Proceedings of ICGeotechnics

IC GEOTECHNICS YOGYAKARTA, 24-26 JULY 2018
Proceeding of International Conference on Geotechnics

“Sustainable Development Challenges in Geotechnics”
Main Organizer:
Universitas Universitas
Atma Jaya Yogyakarta Gadjah Mada
Supporting
Co-organizer: Institution:
Indonesian Society
for Geotechnical Engineering
Proceeding
International Conference on Geotechnics (IC Geotechnics)
Royal Ambarrukmo Plasa

Yogyakarta, 24 – 25 July 2018
Organized by:
Universitas Atma Jaya Yogyakarta
Universitas Gadjah Mada
National Taiwan University of Science and Technology
University of Southern Queensland
Publisher:
IC Geotechnics
Jl. Grafika No. 2 Kampus UGM, Yogyakarta 55281
Phone: +62-274-513665
Proceeding
International Conference on Geotechnics (IC Geotechnics)
International Conference on Geotechnics, 24-26 July, 2018 Yogyakarta, Indonesia
Chair
Luky Handoko
Vice Chair
Ahmad Rifa’i
Reviewer
Ahmad Rifa’i
Adhitya Yogya Purnama
Fikri Faris
Sito Ismanti
Luky Handoko
Jim Shiau
Ou Chang-Yu
Editor
Fikri Faris
Publisher:
IC Geotechnics
Jl. Grafika No. 2 Kampus UGM, Yogyakarta 55281
Phone: +62-274-513665
Foreword
All praise is due to God, the Beneficent, and the Merciful. Through exhaustive collaboration of Atma
Jaya Yogyakarta University, Gadjah Mada University, National Taiwan University of Science and
Technology, and University of Southern Queensland and the support by Indonesian Society for
Geotechnical Engineering, we proudly present the first proceeding of International Conference in
Geotechnic, themed: “Sustainable Development Challenges in Geotechnics”.
2
For the past decade, the field of geotechnics is facing tremendous challenges due to the abrupt
development of construction in infrastructure. The challenges comprise issues related to new technology
of modeling and design, soft soils, ground improvement, tunneling, landslides, environmental issues and
much more. Therefore, good communication and knowledge sharing between stake holders, engineers,
academicians and government are needed to overcome the issues. The objective of International
Conference in Geotechnics is to share the knowledge and experience of the recent advancement and
development of geotechnical engineering all around the world with emphasis on the sustainability issue in
infrastructure development.
We would like to appreciate everyone who has contributed so much to the symposium and in the
preparation of this proceeding. Hopefully, this proceeding can give benefit to society and science,
especially for the field of geotechnics.
Yogyakarta, July 2018

Chief Editor
Fikri Faris
3
Table of Contents
Title page i
Title back page ii
Foreword iii
Table of Contens iv
Pile Load Test Frequently Asked Questions 1
(T.L Gouw, A. Gunawan)
Active Faults Identification for Dam Safety Against Earthquakes 13
(D. Djarwadi)
3D Assessment of Rainfall-induced Slope Movements and Risk Mitigation Strategies 20
(Wei He, Barry Kok, Sangmin Lee)
Bridge Approach Embankments on Rigid Inclusions 29
(M. Rizal & K. Yee)
Design of Rapid Impact Compaction at the New Yogyakarta International Airport 39
(Mathew Sams, Wei He, Jeremy Gamaliel, and Barry Kok)
Effects of Model Scale Due to Displacement Factor for Nailed-slab Pavement System 49
(Anas Puri)
Initial Recommendation Criteria for Distinguishing Between Landslides 54
and Mudflows Based on Several Case Studies in Java and Bali
(B. Widjaja, D. Pascayulinda)
Slope Stability of Metamorphic Rock Based on Limit Equilibrium 61
Method of Poboyo Gold Mine, Palu, Central of Sulawesi
(Sriyati Rahmadani, Ahmad Rifa’i, Wahyu Wilopo, and Kabul Basah Suryolelono)
Deterioration Depth of Cement Treated Clay Under Sulfate Exposure 67
(T. Pradita, L. Handoko, S. Gunawan, and J. T Hatmoko)
Analysis of Failure Base Plate Anchor Flare Stack Foundation and Repair Method 78
(Sulardi)
The Analysis of Rockbolt Reinforcement on the Tunnel 83
by Mohr Coulomb Approach Model and Hardening Soil Model
(Hanindya K. Artati and Dias Dwi Hatmoko)
Effect of Matric Suction Change on Pile Foundation Capacity in Unsaturated Soils 94
(H. Pujiastuti, A. Rifa’i, A. D. Adi, and T. F. Fathani)
4
Pile Load Test Frequently Asked Questions
T. L. Gouw
Associate Professor, Post Graduate Program, Universitas Katolik Parahyangan, Bandung, INDONESIA
gtloffice@gmail.com
A. Gunawan
Department of Civil and Environmental Engineering, Bina Nusantara University, Jakarta, INDONESIA
agunawan@connect.ust.hk
ABSTRACT
There are a few available methods to obtain load-settlement curve of a pile. Likewise, there are many methods to determine the
ultimate pile capacity from a load-settlement curves. Although pile load tests have been widely used over the past decades, there
are still many questions regarding its practice and interpretation. Frequently asked questions include: when does a pile test
considered to have failed? From an economic point of view, a failure in pile load test can cost quite a lot of money. To what load
can the pile be loaded till it is considered to have failed? Can a pile loaded to failure still be used as a working pile? Is pile driving
analyzer (PDA) test reliable? Can PDA test replace static load test? Is it necessary to calibrate PDA test results with static load
test results? Why is PDA test result interpreted as 1 dimensional wave and not 3 dimensional? What is bidirectional pile load
test (also known as O’cell)? When should O’cell be used? Can a pile tested with O’cell be used as a working pile? What are the
differences between kentledge load test, static load test with reaction piles and bidirectional pile load test? Do the three different
pile tests produce the same results? This paper aims to shed light on these questions.
Keywords: Pile static load test, Dynamic load test, Bidirectional test, Ultimate pile capacity, Fail Pile
1 INTRODUCTION • PDA test is interpreted with one dimensional

theory, why not three dimensional?
Foundation piles have been used for over one hundred • Can a PDA test be manipulated?
years. There are many methods to construct the piles.
• What is a bidirectional load test or O’ Cell? When
Likewise, there are also many methods to test the pile
is it necessary to apply this test method?
capacity. In Indonesia, foundation piles are very
• Can a pile tested by bidirectional load test still be
common, and engineers in Indonesia are willing to
used as a working pile?
adopt state-of-the-art testing methods. From the
common kentledge loading test, static load test with • Will a pile tested by kentledge load test, static load
reaction piles, dynamic loading test (also known as test by reaction pile, and bidirectional test give the
PDA - pile driving analyzer), to the more recent one, same results?
bidirectional test, also known as Osterberg Cell or O- The paper first discusses what ultimate pile capacity is,
cell for short. Although these testing methods have the principles of each pile tests, followed by answers to
been widely adopted in Indonesia, there are still the above questions.
questions regarding these testing methods. Often,
engineers have different perspective on the practice of
these testing methods. This paper aims to shed light to 2 ULTIMATE PILE CAPACITY
the following frequently asked questions: Eurocode 7 (BS EN 1997-1, 2004) defines ultimate pile
capacity, also known as ultimate limit states, as
• A pile load test should not be determined as failure compressive or tensile resistance failure of a single or
as the project owner has spent thousands or tens of piles system. However, according to Fellenius (2017),
thousands of US dollars for it. So, in what scenario “Ultimate pile capacity” is a very imprecise concept in
does a pile load test considered to have failed? most soil conditions. This can be clearly seen from a
• Can a pile tested to “failure” still be used as a typical load-settlement curve shown in Figure 1. As
working pile? shown in Figure 1, pile “capacity” continues to increase
• Is PDA test reliable enough to replace static load the further it is loaded. So, when does a pile really fail?
test? Is it necessary to calibrate PDA test results Figure 2 shows the 3 types of load-settlement curve that
with static load test results? can be obtained in field. The first type is a general
failure, which as stated previously, the pile capacity
continues to increase with pile movement. The second
1
International Conference on Geotechnics
type is punching failure with a relatively constant (Fellenius, 2017). The relative movement required is
capacity, in which pile continues to move under independent of pile size, but dependent on soil type and
constant load. The third type is punching failure with a roughness of the pile.
reduction in capacity. Geotechnical pile failure only
occurs when type 2 or type 3 occurs. For type 2 and 3, The magnitude of shaft resistance mobilized is
a true “ultimate” capacity can be defined. It is dependent on the magnitude of relative movement and
important to differentiate the peak and ultimate soil type. This behavior is easily observed from direct
capacity for type 3. These three types of curve obtained shear test results shown in Figure 3. For loose sand or
can be explained with ultimate shaft resistance and toe normally consolidated clay, shaft resistance (shear
resistance. resistance) continues to increase with shear
displacement. Once the displacement is large enough,
the ultimate shaft resistance is reached, in which the
resistance stays constant. For dense sand or overly
consolidated clay, shaft resistance continues to increase
until a peak shaft resistance is reached. From there, the
shaft resistance decreases with movement, until
ultimate shaft resistance is reached.
Figure 1. Typical load-settlement curve from a pile-load test

in medium dense sand.
Figure 3. Shear stress versus shear displacement on dense

and loose sand (Modified after Das, 2014).
2.2 Toe Resistance

Unlike shaft resistance which has an ultimate value, toe
resistance does not. The load-movement of pile toe is a
function of the stiffness and effective stress of the soil.
Figure 4 shows an example of load versus pile toe
movement. The figure shows the unit toe resistance
developed in a 1.5 m and 1.8 m diameter piles against
Figure 2. Three types of load settlement curve: 1 – general
failure, 2 – punching failure with constant capacity, 3 –
their respective movement. It can be seen that even
punching failure with reduction in capacity. after a large movement of 150 mm, which is nearly
10% the pile diameter, the load-movement curve has no
indication of reaching “failure”.
2.1 Ultimate Shaft Resistance
Development of shaft resistance is the consequence of 2.3 Failure in pile
relative movement between the pile and soil. If the pile
The three types of failure shown in Figure 2 can be
settles more than the soil, a positive shaft resistance is
explained from the ultimate shaft and toe resistance.
generated. Whereas, when the soil settles more than the
For general failure, although ultimate shaft resistance
soil, a negative shaft resistance is generated. To fully
has been mobilized, the toe resistance can continue to
mobilize shaft resistance, very small relative
develop resistance as it is further loaded. For punching
movement is required, often only a few millimeters
failure, the pile has majority of its capacity from the
2
shaft resistance. Hence, when the shaft resistance is Figure 1. In this example, the ultimate pile capacity is
fully mobilized, the toe does not have sufficient 2300 kN and 1480 kN based on the Eurocode 7, and
capacity to sustain the current load and self-weight of Indonesian Standard respectively. Based on the 2
pile. For the punching failure with reduction in criteria, 35% difference in ultimate capacity is
capacity, the shaft is surrounded by dense sand or obtained. Naturally, the factor of safety applied for the
overly consolidated clay, while the toe sits on soft or two failure criteria are different. For Eurocode 7, partial
loose soil. When the shaft resistance is mobilized factors design is used, while for Indonesian Standard, a
beyond its peak resistance, the drop in shaft resistance factor of safety of 2.5 is used for deep foundation.
is higher than the increase in toe resistance. Hence, an
overall drop in pile capacity occurs until the ultimate Other than settlement-based failure criteria, there are
shaft resistance is mobilized, and the pile capacity also other failure criterion which attempts to separate
remains constant. the initial relatively gentle part of the curve and the
latter steeper part of the curve. Example of such
criterion includes Hansen 80% and 90% criteria (1963),
Chin-Kondner extrapolation (Chin, 1971; Kondner
1963), Decourt extrapolation (2008), DeBeer
intersection load (1968) and many others.
Before understanding the different types of pile load

test, an engineer needs to understand the basis of
determining ultimate pile capacity. In essence, the so
called ultimate capacity is a definition and hardly a real
failure load! With this knowledge, one can have a
clearer picture on the advantages and disadvantages of
each testing methods, as well as their appropriateness
in different situations.
Figure 4. Unit toe resistance measured on a 1.5 and 1.8 m 3 STATIC LOAD TEST
diameter bored pile constructed in silty sandy clay and
clayey sand (Fellenius, 2017). Static load test is a test whereby tested pile is loaded
axially, either through dead weight (kentledge) or
Punching failure with relatively constant capacity and reaction piles/frames. The choice of pile tested is
punching failure with reduction in capacity is rather usually based on the pile installed in the most adverse
rare. For general failure, there is no well-defined soil conditions. This is to ensure that the obtained
ultimate pile capacity. However, to ease results are most conservative, and there is no
communication between engineers, “ultimate” pile overestimation of pile capacity in other areas.
capacity is required. Hence, failure criterion is used to The main purpose of static load test is to obtain load
determine ultimate pile capacity. versus movement relationship of said pile. From the
results, based on certain failure criterion, the ultimate
2.4 Failure criterion
pile capacity can be derived. In addition to
For pile which experience general failure, failure determination of ultimate pile capacity, creep behavior
criterion need to be established to determine ultimate can also be determined by interpretation of movement
pile capacity. Eurocode 7 states that ultimate state can versus time relationship under constant load. Another
be defined as the load which causes excessive useful information that can be obtained through static
displacement (BS EN 1997-1, 2004). Therefore, the load test is the rebound behavior, when the pile tested
keyword is acceptable displacement, and a lot of is unloaded (BS EN 1997-1, 2004). More useful
standards uses pile settlement magnitude to define information can be obtained by using instrumented
ultimate pile capacity. piles, e.g. shaft resistance and toe resistance (Fellenius,
2017). In the next sections, the procedure of static load
In Eurocode 7, 10% of the pile base diameter of test by kentledge and reaction piles are discussed. Their
settlement is adopted as failure criterion (BS EN 1997- results are also discussed.
1, 2004). For Indonesian standard, the failure criterion
is 25 mm for piles with diameter smaller than 0.8 m, 3.1 Kentledge System
and 4% pile diameter for piles with diameter larger than
0.8 m (SNI 8640:2017). The failure criterion is then Figure 5 and 6 shows a schematic diagram and
used to obtain the ultimate pile capacity, as shown in photograph of static load test with kentledge system,
respectively. A pile capacity can range from a few tons
3
to hundreds of tons. To load a pile to that level,

sufficient reaction force is required. One method to
provide the reaction force is by kentledge. As the name
suggests, it is the stacking up of slabs or blocks of
concrete (kentledge) above cross beams/platform to act
as a counterweight during loading of the pile. ASTM
D1143 state that the total weights should at least be
10% larger than the maximum load that is going to be
applied. The center of the cross beams need to be in the
center of the pile to prevent eccentric loading. Only 1%
eccentric loading is allowed, and eccentric distance of
25 mm. To stabilize the cross beams, temporary cribs, Figure 5. Schematic diagram of static load test with
made of timber or concrete, need to be built. Care must kentledge system (ASTM D 1143, 2007).
be taken in determining the clear distance between the
pile to the cribs. The distance is important as the cribs
take on some of the load provided by the kentledge.
This load is transferred to the soil underneath the cribs,
causing some ground deformation. As stated in Section
2.1, shaft resistance is a function of relative movement
between pile and soil. Therefore, the cribs must be
placed far enough to not let the deformation due to cribs
weight affect the shaft resistance of tested pile. As a
general guideline, ASTM D1143 states that the clear
distance should not be less than 1.5 m.
3.2 Reaction Piles or Anchors Figure 6. Photograph of static load test with kentledge
system (Khmer D&C, 2018).
Alternative to kentledge, reaction piles or anchors can
be used to provide the reaction force for pile loading.
Figure 7 and 8 shows a schematic diagram and
photograph of reaction pile system. Care must be taken
to ensure sufficient resistance from anchors or reaction
piles. Movement of anchors or reaction piles also must
be measured to calculate the net movement of the tested
pile. Another thing to note is the difference requirement
in clear distance between tested pile and its anchors or
reaction piles. ASTM D1143 states that the required
clear distance is 5 times the largest pile/anchor
diameter (can be the test pile or reaction pile) or 2.5 m,
whichever is larger. The required clear distance is
larger than kentledge system because for kentledge Figure 7. Schematic diagram of static load test with anchored
system, the larger the load applied on tested pile, the reaction frame (ASTM D 1143, 2007).
lower the load on the cribs. However, when reaction
piles or anchors are used, the larger the load applied on
tested pile, the larger the opposite load acts on the
reaction piles or anchors.
3.3 Loading procedure

Ideally, the load applied should reach a “failure” that
reflects the ultimate axial static compressive load
capacity of the pile. Care must be taken that the load
applied does not exceed the safe structural capacity of
the pile, or the loading apparatus (hydraulic jack). As
pile capacity changes with time (setup effect, when
strength is gained; relaxation, when strength decreases) Figure 8. Photograph of static load test with reaction piles
(Structville, 2018).
4
a qualified engineer should specify the waiting period In Indonesia, the most commonly used loading
before testing. Apart from the above, for cast-in-place procedure is the second procedure. Loading up to 200%
pile or bored pile, sufficient time should be given for design load, and ultimate pile capacity is taken from 25
the concrete to gain adequate strength. mm settlement for piles with diameter smaller than
0.8 m, and 4% pile diameter for piles with diameter
ASTM D1143 approves 7 loading procedure. The first larger than 0.8 m (SNI 8640:2017).
procedure is called ‘quick test’, the pile is loaded in
increments of 5% of the anticipated failure load. After 3.4 Results, Failures in Execution of Static Load Test
reaching 100% failure load, the pile is unloaded at 5 to Results from a non-instrumented pile static load test
10 equal decrements. Each interval of increment and comes in the form of load-settlement curve (an example
decrement must be between 4 to 15 minutes. The is shown in Figure 1 and 2). From the load-settlement
second procedure is called ‘maintained test’. In this curve, the ultimate pile capacity can be determined
test, the pile is loaded in 25% increments of the design using failure criterion based on local national
load until 200% design load is reached, unless failure standards. However, without any geotechnical
is reached first. Each increment is maintained until the instrumentation, it is impossible to derive the shaft or
movement does not exceed 0.25 mm per hour toe resistance from the load-settlement curve alone.
(minimum 2 hours to verify the movement rate). After The importance of instrumentation is discussed in the
200% design load is reached, maintain the load for 24 next section.
hours unless the 0.25 mm per hour criteria is reached.
After the last loading step, unloading can be initiated at Failures in execution of pile test should not be confused
25% of the maximum test load per unloading step. Each with pile loaded to failure. For execution failure, it
unloading step is 1 hour. If failure is reached (i.e. 0.25 means the inability to produce a load-settlement curve,
mm per hour is not achieved), maintain the maximum or ‘unacceptable’ results. Although failures in static
load possible until the total axial movement equals to load test is highly unlikely if the guidelines laid out by
15% of the pile diameter. Care must be taken to ASTM D1143 are followed, they can still happen.
maintain the load in hydraulic jack, as load may
decrease as the pile deforms. The third procedure is a Failures usually occur due to unexpected malfunction
reloading procedure which can be conducted after the of loading apparatus during testing, e.g. hydraulic jack
second procedure, it is called ‘loading in excess of cannot apply or unable to maintain the correct load.
maintained test’. After the test pile is fully unloaded, Another failure could be due to malfunction of
the pile is reloaded at 50% of the design load in 20 measuring apparatus, e.g. misalignment of dial gauge.
minutes interval until the previous maximum load is Other source of errors is human negligence, such as
reached. From the previous maximum load, the pile is misreading of measurements. When the following
further loaded in 10% increments of the design load ‘accidents’ happen, the results may become
with 20 minutes interval between increments. The 10% inappropriate to be used. Failures can also happen prior
increments are continued until the settlement reached to the load test. For example, miscalculation of the
15% of the pile diameter. If failure does not occur, hold bearing capacity required by the cribs or eccentric
the final load for 2 hours, and unload at 4 equal placement of kentledge. When the aforementioned
decrements in 20 minutes interval. The fourth mistake occurs, the kentledge can experience bearing
procedure, named ‘constant time interval loading capacity failure, as shown in Figure 9. When this
test’ is similar to the first procedure, but the loading happens, it is likely that the test pile could not be used
increments is 20% of design load in 1-hour intervals. for testing, and another pile need to be chosen. Other
Unloading is also conducted in 1-hour intervals. The miscalculation include underestimation of the
fifth procedure is ‘constant rate of penetration test’. kentledge required. More than often, due to
As the name suggest, the pile is loaded at a rate of 0.25 conservatism in design load, and maximum load
to 1.25 mm per minute for cohesive soil; 0.75 to 2.5 applied on test pile being a function of design load, the
mm per minute for granular soil until a total penetration maximum load applied may not be enough to settle the
of 15% of the pile diameter. The sixth procedure is pile more than the required failure criterion. In this
‘constant movement increment test’. In this case, the results from the static load test can become
procedure, the pile is loaded at 1% movement very uninformative, with the exception that the design
increments. The load applied on pile is varied to capacity can be higher than what was initially
maintain every 1% movement until load variation is calculated.
less than 1% of the total load. Likewise, continue
loading of pile until 15% movement is achieved. The Failures in static load tests cost a lot of time and money.
pile is then unloaded at four equal decrements. Accidents that occur in Figure 9 can be prevented and
should not be allowed to occur. Accidents that happen
5
due to loading or measuring apparatus can also be as a restrike on driven piles with a pile driving hammer
reduced by doing frequent maintenance and after a set up period, or on bored piles with a drop
calibration. hammer. Due to the nature of the test, it is much
cheaper and quicker to conduct a PDA test than a static
load test.
4.1 Test Apparatus, Instrumentation and Test

Procedure
For PDA test, the test apparatus only involves impact
device, which can be a pile driver or drop hammer.
Magnitude of impact should be controlled to prevent
damage on the pile cap. Pile cushion, usually plywood,
can be used to reduce the impact on the pile.
Minimum instrumentation required is dynamic

measurement in the form of accelerometer and strain
gage. The dynamic measurement instrument can be
mounted externally on the pile surface or embedded
within the pile. In accordance to ASTM D4945 (2017),
Figure 9. Failure of kentledge system prior to loading. the instruments shall be placed 1.5 times the pile
diameter below the pile head to avoid irregular stress
3.5 Importance of Instrumentation concentrations. The instruments shall be installed in
The most common instrumentation installed in a pile pairs so that they are symmetrically opposite and
are strain gages. Strain gages provides the load equidistant from the pile centroid. This is done to
distribution along the pile shaft. Therefore, the unit remove effects of bending when averaging both
shaft resistance for each strain gage interval can be readings from the instruments. A schematic diagram of
evaluated. In addition, from the most bottom strain a typical instrumentation setup on a bored pile is shown
gage, toe resistance can be estimated. To obtain a more in Figure 10. A photograph of a PDA test on driven pile
accurate toe resistance, pressure cell can be installed at is also shown in Figure 11. Additional accelerometer
the base of pile. and strain gage can be installed anywhere along the pile
shaft, but also have to come in pairs, unless installed in
Instrumentation can also “save” a static load test which the centroid of the pile.
experienced failure from being useless. For example,
when load test needs to be stopped midway due to Prior to conducting PDA test, sufficient time need to be
unforeseen circumstances, it is still possible to obtain allowed for both driven pile and cast-in-place pile to
the shaft and toe resistance developed with movement, gain strength. Instrumentation has to be calibrated, and
such as shown in Figure 3 and 4. As little movement is pile properties, such as wave speed, density and
required to obtain the ultimate shaft resistance. dynamic modulus of elasticity have to be determined.
Pressure cell data can help to indicate the stiffness of For PDA test on driven pile, the test procedure is the
soil below the pile toe. With the help of same as conventional pile driving. During the test,
instrumentation, shaft and toe resistance can be back hammer weight and drop height have to be recorded.
calculated, and design can be verified. Rate of blows (blows per minute) also have to be
recorded in addition to the accelerometer and strain
For large projects, it is very beneficial to conduct static gage readings. As for PDA test on cast-in-place pile, it
load tests with instrumented pile early, prior to the is necessary to flatten the pile top, and reinforce it with
construction/installation of working piles. By back steel shell. Drop hammer with a weight of 1-2% of the
calculating the shaft and toe resistance, initial design desired ultimate test capacity is used to apply impact
can be optimized, and potentially number of piles can on cast-in-place pile, beginning with a low drop height,
be reduced if too conservative or increased if the initial then proceeding to greater drop heights. Indonesian
design capacity was too high. Standard conforms to the guidelines of ASTM D4945
for PDA testing.
4 DYNAMIC LOAD TEST
Pile driving analyzer (PDA) is a tool designed to 4.2 Interpretation and Results from PDA Test
control pile driving, but it can also be used to estimate Figure 12 shows the typical results obtained from PDA
pile capacity by using PDA instruments to conduct test. The force in the pile is interpreted from the strain
dynamic load test. Dynamic load test can be performed gage reading, and velocity from the accelerometer. The
6
results are interpreted using 1 dimensional wave theory 13. In this model, the pile is modelled as discrete
to obtain the ultimate pile capacity. The theory is based lumped mass. Each series of mass are resisted by a
on conservation of energy and momentum. Although spring (represented by Rs in Figure 13) and dashpot
the interpretation is based on the same theory, there are (represented by Rd in Figure 13). The spring represents
many formulations that can be used to interpret the data static soil resistance, while the dashpot represents the
(Samson, 1987). dynamic resistance. For the soil resistance alone, there
a few parameters required, such as ultimate static soil
resistance, quake and viscous damping factor. These
parameters are not easy to determine, especially in
layered soils, and mainly derived empirically. The pile
properties required include wave speed and dynamic
elastic modulus of pile. The hammer weight, drop
height, as well as the cushion’s modulus of elasticity
and coefficient of restitution are required. The cushion
properties change with number of blows, making
interpretation more difficult.
Figure 12. Photograph of a PDA test result (ASTM, 2017).
Figure 10. Schematic diagram of instrumentation for PDA

test (ASTM, 2017).
Figure 11. Photograph of a PDA test (Allnamics, 2018).

Figure 13. Smith model – wave equation analysis (Banerjee
An example of PDA interpretation is by the Smith and Butterfield, 1991).
wave equation analysis. The model is shown in Figure
7
Further complicating the matter, pile capacity changes

with time (set-up effect). Ideally PDA tests should be
conducted after the set-up effect has been completed.
However, it is known that setup effects can go on for at
least a month (Ng et al., 2013), in some soil it can even
goes beyond 100 days (Axelsson, 2000). The
dependencies on many of those parameters above and
the difficulties in determining the dynamic parameters
required make it important to calibrate PDA test results
against static load test, tested after the same period of
time. The Indonesian Standard also emphasize on the
requirement for PDA test to be calibrated against static
load test.
4.3 Failures in Execution of PDA Test

Failure in PDA test typically occur due to poor Figure 15. Example of bad PDA test results
preparation of pile head as shown in Figure 14. Due to
the poor quality of bored pile, and inadequate 5 BIDIRECTIONAL TEST
reinforcement, the pile head is damaged upon impact. Bidirectional test was first used in 1970s (Fellenius,
When such occurrence happens, energy transferred to 2017), and popularize by Dr. Osterberg in 1990s
the pile is erratic, and results cannot be used. This type (Osterberg, 1998). Hence, bidirectional is also known
of failures can be avoided by proper preparation of pile as Osterberg Cell test, or O-cell test. In static load test,
head and by providing good cushion material to avoid the load is applied head-down, while in bidirectional
direct impact on the pile top. Failures can also occur test, as the name suggests, load the pile in two
due to damaged instrumentation. This is especially directions. Figure 16 shows a schematic diagram of
common for instruments embedded in driven pile. For bidirectional test. The test involves placing one or more
externally attached instruments, they can be replaced. sacrificial O-cells at some point within the pile shaft.
The O-cells are considered sacrificial, as they cannot
Other failures are due to ‘bad’ results. Figure 15 shows be retrieved after the test completion. O-cell pressure is
an example of bad PDA test results. This can be seen generated by hydraulic system. When pressure is
from the disparity in trend of force and velocity signal. applied to the O-cell, the cell expands, hence pushing
These types of bad signals can be due to eccentric the upper part of test pile upward, and the lower part
hammering, or instruments which are too loosely or downward. Equipped with proper instrumentation, this
tightly attached. Great care must be under taken so as method, can measure the shaft and toe resistance of the
to obtain good PDA test signals. test pile. O-cell can be used for both cast-in-place and
precast piles.
Figure 14. (left) Bored pile under the impact of drop Figure 16. Schematic diagram of a bidirectional test (ASTM,
hammer; (right) damaged bored pile after impact. 2018).
8
5.1 Test Apparatus, Instrumentation and Installation

of O-cell
There is no guideline yet for bidirectional test in the
Indonesian Standard. For guideline, one can refer to
ASTM D8169 (2018). For O-cell test, the only
apparatus required are the O-cell itself and hydraulic
pump. Required instruments are pile head measuring
device, e.g. digital survey, or reference beam with dial
gauge, O-cell’s top and bottom plate movement
measuring device, can be in the form of electronic
displacement indicator, or telltale. Optional
instrumentation are strain gages along the pile shaft.
For cast-in-place pile, the reinforcement cage is Figure 18. Installation of O-cell in precast pile (YJack,
separated into 2 sections. The first section is welded 2018).
onto the top bearing plate of O-cell, while the second Figure 19 shows the typical results from an O-cell test.
section is welded onto the bottom bearing plate (see During the initial loading phase, the O-cell shows zero
Figure 17). The bearing plate needs to have sufficient movement as it has to overcome the effective weight of
spacing to allow grouting to flow through. the upper length of pile, as well as residual load. As the
For precast pile, the O-cell can be prefabricated with O-cell is further loaded, the shaft and toe resistance
the pile. Alternatively, the precast pile can be separated start to get mobilized until either the shaft or toe
into two sections, and each section can be attached to ultimate resistance is reached. It is also possible for the
the upper and lower bearing plate of O-cell (Figure 18). O-cell to reach its movement or load limit prior to
The installation of precast pile with O-cell attached is reaching either the ultimate shaft or toe resistance. Care
the same as normal precast piles. must be taken when designing the capacity and position
of O-cell.
5.2 Loading, Measurement of O-Cell and Results The upper and lower movement of the bearing plates
Loading of O-Cell is carried out by hydraulic pressure can be measured by a telltale. The total movement
by hydraulic pump from the ground surface (refer to between the top and bottom bearing plates can also be
Figure 16). The applied pressure expands the O-Cell, measured by displacement transducers. The
pushing the upper shaft upward, and the lower shaft and displacement of pile head can be measured by a
toe downward. Naturally, a fracture zone (in the case reference beam or by digital survey. The difference
of cast-in-place pile) is formed during loading of O- between the pile head movement and the top bearing
cell. The fracture zone separates the pile into an upper plate movement is the shortening of pile.
and lower length.
Figure 17. Reinforcement cage welded onto bearing plates of

O-cells (Foundation, 2018). Figure 19. Typical result of an O-cell test (Fellenius, 2017).
9
From the measurements, one can obtain load- 6.2 Using Pile Loaded to Failure as Working Pile
settlement curve for both the pile shaft and pile toe. As discussed in section 2, in most cases, piles will not
Therefore, the shaft and toe resistance can be evaluated reach ‘geotechnical’ failure during pile load test. In
separately. most soils, a pile’s toe resistance continues to increase
the further a pile is loaded. What is meant by pile
5.3 Failures in Execution of O-cell test loaded to ‘failure’ is when the tested pile is loaded
Failures in O-cell test can occur due to malfunctioning beyond a certain failure criterion, e.g. pile head settle
of the O-cell, e.g. the O-cell not expanding. Other more than 4% pile diameter or other definition.
failures can originate from poor placement of O-cell. Therefore, as long as the tested pile is not structurally
Ideally, the O-cell should be placed at a level that damage, a pile loaded to ‘failure’ does not mean it
allows full mobilization of upper length of pile, and as become unusable. In other words, a pile loaded to reach
much toe resistance as possible. Therefore, in most a failure criterion can still be used as working pile as
cases, the O-cell is placed either at the pile toe, or very long as there is no structural damage. Furthermore,
close to the pile toe. However, in cases where the pile when a pile experience unloading, the next time it is
sits on very soft soil, and depends mostly on the shaft loaded, it will show a stiffer response, until the
capacity, it may be necessary to place the O-cell near previous maximum load is exceeded (see figure 20).
the middle of the pile length. Poor placement of O-cell This behavior is very much like loading an over
can mobilize either the ultimate shaft or toe resistance consolidated soil.
too early, hence only obtaining information for 1
resistance instead of the intended 2.
6 DISCUSSION
This section is to answer most frequently asked
questions regarding the loading tests laid out in the
previous sections.
6.1 Fail Pile vs Failing to Get the Intended Results

From the above write up, it is clear that geotechnically
the definition of pile geotechnical ‘failure’ capacity or
ultimate load is a definition determined by the failure
criterion taken in the design stage, specifically by its
magnitude of settlement (movement) at a certain load.
And it does not really mean the pile becomes totally
unusable.
Figure 20. Load-settlement curve in cyclic pile load test
Failures in obtaining the ultimate capacity of tested pile (Trishna, 2018).
can come from various sources, such as
apparatus/instrument errors, or miscalculations in However, one must pay attention when punching
designing the test, e.g. inadequate hydraulic jack failure with reduction of pile capacity occurs (type 3
capacity. failure in Figure 2). When reusing such pile, the
ultimate pile capacity has to be used instead of peak
‘Failures’ can be limited by doing a proper site pile capacity. However, for other piles which weren’t
investigation, and proper design test. By doing so, the loaded, the peak pile capacity can be used as long as
design load can be close to the actual, and the sufficient factor of safety is given to ensure that the pile
kentledge, reaction piles or O’ cell will not be under will not be loaded beyond its peak capacity during its
designed. Apparatus/instruments errors can also be design life.
limited by doing calibration tests prior to installations.
6.3 Reliability of PDA Test
When failures occur midway through the test, some PDA test is a good indicator for capacity of a pile.
useful data may still be obtained with the help of However, as mentioned previously, there are many
instrumentations. It is recommended that all test piles parameters required to interpret a PDA test result. Non-
should be instrumented with strain gages to enable uniform and layered soils making even many more
engineers to optimize their design. parameters need to be estimated. Therefore, it is not
recommended to solely depend on PDA test, but also
static load test to calibrate the parameters of PDA test.
10
In the Indonesian Standards (SNI 8460:2017), a static 6.7 Similarity of Results from Static Load Test, PDA
load test can be replaced by 4 PDA tests. However, Test and Bidirectional Test
only 40% of the required static load tests can be The main objective of conducting pile load tests,
replaced by PDA tests. For example, if 10 static load regardless the methods are to obtain the ultimate
tests are required, 4 static load tests can be replaced by capacity of the tested pile. Despite the difference in the
16 PDA tests, giving a total of 16 PDA tests and 6 static measurement system of the three types of tests,
load tests. theoretically, they should produce the same ultimate
pile capacity. However, practically it is impossible to
6.4 PDA Test with 3-Dimensional Theory obtain the same ultimate pile capacity. This is due to
It is theoretically possible to interpret PDA test with 3- the nature of interpretations from the three types of
dimensional theory. However, it is impractical and tests.
unnecessary. Many more instruments would need to be
installed, and the calculations would become Out of the three tests, static load test (be it kentledge or
unnecessarily complicated. While it is true that the reaction pile system) and bidirectional test should give
wave travelled 3-dimensionally, this is only true for a the closest match in terms of results. This is because
short distance. Hence, the requirement of transducer bidirectional test is essentially a static load test as well.
being placed 1.5 pile diameter away from the pile top. Bidirectional test can provide information on both the
After 1.5 pile diameter down the pile length, the wave shaft and toe resistance, while conventional static load
practically travelled in 1-dimension. test can only produce the total pile capacity. Fellenius
(2017) considers bidirectional test to be superior as
6.5 Manipulation of PDA Test compared to conventional static load tests.
No tests are foolproof from manipulation. Likewise, For PDA tests, with enough calibrations against static
any data can be manipulated, and it depends on the load test, a good match may be found. One should also
ethic of the individual in charge. Example of PDA test note that PDA test were initially intended for driven
manipulation include striking one pile more than the pile. It is unlikely a uniform bored pile (i.e. same
required number of blows and using those extras as diameter throughout the length) can be constructed, and
results from other piles. Manipulation, intended or this affects the dynamic modulus of the pile. Whereas,
unintended, can also be done during interpretation of the dynamic modulus of a precast pile can be tested
test. Reporting the wrong results. prior to its installation. Therefore, obtaining a good
In order to minimize manipulation of any tests, the match between PDA test on bored pile with static load
contractor should be supervised by another body, e.g. test is more difficult. A more accurate pseudo-dynamic
the owner’s consultant, to prevent conflict of interest. test that can be conducted on bored pile is statnamic.
However, to the authors’ knowledge, statnamic has yet
6.6 Using Pile Tested by Bidirectional Test as to be implemented in Indonesia, thus not discussed in
Working Pile this paper.
Similar to that discussed in Section 6.2, a pile tested by
7 CONCLUSIONS
bidirectional test can be reused as working pile.
Although there is a fracture zone in the pile, the upper This paper is intended to answer some of the most
and lower reinforcement cage are connected by the O- frequently asked questions about pile testing. Three
cell’s bearing plates. In addition, the fracture zone is methods of pile test used in Indonesia are explained,
always grouted, making the pile acts as one body. and their sources of failures are also discussed. Ways
to avoid these failures are also given. Hopefully this
It is also very costly to not use a pile tested by paper helps to improve the practice adopted during pile
bidirectional test. Bidirectional tests are common for testing.
large piles, and offshore piles. This is because building
a kentledge system or reaction piles would be too time REFERENCES
consuming and costly. And testing large piles by PDA
Allnamics (2018). Pile Driving Analyzer (PDA).
test are almost impossible. Therefore, due to
http://allnamics.eu/products-2/allnamics-pda-pile-
economical reason, most piles tested by bidirectional
driving-analyser/ [Accessed 3 June 2018]
test are reused as working pile. Of course, this is also
because conducting bidirectional test does not ASTM D 1143/D 1143M – 07 (2007). Standard Test
significantly diminish the tested pile performance. Methods for Deep Foundation Under Static Axial
Tensile Load. DOI: 10.1520/D1143
11
ASTM D 4945 – 17 (2017). Standard Test Method for Mechanics and Foundation Engineering, ASCE, 89,
High-Strain Dynamic Testing of Deep Foundations. 241-242
ASTM D 8169/D 8169M – 18 (2018). Standard Test Kondner, R. L. (1963). Hyperbolic stress-strain
Methods for Deep Foundations Under Bi-Directional response. Cohesive soils. Journal for Soil Mechanics
Static Axial Compressive Load. and Foundation Engineering, ASCE, 89, 115-143.
Axelsson, G. (2000). Long-Term Set-Up of Piles in Khmer D&C Technical Consultant (2018). Static Load
Sand. PhD thesis, Department of Civil and Test. www.khmerdandc.com/products/141380-static-
Environmental Engineering, Royal Institute of load-test [Accessed 31 May 2018]
Technology, Stockholm.
Ng, K W., Rolling, M., and AbdelSalam, S. S. (2013).
Banerjee, P. K. and Butterfield, R. (1991). Advanced Pile Setup in Cohesive Soil. I: Experimental
Geotechnical Analyses: Development in Soil Investigation. Journal of Geotechnical and
Mechanics and Foundation Engineering. CRC Press. Geoenvironmental Engineering, ASCE. DOI:
10.1061/(ASCE)GT.1943-5606.0000751
BS EN 1997-1. (2004). Geotechnical design Part 1:
General rules. Eurocode 7. Osterberg, J. O. (1998). The Osterberg load test method
for drilled shaft and driven piles. The first ten years.
Chin, F. K. (1971). Discussion on pile test. Arkansas Deep Foundation Institute, Seventh International
River Project. Journal for Soil Mechanics and Conference and Exhibition on Piling and Deep
Foundation Engineering, 97, 930-932. Foundations, Vienna, Austria.
Das, B. M. (2014). Principles of Geotechnical Samson, C. H. (1987). Pile Driving Analysis –
Engineering (8th Edition). Cengage Learning, Historical Review, Course on Pile Driving Analysis
Stamford, United States of America. and Shaft Integrity Testing. Texas A & M University,
Texas.
DeBeer, E. E. (1968). Proefondervindelijke bijdrage tot
de studie van het grensdraag vermogen van zand onder SNI 8460:2017 (2017). Standar Nasional Indonesia -
funderingen op staal. Tijdshift der Openbar Verken van Persyaratan perancangan geoteknik. Badan
Belgie. (in Dutch) Standardisasi Nasional.
Decourt, L. (2008). Loading tests: interpretation and Structville (2018). Design of Pile Foundation Using
prediction of their results. Geo Institute Geo Congress, Pile Load Test (Eurocode 7).
ASCE, New Orleans. https://www.structville.com/2018/01/design-of-pile-
foundation-using-pile.html [Accessed 31 May 2018]
Fellenius, B. H. (2017). Basics of foundation design.
Electronic Edition. www.Fellenius.net [Accessed 31 Trishna, B. (2018). Estimating the Load Capacity of
May 2017] Pile. http://www.soilmanagementindia.com/pile-
foundations/load-capacity-of-piles/estimating-the-
Foundation Alliance (2018). Pile Tests – Osterberg load-capacity-of-pile-foundation-soil-
Cell. https://www.foundation- engineering/14285 [Accessed 3 June 2018]
alliance.com/gal_details.php?title=pile-
tests&stitle=compression-load-test&id=75 [Accessed YJack (2018). YJack pile test – Home Page.
5 June 2018] http://www.yjackpiletest.com/ [Accessed 3 June 2018]
Hansen, J. B. (1963). Discussion on hyperbolic stress-
strain response. Cohesive soils. Journal for Soil
12
Active Faults Identification for Dam Safety against Earthquakes
D. Djarwadi
PT Pamapersada Nusantara. Jakarta, INDONESIA
d.djarwadi@yahoo.com
ABSTRACT
Dams were the first structures to be systematically designed against earthquake starting in the 1930s, when the pseudostatic
method was introduced. The dam engineering, geological engineering, seismology engineering and other engineering that
supported and related with the analysis of the dam against earthquake have been largely developed. This progress brought that
the pseudostatic methods was outdated, and a finite element method using time histories was used widely to predict the
deformation of the dam body, appurtenant works to the dam and other structures. This change leads to the identification of the
earthquake sources became very important. Active faults were ones of the earthquake sources that shall be accounted in the
seismic design of the dams.
This paper discussed the identification of active faults on site using the ground penetration radar and electric resistivity that has
been largely used in identification an active faults. Paleoseismic trenching sometimes was also carried out in the case that we
doubt the level of the activity of the faults. Some of the site identification of the active faults using ground penetration radar,
electric resistivity and paleoseismic trenching were presented.
Keywords: Active fault, dam, ground penetration radar, electric resistivity, paleoseismic trenching.
1 INTRODUCTION uncontrolled spill water occurred, while on the

concrete dams in practical, the tensile stress on the
Earthquake was the shaking of the surface of the earth dam during earthquake shall not exceeding the tensile
resulting from the sudden release of energy in the strength of the concrete, in order not to create cracks
earth’s lithosphere that creates seismic waves. that can form a sliding plane in the dam body.
Earthquake can range in magnitude from those that
are so weak that they cannot be felt to those great ICOLD (2016) stated that earthquakes must be
enough to destroy the infrastructures in the surface of defined for analytical purposes of the dams, so that
the earth including dams. appropriate seismic evaluation parameters such as
magnitude, acceleration, spectral ordinates and
Earthquakes are multiple hazards to the dams; it can duration can be selected. The process can be
generate ground shaking that causing a vibration to accomplished by using either a deterministic
the dam body, appurtenant structure of the dams and procedure, or a probabilistic seismic hazard
foundations, it can deformed or shifted of the dam in evaluation. The earthquakes level for which the dam
case an active fault was found on the footprint of the should be defined and analyzed should be as follows;
dams, it can create a wave that may over the crest of a) Safety Evaluation Earthquake (SEE) is the
the dam in case an active fault was on the reservoir maximum of ground motion for which the dam
area, it can create a large landslides on the bank of the should be designed or analyzed. For dam whose
reservoir, it can also create a rock fall which may failure would present a catastrophic damage to the
damage the appurtenant structures of the dams. downstream, the SEE will normally characterized
by deterministically-evaluated MCE or
Even all the earthquakes hazards was been noticed,
probabilistic-evaluated earthquake ground motion
the priority of the dam engineers is to mitigate the
with a 10.000 years return period.
effect of the ground shaking to the dam such as
Deterministically-evaluated earthquakes may be
deformations, stresses, safety factor against sliding
more appropriate in locations with relatively
and stability of the dams. Accordingly, the earthquake
frequent earthquakes that occurred on well
hazard which addressed in the codes and regulations is
identified sources.
limiting the deformations and stresses in order to keep
b) Operating Base Earthquake (OBE) is represents
the dam in the safety conditions even shaking by
the level of ground motion at the dam site for
maximum credible earthquake specifically design for
which only minor damage is acceptable. The
the dams. In the modern seismic analysis of the
return period of 145 years was the practical level
embankment dams, the deformation of the crest dam
of the earthquake parameter.
was limited to a certain values in order not allow an
13
c) Reservoir Triggered Earthquake (RTE) is the earth topographic/landscape or bathymetric express-

maximum level of ground motion capable of ions, generally as structural lineaments, fault scraps,
being triggered at the dam site due to the filling stream offsets, morphological or geological offsets
and presence of the reservoir. The consideration and others morphotectonic evidences as well as may
of the RTE has been reported as generally linked serve as a (tectonic) boundary between two different
to dams higher as 100 m or to large reservoir, or rock formations (Yeats et al., 1997)
located in tectonically sensitive areas.
In geological perspective, a fault is considered active
From the three (3) earthquakes level described above if it shows evidence of movements or seismic
the most important was to design the dams that can activities in Holocene and historical-instrumentally
resist to the SEE earthquake level, and to make sure recorded period, or in the past 11,000 years
that the dam will not failure against the SEE (McClymont, 2001), or large faults which have moved
earthquake level. Damage to the dam against SEE in the latest Pleistocene time (between 11.000 and
earthquake level shall be able to be rectified or 35.000 years, and major fault which have moved
rehabilitated and no uncontrolled release water from repeatedly in Quaternary time (1.8 million years).
the reservoir allowed after the dam experiencing the Although for an extremely high-risk project, such as a
SEE earthquake level. In this case, it was Nuclear-power site, ‘a potentially’-active fault that
recommended to design the bottom outlet and shows any geomorphic-geologic evidences since
spillway gates as critical elements of the dam in (Late) Pleistocene may be taken into considerations
releasing the water from reservoir on the SEE and thus be subjected to more intensive studies.
earthquake level.
From seismological perspective, discrete displace-
In the seismic analysis to the dams, three (3) ments or slips on active fault movements produce
earthquake sources shall be accounted, there were; earthquake events. Thus, fault activities or movements
a) Plates subduction as earthquake source, are represented by earthquake occurrences or their
b) Shallow crustal and background earthquakes from seismicity, though not all movements are seismic,
active faults as earthquake source they are also ‘aseismic’ fault slips. Hence, by plotting
c) Benioff zone as earthquake source. locations of earthquake events helps us to identify
locations of faults. Small earthquake events are
Figure 1 shows the earthquake sources that shall be produced by small slips, and large events are
accounted in the seismic analysis of the dams. produced by larger slips along bigger fault rupture.
Productivity or activity of active faults may also be
indicated by frequencies and magnitudes of related
earthquakes for certain period of time. Bear in mind,
however, since historical data and instrumental
seismic record are often too short compare to major
seismic cycles, then sometimes ‘a quiescence period
of seismicity’ may means ‘a seismic gap’ not because
they are inactive.
A potential hazard of active faults can also be inferred

by a maximum magnitude of earthquake that has been
produced in the past, but it is often not the case since
data of instrumental earthquake records as well as a
historical written record of past earthquake events are
not long enough. Thus, another way to estimate the
maximum credible earthquake magnitude on a given
Figure 1. Earthquake sources that shall be accounted in the fault is by looking at a dimension or a fault length,
seismic analysis of the dam. and its maximum earthquake magnitude is determined
by using an empirical equation that relates a fault size
2 ACTIVE FAULTS to an earthquake magnitude.
A shallow crustal seismic source is basically active Geodetic data, particularly from Global Positioning
fault. A fault is a discontinuity or fracture on the earth Satellite (GPS), is very important sources to study
crust that shows relative movements of the two crustal deformations and active faults. From GPS data
separated blocks of rock assemblages. It can be we can study activities of given faults, and if active,
mapped and studied as it is often illuminated by the we can also estimate how fast the fault moves (=slip
14
rate) or how often large earthquakes can occur. previously from the high resolution DEM, such as
Hence, if we know when the last big earthquake SRTM-90m or SRTM-30m grids, and historical
occurred in certain fault and we know its slip rate and earthquakes occurred around the suspected fault lines
a typical magnitude, then we can estimate how much or zones. The following are the experience of the
strain has been accumulated since the last one, and using of GPR (Ground Penetrating Radar) and
predict the likely hood of the next earthquake may Paleoseismic Trenching methods which commonly
strike again. used to identify and detailed faults.
Determining activity of identified fault is, however, 3 APPLICATION OF GROUND PENETRATING

not always easy. For example, in all old-rocks RADAR FOR FAULT IDENTIFICATION
environment, sometime it is hard to determine
whether the identified fault line is active or not since Ground Penetration Radar (GPR) has been used
there is no young geological deposits involved. Figure successfully to identify the discontinuous in the sub-
2 shown faults on the Tertiary sandstone formation. surface such as geological faults (Chow et al, 2001.,
Khorsandi et al, 2011, Beaupetre et al, 2012, and
Malik et al, 2007, 2012), while a 3D Ground
Penetration Radar reported has been utilized to
visualized the active fault (McClymont et al, 2008).
Ground Penetration Radar transmits a high frequency

electromagnetic pulse (10-10.000 MHz) of energy
into the ground. This pulse radiating downward with
some of its energy reflected back to the surface by
sedimentary horizon. The presence of an obvious
discontinuity and the dragging of reflected pulse
enable the identification of faults. Reflected intensity
pulses are recorded by Monitoring System, and
different intensities shows anomaly in subsurface.
Figure 3 shows the principle of the Ground
Penetration Radar in capturing the anomalies (Annan,
Figure 2. Faults in the Tertiary Sandstone formations. 2009).
An active fault identifications based on morphology is

also limited by a competition between a rate of
movements of a fault and a rate of erosions and
depositions. If a fault slip-rate or a rate of movements
is very low, lower than rates of deposition and/or
erosion, than we may not be able to see surface-
landscape expressions of faults since they will be
erased by erosion or buried under geological deposits.
The fault zones may still be seen, even the slip rates
are low, if they bound two different rocks with
contrast hardness, and so differential erosion rates will
highlight the fault. An accuracy of an active fault
mapping is also limited by a scale of the used base
map. The higher resolution (the larger scale) of the
base map, the more accurate the mapping because we Figure 3. Principle of Ground Penetration Radar in
will be able to inspect smaller morphotectonic capturing the anomalies (Annan, 2009).
features.
Use of high frequency antennas in the range from 1 to
There are many techniques to identify the presence of 2 GHz produces high resolution that can detect small
the active fault even not visible on the ground surface. anomalies, however the depth of imaging is limited to
The Ground Penetration Radar, Electrical Resistivity approximately 1.00 meter, while use of low frequency
(Geo-Electric method) and Paleoseismic Trenching antennas in the range from 10 to 400 MHz can
can be selected to identify based on the site penetrate tens of meters soil condition, however the
conditions. All the methods can be performed based ability to locate small objects or anomalies is limited.
on the fault location that had already identified
15
In principle, one point shoot create one single

electromagnetic waveform that identify the
underground or subsurface layers that are
characterized by their response of reflecting waves as
shown in Figure 4.
Figure 6. Radargram of the location exactly same with

figure 5, but using 80 MHz frequency antenna.
It is clear that using 40 MHz frequency antenna as

Figure 4. Single Ground Penetrating Radar waveform. shown in Figure 5 give a 48 m depth penetration, but
less in the resolution, but using 80 MHz frequency
Depending on antenna type, Ground Penetrating antenna, the depth penetration was only 11,5 meter,
Radar systems are classified air-coupled and ground- but have more detail in the resolution. It also showed
coupled system. An air-coupled, antenna was different fault interpretation from different resolution
approximately 40 cm above the ground surface; allow of the radargram. In this case if the presence of the
faster progress in the rough terrain, while ground- active faults was very important to the structures
coupled system antenna was fully contact to the nearby such as dams, power house or nuclear power
ground. This type was suitable for road surface or flat plant, the examination of the active faults using MLF
ground conditions. frequency antennas was a mandatory.
In practical, the fault identification using GPR can be In Indonesia, the use of Ground Penetrating Radar to
carried out using multi level frequency (MLF) investigate the presence of active faults surroundings
antennas; i.e 15, 40 and 80 MHz, that can be set on a the dam has been successfully to reveal the active
certain frequency to get the best radargram profiles of faults and their strands. The use of Ground
moderate to deep subsurface structures from several Penetrating Radar in proposed Batang Toru Hydro
meters up to 50 meters. There is always a trade Electric Power Plant at the very high seismicity area
between penetration depth and resolution that has to near the Toru and Angkola segments of Sumatra Fault
be set according to targets and site conditions, so that Zone has been successfully revised the seismotectonic
the MLF antennas can solve this problem. Figure 5 model of the faults surroundings the dam. Also the use
show the radargram using 40 MHz frequency antenna, of Ground Penetrating Radar in proposed Tigadihaji
while Figure 6 shows the radargram on the exactly dam in South Sumatra which close to the Kumering
same location with Figure 5 but using 80 MHz segment of the Sumatra Fault Zone reveals the
frequency antenna. thickness of the Ranau Tuff as a product of mega
eruption of the Ranau volcano, and the fault strands
surroundings the dam. In the proposed Batang Toru
Hydro Electric Power Plant, 18 eighteen lines of
Ground Penetrating Radar was utilized in the radius of
100 km from the dam site. An air-coupled MLF
antenna was used to identify the active faults based on
the interpretation of the SRTM maps and earthquake
histories. Even we used MLF antenna, the 40MHz
mostly used in this survey, while if we need more
detail, the MLF antenna then switched to higher
frequency to obtain more clearly the spread of faults,
strands, and zone influenced by faults (fault zone).
Figure 7 shows the air-coupled MLF antennas of
Ground Penetrating Radar for fault identification on
Figure 5. Radargram using 40 MHz frequency antenna. rough terrain.
16
regarding the past earthquake recorded in the trench

walls. A detail stratigraphy and structure due to the
earthquake events shall be documented perfectly to in
order to reveal the earthquake recurrence, predicted
magnitude and slip mechanism.
Akyuz et al (2014) stated that paleoseismic trenching

is a direct observation method of the historical
earthquake provides both exact location and history of
the fault. Exposing faulted modern (or Holocene)
sediments in suitable environments gives invaluable
information about surface-rupturing earthquakes for a
few hundreds to thousands of years.
In Indonesia many paleoseismic trenching has been

Figure 7. Application of air-coupled MLF antennas of GPR done in order to investigate the activity of the fault,
for fault identification on rough terrain. such as Lembang fault in West Java, Kumering fault
as a segment of Sumatra Fault Zone in Lampung, and
4 PALEOSEISMIC TRENCHING Pauh fault in South Sumatra. Pauh fault was located
Paleoseismology is the art of identification of past just 1.85 km on the downstream of the proposed
earthquakes in terms of location and size using Tigadihaji dam, and have around 10 km length. This
geological and geomorphological evidence fault was considered too closed to the dam, and no
(McCalpin, 2009). A detailed paleoseismic study can historically earthquake recorded occurred on this
reveal the history of earthquake events, even dated faults and surroundings. In order to acknowledge the
back thousand years ago. Earthquake parameters from activity of the Pauh fault, predicted of the magnitude
the past such as magnitude, recurrence interval and if the fault ruptured, earthquake recurrence, and the
fault location can be identified (Masana et al; 2001, seismic hazard to the dam. Figure 8 shows the
Klinger et al; 2003, and Jing et al; 2015). A dating location of the Tigadihaji dam and surrounding faults.
from carbon taken from soils on the slip location may
reveal the approximate the age of earthquake events
and also the recurrence of the earthquakes. Also
paleoseismic can help to construct a database which
can be used in seismic hazard assessment.
One of the methods in paleoseismology was digging a

trench perpendicular to the fault line or zone.
Trenching is a multidisciplinary study, which
including geology, geophysics, geomorphology,
geoinformation and history. Before trenching was
executed, pre-field geological interpretation on the
digital maps and geographic study shall be done in
order to locate the approximately ruptured zone of
fault. The length of the trench will be depends on the Figure 8. The Tigadihaji dam location and the surroundings
complexity of the fault zones. The depth of the trench faults
depends on the stability of the walls and thickness of
the soil layers. The location of trench shall be Before the paleoseismic trenching was carried out, the
selected, if possible on the area with low water table location of the Pauh fault was probing using electric
to avoid any collapsing of the wall or flooding due to resistivity method. Scanning results indicated that the
the water infiltration. After excavation completed, the decomposed top soil have a thickness more than 5
trench wall shall be cleaned, and a grids were made in meter, interpreted from the resistivity less than 200
the walls by composing perfect horizontal and vertical Ohm.m. This soil layer probably young alluvial
lines using strong yarns. The grids form a 0.50 or 1.00 deposit and decomposed of Ranau tuff. Below the
m space for detail identification of the trench walls. decomposed soils, the Ranau tuff layer with high
resistivity found up to 80 meters except at the
Trench logs are the fundamental records in
suspected Pauh fault. Below the Ranau tuff, the
paleoseismic trenching. Almost all information
tertiary sedimentary rock was found indicated from
17
low resistivity which range from 2 to 100 Ohm.m. shown in Figure 11. This deformation occurred on the
The most important from the scan result was the Ranau tuff formation, which formed around 31.000
capture of the suspected Pauh fault. Figure 9 indicated years ago, according the carbon dating results. Based
the scanning result which indicated the Pauh fault. on this finding, the Pauh fault shall be accounted as an
active fault, and shall be included in the Seismic
Hazard Analysis for Tigadihaji dam.
Figure 9. Scanning result using electric resistivity

indicating the location of Pauh fault East wall of the paleoseismic trenching
Palaeoseismic trench was carried out on the

approximate the Pauh fault line. Trenching was made
over the line 1 of the geoelectric survey, cut the Pauh
fault perpendicularly. The dimension of the trench are;
West wall of the paleoseismic trenching
20 meter in length, 1.50 in width and 2.00 m in depth.
Logging on the both (east and west) walls were made
Figure 10. Logging of the paleoseismic trenching walls.
using different colours to identify the soil layers and
fault structures. Logging were made to identify
colour, texture, type or nomenclatures of soils.
Trenching in Pauh faults have five (5) different soil
layers and described as follows:
a) Layer 100 was the first soil layer, identified by

black colour was top soil consists of sandy silt
soils, dark brown in colour, many roots found.
b) Layer 200 was the second layer, gravelly sandy
silt with some lithic fragment up to 4 cm, brown
in colour.
c) Layer 300 was the third layer, coarse sand from
Ranau tuff, with some lithic fragment up to 2 cm,
reddish brown in colour.
d) Layer 400 was the fourth layer, coarse sand from Figure 11. Indication of the activity of Pauh Fault in the
Ranau tuff with some sand layers interbedded, Paleoseismic trenching wall.
yellowish brown.
e) Layer 500 was the bottom layer, coarse sand from 5 CONCLUSION
Ranau tuff with some sand layers interbedded, Active fault identification on site in order to
grey colour. acknowledge the seismic hazards to the dam using
Ground Penetrating Radar, Electric Resistiviy and
On the trench walls there is an indication of fracture Paleoseismic Trenching methods has been discussed.
and fault plane that shear the soil layers with the All of these site execution methods shall be done
parallel direction of the Pauh fault. Although the fault based on the guidance of the desk study on the DEM,
structure did not appear on the surface, indication of SRTM, LiDar and bathymetry maps. The conclusions
Pauh fault found at least on the soil layer 400 that can be drawn from this papers are as follows;
shear the soil layer, so that the existence of possible
tectonic fracture structures deforming this soil layer is a) A field study of the active faults shall be
classified as an active fault. Figure 10 shows the conducted especially on the area where the dam
logging on the trench walls. Field observastion results built on the active seismic zone with known and
on paleoseismic trench which made on the Pauh fault suspected active fault obtained from earthquake
indicated that the Pauh fault should be addressed as histories and the interpretation of the DEM,
active fault. Pauh fault indicated that a very small SRTM, LiDar or the bathymetry maps,
movement, reflected by deformation of the soil
formation in the walls of the paleoseismic trench as
18
b) A technique to identify the faults on site using Khorsandi, A., Abdali, M., Miyata, T., & Yuyugi, J,
Ground Penetrating Radar, Electric Resistivity 2011. Application of GPR method due to Active Fault
and Paleoseismic Trenching methods has been Determination in Urban Area. Case Study: North
successfully reveal the presence of the faults, Shahre Ray Fault, South of Tehran, Iran. International
Conference on Environment Science and Engineering.
c) Ground Penetrating Radar images may reveal the Vol.8, pp 148-151. IACSIT Press. Singapore.
fault lines or zones and strands which not visible
from the surface. Combining with earthquake Klinger, Y., Sieh, K., Altunel, E., Akoglue, A., Barka,
histories on that area, the activity of the faults and A., Dawson, T., Gonzales, T., Meltzner, A., &
earthquake recurrence can be judge. Rockwell, T, 2003. Paleoseismic Evidence of
Characteristic Slip on the Western Segment of the
d) The Ground Penetrating Radar results can be North Anatolian Fault, Turkey. Bulletin of the
made to build the fault model or seismotectonic Seismological Society of America, Vol.93, No.6, pp
model of the area which will be used to develop 2317-2332.
the seismic hazard analysis for the structures such
as dams in that area, Malik, J. N., Sahoo, A. K., & Shah, A. A., 2007,
Ground Penetrationg Radar Investigation along
e) The Paleoseismic Trenching method demonstrate Pinjore Garden Fault: Implication toward
to reveal the fault structures, slippages, Identification of Shallow Subsurface Deformation
earthquake recurrence, and can be used as data for along Active Fault, NW Himalaya, India. Current
seismic hazard assessment, Science, Vol.93, No.10, pp1422-1427.
REFERENCES Malik, J. N., Kumar, A., Satuluri, S., Puhan, B., &
Mohanty, A, 2012. Ground Penetrating Radar
Akyuz, H. S., Karabacak, V., & Zabci, C, 2014.
Investigation along Hajipur Fault: Himalaya Frontal
Paleoseismic Trenching. In Encyclopedia of
Thrust-Attempt to Identify Near Subsurface
Earthquake Engineering, Springer-Verlag, Berlin.
Displacement, NW Himalaya, India. International
Annan, A.P, 2009. Electromagnetic Principles of Journal of Geophysics, Vol. 2012, Article ID 608269,
Ground Penetrating Radar. In Ground Penetrating pp. 1-7.
Radar (GPR) Princiles: Theory and Application (Jol,
Masaanaa, E., Villamarin, J. A., & Santanach, P,
H. M Editor).Elsevier. 543p.
2001. Paleoseismic results from multiple trenching
Beaupetre, S., Garambois, S., Manighetti, I., analysis along the silent fault. The El Camp
Malavielle, J., Senechal, G., Chatton, M., Davies, T., (Tarragona northeastern Iberian Peninsula). Acta
Larroque, C., Rousset, D., Cotte, N., & Romano, C, Geologihispanica, Vol.36, no. 3-4, pp 329-354.
2012. Finding the burried record of past earthquakes
McCalpin, J.P, 2009. Paleoseismology. 2nd Edition.
with GPR-based paleoseismology : a case study on the
Academic, Amsterdam.
Hope fault, New Zealand. Geophysic Journal
International. Vol. 189, pp 73-100. McClymont, A. F., Green, A. G., Streich, R.,
Horstmeyer, H., Tronicke, J., Nobes, D. C., Pettinga,
Chow, J., Angelier, J., Hua, J.J., Lee, J.C., & Sun, R,
J., Campbell, J., & Langridge, R, 2008. Vizualization
2001. Paleoseismic event and active faulting: from
of active faults using geometric attreibutes of 3D GPR
ground penetrating radar and high-resolution seismic
data: An example from the Alpine Fault Zone, New
reflection profiles acreoss the Chihsang fault, eastern
Zealand. Geophysics, Vol.73, No.2, pp B11-B23.
Taiwan. Tectonophysics, Vol. 333, pp. 241-259,
Elsevier. Yeats, R. S., Sieh, K., & Allen, C. R, 1997. The
Geology of Earthquakes, New York, Oxford
International Commission on Large Dams (ICOLD),
University Press.
2016. Selecting Seismic Parameters for Large Dams.
Bulletin no.148. Paris.
Jing, L. Z., Shao, Y. X., Klinger, Y., Xie, K., Yuan,

D. Y., & Lei, Z. S, 2015. Variability in magnitude of
paleo earthquakes revealed by trenching and
historical records along the Haiyuan Fault, China,
Journal of Geophysical Research: Solid Earth, pp
8304-8333.
19
3D Assessment of Rainfall-induced Slope Movements and Risk Mitigation

Strategies
Wei He
Geoinventions Consulting Services, Brisbane, Queensland 4119, AUSTRALIA
Wade.he@geoinventions.com.au
Barry Kok
Barry.kok@geoinventions.com.au
Sangmin Lee
Sangmin.lee@geoinventions.com.au
ABSTRACT
Abstract: Severe financial lost is induced due to rainfall-induced movements or shallow landslides of slopes which embed
displacement-sensitive stuctures. The 2-step assessment&design procedure was proposed and practiced, based on strain-based
3D numerical method. On the basis of data collection and desk view, the regional rainfall intensity and duration is compared to
the empirical landslides threshold in the first step. The risk level can be defined in the region to determine if strain-based design
analysis is required. If it is in high-risk zone, the slope needs to be assessed in 3D by using fluid-solid fully-coupled Finite
Element Method (FEM) in step 2. Remediation plan can be proposed to meet the design criteria. The required information such
as empirical threshold of initiating shallow landslides, unsaturated soil parameters estimating method, and fluid-solid fully-
coupled method were provided in this paper. A problematic slope in Brisbane area was assessed by following this procedure. A
critical rainfall event (1 in 2000 AEP) of 150mm/day was applied on the slope surface. 1.114m superficial movement was
observed in the 3D model, which corresponds well with measured movement onsite. The removal of colluvium in the potential
moving area and drainage fingers were proposed to mitigate the risk and justified by further analysis.
Keywords: slope movement; rainfall; 3D modeling; finite element method (FEM); unsaturated soil
1 INTRODUCTION Extensive attentions have been paid on rainfall-induced

slope instability, due to numerous large-scale failure
Rainfall-induced slope movement caused severe cases reported in world-wide (Caine, 1980; Brand,
financial lost due to the detrimental influence on 1984; Au, 1998; Chen, 2005; Borja, 2010). The critical
embedded structures, such as pile foundation and combination of rainfall intensity and its duration is
pipelines (Fig.1). Comparing to other scenarios, a revealed as the essential factors which trigger shallow
globally stable slope with certain level of movement in landslides and debris flows (Zhan, 2014). The
shallow depth can even damage these displacement- established intensity-duration curves are practical for
sensitive structures. However, this is a significant forecasting the potential landslides, but these require
engineering problem which requires further study. well-recorded historical landslides database in different
regions with various soil properties. For a specific site,
Shallow engineering evaluations need to be on the basis of
Landslide
Pipeline
stability analysis. This can be carried out by
introducing unsaturated seepage analysis into
conventional equilibrium limit analysis (Fredlund,
2012; Muntohar, 2010; Sagitaningrum, 2017), or
strength reduction method (Cai, 1998; Cho, 2001).
However, these methods are insufficient for the
prediction of the actual shallow movements. Due to the
Colluvium
complexity of coupling effect, the most common
approach to analyze rainfall-induced movement is to
Fig. 1 A rainfall-induced slope slip with pipeline uncouple the fluid flow and slope-stability problem,
embedded in Queensland, Australia and treat them in a sequential fashion instead (Borja,
20
2010). Recently, Jamei (2015) performed numerical shallow landslide was observed on 12 December in
analysis which took into account the fluid-solid 2008, following an intensive rainfall event. Maximum
coupled effect, but it did not consider 3D effect and superficial velocities of the slope reached a peak of
may underestimate the problem. 100mm/day at the beginning of the monitoring (30
January 2010), following a four-day intense rain
Rainfall thresholds can be defined on physical (approximately 360mm). A theodolite study on slow
(process-based, conceptual) or empirical (historical, hill-slope movement at Gunung Pass site of Cameron
statistical) bases (Guzzetti, 2007). In this paper, the Highland district of Malaysia was conducted to
procedure of evaluating the slope movement was monitor slope movement during two wet seasons in
proposed based on 3D fluid-solid fully coupled finite 2008 (Lateh, 2010). The monitoring results indicated
element method. A real case in Brisbane region was that slope movement was accelerated by each heavy
studied in detail to indicate the capability. rainfall event. The maximum moving velocity was
approximately 140mm/day during the wet season in
2 SLOPE RESPONSE UNDER RAINFALL November 2008 with rainfall intensity up to 60mm/day.
EVENTS Except for site investigation, centrifugal tests can also
be adopted to study the rainfall-induced slope
2.1 Observed rainfall-induced slope movement in the movement (Qian, 2011). Qian prepared two slope
literature models with different initial water content. 52(70)-day
Few cases with measured slope movement due to rain of 30(44)mm/day in intensity was placed on a silty
rainfall events can be found in the literature, although clay model slope, for wet(dry) model. The slope
a great number of landslides after heavy rainfall have movement of 6mm/day in average was observed in the
been reported. Many studies revealed that the wet model, while that of 13mm/day in average was
landslides had close correlation with rainfall events observed in the dry model.
(Chen, 2004; Lateh, 2010). 69.4% of landslides in
The limited data indicated that the superficial
world wide were triggered by precipitation and
movement of slope may reach the velocity of over
infiltration as per the study in Europe (Pinto, 2011).
100mm/day, under rainfall of 60-90mm/day in
Table 1 Reported Rainfall-induced Slope Movement intensity. Note that the available data are based on
either convex hillslope or highway slope, which have
Rainfall Moving less catchment area than a concave hillslope. Therefore,
Location Intensity Velocity Reference larger moving velocity may be observed in concave
(mm/d) (mm/d) hillslopes, which are also preferable locations to embed
Wuhan, - 12 Chen, 2004 pipelines.
China
Calabria, 90 100 Capparelli, 2010 2.2 Empirical rainfall thresholds for shallow
Italy
landslides
Cameron 60 100-140 Lateh, 2010
Highland, Hundreds of rainfall events that resulted in landslides
Malaysia have been collected and analyzed by several
Beijing, 30 (wet) 6 (wet) Qian, 2011 researchers in various regions (Caine, 1998; Chen,
China 44 (dry) 13 (dry) 2005; Guzzetti, 2007). Empirical rainfall thresholds for
Except for the instability cases, three site monitoring the initiation of landslides were proposed in the form
and one centrifugal test result, which recorded both of intensity-duration curves. Soil properties and slope
rainfall intensity and movement development, have topography have not been considered in empirical
been tabulated in Table 1. A highway slope of 450m in thresholds. Thus, each curve needs to be adopted as a
extension and 50m in height was monitored in Wuhan, regional solution, or a lower bound threshold when the
China, from May 2002 to January 2003 (Chen, 2004). database is collected from world-wide.
The soil/rock encountered in the slope included 1.5-
2.0m yellow clay, followed by 30m thick highly According to literature, landslides basically fall into
weathered silty mudstone, followed by sandstone. The four types: shallow landslides, soil slip, debris flow and
drastic movement increase at the rate of approximately lahar. This paper focus on shallow landslides which
12mm/day was observed in July 2002, which occurred may induce unacceptable displacement to embedded
immediately after a heavy rainfall event. However, no structures. Table 2 shows the available empirical
rainfall intensity and duration were reported. thresholds which have taken shallow landslides into
Capparellli reported a large debris slide monitoring in account. The threshold proposed for debris flow only
Calabria, Italy, between September 2009 and August in Indonesia is also included. The thresholds were
2010 (Capparelli, 2010). Before the monitoring plan, a plotted in Figure 2 for better illustration.
21
also locates in this zone. The zone in between is

considered as the high-risk zone.
3 NUMERICAL MODELING STRATEGY

Rainfall weakens an earth slope mainly in three ways:
it increases the water content in the soil, thereby
reduces the apparent cohesion, ca; downhill frictional
drag on the slope, creating a destabilizing effect;
discharge as a surface runoff and erodes the slope
(Borja, 2010). Therefore, unsaturated soil behavior and
fluid-solid coupling effect are required to consider in
the modelling.
3.1 Unsaturated soil behavior and parameters

Matric suction, s=ua-uw, is the independent variable
Fig. 2 Intensity – duration curves proposed for which has significant influence on seepage and
shallow landslides deformation behavior of unsaturated soils. The key
properties include effective stress, yield surface, and
Table 2 Intensity – duration thresholds for the initiation permeability coefficient in unsaturated state.
of shallow landslide
Although the two-independent variable concept has
Area type
Equation
Range been proposed for a long time, Bishop’s effective stress
of D, hr is widely utilized for unsaturated soil modeling, as
World Sh, 𝐼 = 14.82 × 𝐷 −0.39 0.167~500
indicated by,
D
World Sh 𝐼 = 0.48 + 7.2 × 𝐷 −1 0.1~1000
𝐼 = 11.5 × 𝐷 −0.26 1~150
σ′ = (σ − 𝑢𝑎 ) + χ(𝑢𝑎 − 𝑢𝑤 ) (1)
Eastern Jamaica Sh
𝐼 = 4 × 𝐷 −0.45 0.1~150
North Shore Sh where, ' is the effective stress, ua and uw are pore air
Mountains,
Vancouver,
pressure and pore water pressure, respectively;  is the
Canada matric suction coefficient which varies from 0 to 1
Piedmont, NW Sh 𝐼 = 19 × 𝐷 −0.5 4~150 covering the range from dry to fully saturated
Italy conditions.  is often assumed to be equal to the degree
Valzangona, N A 𝐼 = 18.83 × 𝐷 −0.59 24~3360 of saturation, Sr.
Apennines, Italy
Taiwan A 𝐼 = 115.47 × 𝐷 −0.8 1~400 The yield surface of unsaturated soils can be extended
Pyrenees, Spain A 𝐼 = 17.96 × 𝐷 −0.59 > 168 from classical Mohr-Coulomb theory by introducing
Apuane Alps, Sh 𝐿𝑜𝑤𝑒𝑟 𝑡ℎ𝑟𝑒𝑠ℎ𝑜𝑙𝑑: 𝐼 0.1~35
= 26.871 × 𝐷 −0.638
the matric suction, (ua-uw), as indicated in (2).
Tuscany, Italy 𝑈𝑝𝑝𝑒𝑟 𝑡ℎ𝑟𝑒𝑠ℎ𝑜𝑙𝑑: 𝐼
= 85.584 × 𝐷 −0.781 τ𝑓 = c′ + (σ − 𝑢𝑎 ) tan 𝜑′ + (𝑢𝑎 − 𝑢𝑤 ) tan 𝜑𝑏 (2)
N of Lisbon, A 𝐼 = 84.3 × 𝐷 −0.57 0.1~2000
Portugal where, b is the angle indicating the rate of increase in
Indonesia D 𝐼 = 92.06 − 10.68 × 𝐷 2~4
shear strength with respect to a change in matric
Note: Landslides type - Sh, shallow landslide; D, debris flow;
suction, (ua-uw). The value of b appears to be
A, all types.
consistently equal or less than ', and ranges in 7~25.5°
As indicated in Table 2 and Figure 2, shallow according to available testing results (Fredlund, 2012).
landslides can be triggered by lower intensity or shorter The apparent cohesion can be expressed by,
duration during rainfall events, comparing to other
types of landslides. The lower bound of the thresholds c𝑎 = c + (𝑢𝑎 − 𝑢𝑤 ) tan 𝜑𝑏 (3)
is the combination of World-Sh and Vancouver-Sh. The formula (3) indicates that, the apparent cohesion
The zone under the lower bound is the low-risk zone
decreases with the matric suction.
for shallow landslides. The upper bound of the
thresholds is composed of three lines: Apuane Alps- Figure 3 shows the typical stress path of soils in
Sh(Upper), Taiwan-A, and Lisbon-A. The zone above shallow depth during infiltration under rainfall. As
the upper bound has extremely high risk for shallow water content increases, the Mohr circle reduces from
landslides. The threshold of debris flow in Indonesia the large dash circle to the small one. However, the soil
still reaches its critical state due to the reduction of
22
apparent cohesion. The plastic flow after yielding leads The measurement of SWCC data in the laboratory is
to large superficial slope movements. time-consuming and costly. Alternatively, SWCC can
be estimated by using physical properties such as
particle size distribution (PSD), liquid limit (LL),
plastic limit (PL), when the test result is unavailable
(Hernandez, 2011). The widely adopted SWCC model
proposed by Fredlund and Xing (1994) is,
1
𝑆𝑟 = 𝐶(Ψ) 𝑏𝑓 𝑐𝑓
(5)
Ψ
{ln[𝑒+( ) ]}
[ 𝑎𝑓 ]
Ψ
ln(1+ )
ℎ𝑟
𝐶(Ψ) = 1 − 1000000 (6)
ln(1+ )
ℎ𝑟
Fig. 3 Stress path and yield surface moving during
rainfall events (Jamei, 2015) The parameters, af, bf, cf, and hr. can be obtained by (7)-
(10),
In terms of hydraulic behavior, unsaturated soil has
lower permeability coefficient, since air occupies part 2.7
(0.69− )
𝑎𝑓 = 10 1+𝑒(4−0.14𝐺𝐼) (7)
of void in between soil particles. The permeability
coefficient is usually expressed as a relative value 0.78
( )
based on the permeability coefficient in saturated state 𝑏𝑓 = 10 1+𝑒(6.75−0.19𝐺𝐼) (8)
and the matric suction (or negative pore pressure
2
alternatively). The van Genuchten (1980) model is the 𝑐𝑓 = 0.03 + 0.62 × e(−0.82(log 𝑎𝑓−0.57) )
(9)
most widely accepted hydraulic model, which is
660
indicated in (4). ℎ𝑟 = 494 + (10)
1+𝑒 (4−0.19𝐺𝐼)
2
[1−|𝛼ℎ|𝑛−1 (1+|𝛼ℎ|𝑛 )−𝑚 ]
ℎ≤0 In (7)-(10), the group index (GI) can be calculated by,
𝑘𝑟 = { (1+|𝛼ℎ|𝑛 )𝑚/2 (4)
1 ℎ>0 𝐺𝐼 = (𝑃200 − 35)[0.2 + 0.005(𝐿𝐿 − 40)] + 0.01(𝑃200 −
15)(𝑃𝐼 − 10) (11)
where, h is the negative pore pressure head, h=-p/w; 
is the parameter related to the air entry pressure; m=1- where, P200 is the passing the No. 200 sieve; LL is the
1/n; Table 3 lists the fitted van Genuchten model liquid limit; PI is the plastic index.
parameters, n and  based on 72 samples collected
from literature (Ghanbarian, 2010). The weighted plastic index, wPI, is:
Table 3 Typical values of the parameters n and 

𝑃200 ×𝑃𝐼
𝑤𝑃𝐼 = (12)
100
Clay Content
(g/kg)
n  (m-1) 3.2 Fluid-solid coupling method
Material
Min Max Min Max Min Max Analysis that couples the seepage and stress can be
Sand 14 18 2.22 2.56 2.74 2.65 implemented in various ways. The simplest way is to
Loamy sand 23 108 1.33 2.56 4.41 2.35 obtain the pore water pressure distribution by
Sandy loam 70 178 1.12 2.38 4.90 1.27 conducting seepage analysis firstly, then incorporate it
Sandy clay loam 208 349 1.06 1.85 3.92 1.47 into the following stress analysis by introducing pore
Loam 122 260 1.23 1.96 4.90 1.76 water pressure. Such method is the so-called semi-
Silt loam 120 270 1.14 1.25 9.60 1.47 coupled analysis. However, since deformation due to
Silty clay loam 280 390 1.14 1.43 8.82 0.98 stress analysis does not influence the seepage
Clay loam 304 348 1.05 1.64 4.90 0.78 phenomenon inversely, this method may lead to
Sandy clay 352 421 1.10 1.49 4.90 1.76
unacceptable deviation from real situations.
Silty clay 420 460 1.09 1.10 6.37 5.39
Clay 452 452 1.51 1.51 0.88 0.88
Fluid-solid fully-coupled analysis is the two-way
It can be observed that, both the mechanical and coupled method between seepage and solid stress. This
hydraulic models of unsaturated soils are based on method does not follow assumption that steady state
matric suction, which can be estimated by soil water pore water pressure is maintained. Hence, it is suitable
characteristic curves (SWCCs). The SWCCs correlates for simulating the transient seepage phenomenon.
the matric suction, (ua-uw), and water content in soils.
23
Furthermore, comparing to the consolidation analysis, This indicated that the location was in the high-risk
it is possible to define the changes in seepage boundary zone of shallow landslides, thus strain-based design
conditions with time, boundary flow rate etc. These are analysis was required. In the following analysis, 5%
necessary in the modelling of rainfall-induced slope AEP rainfall intensity (150mm/day) with 3-day
movement. duration (1 in 2000 AEP) was adopted to cover the
potential risk.
4 ASSESSMENT PROCEDURE
The influence of rainfall-induced slope movements can
be assessed in two major steps: preliminary assessment;
and strain-based design analysis, based on data
collection. The design of remediation plan can be
incorporated in the assessment procedure. This
procedure is illustrated in Fig. 4.
Data Collection & Desk Review
• Historical rainfall record
• Topography
• Geological profile
• Soil properties (strength, permeability, particle size
distribution curve. . .)
• Design criteria
• Other information required
The critical rainfall No

event exceeds the
triggering threshold?
Fig. 5 Design rainfall depth chart in the location
Yes
Strain-based Design Analysis The geological profile was conservatively simplified to

(3D fully-coupled Analysis or other methods)
a 4-layer stratum: 4.0m colluvium, 4.0m XW basalt,
10.0m HW basalt, and bedrock. The ground surface of
No
Design criteria
exceeded?
the model was generated in MIDAS GTS NX using
borehole elevation records, as illustrated in Fig.6. The
Yes model is 260m in length and 150m in width, with the
Remediation Plan
pipeline incorporated.
Section 1
Reporting
Section 2
Colluvium
Fig. 4 Assessment procedure of rainfall-induced slope XW Basalt
movement Section 3 MW Basalt
Bedrock
Note that collected data should be sufficient to estimate
the unsaturated soil parameters for strain-based design
analysis. The minimum requirements of the parameter Pipeline
estimate include PSD, liquid limit, and plastic limit.
For concave slope, real topography needs to be Fig. 6 3D geometry of the assessed area
modelled in 3D to consider the influence of larger
catchment area. 5.2 Properties of the colluvium
Slope movement or shallow landslides were expected
5 CASE STUDY to occur within the colluvium of 4m in thickness, which
was considered as the key layer. The colluvium in this
5.1 Brief introduction area is typically comprised of very stiff to hard clay.
A concave slope in Brisbane area, Australia, embedded SPT N values in the colluvium varies between 23 and
with utility service pipelines was under concern due to 37. The liquid limit and plasticity limit are 43% and
historical shallow landslides and damage to the 35%, respectively. The typical PSD data of the
services. As per design rainfall depth chart by colluvium is indicated in Fig.7.
Queensland government (Fig. 5), 1-day rainfall depth
for 50% annual exceedance probability (AEP) is
approximately 70mm, which is 3mm/hr in intensity.
24
5.3 Calculated slope movement under rainfall

The contour of the slope movement was obtained from
the 3D fully-coupled numerical analysis, as shown in
Fig. 10. The results indicated a maximum 1.114m
superficial movement occurred due to the intense rain.
The calculated movement is in accordance with
measured in recent shallow moving events.
Fig. 7 Particle size distribution curve of the colluvium
Fig. 10 Slope superficial movement at the end of the

rainfall
Fig. 8 Estimated SWCC of the colluvium
Fig. 11 Development of slope movement at three

sections along the pipeline
The slope movements at three sections (see in Fig.6)

along the pipeline were extracted from the 3D models.
Development of the movements were plotted in Fig. 11.
Fig. 9 Estimated relative permeability coefficient, kr The maximum movement occurred in section 2 (central
section), and the minimum one can be found in section
The SWCC of the colluvium was estimated by using 1 (upper section). Saturated zone was formed after 1-
the PSD and Atterberg limits, as illustrated in Fig.8. day rainfall which accelerated the moving rate in
Taking the parameters, =0.88 and n=1.51, for clay in section 2. In addition, three movement trends indicated
Table 3, then the relative coefficient permeability was that the superficial movement tended to occur in short
estimated and plotted in Fig. 9. The saturated duration under intense rain. However, the intensity
permeability coefficient, ks, is 2.0 × 10-7cm/s for which triggers the movements has not been studied in
horizontal and 1.0×10-7cm/s for vertical. this project.
25
As indicated in Fig. 12, the flow path of water during movement reduction at three sections in two
the rainfall event clearly showed the rainfall-induced remediation scenarios. It can be observed that, the
surface seepage. Comparing to convex slopes and movements in section 2 was completely eliminated as
highway slopes, this concave slope has significantly the colluvium is removed, and displacement in section
larger catchment area. This type of topography has 1 and 3 were significantly reduced. The scenario 3 has
higher chance to trigger large slope movement or greater displacement reduction than scenario 2.
shallow landslides. 3D strain-based numerical model is
required to be adopted for assessment purpose.
(a)
Fig. 12 Flow path of water during the rainfall event
5.4 Risk remediation plan

Two remediation options were proposed in this project:
scenario 2 – removal of the colluvium in section 2; (b)
scenario 3 – install finger drains in section 1 and
replace colluvium in section 3 with free drainage
material based on scenario 2. The original geometry is
the scenario 1.
(c)
Fig. 14 Displacement reduction in each remediation
scenario
In addition, a pipeline element was incorporated in the

scenario 3 model. Pipe-soil interaction was considered
by using the pile elements provided in MIDAS GTS
NX. The results showed the maximum 2.4cm
displacement was induced by slope movement with
remediation. This is a tolerable pipeline displacement
as advised by pipeline structural engineers.
Fig. 13 Slope movement after remediation (Scenario 3)
Fig.13 showed the slope movement contour after

remediation of scenario 3. Figure 14 compared
26
three-dimensional finite element analysis. Computers

& Geotechnics, 23(4), 255-275.
Caine, N. (1980). The rainfall intensity - duration

control of shallow landslides and debris flows.
Geografiska Annaler, 62(1/2), 23-27.
Capparelli, G., Iaquinta, P., Iovine, G. G. R., &

Versace, P. (2012). Modelling the rainfall-induced
mobilization of a large slope movement in northern
calabria. Natural Hazards, 61(1), 247-256.
Fig. 15 Pipeline displacement after remediation
(Scenario 3) Chen, Q., Han J., Ai K. (2004) Monitoring and analysis
of slope deformation along a speedway. Chinese
Journal of Rock Mechanics and Engineering, 23(2):
6 CONCLUSIONS
299-302
The conclusions can be drawn as follows:
Chen, C. Y., Chen, T. C., Yu, F. C., Yu, W. H., &
⑴The 2-step procedure of assessing the influence of Chun-Chieh, T. (2005). Rainfall duration and debris-
rainfall-induced slope displacement & shallow flow initiated studies for real-time monitoring.
landslides was proposed and justified in industry. The Environmental Geology, 47(5), 715-724.
relevant assessing methods were also introduced in
detail. These provide a reference for the problem of Cho S.E., Lee S.R. (2001). Instability of unsaturated
slopes with displacement-sensitive structures soil slopes due to infiltration. Computers and
embedded. Geotechnics, 28(3), 185-208.
⑵The 3D effect of concave slopes is significant due to Fredlund, D. G., & Xing, A. (1994). Equations for the
the large catchment area, which is illustrated in the case soil-water characteristic curve. Canadian Geotechnical
study. The topography of slope needs to be considered Journal, 31(4), 521-532.
in the strain-based design analysis.
Fredlund, D. G. (2012). Unsaturated Soil Mechanics in
⑶Unsaturated soil properties are the key in the rainfall- Engineering Practice. John Wiley & Sons.
induced slope displacement analysis. Attention should
Genuchten, M. T. V. (1980). A closed-form equation
be paid to meet the minimum requirement of data for predicting the hydraulic conductivity of unsaturated
collection, which is inclusive of PSD, LL, and PL. soils. Soil Science Society of America Journal, 44(44),
892-898.
ACKNOWLEDGMENTS
We are grateful to our colleagues who helped us in the Ghanbarian-Alavijeh, B., Liaghat, A., Genuchten, M.
discussion and design stages. T. V. (2010). Estimation of the van genuchten soil
water retention properties from soil textural data.
Pedosphere, 20(4), 456-465.
REFERENCES
Au, S. W. C. (1998). Rain-induced slope instability in Guzzetti, F., Peruccacci, S., Rossi, M., & Stark, C. P.
Hong Kong. Engineering Geology, 51, 1–36. (2007). Rainfall thresholds for the initiation of
landslides in central and southern europe. Meteorology
Borja, R. I., & White, J. A. (2010). Continuum & Atmospheric Physics, 98(3-4), 239-267.
deformation and stability analyses of a steep hillside
slope under rainfall infiltration. Acta Geotechnica, 5, Hernandez, G. T., Zapata, C., Chair, S., Houston, M.,
1–14. & Witczak. (2011). Estimating the soil–water
characteristic curve using grain size analysis and
Brand E. W., Premchitt J., & Phillipson H. B. (1984). plasticity index. Thesis: Arizona State University
Relationship between rainfall and landslides in Hong
Kong. In Proceedings, 4th International Symposium on Jamei, M., Guiras, H., Ben Hamouda, K., Hatira, M., &
Landslides (pp. 377–384). Toronto. Olivella, S. (2008). A study of the slope stability in
unsaturated marly clay soil. Studia Geotechnica et
Cai, F., Ugai, K., Wakai, A., & Li, Q. (1998). Effects Mechanica, 95–106
of horizontal drains on slope stability under rainfall by
27
Jamei, M., Guiras, H., & Olivella, S. (2015). Analysis

of slope movement initiation induced by rainfall using
the elastoplastic barcelona basic model. European
Journal of Environmental and Civil Engineering,
19(9), 1033-1058.
Lateh, H., Muhiyuddin, W. M., Taib, B., & Khan, Y.

A. (2010). Monitoring of hill-slope movement due to
rainfall at gunung pass of cameron highland district of
peninsular malaysia. International Journal of Earth
Sciences & Engineering, 03(1), 974-5904.
Lim, T. T., Rahardjo, H., Chang, M. F., & Fredlund, D.

G. (1996). Effect of rainfall on matric suctions in a
residual soil slope. Canadian Geotechnical Journal,
33, 618–628.
Muntohar, A. S., & Liao, H. J. (2010). Rainfall

infiltration: infinite slope model for landslides
triggering by rainstorm. Natural Hazards, 54(3), 967-
984.
P. Orense, R., Shimoma, S., Maeda, K., & Towhata, I.

(2004). Instrumented model slope failure due to water
seepage. Journal of Natural Disaster Science, 26(1),
15-26.
Pinto, P. S. S. E., Barradas, J., & Sousa, A. (2011)

Lessons learned from case histories on landslides..
Geotechnical Engineering For Disaster Mitigation And
Rehabilitation And Highway Engineering 2011:
Geotechnical and Highway Engineering — Practical
Applications, Challenges and Opportunities: 54-73
Qian J.Y., Zhang G., Zhang J.M. (2011) Centriguge

model tests for deformation mechanicsm of soil slope
during rainfall. Rock and Soil Mechanics, 32(2): 398-
403
Sagitaningrum, F. H., & Bahsan, E. (2017). Parametric

study on the effect of rainfall pattern to slope stability.
MATEC Web of Conferences, 101, 05005. DOI:
10.1051/matecconf/201710105005
Van Sint Jan M, Talloni P (1993) Flujo de sedimentos

del 18 de Juniode 1991 en Antofagosta: La Serena,
Chile. Tercer Congreso Chileno de Ingenieria
Geotecnia 1:247–265
Zhan, L. T., Liu, X. C., Pei, T., & Chen, Y. M. (2014).

Centrifuge modelling of rainfall-induced slope failure
in silty soils and validation of intensity-duration curves.
Chinese Journal of Geotechnical Engineering, 36(10),
1784-1790.
28
Bridge Approach Embankments on Rigid Inclusions
M. Rizal
Rekakarya Geoteknik, Jakarta, INDONESIA,
rizal@rekakarya.com
K. Yee
Regional Synergy Consulting, Kuala Lumpur, MALAYSIA
kenny.yeeks@gmail.com
ABSTRACT
Rapid development in Indonesia calls for a new highway to be constructed in Central Java. Along the highway alignment,
bridges are to be constructed over rivers and existing local roads. Based on the soil conditions, performance requirements,
construction schedule and project budget, ground reinforcement using Controlled Modulus Columns (CMC) was adopted to
support the bridge approach embankments to minimise post construction settlements; and to improve bearing capacity and
slope stability. The CMC system consists of vertical cylindrical grout columns installed in a predetermined grid spacing using
displacement auger. Typically, the CMC column was terminate at stiff layer, which found at depth 15m to 24m. Due to the
huge loading and thick compressible cohesive soil, selection of CMC column spacing and length is importance to ensure CMC
capacity within design. To confirm design termination depth of CMC column a set of drilling instrument were fitted into the
drilling rig. In this paper, 2D numerical modelling were verified by the 3D model and the results are presented. Also presented
in this paper is a brief description of the installation method used in soft ground condition together with a description of the
quality control procedure and acceptance testing. After completion of the CMC works, approach embankments up to 11m were
constructed.
Keywords: ground reinforcement, rigid inclusions, embankment, numerical modeling.
1 GENERAL INFORMATION abutments have already been constructed before

commencement of any ground improvement work, the
1.1 Project Background choice of ground improvement technique is limited to
techniques that are environmental friendly i.e.
In the north-central Java, a new highway of 39 km is techniques having minimum vibration and minimum
being constructed linking Pemalang and Batang. This lateral soil movement during construction works to
new highway forms part of the Trans Jawa Highway. avoid potential damage to the completed bridge
The project site is a paddy field of flat terrain. Figure abutments and the foundations. The solution of
1 shows the project location and the ground Controlled Modulus Columns (CMC) was selected.
improvement areas. CMC columns were installed at nine different
locations adjacent to the bridge abutments (Figure 1).
It is necessary to ensure smooth transition between
Figure 2 shows the location of CMC. A typical CMC
flexible pavement and rigid bridge structure. With a
treatment area is 60m by 30m. The base width of the
thick compressible soft cohesive soil deposit,
embankment is about 60m and the treatment covers a
excessive post construction settlement is a major
distance of 30m from the bridge abutment. The
concern. Also there is potential instability during
embankment height varies from 6m to 11m.
embankment filling works if the bearing capacity is
exceeded. The excessive differential settlement at the
1.2 Ground Conditions
transition area between flexible pavement and rigid
concrete bridge abutment will cause abrupt bump For each treatment area, pre-treatment site
which will cause discomfort to road users and investigation was carried out to ascertain the ground
endanger lifes. Long term maintenance works is conditions. Two numbers of deep boreholes with
required which disrupt the smooth operation of the standard penetration tests (SPT) and one number of
highway and it is a costly affair. in-situ cone penetration test (CPT) were carried out.
Figure 3 shows the typical SPT N-values and cone
Structural solution using RC piles and concrete slab is resistance (qc) values.
expensive. Geotechnical solution using ground
improvement is a viable solution. Since the bridge
29
Generally, the ground condition can be described as Below, a layer of hard clay (average NSPT = 20) to
an upper layer of 6m thick soft alluvium (NSPT  5) of depth of 28m. Following layer is back to firm silt
grey colour low plasticity marine origin overlying layer again (average Nspt = 7) to maximum drilling
firm clay layer (average NSPT = 7) to depth of 18m. depth borehole of 40m.
Pekalongan City
To Pemalang
Project location
To Batang
Figure 1 Project location and ground improvement areas where CMC columns are installed
CMC diameter 0.42m, length 15m

to 24m, Spacing 1.9m to 2.2m
Pile length up to 50m
Figure 2 Typical cross-section of CMC treatment area.
30
sufficiently long consolidation time and time for stage

Nspt qc (MPa)
loading of the embankment and surcharge fill
0 10 20 30 40 50 60 0 4 8 12 16 20 24
0
construction. With time constraint, this solution is not
5 feasible even with close spacing of vertical drains and
4
12 with the addition of reinforcement geotextile at the
5 4 base of the embankment. The rapid loading placed on
8 the soft soils below will cause excessive lateral
9
10 6 movement of the underlying soft soil deposit during
7 at 0m to 18m placement of embankment and surcharge fill which
10 qc<1 MPa
5 may cause deflection of the installed piles supporting
15
8 the abutments.
23
19
Depth (m)
20 21
Based on all the above constraints, an environmental
16 friendly solution of ground reinforcement using
31 Controlled Modulus Columns (CMC) was considered
25 25
28 most suitable. The CMC columns are installed by a
31 non-vibratory soil displacement augering process. The
30 13
8 columns are cement-grouted columns and hence, have
5 no column bulging problem and they are having
5 higher load bearing capacity than any other granular
35 8
9 columns. The columns are 42cm in diameter with a
7 cement grout compressive strength of 20MPa. The
40
column spacing varies from 1.9m to 2.2m square grid
subject to the embankment height.
45
Figure 3. Typical SPT N-values and cone resistance 3 CONTROLLED MODULUS COLUMNS
(qc) values.
3.1 Concept of CMC system
1.3 Performance Specifications
The components in a CMC system consist of a load
The performance specifications are as follow: transfer platform (LTP) of 1.0m thick compacted sand
(1) The maximum allowable residual settlement shall or gravel to facilitate the transfer of fill load on to the
be less than 100mm after the 10 years; columns uniformly. Two layers of reinforced wire
(2) The factor of safety against slope failure shall be mesh are placed inside the LTP layer to provide
not less than 1.5. traction reinforcement. Cylindrical vertical grout
columns (or also known as inclusions) are installed
The traffic loading shall be 15 kPa. below the LTP using displacement auger.
2 CHOICE OF GROUND IMPROVEMENT The process of load sharing mechanism in CMC is

TECHNIQUES illustrated in Figure 4. Since the ratio of stiffness
between CMC and the soil is between 1:1,000 to
The choice of suitable ground improvement 1:10,000 it is necessary to consider the vertical
techniques is governed by environmental constraints, deformation separately for the CMC and the soil. The
a tight construction schedule and the soft ground deformation of a point inside the CMC at a given
conditions. Since the bridge abutments have already initial depth is different from an adjacent point at the
been constructed before commencement of any same depth in the soil. In other words, there exists a
ground improvement work, the choice of ground different field of deformation between the CMC and
improvement technique is limited to techniques that the surrounding soil as explained below:
exhibit minimum vibration and minimum lateral soil
movement during works to avoid potential damage to • Stage 1: Due to the transfer of imposed stress
the completed bridge abutments and the foundations. to the soil (soil) through the load distribution
Vibro stone columns or any casing driven granular layer (sand blanket), vertical deformation
columns will cause excessive ground vibration during (settlement) of the soil (soil) occurs due to
installation works. Also, due to the low shear strength consolidation.
of the underlying soft soils, there will be excessive
column bulging and possible column failure during • Stage 2: As a result of consolidation
loading. Vertical drains and surcharging requires settlement, stress is transferred from the
31
surrounding soil to the CMC. The

deformation at the same given depth (except
at neutral plane) in the soil (soil) is different
from the CMC (CMC) due to different
stiffness (ECMC > Esoil) and that soil > CMC,
negative skin friction is developed in the
CMC.
• Stage 3: At greater depth, the point

deformation CMC > soil resulting in a stress
transfer from the CMC back to the competent
soil. This induces positive skin friction and
Figure 5. Graphs of vertical displacement, shear
base resistance.
stresses and vertical stresses
• Stage 4: Overall, an equilibrium state of load
distribution is achieved where the tip
resistance, friction resistance and soil 3.2 CMC Design
resistance is equals to the total load.
Numerical analysis using Plaxis 2D was carried out to
Figure 5 shows the locations of the neutral plane estimate the deformation and slope stability. The 2D
where point deformation of CMC and soil is the same. analysis was checked against 3D model. The results
At this location, the CMC column carries the show minimum difference.
maximum stress.
Axisymmetric 2D model with long-term stiffness
material is used to determine maximum vertical
settlement and 2D plane strain model with short-term
stiffness is used to excess slope stability.
The analysis is carried out using three different

models that is, a) drained and undrained axisymmetric
b) single plane strain d) full plane strain.
Results of SPT and CPT tests were used for the soil
properties. They are compared with the laboratory test
results.
3.2.1 Long-term settlement

For long term settlement, axisymmetry model was
used.
CMC was modeled as a soil volume and inside the

CMC a “dummy” plate was assigned. In doing so, the
result of axial and shear forces on the CMC can be
accessed.
Figure 4. Design concept for CMC In the field, CMCs were installed in square grid
pattern while in the axisymmetry model, it is circular.
Hence, necessary correction is made based on area
ratio. The result of stress inside the CMC and stress at
CMC head were extracted and compared with plane
strain model. To be conservative strength increase due
to installation effect to the adjacent soil is not taken
into the design. Maximum stress inside the CMC
occurs at the neutral plane, and this value was used to
determine the compressive strength of the cement
grout. Typical example of settlement obtained from an
axisymmetric model is shown in Figure 6.
32
Fictious
CMC as plate CMC
CMC as
Soil
Volume
CMC model
as embedded
beam row
Neutral plane
Figure 7. Single unit plane stain with fictive extension at

CMC head
Stress Inside CMC

Figure 6. Vertical displacement result CMC axisymmetric - 2,000 4,000 6,000
model 4.0
2.0
3.2.2 Short term for stability 0.0
Step 1: Model Single plane strain -2.0
Elevation (mRL)
CMC was modeled as embedded beam row (EBR). -4.0

neutral plane
EBR is not fixed or attached to the soil which allows -6.0
the soil to flow through. EBR can carry axial force -8.0
and bending moment value unlike soil volume. The
-10.0
top and bottom of the CMC was set to be free to
move. The shaft resistance value was pre-defined -12.0
based on initial soil stress. -14.0
CMC
-16.0
𝜏𝑠 = 2𝜋𝑟𝑡𝑚𝑎𝑥 (𝑧) (1) Axisymmetry Model
Plain Strain Model
where s is shaft resistance, r is the column radius, tmax
is shear resistance Figure 8. Comparison stress to CMC in Axisymmetric and
Plain strain model
𝑡𝑚𝑎𝑥 = 𝑅𝑖𝑛𝑡𝑒𝑟 [𝑐 ′ +𝜎 ′ (𝑧) tan 𝜃′ ] (2)
In this model the position of the neutral plane was
Since the EBR is a single line it cannot capture load located at elevation -6.0mRL.
transfer to the column surface to account for arching.
To simulate this, a fictitious CMC head was modeled Step 2: Full plane strain model
by extending the top of the CMC (Figure 7). At the
top tskin was pre-defined based on value obtained from After calibrating the single plane strain model with the
previous axisymmetric model. In EBR CMC spacing, axisymmetric model, a full plain strain model can be
axial skin friction was modelled in an elastic-plastic made with actual embankment shape. In this model
behaviour. The stress inside the CMC was then steel wire mesh and CMC with steel bar were
compared with the axisymmetric model as shown in included. For the steel bar inside the CMC, a limit
Figure 8. plastic moment, Mp of 33 KN.m was assigned based
on steel bar size and the numbers of bars used. For
CMC without steel bar, a very low value of Mp was
used. Figure 9 belows a typical full plane strain
model.
33
15 KPa
Two layer steel

Fill
wire mesh on LTP
FH
𝑅𝑡;𝑇
Figure 11. Results of vertical deformation
CMC with steel

CMC without steel bar inside bar inside
Figure 9. Full plane strain model
Due to the lesser load below embankment slope CMC

length is shorter than the centre. Due to the higher
tensile force at the embankment edge, steel
reinforcement bar was placed inside CMC at a certain
depth.
The stress distribution below the embankment is

gradually reduced with depth, taking an example at
10m depth, the stress reduces 30% and at 25m depth,
stress is reduced by more than half (Figure 10). Due to Figure 12. Results of axial force inside CMC
this, CMC length at the edge is shorter and wider.
Figure 10. Stress distribution below embankment without

CMC
The results of a full plain strain analysis are shown in

Figure 11 to Figure 16 Figure 13. Results of bending moment inside CMC
The results show maximum vertical and horizontal

settlement of 8.5cm and 3.0cm respectively (Figure
11)
34
the requirement of the allowable wire mesh strength

greater than the lateral active force, (RtT > FH).
𝐾𝑎 𝛾𝐻 2
𝐹𝐻 = 2
(3)
where Fh is the lateral active force, ka is the coefficient

of active earth pressure of the embankment,  is the
unit weight of the embankment fill and H is the height
of the embankment at the crest.
And
𝜋𝑑𝑦2
𝑓𝑒
Figure 14. Axial force of the steel wire mesh 𝑅𝑡;𝑇 = 𝛽
. 𝑛. 4
(4)
where fe is the steel yield stress, β is the reduction

factor for the wire mesh soil interaction (= 1.25), n is
the number of steel bars per meter width of mesh, d is
the diameter of steel bar in the longitudinal direction.
3.2.4 CMC with steel reinforcement bar

CMC without steel reinforcement is good in
compression but not in tension. When subject to
embankment loading CMC will experience both axial
and bending moment at the same time. Thus, it is
Figure 15. Slope displacement pattern
necessary to check the capacity of the CMC column in
resisting lateral load. In cases where the lateral force
is large, tensile reinforcement is needed in the CMC
columns. Steel reinforcement is incorporated into the
CMC columns. Excessive lateral force in CMC
columns is normally found near the toe of a high
embankment slope. Calculation for the resistance
capabilities of CMC on lateral loads can be done
according to BS EN 1992-1-1.2004 12.
a) Calculate the axial design load and design

moment.
𝑁𝑒𝑑 = 𝑁𝑝𝑙𝑥 𝑠𝛾𝐺 (5)
𝑀𝑒𝑑 = 𝑀𝑝𝑙𝑥 𝑠𝛾𝐺 (6)

Figure 16. Factor of safety for slope stability
where Ned and Med is the axial and bending moment
3.2.3 Steel wire mesh acting on the CMC respectively. Nplx and Mplx is the
Figure 9 also shows the load transfer platform layer axial and bending moment obtained from Plaxis
reinforced with steel wire mesh. The wire mesh calculation results. S is the column spacing and G is
consists of transversal and longitudinal steel bars. The the factor of safety between 1.2 to 1.4.
longitudinal steel bars enable to “absorb” the lateral
b) Eccentricity e (Figure 17), with the following
forces caused by the active earth pressure of the
equation
embankment. The longitudinal bars also serve to limit
lateral soil displacements and thus any lateral 𝑁
𝑒 = 𝑀𝑒𝑑 (7)
displacement of the CMC. The transversal bars enable 𝑒𝑑
to mobilize friction at the soil-wire mesh interface.
Wire mesh also enhances the efficiency of load
transfer distribution to the CMC thus will minimize
the arching effect. The wire mesh design shall meet
35
If reinforcement is required, then the CMC

reinforcement need to be designed using Med.
4 CONSTRUCTION OF CMC
CMC columns are constructed by soil displacement
using a displacement auger. During auger penetration
drilling, the lower screw section which has a conical
screw-bit shape with variable auger flight pitches
(Figure 19) will cut and loosen the soil and transport
the soil to the displacement body section. The
Figure 17. Illustration of eccentrical load distance displacement body is a cylindrical shape with the
same diameter as the lower screw section which
c) Calculate resistance area (Aref) with following prevents soil from passing through and thus, pushing
equation (Figure 18) (or displacing) the soil towards the borehole wall. The
counter screw section above the displacement body
has opposite direction flight. Soil collapsed from the
above during drilling is brought downward to the
𝐴𝑟𝑒𝑓 = 𝑅 2 (2𝜃 − 𝑠𝑖𝑛2𝜃) (8) displacement body and pushed towards the borehole
𝑒
wall. With this technique, there is minimum spoilt at
𝜃 = arccos( ) the ground surface.
𝑅
where R is the radius of CMC
Counter
screw section
Displacement
body
Lower screw
section
End cap
Figure 19. Full displacement auger
When the drilling auger reaches the design depth,

grout is pumped through a flexible rubber hose and
Figure 18. Illustration of resistance area (Aref)
through the hollow steam attached to the displacement
auger. The grout pumping pressure is monitored and
d) Calculate axial resistance (Nrd) with the following auger-lifting speed is controlled by the CMC rig
equation operator.
𝑁𝑟𝑑 = 𝐴𝑟𝑒𝑓 𝑓𝑐𝑑 (9) Electronic and mechanical sensors are fitted to the
CMC drilling rig to ensure a good column installation.
where Aref is the area resistance and fcd is the grout Figure 20 shows the on-board computer monitoring
strength system fitted with various sensors. A monitor display
is installed inside the cabin to display real-time
e) Check if reinforcement is required: if Nrd>Ned,
monitoring installation parameters. These parameters
reinforcement is not required. If Nrd<Ned,
include:
reinforcement is required.
a) Depth of installation (m)
36
b) Installation and extraction time (sec, min) deeper or shallower than initial design. CMC auger
c) Penetration rate (m/hr) drilling is work by mean of penetration so it can give
d) Rotational torque (Bar) verification to the in situ test. The actual CMC depth
e) Injection grouting pressure (Bar) is determine by live display drilling record by using
f) Auger lifting speed (m/hr) penetration rate and torque.
g) Grout volume (m3)
h) Computed CMC column profile Before commencement of full production work, trial
installation of CMC columns is carried out to calibrate
the operation parameters. The optimum grouting
pressure and the ideal speed of auger retraction during
grout pumping are determined from the trial
installation. Fast auger retraction causes necking of
columns and slow auger retraction causes grout
blockage in the rubber hose. Adequate grouting
pressure recorded indicates the lateral resistance from
the surrounding soil and that the grout has filled the
entire augering and drilling volume.
To ensure flowability of the fresh grout during auger

retraction and grout pumping, high slump grout is
needed. To prevent blockage during auger retraction
each truck delivery is tested for slump value. In this
project slump value between 23± 2cm is adopted.
For the reinforced CMC columns, the reinforcement

steel bars were placed into the columns shortly after
completion of the grouting works.
Figure 20. Onboard computer monitoring system fitted to

the CMC rig
For a 12m length CMC column, it takes about 11

minutes that is, about 5 mins for drilling and about 6
mins for grouting. Shorter columns will take shorter
construction time. The average production is about
600 to 1,400 linear meters per 10-hour working day
per rig. The supply of cement grout and the condition
of the working platform greatly influence the
production rate.
Figure 21 shows a typical installation record. The

drilling rig was operated with torque 0 to 35 bar at
depth 3m and at constant maximum torque of 40bar to
15m. The penetration rate was constant at 400m/hr
from beginning till 13m and gradually drops to 7m/hr
at 15m depth. At 13m below, it was substantial
drilling resistance cause auger rotation decrease
substantially and reducing penetration rate.
This is consistence with the borehole result which

indicates at the first 3m is soft soil (NSPT  4), and
continue with firm soil do the depth 12m. At the depth
13m below, borehole indicate a layer of very dense
sand which make auger difficult to penetrate.
Figure 21. Typical CMC drilling record
The CMC length is pre-determine from Nspt, it is
likely during installation the termination of CMC is
37
5 PLATE LOADING TEST 6 CONCLUSION

Plate load tests (PLT) were carried to maximum load Presented numerical modelling gives the possibility to
of 75 ton or 110% of maximum stress induced in the correctly design 3D problem using 2D plane strain
CMC column. The tests were carried out after 28 model. The result of the back-calculation PLT using
days. A total of nine PLT tests was carried out; one FEM gives a gain in confidence to the predict CMC
for each area of treatment. The objectives of the bearing capacity. Even though no settlement
carrying out the tests were to obtain the load bearing instrument installed, observation shows no indication
capacity of the columns; compare the PLT results with of the differential settlement, soil heaving near the
those obtained from Plaxis analysis; and back slope toe and instability when embankment rises
calculation of design values to refine further the rapidly. This project has demonstrated the successful
numerical analysis. application of CMC column to treat thick layer
compressible soil and high fill embankment close to
A typical PLT results is shown in Figure 22. PLT was the newly constructed bridge abutment within given
carried out for a 42cm diameter column and column performance specification. The project also
length of 20m. The column was installed through a successfully completed within given time frame and
layer of soft to firm silt layer and terminated at hard budget.
clay with Nspt = 20. Since the test column was not
tested to failure, Chin Method was used to determine
the ultimate load capacity. An ultimate load capacity
of 88 tons was obtained. The PLT test was modelled 7 BIBLIOGRAPHY
using Plaxis with a soil stiffness modulus Ey equals to
ASIRI. (2011). Soil improvement with rigid
750 Nspt and soil-CMC interface factor of 0.8. The
inclusions. Paris: IREX.
numerical analysis indicated an ultimate load capacity
of 83 tons. The difference between the two methods is Chin, F. (1970). Estimation of the Ultimate Load of
deemed acceptable. Piles from Tests not Carried to Failure.
Southeast Asian Geotechnical Conference.
Load (tons) Chin, F. (1983). Bilateral plate bearing tests.

0 20 40 60 80 100 Proceedings of the International Symposium
0 on in situ, (pp. 29-33). Paris.
5
head displacement (mm)
Combarieu, O. (1988). Amélioration des sols par

10 inclusions rigides verticales application à
15 l'édification de remblais sur sols médiocres.
Revue Française de Géotechnique, (pp. 44,
20
57-79).
25
FEM Plomteux, C. a. (2000). Embankment construction on
30
PLT extremely soft soils using. Proceedings of the
35 16th Southeast Asian Geotechnical
Conference. Kuala Lumpur.
Figure 22. Load -Settlement result from PLT
Yee, K. (2012). Controlled Modulus Columns (CMC):
The bearing capacity is also compared alpha-cu A New Trend in Ground Improvement and
method and the result is summarised in Table 1.The Potential Applications to Indonesian Soils.
alpha -cu give very close to PLT value. ISSMGE Technical Committee TC 211
International Symposium on Ground
Table 1. Bearing capacity comparison Improvement. Brussels, Belgium: ISSMGE.
Plate
FEM alpha -
Zhao, R. F. (1982). Estimation par les paramètres
Load pressiométriques de l'enfoncement sous
Plaxis cu
test charge axiale de pieux forés dans des sols
Ultimate bearing capacity
87.7 83.3 81.0
fins. Bulletin Laison Laboratoire Central des
(ton) Ponts et Chaussees, (pp. 119, 17-24).
38
Design of Rapid Impact Compaction at the New Yogyakarta International

Airport
M. Sams
Geotechnical Engineer, Geoinventions Consulting Services Pty Ltd, Brisbane, AUSTRALIA
info@geoinventions.com.au
W. He
Changsha University of Science and Technology, Changsha, CHINA
J. Gamaliel
M. Mueller
Engineering Manager, Vantage Commerce Pte. Ltd, Singapore, SINGAPORE
m.mue@vantage-commerce.com
B. Kok
Technical Director, Geoinventions Consulting Services Pty Ltd, Brisbane, AUSTRALIA
ABSTRACT
Rapid Impact Compaction has been adopted as part of the ground improvement works to mitigate potential seismic liquefaction
at the New Yogyakarta International Airport. The surface soil material vulnerable to liquefaction is comprised of up to eight
metres of very loose to loose sand, typical to the coastal areas of the region. The design of the ground improvement has been
conducted using finite element modelling and the soil composite block approach. In the FEM, the dynamic loading of the
compaction is modelled with the existing soil, and the influence area of improvement can be predicted. With the composite block
approach, the geotechnical parameters of the block are required to correlate to the SPT and CPT requirements to mitigate
liquefaction. Based on this, a design spacing of S/D=1.17 was stipulated, with two phases and two cycles. A trial of this design
configuration has been conducted on site with CPT and SPT testing, to demonstrate the effectiveness of the design. This has
indicated substantial improvement from the initial state to the first compaction cycle, and reduced improvement from the first
cycle to the second cycle. In terms of SPT results, a 200-300% improvement at the surface and 20-43% improvement at 8m
depth was measured. An preliminary empirical design methodology has developed based on the test results. With future testing
and further development of the numerical modelling, this can form a robust and optimised design approach for rapid impact
compaction.
Keywords: rapid impact compaction, liquefaction, airport, soil composite block.
1 INTRODUCTION Indonesia is particularly seismically active, with

several substantial seismic events per year. For the
Ground improvement is often a significant part of large geotechnical designer, this has a number of
construction projects in regions with unfavourable implications. For the loose iron sands along the south
ground conditions. This is especially the case in many of the Java island, liquefaction is a project risk to be
parts of south-east Asia where the active seismic addressed by the designer.
conditions become more dominant in the geotechnical
engineering design.
39
As part of a wider effort to increase tourism for The methodologies to determine the degree of
Yogyakarta and the surrounding region, a new improvement are rare and often are only specific to
international airport is to be constructed approximately each project. In addition to this, they are often based on
40km south-west of Yogyakarta city. This New the composite block approach. This is due to the
Yogyakarta International Airport (NYIA) will have a difficulty of this three-dimensional, dynamic problem
3.6km runway and full terminal facilities that have a and the number of variables that impact the result.
footprint of up to 107,000m2. These include: initial ground conditions, fines
(<0.075mm) content, impact energy of the machinery,
The existing ground has been deemed a liquefaction spacing, depth of improvement, water table level.
risk by the designer. Therefore, ground improvement
was required to sufficiently mitigate the potential for The outcome for the designer is required to be the
liquefaction. The design criteria to achieve this geometry configuration (RIC diameter D, and spacing
improvement is summarised in Table 1. These S), and termination criteria for the machine. The design
comprise of Standard Penetration Tests (SPTs), Cone of the ground improvement for the NYIA required a
Penetration Tests (CPTs), and Relative Density (Dr). diameter of 1.5m, primary spacing of 3.5m, and overall
spacing after secondary phase of 1.75m. See Figure 2
Table 1. Acceptance criteria for ground improvement for an illustration of the spacing. The termination
Depth Dr% (Relative
SPT-N CPT - qc (MPa) criteria used were:
(m) Density)
2 16 7 >70 a) Maximum cumulative deflection of 1000mm
4 22 9 >70 b) Minimum deflection per blow of 10mm
6 26 13 >70
c) Maximum number of blows per column of 60
8 30 15 >70
The purpose of this paper is to document the design
Rapid Impact Compaction (RIC) was selected as the process that has been adopted, and the outcomes of a
ground improvement method for the terminal area. This RIC trial that has been undertaken on-site. In addition
was due to the sand ground conditions and required to this, the difficulties and limitations of the design
improvement depth in the ideal range of RIC. This process are discussed, with suggested future work to
machine uses a high frequency, low drop height develop a reliable design method.
hammer to achieve the compaction. A photograph of
the RIC machinery is shown in Figure 1. This RIC 2 DESIGN METHODOLOGY
attachment is comprised of a 9t hammer & 4t foot.
A typical diagram of the design problem is shown in
Figures 2 and 3. As discussed, the effectiveness of the
RIC is a function of: initial soil conditions, spacing
ratio (S/D), and the compaction energy (hammer
weight, drop height). In the design case, the initial soil
conditions can be assumed constant (i.e worst case
conditions), and the compaction energy can remain
constant (consistent machinery and operation).
Therefore, the objective of the design is determining
the required spacing ratio.
Figure 1. RIC attachment mounted on excavator
40
d) Step 1 - The soil composite block approach is used

to find the RIC spacing required to meet SPT-N
requirements.
e) Step 2 - The numerical modelling is used to
determine confirm that the influence area of the
proposed spacing is appropriate.
f) Step 3 - A trial of the proposed RIC configuration
with SPT and CPT is conducted on site to assess
the performance of the design configuration.
g) Step 4 - If the performance of the RIC is
substantially different than predicted, a back
analysis is to be conducted to find the errant input
parameter/s.
3 SOIL COMPOSITE BLOCK APPROACH

The soil composite block approach has been used in
Figure 2. Typical plan view of the RIC design problem
ground improvement for decades as a convenient and
intuitive way of accounting for the spacing and the
Design of the RIC ground improvement has been initial ground conditions. In particular, the design
undertaken using a combination of numerical methodology documented by Goughnour et al (1991)
modelling and conventional empirical method. The and Poulos (2002) has been followed.
numerical modelling is conducted using finite element
analysis with ABAQUS v6.12 (Dassault Systèmes This approach allows the designer to treat the improved
Simulia, 2014). The conventional empirical method ground layer as one homogenous block. Design
uses a composite soil block approach. parameters are calculated using a weighted average of
based on the proportion of each of the component
materials. In this approach, the ground improvement by
RIC is assumed to act as columns with diameter D.
Using this approach, the design parameters (unit weight

and friction angle) of the equivalent soil block must
correlate to the SPT-N criteria required for the
liquefaction mitigation. Therefore, the spacing of the
RIC columns can be adjusted until this requirement is
met. The input parameters for the existing ground and
RIC ‘column’ materials are as shown in Table 2.
Table 2. Acceptance criteria for ground improvement

Figure 3. Typical side view of the design problem Unit Young’s Effective
Material Weight Modulus Friction Angle
(kN/m3) (MPa) (deg)
Using these approaches the spacing of the RIC columns
Existing Ground 18 10 28
can be determined to achieve the required improvement
RIC ‘Columns’ 22 50 35
to the vulnerable layer of soil. Following the design
analysis, a trial has been conducted to verify the design The parameters of the equivalent block can be
configuration. This also allows the designer to collect calculated using Equation (1) and (2), where A can be
data for design/model calibration and future design the following: unit weight γ, effective cohesion c’, and
optimisation. effective friction angle tan(φ). k is the improved area
ratio. Equations (1) and (2) have been found to
The design process has generally followed the consistently underestimate Young’s Modulus E.
following process: Therefore, this can be estimated using Equations (3)
and (4), where kE is the modified area ratio, using the
pattern coefficient ς. The calculated composite block
parameters for a range of S/D are shown in Figure 4.
41
Acomposite = kARIC + (1 − k ) Asoil (1)  ' = 27.1 + 0.3N 60 − 0.00054N 60 2 (5)
D 2 This requirement has been shown in Figure 3 as the

k= (2) ‘Target’ line. This indicates that the overall spacing
4S 2 ratio is required to be less than 1.6.
 E
Ecomposite = E RIC  k E + soil

(
 1 − k E 2 ) (3)
4 NUMERICAL MODELLING OF RIC
 E RIC  The RIC has been modelled with finite element
2
analysis in the ABAQUS software. A simplified 2D
D plane strain approach is used, similar to that shown in
kE =    (4)
S Figure 3. Three RIC impact sites are modelled - two
primary which occur first and one secondary which
ς = 0.78 (square), ς = 0.91 (triangular) occurs last. The soil material is simulated using a strain
hardening model. The boundary conditions on the base
36 40
Friction Angle 
and sides are modelled with infinite elements.
35 Young's Modulus E
Target Friction Angle
The dynamic loading caused by the hammer drop is
Young's Modulus E (MPa)
30
34
Friction Angle  (deg)
modelled using a repeating surcharge with time

33 function. This function is developed based on the
20
32
expected hammer drop frequency of the machinery.
Using a 9t hammer and a 1.2m drop height, a surcharge
31
10 of 135kN is applied for 0.1 second, every second. The
30 surcharge is assumed to increase from zero to peak and
back to zero in 0.1s.
29 0
1.0 1.5 2.0 2.5 3.0 3.5
S/D A number of models have been assessed with the
overall S/D varying from 1.17, 1.5, and 2. The
Figure 4. Estimated Parameters of Composite Block surcharge application and initial soil conditions have
been kept constant. The contour charts of equivalent
The improved ground is required to achieve an SPT-N plastic strain (PEEQ) for each S/D case are shown in
value of at least 16, as per Table 1. This has been Figure 5. This represents the influence area of the
correlated to friction angle using Equation (5), from compaction, similar to the research by Gu and Lee
Peck, Hanson, & Thornburn (1974). Using Equation (2002). Note that the each model is 10m high. As
(1), an SPT-N of 16 correlates to a friction angle of expected, the maximum influence is at the base of the
approximately 32 degrees. The ground improvement surcharge and extends down in a bulb shape.
composite block is required to achieve this friction
angle.
PEEQ
(Avg: 75%)
+3.421e+00
+3.138e+00
+2.854e+00
+2.571e+00
+2.287e+00
+2.004e+00
+1.721e+00
+1.437e+00
+1.154e+00
Z X
42
PEEQ
(Avg: 75%)
+1.929e+00
+1.770e+00
+1.611e+00
+1.451e+00
+1.292e+00
+1.133e+00
Z X
PEEQ
(Avg: 75%)
+1.781e+00
+1.634e+00
+1.488e+00
+1.341e+00
+1.194e+00
+1.047e+00
Z X
Figure 5. Equivalent plastic strain (PEEQ) contour charts for S/D=1.17 (top), S/D=1.5 (middle), S/D=2 (bottom). Note that the
height of model is 10m
These results indicate that substantial improvement The results of the SPT testing have been summarised in
extends to depths of up to 6m. In the S/D=1.17 case, the Table 5. The SPT and CPT test results are shown in
improvement is almost constant to this 6m depth. In the Figures 7 and 8. Note that the test numbers (SPT3,
S/D=2 case, there is substantial gaps in the influence CPT2 etc) are shown in Figure 6. The project
zone between the columns at 3-6m depth. At S/D=1.5, acceptance criteria have been shown for this testing.
these gaps are relatively minor and may not have a This testing shows a substantial improvement from the
significantly adverse impact on the outcome. However, initial state to after the RIC.
the impact of this is still unclear. Overall, the results of
the numerical modelling are consistent with the
outcome of the composite block approach, in
recommending an S/D<1.6.
5 ON-SITE RIC TRIAL

Based on the outcomes of the numerical and empirical
assessment, a design configuration was proposed
comprising 1.5m diameter RIC columns with an overall
1.75m spacing in a triangular pattern. This was
achieved with a primary stage using 3.5m spacing in a
triangular pattern, and a secondary stage that placed an
RIC column in the centre of each 3.5m triangle.
As part of the design process, an on-site trial was

specified to verify the performance of the RIC
configuration. This trial was conducted using two
cycles of the primary and secondary stages. SPT testing Figure 6. Configuration for the RIC trial
was conducted prior to, and after each cycle. CPT
testing was conducted before testing and after cycle 2. In terms of the SPT scores, the degree of improvement
The adopted RIC configuration and testing is indicated after the first cycle is >250% at 2m depth, and reduces
in Figure 6. with depth. Further improvement occurs during cycle
43
2, but on a reduced scale. This is summarised in Table the SPT conducted after cycle 2 yielded a lower result
3. than after cycle 1. This can also be observed in Figure
7 - ‘Between Columns’. It is expected that this is due
Note that at 6m depth, the improvement from ‘initial to to the inherent variability and inaccuracy of the SPT
cycle 2’ is lower than ‘initial to cycle 1’, indicating that testing, and a statistically low number of tests.
Centre of Column (Primary) Centre of Column (Secondary) Between Columns

2 2 2
BH18A (Initial) BH18A (Initial) BH18A (Initial)
SPT0 (Initial) SPT0 (Initial) SPT0 (Initial)
SPT1 (Cycle 1) SPT2 (Cycle 1) SPT3 (Cycle 1)
SPT4 (Cycle 2) SPT5 (Cycle 2) SPT6 (Cycle 2)
Requirements Requirements Requirements
3 3 3
4 4 4
Depth (m)
Depth (m)
Depth (m)
5 5 5
6 6 6
7 7 7
8 8 8
0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70
SPT SPT SPT
Figure 7. SPT results at each location for all cycles
overall trend of improvement due to RIC. That is, there

CPT4 (Initial)
CPT1 (cycle 1)
CPT2 (cycle 2)
is dramatic improvement from the initial state to cycle
0 CPT3 (cycle 2)
Requirements 1, and a reduced improvement from cycle 1 to cycle 2.
The degree of improvement gradually reduced with
depth, but is still evident even at 8m depth.
2 Table 3. Average degree of improvement in terms of SPT

Depth (m) Initial to Cycle 1 Initial to Cycle 2
Depth (m)
2 289.7% 402.6%
4 89.7% 141%
4 6 220.5% 169.2%
8 42.9% 42.9%
The improvement observed in cone resistance (qc) from

6
the CPTs, as shown in Figure 8, is also substantial. As
the tests were limited to a 2.5t CPT, these all
encountered refusal before 2m depth.
0 5 10 15 20 25
CPT qc (MPa)
Figure 8. Results of the CPT testing
All SPT results, regardless of position, have been

graphed against depth in Figure 9. A logarithmic line
of best of fit is shown. This graph demonstrates the
44
0 6 BACK-CALCULATED DESIGN
Initial METHODOLOGY FOR SOUTH JAVA IRON
Cycle 1
Cycle 2 SANDS
Requirements
Following the on-site trial of the RIC, further analysis
2
of the testing results has been conducted. This has been
used to determine a preliminary method for the design
of RIC in South Java iron sands.
Depth (m)
4 The design of RIC will generally require achieving SPT

and CPT targets, such as those criteria in Table 1 for
the project discussed in this paper. Therefore, the
design methodology of RIC must be able to predict the
6
SPT results (SPTfinal) following the RIC treatment,
based on information that will typically be available at
Cycle 2
the design stage. This includes SPT testing at the initial
Initial Cycle 1
condition (SPTinitial), distance (from the centre of RIC
8
column) ratio (s/D), and depth ratio (z/D). This can be
0 20 40 60 summarised, as in Equation (5).
SPT
 s z
Figure 9. Overall trend of the SPT results SPT final = f  SPTinitial , ,  (5)
 D D
Table 4. Summary of the SPT Results for the RIC Trial

Depth SPT Cycle 0 (Initial) Cycle 1 Cycle 2
(m) Required BH18A SPT0 SPT1 SPT2 SPT3 SPT4 SPT5 SPT6
2 16 5 8 26 28 22 30 34 34
4 22 16 10 28 24 22 32 32 30
6 26 12 14 42 35 48 34 34 26
8 30 24 >60 >60 >60 >60 >60 >60 >60
70
60
In Figure 10, all SPTs are graphed against the

correspondingly positioned SPT in the preceding RIC 50
cycle. So, in the cycle 1 stage - the cycle 1 SPTs are

‘final’ and the SPTs at the initial state are ‘initial’. In 40
SPTfinal
the cycle 2 stage - cycle 2 SPTs are ‘final’ and the SPTs
from cycle 1 are ‘initial’. The solid line is 1V:1H, 30
which represents where the initial and final SPTs are
equal.
20
10
All SPT Data

Initial=Final Line
0
0 10 20 30 40 50 60 70
SPTinitial
Figure 10. Initial SPT against final SPT results
45
Results that fall below this line indicate that the SPT at z/D=2. If the initial SPT (SPTinitial) is 15, then the
result is lower than from the previous cycle, a condition estimated post-RIC SPT value (SPTfinal) is estimated to
which should not occur. Thus, it is logical that all points be 28. Note that the red shaded area indicates where
should be above the line; the further above the line the SPTinitial is less than SPTfinal, which should be an invalid
higher the degree of improvement. result.
The chart shows that the degree of improvement is high Contour Plot for SPTfinal
40
when the initial SPT is low, and reduces as the initial 30
40 50
SPT increases. The data points have been fitted with a

35
logarithmic line, as this is found to be the best fit, the
regression equation and fit coefficient are shown.
40 50
30
Based on this regression, the general equation for 30
SPTfinal can be expressed as in Equation (6), where the

α, k and c coefficients are dependant on s/D and z/D. 25
SPTinitial
SPT final =  k ln(SPTinitial ) + c 
30 40 50
(6) 20
z
k = −2.43 + 10.5 15 20
(7) 30 40 50
D
z 10
c = 14.47 − 15.4 (8) 20
D 30
40 50
10
5
Figure 11 is a reproduction of Figure 10, with the data 1 2 3 4 5
points separated by the depth of the test. Based on this, z/D
the k and c coefficients are found to be related to z/D
by Equations (9) and (10) respectively. Figure 12. Contour Chart to estimate Post-RIC SPT values
70 Note that the estimate provided using Figure 12 is only

applicable directly beneath the RIC columns. The post-
60 RIC SPT result can also be expected to reduce between
the columns. Thus, the spacing reduction factor α, as
50 shown in Figure 13, can be used to suitably reduce the
result from Figure 12 based on the distance ratio from
40
the centre of the RIC column s/D. Continuing the
previous example, the user can predict the spacing
SPTfinal
reduction factor at s/D=2. In this instance, α=0.9.

30
Therefore, the estimated SPTfinal using Figures 12 and
13 at z/D=2, s/D=2 is 25.2.
20
z/D=1.33
10 z/D=2.67
z/D=4
z/D=5.33
Initial=Final Line
0
0 10 20 30 40 50 60 70
SPTinitial
Figure 11. Initial SPT against final SPT results, separated

by depth
Using Equations (6), (7), and (8), the contour chart,

Figure 12, has been generated that allows the user to
estimate SPTfinal based on SPTinitial and the depth ratio
(z/D). For example, the post-RIC SPT can be estimated
46
Spacing Reduction Factor -  2D plane strain modelling will likely produce an

2.0
0.80 aggressive result, as the pressure is overestimated as a
0.85
strip load, and the impact energy only gets dissipated
1.8
0.85 laterally, and not three dimensionally as in the real
1.6 0.80
case. In addition to this, 2D axisymmetric models are
0.85
unable to simulate the spacing and staging of the RIC.
1.4 0.90
0.85 The results of real project testing are being used to
0.90
1.2 develop a back-calculated empirical design
s/D
0.90 methodology. Note that this approach has only taken

1.0 data from the trial at NYIA with the corresponding
0.90 0.95 machine and soil conditions. Thus, it is likely only
0.8 0.95
applicable with similar conditions.
0.95
0.6
RIC is concluded to be a very effective ground
0.95
improvement method when used in sandy material, as
0.4
1.00
at the NYIA and the South Java iron sands generally.
1.00
1 2 3 4 5 6
z/D
REFERENCES
Dassault Systèmes Simulia, 2014. Abaqus 6.12
Figure 13. Contour plot for spacing reduction factor α Analysis User’s Guide.
7 CONCLUSION Goughnour, R. R., Sung, J. T., & Ramsey, J. S. (1991).

Using the soil composite block approach, the spacing Slide correction by stone columns. In Deep Foundation
ratio is required to be less than 1.6 to achieve the Improvements: Design, Construction, and Testing.
material criteria. The results of the numerical ASTM International.
modelling indicated that the likely upper limit of the
Gu, Q. and Lee, F.H., 2002. Ground response to
spacing ratio was 1.5. Based on this, the design and the
dynamic compaction of dry sand. Geotechnique, 52(7),
RIC trial conducted on site used a spacing ratio of 1.17.
pp.481-493.
The results of the trial exceeded expectations with
Peck, R.B., Hanson, W.E. and Thornburn, T.H., 1974.
improvement of greater than 200% measured in the
Foundation engineering (Vol. 10). New York: Wiley.
first 4m, and up to a 42% to 8m depth, in terms of SPT
value. The CPT tests also measured a substantial Poulos, H.G. (2002), Research Report on Foundation
improvement, with cone resistance increasing by Design for Very Soft Clays, Coffey’s Technical
several multiples. Innovation Group.
The soil composite block is a simple design approach
that can take into account a number of the important
parameters and provides a good estimate of the
spacing. However, it is very basic approach that is
ineffective at analysing a layered soil profile, and
provides no indication of the depth of improvement. In
addition to this, a number of assumptions are required
by the designer including the analytical treatment of
RIC as perfect columns and the corresponding design
parameters of these columns.
Therefore, a numerical modelling approach is under

development to provide a more robust design solution,
which can take into account more of the required inputs
and provide an optimised solution. It is noted that RIC
is inevitably a three dimensional process. Therefore the
47
Effects of Model Scale Due to Displacement Factor for Nailed-slab Pavement

System
Anas Puri
Department of Civil Engineering, Universitas Islam Riau, Pekanbaru, INDONESIA
anaspuri@eng.uir.ac.id
ABSTRACT
The Nailed-slab pavement system can be analyzed by using equivalent modulus of sub grade reaction. This modulus consists
of modulus of subgrade reaction contributes by slab and additional modulus of subgrade reaction contributes by pile which
installed under the slab. The displacement on the surface of loading plate is always larger than the relative displacement
between pile and soil. Since the relative vertical displacement between the pile and surrounding soil is very small, ultimate pile
shaft resistance has not been fully mobilized yet. The mobilized unit pile shaft resistance can be reduced by applying
dispalcement factor. Some researcher conducted small scale test and others conducted full scale test. This research is aimed to
study the effect of different model scale to the dispalcement factor. Results show that the model scale affects the value of
dispalcement factor. Displacement factor for full scale model very small than small scale model. In other case, pile length can
also affect this factor.
Keywords: rigid pavement, soft clay, Nailed-slab System, modulus of subgrade reaction, displacement factor, model scale.
1 INTRODUCTION resistance (Qs) has not been fully mobilized yet. The
mobilized unit pile shaft resistance can be reduced by
Nailed-slab pavement system is a method to improve applying dispalcement factor ( = s /0).
the performance of rigid pevement on soft subgrade in
case this system constructed on the ground (Puri, Hardiyatmo (2011a) conducted small scale model test
2015a, 2015b, 2016, 2017a, 2017b; Puri, et.al 2011a, (1 : 5) on soft clay and proposed curve of inverse
2011b, 2012a, 2012b, 2013a, 2013b, 2013c, 2013d, displacement factor. Puri (2015) also proposed similar
2015, 2015; Hardiyatmo, 2008, 2009, 2011a, 2011b; curve based on small scale model test (1 : 5) on soft
Dewi, 2009; Nasibu, 2009; Taa, 2010; Somantri, clay. Hardiyatmo continued to propose the same curve
2013; Diana, et.al, 2016, 2017). The slab have double of inverse displacement factor based on full scale
functions; as a pile cap, and as a pavement slab at the model on stiff clay (Hardiyatmo, 2011b). Puri (2017a)
same time. This system can be also as a soil proposed a curve of displacement factor for soft clay
reinforcement if it is contructed under embankment based on full scale model.
(Waruwu, et.al 2017).
This research is aimed to study the effect of different
Hardiyatmo (2011a) proposed the method to analyze model scale to the dispalcement factor.
Nailed-slab pavement system by using equivalent
modulus of sub grade reaction. This modulus consists 2 DISPLACEMENT FACTOR
of modulus of subgrade reaction contributes by slab
and additional modulus of subgrade reaction The analytical approach in determining the equivalent
contributes by pile which installed under the slab. Pile modulus of subgrade reaction (k’) is given as follows
and slab should be connected monolithically (Puri, (Hardiyatmo, 2011a; Dewi, 2009; Puri et al., 2012a):
2015). k ' = k + k (1)
According to Hardiyatmo (2011a), the displacement Where k : modulus of subgrade reaction from plate
of the loading plate is different from the relative load test (kN/m3) and k : additional modulus of
displacement between soil and pile. When the pile subgrade reaction due to pile installation under slab
moves down due to loading, the soil under the plate (kN/m3). The modulus of subgrade reaction from a
also goes down. The displacement on the surface of plate load test (k) is usually taken by using a circular
loading plate (s) is always larger than the relative plate, and it should be corrected to the slab shape of
the nailed slab. The secant modulus is recommended.
displacement between pile and soil (0). Since the
relative vertical displacement between the pile and Hardiyatmo (2011a) proposed Eq. (2) in determining
surrounding soil is very small, ultimate pile shaft the additional modulus of subgrade reaction (k). The
49
relative displacement between the pile and soil is 12

L= 1,0 m
considered.
10 L= 1,5 m
L= 2,0 m
 0 As
Ratio of s/o
k = (a d cu + p0 K d tan  d ) (2)
8
 s2 s 2
6
Where δ0 : relative displacement between pile and soil 4
(m), δs : deflection of surface of slab (m), As : surface
area of pile shaft (m2), s : pile spacing (m), ad : 2
adhesion factor (non-dimensional), cu : undrained 0
cohesion (kN/m2), po’ : average effective overburden 0 0.001 0.002 0.003 0.004 0.005
pressure along pile (kN/m2), Kd : coefficient of lateral
Slab deflection, s (m)
earth pressure in pile surroundings (non-dimensional),
and d : soil internal friction angle (degree). Figure 2 Relationships of s/0 ratio vs. slab deflection
(Hardiyatmo, 2011b) based on full scale model test.
Hardiyatmo (2011a) published Figure 1 the inverse of
dispalcement factor (δ0/δs) based on small scale tests
(1 : 5). The model used was 4 cm pile diameter and 40 0.35
cm in length. Hardiyatmo (2011b) re-published the
Displacement factor, =0/s

relation between δ0/δs and slab deflection for a full- 0.30
scale model (Figure 2) while the pile and slab were 0.25
connected by a bolt. The pile diameter was 20 cm, and
0.20
the length of the pile varied between 1.0 m and 2.0 m.
Puri (2015) developed the curve of dispalcement 0.15
factor (δ0/δs) based on small scale tests (1 : 5). The
0.10
model used was 4 cm pile diameter and 40 cm in
length. dispalcement factor (δ0/δs) based on small 0.05
scale tests (1 : 5). The model used was 4 cm pile 0.00
diameter and 20 cm in length. 0.00 0.01 0.02 0.03
s/D Ratio
Puri (2017a) peoposed a curve of displacement factor
( = s /0) as shown in Figure 3, based on full scale Figure 3 Curve of displacement factor,  based on full
model on soft clay. The pile and slab was connected
scale model test (Puri, 2017a).
monolitichally.
0.35
0.30
Displacement factor,
Soft clay (Puri,

2017)
0.25
=0/s
Stiff caly (based

0.20 on Hardiyatmo)
0.15 Poly. (Soft clay

(Puri, 2017))
0.10
Poly. (Stiff caly
0.05 (based on
Hardiyatmo))
0.00
0.00 0.02 0.04
Ratio s/D
Figure 4 Displacement factor for soft and stiff clay
based on full scale model test (Puri, 2017b).
Figure 1 Relationships of s/0 ratio vs. slab deflection
based on small scale model test (Hardiyatmo, 2011a). Figure 4 shows the displacement factor for stiff clay
based on Hardiyatmo (2011b)—only for 1.50 m pile
length—and combined to Puri (2017a). It seen that
Based on Hardiyatmo work (2011b) and Dewi (2009), there is no differentiation between both soil
the curve of displacement factor is shown in Figure 4 consistency up to 0.01 in s/D ratio. Significant
(Puri, 2017b). defferentiation came up after 0.01 in s/D ratio. It is
can be also concluded for both cases that there is no
50
effect of soil consistency at least up to 0.01 in s/D (δ0/δs), and Hardiyatmo (2011b) for small scale and
ratio. Allthought there are some differents between full scale model test respectively. Those inverse of
both cases. Puri (2017a) used lean concrete, larger and dispalcement factor (δ0/δs) re-inverse to dispalcement
rectangular slab while Hardiyatmo (2011b) used factor ( = δs/δ0). Data also taken from Puri (2017a,
circular and smaller slab without lean concrete. And 2017b) for dispalcement factor ( = δs/δ0) for full
there was also differed on slab-pile connection type. scale model test. Those all of dispalcement factor
These factors can be neglected, because of the relative were compare eacth other and to find the effect of
displacement between pile and soil was response of model scale due to this factor. The structural
these conditions. dimension of model and soil data are presented in
Table 1. Hardiyatmo (2011a) and Puri (2015) are for
3 RESEARCH METHOD small scale model in soft clay. Hardiyatmo (2011b)
and Puri (2017) are for full scale model in stiff clay
Data was taken from work of Hardiyatmo (2011a) and
and soft clay respectively.
Puri (2015) in form of inverse dispalcement factor
Table 1 Nailed-slab dimension and soil data

Hardiyatmo Hardiyatmo
Description Unit Puri (2015) Puri (2017)
(2011a) (2011b)
Pile:
Diameter, D cm 4 4 20 20
Length, Lp cm 40 20 150 150
Slab:
Width, B cm 20 20 120 120
Thickness, h cm 4 4 15 15
Shape Rectangular Rectangular Circular Rectangular
Soil:
Undrained cohesion kPa 21 21 60 20
Soil Classification CH CH CH CH
4 RESULTS AND DISCUSSION 2.50

Dispalcement factor ( = δs/δ0) for small scale model R² = -5.356
R² = 0.9439 Scale 1 : 5
test is shown in Figure 5. The maximum  value tend 2.00 (Hardiyatmo,
2011a)
to be similar between shorter pile and longer pile, Scale 1 : 5
while the ratio of s/D for shorter pile tend to be in 1.50 (Puri, 2015)-fsi
=0/s
small range compare to longer pile. Poly. (Scale 1 :

1.00 5 (Hardiyatmo,
There are very significant effect of model scale as 2011a))
Poly. (Scale 1 :
shown in Figure 6. Displacement factor for full scale 0.50 5 (Puri, 2015)-
fsi)
model very small than small scale model. It can
influence the result of analysis. Displacement factor 0.00
0.00 0.05 0.10
for small scale model should be used for small scale
and the same thing for full scale model. According to Rasio s/D
Hardiyatmo (2011a) and Waruwu et.al (2017), by
using inverse of displacement factor for small scale in Figure 5 Displacement factor for soft clay based on
calculation of small scale Nailed-slab model was good small scale model test.
agreement with the observed deflection. According to
Puri (2017a), displacement factor for full scale model
in calculation of full scale Nailed-slab model was very
good agreement with the observed deflection.
51
Nasibu, R. (2009) Study on Modulus of Subgrade Reaction Due to

2.50 Effect of Pile Attached Under Plate (Loading Test on Full Scale).
Master’s Thesis, Graduate Program Gadjah Mada University,
Scale 1 : 1 (Puri,
2017)
Yogyakarta, Indonesia
2.00
Scale 1 : 1 Puri, A., 2015a, Perilaku Perkerasan Sistem Pelat Terpaku pada
(Hardiyatmo,
2011b) Lempung Lunak, Disertasi, Program Doktor Teknik Sipil Fakultas
1.50 Scale 1 : 5
(Hardiyatmo,
Teknik Universitas Gadjah Mada, Yogyakarta
=0/s
2011a)
1.00
Scale 1 : 5 (Puri, Puri, A. (2016) Behavior of Uplift Resistance of Single Pile Row
2015)-fsi
Nailed-slab Pavement System on Soft Clay Sub Grade, Proc. The
Poly. (Scale 1 : 1
(Puri, 2017))
3rd Asia Future Conference (AFC), Kitakyushu, Japan, 29 Sept-3
0.50 Oct. 2016.
Poly. (Scale 1 : 1
R² = 0.588
(Hardiyatmo,
2011b)) Puri, A. (2017a) Developing the Curve of Displacement Factor for
0.00 Poly. (Scale 1 : 5 Determination The Additional Modulus of Sub Grade Reaction on
0.00 0.05 0.10 (Hardiyatmo,
Nailed-slab Pavement System, International Journal of
Rasio s/D
2011a))
Technology, Vol. 1, pp. 1117-1126. ISSN 2086-9614.
Puri, A. (2017b) Differentiation of Displacement Factor for Stiff
Figure 6 Displacement factor based on full scale and Soft Clay in Additional Modulus of Subgrade Reaction of
model test. Nailed-slab Pavement System, Proc. The 2nd International
Conference on The Future of ASEAN (icofa), Vol. 2, pp. 927-933.
Springer. https://link.springer.com/chapter/10.1007/978-981--10-
5 CONCLUSIONS 8471--3_92.
Displacement factor was discussed by considering the Puri, A., Hardiyatmo, C. H., Suhendro, B., dan Rifa’i, A. (2011a)
model scale. It can be concluded that the model scale Experimental Study on Deflection of Slab which Reinforced by
Short Friction Piles in Soft Clay. In: Proceedings of the 14th
affects the value of dispalcement factor. Displacement Annual Scientific Meeting (PIT) HATTI, Yogyakarta, 10-11
factor for full scale model very small than small scale February, Indonesia, pp. 317-321
model. In other case, pile length can also affect this
Puri, A., Hardiyatmo, H.C., Suhendro, B., dan Rifa’i, A. (2011b)
factor. It is recommended using each displacement Contribution of Wall Barrier to Reduce the Deflection of Nailed-
factor approriate with dimension scale of Nailed-slab. slab System in Soft Clay . In: Proceedings of the 9th Indonesian
Geotech. Conf. and 15th Annual Scientific Meeting (KOGEI IX &
PIT XV) HATTI, Jakarta, 7-8 December, Indonesia, pp. 299-306
ACKNOWLEDGMENTS
Puri, A., Hardiyatmo, H.C., Suhendro, B., dan Rifa’i, A., 2012a.
Author to thanks for supporting by Universitas Islam Determining Additional Modulus of Subgrade Reaction Based on
Riau, Pekanbaru. Tolerable Settlement for the Nailed-slab System Resting on Soft
Clay. IJCEE-IJENS, Volume 12(3), pp. 32-40
REFERENCES Puri, A., Hardiyatmo, H.C., Suhendro, B., dan Rifa’i, A. (2012b)
Dewi, D.A. (2009) Study on Effect of Single Pile Due to the Application of the Additional Modulus of Subgrade Reaction to
Value of Equivalent Modulus of Subgrade Reaction from Full- Predict the Deflection of Nailed-slab System Resting on Soft Clay
scale Loading Tests. Master’s Thesis, Graduate Program Gadjah Due to Repetitive Loadings. In: Proceedings of Pertemuan Ilmiah
Mada University, Yogyakarta, Indonesia Tahunan ke-16 (PIT) HATTI, Jakarta, 4 December, Indonesia, pp.
217-222
Diana, W., Hardiyatmo, H.C. dan Suhendro, B., 2016, Small-scale
Experimental Investigation on the Behaviour of Nailed Slab Puri, A., Hardiyatmo, H.C., Suhendro, B., dan Rifa’i, A. (2013a)
System in Expansive Soil, In AIP Conference Proceedings 1755, Pile Spacing and Length Effects Due to the Additional Modulus of
060002 (2016). https://doi.org/10.1063/1.4958493. Akses tanggal Subgrade Reaction of the Nailed-slab System on the Soft Clay. In:
24 Februari 2018. Proceedings of the 13th International Symposium on Quality in
Research (QiR), Yogyakarta, 25-28 June, Indonesia, pp. 1032-
Diana, W., Hardiyatmo, H.C. dan Suhendro, B., 2017, Effect of 1310
Pile Connections on The Performance of The Nailed Slab System
on The Expansive Soil, International Journal of GEOMATE, Puri, A., Hardiyatmo, H.C., Suhendro, B., dan Rifa’i, A. (2013b)
April, 2017, Vol. 12, Issue 32, pp. 134-141. Deflection Analysis of Nailed-slab System which Reinforced by
Vertical Wall Barrier Under Repetitive Loadings. In: Proceedings
Hardiyatmo, H.C. (2008) Nailed-slab System for Reinforced of the 6th Civil Engineering Conference in Asian Region
Concrete Slab on Rigid Pavement. In: Proceedings of the National (CECAR6), Jakarta, 20-22 August, Indonesia, pp. TS6-10-TS6-11
Seminar on Appropriate Technology for Handling Infrastructures,
MPSP JTSL FT UGM, Yogyakarta, 12th April, Indonesia, pp. Puri, A., Hardiyatmo, H.C., Suhendro, B., dan Rifa’i, A. (2013c)
M1-M7 Application of the Method of Nailed-slab Deflection Analysis on
Full Scale Model and Comparison to Loading Tests. In:
Hardiyatmo, H.C. (2011a) Method to Analyze the Deflection of Proceedings of Konferensi Nasional Teknik Sipil ke-7
the Nailed Slab System, IJCEE-IJENS, Volume 11(4), pp. 22-28 (KoNTekS7), Universitas Negeri Sebelas Maret, Surakarta, 24-26
Hardiyatmo, C.H. (2011b) Designing of Pavement Roads and Soil October, Indonesia, pp. G201-G211
Investigation: Flexible Pavement, Rigid Pavement, Modified Puri, A., Hardiyatmo, H.C., Suhendro, B., dan Rifa’i, A. (2013d)
Chicken Foot Foundations, Nailed-Slab System. Gadjah Mada Behavior of Fullscale Nailed-slab System with Variation on Load
University Press, Yogyakarta, Indonesia. Positions. In: Proceedings of the 1st International Conference on
52
Development Infrastructure (ICID), UMS, Solo, 1-3 November,

Indonesia, pp. 26-36
Puri, A., Hardiyatmo, H.C., Suhendro, B., dan Rifa’i, A. (2015)
Pull out Test of Single Pile Row Nailed-slab System on Soft Clay,
Proc. The 14th International Conference on Quality in Research
(QiR), Universitas Indonesia, Lombok, 10-13 August, pp. 63-68
Somantri, A.K. (2013) Kajian Lendutan Pelat Terpaku pada Tanah
Pasir dengan Menggunakan Metode Beam on Elastic Foundation
(BoEF) dan Metode Elemen Hingga. Master’s Thesis, Graduate
Program Gadjah Mada University
Taa, P.D.S. (2010) Effects of Installation of Group Pile Due to
Slab Uplift of Nailed-slab Resting on Expansive Subgrade.
Master’s Thesis, Graduate Program Gadjah Mada University,
Yogyakarta, Indonesia.
Waruwu, A., Hardiyatmo, H.C., dan Rifa’i, A., 2017, Behaviour
of Nailed-slab Syatem on Peat Soil Under Loading, In The 1st
Warmadewa International Conference on Architecture and Civil
Engineering, Bali, 23th October 2017, pp. 91-97.
53
Initial recommendation criteria for distinguishing between landslides and

mudflows based on several case studies in Java and Bali
B. Widjaja
Department of Civil Engineering, Parahyangan Catholic University, Bandung, INDONESIA
widjaja@unpar.ac.id
D. Pascayulinda
Department of Civil Engineering, Parahyangan Catholic University, Bandung, INDONESIA
devinaaap@gmail.com
ABSTRACT
Mass movement is a type of disaster that commonly causes fatalities in Java and Bali. In this study, two types of mass movement,
namely, landslide and mudflow, are introduced. This paper presents an assessment of 10 cases of mass movement in Java and
Bali. The objectives of this research are as follows: (1) to create a database for mass movement from 2013 to 2018; (2) to propose
initial criteria for landslides and mudflows based on soil type, solid concentration by volume, and width-to-length ratio; and (3)
to determine the geomorphology of slopes. To achieve these objectives, data were collected from previous mass movement
studies in Indonesia. Silty-type mass movements tend to be mudflows rather than landslides. The recommended ranges for solid
concentration by volume for mudflows and landslides are 0.33–0.4 and 0.43–0.53, respectively. The width-to-length ratios are
within the range of 0.02–0.2 for mudflows and 0.34–0.54 for landslides. Mudflows may be triggered on all three shapes of source
areas (i.e., planar, divergent, and convergent). By contrast, landslides may be initiated only on planar and divergent shapes.
Hence, this study provides early recommendation criteria to distinguish between landslides and mudflows based on mass
movement events in Indonesia.
Keywords: landslide, mudflow, criteria
1 INTRODUCTION liquid state. The water content of a landslide may

be lower than the liquid limit and may occur in a
A mass movement is a geological event in which soil plastic state (Abbot, 2004).
or rock moves out of slopes due to the influence of c. Solid concentration by volume
gravity. Among natural disasters, mass movement This parameter classifies mass movements based
causes the highest number of fatalities in Java and Bali. on sediment concentration. Solid concentration by
Mass movement can be classified into two types: volume (Cv) was proposed by OʹBrien and Julien
mudflow and landslide. A mudflow occurs when soil is (1988). The value of Cv is defined as follows:
in a saturated condition as a result of high-intensity rain
events. By contrast, a landslide occurs because of an 𝑉
𝐶𝑣 = 𝑉 +𝑠𝑉 , (1)
unstable slope when the driving force is greater than the 𝑠 𝑤
retaining force. 1
𝐶𝑣 = , (2)
1+𝑒
2 MATERIAL AND METHODS
1
𝐶𝑣 = . (3)
1+𝐺𝑠 .𝑤
2.1 Classification of mass movement
The following criteria can be used to categorize a mass where Vs is volume of solid part, Vw is volume of
movement either as a mudflow or a landslide. water, e is void ratio, Gs is specific volume, and w
is water content. Equation (3) is valid for saturated
a. Flow velocity soil.
The flow velocity of a mudflow can be very rapid d. Width-to-length ratio
(higher than 0.05 m/s) to extremely rapid (higher This criterion was recommended by Liu and Mason
than 5 m/s) depending on flow viscosity. The flow (2009). Width refers to the average width of a
velocity of a landslide is lower than 0.05 m/s mudflow, whereas length refers to the
(Cruden and Varnes, 1996; Varnes, 1978). transportation length from the source area to the
b. Viscous liquid state end of the deposition area. The width-to-length
The water content of a mudflow is equal to or ratio should range from 0.05 to 0.3.
higher than the liquid limit and occurs in a viscous
54
2.2 Slope shape cultivation area for fields and gardens. The change in
Slope shape considerably influences slope stability in a land functions is assumed to be one of the triggers of
steep terrain by concentrating or dispersing surface and the mudflow. Two springs in the source area and heavy
primarily subsurface water in a landscape. Sidle (2006) rains that occurred over the past few days resulted in a
identified three basic hydrogeomorphic slope units in saturated soil condition.
assessing terrain stability: divergent, straight, and
convergent.
Figure 1. Three common types of slope shape (Sidle, 2006)
Slope shape is determined to analyze whether a

Figure 1. Mudflow in Pakuon Village
correlation exists between slope shape and mass
movement type. Such determination is achieved by
using terrain mode from satellite photos, and then the Table 1. Soil parameters in Pakuon Village
contour form of the source area of the mass movement Mass Movement Type Mudflow
is classified according to slope shape as convergent, Date January 18, 2013
divergent, or planar. Coordinates 6°44ʹ29ʺS
107° 4ʹ20ʺE
2.3 Soil classification Water Content 66.41
Soil classification is performed via laboratory testing Specific Gravity 2.55
using the Atterberg limits, which include liquid limit Liquid Limit 66
Plastic Limit 47.52
and plastic limit tests. The liquid limit is the water
Plasticity Index 18.48
content in which soil condition changes from plastic to Liquidity Index 1.02
liquid. The plastic limit is the lowest water content in
which soil begins to become plastic. Soil types are
classified to identify the type of soil that is generally Mukapayung Village, Cililin, Bandung Barat
associated with the occurrence of mudflows and
landslides. On March 25, 2013, a mudflow occurred on Nagrok
Village Hills in Mukapayung Village, Cililin District,
3 RESULTS AND DISCUSSIONS West Bandung Regency, West Java (Figure 3 and
Table 2). This hill has a height of over 150 m with a
3.1 Description of the case studies slope of ±40°. Local residents cultivate fields and
gardens on this hill. The mudflow was suspected to
This research used 10 case studies of landslide and
occur due to changes in land functions and heavy rains
mudflow events in Java and Bali from 2013 to 2018
that occurred over the past few days.
(Riyanto, 2018; Suherman, 2018; Prakoso, 2017;
Johan, 2017; Sindaka, 2016; Deasmara, 2015; Naba,
2016; Krisdanto, 2014; Wibisono, 2014; Wahyuning,
2013). Nine case studies were conducted in Java,
whereas only one case study was performed in Bali.
These case studies are specifically described as
follows.
Pakuon Village, Sukaresmi, Cianjur, Jawa Barat
On January 18, 2013, a mudflow occurred on the hills

of Mount Karanggantung Beunying in Legok Lebe,
Pakuon Village, Sukaresmi District, Cianjur Regency,
West Java (Figure 2 and Table 1). This hill has a height
of over 150 m with a slope of ±30°. It is used as a Figure 2. Mudflow in Mukapayung Village
55
Table 2. Soil parameters in Mukapayung Village water sources around the mudflow site were assumed
Mass Movement Type Mudflow to have triggered the mudflow.
Date March 25, 2013
Coordinates 7°0ʹ27ʺS
107°28ʹ33ʺE
Water Content 64.36
Specific Gravity 2.74
Liquid Limit 58
Plastic Limit 29.93
Liquidity Index 1.23
Parakan Muncang, Cimanggung, Sumedang
On March 17, 2014, a landslide occurred in the Parakan Figure 4. Mudflow in Honje Village
Muncang area, Cimanggung Subdistrict, Sumedang
Regency, West Java (Figure 4 and Table 3). This Table 4. Soil parameters in Honje Village
landslide damaged a building being constructed at the Mass Movement Type Mudflow
foot of the landslide area. The landslide occurred at a Date July 13, 2014
slope of ±35°. Coordinates 7°30ʹ6ʺS
108°8ʹ48ʺE
Water Content 69.64
Liquid Limit 68.55
Plastic Limit 36.64
Jemblung Village, Sampang, Karang Kobar,

Banjarnegara
Figure 3. Landslide in Parakan Muncang On December 12, 2014, a couple of mass movements
were reported in Jemblung Hamlet, Sampang Village,
Karang Kobar Subdistrict, Banjarnegara Regency,
Table 3. Soil parameters in Parakan Muncang
Central Java (Figure 6 and Table 5). First, a mudflow
Mass Movement Type Landslide occurred on the west side of Mount Telagalele, which
Date March 17, 2014 was followed by a landslide on the east side of the
Coordinates 6°55ʹ54ʺS mountain. This occurred at a slope of ±35°. Land use
107°48ʹ40ʺE
around the mudflow area, which is mostly plantations
Water Content 51.04
Specific Gravity 2.6 and cultivated croplands such as rice, banana, and
Liquid Limit 66.64 pepper fields, cannot absorb water easily.
Plastic Limit 29.28
Honje Village, Parung Ponteng, Tasikmalaya
On July 13, 2014, a mudflow was reported in Honje

Village, Parung Ponteng District, Tasikmalaya
Regency, West Java (Figure 5 and Table 4). The
mudflow occurred at a slope of ±30°. This slope is used
for rice fields and mixed gardens interspersed with
jambon trees. High-intensity rain events over the past Figure 5. Mass movement in Jemblung Village
three days before the incident and several ponds and
56
Table 5. Soil parameters in Jemblung Village Before the landslide, continuous rain occurred for 3
Mass Movement Type Mudflow and days, which was suspected to have triggered the
Landslide landslide. This ±30° slope functions as wood forests
Date December 12, 2014 and plantations of food crops, such as cassava, for local
Coordinates 7°16ʹ48ʺS residents.
109°43ʹ12ʺE
Water Content 72.99 and 41.75
Liquid Limit 64.83
Plastic Limit 40.32
Liquidity Index 1.33 and 0.06
Margamukti Village, Pangalengan, Bandung
On May 5, 2015, a mudflow was reported at the foot of

Mount Bedil in Margamukti Village, Pangalengan
District, Bandung Regency, West Java (Figure 7 and
Table 6). Heavy rainfall triggered the mudflow, which
occured at a slope of ±45°–50°. The land around the
slope used to be pine forests that have been converted
to elephant grass pasture.
Figure 7. Landslide in Karangrejo Village
Table 7. Soil parameters in Karangrejo Village

Mass Movement Type Landslide
Date June 18, 2016
110°2ʹ48ʺE
Figure 6. Mudflow in Margamukti Village Water Content 47.49
Liquid Limit 73.62
Table 6. Soil parameters in Margamukti Village
Plastic Limit 37.95
Mass Movement Type Mudflow Plasticity Index 35.67
Date May 5, 2015 Liquidity Index 0.27
107°37ʹ52ʺE
Water Content 73.84 Songan Village, Kintamani, Bangli
Liquid Limit 77.55 On February 10, 2017, a landslide occurred in Songan
Plastic Limit 44.44 Village, Kintamani District, Bangli Regency, Bali
Plasticity Index 33.11 (Figure 9 and Table 8). This landslide was recorded at
Liquidity Index 0.89 a slope of ±30°–40°, and it started at Bubung Pegat
Road. The soil composition in Songan Village is
mostly gravel and silt sand or clay. In addition, Songan
Karangrejo Village, Loano, Purworejo
Village is located on the mainland, adjacent to Mount
On June 18, 2016a landslide occurred on the valley of Batur, which is one of the active volcanoes in
Kali Ginting, Karangrejo Village, Loano District, Indonesia. Therefore, a layer of volcanic products, such
Purworejo District, Central Java (Figure 8 and Table 7). as lava sediments, is also present. This content makes
Karangrejo is classified as vulnerable to landslide the soil fertile, and thus, local residents use the slopes
disasters because its territory is largely composed of to plant gardens and fields.
peat soil and illegal logging is rampant in the area.
57
Figure 9. Mudflow in Banaran Village
Table 9. Soil parameters in Banaran Village

Mass Movement Type Mudflow
Date April 1, 2017
Figure 8. Landslide in Songan Village Coordinates 7°51ʹ4ʺS
111°40ʹ43ʺE
Table 8. Soil parameters in Songan Village Water Content 55.59
Mass Movement Type Landslide Specific Gravity 2.72
Liquid Limit 53.93
Date February 10, 2017
Plastic Limit 45.42
115°24ʹ59ʺE
Water Content 33
Liquid Limit 36.1
Plastic Limit 18.62
Banaran Village, Pulung, Ponorogo
On April 1, 2017, a mudflow occurred in Banaran

Village, District Pulung, Ponorogo Regency, East Java
(Figure 10 and Table 9). Banaran Village is located
around the Wilis Mountains, which have a composition
of clay and loose clay soil. The Banaran cliff area is
used as fields for growing crops, such as ginger plants.
One of the causes of the mudflow is the conversion of
land use from pine forests to ginger fields. The
mudflow incident in Banaran Village occurred at a
slope of ±40°.
Figure 10. Mudflow in Pramen Village
Pramen Village, Bantar, Wanayasa, Banjarnegara
On January 7, 2018, a mudflow occurred in Bantar Table 10. Soil parameters in Pramen Village
Village, Wanayasa District, Banjarnegara Mass Movement Type Mudflow
Regency,Central Java (Figure 11 and Table 10). Many Date January 7, 2018
previous landslides have occurred on the eastern part of Coordinates 7°17ʹ34.47ʺS
the mudflow site. The mudflow was triggered by high 109°44ʹ48ʺE
Water Content 74.11
rainfall intensity in the past week, strong erosion, and
land use conversion from forests to fields and salak Liquid Limit 75.75
gardens. The mudflow occurred at a slope of ±30°–40° Plastic Limit 41.75
and caused the district road from Bantar Village to Plasticity Index 34
Suwidak Village to collapse. Liquidity Index 0.95
58
3.2 Soil type 3.4 Width-to-length ratio

The classification system used for soil type is the The data for the width and length of the mass
Unified Soil Classification System, which was movements are obtained from satellite photos, disaster
established based on the results of laboratory tests. reports, and previous studies.
As seen on Table 11, mudflow can generally occur in These data have an outlier, namely, Honje Village,
soil type MH (high plasticity silt), whereas landslide which is within the B/L range for landslides. However,
can occur in soil types MH (high plasticity silt), CH its original condition is mudflow. Width-to-length ratio
(high plasticity clay), and SC (sandy clay) . influences the process of determining the type of mass
movement that occurs. For mudflow, B/L is within the
Table 11. Soil classification analysis range of 0.02–0.20; for landslide, B/L is within the
range of 0.34–0.54 (Table 13).
Soil Mass Movement
No. Location
Type Type
1 Pakuon Village MH Mudflow Table 13. Width-to-length ratio
2 Mukapayung Village MH Mudflow Mass
No
Location L B B/L Movement
3 Parakan Muncang CH Landslide .
Type
4 Honje Village MH Mudflow 1 Pakuon Village 368 20 0.05 Mudflow
Landslide and Mukapayung
5 Jemblung Village MH 2 1000 20 0.02 Mudflow
Mudflow Village
6 Margamukti Village MH Mudflow 3 Parakan Muncang 147 79 0.54 Landslide
7 Karangrejo Village MH Landslide 4 Honje Village 120 64 0.53 Mudflow
8 Songan Village SC Landslide 300 140 0.47 Landslide
5 Jemblung Village
9 Banaran Village MH Mudflow 500 70 0.14 Mudflow
10 Pramen Village MH Mudflow Margamukti
6 1000 200 0.2 Mudflow
Village
7 Karangrejo Village 225 83 0.37 Landslide
8 Songan Village 134 45 0.34 Landslide
3.3 Solid concentration by volume (𝐶𝑣 ) analysis 9 Banaran Village 1100 200 0.18 Mudflow
OʹBrien (1988) recommended the classification of the 10 Pramen Village 650 89 0.14 Mudflow
𝐶𝑣 value for mudflow to be within the range of 0.45– Note: L = length of mass movement
B = average width of mass movement
0.55, whereas that for landslide should be higher than
B/L = width-to-length ratio
0.55.
The value of the recommended classification for solid

3.5 Slope shape
concentration by volume (𝐶𝑣 ) in the 10 case studies of
mass movement in Java and Bali can be within the Slope form is determined by analyzing the contour of
range of 0.33–0.40 for mudflow and 0.43–0.53 for the source area for each case study. Slope shape of 10
landslide (Table 12). cases can be seen at Table 14.
Table 14. Slope shape

Table 12. Values of solid concentration by volume
Mass
Mass Movement No Slope
No. Location 𝐶𝑣 Location Movement
Type . Shape
Type
1 Pakuon Village 0.371 Mudflow 1 Pakuon Village Mudflow Convergent
2 Mukapayung Village 0.362 Mudflow Mukapayung
2 Mudflow Straight
3 Parakan Muncang 0.430 Landslide Village
4 Honje Village 0.363 Mudflow 3 Parakan Muncang Landslide Straight
0.467 Landslide 4 Honje Village Mudflow Divergent
5 Jemblung Village Landslide
0.334 Mudflow
5 Jemblung Village and Divergent
6 Margamukti Village 0.327 Mudflow
Mudflow
7 Karangrejo Village 0.451 Landslide 6 Margamukti Village Mudflow Straight
8 Songan Village 0.530 Landslide 7 Karangrejo Village Mudflow Divergent
9 Banaran Village 0.398 Mudflow 8 Songan Village Landslide Straight
10 Pramen Village 0.353 Mudflow 9 Banaran Village Mudflow Straight
10 Pramen Village Mudflow Straight
59
Landslides tend to occur on planar and divergent slope Liu, J. G. & P. J. Mason, 2009. Essential Image
shapes, but do not occur on a convergent slope shape. Processing and GIS for Remotes Sensing. West Sussex:
By contrast, mudflows can occur on divergent, planar, Wiley-Blackwell.
and convergent slope shapes.
Naba, S. M. B., 2015. Simulasi Terjadinya Pergerakan
4 CONCLUSIONS Tanah di Dusun Jemblung, Banjarnegara dengan
Bantuan Program FLO-2D. Bandung: Universitas
In this study, mudflows generally occur in high Katolik Parahyangan.
plasticity silt soil type, whereas landslides can occur in
high plasticity silt, high plasticity clay, and sandy clay O'Brien, J. S. & Julien, P. Y., 1988. Laboratory
soil types. Analysis of Mudflow Properties. J. Hydraul. Eng., pp.
877-887.
The recommendation for the classification of solid
concentration by volume values is within the range of Prakoso, J. S., 2017. Perbandingan Hasil Analisis
0.33–0.40 for mudflows and 0.43–0.53 for landslides. Pergerakan Tanah Menggunakan Model Bingham dan
This result differs from the recommendation proposed Model Voellmy : Studi Kasus Desa Songan, Bali.
by OʹBrien, in which the value for mudflows is within Bandung: Universitas Katolik Parahyangan.
the range of 0.45–0.55 and that for landslides should be
higher than 0.55. Riyanto, N., 2018. Simulasi Pergerakan Tanah pada
Studi Kasus Longsor di Bantar, Banjarnegara
The width-to-length ratio for mudflows is within the Menggunakan Program Flo-2D. Bandung: Universitas
range of 0.02–0.20 and that for landslides is within the Katolik Parahyangan.
range of 0.34–0.54. The range for mudflows is nearly
Sidle, R. C. & Ochiai, H., 2006. Landslides (Processes,
similar to that proposed by Liu and Mason, i.e., 0.05–
Prediction, and Land Use). Washington: AGU Books
0.3.
Board.
Landslides tend to occur on planar and divergent slope Sindaka, J. K., 2016. Penentuan Parameter Reologi
shapes, whereas mudflows can occur on divergent, dengan Fall Cone Penetrometer, Mini Vane Shear dan
planar, and convergent divergent slope shapes. Flow Box untuk Menjelaskan Perilaku Pergerakan
Tanah Longsoran di Pangalengan. Bandung:
REFERENCES Universitas Katolik Parahyangan.
Abbot, P.L., 2004. Natural Disaster. 4th ed. New York,
Suherman, T., 2018. Trend Variasi Perubahan
USA: Mc-Graw Hill.
Parameter Reologi Beberapa Sampel di Lokasi
Cruden, D.M. & Varnes, D.J., 1996. Landslide types Longsoran Ponorogo Jawa Timur. Bandung:
and processes. Landslides: investigation and Universitas Katolik Parahyangan.
mitigation. A.K. Turner and R.L. Shuster R.L., eds.,
Varnes, D. J., 1978. Slope Movement and Types of
Transp. Res. Board., Spec., pp. 36-75.
Processes in Landslides, Analysis and Control
Deasmara, G., 2015. Penentuan Parameter Rheology Transportation Research Board. Washington D.C.:
Menggunakan Flow Box Test di Parungponteng National Academy of Sciences.
Tasikmalaya. Bandung: Universitas Katolik
Wahyuning, A., 2013. Simulasi Mudflow di Sukaresmi
Parahyangan.
- Cianjur Menggunakan Program FLO-2D. Bandung:
Johan, A., 2017. Pemodelan Longsoran pada : Studi Universitas Katolik Parahyangan.
Kasus Longsoran Desa Karangrejo dengan Flo-2D
Wibisono, D., 2014. Penentuan Nilai Kuat Geser dan
dan RAMMS. Bandung: Universitas Katolik
Viskositas pada Kaolin dan Tanah Longsor Parakan
Parahyangan.
Muncang dengan Uji Geser Baling-Baling di
Krisdanto, M. A., 2014. Simulasi Pengaruh Perubahan Laboratorium. Bandung: Universitas Katolik
Kadar Air Terhadap Longsoran Mudflow Parahyangan.
Menggunakan Program Flo-2D : Studi Kasus
Mukapayung - Cililin. Bandung: Universitas Katolik
Parahyangan.
60
Stability of Metamorphic Rock Slope Based on Limit Equilibrium Method at

Poboya Gold Mine, Palu, Central of Sulawesi
Sriyati Ramadhani
Department of Civil and Environmental Engineering, Universitas Gadjah Mada, Yogyakarta, INDONESIA
sriyati.ramadhani@mail.ugm.ac.id
*Ahmad Rifa’i
Email: ahmad.rifai@ugm.ac.id
Wahyu Wilopo
Department of Geological Engineering and Center for Disaster Mitigation and Technological Innovation (GAMA-InaTEK),
Universitas Gadjah Mada, Yogyakarta, INDONESIA
Email: wilopo_w@ugm.ac.id
Kabul Basah Suryolelono

Email: kabulbasah@yahoo.com
*Corresponding Author
ABSTRACT
Poboya is considered as one of the traditional gold mine in Palu City. There were several landslides or ground movements
occurred due to slope cuts and underground excavation by miners which unwittingly lead to force changes on the slopes that
result in disruption of slope stability. The purpose of this study is to identify the stability of metamorphic rock slope based on
the limit equilibrium method at Poboya gold mine. The study was conducted at Mantikulore district of Palu City. The limit
equilibrium method was used for numerical analysis with the Slide 7.0 software. The result of the analysis shows that the
condition of the existing slope is still safe. However, after changing of slope geometry and adding earthquake load the safety
factor is decrease. The results explain that the mining or excavating method of the slope are safe if there is no earthquake,
however with the addition for earthquake loads of 0,75 g slope conditions become critical. Slopes become vulnerable to ground
movement.
Keywords: Slope stability, metamorphic rock, equilibrium limit.
1. INTRODUCTION mining of material carried out by miners on the slopes

Mining is the activity of digging and taking the of the hillsides as well as the weathering of rocks
valuable metal and non-metal minerals from the earth. existence and earthquake caused by Palu Koro fault
Poboya Mountains is one of the gold mines located in through of Mantikulore district. Based on the above
Palu City. Miners conduct quarrying and slope slicing description, it is necessary to analyze the stability of
to extract gold ores every day. One of several aspects metamorphic rock slopes in Poboya gold mine.
must be taken into consideration is the safety aspect. The purpose of this research is to identify the stability
This aspect considered very important to slope stability of metamorphic rock based on the limit equilibrium
since it has many negative impacts, especially the method by using Slide 7.0 software. The research was
social insafety impact. Ground movements in Poboya conducted in Poboya gold mine which is
gold mining area are enormous, as it is based on the geographically located in Mantikulore district, Palu
existence of a large rock outcrop, the structure of the City, Central of Sulawesi Province. (Figure 1)
material in the form of soil and rocks, the slope is
between 20% to <45%, and the existence of erosion
grooves (BPBD, 2014).
The trigger factors of the ground movement occurred
in Poboya gold mine area were the excavation and
61
mineralization. Metamorphic rocks of Toboli

Formation is the dominant host rock of mineralization.
The mineralized Zone is associated with quartz-
carbonate vein and vein breccia within dominantly
banded, chalcedonic, and bladed textures. The vein
contains multiple episodes of quartz-carbonate phase
(Wajdi, et al, 2011).
The geological structure of the Poboya region in the
form of a fault, the geological structure (cesarean)
formation is the transportation route of minerals in the
hydrothermal environment. This is reinforced by the
formation of a fault structure on the eastern side that
extends from north to south, and the western side which
is suspected to be a settling process by hydrothermal
solution through fracture as a controller of the path of
mineralization process in the area. The area of Poboya
and its surroundings is dominated by quartz rocks as a
gold mineral carrier contained in quartz veins. Quartz
rocks are the main type of rock that forms a fault
structure and become an important factor in the
formation of gold (Junaedy, et al, 2016).
The rock’s composition of Poboya area based on
stratigraphic sequence from old age to young
(Sukamto, et al. 1973).
a. Metamorphic rock
These rocks are found around the eastern border of the
Fig. 1. The geology map of Palu city regional study area which is also Palu district boundary with
(Sukamto, et al., 1973, modified). Parimo. Schist rocks are generally clogged with more
intensive surface weathering. The areas composed by
these rocks are generally vegetated, so that relatively
2. THE GEOLOGY CONDITION OF POBOYA intact rock physical conditions can only be observed
The geological conditions of the island of Sulawesi and around the river cliffs in the northeast or upstream of
the surrounding area can be divided into two major the Poboya River.
parts, namely the western and eastern parts. The b. Intrusion rocks
western part includes the southern arm, the middle and The intrusion rocks in the study site are found in
extends into northern arm of Sulawesi called arc relatively small masses as well as local intrusive bodies
volcanism West Sulawesi. The eastern section consists that infiltrate the older rocks. The physical properties
of the southeastern arm and the eastern arm of of the rocks are fractured, fragmented a partly
Sulawesi, formed by two large groups of rocks that weathered.
include; continental pieces and ocean fragments, these c. Sedimentary rock type molasses
two distinct rock pieces originally covered by the These rocks consisted of conglomerate, sandstone, silt,
molecular sediments known as the molasses of and clay. The rocks spread in the study site is very large
Sulawesi (Surono, et al, 1948). and considered as dominant composer about 90%
The main fault of Palu and its surrounding areas is the dominates the distribution of composer rock. The
Palu-koro fault. This fault runs north-west-south nature of bedding that is not constantly on this rock in
direction. Palu City is allegedly located between two some places can be observed mainly in the cliffs
segments of Palu fault that resulted in the formation of Poboya River.
Palu valley. Other structures are Pasangkayu fault, d. Alluvial rocks
Palintuna fault, Sausu fault, Parigi fault, Pasangkayu This material is the compiler of Poboya river region
fault and valley formation (Surono, et al, 1948). which is characterized by terrain morphology. The
Poboya located in the eastern zone of the main basin composition of this alluvial fragment consists of schist,
connected to Palu Koro fault, which is a part of the gneiss, diorite, basalt, and granite.
Northern Neogene of Sulawesi Island (Kavalieris, et al,
1992).
Structurally, Palu-Koro Fault has formed a host
structure and fluid conduits (pathways) for the
62
3. LIMIT EQUILIBRIUM ANALYSIS carried out by dividing the sloping ground masses into
Limit equilibrium analysis is a limit equilibrium several slices considered as a unique shear.
method of potentially moving masses by comparing
both driving and holding forces along the landslide 4. SLOPE STABILITY
plane, illustrating the active forces on the slip. The Hoek (1991) proposed a safety factor for the design of
comparison of the two forces will result in safety factor various types of rocks in engineering problems ranging
value (SF) of the slope, with limit equilibrium from safety factor values greater than 1 for extreme
conditions to be achieved when the value of SF = 1 (de loading, seismic and gravity analysis to dam design, the
Vallejo and Ferrer, 2011). safety factor value is greater than 1.5 for a permanent
According to Abramson, et al., (2002) the slope rock slopes (Table 1) and the safety factor value is
stability analysis based on limit equilibrium method is greater than 2 for block fall-out in the tunnel.
Table 1. Safety factor criterion (Hoek, 1991)

Safety Factor (SF) Occurrence
F<1 Collapse
1 ≤ F < 1,5 A critical condition (not significant in design)
F ≥ 1,5 Stable condition (the design above the critical value)
5. RESEARCH METHOD region based on ground vibration acceleration map

The analysis of slope stability were using the limit (Pusgen 2017) is 0,75 g. This slope stability analysis
equilibrium with the help of Slide 7.0. The limit was performed at 3 (three) point location slopes i.e.
equilibrium method in this study uses the concept of location 1 (gneiss) located at the coordinate point
safety factor (SF) value in analyzing the two modeling 828098 mE and 9904512 mS, location 2 (schist)
of an existing slope modeling and slope modeling with located at the coordinate point 828182 mE and
various slopes. The type of surface failure used in this 9904498 mS and location 3 (schist) at the coordinate
study is non-circular with the material model using point 828249 mE and 9904446 mS, where the three
Mohr-Coulomb failure criteria and modeled as locations are the miners do a lot of excavation activities
anisotropic. The study area is prone to earthquake, and slope cutting to take rocks containing gold ore. The
therefore the earthquake load was include in the input parameters required for slope stability analysis
analysis. Value of PGA used in this study for Poboya using the slide program can be seen in Table 2
Table 3. Parameter Input used in Slides 7.0.

Parameter Loc. 1 Loc. 2 Loc. 3
Material Name Gneiss Schist Schist
Unit Weight (kN/m3) 23 25 25
Cohesion (kPa) 178 129 255
ϕ ( o) 23 25 31
Tensile Strength (kPa) 2300 8270 7170
6. RESULTS AND DISCUSSION

The analysis of slope stability in this research by using height of 32 m with an overall slope of 270. The results
the limit equilibrium method consists of two kinds of of slope stability analysis for an existing condition can
analysis, both are slope stability analysis with an be found in Table 3. Based on the findings in Table 3,
existing slope modeling and slope modeling based on it shows that the value of safety factor at location 1 for
slope variation. The findings are as follow: the static load is 2,237 and the dynamic load is 0,726.
The results explain that for existing slopes at location 1
a. An existing condition slope modeling in stable condition if no earthquake occurs due to SF>
An existing slope stability analysis is a slope stability 1.5 and possibly will collapse in case of earthquake due
analysis based on the initial shape of the slope model to SF <1 (Hoek, 1991).
of Poboya gold mine. There are 3 locations to be
analyzed on slopes in Poboya gold mine. Location 1 The analysis results for location 2 in Table 3 show that
has a slope height of 61m with an overall slope of 200, the factor value is safe for static loads of 2.579 and the
location 2 has a slope height of 33 m with an overall dynamic load is 0.809. The findings explain that the
slope of 190, and meanwhile, location 3 has a slope existing slope at location 2 will be stable if there is no
63
earthquake due to SF > 1.5 and possibly will collapse condition in case of earthquake due to SF <1,5 (Hoek,
in case of earthquake due to SF <1 (Hoek, 1991). The 1991). Based on those findings, the existing slopes of
analysis results in Table 3 for location 3 indicate that the Poboya gold mine are in stable condition in the
the safety factor value for the static loads of 3.106 and absence of an earthquake and are likely to collapse or
the dynamic load is 1.258. The findings explain that for in critical condition in the occurrence of an earthquake.
existing slopes at location 2 are stable if there is no The appearance of field conditions and the existing
earthquake due to SF > 1.5 and possibly in critical slope modeling analysis results can be seen in Table 4.
Table 3. The result of slope stability simulation with the limit equilibrium method for an existing condition.
Slope Slope Safety Factor (SF)
Location Lithology height Angle
Static
(m) (o) Dynamic
1 Gneiss 61 20 2,237 0,726
2 Schist 33 19 2,579 0,809
3 Schist 32 27 3,106 1,258
Table 4. Field Condition Appearance and analysis results of existing slope modeling.
location Location condition Slope modeling
Static Dynamic
1
b. Slope modeling value for sloping variation based on limit equilibrium

Slope modeling is a slope stability analysis of geometry method can be explained below as follows:
changes in slope by analyzing the variation of slopes. 1. Location 1 has a height slope of 61 m with new
The result of slope stability analysis for new design design slope angles of 16o, 18o, 22o and 24o. Table 5
slope model can be seen in Table 5. The safety factor shows that the safety factor values of static loads at
64
location 1 with slopes of 16o, 18o, 22o and 24o are 3. Location 3 has a height slope 32 m with new design
2,707, 2,444, 2,062 and 1,905 respectively, this slope angles of 23o, 25o, 29o and 31o. Based on Table
indicated that slope is still in safe condition in the 5, it shows that the safety factor value for static
absence of earthquake due to the value SF > 1,5 loads at location 3 with slopes of 23o, 25o, 29o and
(Hoek, 1991). Meanwhile, for dynamic loads, the 31o are 3,587, 3,277, 2,979 and 2,761 respectively,
safety factor values at location 1 with slopes of 16o, this denotes that slope is still in safe condition in the
18o, 22o and 24o are respectively 0,767, 0,735, 0,690 absence of earthquake due to the value SF > 1,5
and 0,670, this denotes that slope is unsafe in the (Hoek, 1991). For the dynamic loads, the safety
occurrence of an earthquake, for the value SF < 1 factor value at location 3 with slopes of 23o, 25o, 29o
(Hoek, 1991). and 31o are respectively 1,271, 1,269, 1,238 and
2. Location 2 has a height slope of 33 m with new 1,218, this indicates that slope is in the critical
design slope angles of 15o, 17o, 21o and 23o. Based condition if earthquake occurs, for the value of SF
on Table 5, it shows that the safety factor values of < 1,5 (Hoek, 1991).
static loads at location 2 with slopes of 15o, 17o, 21o
and 23o are 3,197, 2,900, 2,335 and 2,176 The analysis result of slope stability on both of the
respectively, this explains that slope is still in safe existing slope and the geometry changing slopes
condition in the absence of earthquake due to the showed that slopes in these 3 locations of Poboya gold
value of SF > 1,5 (Hoek, 1991). Meanwhile, for the mine are in stable condition if the miners do slaughters
dynamic loads, the safety factor value at location 2 or excavation if no seismic occur, but after the addition
with slopes of 15o, 17o, 21o and 23o are respectively of the earthquake load of 0,75g the value of the safety
0,832, 0,810, 0,788 and 0,782, this indicates that factor decreases so that the slope conditions become
slope is unsafe in the occurrence of earthquake, for unsafe and critical.
the value of SF < 1 (Hoek, 1991).
Table 5. The result of slope stability simulation with limit equilibrium method for variation of the slope.
Slope Slope Safety Factor (SF)
Location Lithology Angle
height (m) (o) Static
Dynamic
16 2,707 0,767
18 2,444 0,735
1 Gneiss 61
22 2,062 0,690
24 1,905 0,670
15 3,197 0,832
17 2,900 0,810
2 Schist 33
21 2,335 0,788
23 2,176 0,782
23 3,587 1,271
25 3,277 1,269
3 Schist 32
29 2,979 1,238
31 2,761 1,218
7. CONCLUSIONS adding earthquake loads of 0,75g, the slope condition

The result of slope stability analysis based on the limit became unsafe and critical up to sloping degrees of 16o,
equilibrium method indicates that the existing slopes at 15o, and 23o. Based on the findings, it is necessary for
the three different location are in safe condition in the activities of excavation and cutting at Poboya gold
absence of an earthquake with the safe factor value is mine to redesign slopes to be slighter in order to reduce
>1.5. However, after the adding of earthquake loads of landslides of slopes in case of earthquake.
0,75g, the safety factor value decreased <1, hence, the
slope condition became unsafe and critical. The results ACKNOWLEDGMENTS
also shows that for slope conditions with sloping The authors highly appreciate the Ministry of Research
variation model, it was obtained that location 1, 2 and and Technology and Higher Education for funding this
3 are in safe condition up to sloping degrees of 24o, 23o, research in the form of postgraduate scholarship. Many
and 31o if no earthquake occurred. However, after the thanks are also given to the research team for their
65
significant contribution during data collection in the

research field. We also thank the head and the
technicians of the Soil Mechanics Laboratory as well
as the Structure Laboratory of Gadjah Mada University
for their assistance during the laboratory study.
REFERENCES
Abramson, L. W., Lee, T. S., Sharma, S., & Boyce, G.
M. (2002). Slope Stability and Stabilization
Methods (2nd ed ed.). New York: John Wiley
& Sons, Inc.,.
BPBD, K. P. (2014). Laporan Dala Banjir Sungai
Pondo. Kota Palu.
Carlile, J. C. (1983). Geology, Exploration
Geochemistry and Mineralization of the
Tombolilato District,. Sulawesi, Indonesia.
de Vallejo, L. I., & Ferrer, M. (2011). Geological
Engineering. London, New York: CRC Press
Taylor dan Francis Group.
Hoek, E. (1991). When is a Design in Rock
Engineering Acceptable. Aachen: Proceedings
of The 7th International Congress on Rock
Mechanics, 3, 1485-1497.
Junaedy, M., Efendi, R. and Sandra, S., Studi Zona
Mineralisasi Emas Menggunakan Metode
Magnetik Di Lokasi Tambang Emas
Poboya. Online journal of Natural
Science, Vol. 5 (No. 2): 209-222, August 2016.
http://dx.doi.org/10.22487/25411969.2016.v5.
i2.6708
Kavalieris, I., Van, L. T., & Wilson, M. (1992).
Geological setting and styles of mineralization,
north arm of Sulawesi Indonesia. Journal of
Southeast Asian Earth Science, 7, 113-129.
Pusgen. (2017). Peta Sumber dan Bahaya Gempa
Indonesia 2017. Jakarta: Pusat Studi Gempa
Nasional.
Soekamto, Sumadirdja, H., Suptandar, T.,
ardjoprawiro, S., & Sudana, D. (1973). Peta
Geologi Tinjau Lembar Palu, Sulawesi.
Bandung: Pusat Penelitian dan Pengembangan
Geologi.
Surono, 1948, (editor) & Hartono, U. (Udi), (editor.)
2013, Geologi Sulawesi, Cetakan pertama.
(Menteng, Jakarta LIPI Press, November
2013).
Wajdi, M. F., Santoso, B., & Kusumanto, D. (2012).
Metamorphic Hosted Low Sulphidation
Epithermal Gold System. Majalah Geologi
Indonesia, 27 No. 2, 131-141.
66
Deterioration Depth of Cement Treated Clay under Sulfate Exposure
T. Pradita, L. Handoko*, S. Gunawan, and J. T. Hatmoko

Department of Civil Engineering, Universitas Atma Jaya Yogyakarta, Yogyakarta, INDONESIA
*Corresponding author: luky.handoko@uajy.ac.id
ABSTRACT
One of a major problem in geotechnical field is soft soil because of its low strength. In order to be able to build a structure above
soft soil, soil improvement has to be done. Deep mixing is one of soil improvement implementation. Deep mixing is done by
forming soil pile of cement treated clay with in-situ mixing. However, when soft soil is located in marine area that is considered
as an extreme environment, the seawater may cause corrosion to the soil pile. Sulfate is one of the chemical content in seawater
that is corrosive. Thus, the objective of the paper is to know about how severe the deterioration of soil pile with variation of
cement content (cc), water content (wc), and curing time (t) after the pile exposed to sulfate. Magnesium sulfate is used to
represent sulfate with the content of 10% which is uniform for all piles. The exposure is represented by the immersing of soil
pile onto soil-sulfate mixture for 7, 14, 28, and 56 days. The deterioration of the pile is investigated using penetration test. The
output of penetration test is in the form of cone penetration resistance (R) vs penetration depth (d) graph. The result of the test
shows that deterioration depth decreases as the increase of cement content, and the decrease of water content. Furthermore, the
on the undeteriorated zone, the resistance (Rreff) increases along with the increase of cc. In addition, Ca2+ and Mg2+ ion
investigation is also done in order to know the content of ion from the surface of the pile up to 1.5 cm below. It shows that the
deeper, the more Ca2+ exists. In contrary to that, Mg2+ decreases as the depth increases. This indicates deterioration at the surface
area which exposed to seawater.
Keywords: Sulfate Exposure, Cone Penetration Resistance, Deterioration Depth, Magnesium Sulfate.
1 INTRODUCTION Thus, a soil improvement is needed to strengthen the

ground. Soil-cement mixed piles are widely used to
Soft soil is one of a common concern in civil improve soft subsoil (Cui et al., 2016). Soil pile can be
engineering area. In Indonesia, soft soil can be found in erected by deep-mixing the ground. The stabilized soil
Java. For example, in North Java such as Jakarta, will have a higher strength, lower permeability and
Semarang, and Surabaya mainly consist of soft clay lower compressibility than the native soil. However, in
with a very little content of sand (Adi Tirta et al., 2018). regions with high-salinity groundwater, the soil-
Soft soil especially in the form of clay can be found cement is susceptible to deterioration (Cui et al., 2016).
easily in the marine area. The natural water content of High-salinity is one of marine area indication.
marine clay that is generally higher than the liquid limit Moreover, marine soil at the seabed is mixed with the
creates large void in the soil and thus making marine seawater containing sulfate above it, resulting in the
clay weak (Rao et al., 2011). deterioration of the soil pile. Figure 1 shows the cross
section of soil pile placement. Seawater above will be
Marine area or deep ocean is considered to be one of an
infused into the soil below causing the sulfate attack
extreme environment where hydrostatic pressure
inside the soil occurs radially from every directions.
occurs and mineral solute exists. Additionally, marine
area is considered as an extreme environment because
the seawater is having the possibility to cause corrosion
due to harmful chemical contained in it. One of the
most aggressive chemicals contained in seawater that
affect the long term durability of cemented material is
sulfate. Sulfate has 7.7% part out of the whole
composition of seawater, and magnesium takes 3.65%
portion. Sulfate attack has been reported to be a cause
of damage to concrete for over a century, and
considered as one of the major deteriorative problems
(Ganjian & Pouya, 2005).
When a structure is built above weak marine soft soil,

the soil will not be able to support the upper structure. Figure 1. Pile Placement Cross-Section
67
Table 1. Specimen Mix Design

Soil
Cement Cement Curing Time, t (days)
Water Penetration
Content, Content, C
Content, Test
cc (%) (kg/m3) 7 14 28 56
wc (%)
12 115.86 √ √ √ √ √
60
14 134.28 √ √ √ √ √
20 120 √ √ √ √ √
120 22 131.47 √ √ √ √ √
24 142.84 √ √ √ √ √
Sulfate attack leads to the redundant of the compressive

strength of cement blended material (Makhloufi et al.,
2016). Furthermore, the deterioration by sulfate attack
is characterized by a swelling of material due to the
formation of expansive products, which lead at a long
term to decohesion of this material and consequently a
degradation of its mechanical properties (Al-Dulaijan,
2007). Also, the deterioration decreased the effective
diameter of the soil-cement pile. The smaller the pile
diameter, the more obvious the difference in bearing
capacity between the deteriorated and undeteriorated Figure 2. Specimen Cast Dimension
piles (Cui et al., 2016). Thus, soils containing sulfate
should not be used for soil-cement stabilization 60% of liquid limit. Thus, in order to investigate about
(Sherwood, 1957). the strength behavior of high-water-content soil, the
targeted water content is both 60% and twice liquid
Salt-rich soft soils have not only general characteristics limit that is 120%. In the specimen making, water-soil
of common soft soils, but also contain high contents of mixture has to be homogenous confirmed.
Mg2+, Cl−, and SO42− (Xing et al., 2009). Hara et al.
(2014) identified that the most deteriorative content of Magnesium sulfate (MgSO4) was used to represent
seawater is magnesium sulfate. However, magnesium extreem environment. The content of MgSO4 is 10 %,
sulfate does affect unconfined compression strength of meaning in every 1 kg of water there is 10 gram MgSO4
specimens. UCS value reaches the peak value when the powder. It is mixed directly with the soft soil because
MgSO4 content is 4.5g/kg (Han et al., 2015). later the mixture will be used both for the specimen
Otherwise, as the content gets higher than 4.5g/kg, the mixture and the soaking solution. The immersion or
UCS value decreases. Therefore, to know the severity, curing times are 7, 14, 28, and 56 days.
MgSO4 is used as the deteriorative substances and the
Ordinary Portland cement is used as a binder for the
effect is studied in this paper. In order to tell about
specimen. The content variations of OPC are 12%,
deterioration severity, penetration test is done. The
14%, 20%, 22%, and 24%. The variations refer to
output of the test is in the form of Tip or Cone
active zone (Zhang et al., 2013) which is suitable to be
Resistance (R) vs Penetration Depth (d).
used for deep mixing is approximately more than 115
kg/m3.
2 METHOD AND MATERIALS
2.2 Specimen Preparation
2.1 Materials
Figure 2 shows the molding method of specimens. 8 cm
Sidoarjo Mud or Lumpur Sidoarjo (Lusi) is a result of Ø x 8 cm height cast is purposed to ease penetration test
mud volcano eruption in Sidoarjo, East Java. Lusi is implementation because by an adequate surface area,
very soft clay with liquid limit of 58.44%%, plastic the penetration will be more accurate since soil has
limit of 30.77%, and shrinkage limit of 22.27%. Lusi more area to distribute tension from the cone tip. 20
has 2.71 specific gravity and contains coarse grain cylinders are prepared in total in order to make
about 15.53%, silt 54.47%, and clay 30%. Since Lusi specimens by variation in Table. 1. Water content (wc)
has some similar properties with marine clay, Lusi is is set into 60% and 120%. Where 120% is set to be
used in the research.Lusi that is used as the soft soil has
68
twice the liquid limit of 60%. The variations of cement 3 CEMENT REACTIONS
content (cc) are 12% and 14% for the 60% wc Cement is manufactured through a closely controlled
specimens and 20%, 22%, and 24% for 120% wc chemical combination of calcium, silicon, aluminum,
specimens. Cement content is weighed based on the iron and other ingredients. There are some components
determined percentage towards soil weight (Ws). The of OPC that have their own role. Tricalcium aluminate,
focus is to compare several identical specimens that are C3A releases a lot of heat during the early stages of
cured at different periods. Thus, curing times (t) are hydration, but has little strength involvement. Gypsum
varied into 7, 14, 28, and 56 days. Since the laboratory slows down the hydration rate of C3A. Cement low in
specimens represent the real soil piles in marine area, C3A is sulfate resistant cement
magnesium sulfate (MgSO4) is added into the mixture (SRC). Tricalcium silicate, C3S hydrates and hardens
to indicate seawater on it. MgSO4 is used in the form of fast. It is largely responsible for OPC’s initial set and
powder as much as 10% of weight of water (Ww). Ww early strength gain. Dicalcium silicate, C2S hydrates
can be obtained by using Equation (1). and hardens slowly. It is mostly responsible for
strength gain after one week. Ferrite, C4AF is a fluxing
Ww = Ws .wc (1) agent which reduces the melting temperature of the raw
materials in the kiln (from 3,000° F to 2,600° F). It
While Ws can be obtained by: hydrates fast, but does not contribute much to the
strength of the cement paste (Hekal et al., 2002).
G s .V . w
Ws = (2)
   3.1.1 Water-Cement Reaction
1 +  G s. . c c 
 + (wc .G s ) 
  G  When cement is mixed with water, hardening reaction
   c   will happen. There are 2 types of reactions occur,
hydration and pozzolanic reaction. Hydration occurs
Where Ww is weight of water, Ws is weight of soil, and only between cement and water, while pozzolanic
wc is water content, Gs is soil specific gravity, V is total reaction occurs among cement and soil particles such
volume, ρw is water density (1000 kg/m3), cc is cement as silica and alumina.
content, and Gc is cement specific gravity. The mixing
duration is kept for 10 minutes uniformly for every C3S hydration:
specimens so that the mixture is mixed homogenously. 2 C3S + 6H→ C3S2H3 +3 Ca(OH)2 (3)
Before pouring the mixture into the mold, the bottom
part of the mold has to be sealed so that the liquid C2S hydration
mixture does not leak. The surface molded mixture has
2 C2S + 4H→ C3S2H3 + Ca(OH)2 (4)
to be flat in order to ease the penetration test afterwards.
After the surface flatten using spatula, specimen’s
surface is covered with porous paper in order not to mix C3S and C2S are the most important components
the liquid mixture with the immersing solution. The because they are responsible for strength. They require
porous paper has to be sealed at the folded section so approximately the same amount of water from
that the paper does not release. Specimen is then put hydration, but C3S produces more than twice of
inside a PVC that acts as an immersing container. The Ca(OH)2 that C2S hydration produces. From the
immersing solution is the mixture of soil, water, and hydration process, the product Ca(OH)2 will be
magnesium sulfate. MgSO4 content in this solution is triggered to react with soil’s substances as what shown
in accordance with the content in specimen mixture in Equation (5) to Equation (8). The reactions are
which is 10% Ww. Solution is then poured inside PVC included in pozzolanic reaction unity.
until it reaches 10 cm height above the surface of the
specimen. The opening of PVC is then covered by CaO + H2O → Ca(OH)2 (5)
plastic to prevent evaporation and to make wc always in
the stable content. The specimens are then cured for 7, Ca(OH)2 → Ca++ + 2(OH)– (6)
14, 28, and 56 days. Penetration test is performed right
after the curing time. Furthermore, chemical analysis is Ca(OH)2 + SiO2 → CSH (7)
also done in order to know about the content of ion
calcium and magnesium from the surface of the Ca(OH)2 + Al2O3 → CAH (8)
specimen, which is directly exposed to sulfate solution,
up to certain depth. Equation (5) is the lime hydration that occurs for the
first time. Cement has abundant amount of calcium and
therefore will react once it is mixed with water. Cation
exchange on Equation (6) occurs after lime hydration.
69
Equation (7) and Equation (8) shows the first stage of Cone Resistance (N)
pozzolanic reaction. Calcium hydroxide together with
0 100 200 300
silicon dioxide that is originated from the soil, will
form C-S-H or calcium silicate hydrate. Moreover,
0
calcium hydroxide will also produce C-A-H or calcium dn
aluminate hydrate in its reaction with aluminum oxide 5
Penetration Depth (mm)

that is also one of the composition of soil.
Ca2+ + 2(OH) + SiO2 + Al2O3 → C-A-S-H (9)

10
Equation (9) is the second stage of pozzolanic reaction.
The product is calcium aluminate silicate hydrate or C- 15
A-S-H that has higher strength compared to the
previous reactions.
20
3.2 Deterioration Caused by Seawater
Deterioration is the process of a cemented material to Rreff
become progressively weaker or worse. The severity of 25
deterioration can be caused of the reaction between Laboratory Data
MgSO4 with cement hydrates, and the repetitive Fitting Line (Equation 14)
crystallization cycles of MgSO4.nH2O by drying – 30
immersion of the hardened pastes that can produce Figure 3. Penetration Test Specimen 60% wc, 14 cc for 56-
internal stresses in pores leading to the formation of days Curing Time
cracks. The reaction of MgSO4 with cement hydrates is
shown in Equation (10) and Equation (11). 4 RESULT AND DISCUSSION
MgSO4 (aq) + Ca(OH)2 → CaSO4 . 2H2O + Mg(OH)2 4.1 Penetration Test
(10)
Figure. 3 shows a penetration curve of 60% wc, 14 cc
specimen that was cured for 56 days. The dots
MgSO4 (aq) + C-S-H → CaSO4 . 2H2O + M-S-H
symbolize the data that is gotten from laboratory
(11) research while the solid line represents the fitting line.
Magnesium sulfate will replace the existence of In this research, fitting line is proposed using Equation
calcium and form gypsum, which is a low in hardness (14).
and therefore reducing the strength of cemented-soil.
Furthermore, the formed gypsum from the above two Rreff
reactions reacts with calcium aluminate hydrate R = Rreff − (14)
(C4AH13) and calcium monosulfoaluminate hydrate
(C4ASH12) to form ettringite as mentioned below:
(1 + (  d ) )
 
C4AH13 + 3CaSO4. 2H2O + 14H2O → C3A. 3CaSO4. Where R is tip cone resistance, Rreff is reference cone
resistance, d is penetration depth (mm), and α,β,γ are
3H2O + Ca (OH)2 (12)
fitting constants. dn is the deterioration depth. All
contants are obtained from Matlab program.
C3A. CaSO4. 12H2O + 2CaSO4. 2H2O + 16H2O →
C3A. 3CaSO4. 32H2O (13) The closest the gap between fitting line and laboratory
data, the more accurate the result. It indicates that the
The formed products are M-S-H as a result of experiment fulfills the initial hypothesis which was
decalcification of C-S-H, gypsum, and ettringite. All of made using Equation (14). At some point, the
these lead to a decrease of compressive strength and the resistance becomes constant. It indicates that the
last two products can cause crack formation (Hekal et specimen has reached its maximum resistance. Figure.
al., 2002). Similar with gypsum, ettringite also has 4 shows the results of penetration test of 120% wc that
weak characteristic. The difference is that ettringite is cured for 7, 14, 28, and 56 days with variation of
gives swelling effect to the cemented material. cement content of 20%, 22%, and 24%. The longer the
immersing period, the more strength gained. It is
proved by the increase of cone resistance along with the
rise of curing term. Figure. 5 shows the penetration test
70
result of 60% samples. The increment of immersing However, although the specimens are set in such way
period should indicate the increase of cone resistance. to fulfill the planned contents, they may not have the
However, there is a peculiarity in Figure. 5 (a) that the exact value. For instance, for 120% wc specimen, the
resistance of 28 days-cured specimen is that the water content that is added between one and another is
resistance of 28 days-cured specimen is lower than 14 not necessarily the same (exact 120%) because
days-cured one. It is also clear that the transition specimens are mixed in different time. The same
deterioration depth of 14 days-cured sample is more applies with cement content. All the uncertainties are
sudden compared to the 7 days specimen. Transition caused by human error during mixing, and thus some
deterioration depth is a condition when fully cannot be compared one and another. The imprecise
deteriorated zone proceed to switch into the data can be seen in Table 2, there is some
undeteriorated zone by some slight increase of unsynchronized data where the pattern of water content
resistance. In contrary to this condition, Figure. 5 (b) is random as the curing time increases. It is irregular
shows the reasonable result where the resistance that one specimen has scattered data and the other has
increases along with the immersing period. Even so, the linear data. For example for specimen 120% wc 20 cc,
14 days-cured samples still shows a significant water content in 14 days curing time is significantly
transition from fully deteriorated zone into the higher compared to the 7 days curing time. The pattern
undeteriorated one. should have been in the form of wc decrease as the
curing time increases. Furthermore, there is an odd
Due to some peculiarities occur on the test results, there value contained on specimen 120% wc 22 cc 7 days-
is a possibility that water content is the one in charge cured where the bottom part of specimen which is not
of the error. Table 2 and 3 show the water content of directly exposed by soaking mixture contains higher
every specimen theoritically and practically after the water content compared to the surface area and the
curing time. Theoritical water content can be calculated theoretical one. Similar condition is also found in Table
by Equation (15). 3, where the bottom-part wc in specimen 60% wc 14 cc
28 days-cured is higher than the surface area. By all the
Ww
wct = (15) deviate data, it cannot be assured that the dose of each
Wc + Ws material measured is precise. Thus, in order to gather
the conclusion, some specimens that are not feasible
Where 𝜔𝑐𝑡 is theoretical water content, Ww is weight of need to be excluded. The excluded specimens are 60%
water, Wc is weight of cement, and Ws is weight of soil. wc 14 cc 28 days and 60% wc 12 cc 14 days.
Cone Resistance, R (N) Cone Resistance, R (N)

0 50 100 0 5 10 15
0 0
7 days 7 days
14 days 14 days
28 days 0.5 28 days
5 56 days
56 days
Fitting Line
1
10
Penetration Depth, d (mm)
1.5
15 2
2.5
20
25
3.5
30 4
(a)
71

0 50 100 150 0 10 20 30
0 0
7 days 7 days
14 days 14 days
28 days 0.5 28 days
5 56 days
Fitting Line 56 days
1
10

1.5
15 2
2.5
20
25
3.5
30 4
(b)

0 50 100 150 0 10 20
0 0
7 days 7 days
14 days 14 days
28 days 0.5 28 days
5 56 days
56 days
Fitting Line
1
10
1.5
15 2
2.5
20
25
3.5
30 4
(c)
Figure 4. Penetration test results for specimen with wc =120%; cc = 20% (a), cc, = 22% (b) and cc = 24% (c)
72

0 50 100 150 200 250 0 10 20 30
0 0
7 days 7 days
14 days 14 days
28 days 0.5
28 days
5
56 days 56 days
Fitting Line 1

10
1.5
15 2
2.5
20
25
3.5
30 4
(a)

0 100 200 300 0 10 20 30 40 50
0 0
7 days 7 days
14 days 14 days
28 days 0.5 28 days
5
56 days 56 days
Fitting Line 1
10
1.5
15 2
2.5
20
25
3.5
30 4
(b)
Figure 5. Penetration test results for specimen with wc =60%; cc = 12% (a) and cc = 14% (b)
73
Table 2. Water content of specimen with wc = 120%
Theoretical Final Water

Curing
Water Content, wcf (%)
Time, t Specimen
Content,wct
(days) Surface Bottom
(%)
120% wc, 20 cc 101.765 91.339 86.967
7 120% wc, 22 cc 100.097 95.284 113.302
120% wc, 24 cc 98.483 89.171 86.207
120% wc, 20 cc 101.765 100.610 90.595
14 120% wc, 22 cc 100.097 96.349 87.645
120% wc, 24 cc 98.483 94.192 83.063
120% wc, 20 cc 101.765 92.466 85.286
28 120% wc, 22 cc 100.097 86.135 82.609
120% wc, 24 cc 98.483 82.704 77.413
120% wc, 20 cc 101.765 83.165 86.644
56 120% wc, 22 cc 100.097 79.374 79.284
120% wc, 24 cc 98.483 85.521 79.448
Table 3. Water content of specimen with wc = 60%
Theoretical Final Water

Curing Content, wcf (%)
Water
Time, t Specimen
Content,
(days) Surface Bottom
wct (%)
60% wc, 12 cc 55.161 54.079 51.079
7
60% wc, 14 cc 54.193 50.442 48.966
60% wc, 12 cc 55.161 48.154 48.772
14
60% wc, 14 cc 54.193 48.080 44.316
60% wc, 12 cc 55.161 49.787 48.901
28
60% wc, 14 cc 54.193 46.862 47.643
60% wc, 12 cc 55.161 47.037 43.237
56
60% wc, 14 cc 54.193 39.252 35.281
350 4.2 Rreff Correlation

60% wc 12 cc
60% wc 14 cc
Figure. 6 shows the correlation between Rreff and
Reference Cone Resistance, Rreff (N)
300 immersing period. The Rreff increases as the curing time

120% wc 20 cc
120% wc 22 cc gets longer. The result above satisfies the theory that
250 are stated by some researchers. The test result shows
120% wc 24 cc
that the more cement added into the mixture, the larger
200 the Rreff which is in line with Hara et al. (2014). The
maximum Rreff occurs in the undeteriorated zone which
150
is not reached by the sulfate. Rreff increases as the
100 forming of pozzolanic reaction. This is because all the
components of cement have reacted with water, thus,
50 making the mixture harder. Rreff will be reached its
maximum at the undeteriorated zone. The condition is
0 caused by the immersing of specimen inside soil-
0 20 40 60 sulfate mixture that causes deterioration from the
Immersing Period, t (days) surface of the specimen up to its certain depth. Because
sulfate immersing is done, then the specimen is divided
Figure 6. Correlation of Refference Cone Resistance and into deteriorated zone and undeteriorated zone. By all
Immersing Period
74
Immersion Time, t (days) Immersion Time, t (days)

0 10 20 30 40 50 60 0 10 20 30 40 50 60
0 0
Deterioration Depth, dn (mm)

(a.1) (a.2)
120% wc - 20% cc 120% wc - 22% cc
1 1
2 2
3 3
4 4
Immersion Time, t (days) Immersion Time (days)

0 10 20 30 40 50 60 5 15 25 35 45 55 65
0 0
(a.3) (b)
Deterioration Depth dn (mm)

120% wc - 24% cc
1
1
2
2
3
3 60% wc - 12% cc
4
60% wc - 14% cc
4 5
Figure 7. Deterioration Depth for specimen with wc = 120% (a) and wc = 60% (b)
means, the longer the age of the specimen, the larger where magnesium sulfate replaces the position of
Rreff value due to pozzolanic reaction. If the specimen calcium and form gypsum is also high. This explains
was not immersed by sulfate, the cone or tip resistance the condition of a specimen that has high Rreff and dn
(R) will be approximately constant from the surface when immersed in a long period. Figure. 7 (b) which
because there is no deterioration. shows the dn of 60% wc specimens has proper data
where 14 % cc specimens have lower dn compared to
4.3 Deterioration Depth (dn) the 12 % ones. Furthermore, the data shows that the
A certain penetration depth can be considered as longer the curing time, the deeper the dn. Therefore, the
deterioration depth if there is no resistance or/and there result is suitable with Hara et al. (2014) and Yang et al.
is resistance that is close to zero. In line with strength (2016).
increment principal, the longer the immersing term, the
deeper or bigger the depth time increases. However, in 4.4 Chemical Analysis (Ca2+ and Mg2+ Content)
Figure. 7 (a.2), specimen 120% wc 22 cc shows a slight Figure. 8 shows the Ca2+ and Mg2+ content inside 120%
decrease of dn on curing time day 56. Figure. 7 (b) wc 24 cc specimen that is cured for 72 days. The ion
shows the excellent result of dn. The 12% cc specimen content checking is done for every 0.5 cm depth from
has deeper dn than 14% cc specimen. This corresponds the specimen surface for maximum 1.5 cm. The result
to Yang et al. (2016) where cement content is not only indicates significant increase of calcium along with the
increases the soil strength but also reduce the increase of sample depth. As for the opposite, there is
deterioration depth. a significant reduction of magnesium as the depth goes
deeper. The result is in line with Yang et al. (2016).
Deterioration depth is highly depending on sulfate and Due to the sulfate immersion, directly-contacted
cement content. Based on Yang et al. (2016) and Hara surface has more magnesium compared with specimen
et al. (2014), cement content decreases deterioration. in deeper area. Sulfate is not penetrated into the
The higher the cement content, the more shallow the undeteriorated zone. Thus, the deeper the segment,
deterioration depth will be. Generally, dn increases as magnesium content is smaller. Magnesium ion is
curing time gets longer in Figure. 7 (a). Pozzolanic abundant in deteriorated area because it replaces the
reaction has more time to occur during the curing time. position of calcium and forms gypsum which makes the
Its final product that is calcium aluminate silicate specimen weak. The reaction of replacement is shown
hydrate has strong behavior. Although the chance of C- in Equation (10) and Equation (11).
A-S-H formation is high, but the deterioration reaction
75
Consentration of Ca2+ / Mg+ (%) Al-Dulaijan SU, 2007. Sulfate resistance of plain and
0 1 2 3 4 5 blended cements exposed to magnesium sulfate
0 solutions. Construction and Building Materials 21,
Ca2+ 1792-802.
Mg+ Cui X, Zhang N, Li S, Zhang J, Tang W, 2016.
Depth from the surface (mm)
Deterioration of Soil-Cement Piles in a Saltwater

5 Region and Its Influence on the Settlement of
Composite Foundations. Journal of Performance of
Constructed Facilities 30, 04014195.
10 Ganjian E, Pouya HS, 2005. Effect of magnesium and
sulfate ions on durability of silica fume blended mixes
exposed to the seawater tidal zone. Cement and
Concrete Research 35, 1332-43.
15
Han P-J, Wang S, Chen FY, Bai X-H, 2015.
Mechanism of cement-stabilized soil polluted by
magnesium sulfate. Journal of Central South
20 University 22, 1869-77.
Figure 8. Ca2+ and Mg2+ Content on 120% wc 24 cc 72 Hara H, Suetsugu D, Hayashi S, Matsuda H, 2014.
days-cured Speimen Deterioration Progress of Cement-Treated Ariake Clay
under Seawater. Journal of the Society of Materials
However, calcium content gets bigger as the depth Science 63, 49-54.
increases. It is because the deeper the segments, the less
magnesium sulfate penetrates onto it. Thus, when there Hekal EE, Kishar E, Mostafa H, 2002. Magnesium
is less calcium replaced by magnesium sulfate, less sulfate attack on hardened blended cement pastes under
gypsum is formed. different circumstances. Cement and Concrete
Research 32, 1421-7.
5 CONCLUSIONS Makhloufi Z, Aggoun S, Benabed B, Kadri EH,
Generally, the result of this research lead to conclusion M.Bederina, 2016. Effect of magnesium sulfate on the
that deterioration of cement treated clay occurs due to durability of limestone mortars based on quaternary
extreme environment which is represented by MgSO4. blended cements. Cement and Concrete Composites
Water content, cement content and period of exposure 65, 186-99.
with MgSO4 is the main parameters which are affecting Rao DK, Raju GVRP, Sowjanya C, Rao JP, 2011.
the deterioration depth. Deterioration gets deeper with Laboratory Studies on the Properties of Stabilized
increase of water content, decrease of cement content Marine Clay from Kakinada Sea Coast, India.
and the longer the exposure periods. The reduction of International Journal of Engineering Science and
the Ca2+ near surface which exposed to MgSO4 Technology 3, 421-8.
validates the occurrence of deterioration.
Sherwood PT, 1957. The Stabilization with Cement of
Weathered and Sulphate-Bearing Clays. Géotechnique
ACKNOWLEDGMENTS 7, 179-91.
The authors express their gratitute to Institute Of
Xing H, Yang X, Xu C, Ye G, 2009. Strength
Research And Community Service of Universitas Atma
Jaya Yogyakarta for the financial support under characteristics and mechanisms of salt-rich soil–
internal funding scheme No. 079/HB- cement. Engineering Geology 103, 33-8.
PEN/LPPM/III/2018 led by Dr. Luky Handoko. Yang J, Yan N, Liu Q, Zhang Y, 2016. Laboratory test
on long-term deterioration of cement soil in seawater
REFERENCES environment. Transactions of Tianjin University 22,
132-8.
Adi Tirta B, Nugroho MA, Parahita B, Jaspers Focks
D-J, 2018. Observational Method for More Reliable Zhang RJ, Santoso AM, Tan TS, Phoon KK, 2013.
Settlement Prediction for Reclamation on the Holocene Strength of high water-content marine clay stabilized
Marine Clay Deposit in Jakarta Bay. MATEC Web by low amount of cement. Journal of Geotechnical and
Conf. 147, 07007. Geoenvironmental Engineering 139, 2170-81.
76
Analysis Of Failure Base Plate Anchor Flare Stack Foundation

and Repair Method
Sulardi
Civil Engineering Departement, Technical Faculty, Tridharma University, Balikpapan,
East Kalimantan, Indonesia
Sulardikm61@yahoo.com
ABSTRACT
The problem faced is the sub standard and unsafe condition on the flare stack equipment of PT.Pertamina Refinery Unit V (RU
V) Balikpapan. The problem is the removal of base plate bolts due to base plate flare stack deformation and tenuous. This
study aims to describe the root causes of problems and methods of improvement. Research method with case study approach
method and is success story improvement of sub standard condition and unsafe condition on base plate of flare stack
equipment. The method of causal factor analysis was done by the fish bone method, while the root cause determination was
done by the five why's method. The repair results show after the repair and returned in the condition of safe condition and
conditions according to the standard, the flare stack can be re-operated properly and safely. With the results of these
improvements also improved methods used can be replication to overcome similar problems elsewhere.
Keywords: Base plate, anchor bolt, flare stack.
1 INTRODUCTION safe riser flare stack structure. Each guy wire is

mounted three positions by forming a 120º angle and
1.1 Problems attached to a reinforced concrete foundation
One of the most important equipment to support (deadman) with a pile. The riser stack is a 66 inch
refinery operations is the flare stack tower. Flare has a diameter steel pipe and at 42 meters the dimension is
function to burn offspecification gas (Off gas) from reduced to 48 Inch. Base plare riser flare stack is a
the processing at the refinery so as not to pollute the 14mm round thick steel plate with a width of 10 cm
air environment. The operation of the flare stack as above the steel pipe and supported by a sturdy
described above is in compliance with the foundation with reinforced concrete structure and
Government Regulation (PP) Regulation No. 11 Year reinforced with pile. Base plate bond to concrete
1973 Chapter 21 Article 39 paragraph (2) is prohibited foundation with 1 inch diameter armor bolts.
to remove toxic gases and toxic materials into the air
(3) Disposal of gases and other materials into the air
through the chimney combustion shall not contain 1
certain substances exceeding the specified levels, and
(4) flammable and unused gases when disposed of in 2
the air shall be burned. The urgency of the existence
of the flare stack at the refinery is an important tool in
supporting the environmental conservation process.
Consequently, if flare stack operations are disrupted
then refinery operations should be stopped with the
impact of reduced fuel oil production and fuel
3
shortages in the central and eastern parts of Indonesia.
In general the construction of the flare stack consists

Figure 1. Flare stack Balikpapan II Refinery
of three main parts: the riser stack (1), the guy wire
rope (2) and the riser stack (3) foundation as shown in
Figure 1. The three components are an inseparable The condition of the flare stack was discontinued from
unity. If one of the components of the equipment is its operation due to the finding of sub standard
damaged then it must be repaired or replaced condition and unsafe condition after the disturbance of
immediately. Guy wire is 32 mm in diameter and instrument control equipment. Sub standard and
mounted in three positions at 40 meters, 71 meters and unsafe condition conditions are deformation base plate
104 meters respectively so as to support the stable and
78
foundation riser stack on the north side resulting the ultrasonic ultrasonic concrete foundation results
base tilts (open) as high as 4 Cm and resulted also 4 showed ultrasonic pulses. 3.43 km / s with good
(four) bolt armature broken and pulled from its quality concrete category. The results of the length /
position and 2 (two) bolt of armature condition has condition of anchorage length is known to be 376 mm
been graduated as shown in Figure 2. Under the in length, embedded 300 mm in concrete stack flare
conditions described above, based on the results of the foundation, where 4 Pcs anchorage broke, 2 angkur
engineering inspection of PT.Pertamina RU V mulur. From the above condition analysis it is
Engineering Team in cooperation with the predicted that the cause of deformation of base plate
Construction Research Institute for Research and of flare stack riser is due to the effect of excess
Application of Technology (LUK-BPPT) stated that pressure causing the transfer pipeline to be pushed
the Balikpapan II Flare Stack Equipment in the towards the south and pushing the riser flare stack and
Condition grade Damaged , Sub standard and Unsafe causing the riser flare stack tilted (Figure 3). The
condition. The results of the inspection techniques boost of the transfer line pipe resulted in the rising
also recommend that the flare stack should be stopped flare stack tilted and lifted the north side plate base
from its operation and repaired so that the flare stack resulting in a 4 Pcs rupture of the anchors bolt and the
can function again as before. 2 Pcs bolt of the base plate bolt.
Baut angker mulur
Baut angker putus
Figure 3 Tail effect due to excess pressure
Figure 2. Deformation of the base plate and the removal of 1.3 Research objectives
an armature thread The objective of the study was to provide an overview
of the method of improvement of the base plate of the
The problem faced is not yet known exactly the cause foundation and the replacement of base plate binding
of deformation of base plate riser flare stack and the armature. The study was limited to the analysis of
absence of suitable repair method and appropriate use. causes of base plate deformation problems and
Bearing in mind that repair work must be done with method of improvement on the base plate and the
high accuracy and method accuracy to avoid the risk method of replacing armature-binding base plate base
of failure and workplace accidents. of riser flare stack. Methods of improving the position
of the transfer line pipeline and the setting up of the
1.2 Cause of Problem wire rope buffer for the riser flare stack are not
To determine the cause of the problem is done visual discussed in this study because they have been
inspection with survey result and visual inspection of handled by other research teams.
6 anchor bolt of base flange flare holder, pushed riser
flare stack by 66 Inch line transfer line toward south, The assumptions developed in this research are
shifted line transfer line 66 Inch to south as far as 30 knowing the underlying cause of the base plate
cm and slope on the riser flare stack followed by the deformation and the removal of the base plate flare
saggy southern guy wire and the rising voltage of the stack bolts due to the tail process effect of excessive
guy wire in the opposite direction. The results of the pressure resulting in deformation of the base palte, the
male wire rope diameter measurement showed no breaking and the extension of the armature bolts
significant diameter reduction. The result of binder base then the focus of the method of
examination of foundation concrete on the surface did improvement is how the method of replacing base
not appear any signs of damage and hammer test plate and bolts armor base plate binding. If the repair
result showed concrete compressive strength. 391 - method is known then the problem can be solved. For
515 kg / cm2 with good quality of concrete. The
79
this reason, this research is conducted in order to find

the answer to the problems faced by Pertamina RU V.
2. METHODS
This study was conducted as a follow-up to the

findings of the condition of damage, unsafe condition
and sub standard on base plate and armature-tilt
binder base plare riser flare stack. The research
method using case study method of base plate
foundation damage and removal of armature bundle
base plate riser flare stack. This paper is a report of
succes story of appropriate technology application for
method of base plate repair and replacement of
binding armature base binder riser flare stack which
has been done on turn around and is the first method
used in environment PT.Pertamina. The base plate
repair work and the replacement of the armature bolts
can only be done on total black out (TBO) condition
where all refinery operating units are not operated
except the power plant unit and its supporting utility
unit.
Figure 4. The condition of riser flare
2.1 Analysis of Fenite Element Methods (FEM) stack after driven and FEM results
Against the shifting conditions of the NPS 66 Inch The results of a series of tests conducted include
transfer line pipeline, the riser flare stack was pushed visual, non destructive test (NDT) and analysis with
as far as 30 cm towards the south and the armature Fenite Element Method (FEM) with the following
base plate of riser flare stack was extracted and the results:
stretch was analyzed by Fenite Element Method a. The north side plate base is raised as high as 3 cm
(FEM) method. From the FEM analysis (Figure 4) and the pulling of 4 anchor bolts and the 2 anchor
there is a voltage distribution on the outer side of the bolt flume base flange holders.
flare plate structure driven 600 mm by the transfer b. Pushing flares by 66 Inch line transfer line towards
line pipe and then back again but the NPS 66 Inch line the south.
transfer line can not return to its original position and c. Shifting line transfer line 66 Inch towards the south
still shift as far as 30 cm to the south. as far as 30 cm
d. There is a slope in the flare and followed by the
saggy southern guy wire and the rising voltage of
the guy wire in the opposite direction.
e. The result of flare stack structure analysis using
FEM shows that the stress distribution is
concentrated at the bottom of the flare structure on
the south side.
f. The southern flare structure, the stress distribution
that occurs on the inner side of the plate is greater
than the voltage occurring on the outer side of the
plate.
2.2 Material
a. Plate thick 16 mm, dimensions as needed
b. Anchor bolt, type unhead continusous threat
adhessive bonded
c. Other materials as needed on site
2.3 Equipment
a. Concrete drilling machine
80
b. Concrete construction work tools r. Conduct base plate preparation and skirt plate riser
c. Steel construction work tool flare stack that has been reinforced using power
d. Metal cutting and welding tools tools so that free of corrosion
e. Safety equipment and personal protective s. Perform base layer coating and skirt riser flare
equipment (PPE) stack with mastic tolerance coating specification
f. Other auxiliary work tools as needed in the field. with min thickness. 300 micron DFT.
t. The repair work is done, the flare stack is ready to
2.4 Method of Improvement be operated again.
a. Repositioning of NPS 66 Inch line transfer has
been completed 3. RESULT
b. The repositioning of the verticality riser flare stack
has been completed with a verticality of 1/500 L Deformation of base plate riser flare stack north side,
c. Setting guy wire rope has finished the removal of 4 Pcs base arnchor bolt and 2 Pcs
d. Base plate riser flare stack has seatle in position stretching bolts base bundle anchor believed to be due
e. Attach the balancing guy wire to the elevation to the impulse of flare stack riser at elevation of 11.5
position of 11.5 meters as much as 4 Pcs and tie it meters. The cause of the rise of the flare stack riser is
to the reinforced concrete deadman mounted firmly due to the shift of the line transfer line of the NPS 66
and securely Inch line which is predicted as far as 60 Cm and
f. Prepare the armor bolts as much as 8 Pcs him. 1.1 / shifted back as far as 30 Cm so that the position of the
4 Inch with type unhead continusous threat transfer line pipeline NPS 66 Inch shifted at position
adhessive bonded with length 40 cm 30 Cm towards the south. Structural analysis using the
g. Tests have been performed on the type of unhed Fenite Element Method (FEM) was due to the stress
continuous therat adhessive bonded armor and distribution on the outer side of the flare plate
tensile tests have been carried out with a crane structure due to being pushed as far as 30 cm towards
pulling capacity of 50 tons, damage occurred in the the south and followed by a riser flasre stack, riser
concrete with crack indication flare stack deformation and the removal of base plate
h. Prepare replacement base plate with thick steel base anchors.
plate material specifications. 16 mm, shape and
configuration adjust to existing base Result of base plate repair base and replacement of
i. Base the base in the skirt plate by installing a base plate binding with result as follows:
continuous unhindered therat adhessive bonded 6 a. Examined compressive strength concrete of
Pcs, tie it well and firmly existing foundation with compressive strength of
j. Cut the base plate binder armor bolts surface concrete. 391-515 Kg/Cm2, condition of
k. Cut the base plate with an angled cut shape foundation concrete in good category
(groove 60º) to facilitate the connection with a b. Concrete improvements have been made using
replacement base plate chipping methods against spalling and carbonated
l. Clean the surface of the base plate base concrete, concrete surfaces with microconcrete material
broken parts / spalling fix with microconcrete so specifications with a compressive strength of 650
that the surface is flat Kg /Cm2.
m. Install replacement base plate and weld tig at 8
point (mounting position of successor anchor bolt)
n. Drill plate base plate dia drill diameter 1.1 / 2 Inch
with 42 Cm drill depth penetrate into the
foundation concrete core
o. Clean inner diamater drill holes with a swivel
brush and spray with pressurized wind so that it is
clean to dirt
p. Attach the continuous threat adhessive bonded
bumps of 8 Pcs and allow chemical bonded to
harden for 24 hours (as recommended by
manufacturer)
q. Weld the base connection of the existing base with
a replacement base plate and at the same time Figure 5. The condition of base plate and skirt plate
binding the bolts of the armature tie so that it is riser flare stack after repair
really tight with evenly
81
Method of repairing base plate riser flare stack and Government Regulation (PP) No. 11 of 1979
method of replacement of base plate binding anchors concerning Work Safety on Oil and Gas Purification
with type of unhed continuous therat adhessive and Processing.
bonded that has been successfully applied to
overcome problems in PT. Pertamina RU V PKM Ires 1907, 2015, Method of Reposition Pipe
Balikpapan has been standardized and registered as a Transfer Line F-1-01AB, PT. Pertamina RU V,
standard work procedure (TKI) with No.C-001 / Balikpapan.
E15142 / 2017-S9 Rev.2 and this standard working
method has also been replicated to conduct similar Sulardi, 2016, Overcoming Adversity Installation
problems in PT.Pertamina other units. Base Plate Equipment Equipment Installation With
Unhead Continuous Threaded Adhessive Bonded
Other improvements result as follows: Method, Portal Komet,
a. Has been done reposition pipeline transfer line http://ptmpeab81.pertamina.com/komet/searchResult.
NPS 66 Inch with roll method of rolling and aspx?.ptm Codefication No. 16092004
withdrawal with tool block block capacity of 50
tons so that the transfer line pipe sat well in the Sulardi, 2016, Method of Repair Skirt Plate Riser
original position above saddle support Flare Stack With Doubling Plate Method, Portal
b. Resetting verticality riser flare stack has been done Komet, http://ptmpeab81.pertamina.com/komet/
by setting back buckle turn position with searchResult.aspx?.ptm Codefication No. AC
verticality position of riser flare stack is 1/625 0001/0002
<1/500 L (API standard)
c. Skirt plate riser flare stack structure has been done Sulardi, 2014, Guy Wire Flare Stack V-19-01
by doubling plate method with the shape, Installation Method With Roll Shear and Air Winch In
dimension and configuration adjust the existing Roller In Pertamina RU V Balikpapan, Portal Komet
skirt plate conditions. http://ptmpeab81.pertamina.com/komet/searchResult.
aspx ? .ptm Codefication No. AB 9999
4. CONCLUSION
Sulardi, 2015, Transfer Reposition Method NPS 66
Based on the description of the problem and visual Flare Stack V-19-01 With Roll Shear Method In
examination results, non destructive test and Fenite Pertamina RU V Balikpapan, Portal Komet,
Element Method (FEM) analysis can be concluded: http://ptmpeab81.pertamina.com/komet/searchResult.
a. The cause of the shift in the NPS 66 Inch line aspx ?. ptm Codefication No. AC 0003/0004
transfer line resulted in the deformation of the base
on the north side and the removal of the base plate TKI No. C-001 / E15142 / 2014-S9 Rev.2, 2014,
riser stack is the tail effect due to excess pressure Transfer Line Pipe Reposition Method with Shear Roll
on the internal pipeline transfer line and Withdrawal Method, PT. Pertamina RU V,
b. Repair of base plate riser flare stack by replacing Balikpapan.
with new base using thick steel plate 16 mm and
replacement base bolts with unhold continuous
therat adhessive bonded type proved suitable,
suitable and safe flare stack re-operated without
accident (zero incident).
5. REFERENCE
API Standard 521, 2014, Pressure-Relieving and

Depressuring Systems, Six Edition, New York, USA
Document No.1569-001, Job No. 1569, 1996, Data

Sheet of Flare Stack, Balikpapan Expansion Program,
PT.Pertamina UP V, Balikpapan
LUK-BPPT, 2012, Assessment Report Guyed Support

Flare Stack V-19-01, BPPT, Jakarta
82
THE ANALYSIS ON ROCKBOLT REINFORCEMENT ON THE TUNNEL
BY APPROACH OF MOHR COULOMB MODEL AND HARDENING SOIL MODEL
(Case Study: The double - track railway Tunnel, Notog, Purwokerto, Banyumas
Regency, Central Java)
ABSTRACT
Hanindya K Artati., ST., MT1

Dias Dwi Hatmoko2
The increasing population has led to the need for an alternative transportation such as
the development in railway sector. One of the forms of railway development from PT. KAI
(Indonesia Railway) is the development of Double Track Railway viewed as the most
effective and efficient alternative. The railway lane in the hilly area uses a tunnel with the
purpose to prevent many spatial changes. Considering the lane passing through the tunnel,
then, the sufficient stability of tunnel is deemed necessary. For this, the development of
technology in infrastructure in the form of the reinforcement additionsto achieve the stability
as planned is needed. One of the reinforcement used is wire meshand rock bolt. This paper
will discuss about the analysis on the rock bolt reinforcement in tunnel using Mohr Coulomb
Model and Hardening Soil Model as the alternative solution in the double-track railway.
In this discussion the data were derived from case studies taken at tunnel point B3 on
double track railway, Notog, Purwokerto, Banyumas District in Central Java Province.The
analysis was performed by approaching Mohr-Coulomb and Hardening Soil models by using
finite element method on plaxis.The alternative use of wiremesh and rockbolt variations as
the reinforcement makesthe analysis comprehensiveto obtain an effective approach to
retrofitting the tunnel, especially in Purwokerto area.
The results of the analysis showed an increaseof the level of Tunnel's Safety Factor.
The safety factorin the existing condition withoutany reinforcement was equal to 2,3513
while the one with reinforcement of wiremesh and variation of rockbolts 1 safety factor
increased to 2,610 while with the variation of RB3 and RB 5 respectively reached 4,3022 and
4 , 6473.
Keywords: Rockbolt, Tunnel, Mohr Coulomb Model, Harderning Soil Model, Deformation,
Safety Factor.
83
THE ANALYSIS ON ROCKBOLT REINFORCEMENT ON THE TUNNEL
BY APPROACH OF MOHR COULOMB MODEL AND HARDENING SOIL MODEL
(Case Study: The double - track railway Tunnel, Notog, Purwokerto, Banyumas
Regency, Central Java)
Hanindya K Artati., ST., MT1

Dias Dwi Hatmoko2
BACKGROUND
The rapid population growth in Special District of Yogyakarta and Central Java
becomes a challenge in transportation services. Itis noted that the annual rate of growing
population is approximately 0,81% . Related to the higher growing population and economic,
the alternative transportation services are required to be reliable and to take a crucial
roletherefore the fast, effective, and efficient transportations are needed, one of which is the
development of New Yogyakarta International Airport (NYIA) in Kulonprogo. The
development of NYIA is also supported by railway transportation facility connecting the city
points surrounding the Kulonprogo Regency including the railway route of Jogja-
Purwokerto.This alternative transportation in turn has created new motivation and challenge
for the railway development not to disturb the existing settlement area purposely to minimize
the spatial change. The advanced technology in railway infrastructure therefore is deemed
necessary in the development of double tract railway track passing through the tunnels in
Purwokerto hills, Central Java. This paper focuses on tunnel at profile cut B3 likes in figure
1, the tunnel analysis on rock bolt reinforcement using Mohr Coulomb model and Hardening
Soil model as the alternative solution in track double track railway.
Figure 1. Profile Cut B-3
84
Tunnel And Rock bolt
The measurement of the tunnel stability, as stated by Hoek and Brown (1980) is essential to
identify to what extent the load bearing is in comparison to the stones that bear it on a tunnel.
The reinforcement of stone restraint commonly uses the rockbolt as shown in figure below. It
can be explained that the safety factor is the restraint comparison (τ) towards the load
working (σ or w) in which the restraint is dependent upon the number of bolts (n) and tensile
strength/rockbolt.
𝜏 𝑛𝑥𝑐
𝐹𝐾 = σ = 𝑊
where
• FK > 1,0 tunnel in secure state
• FK = 1,0 critical tunnel
• FK< 1,0tunnel in insecure state
In detail, based upon SNI the Safety number can be categorized as follows Table 1:
Table 1. The Estimation of Safety Factor (SF)
Safety Factor Vulnerability of Soil Movement

≤ 1.2 Height: Soil Movement frequently occurs
1.2 < Safety Factor ≤ 1.7 Medium: Soil Movement probably occurs
1.7 < Safety Factor ≤ 2.0 Low: Soil Movement seldom occurs
>2.0 Very Low: Soil Movement rarely occurs
Source: Ward. R. 1978 in SNI 13-7124-2005, 2005
Mohr Coulomb Model
The basic principle of Mohr Coulomb is the elastic-perfectly plastic in which the stretch and
the change of stretch occurred is divided into two parts: elastic and plastic. See in Figure 2,
85
Figure 3 Basic Concept of Elastic-Perfectly Plastic of Mohr Coulomb
The parameters of the function of the plastic are the value of cohesion (c) and shear angle (ϕ)
in which the functions of the melts formed hexagonal conus in its main strain. Meanwhile,
the potential function of other plastics is the dilatation angle dilation (ψ ) that is used to
model the addition of volumetric tensile of positive plastic in detailed on the solid soil.
Figure 4 Melting Field of Mohr Coulombin Main Shear Room (c=0)
Hardening Soil Model
Hardening soil model is different from other elastic model. This model is more elastoplasticin
which there is no any dependency between the stiffness in Triaxialand stiffness in the
odometer test.
86
ref
Figure 5 Definition ofE50 from the result of the odometer test
In the figure below, it can be seen that the melting field due to the shear and cap formed a
hexagonal field in the main strain from the fall of Mohr coulomb. For the melting field
caused by the shear, then it would increase to later on approach the line of fall of Mohr
coulomb. Meanwhile, in “cap” it developed dependent upon the pre-consolidation load
working.
Figure 6 Yield surfaces of hardening soil model in p’-q plane
Figure 7 The representation of total yield contour of hardening soil model in principal stress
space for cohesionless soil
Earthquake
87
Earthquake Distribution Analysis was done based upon the Map of Bedrock
Acceleration Region in Short Period (Ss)in which the spectrum value of earthquake response
(C)was based on SNI-1726-2012.To obtain the value of spectrum response, then it needs to
first identify the parameter of mapped acceleration including the bedrock acceleration in short
period (Ss) and bedrock acceleration in 1-seond period (S1) as seen in figure below.
Figure 8 Map for the Bedrock Acceleration Region in Short Period (Ss)
(Source: National Standardization Board SNI, 1726-2013)
Figure 9 Map of Bedrock Acceleration Region in 1-Second Period (S1)

(Source: National Standardization Board SNI, 1726-2013)
METHODOLOGY
The methodology used included:
1. Primary and secondary data collection. The primary data was obtained from the
laboratory test.
2. Literature study to the previous researches aimed to obtain the enrichment in writing
this paper
3. Analysis on the data that would be used in modelling
4. Modeling the construction phase that would be done without any reinforcement. The
soil model was done using Mohr Coulomb Model and Hardening Soil Model.
5. Analysis on the condition stability of the existing tunnel condition
88
6. Analysis on the reinforcement for the tunnel stability using wire meshand alternative
rockbolt (1, 3, and 5).
7. Result of analysis to draw conclusion between the Mohr Coulomb model and
Hardening Soil model.
ANALYSIS AND RESULT
Analysis Mohr Coulomb Model
The analysis was done started from the use of the element method to the use of plaxis
software. The result obtained from the Mohr Column model is presented as follows.
Figure 10 Result of total displacement without reinforcement of tunnel using

mohr coulomb model
Figure 11 Result of total displacement in alternative reinforcement wiremesh+RB5

of tunnel using mohr coulomb model
From the analysis on the tunnel using Mohr coulomb model, it has been found the result of
safety factor at excavation stage construction is 2,353. And the result of safety factor with
89
wiremesh reinforcement is 2,428, and with added alternative rockbolt shown RB1 is 3,0217 ;
RB3 = 4,303; and RB5 = 4,647.
Meanwhile, for the displacement in the condition shown without reinforcement 0,02065 m;
with wiremesh 0,01831 m; and with alternative rockbolt in RB1 displacement shown 0,01709
m; RB 3 = 0,01551 m; and RB 5 = 0,01413 m.
Analysis Harderning Soil Model
The analysis was done started from the use of the element method to the use of plaxis
software. The result obtained from the Hardening Soil model is presented as follows.
Figure 12 Result of total displacement without reinforcement of tunnel using

hardening soil model
Figure 13 Result of total displacement in alternative reinforcement wiremesh+RB5

of tunnel using hardening soil model
From the analysis on the tunnel using hardening soil model, it has been found the result of
safety factor without reincforcement is 2,3378. And the result of safety factor with wiremesh
90
reinforcement is 2,3017, and with added alternative rockbolt shown RB1 is 2,4261 ; RB3 =
3,8436; and RB5 = 4,653.
Meanwhile, for the displacement in the condition shown without reinforcement 0,02901 m;
with wiremesh 0,02114 m; and with alternative rockbolt in RB1 displacement shown
0,01773m; RB 3 = 0,01537 m; and RB 5 = 0,0146 m.
Result of Mohr Coulomb Model and Harderning Soil Model approach
From the analysis results obtained can be seen in the following graph between mohr coulomb
model and hardening soil model
Table 2. Result of safety factor using mohr coulomb model and hardening soil model
Figure 14 Grapf of safety factor using mohr coulomb model and hardening soil model
While the displacement shown in figure 15:
91
Table 3. Result of displacement using mohr coulomb model and hardening soil model
Figure 15 Grapf of safety factor using mohr coulomb model and hardening soil model
CONCLUSION
From the analysis on the tunnel using Mohr coulomb model, it has been found the
result of safety factor without any reinforcement at 2,352, and after giving the reinforcement,
the wire mesh safety factor experienced an increase to be2,428 With the alternative of the
addition of rockbolt RB1, RB3, and RB5 safety factor respectively experienced the increase
of 3,0217; 4,303; 4,647. Meanwhile, for the displacement in the condition without any
reinforcement showed the value of 0,02065 and with the wiremesh reinforcement is 0,01831;
and with addition alternative rockbolt of RB1, RB2, and RB 3 respectively showed 0,01709;
0,01551; and 0,01413 Thus, it can be concluded that the use of analysis on the finite element
of Mohr Coulomb model, the tunnel experienced the increase of safety factor level and the
decrease in the soil deformation surrounding the tunnel.
For the hardening soil model, it can be seen that safety factor without any
reinforcement was at 2,3378, and after given the reinforcement of wire mesh, safety factor
experienced an increase to be 2,3017. Given the alternative of addition of rock bolt RB1,
RB3, and RB5 the safety factor respectively experienced the increase of 2,4261; 3,8436; and
4,653 For the displacement surrounding the tunnel, in the condition without any
reinforcement the level was at 0,02901 and when given the addition wiremesh is 0,02114 of
92
rock bolt RB1, RB2, and RB3 respectively was at 0,01773; 0,01537, and 0,0146. Thus, it can
be concluded that using the analysis of finite element,hardening soil model on tunnel
experienced an increase of safety factor and the decrease in the soil deformation surrounding
the tunnel. But the value of the safety factor is still smaller than the mohr coulomb method.
As for the displacement value is still greater than the mohr coulomb method as shown in the
figure 14 and 15.
In the analysis on the value in the Mohr coulomb model and hardening soil model,
though not showing a significant difference, it can be seen that the value of hardeningsoil
model was lower than that of Mohr Coulomb Model. This was in relation to that the analysis
with hardening soil model, the measurement considered the soil stiffness for the primary
𝑟𝑒𝑓
load, in which the value of 𝐸𝑜𝑒𝑑 (tangent stiffness for primary odometer loading) approaching
𝑟𝑒𝑓 𝑟𝑒𝑓
to the existing field condition. In addition, there were the parameters of𝐸50 and 𝐸𝑢𝑟 in which
𝑟𝑒𝑓 𝑟𝑒𝑓
𝐸𝑢𝑟 = 3. 𝐸50 . The measurement of stiffness was assumed in the unloading or initial
condition. This then made the result of the analysis on hardening soil more approached to the
carefulness more/critical condition of tunnel stability.
REFERENCES
Brinkgreve, R.J.B. 2002.Reference Manual V.8 Plaxis (manual Plaxis). A. A Balkema.

Netherlands.
Das, Braja M. 1995. Mekanika Tanah (Prinsip-Prinsip Rekayasa Geoteknis).
Erlangga. Surabaya.
Hudson, John A. 1997. Rock Mechanics. London: Pergamon
Minister of Public Work and Housing.(2015). Metode Perencanaan Penggalian dan Sistem
Perkuatan Terowongan Jalan Pada Media Campuran Tanah-Batuan. Jakarta:
Ministry of Public Work and Housing
93
Effect of Matric Suction Change on Pile Foundation Capacity

in Unsaturated Soils
H. Pujiastuti
Doctoral Candidate at Department of Civil and Environmental Engineering, Universitas Gadjah Mada, Yogyakarta,
INDONESIA
heni.pujiastuti@ugm.ac.id
A. Rifa’i
ahmad.rifai@ugm.ac.id
A. D. Adi
adhadhi2@yahoo.com
T. F. Fathani1,2
1
2
Center for Disaster Mitigation and Technological Innovation (GAMA-InaTEK), Universitas Gadjah Mada, Yogyakarta
tfathani@ugm.ac.id
ABSTRACT
The capacity of pile foundation driven into unsaturated soils is controlled by undrained cohesion. The hydrological process and
seasonal changes lead to the changes in the degree of saturation, matric suction, and cohesion of the soil. This research analyzes
the influence of the change of undrained cohesion due to the change of matric suction to the pile capacity on unsaturated sandy
clay. The calculation of pile capacity is also applied for unsaturated kaolinite clay as a case study. The evaluation is then
conducted on these two soil types i.e. unsaturated kaolinite clay and unsaturated sandy clay. Matric suction is measured using
the filter paper method, whereas the undrained cohesion is obtained from a laboratory bearing test for unsaturated kaolinite clay
and unconsolidated-undrained triaxial test for unsaturated sandy clay. The calculation of pile capacity was simulated on the pile
foundation with 0.4 m in diameter, 5 to20 m in length, with the matric suction varied from 10 to 1000 kPa. The results show that
the skin friction component of kaolinite clay increased significantly more than sandy clay, while the end bearing component of
sandy clay increased significantly more than kaolinite clay. The total pile capacity increased non-linearly with the increasing of
matric suction. The total pile capacity increased significantly until the matric suction of 50 kPa, furthermore the curve of the
relationship between total pile capacity and matric suction begins to curve on the matric suction of 200 kPa. The total pile
capacity driven into unsaturated sandy clay are influenced by both the cohesion and internal friction of the soil.
Keywords: unsaturated clay, undrained cohesion, filter paper, skin friction, end bearing
1 INTRODUCTION bearing and total pile capacity) that is driven into

unsaturated soils, is controlled by the undrained
The hydrological processes including the precipitation, cohesion (Uchaipichat, 2012), whereas unsaturated
evaporation, evapotranspiration, and the seasoning cohesion is controlled by matric suction (Pujiastuti et
change from wet season to dry season, resulting in the al., 2018).
decrease of groundwater level. This phenomenon may
cause the decrease of the degree of saturation and the Several studies have been conducted to calculate the
increase of matric suction of the soil above pile capacity driven into unsaturated soils. Fattah et al.
groundwater level. (2014) used the numerical methods, while Vanapalli
and Taylan (2012) analyzed the equation and then
The pile foundation is generally used to support the validated the pile capacity by laboratory testing.
building or transmit the upper structural load to the Furthermore Chung and Yang (2014) used the
hard soil that is very deep (Gaaver, 2013). The soil laboratory test and then verified the calculation of pile
which is located above the groundwater level is known capacity by finite element method. Carvalho and Rocha
as unsaturated soil (Fredlund and Rahardjo, 1993). The (2013) implemented several theories and validated the
pile foundation capacity (includes skin friction, end results by the field load test.
94
The behavior of single pile on unsaturated clay by where 𝑄𝑠 is the ultimate skin friction capacity of the
incorporating the finite element method was pile, 𝑆𝑢 is undrained shear strength, 𝐴𝑠 is the area of
investigated by Fattah et al. (2014). The soil parameters shaft in contact with the soil , and 𝛼 is the adhesion
were taken from the laboratory testing. The factor. American Petroleum Institute (API, 1984)
measurement of matric suction was conducted using suggests the relationship between 𝛼 dan the undrained
the filter paper method. The data fitting was evaluared cohesion (𝑐𝑢 ) as expressed in Equation (2). This
using the Soil Vision to describe Soil Water formula can be used as an alternative to estimate the
Characteristic Curve (SWCC) and H-modulus adhesion factor (𝛼).
function. Clay soil samples were taken from three
locations in the city of Baghdad, Iraq. The results
 = 1 for cu < 25 kPa
showed that the proposed method to define H-modulus
 cu − 25kPa 
obtained a satisfactory result on the analytical  = 1 − 0,5  for 25 kPa< cu < 75 kPa
procedure of pile foundation capacity in unsaturated  50kPa 
soil.
 = 0.5 for cu > 75 kPa (2)
The skin friction capacity of a small scale single pile on
compacted unsaturated clay due to static axial load was where 𝑐𝑢 is similar to 𝑆𝑢 . The contribution of undrained
evaluated by Chung and Yang (2014) using finite cohesion to the ultimate end bearing capacity of the pile
element approach. The disturbed laterite soil samples expressed in Equation (3).
were taken from 1 m below the ground level on the
hillside of Hsinchu in northern Taiwan. It was found Qb = S u N c Ab (3)
that the loading behavior of a small-scale single pile on
unsaturated soil indicated a nonlinear response. The where 𝑄𝑏 is the ultimate end bearing capacity, 𝐴𝑏 is
ultimate skin friction capacity of the pile has decreased cross sectional area, 𝑁𝑐 is bearing capacity factor
from 52 % to 5 % when the water content was increased (equal to 9 for pile).
from 15 % to 21 %.
2.2 The Contribution of Undrained Cohesion and
In this paper, laboratory tests were conducted to obtain Internal Friction Angle
the soil properties. The filter paper contact method was
used to measure the matric suction, meanwhile the For the soils having both the undrained cohesion and
shear strength of unsaturated sandy clay was examined internal friction angle, the ultimate skin friction
based on a series of unconsolidated-undrained triaxial capacity of the pile considers the contribution of
test in order to obtain the relationship between the undrained cohesion and internal friction angle, as
cohesion, internal friction angle, and matric suction of expressed in Equation (4).
the soil. A laboratory bearing test was performed to
obtain the relationship between undrained shear Qs = (cu As ) + ( K s p0 tan As ) (4)
strength and matric suction for unsaturated kaolinite
clay. The relationship between shear strength
where 𝐾𝑠 is a coefficient of horizontal soil stress
parameters and matric suction was used as the input to
depends on soil condition, 𝑝̅𝑜 is the average of effective
the formula for calculating the ultimate pile capacity.
overburden pressure at pile base, 𝛿 is angle of interface
along the pile and soil. The contribution of undrained
2 DETERMINING THE ULTIMATE TOTAL PILE cohesion and internal friction angle to the ultimate end
CAPACITY bearing capacity of the pile is expressed in Equation
Ultimate total pile capacity is calculated from the sum (5).
of ultimate skin friction and ultimate end bearing
capacity, which are determined as follows. Qb = Ab (1.3cu N c + Pb' N q + 0.3dN  ) (5)
2.1 The Contribution of Undrained Cohesion
For the cohesive soil having no internal friction, the where 𝑃′𝑏 is the effective vertical pressure at pile base,
ultimate skin friction capacity of pile foundation is 𝑁𝑐 , 𝑁𝑞 , 𝑁𝛾 is bearing capacity factor,  is unit weight of
calculated based on the contribution of undrained soil, d is pile diameter.
cohesion expressed in Equation (1).
Qs = S u As (1)
95
3 EXPERIMENTAL STUDY between the undrained shear strength (Su) and matric
suction (s), which can be expressed by Equation (6).
3.1 Soil Properties
In this study, the ultimate pile foundation capacity is S u = 43.026s 0.3488 (6)
estimated in two soil types, i.e. unsaturated kaolinite
clay and unsaturated sandy clay. A series of laboratory The undrained cohesion in unsaturated sandy clay is
tests on kaolinite clay have been carried out by obtained by UU Triaxial Test conducted by Pujiastuti
Uchaipichat and Man-koksung (2011). Meanwhile, a et al. (2018). The result shows that the relationship
series of laboratory tests on unsaturated sandy clay between undrained cohesion (𝑐𝑢 ) and matric suction
have been carried out by Pujiastuti et al. (2018). The (s) can be expressed by Equation (7) and Figure 1.
results of the laboratory tests on both type of soils are
shown in Table 1. cu = 23.767s 0.1438 (7)
Table 1. The properties of kaolinite clay and sandy clay 80
Undrained cohesion (kPa)

70
Properties Kaolinite clay Sandy clay
Specific Gravity, Gs 2.72 2.62 60
Liquid Limit, LL (%) 52 39.83
50
Plastic Limit, PL (%) 31 24.07
Maximum Dry Density, 14.1 15.1 40
MDD (kN/m3) 30
Optimum Moisture 27.5 23
Content, OMC (%) 20
10
3.2. Calculation of the Pile Capacity 0
A circular pile foundation with a diameter of 0.4 m, is 0.1 1 10 100 1000 10000
Matric suction (kPa)
driven into unsaturated kaolinite clay and unsaturated
sandy clay. The pile length variations are 5 m, 10 m, 15 Figure 1. The relationship between undrained cohesion
m, and 20 m. The variations of matric suction are 10 and matric suction based on laboratory test
kPa, 50 kPa, 100 kPa, 200 kPa, 400 kPa, 600 kPa, 800
kPa, and 1000 kPa. The matric suction and undrained 4.3 Internal Friction Angle
shear strength of soils is assumed to be constant
throughout the pile. The ultimate skin friction capacity The internal friction angle (𝜑) in unsaturated sandy
and the ultimate end bearing capacity of the pile are clay is obtained by UU Triaxial Test conducted by
calculated by using Equations (1) and (3) for kaolinite Pujiastuti et al. (2018). The relationship between
clay soils, whereas Equations (4) and (5) are for sandy internal friction angle (𝜑) and matric suction (s) is
clay. expressed by Equation (8) and Figure 2.
4 RESULT AND DISCUSSION  = 3.339s 0.3211 (8)
4.1 Soil Water Characteristic Curve (SWCC) 40

The Soil Water Characteristic Curve (SWCC) is a 35
curve which states the relationship between the degree 30
of saturation and matric suction of the soil. The filter
25
paper method is used to measure the matric suction on
20
𝜑
both types of soil. The unsaturated kaolinite clay has

the air entry value (AEV) of 700 kPa and the residual 15
suction of 7.000 kPa, whereas the unsaturated sandy 10
clay has the AEV of 21.5 kPa and residual suction of
5
70.000 kPa.
0
4.2 Undrained Cohesion 0.1 1 10 100 1000 10000
The undrained shear strength in an unsaturated clay
was obtained from back analysis base on the laboratory
Figure 2. The relationship between internal friction
bearing tests conducted by Uchaipichat and Man-
angle and matric suction based on laboratory test.
koksung (2011). The result provides the relationship
96
Furthermore, the relationship between tangent of 2500

Skin friction (clay)
internal friction (tan 𝜑) with matric suction (s) can be
Ultimate pile capacity (kN)

End bearing (clay)
expressed by Equation (9) and Figure 3. 2000 Total pile capacity (clay)
tan  = 0.0575s 0.3336 (9) 1500
0.7
1000
0.6
500
0.5
0.4 0
0 200 400 600 800 1000 1200
0.3
0.2
(a)
0.1
2500
0.0 Skin friction (sandy clay)

0.1 1 10 100 1000 10000 End bearing (sandy clay)
2000 Total pile capacity (sandy clay)
Figure 3. The relationship between tan 𝜑 and matric 1500

suction based on laboratory test.
1000
4.4 Pile Capacity
For pile foundation driven into unsaturated kaolinite 500
clay, the ultimate skin friction capacity was calculated
using Equation (1), while 𝑆𝑢 as a function of the matric 0
suction is determined by Equation (6), and α is 0 200 400 600 800 1000 1200
determined by Equation (2). The ultimate end bearing Matric suction (kPa)
capacity was calculated using Equation (3) and the
ultimate total pile capacity is the sum of the result of (b)
calculation by Equations (1) and (3).
Figure 4. Ultimate pile capacity with the variation of
For pile foundation driven into unsaturated sandy clay, matric suction on L = 5 m; (a) Kaolinite clay; and (b)
the ultimate skin friction capacity was calculated using Sandy clay.
Equation (4), while 𝑐𝑢 as a function of the matric 4000
suction is calculated by Equation (7), 𝛼 is determined Skin friction (clay)
End bearing (clay)

by Equation (2), 𝐾𝑠 is assumed to be equal to 1 Total pile capacity (clay)
(Tomlinson, 2004), 𝛿 is determined in the same way as 3000
𝜑 (Tomlinson, 2004). In this case, tan 𝛿 is a function
of matric suction and determined using Equation (9).
2000
The ultimate end bearing capacity was calculated using
Equation (5). The bearing capacity factors (𝑁𝑐 ,
𝑁𝑞, 𝑁𝛾 ) are determined by inputting 𝜑 value on the 1000
Terzaghi graph (Tomlinson, 2004). The internal
friction angle as a function of matric suction is
calculated using Equation (8). The calculation results 0
0 200 400 600 800 1000 1200
of ultimate skin friction, ultimate end bearing and
ultimate total capacity of the pile are shown in Figure
4-7.
(a)
97
4000 7000
Skin friction (sandy clay)

End bearing (sandy clay) 6000
Total pile capacity (sandy clay)
3000
5000
4000
2000
3000 Skin friction (clay)
End bearing (clay)
2000
1000 Total pile capacity (clay)
1000
0 0
0 200 400 600 800 1000 1200 0 200 400 600 800 1000 1200
Matric suction (kPa) Matric suction (kPa)
(b) (a)
Figure 5. Ultimate pile capacity with the variation of 7000

matric suction on L=10 m; (a) Kaolinite clay; and (b) 6000 End bearing (sandy clay)

Sandy clay. Total pile capacity (sandy clay)
5000
5500
5000 4000
4500
3000
4000
3500 2000
3000
1000
2500
Skin friction (clay)
2000 End bearing (clay) 0
1500 Total pile capacity (clay) 0 200 400 600 800 1000 1200
1000 Matric suction (kPa)
500
0 (b)
0 200 400 600 800 1000 1200
Matric suction (kPa) Figure 7. Ultimate pile capacity with the variation of
matric suction on L=20 m; (a) Kaolinite clay; and (b)
(a) Sandy clay.
5500
5000 Based on the calculation results on the ultimate pile
End bearing (sandy clay)

4500 capacity ( Figure 4-7), it is clarified that the ultimate
Total pile capacity (sandy clay)
4000 skin friction, the ultimate end bearing and the ultimate
3500 total pile capacity increased with the increasing of
3000 matric suction on both types of the soil. The ultimate
2500 total pile capacity and the ultimate skin friction on
2000 unsaturated kaolinite clay is higher than in unsaturated
1500
sandy clay.
1000
500 The physical behavior of the long piles (L = 15 m and
0 L = 20 m) are presented in Figure 6-7. The skin friction
0 200 400 600 800 1000 1200
is more dominant than the end bearing on both the types
of the soil. With the matric suction of 200 kPa, the skin
friction began to curve significantly on the kaolinite
(b)
clay, while on the sandy clay is rather significant. The
Figure 6. Ultimate pile capacity with the variation of end bearing capacity tends to be flat on the kaolinite
matric suction on L=15 m; (a) Kaolinite clay; and (b) clay but on the sandy clay increased rather significant.
Sandy clay. The difference between the total pile capacity and the
98
end bearing capacity is high. It is stated that the capacity is dominated by the skin friction. A significant
dominant role of skin friction a long pile capacity. contribution from undrained cohesion and internal
friction components can be found on the long piles. For
The physical behavior of the short piles (L=5 m dan the short piles with low matric suction (below 400 kPa),
L=10 m) driven into the kaolinite clay show that the the ultimate total pile capacity is dominated by the skin
skin friction is more dominant than the end bearing. friction, but in higher matric suction, the ultimate total
With the matric suction of 200 kPa, the skin friction pile capacity is dominated by the end bearing capacity
began to curve rather significantly, while the end with the dominant constribution of the internal friction
bearing capacity tends to be flat. The difference angle of the soil.
between pile capacity and end bearing capacity is rather
high. This states the role of the skin friction is stronger ACKNOWLEDGMENTS
than the end bearing. The physical behavior of the short
piles driven into sandy clay shows that on a low matric The authors would like to express their sincere
suction, the value of the skin friction tends to be higher gratitude to the Directorate General of Science and
than the value of the end bearing. When the matric Technology and Higher Education, Ministry of
suction value of 200 kPa, the skin friction tends to be Research, Technology and Higher Education, of the
flat while the end bearing value tends to increase and Republic of Indonesia for the financial support by the
exceed the skin friction value that occurs when the Education Scholarship of Postgraduate Domestic
matric suction greater than 400 kPa. The difference Programs.
between the total pile capacity and the end bearing
capacity is slightly different. This states the role of end REFERENCES
bearing is very dominant on a short pile capacity. API, 1984, Recommended practice for planning
designing and construction fixed offshore platforms,
In this research, there is a necessity to validate the 14th Edn. APIRP2A, American Petroleum Institute,
calculation results using the same soil, however this
Dallas, TX.
research compares the calculation of pile capacity on
kaolinite clay and sandy clay. In the future, an Carvalho D. and Rocha A.P.J., 2013, Uplift behavior of
experimental testing of sandy clay is required to bored piles in tropical unsaturated sandy soil,
validate the results of proposed estimation. Proceedings of The 18th International Conference on
Soil Mechanics and Geotechnical Engineering, Paris.
5 CONCLUSIONS Chung S.H. and Yang S.R., 2014, Loading behavior of
This research examines the calculation of the circular small scale single pile in unsaturated clayey soil,
pile foundation driven into unsaturated soils. The Material Research Innovation, Vo.18.
ultimate skin friction, the ultimate end bearing and the Fattah M.Y., Salim N.M and Mohsin I.M., 2014,
ultimate total pile capacity were calculated using the Behavior of single pile in unsaturated clayey soils, Eng.
equations considering the undrained cohesion for & Tech. Journal, Vol. 32, Part (A), No. 3.
kaolinite clay and incorporating the cohesion and
internal friction angles for sandy clay. Fredlund D.G. and Rahardjo H., 1993, Soil mechanics
for unsaturated soils, John Wiley & Sons, Inc., New
The undrained cohesion was obtained from laboratory York.
test using the UU Triaxial Test and the laboratory
Gaaver K.E., 2013, Uplift capacity of single piles and
bearing test. Matric suction was measured using the
pile groups embedded in cohesionless soil, Alexandria
filter paper method for determining Soil Water
Engineering Journal, 52, 365-372.
Characteristic Curve (SWCC). The relationship
between undrained cohesion and matric suction yields Pujiastuti H., Rifa’i A., Adi A.D. and Fathani T.F.,
the undrained cohesion equation as a function of the 2018, The Effect of Matric Suction on The Shear
matric suction as well as the internal friction angle of Strength of Unsaturated Sandy Clay, International
soil and matric suction. Journal of GEOMATE, Jan., 2018 Vol.14, Issue 42,
pp.112-119.
The results show that the ultimate skin friction, the
ultimate end bearing and the ultimate total pile capacity Uchaipichat A., 2012, Variation of Pile Capacity in
increased with the increasing of matric suction. In the Unsaturated Clay Layer with Suction, EJGE, Vol.17
kaolinite clay, the ultimate total pile capacity is Uchaipichat A. and Man-koksung E., 2011, Variation
dominated by the skin friction contribution from of Ultimate Bearing Capacity of Unsaturated Clay with
undrained cohesion components both on long piles and Suction, ARPN Journal of Engineering and Applied
short piles. In the sandy clay, the ultimate total pile Sciences.
99
Tomlinson M.J., 2004, Pile design and construction

practice, Fourth Edition, Taylor & Francis Group e-
library, London and New York.
100

Proceedings of ICGeotechnics

Diunggah oleh

Informasi Dokumen

Deskripsi Asli:

Hak Cipta

Format Tersedia

Bagikan dokumen Ini

Bagikan atau Tanam Dokumen

Opsi Berbagi

Apakah menurut Anda dokumen ini bermanfaat?

Apakah konten ini tidak pantas?

Hak Cipta:

Format Tersedia

Proceedings of ICGeotechnics

Diunggah oleh

Hak Cipta:

Format Tersedia

IC GEOTECHNICS YOGYAKARTA, 24-26 JULY 2018

Proceeding of International Conference on Geotechnics

Royal Ambarrukmo Plasa

Yogyakarta, July 2018

Pile Load Test Frequently Asked Questions

1 INTRODUCTION • PDA test is interpreted with one dimensional

Figure 1. Typical load-settlement curve from a pile-load test

Figure 3. Shear stress versus shear displacement on dense

2.2 Toe Resistance

Before understanding the different types of pile load

to hundreds of tons. To load a pile to that level,

3.3 Loading procedure

4.1 Test Apparatus, Instrumentation and Test

Minimum instrumentation required is dynamic

Figure 12. Photograph of a PDA test result (ASTM, 2017).

Figure 10. Schematic diagram of instrumentation for PDA

Figure 11. Photograph of a PDA test (Allnamics, 2018).

Further complicating the matter, pile capacity changes

4.3 Failures in Execution of PDA Test

5.1 Test Apparatus, Instrumentation and Installation

Figure 17. Reinforcement cage welded onto bearing plates of

6.1 Fail Pile vs Failing to Get the Intended Results

Active Faults Identification for Dam Safety against Earthquakes

1 INTRODUCTION uncontrolled spill water occurred, while on the

c) Reservoir Triggered Earthquake (RTE) is the earth topographic/landscape or bathymetric express-

A potential hazard of active faults can also be inferred

Determining activity of identified fault is, however, 3 APPLICATION OF GROUND PENETRATING

Ground Penetration Radar transmits a high frequency

An active fault identifications based on morphology is

In principle, one point shoot create one single

Figure 6. Radargram of the location exactly same with

It is clear that using 40 MHz frequency antenna as

regarding the past earthquake recorded in the trench

Akyuz et al (2014) stated that paleoseismic trenching

In Indonesia many paleoseismic trenching has been

One of the methods in paleoseismology was digging a

Figure 9. Scanning result using electric resistivity

Palaeoseismic trench was carried out on the

a) Layer 100 was the first soil layer, identified by

Jing, L. Z., Shao, Y. X., Klinger, Y., Xie, K., Yuan,

3D Assessment of Rainfall-induced Slope Movements and Risk Mitigation

1 INTRODUCTION Extensive attentions have been paid on rainfall-induced

also locates in this zone. The zone in between is

3 NUMERICAL MODELING STRATEGY

3.1 Unsaturated soil behavior and parameters

Table 3 Typical values of the parameters n and 

The critical rainfall No

Strain-based Design Analysis The geological profile was conservatively simplified to

5.3 Calculated slope movement under rainfall

Fig. 7 Particle size distribution curve of the colluvium

Fig. 10 Slope superficial movement at the end of the

Fig. 8 Estimated SWCC of the colluvium

Fig. 11 Development of slope movement at three

The slope movements at three sections (see in Fig.6)

Fig. 12 Flow path of water during the rainfall event

5.4 Risk remediation plan

In addition, a pipeline element was incorporated in the

Fig.13 showed the slope movement contour after

three-dimensional finite element analysis. Computers

Caine, N. (1980). The rainfall intensity - duration

Capparelli, G., Iaquinta, P., Iovine, G. G. R., &

Jamei, M., Guiras, H., & Olivella, S. (2015). Analysis

Lateh, H., Muhiyuddin, W. M., Taib, B., & Khan, Y.