Professional Documents
Culture Documents
NOVEMBER 2006
iii
ACKNOWLEDGEMENTS
Besides, I would like to take this opportunity to thank all lecturers and staff of
UTM, my friends especially Mr. Michael Choy (Taisei Corporation), Mr. Sridar
Krishnan (Jimah Energy Venture), Mr. Selvam and Mr. Ngui Wei Chiun and to all
whom had been involved in the success of the completion of this report.
Last but not least, I would like to convey my thanks to my beloved parents
for their guidance and support throughout my studies.
v
ABSTRACT
Due to variation in soil layers, it is not easy for engineer to be assured that
theoretical design of piles comply with the actual site condition. Thus, every design
of piled foundations carries its own uncertainty and risk. This project evaluates the
applicability of eight methods to predict the ultimate bearing capacity of spun driven
friction piles. Analyses and evaluations were conducted on four piles of different
sizes and lengths that failed during pile load testing. The load test interpretation
methods, pile driving formulae, as well as the Meyerhof method (static analysis)
were used to estimate the bearing capacities (Qp) of the investigated piles. The
failure loads were the maximum measured load carrying capacities (Qm) from pile
load test. The pile capacities determined using the different methods were compared
with the measured pile capacities obtained from pile load tests. Three criteria were
selected as basis of evaluation: the best fit line for Qp versus Qm, the arithmetic mean
and standard deviation for the ratio of Qp/Qm, and the cumulative probability for
Qp/Qm. Results of the analyses show that the best performing method is Butler and
Hoy method (load test interpretation method). This method is ranked number one
according to the mentioned criteria.
vi
ABSTRAK
TABLE OF CONTENTS
DECLARATION ii
DEDICATION iii
ACKNOWLEDGEMENT iv
ABSTRACT v
ABSTRAK vi
TABLE OF CONTENTS vii
LIST OF TABLES xi
LIST OF FIGURES xii
LIST OF SYMBOLS xiv
LIST OF APPENDICES xvi
1 INTRODUCTION 1
1.1 Background of the study 1
1.2 Objectives 2
1.3 Scope of study 3
1.4 Importance of study 4
2 LITERATURE REVIEW 5
2.1 Foundations on Problematic Soils 5
2.2 Deep Foundations 6
2.2.1 Driven Piles 7
2.2.2 Changes in Cohesive Soils 7
2.2.3 Changes in Granular Soils 8
2.3 Pile Load Testing 9
viii
3 METHODOLOGY 42
3.1 Introduction 42
3.2 Data Collection 42
3.3 Compilation of Data 43
3.3.1 Soil Data 44
3.3.2 SPT Data 44
3.3.3 Piling Records 44
3.3.4 Pile Load Tests Reports 44
3.4 Data Analysis 45
3.5 Comparison of the Results 45
3.6 Evaluation of Methods 46
3.6.1 Best Fit Line Equation 46
3.6.2 Cumulative Probability 47
3.6.3 Mean () and Standard Deviation ()
of Qp/Qm 48
3.7 Conclusion and Recommendation 49
4 CASE STUDY 50
4.1 Location of Study 50
4.2 Piled Foundations 52
4.3 Static Pile Load Test 53
4.4 Pile Instrumentation 53
x
5 ANALYSIS OF RESULTS 56
5.1 General Presentation 56
5.2 Characterization of the Investigated Piles 56
5.3 Failure Criteria 57
5.4 Predicted Versus Measured Pile Capacity 57
5.5 Evaluation of Methods 62
5.5.1 Best Fit Line Equation 62
5.5.2 Cumulative Probability (CP) 65
5.5.3 Mean () and Standard Deviation ()
of Qp/Qm 70
5.5.4 Overall Performance 70
5.6 Discussion 70
REFERENCES 78
APPENDICES 84-114
xi
LIST OF TABLES
piles 24
2.3 Summary of Briaud et als statistical analysis for H-piles 28
2.4 Variation of CN with v 30
2.5 Variation of undrained shear strength, cu with SPT N-value 32
2.6 Value of C for different types of hammers 35
2.7 Value of for different types of hammers 35
2.8 Value of n for different types of hammers 36
2.9 Allowable total settlements, a for foundation design 38
2.10 Allowable angular distortion, a 39
4.1 Spun pile properties 52
5.1 Summary of pile failure criterion 57
5.2 Summary of piles investigated 59
5.3 Evaluation of the performance of the prediction methods
considered in this study 69
5.4 Summary of discussion 73
xii
LIST OF FIGURES
LIST OF SYMBOLS
LIST OF APPENDICES
CHAPTER 1
INTRODUCTION
(i) When upper soil layers are weak and unable to support the structural loads.
(ii) When underground water level is not constant.
(iii) When upper soil layers are susceptible to large settlement.
(iv) When the structure is subjected to lateral loads.
There are many different types of pile in use today, such as timber piles,
concrete piles, steel piles, composite piles and others. The choice of pile type for a
particular job depends upon the combination of all the various soil conditions and the
magnitude of the applied load; for example, timber piles are usually used in water
structure while precasted concrete piles are usually used in housing estate.
2
Current practice of pile design is based on the static analysis for example
Meyerhof Method, Vesic Method and Coyle & Castello methods. Due to the
uncertainties associated with pile design, field tests (pile load tests) are usually
conducted to verify the design loads and to evaluate the actual response of the pile
under loading. Static pile load tests are a verification tool for pile design and they
cannot be a substitute for the engineering analysis of the pile behavior. Maintained
Load Test Method (ML Test) is considered as the standard method by Jabatan Kerja
Raya (JKR). This test however takes 2-3 days to complete. Due to the long period
of time needed to conduct ML Test, it contradicts with the current construction
industry practice which is time-saving. Hence, Dynamic Load Test (DLT) especially
Pile Driving Analyzer (PDA) is gaining popularity in construction industry.
However, ML Test should have the final say on the ultimate bearing capacity of
piles.
1.2 Objectives
(i) To determine the ultimate bearing capacity of piles from illustrated full-scale
pile load tests.
(ii) To predict and calculate the bearing capacity of pile from static analysis, pile
driving formulae, and interpretation method.
(iii) To identify the most accurate method to predict pile bearing capacity by
comparing the predicted and calculated results with the actual results from
pile load tests.
This study is only considering the carrying capacity of spun piles of different
sizes driven to set. Other pile types such as timber piles and steel pipes were not
covered in the analyses. Four sets of data were acquired from Taisei Corporation.
Their testing program was conducted in Mukim Jimah power plant on November
2005. Square concrete piles are obsolete in this study due to different load
transferring mechanism (Hani and Murad, 1999). Only spun friction piles that tested
to failure are considered in this study.
Data acquired includes soil investigation reports, piling reports and pile load
tests reports. Soil investigation reports revealed the soil strata at the site and the
soils parameters, piling information and depth at which the piles set was revealed
from piling records while pile load tests reports gave the actual carrying capacity of
the piles.
Static analysis formulae and pile driving formulae are not recommended as
the sole means of determining the acceptability of a pile, except on small jobs
(Fleming, 1985). These analyses do not describe the complex mechanics of pile
driving in rational way and interaction between pile and the surrounding soil is
poorly modeled. Thus, it is important to determine accuracy from these formulae
through comparison with actual bearing capacity from site. The differences can be
used as a guideline when pile load tests are not able to be conducted.
The problems with many of the interpretation methods are that they are either
empirical methods or are based on set deformation criteria. Several methods are also
sensitive to the shape of the load-settlement curve and it is preferable to use a
considerable number of load increment to define the shape clearly; for example,
Chins Method assumes the load-deformation curve is hyperbolic and is an empirical
method. An engineer may have difficulty in choosing the best method to interpret
the static load test data. This study is able to help an engineer to identify the
suitability of the proposed interpretation methods to predict the ultimate bearing
capacity of spun piles driven to set. Moreover, through the analyses, the most
appropriate method is identified.
5
CHAPTER 2
LITERATURE REVIEW
The most common of these problematic soils are the soft, saturated clays and
silts often found near the mouths of rivers, along the perimeter of bays, and beneath
wetlands. These soils are very weak and compressible, and thus are subject to
bearing capacity and settlement problems. These soils also frequently include
organic material in which will aggravate these problems.
Areas underlain by soft soils frequently below mean sea level, and thus are
subjected to flooding. Therefore, it is necessary to raise the ground by placing fill.
However, the weight of the fill frequently causes large settlement. For example,
Scheil (1979) described a building constructed on fill underlain by varved clay in the
Hackensack Meadowlands of New Jersey. About 250 mm of settlement occurred
during placement of the fill, 12 mm during construction of the building, and an
additional 100 mm over the following ten years.
In seismic areas, loose saturated sands can become weak through the process
of liquefaction. Moderate to strong ground shaking can create large excess pore
water pressures in these soils, which temporarily decrease the effective stress and
shear strength. Seed (1970) described the phenomenon occurred in Niigata, Japan
during 1964 earthquake. Many buildings settled more than 1 m, and these
settlements were often accompanied by severe tilting.
6
There are many types of pile in use today, with varying geometry which
depends upon imposed loading and soil conditions. Generally, piles are classified
according to the nature of load support (friction and end-bearing piles), the
displacement properties (full-displacement, partial-displacement, and non-
displacement piles), and the composition of piles (timber, concrete, steel, and
composite piles). The choice of pile for a particular job depends upon the
combination of all the various soil conditions and the magnitude of the applied load.
Besides its technical aspects, economical factor should also be a consideration.
The behavior of the pile depends on many different factors, including pile
characteristics, soil conditions and properties, installation method, and loading
conditions. The performance of piles affects the serviceability of the structure they
supported. In this study, only driven piles (displacement piles) are discussed.
7
Most piles are driven into the ground by hammer or vibratory drivers. In
special circumstances, piles can also be inserted by jetting or pre-boring. In the
driving operation, a cap is attached to the top of the pile. A cushion may be used
between the pile and the cap. This cushion has the effect of reducing the impact
force and spreading it over a longer time. Pile is driven until it finds its bearing
layer. Usually, the driving is halted when the penetration is less than 25 mm per 10
blows.
Driven piles also known as displacement piles based on the nature of their
placement. Driven piles caused some soil to move laterally; hence, there is a
tendency for the densification of soil surrounding them. Concrete piles and closed-
ended pipe piles are high-displacement piles. However, steel H-piles displace less
soil laterally during driving, and so they are low-displacement piles.
Piles wobble during driving, thus creating gaps between them and the soil.
Tomlinson (1987) observed the gap extending to a depth of 8 to 16 diameters below
the ground surface. Piles subjected to applied lateral loads also can create gaps near
the ground surface. Therefore, the side friction in this zone may be unreliable,
especially in stiff clay (Coduto, 2001)
As a pile is driven into the ground, the soil below the toe must move out of
the way. This motion causes both shear and compressive distortions. These
distortions are greatest around large displacement piles. Cooke and Price (1973)
observed the distortion in London Clay as a result of driving a 168 mm diameter
closed-end pipe pile. The soil within radius of 1.2 pile diameters from the edge of
the pile was dragged down, while that between 1.2 and 9 diameters moved upward.
8
Besides, this remolding of the clay changes its structure and reduces its strength to a
value near its residual strength.
Pile driving also compresses the adjoining soils. If saturated clays are
present, this compression generates excess pore water pressure. The greatest
compression occurs near the pile toe, so the ratio of excess pore water pressure over
original vertical effective stress (ue/v) in that region may be as high as 3 to 4
(Airhart et al., 1969). Poulus and Davis (1980) also suggest that ue/ v may be as
high as 1.5 to 2.0 near the pile, gradually diminishing to zero at a distance of 30 to 40
pile radii.
For most clay, the excess pore water pressures that develop around a single
isolate pile completely dissipate in less than one month, with corresponding increases
in load capacity (Soderberg, 1962). This is due to the dissipation of water,
thixotropic effect and consolidation. However, in pile groups, the excess pore water
pressures may require a year or more to dissipate.
Soil compression from the advancing pile will generate excess pore water
pressures in loose saturated sands. However, sands have a much higher permeability
than clays, so these excess pore water pressures dissipate very rapidly. Thus, the full
pile capacity develops almost immediately.
Based on the observation by Coduto (2001), some local soil expansion can
occur when driving piles through very dense sands. This temporarily generates
negative pore water pressures that increase the shear strength and make the pile more
difficult to drive. This effect is especially evident when using hammers that cycle
rapidly. Suggestion by Coduto to partially or wholly negate this effect is by
predrilling or jetting to install the pile.
9
The pile load test program should be considered as part of the design and
construction process, and not carried hurriedly in response to an immediate
construction problem (Fleming, 1985). Pile tests may be performed at various stages
of construction, i.e. prior to construction and during construction. A large amount of
information can be obtained from properly planned tests. This useful information
may lead to refinement of the foundation design with consequent possible cost
saving and certainly greater assurance of the satisfactory performance of the
foundation.
Three types of tests have been recommended by the JKR, namely Maintained
Load Test (ML Test), Constant Rate of Penetration Test (CRP Test) and Pile Driving
Analyzer (PDA). These tests are performed based on the JKR specification or BS
8004. The standard procedures are explained in the later part of the report.
The period of time which the test should be carried out in various soils is
mentioned by Bowles (1996). Piles in granular soil are often tested 24 to 48 hrs after
driving when load arrangements have been made. This time lapse is sufficient for
excess pore water pressure to dissipate. Pile in cohesive soils should be tested after
sufficient lapse for excess pore water pressure to dissipate. This time lapse is
commonly in the duration of 30 to 90 days in order for cohesive soil to gain some
additional strength from thixotropic effects.
10
The failure of load test happened when either/all the following conditions are
observed at the site:
Residual
Settlement
Total
Settlement
After the load has been applied and removed, as above, the pile is reloaded to
the test load in increments of 50% of the design load until two times the design load.
20 minutes are lapsed between load increments. Then the load is increased in
increments of 10% of design load until failure, allowing 20 minutes between load
increments. The main disadvantage of this test is that it is time consuming. A typical
test period may last 40 to 70 hours or more.
In this test method, the pile is loaded in 20 increments to 300% of the design
load (each increment is 15% of the design load). Each load is maintained for a
period of 5 minutes with reading taken every 2.5 minutes. Load increments is added
until continuous jacking is required to maintain the test load or test load has been
reached. After a 5 minutes interval, the full load is removed from the pile in four
equal decrements with 5 minutes between decrements.
This method is fast and economical. Typical time of test by this method is 3
to 5 hours. This test method represents more nearly undrained conditions. This
method cannot be used for settlement estimation because it is a quick method.
12
According to Han (1999), the static test is considered as the reference test
because it is the one that corresponds the most with the way that the load is applied
in reality (duration, loading rate and type of loading). The static test is generally
regarded as the definitive test against which other types of tests are compared. These
elements are obviously the best advantages of this kind of tests.
The data obtained are directly interpretable because they are linked to the
acceptance criteria (maximum settlement and authorized stiffness and/or design
load). Another reason is that the main interpretations were created with respect to
this kind of test. As such, all the other methods tried to predict response comparable
to the load settlement produced by the static load test. Finally, the measurements are
generally independent of the pile material properties.
Since the static load test is very closely related to the reality, the time needed
to carry out is relatively long (Han, 1999). This duration is costly in term of money
and contract planning. Besides, to create the actual condition of loading slow
loading rate is imposed. The load is applied high enough to get closer to the real
load to be applied to the foundation. So the mobilization of this load and of this
associated reaction is strongly expensive regarding to the obtained result (one pile
tested).
The reaction supplied for the applied loading (kentledge, reaction piles,
ground anchors) generates some associated effects or interaction with pile that
perturb the interpretation of the results. These stresses will increase the shaft friction
and the base capacity. The pile settlement is reduced and the pile head stiffness is
also overestimated.
13
There are many procedures for static pile load testing. Due to the different
loading paths, any pile subjected to the various tests will exhibit a different load-
settlement response influencing the conclusions because the results are influenced by
loading history (steps and duration).
The ultimate load of pile is usually not well defined. Based on Fleming
(1985), two simple criteria which can be used to define ultimate load are the load at
which settlement continues to increase without further increase in load and the
load causing settlement of 10% of the pile diameter (base diameter). The latter limit
is likely to give a low estimate of the ultimate load as it is unlikely that general
yielding of the soil around the pile will have been initiated.
There are many methods that can be used to predict the ultimate load of piles.
The procedure of these methods is discussed at the latter part of the report.
This method was originally recommended for the QM test method. The main
advantage of this method is that the limit line BC can be drawn before starting the
test. Therefore, it can be used as one of the acceptance criteria for proof-tested
contract pile.
Based on the study by Tolosko (1999) on 63 piles, he found out that the ratio
of Davissons Method and designated static analysis is in the range of 0.9 to 1.1.
According to Bachand (1997), Davissons Method has the advantage of deterministic
(and hence objective), while being able to consider pile properties and geometry,
hence the tip size on failure zone.
Chins Method was first introduced in 1971 (Tolosko, 1999). This method is
shown in Figure 2.3 and consists of the following steps. The /Qva (settlement/load)
versus (settlement) plot is drawn, where is the settlement and Qva is the
corresponding applied load. The predicted failure load, Qp is then equal to 1/C1
15
where 1/ C1 is the gradient of the slope. The relationships given in the Figure 2.3
assume that the load movement curve is approximately hyperbolic.
This method of ultimate load interpretation is applicable for both the QM and
SM tests, provided that the constant time increments are used during the test. In
selecting the straight line from the points, it should be understood that the data points
do not appear to fall on the straight line. This method may not provide realistic
failure for tests carried out as per ASTM Standard Method because it may not have
constant time load increments.
steps. Load and movement is drawn on logarithmic scales. These values then will
fall on two straight lines. The predicted failure load, Qp is then defined as the load
that falls at the intersection of these two straight lines (De Beer, 1971). This method
was originally proposed for maintained load test, such as SM and QM test.
and consists of the following steps. The and curve is drawn, where is the
Qva
settlement and Qva is the load. Predicted failure load ,Qp and failure movement u are
then given as follows:
17
1
Qp = C1C 2 (2.1)
2
u = C2/C1 (2.2)
All the terms are defined in Figure 2.5. This method assumes that the load-
movement curve is approximately parabolic. The method is applicable for both QM
and SM tests. The failure criteria agree well with the plunging failure. However, the
plot and calculations could not be performed in advance of the test loading. This
method interpretation is not suitable for test methods that include unloading cycles
where plunging failure is not achieved (Nor Azizi, 2003).
Fuller and Hoys Method or also known as single tangent method was first
proposed in 1976 (Spronken, 1998). This method consists of the following steps. A
load-movement curve is drawn as shown in Figure 2.7. The predicted failure load Qp
on the curve is determined where the tangent on the load-movement curve is sloping
at 0.1 mm/kN.
This method is applicable for QM and SM test. The main disadvantage with
this method may be that it penalizes the long piles because they will have larger
elastic movements and therefore 0.1 mm/kN slope will occur sooner (Spronken,
1998).
19
Figure 2.7 Load-movement curve of Fuller and Hoys, and Butler and Hoys
Methods (Nor Azizi, 2003)
Butler and Hoys Method or also known as double tangent method was first
proposed in 1977 (Spronken, 1998). As shown in Figure 2.7, this method consists of
the following steps. The load-movement curve is drawn. The failure load is then the
intersection of the 0.1 mm/kN slope line with either the initial straight portion of the
20
curve or the line parallel to the rebound curve or the elastic line starting from the
origin. This method is applicable for the QM and SM test.
Hani and Murad (1999) mentioned that Butler and Hoys Method is the
primary load test interpretation method used by Louisiana Department of
Transportation and Development (LDOTD). They concluded that this method
generally underpredicted the bearing capacity of prestressed concrete piles based on
35 driven friction square concrete piles.
(i) For sand, since cohesion, c is equal to zero, the equation for load carrying
capacity at pile point, Qt = ApvNq where Ap is area of the pile, v is
effective overburden stress and Nq is bearing capacity factor.
(ii) Soil friction angle, is determined.
(iii) The Lb/D ratio for the pile is determined in which Lb is length of pile
embedded into bearing stratum and D is the width or diameter of pile.
(iv) (Lb/D)critical is obtained through Figure 2.8. Lb/D ratio should not exceed
(Lb/D)critical. If otherwise, (Lb/D)critical ratio will be used as design parameter.
(v) The appropriate value of Nq corresponding to the given Lb/D ratio or
(Lb/D)critical ratio is determined from Figure 2.8.
(vi) The Nq value calculated in Step (v) is used to obtain the Qt.
21
(vii) The value of Qt obtained in Step (vi) should not exceed the limiting load
carrying capacity at pile point given as Ap50Nq .
Figure 2.8 Critical embedment ratio and bearing capacity factors for various soil
friction angles (after Meyerhof, 1976)
Based on field observations, Meyerhof (1976) also suggested that the ultimate
point resistance, Qt, in a homogeneous granular soil (L = Lb) can be obtained from
standard penetration numbers as
where N is an average standard penetration number (about 10D above and 4D below
the pile point).
22
Coyle and Castello (1981) have analyzed 24 large-scale field load tests of
driven piles in sand. They have shown that equation Qp = ApvNq + favpL can
predict the ultimate load with an error band 30% with a majority falling within an
error band of 20%. Based on this study, Coyle and Castello correlated Nq with
embedment ratio (L/D). Figure 2.9 shows the values of Nq for various of embedment
ratios (L/D) and friction angle ( ) of the soil.
Qs = pLf av (2.4)
where p is perimeter of the pile section, L is the pile length of the soil boundary and
fav is unit friction resistance at any given depth.
Meyerhof indicated that the average unit frictional resistant for driven high-
displacement piles can be obtained from corrected average standard penetration
resistance values as fav = 2Ncor. Thus,
Qs = pL 2 N cor (2.5)
where Ncor is corrected average standard penetration resistance values of the soil
layer.
There are several more researchers studied on the average frictional resistance
such as Broms (1965), Aas (1966) and Meyerhof (1976). All of them agree that
average frictional resistance can be represented by Kvtan where K is earth
friction angle. However, the values of the parameters are varies from one researcher
to another researcher.
24
Coyle and Castello (1981) correlated that earth pressure coefficient, K with
embedment ratio (L/D) and friction angle ( ) of the soil as shown in Figure 2.9.
This chart is designed based on assumptions that = 0.8 .
Broms (1965) suggested the values for K in granular soils as in Table 2.1
while Aas (1966) proposed the values of as in Table 2.2:
Table 2.1 : Values for earth pressure coefficient, K in granular soils (Broms, 1965)
Type of Pile Loose Sand Dense Sand
Steel 0.5 1.0
Concrete 1.0 2.0
Timber 1.5 3.0
(Aas, 1966)
Type of Pile Soil-Pile friction angle,
Steel 20
Concrete 0.75
Timber 0.66
Note : is the friction angle of soil
Qt = 9 c u A p (2.6)
f = c u (2.7)
Tomlinson (1971) showed that the adhesion factor is influenced by other soils
overlying the stiff London Clay. Overlying soft clay results in smaller adhesion
factors, whereas overlying granular soils give greater factors.
where v is mean effective vertical stress for the entire embedment length, cu is the
mean undrained shear strength ( = 0 concept) and the value for can be estimated
from Figure 2.11.
27
Downdrag force can cause differential settlement, and, in severe cases, these
loads may pull some of the piles out of their caps (Bozozuk, 1981) because of the
variations in soil properties, fill thickness, and other factors. Downdrag force is
more pronounced in the soft soils and the overlying fills. One of the method to
reduce this effect is by coating the bitumen in downdrag zone.
28
The bitumen acts like a rubber to reduce the coefficient of friction. This
method is very effective, so long as the pile is not driven through an abrasive soil,
such as sand, that might scrap off the bitumen coating.
However, this method will present another problem which is reduced the skin
friction of the pile. Hutchinson and Jensen (1968) described a reinforced concrete
test piles driven into soft silty clays at Khorramshar, failed to reach predicted pile
bearing capacity by wide margin. They made the interesting observation that the
skin friction at the pile/soil interface had been considerable weaken by the 1-2 mm
thick soft bitumen applied to the piles to protect them from acid attack. The skin
friction developed on the coated piles was only 30 to 80% of that on uncoated piles.
Thus, Engineers should investigate before any method is implemented.
2.5.6 Comparison of Static Analysis Results with Pile Load Test Results
Briaud et al. (1989) reported the results of 28 axial load tests performed by
the U.S. Army Engineering District (St. Louis) on impact-driven H-piles and pipe
piles in sand during the construction of the New Lock and Dam No. 26 on the
Mississippi River. Briaud et al. made a statistical analysis to determine the ratio of
theoretical ultimate load to measured ultimate load. The results are summarized in
Table 2.3.
Table 2.3 : Summary of Briaud et al.s statistical analysis for H-piles (Das, 2004)
Method Qt Qs Qu
CV CV CV
Coyle and Castello 2.38 1.31 0.55 0.87 0.36 0.41 1.17 0.44 0.38
(1981)
Briaud and Tucker 1.79 1.02 0.59 0.81 0.32 0.40 0.97 0.39 0.40
(1984)
Meyerhof (1976) 4.37 2.76 0.63 0.92 0.43 0.46 1.68 0.76 0.45
Note : is mean, is standard deviation, and CV is coefficient of variation
29
Meyerhof (1976) also provided the results of several field load tests on long
piles (L/D 10), from which the derived values of Qt were calculated. He concluded
from the results that for a given friction angle, the magnitude of Qt can deviate
substantially from that given in the static analysis method. Again, this shows the
uncertainty in predicting the bearing capacity of piles.
The current practice in using the N-value for design is to use an average N-
value but in the zone of majoring stress. For pile foundations, there may be merit in
the simple average of blow count N for any stratum unless it is very thick. It is more
accurate better to subdivide the thick stratum into several strata and average the N
count for each division (Ramli, 2005). Average corrected N-value can be computed
and then averaged. The average N-value is correlated with empirical formula to
obtain the soil parameters and finally the bearing capacity.
30
Kovac and Salomone (1982) found that the energy impact to the sampler
range about 30 to 80% while Riggs (1983) obtained energy input from 70 to 100%.
Therefore, the raw SPT data need to be improved by applying certain correction
factors.
20
Ncor = 0.77 N log (for v 23.9 kN/m2) (2.9)
0.0105 ' v
where CN is a correction factor, the variation of which with v is given in Table 2.4.
1 2
9.78 while Skempton (1986) porposed CN = .
'v 1 + 0.01 , v
For granular soils, the corrected N-value can be used to estimate the effective
friction angle of the soil, . Wolff (1989), based on research by Peck, Hanson and
Thornburn in 1974 has produced an empirical formula to correlate friction angle with
Ncor. The formula is shown as:
Kulhawy and Mayne (1990), based on the work by Schmertmann in 1975 has
approximate an empirical formula to estimate the friction angle:
Nf
= tan 1 (2.12)
'v
12.2 + 20.3 p
a
Table 2.5 : Variation of undrained shear strength, cu with SPT N-value (BS 5930)
SPT N-value Consistency Undrained shear
strength, cu (kN/m2)
Less than 4 Very soft Less than 20
4-10 Soft 20-40
10-30 Firm 40-75
30-50 Stiff 75-150
More than 50 Very stiff More than 150
Besides BS 5930, Stroud (1974) based on the results of undrained triaxial test
suggested that cu = KN where K is a constant in the range of 3.5 6.5 kN/m2. Stroud
found that the average value for K is about 4.4 kN/m2. Hara et al. (1971) also
suggested that cu = 29N0.72.
Many attempts have been made to determine the relationship between the
dynamic resistance of pile during driving and the static load-carrying capacity of the
pile. These intended relationships are called pile driving formulae and have been
established empirically or theoretically. According to Simon nad Menzies (2000).
Much discussion has been generated, for example, ASCE (1951), Chellis (1941),
Cummings (1940), and Greulich (1941). Conflicting opinions have been expressed.
Simons and Menzies (2000) suggest that the Janbu formula and the Hiley
formula are convenient to use and give reasonable predictions of the ultimate bearing
33
capacity of driven piles in granular soils. Das (1986) also suggested Engineering
News Record (ENR) formula besides above mentioned formulae. His reason is that
ENR formula which was introduced during nineteenth century has gone through
several revisions over the years and is acceptable for prediction of the ultimate load.
Das also suggested a factor of safety (FOS) of 4 - 6 should be recommended to
estimate the allowable load. However, McCarthy (1998) has different opinion. He
suggested that the use of ENR formula should be discouraged because it does not
have application for existing pile driving methods.
Janbus Formula was first introduced in 1953 (Das, 1999). The ultimate
bearing capacity can be calculated based on the following formula:
W R H
Qp = (2.13)
Ku S
e
0.5
Ku = C d 1 + 1 + (2.14)
d
C
Wp
Cd = 0.75 + 0.15 (2.15)
WR
W R HL
e = (2.16)
Ap E p S 2
ENR formula is derived on the basis of the work-energy theory. This means
that energy imparted by the hammer per blow is the summation of pile resistance
times penetration per hammer blow (Das, 1999).
35
The original ENR Formula has been revised for several times due to its
irrational prediction on ultimate bearing capacity. The Michigan State Highway
Commission (1965) undertook a study to obtain a rational pile-driving equation. At
three diverse sites, a total of 88 piles were driven. Based on these results, a modified
ENR had been adopted. According to the revised ENR formula (Das, 1999), the pile
resistance is the calculated ultimate load, Qp and can be expressed as:
W R H W R + n 2W P
Qp = x (2.17)
S +C WR + W p
The values of C recommended are shown in Table 2.6 while the efficiencies
of various pile driving hammers, , are given in Table 2.7.
Coduto (2001) explains that the foundation should not only based on strength
requirements but also the serviceability requirements. The following sections will
discuss on these requirements.
37
Geotechnical strength requirements are those that address the ability of the
soil or rock to accept the loads imparted by the foundation without failing. The
strength of soil is governed by its capacity to sustain shear stresses. In other words,
if the shear strength of the soils is greater than the shear stresses, then the foundation
is safe. Geotechnical strength analyses are almost always performed using allowable
stress design (ASD) methods.
Foundation that satisfies strength requirements will not collapse, but they still
may not have adequate performance. For example, they may experience excessive
settlement. Therefore, there are second performance requirements, which are known
as serviceability requirements.
38
2.8.2.1 Settlement
a (2.18)
Table 2.9 : Allowable total settlements, a for foundation design (Coduto, 2001)
Type of structure Allowable total settlement, a (mm)
Office building & house 12 50 (25 is the common value)
Heavy industrial building 25 75
Bridges 50
Da = aSc (2.19)
2.8.2.2 Heave
criteria for heave are the same as those for settlement. However, if some foundations
are heaving while others are settling, then the differential is the sum of two.
2.8.2.3 Tilt
Studies of waterfront structures have found that steel is lost at rate of 0.075 to
0.175 mm/yr (Whitaker, 1976). This corrosion occurs most rapidly in the tidal and
splash zones and can also be very extensive immediately above the sea floor. It
becomes almost negligible at depths of more than about 0.5 m below the sea floor.
Such structures may also prone to abrasion from moving sand, ships, floating debris,
and other sources. It is common to protect such foundations with coatings or jackets.
CHAPTER 3
METHODOLOGY
3.1 Introduction
The collected data include soil investigation reports, piling records, and pile
load test records. Data was acquired from Taisei Corporation. They conducted their
tests in Mukim Jimah Power Plant on November 2005. Spun piles of different sizes
driven to set were considered in this study.
43
Data Collection
Compilation of Data
a) soil data
b) SPT data
c) piling records
d) pile tests data
Evaluation of Methods
Conclusion
The information from the data collected was compiled. The information and
data regarding the project, soil stratification and properties, pile characteristics, load
test data, SPT profiles, etc. were processed and transferred from each load test report
to tables, forms, and graphs.
44
The soil data consists of information on the soil boring location (station
number), soil stratigraphy and other information. From soil stratification, the
predominant soil type was qualitatively identified (cohesive or cohesionless). The
importance of this identification is addressed in the analysis section.
Pile load test report consist of date of loading, applied load with time, pile
head movement, pile failure under testing, and etc.
45
only on friction piles that failed (plunged or showed large settlement) under load
testing.
(i) R1 - the equation of the best fit line of predicted versus measured capacity
(Qp/Qm) with the corresponding coefficient of determinations (R2);
(ii) R2 Qp/Qm at 50 and 90 percent cumulative probability (P50 and P90);
(iii) R3 mean () and standard deviation () calculations of Qp/Qm
For each prediction method, analysis was conducted to obtain the best fit line
for the predicted/measured pile capacities (Figure 3.2). The relationship Qp/Qm and
the corresponding coefficient of determination (R2) were determined for each
method. In this criterion, the method is better when Qp/Qm is closer to unity. If there
are two or more methods that produced the same answer, the method with R2 closer
to one will be considered as the better method.
47
Long et al. (1999) used the cumulative probability (CP) value to quantify the
ability of different methods to predict the measured pile capacity. The concept is to
sort the ratio Qp/Qm for each method in an ascending order. The smallest Qp/Qm is
given number i = 1 and the largest is given i = n where n is the number of piles
considered in the analysis. The cumulative probability value for each Qp/Qm is given
by:
i
CPi = (3.1)
n +1
The cumulative probability curves were used to determine the 50 percent and
90 percent cumulative probability values (P50 and P90) (Figure 3.3). The pile capacity
prediction method with P50 value closer to unity and with lower P50 P90 range is
considered the best. P50 is considered as the average value of Qp/Qm while P50 P90
is considered as the value examining the consistency of the data.
48
The ratio of predicted to measured ultimate pile capacity (Qp/Qm) was the
main variable considered in the analyses. This ratio (Qp/Qm) ranges from 0 to with
an optimum value of one. The methods underpredicts the measured capacity when
Qp/Qm < 1 and it overpredicts the measured capacity when Qp/Qm > 1. The mean and
standard deviation of Qp/Qm are indicators of the accuracy and precision of the
prediction method. An accurate and precise method gives mean (Qp/Qm) = 1 and
standard deviations (Qp/Qm) = 0, respectively, which means that for each pile, the
predicted pile capacity equals to the measured one. This case is ideal, however, in
reality the method is better when mean (Qp/Qm) is closer to one and standard
deviation (Qp/Qm) is closer to 0.
Qp 1 n Qp
= (3.2)
Qm n i =1 Qm
2
Qp 1 n Qp
=
n 1 i =1 Qm
(3.3)
Qm
The method with the lowest rank index is considered as the most accurate
method and recommended for pile design practice. Recommendations regarding the
proposed methods and further improvement for the study were also included in this
section.
50
CHAPTER 4
CASE STUDY
The site of this study is located at Mukim Jimah. Mukim Jimah is located
east of the mouth of the Sepang River and off the Kuala Lukut shoreline in the state
of Negeri Sembilan in west peninsular Malaysia. It lies at an elevation of between 0
m and 5 m below the Malaysian Land Survey Datum (MLSD, approximate Mean
Sea Level). Reference to the geological map of the site and its surroundings
(Geological Survey Malaysia, 1985) shows that the site is underlain by very soft to
soft clays, organic soils and very loose to loose sands presumably deposited during
the Pleistocene and Holocene Epochs of the Quaternary Period. The solid geology of
the site consists of meta-sedimentary rocks (Phyllite, Schist, Slate and Sandstone) of
the Devonian Period (Krishnan and Lee, 2006).
In the site, the spun piles were driven by 11 tonne BSP hydraulic hammer
(pile no. TP3C) and 9 tonne Junttan hydraulic hammer (pile no. TP5, TP9, and
TP10). Pre-boring was also carried out on upper 12 m of piles TP3C and TP5. Table
4.1 summarized the properties of the mentioned piles.
Two types of static pile load test had been conducted on the site. They are
normal maintained load test and quick maintained load test. Normal maintained load
tests were conducted on three piles in the site (TP3C, TP5 and TP9) while quick
maintained load test was conducted on TP10. The detail procedures for these tests
had been elaborated in Sections 2.3.1.1 and 2.3.1.2 respectively.
For the instrumented test spun piles (TP3C, TP5 and TP9), pile
instrumentation was conducted using Vibrating Wire (VW) Extensometers
monitoring system for determining axial loads and movements at various levels
down the pile shaft including the pile base level using Global Strain Extensometer
(GloStrExt) Instrumentation method (Mukim Jimah Site Report, 2005).
During static load testing, the load intrigued pile deformation resulted in
relative movement between each and every two anchored intervals. The relative
movement causing changes in strain gauge wire tension and corresponding change in
its resonant frequency of vibration. To measure the resonant frequency, Glostrext
strain gauge wire was plucked using electromagnetic coil connected through a signal
cable to data logger, which also measures the frequency and displays the shortening
reading and strain reading. The strain reading is essential for calculation of stress
value at each anchor level.
54
Figure 4.3 Instrumentation details for static axial compression load tests
(Krishnan and Lee, 2006)
The pile top settlement was monitored using two different instruments,
survey precise level instrument (Figure 4.4) and Linear Vertical Displacement
Transducers (LVDT).
In the first instrument, pile top was affixed with vertical scale rule and
sighted by a survey precise level instrument with the use of TP3B and TP4 as
temporary benchmark reference for correction purposes. JKR specification
mentioned that the apparatus should be placed on the stable ground and six meters
from the reaction system.
As mentioned in Section 4.2, normal and quick maintained load tests were
conducted on the test piles. The tests were conducted using kentledge reaction
system (Figure 4.4). In this method, kentledge was placed onto a test frame and cribs
which rest upon the ground.
In the setup, a hydraulic jack was used to provide the load by acting against
the main beam. The hydraulic jack was operated by electric pump. Calibrated VW
Load Cell was used to indicate the applied load. The load cell was placed between
the jack and the kentledge framework and a pressure gauge linked to the hydraulic
pump.
Besides manual precise level survey level, all other instruments were logged
automatic using Micro-10x Datalogger and Multilogger software, at 2 minutes
interval during loading and unloading steps. All the instruments were calibrated
before were used in the test programs.
Figure 4.4 Typical static axial compression load tests setup (Krishnan and Lee,
2006)
56
CHAPTER 5
ANALYSIS OF RESULTS
Four spun piles were considered in the current study. However, the analyses
were conducted on the piles that were failed during the pile load test. A summary of
the characteristics of the investigated pile is presented in Table 5.2. All the piles are
friction piles based on the data given.
57
The predicted ultimate bearing capacity (Qp) is the sum of pile tip capacity
(Qt) and pile shaft resistance (Qs). The pile capacities Qt, Qs, and Qp predicted by the
interpretation methods, pile driving formulae, and Meyerhof method are compared
with Qm in Figures 5.1 to 5.8. Based on the graph, it is observed that prediction of
pile ultimate bearing capacities by Fuller and Hoys, and Butler and Hoys method are
58
near to the exact bearing capacities measured at the site whereas the other six
methods showed deviation from the measured bearing capacitites.
14000
12000
10000
Capacity (kN)
8000 Actual
6000 Predicted
4000
2000
0
TP3C TP5 TP9 TP10
Pile Number
7000
6000
5000
Capacity (kN)
4000 Actual
3000 Predicted
2000
1000
0
TP3C TP5 TP9 TP10
Pile Number
Figure 5.2 Comparison of measured and predicted pile capacity (Brinch Hansen)
59
7000
6000
5000
Figure 5.3 Comparison of measured and predicted pile capacity (Fuller and Hoy)
7000
6000
5000
Capacity (kN)
4000 Actual
3000 Predicted
2000
1000
0
TP3C TP5 TP9 TP10
Pile Number
Figure 5.4 Comparison of measured and predicted pile capacity (Butler and Hoy)
7000
6000
5000
Capacity (kN)
4000 Actual
3000 Predicted
2000
1000
0
TP3C TP5 TP9 TP10
Pile Number
Figure 5.5 Comparison of measured and predicted pile capacity (De Beer)
61
7000
6000
5000
7000
6000
5000
Capacity (kN)
4000 Actual
3000 Calculated
2000
1000
0
T3C TP5 TP9 TP10
Pile Number
7000
6000
Capacity (kN)
5000
4000 Actual
3000 Calculated
2000
1000
0
TP3C TP5 TP9 TP
Pile Number
Figures 5.9 to 5.16 show the best fit equation and coefficient of determination
of each method. The summary of the results for each method and their ranks in this
criterion is shown in Table 5.3.
P re d ic t e d P ile C a p a c it y , Q p (k N )
14000
12000 QP= 1.7793 Qm
10000 R2 =0.98
Perfect fit
8000
6000
4000
2000
0
0 2000 4000 6000 8000 10000 12000 14000
7000
5000
Perfect fit
4000
3000
2000
Qp = 0.4721 Qm
1000 R 2 =0.90
0
0 1000 2000 3000 4000 5000 6000 7000
Figure 5.10 Predicted (Brinch Hansen Criterion) versus measured pile capacity
7000
Predicted Pile Capacity, Qp (kN)
6000
5000
Perfect fit
4000
3000
2000
Qp = 1.0182 Qm
1000 R2 =0.99
0
0 1000 2000 3000 4000 5000 6000 7000
Figure 5.11 Predicted (Fuller and Hoys Method) versus measured pile capacity
7000
Predicted Pile Capacity, Qp (kN)
6000
5000
Perfect fit
4000
3000
2000
Qp = 0.9849 Qm
1000 R2 =0.99
0
0 1000 2000 3000 4000 5000 6000 7000
Figure 5.12 Predicted (Butler and Hoys Method) versus measured pile capacity
64
7000
5000
Perfect fit
4000
3000
2000
Qp = 0.6283 Qm
1000 R 2 =0.99
0
0 1000 2000 3000 4000 5000 6000 7000
Figure 5.13 Predicted (De Beers Method) versus measured pile capacity
7000
Calculated Pile Capacity, Qp
6000
5000
Perfect fit
4000
(kN)
3000
2000
Qp = 0.5061 Qm
1000
R 2 =0.92
0
0 1000 2000 3000 4000 5000 6000 7000
7000
Calculated Pile Capacity, Qp
6000
5000
Perfect fit
4000
(kN)
3000
2000
Qp = 0.3596 Qm
1000
R 2 =0.99
0
0 1000 2000 3000 4000 5000 6000 7000
7000
5000
Perfect fit
4000
3000
2000
Qp = 0.4633Qm
1000 R 2 =0.62
0
0 1000 2000 3000 4000 5000 6000 7000
Inspection of Figures 5.9 to 5.16 (Qp/Qm plots) shows that Butler and Hoy
method has best fit equation Qp = 0.9849Qm with R2 = 0.99. This method tends to
underpredict the measured pile capacity by an average of 1 percent. Therefore,
Butler and Hoy method ranks number one according to this criterion and is given R1
= 1 (R1 is the rank based on this criterion). The Fuller and Hoy method with Qp =
1.0182Qm (R2 = 0.99) tends to overpredict the measured capacity by 2 percents and
therefore ranks number 2 (R1 = 2). According to this criterion, Brinch Hansen, De
Beer, Janbu, ENR, and Meyerhof methods tend to underpredict the measured
ultimate pile capacity, while Chin method tends to overpredict the measured ultimate
pile capacity. The Chin method showed the inaccurate performance with Qp =
1.7793Qm (R2 = 0.98) and therefore was given R1= 8.
Figures 5.17 to 5.24 show the values of P50 and P90 of each method. The
summary of the results for each method and their ranks in this criterion is shown in
Table 5.3.
66
2.5
2.10
2
1.65
1.5
Qp/Qm
1
0.5
0
0 20 40 60 80 100
Cumulative probabiity (%)
1 0.97
0.9
0.8
0.7
0.6 0.55
Qp/Qm
0.5
0.4
0.3
0.2
0.1
0
0 20 40 60 80 100
Cumulative probabiity (%)
Figure 5.18 Cumulative probability plot for Qp/Qm (Brinch Hansens Criterion)
1.2 1.12
1
1.00
0.8
Qp/Qm
0.6
0.4
0.2
0
0 20 40 60 80 1 00
Cumulative probabiity (%)
Figure 5.19 Cumulative probability plot for Qp/Qm (Fuller and Hoys Method)
67
1.2 1.08
1
0.96
0.8
Qp/Qm
0.6
0.4
0.2
0
0 20 40 60 80 100
Cumulative probabiity (%)
Figure 5.20 Cumulative probability plot for Qp/Qm (Butler and Hoys Method)
1
0.9
0.8
0.7 0.68
0.6
Qp/Qm
0.62
0.5
0.4
0.3
0.2
0.1
0
0 20 40 60 80 100
Cumulative probabiity (%)
Figure 5.21 Cumulative probability plot for Qp/Qm (De Beers Method)
1 0.92
0.9
0.8
0.7
0.6
Qp/Qm
0.51
0.5
0.4
0.3
0.2
0.1
0
0 20 40 60 80 100
Cumulative probabiity (%)
1
0.9
0.8 0.75
0.7
Qp/Qm 0.6
0.5
0.4 0.36
0.3
0.2
0.1
0
0 20 40 60 80 1 00
Cumulative probabiity (%)
1
0.9 0.83
0.8
0.7
0.6
Qp/Qm
0.50
0.5
0.4
0.3
0.2
0.1
0
0 20 40 60 80 1 00
Cumulative probabiity (%)
Table 5.3 : Evaluation of the performance of the prediction methods considered in this study
Pile Capacity Best fit calculations Arithmetic calculations Cumulative probability Rank of the methods based on
Prediction Methods calculations their performance
The summary of the results for each method and their ranks in this criterion is
shown in Table 5.3. In this criterion, the arithmetic mean () and standard deviation
() of the ratio Qp/Qm values for each method were calculated. The best method is
the one that gives a mean value closer to one with a lower standard deviation, which
is the measure of scatter in the data around the mean. According to this criterion,
Fuller and Hoy method with = 0.998 and = 0.0022 ranks number one (R2 = 1)
followed by the Butler and Hoy method (R2 = 2). Brinch Hansen, De Beer, Janbu,
ENR, and Meyerhof methods have < 1, which means that these methods on
average are underpredicting the measured pile capacity. On the other hand, Chin
method has > 1, which means that these methods on average are underpredicting
the measured pile capacity.
5.6 Discussion
in predicting the ultimate measured load carrying capacity of spun piles. It is strongly
recommended that this method is implemented in design and analysis of spun piles.
Although Butler and Hoy, and Fuller and Hoy showed accuracy in predicting
the ultimate load carrying capacity of spun piles, it is recommended that a factor
should be introduced due to uncertainties and variations in soil profiles. The factor
recommended is 1.1 as Fuller and Hoys method overpredicted the ultimate bearing
capacity by 2 percent. Besides, one should be careful in determine the 0.1 mm/kN
slope on load-movement curve as it is not clearly defined especially in longer piles.
It is recommended that the same personnel should be responsible in determining the
tangent when these two methods are implemented as the judgment will be consistent.
De Beer method defined the ultimate load falls at the intersection of the two
straight lines. However, not all the intersections of these lines are clearly defined.
Therefore, some judgment may be required in determine these intersections. It is
recommended that the same personnel should be responsible in determining the
intersection point as the judgment will be consistent. This method may be used as
supplementary method when the load is more than 1000 kN (accuracy is around
60%).
Brinch Hansens 80 percent criterion is not suitable for test methods that
include unloading cycles or where plunging failure is not achieved. The result of this
study support the statement as the performance of this method is average (accuracy
of 60%). Therefore, this method is not recommended for load tests that include
unloading cycles. However, it can be used as an additional method when the load is
not more than 1000 kN as the accuracy is relatively higher (83%) than the average
accuracy.
Chin method shows least competency in predicting the ultimate load of spun
piles driven into Mukim Jimah soils. Part of the reason is that the load tests carried
out in the site did not have constant time load increments. As a result, it affected the
accuracy of this method as Chin method is understood dependant on the time load
72
increments. Hence, it is not recommended for load tests that have different time load
increment. However, the suitability of this method for load tests which have constant
time load increments is subjected to further study. A factor of 2 is recommended as
Chins method overpredicted the ultimate bearing capacity by 78 percent.
Among the two pile driving formulae evaluated in this study, Janbu formula
shows better accuracy than ENR formula. However, both showed low accuracy with
accuracy less than 55%. The sources of this problem may include the followings:
(i) The pile, hammer, and soil types used to generate the formula may not be the
same as those at site where it is being used. This is probably one of the major
reasons for the inaccuracies in ENR Formula.
(ii) The hammers do not always operate at their rated efficiencies.
(iii) The energy absorption properties of cushions can vary significantly.
(iv) The formulae do not account for flexibility in the pile.
(v) There is no simple relationship between the static and dynamic strength of
soils.
Because of these difficulties, there is little need to continue using pile driving
formulae unless there are some correction in these formulae. For recommendation,
the formula may be modified to suit the condition of the site or at least the particular
area. Besides, the coefficient of efficiencies for Janbu and ENR formula should be
restudied as the equipments which are used today are certainly more efficient.
Finally, the formulae should include some parameters on pile flexibility and
relationship between static and dynamic strength of soils. However, pile driving
formulae may be used for initial design as it gives engineer some idea on the pile
bearing capacity.
Based on the data, the piles are considered as friction piles. However,
calculations based on Meyerhof method suggested that the piles are end bearing
piles. Meyerhof method overpredicted the end bearing of piles but on the other hand,
underpredicted the skin resistance. These differences were down to the soil
73
parameters used in the design procedure. As mentioned, these soil parameters were
obtained from empirical formulae based on average SPT N-values. These empirical
formulae may not be suitable for Mukim Jimah soils as the parameters are irrelevant.
Laboratory tests should be conducted to obtain the actual soil parameters or at least
proven empirical formulae in Malaysian soils should be implemented. Meyerhof
method with the soil parameters based on the average N-value may only be used as
the preliminary design. This method is found to underpredict the ultimate load.
Table 5.4 summarized the results of discussion and comments.
CHAPTER 6
5.1 General
5.2 Conclusion
(i) Based on the results of this study, Butler and Hoy method shows the best
capability in predicting the measured load carrying capacity of spun piles.
Fuller and Hoy method also shows competency in predicting the ultimate
load carrying capacity of piles. Other methods such as De Beer, Brinch
Hansen showed an average accuracy in predicting the ultimate carrying
capacity of spun piles. Chin method is found to be the least suitable in
predicting ultimate load carrying capacity.
(ii) It is concluded that six out of eight methods considered in the study
underpredicted bearing capacity of spun piles. These methods are Butler and
Hoy, Brinch Hansen, De Beer, Meyerhof, Janbu, and ENR method. Except
76
for Butler and Hoy method, the other five methods tend to underpredict the
ultimate load in the range of forty to sixty percent. Butler and Hoy method
underpredicts the ultimate load by one percent. The two methods that
overpredict the ultimate carrying capacity of piles are Fuller and Hoy and
Chin method. The margin of error by Fuller and Hoy is small (around two
percent). However, Chin method overpredicted almost two times the
measured ultimate load.
(iii) Fuller and Hoy, Butler and Hoy and De Beer method show its consistency in
predicting the ultimate bearing capacity of the spun piles. It is shown in
standard deviation () column as all of these methods have less than 0.004.
Meyerhof and Chin method shows less consistency in predicting the ultimate
load as for meyerhof method is 0.05 while the standard deviation for Chin
method is 0.09.
(iv) Butler and Hoy method is the recommended method for pile design practice
as it is precise and consistent in predicting the spun piles capabilities
5.3 Recommendation
(i) At least 10 data are needed for this type of study. This is essential for
cumulative probability calculations (CP) as CP at ninety percent is needed for
analysis. For the current study, the value of CP ninety percent is based on
interpolation.
(ii) For further studies, driven square precasted concrete piles and bored piles
may be included. End bearing piles may be useful for improvement of this
study.
(iii) Other load test interpretation methods such as Davissons, Brinch Hnasen
90% Criterion and Mazurkiewicz Methods may be included in the discussion.
(iv) Pile driving formula used by various companies in the industry such as
Pilecon and Hileys formulae may be included for further discussion.
77
(v) Other static analysis formulae such as Vesic, and Coyle and Castello may be
included for further investigation.
(vi) Detail laboratory tests reports should be included for further studies.
(vii) Other evaluation methods such as histogram and log normal probability
distribution may be included to further evaluation of the methods.
78
REFERENCES
Airhart, T.P. and et. al. (1969). Pile-Soil System Response in a Cohesive Soil.
Performance of Deep Foundations. Philadelphia: ASTM. 264-294.
Bachand, M.L.Jr. (1999). Express Method of Pile Testing by Static Cyclic Loading.
University of Massachusetts Lowell: Master Thesis.
Bowles, J.E. (1996). Foundation Analysis and Design. 5th edition. New York:
McGraw-Hill Companies Inc. 167-181.
Briaud, J.L. et. al. (1989). Analysis of Pile Load Test at Lock and Dam 26.
Proceedings of Foundation Engineering: Current Principles and Practices.
American Society of Civil Engineers. Vol. 2. 925-942
British Standard Institution (1987). British Standard 5930: Code of Practice for Site
Investigation. London.
Broms, B.B. (1965). Methods of calculating the Ultimate Bearing capacity of Piles:
A Summary. Journal of the Geotechnical Engineering Division. American
Society of Civil Engineers. Vol. 91, No. GT3. 187-222.
Butler, H.D., and H.E. Hoy (1977). The Texas Quick-Load Method for Foundation
Load Testing-User's Manual. Report No. FHWA-IP-77-8.
Coduto, D.P. (2001). Foundation Design: Principles and Prctices. 2nd edition. New
Jersey: Prentice Hall. 24-35.
79
Coyle, H.M., and Castello, R.R. (1981). New Design Correlations for Piles in Sand.
Journal of the Geotechnical Engineering Division. American Society of
Civil Engineers. Vol. 107, No. GT7. 965-986.
Fleming, W.G.K., Weltman, A.J. and Randolph, M.F. (1985). Piling Engineering.
New York: John Wiley & Sons. 301-340.
Fuller, F.M., (1983). Engineering of Pile Installations. New York: McGraw Hill
Book Co. 170-181.
Han, Y. (1999). Axial Load Test and Analysis for Open-Ended Steel Tubular Piles
Driven into Weathered Rock. University of Ottawa: Master Thesis.
Hani, H.T. and Murad, Y.A.F. (1999). Evaluation of Capacity of Piles from Cone
Penetration Data. Unites States: Louisiana Transportation Research Center.
Hara, A., Ohata, T., and Niwa, M. (1971). Shear Modulus and Shear Strength of
Cohesive Soils. Soils and Foundations. Vol. 14, No. 3. 1-12.
80
Hutchinson, J.N., and Jensen, E.v. (1968). Loading Tests on Piles Driven into
Estuarine Clays at Port of Khorramshav, and Observations on the Effect of
Bitumen Coatings on Shaft Bearing Capacity. Publication No. 78,
Norwegian Geotechnical Institute, Oslo. 119-126.
Kovacs, W.D.,, and Salomone, L.A. (1981). Energy Measurements in the SPT.
Building Science Series 135. Washington, D.C: National Bureau of
Standards.
Kulhawy, F.H., and Jackson, C.S. (1989). Some Observations on Undrained Side
Resistance of Drilled Shafts. Proceedings, Foundation Engineering: Current
Principles and Practices. American Society of Civil Engineers. Vol. 2.
1011-1025.
Liao, S.S.C., and Whitman, R.V. (1986). Overburden Correction Factors for SPT in
Sand. Journal of Geotechnical Engineering. American Society of Civil
Engineers. Vol. 114, No. 3. 373-377.
McCarthy, D.F. (1998). Essentials of Soil Mechanics and Foundations. 5th edition.
New York: Prentice Hall Inc. 147-149, 497-503
Meyerhof, G.G. (1959). Compaction of Sands and Bearing Capacity of Piles. Journal
of the Geotechnical Engineering Division. American Society of Civil
Engineers. Vol. 85. 1-29.
Nor Azizi Yussoff (2003). Foundation Engineering: Lecture Notes. Batu Pahat:
Kolej Universiti Teknologi Tun Hussein Onn. (Unpublished).
Peck, R.B., Hanson, W.E., and Thornburn, T.H. (1974). Foundation Engineering.
New York: John Wiley and Sons, Inc. 514.
Poulos, H.G. and Davis, E.H. (1980). Pile Foundation Analysis and Design. New
York: John Wiley and Sons, Inc.
Scheil, T.J. (1979). Long Term Monitoring of a Building over the Deep Hackensack
Meadowland Varved Clays. Presented at Converse Ward Davis Dixon, Inc.
Technical Seminar. California.
Skempton, A.W. (1986). SPT Procedures and the Effects in Sands of Overburden
Pressure, Relative Density, Particle Size, Aging and Overconsolidation.
Geotechnique. Vol. 36, No. 3. 425-447.
Tomlinson, M.J. (1971). Some effects of Pile Driving on Skin Friction. Proceedings
of Conference of Behavior of Piles. London: ICE. 107-114.
Tomlinson, M.J. (1987). Pile Design and Construction Practices. 3rd edition.
London: Palladian Publication.
Tolosko, T.A. (1999). Extrapolation of Pile Capacity from Non-Failed Load Test.
University of Massachusetts Lowell: Master Thesis.
83
Vijayvergiya, V.N., and Focht, J.A., Jr. (1972). A New Way to Predict Capacity of
Piles in Clay. Proceedings of Conference of Offshore Technology. Houston:
Forth Offshore Technology. Conference Paper 1718.
Whitaker, T., (1976). The Design of Piled Foundations. 2nd edition. Oxford:
Pergamon Press. 135-150.
Appendix A1
Summary of Average Pile Top Settlement for Test Pile TP3C
85
Appendix A2
Bearing Capacity of Test Pile TP3C from Load Test Interpretation Method
Chin's Method
0.008
0.006
0.005
0.004
0.003
0.002
0.001
0
0 5 10 15 20 25 30 35 40 45
Settlement (mm)
0.003
Settlement^0.5/Load (mm^0.5/kN)
0.002
y = -0.00002x + 0.00169
0.001
0
0 10 20 30 40 50
Settlement (mm)
6000
5000
Load (kN)
4000
3000
2000
1000
0
0 10 20 30 40 50
Settlement (mm)
From the graph, it is estimated the ultimate bearing capacity is 6600 kN.
6000
5000
Load (kN)
4000
3000
2000
1000
0
0 10 20 30 40 50
Settlement (mm)
From the graph, it is estimated the ultimate bearing capacity is 6400 kN.
87
De Beer's Method
10000
Load (kN)
Qp
1000
100
1 10 100
Settlement (mm)
From the graph, it is estimated the ultimate bearing capacity is 4300 kN.
88
Appendix A3
Bearing Capacity of Test Pile TP3C from Pile Driving Formulae
Janbu Formula
W R H
Janbu formula, Qp =
Ku S
= 3787 kN
e
0.5
where Ku = C d 1 + 1 + = 19.9
d
C
Wp
Cd = 0.75 + 0.15 = 0.98
WR
WR HL
e = = 366
Ap E p S 2
WR H WR + n 2WP
ENR formula, Qp = x
S +C WR + W p
= 2392 kN
89
Appendix A4
Bearing Capacity of Test Pile TP3C from Static Analysis (Meyerhof Method)
0 9.0 m
Loose sand, average unit weight, avg = 16.5 kN/m2
Navg = 3
9.0 m 25.8 m
Soft clay, average unit weight, avg = 17.5 kN/m2
Navg = 4
Cu = 20 kN/m2
25.8 m 38.7 m
Medium dense sand, average unit weight, avg = 18.75 kN/m2
Navg = 17
38.7 m 42.0 m
Very dense sand, average unit weight, avg = 18.75 kN/m2
Navg = 176
Since Qtl < Qtu, the point bearing capacity, Qt is 2566 kN.
Appendix B1
Summary of Average Pile Top Settlement for Test Pile TP5
93
Appendix B2
Bearing Capacity of Test Pile TP5 from Load Test Interpretation Method
Chin's Method
0.01
0.008
Settlement/Load(mm/kN)
0.007
0.006
0.005
0.004
0.003
0.002
0.001
0
0 10 20 30 40 50
Settlement (mm)
0.004
Settlement^ /Load (mm^ /kN)
0.5
0.003
0.002
y = -0.00002x + 0.00195
0.5
0.001
0
0 10 20 30 40 50
Settlement (mm)
6000 Qp
5000
4000
Load (kN)
3000
2000
1000
0
0 10 20 30 40 50
Settlement (mm)
From the graph, it is estimated the ultimate bearing capacity is 5400 kN.
5000
4000
Load (kN)
3000
2000
1000
0
0 10 20 30 40 50
Settlement (mm)
From the graph, it is estimated the ultimate bearing capacity is 5200 kN.
95
De Beer's Method
10000
Load (kN)
Qp
1000
100
1 10 100
Settlement (mm)
From the graph, it is estimated the ultimate bearing capacity is 3000 kN.
96
Appendix B3
Bearing Capacity of Test Pile TP5 from Pile Driving Formulae
Janbu Formula
W R H
Janbu formula, Qp =
Ku S
= 2682 kN
e
0.5
where Ku = C d 1 + 1 + = 105.6
d
C
Wp
Cd = 0.75 + 0.15 = 0.97
WR
WR HL
e = = 11299
Ap E p S 2
= 1875 kN
97
Appendix B4
Bearing Capacity of Test Pile TP5 from Static Analysis (Meyerhof Method)
0 10.5 m
Loose sand, average unit weight, avg = 16.5 kN/m2
Navg = 3
10.5 m 26.0 m
Soft clay, average unit weight, avg = 17.5 kN/m2
Navg = 2
Cu = 10 kN/m2
26.0 m 43.0 m
Dense sand, average unit weight, avg = 18.75 kN/m2
Navg = 32
= 33
99
Since Qtl < Qtu, the point bearing capacity, Qt is 187 kN.
Appendix C1
Summary of Average Pile Top Settlement for Test Pile TP9
101
Appendix C2
Bearing Capacity of Test Pile TP9 from Load Test Interpretation Method
Chin's M e thod
0.02
y = 0.00017x + 0.00630
Settlement/Load (mm/kN)
0.01
0
0 5 10 15 20 25 30 35 40 45 50
Settlement (mm)
0.006
0.005
^ /kN)
0.5
0.004
ent^ /Load(mm
0.002
0.001
0
0 10 20 30 40 50
Settlement (mm)
4000 Qp
3500
3000
2500
Load (kN)
2000
1500
1000
500
0
0 10 20 30 40 50
Settlement (mm)
From the graph, it is estimated the ultimate bearing capacity is 3300 kN.
4000 Qp
3500
3000
2500
Load (kN)
2000
1500
1000
500
0
0 10 20 30 40 50
Settlement (mm)
From the graph, it is estimated the ultimate bearing capacity is 3200 kN.
103
De Beer's Method
10000
Qp
Load (kN)
1000
100
1 10 100
Settlement (mm)
From the graph, it is estimated the ultimate bearing capacity is 2100 kN.
104
Appendix C3
Bearing Capacity of Test Pile TP9 from Pile Driving Formulae
Janbu Formula
W R H
Janbu formula, Qp =
Ku S
= 1545 kN
e
0.5
where Ku = C d 1 + 1 + = 12
d
C
Wp
Cd = 0.75 + 0.15 = 0.9
WR
WR HL
e = = 128
Ap E p S 2
= 1078 kN
105
Appendix C4
Bearing Capacity of Test Pile TP9 from Static Analysis (Meyerhof Method)
0 6.8 m
Loose sand, average unit weight, avg = 16.5 kN/m3
Navg = 3
6.8 m 19.0 m
Soft clay, average unit weight, avg = 17.5 kN/m3
Navg = 2
cu = 10 kN/m2
19.0 m 39.0 m
Medium dense sand, average unit weight, avg = 18.75 kN/m3
Navg = 13
39.0 m 45.0 m
Very dense sand, average unit weight, avg = 20.5 kN/m3
Navg = 165
Since Qtl < Qtu, the point bearing capacity, Qt is 1917 kN.
Appendix D1
Summary of Average Pile Top Settlement for Test Pile TP10
109
Appendix D2
Bearing Capacity of Test Pile TP10 from Load Test Interpretation Method
Chin's Method
0.06
0.04
0.03
0.02
0.01
0
0 5 10 15 20 25
Settlement (mm)
0.025
0.024
0.023
0.022
0.021
Settlement^0.5/Load (mm^0.5/kN)
0.02
0.019
0.018
0.017
0.016
0.015
0.014
0.013 y = -0.00015x + 0.01242
0.012
0.011
0.01
0.009
0.008
0.007
0.006
0.005
0.004
0.003
0.002
0.001
0
0 5 10 15 20 25
Settlement (mm)
500 QP
450
400
350
Load (kN)
300
250
200
150
100
50
0
0 5 10 15 20 25
Settlement (mm)
From the graph, it is estimated the ultimate bearing capacity is 420 kN.
500
QP
450
400
350
Load (kN)
300
250
200
150
100
50
0
0 5 10 15 20 25
Settlement (mm)
From the graph, it is estimated the ultimate bearing capacity is 390 kN.
111
De Beer's Method
1000
QP
Load (kN)
100
10
1 10 100
Settlement (mm)
From the graph, it is estimated the ultimate bearing capacity is 230 kN.
112
Appendix D3
Bearing Capacity of Test Pile TP10 from Pile Driving Formulae
Janbu Formula
W R H
Janbu formula, Qp =
Ku S
= 350 kN
e
0.5
where Ku = C d 1 + 1 + = 1.7
d
C
Wp
Cd = 0.75 + 0.15 = 0.8
WR
WR HL
e = = 0.15
Ap E p S 2
= 272 kN
113
Appendix D4
Bearing Capacity of Test Pile TP10 from Static Analysis (Meyerhof Method)
0 7.8 m
Loose sand, average unit weight, avg = 16.5 kN/m2
Navg = 3
7.8 m 19.0 m
Soft clay, average unit weight, avg = 17.5 kN/m2
Navg = 3
cu = 15 kN/m2
19.0 m 41.2 m
Medium dense sand, average unit weight, avg = 18.75 kN/m2
Navg = 26
41.2 m 46.0 m
Very dense sand, average unit weight, avg = 18.75 kN/m2
Navg = 178
This pile is carry by skin resistance alone as the soft clay is not capable of generating
end bearing for the pile. Thus, Qp = Qs.