of project geological
asset data and designs into digital models and digital models,
twins, to complete the data cycle we need to look at 3. Interrogation of data during construction and
integrating construction data directly back into our visualisation of piling data into BIM as-built models.
models. Currently construction data collection and
automation are typically targeted around increased
productive to reduce construction costs and programme. 4.2 Quality control reviews
One of our next steps as an industry must be integrating
these digital systems directly into our digital design Quality reviews of piling records by designers are typically
models to validate geotechnical models, improve quality complete at the end of the project or when zones of the
control systems, manage geotechnical risks and provide work are completed. Designers normally check grout
efficiencies in automated development of as-built records. pressure and embedment depths to confirm that the as-
built records meet the design requirements. Smart Piles
Collection and use of data from sensors, aka the Internet uses automated processes to review piling construction
of Things (IoT), is already rapidly occurring within other records to verify the quality of each pile as it is constructed
industries such as healthcare, manufacturing, agriculture to provide early warning of construction issues (Figure 5).
and energy. The construction industry is not as advanced, The system effectively completes all necessary QA
but the use of automation and machine learning checks, reducing input from engineers reviewing logs and
processes are advancing rapidly. The industry already providing confidence in the works.
has many established systems ranging from data collation
on automated monitoring platforms, robotics in 4.3 Verification of project design models
construction (e.g. drilling for tunnel ferrules), autonomous
mining trucks, UAV surveys, tracking materials/supplies Manual reviews of piling records will often look at changes
and GPS positioning of plant. in drilling penetration/torque to help validate ground
models and design assumptions. This is particularly
Adopting IoT construction solutions has the potential to important in lateral loading designs including lateral
improve workplace safety (e.g. wearable IoT devices, fall spreading associated with liquefaction, where the depth
detection apps), resources management, reporting and to a firm layer is a key design assumption. On sites with
maintenance, reduce insurance premiums, mitigate variable geological profiles, pile layouts may change
wastage and theft. However, the industry also faces across the site according to changes in subsurface
challenges including employee privacy, data security, profiles.
cost vs benefit and collection of all data vs useful data.
For the time being it is likely that advances in IoT Smart Piles uses automated process to review logs during
construction solutions will come from large building and construction, enabling changes in construction conditions
infrastructure projects where the benefit will outweigh the to be assessed immediately once pile records are
effort to implement new systems. completed and integrated back into the geological
models. This is a more proactive approach to identifying
4.1 Smart piles ground changes and managing ground risks, particularly
on complex sites.
Smart Piles is an innovative product which has been
developed by Arup as a tool for collating and using data 4.4 Interrogation of digital pile data
from piling rigs. The tool collects site data from piling rigs
to visualise progress, undertake quality checks and Once pile data is available in a digital format, it opens new
integrate construction record data back into project opportunities for construction reviews. Within the piling
geotechnical models. Research within Arup has played a industry some organisations are using digital piling
fundamental role in identifying and leveraging business records to review construction programmes and assess
opportunities, with a structured, global approach to the rate of piling productivities.
research engaging over 1,900 staff across 500 research
projects annually (Arup, 2021). Smart Piles can take the automated quality control and
design verification process and integrate them spatially
Most modern piling rigs have sensors which record data into digital models. This enables the automated quality
with depth for piling, soil mixing and ground improvement and design verification processes and checks to visually
activities. Examples of data types include penetration communicated across the site, enabling both issues to be
depth, inclination, drilling torque, rate of progress and address early in the construction process and
concrete pressure (see example in Figure 1). opportunities for refinement to be identified early enough
Traditionally these records are provided as hard copy so that they can have a material impact on the works.
PDFs for a review of quality (embedment depth, grout
pressure etc) or as a record of construction. However, Smart Piles also enables digital as-built models of the
this digital information can also be uploaded to a cloud- piling works to be produced in an automated manner,
based server relatively easily akin to an Internet of Things providing cost savings and improved quality to produce
(IoT) approach, opening various avenues of digital LoD 500 BIM models.
manipulation.
Smart Piles is a step towards closing the gap from ‘design’
Within the construction industry, many operators have to ‘as-built’, using modern digital technologies in an
already similar approaches to extract digital piling rig data automated and innovative way.
Go to table of contents
56
Using geotechnical innovation to reduce project risk C. Lyons, C. Cairns
Figure 5. Example of a piling rig installation records where checks can be completed on consistency or variations in the
data.
Go to table of contents
57
Using geotechnical innovation to reduce project risk C. Lyons, C. Cairns
REFERENCES
Go to table of contents
Design methodology and input parameters applicable to foundation design for large
complex towers
Y.Y. Tay1, BEng (Hons), PhD, CEng, FICE, FIEAust, CPEng; A. K. C. Smith2, MA, PhD, MBA, CEng, FICE, CGeol, FGS;
and C. Haberfield3 BSc, BE (Hons), PhD, FIEAust, CPEng, RPEQ
1, 3
Golder Associates Member of WSP, Building 7, Botanicca Corporate Park, 570-588 Swan Street, Richmond, Victoria 3121, Australia;
email: ytay@golder.com.au, chaberfield@golder.com.au
2
Coffey Geotechnics Ltd., Gardner House, Hornbeam Park Avenue, Harrogate, UK, HG5 8NA, email andrew.smith@coffey.com
ABSTRACT
As buildings are progressively getting taller, traditional methods of design that generally relied on considerations of the
vertical load-carrying capacity of the foundation system, assessed by empirical methods and a lumped factor of safety,
have been largely replaced by serviceability-based methods of design which typically result in an optimised foundation
design. Serviceability-based designs typically rely on powerful commercial software packages to enable advanced
numerical analysis of foundation systems. This paper briefly discusses case studies of foundation design processes
including soil-structure interaction analyses adopted for the serviceability design of tall towers. In order to obtain accurate
building movement prediction from complex computer analysis, it is imperative that materials and ground stiffness
properties be accurately characterised and measured. This paper presents ground stiffness properties measured from
various types of tests at different strain levels (i.e., geophysical testing, pile load tests, pressuremeter and laboratory tests)
that have been adopted as input parameters in the numerical analyses. The higher allowable shaft friction values from
serviceability analysis compared to those from traditional methods, are further justified on the basis of bi-directional static
pile testing.
2 SUBSURFACE CONDITIONS
Go to table of contents
59
Design methodology and input parameters applicable to foundation design for large complex towers Y. Y. Tay, A. K. C. Smith, and C. Haberfield
qc profile with depth, ELAB values of the four plots tend not assessment, GHH from pressuremeter presented in the
to increase significantly with depth (Figure 2). following sections have been corrected to GVH assuming
anisotropy is due to the differences in normal stress acting
In-situ Young’s modulus (EPMT-H) was measured by in different directions. Therefore, stiffness in any direction
pressuremeter tests with unload reload cycles. The is a function of K0 as shown by the following equations
subscript “H” used here refers to measurements taken in from first principles:
the horizontal direction. Unload-reload EPMT-H-ur values GHH = EH / [2(1+ νHH)] (1)
from pressuremeter tests were about 4 times higher than
the initial EPMT-H-ini (Figure 2), the main reason being that For undrained expansion,
the latter was an elasto-plastic measurement where part νHH = 1 - K0 / 2 (2)
of the expansion was irrecoverable. where K0 = EH / EV (3)
Intuitively, EPMT-H from pressuremeter tests which test a EPMT-V is conservatively assumed to be EPMT-H / 2 in this
larger rock mass compared to intact rock samples tested paper, as K0 measured from pressuremeter tests was
in laboratory, should be smaller than ELAB because of the greater than 1.5. Correction of anisotropy of rock may
scaling effect and the fact that the bigger rock mass is partly explain higher EPMT-H compared to ELAB.
likely to consist of more imperfections, noting rock mass
in Dubai do not have much fractures. However, Figure 2 3.3 Stiffness and strain relationship
shows initial EPMT-H-ini to be around 2.5 times higher than
ELAB and unload-reload EPMT-H-ur to be around 9 times Cavity strain measured from the expansion of borehole in
higher than ELAB. In addition, EPMT-H increases with depth pressuremeter tests was converted to shear strain. Shear
and the difference between ELAB and EPMT-H is more strain is the constant area ratio as shown in Figure 3,
pronounced at depth. It is likely that ELAB measurements therefore shear strain of the pressuremeter tests is
were underestimated as a result of the lack of determined from (4) below:
confinement in UCS testing and that the calcisiltite
samples extruded from deeper depths were disturbed as Shear strain = 1 – [ 1/ (1+ɛc)2 ], (4)
they swelled. Other explanations of the difference where ɛc is the cavity strain.
between ELAB and EPMT-H, aside from the above and
scaling effects are likely anisotropy of rock and test strain
levels discussed in the following sections.
3.2 Correction for anisotropy Large scale bidirectional static load tests to failure, using
Osterberg cell (O-cell), were carried out at the four plots
The rock mass in pressuremeter tests is loaded in the for the assessment of pile-settlement behaviour and pile
horizontal plane, and the tests give horizontal shear ultimate skin friction. Osterberg cell tests are increasing
stiffness (GHH). Whilst this may be directly applicable to in popularity in Dubai. They eliminate many of the safety
the analysis of radial compression or pile lateral risks of static load tests using kentledge or tension piles.
movement assessment, it may not be applicable for They also have the significant advantage of enabling test
axially loaded piles where the piles and rock are deformed piles to be short enough such that full mobilisation of skin
in the vertical plane. In the absence of direct friction between the piles and ground (i.e. failure) can be
measurements of GHH and vertical shear stiffness (GVH) achieved. The full range of the anticipated pile depths can
measured from geophysical testing for anisotropy
Go to table of contents
60
Design methodology and input parameters applicable to foundation design for large complex towers Y. Y. Tay, A. K. C. Smith, and C. Haberfield
be tested by the simple expedient of having O-cells at in the model to that measured. ETP and average shear
different depths. strain along the pile, obtained from the model at pile head
settlement of 1% pile diameter are included in Figure 4.
The ETP values fit within the range of the values
determined from other tests carried out.
A total of 7 no. pile tests were carried out at the four plots.
Movements of the upper and lower segments of each test
pile were measured as the segments were jacked apart
by the O-cell. These movements were converted into an
equivalent top-loaded pile-settlement curve for each test.
Stiffness ETP from the pile tests were back-analysed by
Figure 6. Photo showing lowering of reinforcement cage
numerical modelling a single isolated pile using Plaxis 2D
with 0.5m thick polystyrene ‘soft toe’ attached to the
(axis-symmetry model) and matching the pile-settlement
base
Go to table of contents
61
Design methodology and input parameters applicable to foundation design for large complex towers Y. Y. Tay, A. K. C. Smith, and C. Haberfield
4.2 Measured pile skin friction test values, which would mean that (5) yields conservative
ultimate skin friction and/or qc values measured from UCS
Average mobilised skin friction values along the upper tests were underestimated.
and lower test segments of each test are shown in Figure
7. Test piles at Plots A17 and E14/15 had longer upper 5 DESIGN METHODOLOGY
test segments (10m) than the lower segments (5m), the
average measured values were lower for the upper The design methodology adopted for the four plots follows
segments as they were not fully mobilised. The maximum two slightly different but related means of achieving
measured skin friction (near full mobilisation) at each site optimised foundation solutions by justifying serviceability
ranged between 1250 and 1700 kPa. analysis, also associated with higher stiffness values
derived in part from pile testing; and by justifying the use
Expressions for ultimate skin friction as a function of qc of higher allowable shaft friction values on the basis of pile
have been derived by several authors, some of which are testing to failure.
shown in Table 1. These can typically be normalised, as
shown in (5): 5.1 Serviceability assessment
Ultimate skin friction = 𝑥𝑥𝑥𝑥 √(qc)0.5 (5) For serviceability-based design, soil-structure interaction
Where 𝑥𝑥𝑥𝑥 = 0.25 to 0.4 modelling of foundations is key. Soil-structure interaction
qc = unconfined compressive strength essentially requires compatibility between a structure, its
foundation and surrounding soil/ rock. This requires two
Table 1. Correlations of ultimate skin friction and qc key fundamentals to be achieved, which is load
equilibrium amongst the structure, foundation system and
Correlations Reference
soil; and compatibly amongst displacement of structure,
0·375 (qc)0·515 Rosenberg and Journeaux pile head and ground at the interfaces. The approaches
(1976) to undertake soil-structure interaction assessment are:
0·22 (qc)0·6 Meigh and Wolski (1979)
ab (qc) Williams and Pells (1981) 1. 3D finite element modelling which incorporates the
ground, foundation and structure.
(0·20 to 0·30) (qc)0·5 Horvath et al. (1983) 2. Split the ground, foundation and structure into two
0·45 (qc)0·5 Rowe and Armitage (1987) models (structural and geotechnical models) and
0·15 (qc) Carter and Kulhawy (1987) iterate between the models until compatible
displacements at the interface of the models have
(0·15 to 0·2) × (qc)0·5 Reese and O’Neil (1988)
been achieved.
(0·40 to 0.8) × (qc)0·5 Zhang and Einstein (1998) 3. 3D finite element modelling which incorporates the
ground and foundation system; and column/ wall
loads input as point/ line loads on the foundation
system.
Go to table of contents
62
Design methodology and input parameters applicable to foundation design for large complex towers Y. Y. Tay, A. K. C. Smith, and C. Haberfield
Go to table of contents
63
Design methodology and input parameters applicable to foundation design for large complex towers Y. Y. Tay, A. K. C. Smith, and C. Haberfield
7 ACKNOWLEDGEMENTS
REFERENCES
Go to table of contents
Optimisation of temporary support design for the Northern portal cut & cover tunnel
J. Zeerak1, Dr. M. Wei2, J Roper3, B Clarke4
1
Associate Principal-Geotechnics, EIC Activities Pty Ltd, 567 Collins Street, Melbourne, Australia; email: Jawad.Zeerak@eicactiv.com
2
Technical Principal-Geotechnics, EIC Activities Pty Ltd, 567 Collins Street, Melbourne, Australia; email: Michael.Wei@eicactiv.com
3
Principal-Structures, EIC Activities Pty Ltd, 567 Collins Street, Melbourne, Australia; email: Jeff.Roper@eicactiv.com
4
Engineering Manager, John Holland Group, 180 Flinders Street, Melbourne, Australia; email: Ben.Clarke@jhg.com.au
ABSTRACT
The West Gate Tunnel Project is a city-shaping project that will deliver a vital alternative to the West Gate Bridge, provide
quicker and safer journeys, and remove thousands of trucks off residential streets. Delivery of the WGTP project is currently
underway by a joint venture of CPB Contractors and John Holland Group. Northern Portal cut and cover tunnel is one of
the major structures on the project which requires excavations to a depth of 22 m to allow the launch of the twin 15.6m
diameter Tunnel Boring Machines (TBM). Design of the retention system comprised 900 – 1500 mm diameter secant pile
wall supported with multiple levels of temporary steel struts. A detailed Soil-Structure Interaction (SSI) analysis together
with a review of proposed construction methodology indicated that a two-level propping arrangement as opposed to
propping at three levels, which is what would normally be expected for a structure of this scale, would be adequate. The
opportunity to remove one level of the proposed steel struts with the potential for a significant reduction in materials and
time was considered critical to the completion of this critical path structure. Removal of one level of props would result in
a reduction of steel tonnage in excess of 1,000 tons, in addition to improving constructability, productivity and safety. This
paper discusses technical aspects of the analysis which enabled the development of the above optimised solution. In
addition, the results of the instrumentation and monitoring and performance of the constructed portal structure will be
discussed.
Keywords: WGTP, Cut & Cover Tunnel, Deep Excavation, Secant Pile Wall, Soil Structure Interaction, FEM, Value
Engineering
1 INTRODUCTION the launch of twin TBM machines for the bored tunnels
and forms the final cut & cover tunnel (tunnel portal) after
The West Gate Tunnel Project (WGTP) is a city-shaping completion of the TBM launch and construction of the
project that will deliver a vital alternative to the existing permanent structural lining. The portal structure is over
West Gate Bridge, provide quicker and safer journeys, 330m in length and up to 22.2 m in depth at the interface
and remove thousands of trucks off residential streets. In of the cut & cover tunnel and the TBM tunnel. Temporary
addition to the construction of twin large diameter (15.6 support for the northern portal excavation comprised the
m) tunnels the project will deliver: following retention systems:
• Widening of the West Gate Freeway from 8 to 12 • Secant pile walls supported with heavy steel strutting
lanes & waling
• Multiple crossings across the Maribyrnong River, • Anchor supported sheet pile walls
connected to an elevated viaduct along Footscray Rd, • Base slab
and • Tension piles
• Multiple bridges, entry and exit ramps across the
eastern and western zones of the project. Propped secant pile wall was adopted for the deeper parts
of the cut & cover tunnel at the interface with the TBM
The project is currently in the delivery phase in Melbourne tunnels, transitioning to the anchored sheet pile wall as
by a construction Joint Venture of CPB Contractors and the excavation depth reduces towards the open trough
John Holland Group. structure. Multiple design types were incorporated in the
design to suit excavation depths and ground conditions
2 NORTHERN PORTAL CUT & COVER TUNNEL along with the portal structure. The portal structure was
designed as a tanked structure (undrained) for the full
The Northern Portal (NP) cut and cover tunnel at WGTP length of the portal to minimise disturbance to the regional
is one of the major packages of the works on the project. groundwater regime. Figure 1 presents a general layout
The portal is used in the temporary condition to facilitate of the northern portal.
Go to table of contents
65
Optimisation of Temporary Support Design for the Northern Portal Cut & Cover Tunnel J. Zeerak, M. Wei, J. Roper and B. Clarke
It is worth noting that the northern portal cut & cover through the foundation rock which was expected to be
tunnel is a major structure comprising propped secant pile minimal.
wall in the temporary and a structural lining wall
constructed on the inside in the long term. The two
structure types combined forms the cut & cover tunnel
with a significant interaction between the two structure
types. Discussions in this paper are limited to the analysis
conducted for the structure in the temporary condition
only before construction of the permanent structural
lining. It is further noted that the discussion presented in
this paper are based on the analysis, which are limited in
nature conducted as part of a value engineering and
optimisation exercise, detailed design of the structure
were undertaken by others.
2.1 Geological Conditions & Geotechnical Figure 2. Geological Survey of Victoria, Melbourne
Parameters Mapsheet, (1:63,360)
Figure 2 presents surface geology of the general project Geotechnical parameters adopted in the finite element
area highlighted on an extract from the Geological Survey analysis were derived from site investigation data,
of Victoria’s Melbourne Mapsheet. laboratory testing, published literature and past
experience. Parameters for the soil units were generally
The geological conditions at the Northern Portal consists derived from laboratory testing, typical correlations, and
of an upper unit comprising man-made Fill overlying a thin past experience. While rock mass parameters for use in
veneer of the soft and compressible Coode Island Silt the design of excavation support were assessed using the
(light green in Figure 2, not shown in Figure 3). These generalised Hoek-Brown strength criterion using the
thinner units are underlain by a more substantial layer of Roclab (RocData) software by Rocscience. Adopted
Quaternary alluvial outwash, and deeper alluvial infill to a parameters are summarised in Table 1.
paleochannel crossing the northern end of the Northern
Portal (green/dark green in Figure 3). Variable
thicknesses of the Brighton group unit (yellow in Figure 3)
are encountered in parts of the northern portal. The older
volcanics rock (pink in Figure. 3) comprises the bedrock
in the area of the portal where most of the piling and
excavation works within the deeper ends of the northern
portal occur.
Go to table of contents
66
Optimisation of Temporary Support Design for the Northern Portal Cut & Cover Tunnel J. Zeerak, M. Wei, J. Roper and B. Clarke
In reality, the soil behaves in a lot more complex manner 3 NORTHERN PORTAL RETENTION DESIGN
than that in the modelling. For the purposes of covering
critical cases in terms of forces on the retention structure The adopted retention system for the majority of the cut &
and support elements, various sensitivity checks were cover structure comprised 900 mm to 1500 mm secant
carried out with changing soil properties from undrained piles propped with multiple layers of steel struts. A secant
to drained at various constructing stages. The impact of pile wall option was required to provide a watertight
the above sensitivity checks on the overall design was structure as required by the project technical
found to be negligible in this case. specifications. The larger diameters 1200 mm and
1500mm diameter piles were adopted for the deeper parts
Excavation works are essentially ‘unloading’ problems. of the excavation in a hard-hard pile configuration, while
This behaviour is best captured using the constitutive soil the smaller 900 mm diameter piles were adopted for the
model Hardening Soil (HS) model in a finite element shallower excavation depths with a hard-soft
analysis e.g. PLAXIS. The HS model is an advanced soil arrangement. It is noted that a hard-hard pile arrangement
model that is able to generate a more realistic soil was incorporated in the design for constructability
response in terms of non-linearity, stress dependency and reasons i.e. to reduce reinforcement congestion and ease
inelasticity of soils. The model requires three stiffness of installation.
parameters (E50/Eoed/Eur) as shown in the E’ column in
Table 1. In the analysis, in general an unload/reload The proposed design of the propping for the secant pile
modulus (Eur) of 3 x Secant modulus (E50) was adopted wall comprised 3 levels of steel strutting connected to a
based on literature and common practice. Although waler system against the secant pile walls. The largest
experience and some cases studies in literature indicate steel struts comprised double 1200WB steel sections
that this ratio can be significantly higher than 3 in some connected to a double 1200WB walers system attached
cases. This may also explain the better than predicted to piles. Steel struts were supported in the middle using a
deflection results for the structure which will be discussed row of king posts supported on bored piles socketed to
in later sections. rock. The initial strutting layout at S2 level is shown in
Figure 4 with a typical section shown in Figure 5.
Go to table of contents
67
Optimisation of Temporary Support Design for the Northern Portal Cut & Cover Tunnel J. Zeerak, M. Wei, J. Roper and B. Clarke
The assessment criteria relevant to the design of retention walls as shown in Figure 7, discussion on the interaction
system in the temporary condition included: of permanent and temporary structure will be limited to the
- Stability during excavation above.
- Wall deflection and settlement behind the wall
- Crack widths
- Strength checks
Go to table of contents
68
Optimisation of Temporary Support Design for the Northern Portal Cut & Cover Tunnel J. Zeerak, M. Wei, J. Roper and B. Clarke
It is also noted that small section of the permanent • Settlement markers behind the excavation at regular
structural lining was constructed in early stages after intervals
casting of the base slab which was called ‘Upstand Walls’. • Groundwater monitoring wells; and
The analysis also showed that it was critical that the • Strain gauges on selected steel props to monitor
partial upstand walls were constructed prior to removal of forces on the steel struts.
the S3 prop. These upstand walls would prop the
retention piles at a higher level and therefore reduce The I&M requirement was a critical part of the design
additional forces generated in the piles. Analysis allowing the wall and strutting system to be monitored
indicated that this would result in an increase of forces during progressive excavation and installation of strutting
within the base slab and upstand walls slightly, however, system. This was in particular critical where the S2 level
the benefits would outweigh the small increase in forces. props were omitted. The recommendation was that
The above were relatively simple, yet critical modifications retention piles and S1 level props would be monitored
required to enable a two-level strutting system for spans continuously, and wall deflections and prop forces
of over 15 m between the two props which were compared against design predictions.
considered unprecedented in author’s experience.
The monitoring system would allow for early intervention
In addition to the above, various refinements and should the monitoring results indicate a need based on
improvements were also made in the finite element the wall performance during construction. Accordingly,
modelling resulting in improvements of the overall wall deflections were monitored continuously during bulk
analysis outcome. In the recent versions of PLAXIS 2D, excavations and installation of the steel props until base
the software allows the input of flexural stiffness as slab excavation. Based on the PLAXIS analysis, a
‘elastoplastic’ moment-curvature diagram instead of the maximum wall deflection of 55mm was expected in the
more ‘traditional’ way of defining flexural stiffness of the final stage, upon removal of the S3 prop after the base
plate elements as a linear elastic material. With the slab had been constructed. Inclinometer readings
improved moment-curvature (M-K) input feature, it was obtained during construction indicated maximum wall
found that forces generated within the retention piles were deflections of 20mm which was well below the design
more optimised and considerably lower than those which predicted value and set trigger levels. Typical inclinometer
would have been obtained using the traditional ‘EI’ values profiles showing horizontal wall deflections are shown in
reduced by a factor to account for short- and long-term Figure 9.
cracking of the concrete. The combined effects of the
above led to the possibility of the S2 (middle props) being
removed from the design leaving two levels S1 and S3.
Although there were differences in the pile forces at
individual stages, PLAXIS analysis indicated that
retention pile structural actions envelopes generally
remained similar between a two-strut configuration and a
three-strut system. Proposed prop sizes also remained
adequate for the slightly increased loads with the S2
props removed. Figure 8 shows a view of the northern
portal temporary works completed.
5 INSTRUMENTATION AND MONITORING Similarly strain gauges installed on the steel struts were
RESULTS monitored during construction to obtain prop forces and
comparison against design predictions. Monitoring results
Given the size and scale of the portal structure, a detailed indicated strut forces within the range predicted in the
and comprehensive Instrumentation and Monitoring PLAXIS analysis. There were variations in the peak forces
regime was recommended as part of the design. The as shown in Figure 10. These fluctuations in the forces
following instruments were installed on piles and struts were attributed to the changes in the atmosphere
along the portal: temperature levels causing expansion and contraction of
the steel struts. Progressive and real-time monitoring
• In Place Inclinometers at regular intervals results obtained at each stage of the construction
• Reflectorless prims at multiple levels on the piles provided confidence in the suitability and adequacy of the
design.
Go to table of contents
69
Optimisation of Temporary Support Design for the Northern Portal Cut & Cover Tunnel J. Zeerak, M. Wei, J. Roper and B. Clarke
Observations made from the monitoring results were that methodology indicated possibility of removal of one level
significantly lower wall deflections were recorded of props for a major portal structure on the WGTP. The
compared to those predicted in the design and calculated SSI was completed using the finite element software
in PLAXIS analysis. This is in part attributed to some level package PLAXIS 2D. Results of the FE analysis indicated
of prudent conservatism in the design as would expected that a two-level strutting arrangement opposed to
for the scale and size of the excavation described in this propping at three levels, would provide adequate support
paper. However, other observations and lessons learnt to the 22 m deep excavation at the portal structure. Level
relating to better-than-expected performance of the of the lowest strut (along with other improvements
excavation support system, particularly the retention piles discussed in this paper) was found to be critical in
could be attributed to factors stated below: optimising the temporary support requirements for the
• Stiffness parameters for various rock units were portal structure. A thorough and continuous
assessed using the Roclab software using lower instrumentation and monitoring regime prescribed as part
bound and upper bound test results. In the PLAXIS of the design, enabled a close monitoring of the
analysis, typically lower bound parameters were used performance of the structure and provided confidence
given the scale and size of excavation, associated during construction with the suitability and adequacy of
risks and to cover inherent variability and uncertainty the proposed design. Results of the ongoing monitoring
associated with ground conditions. It appears, that indicated a generally better performance of the retention
encountered ground conditions have responded design than expected. This may be attributed to the 3D
considerably better than those assumed in the nature of the structure towards the deeper ends of the
analysis which has resulted in better performance of portal which was not captured in the 2D analysis and
the retention system. Greater levels of confidence can encountering better ground conditions in particular better
be reached with additional targeted geotechnical rock stiffness than those assumed in the design. This
investigations and laboratory testing to optimise underlines the significance of adequate site investigation
parameters further during design stages. and testing to enable detailed assessment and adoption
of refined geotechnical parameters. The adoption of an
• As noted earlier, an unload/reload (Eur) modulus of optimising strutting design at two levels as opposed to
3xE50 was adopted in the design. It is also possible three levels which would normally be seen reasonable for
that in this Eur/E50 ratio is higher than 3, potentially excavation depths such as those required at the northern
greater than 5. Pressuremeter testing can be portal, provided significant cost savings to the project in
undertaken to obtain more accurate estimates of the addition to an efficient, smoother and safer construction
Eur/E50 ratio. program.
Go to table of contents
SESSION 4
THE ROLE OF DESIGN
IN INFRASTRUCTURE
PROJECTS
Session 4 – The Role of Design in Infrastructure Projects
Supporting Innovative Design and Construction
S. R. Taylor1, CPEng, FIEAust.
1
Major Road Projects Victoria, Level 25, 180 Lonsdale Street, Melbourne; email scott.taylor@roadprojects.vic.gov.au
ABSTRACT
Innovation is at the core of the engineering profession. Innovation is driven largely by the need to increase efficiency,
reduce costs, or respond to increasing complexity. In today’s context, these drivers appear to be converging, with
efficiency, cost and increased complexity almost a baseline for all projects, and the impacts of climate change, sustainability
and the circular economy a significant influence on the future of transport infrastructure. To seek the benefits of innovation
on transport infrastructure projects, Major Road Projects Victoria’s (MRPV) aims to facilitate the minimisation and removal
of barriers and obstacles to innovation. The barriers to innovation include a risk-adverse culture, limited capacity and
capability of resources (both within industry and government), leadership, regulatory requirements and a bureaucratic
culture, and rewards and incentives for the implementation of innovation. Through a series of new initiatives, this
presentation will outline how MRPV is supporting innovation in design and construction of major road projects in Victoria.
To address barriers associated with risk-aversion and capacity, MRPV have implemented a new delivery model that
focuses on a program of projects with incentives for innovative solutions. For barriers associated with leadership,
regulatory requirements and bureaucratic culture, MRPV is leading a program of modernising and updating standards and
specifications including trialling intelligent compaction, and creating of a new technical specific for recycled organics.
Go to table of contents
72
Supporting Innovative Design and Construction S. R. Taylor
projects within different organisations. This means the While this may be a reality of government entities, this is
risk of losing innovation done by a project is significant, not such a barrier for private industry who are more
as once the project finishes and the team is disbanded, willing to invest in innovation without the certainty of
there is little incentive or desire for the project learns to success. Therefore, to effectively remove this barrier,
pass their innovations onto the next project. In the government and industry need to partner in a manner
transport infrastructure sector this is largely driven by the that makes best use of the limited resources, while still
competitive nature of projects, where organisations will allowing individuals the freedom to explore new ideas or
tend to protect innovations from the broader market. concepts.
For the full benefits of any innovation to be realised by 3.3 Failure of Leadership
the broader industry, it requires an industry wide agency
to oversee innovation for the industry. This is a role Wipulanusat et al. (2019) comment that leadership plays
State Governments in Australia have significant part to an important role in fostering a culture of innovation.
play. Governments set the regulatory and policy “Leaders must find mechanisms to encourage the
environment within which industry operates. They are generation, adoption and implementation of
also the major purchaser of civil infrastructure within innovations,” (Wipulanusat et al. 2019). While the
Australia. Therefore, it is within government’s interest to leaders don’t need to be the creative force behind
ensure innovations succeed at the industry level so that innovations, they are critical in supporting and
their benefits can be fully realised. championing innovation to ensure its success.
Capacity can be considered in both the context of people 3.5 Few Rewards or Incentives
and funding. Often innovation will require development
and funding without knowing the outcome of the Wipulanusat et al. (2019) comment that in government
research being undertaken. Government’s, however, innovators rarely received feedback or a reward for their
operate on clear funding cycles and productive use of success, however “if the innovation fails or does not
resources, leaving little space for the ability to innovate. prove to be efficient, the innovators are responsible for
all the costs.” When translated into the transport sector,
Go to table of contents
73
Supporting Innovative Design and Construction S. R. Taylor
particularly for traditional D&C delivery models, there is of the project, and then the Project Delivery Phase once
little incentive for contractors and designers to share contracts are executed. A designer is engaged by the
innovations beyond their projects, meaning they can be contractor in a sub-contract agreement similar to a
short lived and not widely adopted. traditional D&C model. The appointment is made after
the award of the project to a contractor and before the
Therefore, a key feature of delivery models that foster a start of the Project Development Phase, such that
culture of innovation are those that contain some MRPV, the contractor and design all work collaboratively
element of reward and recognition for new technology, through both phases of the delivery model.
processes, or systems.
At its core, the PDA approach features elements of both
3.6 Bureaucratic Culture the Alliance, and Design and Construct (D&C) project
delivery models. The initial Project Development Phase
The barrier related to a bureaucratic culture relates is similar to Alliancing, whereby scope, risks and
strongly to leadership and the risk adverse culture within opportunities are shared and resolved with input from
government. Governments tend to be hierarchical in MRPV, contactors and designers. Once the contract is
decision-making, as well as restrained in their promotion executed, the model converts to more of a traditional
of innovation due to their inherent tendency towards D&C, however payment is through an Incentivised
regulation and certainty (Wipulanusat et al. 2019). Target Cost (ITC) payment mechanism, which
reimburses direct costs and includes cost and non-cost
To overcome the barrier of a bureaucratic culture, MRPV incentives through a performance regime. However, a
has partnered with industry to delivery projects through significant difference of the PDA model from current
the Program Delivery Approach (PDA) which has forms of delivery models is the projects are managed as
enabled projects to offer innovative solutions that a program of works enabling a more sustainable
address performance-based regulations, standards, and contractor and designer market.
specifications. This ensures industry suppliers with new
technology, process or systems can successfully The benefits of the PDA model for industry include;
implement their innovative solutions within the
constraints of government requirements. • a more sustainable supply of projects for industry
partners and their supply chain;
4 INITIATIVES TO ENCOURAGE INNOVATION • creation of long-term competition based on
performance rather than short term, cost driven
To increase and encourage new technology, systems, competition;
and processes within road projects, MRPV has been • improved cost certainty through the reimbursement
working on a number of initiatives that aim to remove and of all direct costs;
reduce the barriers to innovation. The next sections • better sharing and allocation of risk to the party best
outline two key initiatives, namely the Program Delivery able to manage the risk;
Approach (PDA) used to deliver projects, and a program • streamlined procurement processes leading to
of modernising and updating standards and reduce procurement time and costs;
specifications. • streamline procurement processes and, in so doing,
maximise efficient engagement of contractors and
5 PROGRAM DELIVERY APPROACH consultants of all tiers to perform works;
• enable more Front End Engineering and Design
In mid-2020, MRPV implemented a new project delivery (FEED), investigations and assessment of project
model called the Program Delivery Approach (PDA). specific risks; and,
The PDA offers the ability for a portfolio approach to the
• better integrated project planning and project
delivery of projects. The objectives of the PDA are to;
delivery through collaboration between government
and industry.
• Create a more sustainable contractor market, by
engaging with the appropriate industry partner When considering barriers to innovation, the PDA model
based on their capacity and capability. has a number of key features that foster and enable
• Improve the efficiency of project procurement creativity and allow new technology to be more easily
resulting in saving time and minimising State and adopted by projects, as well as shared with the broader
contractor costs. industry. The key features of the PDA model to enable
• Improve project outcomes, contractor performance innovation are explored further in the following sections
and optimise value through a more collaborative and include increased collaboration, risk mitigation and
approach to procurement and delivery, by sharing, and incentivisation for innovation.
incentivising contactor performance, both financially
and via future opportunities. 5.1 Collaboration
The PDA model awards projects to pre-qualified A significant feature of the PDA model is the closer
contractors and designers based on their capability, collaboration between industry (both contractors and
capacity, past performance, and ability to deliver value- designers) and government, as well as the supply chain
for-money solutions. These pre-qualified companies partners critical to the projects delivery.
have been assigned to MRPV’s Construction and Design
Panels. The panels divide industry partners into broad By engaging a single team through the project
groups based on their capability. development phase, risk and opportunities can be
rigorously assessed and mitigated, which would not
Once selected, a contractor will follow a two-staged normally be achieved in a more traditional D&C projects.
delivery model under the PDA, entering into a Project Further, the collaboration between government and
Development Phase to determine the scope and value
Go to table of contents
74
Supporting Innovative Design and Construction S. R. Taylor
industry means that risks can be assigned to the party of Transport (DoT), who are the owner and operator of
best able to mitigate those risks. It also enables a the assets delivered by MRPV.
greater adoption of innovation, with industry and key
suppliers able to bring forward ideas, with MRPV A significant barrier to new products and new
facilitating and enabling the adoption of the innovation. technologies is that standards and specifications don’t
allow or even consider the application of these emerging
5.2 Risk Mitigation and Sharing materials or concepts. This is a result of standards and
specification being largely prescriptive in nature, where
Risk and risk management is dealt with very differently the minimum requirements define the inputs and
in the PDA model when compared with traditional D&C describe what must be done to achieve the
contracting. Considering only a single contractor and requirements. Prescriptive standards and specifications
designer is taken through the project development phase are best suited to situations where a high level of control
with MRVP to develop a target outturn cost (TOC), it is required and appetite for risk is low. However, under
enables a shared understanding and contribution to risk this framework, changes in technology or methodologies
mitigation. The eventual aim of the TOC development is can not be easily adopted as it is unlikely the prescriptive
to understand and eliminate risk by utilising resources, standards or specification would have accounted for
knowledge and relationships across both government future innovations.
and industry. Traditionally, in a D&C environment a
contractor would assign a cost to items or scope they In an innovative culture, regulations (including standards
could not effectively mitigate or eliminate the risk. and specifications) should be flexible and adaptive to
Whereas, in the PDA model, with input from MRPV, enable changes in technology and new practices. This
project teams work to identify and mitigate or eliminate is achieved using performance-based standards and
all risk during the TOC development. specifications which focus on the requirement outcome.
Under this framework, products much achieve a
When considering the barriers to innovation, this model minimum performance using which ever means they
enables novel problem solving and innovative solutions wish. Typically, the performance-based standard or
to reduce or remove risk during the development phase specification will require a series of tests to be
of the project. Further, MRPV is able to draw on undertaken to verify the product, process or system
innovative solutions from across the industry to facilitate achieves the minimum requirement. The risk with this
solutions for individual projects. approach is that the suite of performance requirements
may not account for a behaviour or outcome that was not
5.3 Incentives for Innovation intended to be achieved.
The final feature of the PDA model that encourages In practice, it is very difficult to write a standard or
innovation is the use of financial and non-financial specification entirely using a performance-based
incentives for those projects that develop and adopt framework. Therefore, what more commonly happens to
innovative solutions. Specifically, there is a KPI (key enable an innovative culture, is that standards and
performance indicator), for both contractors and specifications are written predominantly with
designer who develop innovative initiatives for not only performance-based requirements and use more
the project, but that can be adopted more broadly in the prescriptive-based requirements for elements deemed
industry. The assessment of the innovation KPI is based high risk.
on the type of innovation(s) adopted, and defined as;
Looking back at the barriers to innovation, the move
• Continuous improvement is defined as an towards largely performance-based standards and
initiative which is a continuous improvement specifications enables removal of barriers associated
initiative within the project. This is the lowest level of with risk-aversion, regulatory requirements and
innovation that will trigger the KPI. bureaucratic culture. The basic premise of defining the
• Innovation adoption is an initiative which is performance outcome rather than prescriptive
adopted from another project. This is the starting requirements means the basic elements that have led to
point for more broader industry wide innovation and risk-aversion, regulatory requirements and bureaucratic
encourages teams to look at the practices of other culture are retained, in that risk and uncertainty are
projects to implement a new technology, process or managed, however the performance-based approach
system. enables suppliers to propose new technologies,
• Industry application is the highest level achievable techniques and processes.
within the KPI and is defined as an initiative which
has industry wide application. 6.1 Key Standards and Specifications
Contractors and designers are judged on the number The range of standards and specifications MRPV is
and type of innovations adopted by the project, and given working on with DoT is diverse and covers the full range
short term financial incentives, as well as longer non- of technical disciplines within the road transport sector.
financial incentives related to their ongoing performance. It includes new and updated standards and
specifications for plastic noise walls and slip form
6 MODERNISING STANDARDS AND barriers, as well as guidelines for application of spray
SPECIFICATIONS seal pavements. However, the focus of this paper is a
couple of initiatives associated with geotechnical aspects
The second key initiative MRPV are undertaking to drive of road infrastructure, including trialling the use of
innovation at the industry level is through a program of intelligent compaction for road pavement, and creating a
modernising and updating standards and specifications. new technical specification for recycled organics.
It is acknowledged that much of the work associated with
this initiative is done in conjunction with the Department
Go to table of contents
75
Supporting Innovative Design and Construction S. R. Taylor
6.1.1 Intelligent Compaction prescriptive standards and specifications can limit the
type of materials used in civil infrastructure projects. To
Compaction is the process whereby material is placed in conform to a highly prescriptive-based standard or
layers and compressed to increase the density and specification, recycled materials must achieve the same
uniformity of the pavement material. Compaction is inputs as their virgin material counterparts which can be
achieved by a roller and applied to all layers within a road unrealistic.
profile (e.g. subgrade through to asphalt).
In the landscaping discipline, this is particularly relevant
Currently, the most widely used technique used to for recycled organic materials. The Australian Standard
confirm the achievement of compaction is a combination AS 4454 (2012) Composts, Soil Conditioners and
of in-situ spot test with a nuclear gauge density device Mulches specifies the “requirements for organic products
and proof rolling. These methods are limited in that they and mixtures of organic products that are to be used to
only evaluate a small portion of the entire road amend the physical and chemical properties of natural or
composition (e.g. approximately 1 %) and as a result artificial soils and growing media.” The standard is
may miss weak or unqualified compaction areas. largely a prescriptive-based set of requirements, where
physical and chemical requirements of nominated virgin
Intelligent Compaction (IC) is a compaction technology materials are described.
that uses “vibratory rollers equipped with the real-time
kinematic (RTK) Global Positioning System (GPS), Recycled organic materials don’t easily conform with
roller-integrated measurement system (normally requirements of their virgin counterparts. Specifically,
accelerometer-based), feedback controls, and onboard recycled organic materials can vary in quality largely due
real-time display of all IC measurements,” (Chang et al., to the (i) the extent of impurities within the recycled
2011). It is starting to be adopted in Europe, Japan and material (e.g. plastics, paper etc), and (ii) the extent of
North America. processing time allowed to create the compost or soil
conditioner. To enable recycled organic materials to be
The benefits of IC include (Chang et al., 2011); used on civil infrastructure projects, MRPV in
collaboration with DoT and ARRB, are in the process of
• IC allows mapping of each compaction layers to developing a technical specification for recycled
enable real-time identification of weak spots for organics. The intent is to define the performance
corrective actions prior to the compaction of the requirements for a range of uses of recycled organic
upper layers. material, that link back to the classifications used within
• IC provides the means to maintain a consistent AS 4454 (2012). These include for landscaping (e.g.
rolling pattern for 100% coverage of a construction planting, hydro-mulch, turf topdressing, soil conditioning,
area. fertiliser), erosion control (e.g. soil compaction
• IC enables consistent rolling patterns under lower mitigation, filter berms, and compost blankets), and
visibility conditions such as night paving operations. biofiltration (e.g. filter media). By developing a technical
• IC technology enables digital record collection for specification, the innovative use of recycled organics is
future investigation. available for all projects within the industry.
Go to table of contents
76
Supporting Innovative Design and Construction S. R. Taylor
REFERENCES
Go to table of contents
77
ABSTRACT
Building Information Modelling (BIM) is one of the important processes being adopted by the construction industry as it
provides a collaboration platform in conjunction with technical standards for interoperability over the lifecycle of an asset.
However, geotechnical analysis engaging numerical tools has yet to leverage the BIM benefits due to the lack of effective
interoperability means, which not only results in unnecessary remodelling and rework at a cost of labour and computational
waste, but also with possibility of errors, misinterpretation and omission of information. Using a trinocular underground
station as an example, a workflow underpinned by heuristic techniques is proposed to enhance the interoperability between
a BIM design authoring tool (Revit) and a numerical modelling tool (FLAC3D) for geotechnical analysis. A BIM-based
multiple LoD (levels of detail) model framework is proposed to represent different information requirements for different
purposes of BIM use throughout the project lifecycle. Leveraging the associated geometry and semantics, techniques of
parametric modelling, data manipulation via visual and traditional programming are engaged to bridge BIM and numerical
modelling for geotechnical analysis on different design scenarios. The simulated results are visualised through a backward
automation cycle to Revit for design optimisation. The presented solution offers and automates an error-free design-to-
design workflow solution and therefore enables efficient exploration of design scenarios and design optimisation.
Keywords: Building Information Modelling (BIM), numerical modelling, automation, underground, metro station
Go to table of contents
BIM to Numerical Modelling Interoperability for Geotechnical Design of Underground Metro Station M. Q. Huang, X. L. Chen, T. J. Sia and Q. B. Zhang
Figure 1. Demonstration of creating synthetic borehole and cross-section (Top: the main interface for user
inputs including selection of model of interest, locating synthetic borehole and viewing custom-defined cross-
section; bottom left: the resulted information of synthetic borehole including relevant geographical and
geological data; bottom right: the cross-sectional geological information). (Huang et al., 2021)
A multi-arch cavern is a common structure adopted for common means to simulate different engineering
underground metro stations with columns and walls scenarios of underground excavation and structure short-
providing support to the wide-span. Examples of stations term to long-term stability. Since BIM models consist of
using such structure include Moscow Mayakovskaya geometry and semantics, BIM could act as a pre-modeller
station (Shilin et al., 2016), Tokyo Kiyosumi-shirakawa and post-viewer for numerical modelling, the BIM-
Station (Konda, 2003), and Beijing Badaling Great Wall numerical modelling interoperability, therefore, allows an
station (Li et al., 2020). Figure 2 illustrates an example of error-free exchange of information between the two
typical triple arched cavern and support systems. The interfaces.
construction methods or philosophies, such as sequential
excavation method (SEM), sprayed concrete lining (SCL),
New Austrian Tunnelling Method (NATM), and the
‘Analysis of the Controlled Deformation in Rocks and
Soils (ADECO-RS)’ emphasised the significance of
excavation and construction with adequate observation
and support measures. The components prescribed in
these methods should be modelled. Their mechanical
properties should be taken care of given the information
modelling approach is undertaken.
The stability analysis of the cavern structure is essential Figure 2. Example of a typical triple-arched cavern and
to geotechnical engineering. With computer software and support systems.
information technology becoming easily accessible,
numerical modelling methods such as the Finite Element BIM encompasses a set of tools, technologies and
Method (FEM), Finite Difference Method (FDM) and processes ranging from parametric modelling (Ninic et al.,
Discrete Element Method (DEM) have been used as
Go to table of contents
79
BIM to Numerical Modelling Interoperability for Geotechnical Design of Underground Metro Station M. Q. Huang, X. L. Chen, T. J. Sia and Q. B. Zhang
2021) to project lifecycle management (Wang and Zhang, setting of meshing parameters, initial and boundary
2021) that are driving the digital transition of the conditions, and excavation and ground support
Architecture, Engineering and Construction (AEC) properties.
industries. Many of the benefits suggested by adopting
BIM and digital information technologies, such as 3.1 Multi-LoD station BIM
improved visualisation and cross-disciplinary
communication and collaboration, are desired for The station model adopts the concept of “Level of Detail
underground construction. For example, the downstream (LoD)”, which specifies what objects are included in the
application of BIM has been investigated for tunnels. In model and the degree of detail with which each object is
Ninić et al.(2018, 2020 a and b), BIM-to-FEM and BIM-to- modelled (Sacks, 2018).
IGA workflows were developed, incorporating machine
learning techniques, to deliver solutions that focus on
automating model generation and simulation for
numerical analysis on shield tunnelling induced
settlement and decision aid on parameter optimisation in
soil condition. Fabozzi et al. (2021) has also studied BIM-
to-FEM and FEM-to-BIM interoperability via the use of
several software tools, including commercially available
rail design modelling package.
Go to table of contents
80
BIM to Numerical Modelling Interoperability for Geotechnical Design of Underground Metro Station M. Q. Huang, X. L. Chen, T. J. Sia and Q. B. Zhang
parameters. The to-be proposed methodology largely parametric families are created for this study, mainly
relies on computational design concepts, which have concerning the geotechnical analysis that will be
been in use for architectural designs since emerging in performed. Example of an outer shell and rock bolts as
the 60s, and has more or less influenced the development forms of excavation support are illustrated in Figure 5.
of BIM tools. Both parametric design and generative The triple arched primary lining as the outer shell is
design are popular terms considered in the scope of essentially created by extrusion, which is a simple
computational design (Caetano et al., 2020). Establishing parametric modelling procedure.
a component (referred to as a family in the selected BIM
software), for example a bolt, based on constraints of
parameters and rules denotes parametric design for a
single element.
Go to table of contents
81
BIM to Numerical Modelling Interoperability for Geotechnical Design of Underground Metro Station M. Q. Huang, X. L. Chen, T. J. Sia and Q. B. Zhang
(b)
Go to table of contents
82
BIM to Numerical Modelling Interoperability for Geotechnical Design of Underground Metro Station M. Q. Huang, X. L. Chen, T. J. Sia and Q. B. Zhang
numerical modelling loop for geotechnical analysis. Tunnelling and Underground Space Technology 108,
Currently, the exchange module is developed for the 103677.
specific interfaces that it is connecting, this means that Koch, C., Vonthron, A., König, M., 2017. A tunnel information
modelling framework to support management, simulations
certain adaptations are required if other software tools are
and visualisations in mechanised tunnelling projects.
employed for BIM and numerical simulation. Another Automation in Construction 83, 78-90.
limitation the current exchange module may suffer from is Konda, T., 2003. Reclaiming the underground space of large
the level of complexity in meshing. The success of FEM cities in Japan, Proceedings of the ITA World Tunnelling
or FDM simulation largely relies on the quality of meshes Congress, pp. 1-13.
while examples used to testify the workflow have only Li, R., Zhang, D., Fang, Q., Liu, D., Luo, J., Fang, H., 2020.
involved relatively simple geometrical structures without Mechanical responses of closely spaced large span triple
comprehensive set of interfaces. A potential approach tunnels. Tunnelling and Underground Space Technology
105, 103574.
that would be investigated in the next stage is to engage
Ninic, J., Alsahly, A., Vonthron, A., Bui, H.-G., Koch, C., König,
a meshing program as a middle-ware or to experiment M., Meschke, G., 2021. From digital models to numerical
with the octree mesh in FLAC3D. Ultimately, proper analysis for mechanised tunnelling: A fully automated
translators could be developed based on BIM data design-through-analysis workflow. Tunnelling and
schema, such as the industry foundation class (IFC). Underground Space Technology 107, 103622.
There is yet standard for underground construction based Ninić, J., Bui, H.G., Meschke, G., 2020a. BIM-to-IGA: A fully
on IFC while research and development efforts are automatic design-through-analysis workflow for
ongoing by buildingSMART, an international industry segmented tunnel linings. Advanced Engineering
Informatics 46, 101137.
body aiming at improving interoperability between
Ninic, J., Giang Bui, H., Meschke, G., 2018. Parametric Design
software applications used in the construction industry, and Isogeometric Analysis of Tunnel Linings within the
cooperating with other organisations focusing on Building Information Modelling Framework. CEUR
subsurface utilisation, such as the International Tunneling Workshop Proceedings.
and Underground Space Association (ITA-AITES). Ninić, J., Koch, C., Vonthron, A., Tizani, W., König, M., 2020b.
Integrated parametric multi-level information and
5 CONCLUSION numerical modelling of mechanised tunnelling projects.
Advanced Engineering Informatics 43, 101011.
Sacks, R., Eastman, C., Lee, G., Teicholz, P., 2018. BIM
This study aims to provide a solution to tackle the problem
handbook: A guide to building information modeling for
of lacking open standard data models that cover the owners, designers, engineers, contractors, and facility
context-specific knowledge required for underground managers. John Wiley & Sons.
construction. By leveraging the parametric modelling Shilin, A., Kirilenko, A., Znajchenko, P., 2016. Complex
technique of BIM, the workflow could mitigate the efforts reconstruction project of Mayakovskaya metro station in
of numerically retrofitting and re-modelling at feasibility the centre of Moscow. Structural Analysis of Historical
and earlier design stage. The enabled a design-to-design Constructions—Anamnesis, Diagnosis, Therapy, Controls;
workflow with the reversed visualisation loop creates Van Balen, K., Verstrynge, E., Eds, 1736-1741.
Wang, G., Zhang, Z., 2021. BIM implementation in handover
opportunity for backanalysis and design optimisation.
management for underground rail transit project: A case
study approach. Tunnelling and Underground Space
6 ACKNOWLEDGEMENTS Technology 108, 103684.
Zhu, H., Wu, W., Li, X., Chen, J.Q., Huang, X.B., 2017. High-
This work was financially supported by the ARC precision acquisition, analysis and service of rock tunnel
Nanocomm Hub (IH150100006). The first author would information based on iS3 platform. Chin. J. Rock Mech.
like to acknowledge the Australian International Eng. 36. 2350–2364.
Postgraduate Research Scholarship and Monash
Graduate Scholarship. The technical information and
photos used in this paper are from the Metro Tunnel
Project in Melbourne (https://metrotunnel.vic.gov.au), and
the models are partially supported by the Government
and Industries.
REFERENCES
Go to table of contents
83
ABSTRACT
This paper reviews the best practices of trackbed design for railway projects. Various existing methods have been studied
and recommendations for more economical design are provided. The analytical/empirical methods from various standards
such as UIC, AREMA, British Rail, and Australian standards, as well as other commonly used methods such as Raymond
and Li-Selig are compared based on a typical track embankment cross section. The outcome was then evaluated against
2D and 3D numerical models. Incorporating numerical methods is shown to render considerable reductions in the required
prepared subgrade/structural fill materials and allow for assessment of long-term design issues, such as subgrade shear
failure due to excessive plastic deformations.
Go to table of contents
On best practices for trackbed design N. Yousefpour
Go to table of contents
85
3.5 Li-Selig Method granular layer thickness. This normally causes a heave
at trackside which can block the proper drainage. Water
The Li-Selig Method (1998) [5,6] can be used based on coming from the granular layer can be trapped in the
the allowable stress at the subgrade surface (allowable subgrade depression areas which can aggravate the
cumulative plastic strain), as well as the allowable failure.
subgrade deformation. Figure 5 provides examples of
design charts. In these charts, H is granular material 2) Excessive Plastic Deformation due to Repeated
thickness. Iԑ and Ip are referred to as the strain influence Loading: This can cause ballast pockets under the track
factor and the deformation influence factor, respectively: (Figure 7) and happens predominantly in soft soil
subgrades.
ARTC and ASA [7,8], VLine [9] and MTM [10] standards
provide relatively similar guidelines for trackbed design. A
minimum of 0.3m ballast and 0.15m subballast or capping
is recommended (total granular layer thickness of 0.45m).
Typical thickness/depth for the prepared
subgrade/structural fill layer is provided based on the
CBR of the natural subgrade as presented in Table 3.
Go to table of contents
On best practices for trackbed design N. Yousefpour
An example is worked based on the discussed methods An Elasto-Plastic, Mohr-Coulomb constitutive model was
in this section to compare the required thickness of the implemented, adopting material parameters as per Table
granular layer (H=Hb+Hs) from different methods. The 5. The water table was not considered within the model,
example is based on an axle load of 160 kN. It is assumed assuming a dry subgrade (zero pore water pressure).
that granular layer will lay on an underlaying layer with
QS2 soil quality with CBR~5% (~150kPa). Table 4
provides the estimated required thickness ranging
between 0.4m to 0.5m.
4.2 Geometry
As shown in the analytical example above, this condition Figure 9. Train loads
renders a structural fill of 0.5m according to Australian
standards. However, in this case, a 0.3m of prepared 4.4 Results
subgrade from stabilised subgrade material with CBR=15
has been proposed to replace the structural fill to save on A summary of model outputs, including stress and strain
cost. Using a Ls-Dyna model, the proposed design is distribution, as well as deformations and plastic strains
being validated by checking the stress and strain are provided below (see Figures 10 to 14). The shear
developed in the subgrade and by checking for bearing stress within the subgrade is below 15 kPa and the
failure. vertical stress is below 50 kPa, the subgrade allowable
stress, hence no plastic stain is observed. The
The train loading assumes a 23 tons axle load (two deformation at the sleeper level shows to be less than 5
bogies, four axles per car), resulting in 450kN/m linear mm, which is below the common acceptable criteria of 10
load for each wheel applied on top of the rail track over mm (1/2 in). This validates the proposed design of 0.3m
the length of a sleeper (0.25m) (see Figure 9). prepared subgrade with higher CBR, instead of the 0.5m
structural fill with CBR=8 required by Australian
standards.
Go to table of contents
87
Results of LS-Dyna model was then compared with an maximum shear stress is around 15 kPa in the subgrade
equivalent Plaxis 2D model of the longitudinal cross and around 27 kPa in the prepared subgrade layer,
section. As shown in Figure 15 and 16, the 2D model whereas Plaxis shows maximum of 12 kPa in subgrade
shows a maximum stress of 140 kPa in ballast, about 20% and 40kPa in the prepared subgrade. The 2D model
lower stress as compared with 3D LS-Dyna. Similarly, underestimated the maximum stress, hence the
looking at the shear stress distribution, from Ls-Dyna the deformations, in the trackbed layers.
5 CONCLUSIONS
Figure 11. Vertical deformations Among the discussed analytical design methods, Li-
Selig is the only method that considers both allowable
stress and deformation at subgrade with respect to the
train loading cycles, providing a more reliable
trackbed design. Incorporating 3D numerical
Go to table of contents
On best practices for trackbed design N. Yousefpour
These methods also allow for identifying potential long- Li, D. and Selig, E., Method for Railroad Track Foundation
term issues with bearing failure or excessive plastic strain Design. I: Development, Journal of Geotechnical and
due to cyclic loading. Although it is common for clients to Geoenvironmental Engineering, ASCE, 316-322, 1998.
Selig, E. and Waters, J. M., Track Geo-technology and
demand compliance with standards, there is often
Substructure Management, Thomas Telford, 1995.
provisions for applying more rigorous design methods ARTC (Australien Rail Track Corporation), Track & Civil Code of
beyond the standards by getting client approvals for Practice, Railway Earthworks - Supplementary Appendix,
derogations in early phases of a project. ETC-08-02, 2006.
ASA (Asset Standards Authority) Earthworks and Formation
6 ACKNOWLEDGEMENTS Standard, Version 1.0, 2018. Supplementary Appendix to
ARTC.
This study was sponsored by Arup University. VLine, Network Infrastructure Standard (NIST) 2659, Earthworks
and Drainage Standard, 2019.
MTM (Metro Train Melbourne), Earthwork and Formation
REFERENCES Specification, L1-CHE-SPE-178, Version 2.0, 2018.
Go to table of contents
NOTES
Go to table of contents
96 2021 AGS VICTORIAN SYMPOSIUM INNOVATIONS IN GEOTECHNICAL DESIGN
Go to table of contents
2021 AGS VICTORIAN SYMPOSIUM INNOVATIONS IN GEOTECHNICAL DESIGN 97
Go to table of contents
2021 AUSTRALIAN GEOMECHANICS SOCIETY
VICTORIAN SYMPOSIUM
Innovations in Geotechnical Design
VICTORIA CHAPTER