GUIDELINES

for the
INTERPRETATION AND
ANALYSIS
of the
STATIC LOADING TEST

A Continuing Education Short Course Text

DEEP FOUNDATIONS INSTITUTE

GUIDELINES
for the
INTERPRETATION AND ANALYSIS
of the
STATIC LOADING TEST
Prepared by the
CONTINUING EDUCATION COMMITTEE
of
DEEP FOUNDATIONS INSTITUTE
By: BENGT H. FELLENIUS, P.E.
ANNA Geodynamics, Inc.
and
University of Ottawa
Ottawa, Ontario, Canada
First Edition

A Continuing Education Short Course Text

Deep Foundations Institute
P.O. Box 281

Sparta, NJ 07871
Copyright 1990

Deep Foundations Institute

Printed in USA

TABLE OF CONTENTS
1. INTRODUCTION
2. EXECUTION OF THE STATIC LOADING TEST
. 1 Introduction
.2 Testing Arrangement
.3 ASTM Testing Procedures
.4 Reporting of Results
3. SAFETY CONSIDERATIONS
. 1 Safety to persons
.2 Safety to the Test
.3 Point of warning
4. INTERPRETATION OF FAILURE LOAD
5. FACTOR OF SAFETY AND ACCEPTANCE CRITERIA
. 1 Factor of Safety
.2 Acceptance Criteria
6. INSTRUMENTATION OF THE PILE
. 1 Introduction
.2 Telltale Instrumentation
.3 Strain Gage Instrumentation
.4 Load Cells
7. DETERMINATION OF 'ELASTIC MODULUS
. 1 Basic principles of stress-strain analysis
.2 Actual test results
.3 Mathematical relations
.4 Example from a pile with non-constant modulus
8. INTERPRETATION AND EVALUATION OF TELLTALE DATA
. 1 Basic analysis
.2 Leonards-Lovell method for load distribution
.3 Example of Leonards-Lovell analysis
9. INFLUENCE OF RESIDUAL COMPRESSION
.1 Residual compression in a Leonards-Lovell analysis
.2 Residual compression from a push-pull test combination ...
11. REFERENCES
12. EXAMPLES

ii

Page
1
1
1
2
4
8
8
8
10
10
11
19
19
20
21
21
22
25
26
27
27
28
30
32
34
34
35
38
39
39
41
42
45

1. INTRODUCTION
The design of pile foundations is much more commonly verified by
means of a full-scale test, than is the design of other foundation units.
The reason is not that our knowledge of pile behavior is more uncertain
than our knowledge of other foundation types making the verification
necessary, but more that the loads in a structure are more concentrated
to single foundations in a structure founded on piles as opposed to structures on footings or mats. Therefore, should a pile cap fail or move, the
adverse consequence of this is often drastic, as the piled structure has
little freedom to transfer its need for support to other foundation units.
Consequently, it becomes important to assure the design of piled foundations.
In many, maybe in most instances, the static pile loading test is routine
and geared toward determining an at least capacity, only. However, in
these times there is an ever increasing liability of the professional, demands for increased economy of the foundation, and frequent lack of the
involvement in the test of the experienced old-timer exercising good
judgment. Furthermore, modern pile design is leaving the single, simple
concept of capacity and requires more information from the test to assist
in determining aspects of long-term behavior and settlement. Therefore,
even the straight forward, routine static loading test requires improved
planning, execution, and analysis.
These guidelines are written with the objective of presenting views
on the execution and analysis of the static loading test as it should be
performed in routine situations and what to consider when expanding
the test to provide more answers to the design engineer than just addressing the total capacity of the pile.
2. EXECUTION OF THE STATIC LOADING TEST
2.1 Introduction
For many good reasons, a static loading test must be carried out in
accordance with good engineering practice and under experienced
supervision. In North America and in most parts of the world, this
means that the test must be in agreement with the recommendations
in standards published by the American Society for Testing and Materials, ASTM. For static axial testing of piles, the ASTM has published two
standards, one for testing in compression (push) and one for testing in
tension (pull) with designations D-1143-81 and D-3689-83, respectively1). The aspects presented in this guide apply in equal degree to
testing in push as well as in pull.
1

The ASTM has also published standards for lateral and dynamic testing of piles
with the designations D 3966 and D 4945, respectively.
1

In essence, each ASTM standard contains two parts: the first presents the arrangement of reaction, minimum distances, measuring devices to use, etc., and the other presents procedures of applying the
load and measuring the responsethe actual testing procedure.
FIRST PART
Apparatus for applying load
Apparatus for measuring movement
SECOND PART
Loading procedures
Procedures for measuring movement
Safety requirements
Report
The first part of the ASTM standard provides very little leeway in
how to arrange a test and what requirements to follow. However, and
despite that the recommendations are very lenient, they are often, too
often, violated. Some of the more important aspects of the recommended arrangements will be discussed below. However, it is absolutely necessary for the person in charge of a static loading test to be
familiar with all of the applicable ASTM standard.
The second part of the ASTM standard allows considerable flexibility in the choice of procedure for applying the load. The standard
addresses both the slow and the quick maintained-load methods, as
well as the constant-rate-of-penetration method.
2.2 Testing Arrangement
The ASTM standards are very detailed and yet concise and it would
be impractical to summarize the various clauses. However, a few important points will be made as emphasis to the requirements given in
the standards.
Normally, the test load is produced by a hydraulic jack activated by
a pump. The pressure, as registered by means of a pressure gage
(manometer), is used to determine the magnitude of the applied load.
To ensure adequate accuracy, the pressure gage for the test should be
selected so that the maximum pressure applied at the test is about 80
percent of the gage range.
As an option in the ASTM standards, use of a separate load cell is
suggested for determining the load. However, the pressure in the jack
is a very inaccurate measure of the load applied. Even when using a
properly calibrated manometer and jack system, the load determined
from the jack pressure usually overestimates the applied load, and
more often than not, the error is in the range of 10% through as
much as 20% and more (Fellenius 1980; 1984). For most engineering
projects, a potential load error of this magnitude is unacceptable.
Therefore, a load cell should always be used even for routine projects
2

Fig. 2.1 Illustration of error of

applied load

and the jack pressurewhich must be recordedserves as a back-up

value.
As a representative example of what, to expect from the equipment
used by the industry today, Fig. 2.1 shows the difference between the
load determined from the jack pressure and the load determined by
the load cell, as plotted against the load given by the load cell.
The reasons for the load error is that the jacking system is required
to do two things at the same time, i.e., both provide the load and
measure it, and the jack having moving parts is considerably less accurate than without moving parts. Also, to extend the jack piston, friction
has to be overcome and part of the jack pressure is used for this
purpose. Many measurement results similar to those shown in Fig. 2.1
make it obvious that if one wants to ensure that the error in the applied
load is not too large, a load cell must be used. A calibration of the jack
and pressure gage (manometer) for one pile is not applicable when
performed on even a neighboring test pile. When calibrating testing
equipment in the laboratory, it is ensured that no eccentric loadings,
bending moments, or temperature variations influence the calibration.
However, in the field, all of these factors are at hand to influence the
test results. The extent of the error will be unknown unless a load cell
is used.
Naturally, many structures are safely supported on piles which have
been tested with erroneous loads, and, as long as we are content to
stay with the old rules, loads, and piling systems, we do not need to
improve the precision. The error is included in the safety factor. That
is why factors as large as 2.0 and 2.5 are applied and such numbers
are really more ignorance factors than safety factors. However, if we
want to economize and continue to increase the allowable loads, as
our geotechnical know-how increases, we cannot accept potential er3

rors as large as 20%. Therefore, use of a load cell to monitor the load
applied in a loading test is imperative.
The fact that a load cell is used is no guarantee for precise loads.
Some load cells are very sensitive to eccentric loading and to temperature variation and are, therefore, unsuitable for field use.
It must be remembered that the minimum distances from the supports of reference beam to the pile and the platform, etc., as recommended in the ASTM Designation, are really minimum values, which
most often do not give errors of much concern for ordinary testing, but
which are too short for research or investigative testing purposes.
The measuring of movement of the pile head is normally determined
in relation to two reference beams. The most common shortcoming of
a test is that the reference beam is not arranged in accordance with
the ASTM standards: the supports of the beams, and therefore also the
measured movements, are influenced by the reaction load or reaction
system; the sun is let to shine on the beams; the two beams are connected instead of independently supported; no smooth bearing surface, such as glass, is used for the dial-gages; the gage stems are too
short; all gages are adjusted simultaneously causing a loss of test
continuity, etc. Before starting the test, the person in charge and responsible for the test must ensure and verify that the test set-up is
in conformity with all aspects of the recommendations given in the
applicable ASTM standard.
2.3 The ASTM Testing Procedures
Until recently in North America, the most common test procedure
has been the slow maintained-load procedure referred to as the "standard loading procedure" in the ASTM D-1143 Standard in which the
pile is loaded in eight equal increments up to a maximum load, usually
twice the predetermined allowable load.
The "standard loading procedure" is often thought of as the ASTM
procedure. However, the ASTM D1143-81 and D 3689-83 Standards
present six additional procedures of applying the load. Of these, the
first three are variations of the slow maintained load procedure. The
remaining three are: the constant-rate-of-penetration (C.R.P.) procedure, the quick-test procedure, and the constant-movement-increment
procedure.
In the "standard loading procedure", each increment is maintained
until a minimum movement is reached, commonly referred to as the
"zero movement". The minimum movement is defined as 0.01 in/h or
0.002 in/10 min). The final load, the 200 percent load, is maintained
for 24 hours. The "standard procedure" is very time consuming requiring from 30 to 70 hours to complete. It should be realized that the
words "zero movement" are very misleading, as the movement rate of
4

0.01 in/h is equal to a movement as large as 7 ft/yr, well beyond any

conceivable yearly settlement rate.
The "standard procedure", also called "Slow Maintained-Load Test"
or just "Slow Test", can be speeded up by using the method proposed
by Mohan et al. (1967), where the load (jack pressure) is allowed to
reduce to and stabilize at a lower value rather than being maintained
by pumping. The stabilized value is taken as the load applied to the
pile.
Housel (1966) proposed that each of the eight increments be maintained exactly one hour whether or not the "zero" movement has been
reached (called the constant-time-interval-loading procedure).
Applying the load at equal time intervals allows an analysis of movement with time, which is not possible with the "standard procedure":
For each load increment, plot the magnitude of movement obtained
during the last 30 minutes of the one-hour load duration versus the
applied load. Initially, the values will fall on a more or less straight line.
At one load level, however, two approximately straight lines will be
obtained. Provided that the test has approached failure, that is. The
intersection of the two lines is termed yield value.
A test according to Housel's procedure takes a full day to perform.
The points on the curve are still very few, but Housel's procedure is a
definite improvement of the "standard procedure" and it is one of the
seven optional procedures in the ASTM Designation D-1143. However,
it is better to apply, say, 16 equal increments of a half hour duration
as opposed to the "standard" 8 equal increments of one hour duration;
the rate of loading is the same, but the load-movement curve is better
defined. A yield value similar to the one obtained from the movement
during the last 30 minutes of the one-hour increment can be evaluated
from the movement during the last 15 minutes of the 30-minute increment provided that readings are taken often enough and that they are
accurate. But why stop at 16 increments applied at every 20 minutes,
when 32 increments are applied every 15 minutes determine the load
deformation curve even better? The load is still applied at a constant
rate in terms of tons per hour and no principal change is made. An
additional benefit is that a small increment will not shock the soil and
change the load transfer characteristics in contrast to the effect of a
large increment applied quickly.
Actually, the duration of each load is less important, be it one hour
or 15 minutes. The importance is that the duration of each load is the
same. From this realization, we can progress to the one that even
shorter time intervals are possible without impairing the test. Further,
by using as short time intervals as practically possible, the influence
on the results of time dependency is reduced. When it is desirable to
study the time dependency, drained test conditions, creep aspects,
etc., the test duration should be measured in weeks, months, or even
5

years. A 48-hour or 72-hour test does not give any information on timedependent behavior of the pile and results only in confusion.
A test which consists of load increments applied at constant time
intervals of 5, 10, or 15 minutes, is called Quick Maintained-Load
Test or just "Quick Test" and is from both technical, practical, and
economical views superior to the Slow Test. This procedure is also
included in the mentioned two ASTM standards.
A Quick Test should aim for at least 20 load increments with the
maximum load determined by the amount of reaction load available or
the capacity of the pile. In routine proof tests, the maximum test load
is commonly chosen to 200 percent of the intended allowable load.
For most tests, however, it is preferable to carry the test beyond the
200 percent value.
As to time intervals, for ordinary test arrangements, where only the
load and the pile head movement are monitored, time intervals of 5
minutes are suitable and allow for taking 2 to 4 readings for each
increment. The ASTM standards permit intervals of time between load
increments as short as 2 minutes. While no technical disadvantage is
associated with a very short time interval as long as the intervals are
equal, unless data acquisition apparata are employed having a rapid
scanning capability, practical difficulties arise when using intervals
shorter than 5 minutes.
When testing instrumented piles, where the instruments take a while
to read (scan), the time interval may have to be increased. To go beyond
15 minutes, however, should not be necessary. Nor is it advisable,
because of the potential risk for influence of time dependent movements, which may impair the test results. Usually, a Quick Test is
completed within two to five hours.
A test which has gained much use in Europe is the constant-rate-ofpenetration test (C.R.P. test), first proposed internationally for piles by
Whitaker (1957; 1963) and Whitaker and Cooke (1961). Manuals on
the C.R.P. test have been published by the Swedish Pile Commission
(1970) and New York Department of Transportation (1974). In the
C.R.P. test, the pile head is forced to move at a predetermined rate,
normally 0.02 in/min (0.5 mm/min), and the load to achieve the movement is recorded. Readings are taken every two minutes and the test
is carried out to a total movement of the pile head of two to three
inches (50 to 75 mm) or to the maximum capacity of the reaction
arrangement, which means that the test is completed within two to
three hours.
The C.R.P. test has the advantage over the Quick Test that it enables
an even better determination of the load-movement curve. This is of
particular value in testing shaft bearing piles, when sometimes the
force needed to achieve the penetration gets smaller after a peak
value has been reached. It also agrees with the testing in most other
6

engineering fields, which regularly use the C.R.P. procedure to determine strength and stress-strain relations. A C.R.P. test is best performed with a mechanical pump that can provide a constant and nonpulsing flow of oil. Ordinary pumps with a pressure holding device,
manual or mechanical, are less suitable because of unavoidable loading variations. Also, the absolute requirement of simultaneous reading
of all load and deformation gages (changing continuously) could be
difficult to achieve without a well trained staff. For these reasons, the
Quick Maintained-Load Test is preferable. For instrumented piles, a
C.R.P. test is not suitable unless used with a very fast data acquisition
unit.
A fourth test procedure is cyclic testing. For details cyclic procedures, see Fellenius (1975), and references contained therein. In routine tests, cyclic loading, or even single unloading and loading phases
must be avoided. It is a common misconception that unloading a pile
every now and then according to some more or less "logical" scheme
will provide information on the toe movement. That it will not, but it
will result in a destruction of the chances to analyze the test results
and the pile load-movement behavior. In non-routine tests and for a
specific purpose, cyclic testing can be used, but then after completion
of an initial test and when having the pile instrumented with at least a
telltale to the pile toe.
To emphasize: there is absolutely no logic in believing that anything
of value can be obtained from cyclic testing consisting of one or a few
occasional unloadings, or one or a few resting periods at certain load
levels, when considering that we are testing a unit that is subjected to
the influence of several soil types, is already under stress of unknown
magnitude, exhibits progressive failure, etc., and when all we know is
what we apply and measure at the pile head, while we really are interested in what happens at the pile toe.
The constant-movement-increment-loading procedure is rather special and of little interest to engineering practice.
Unloading procedure. When unloading the pile, a simple procedure
is recommended, as follows: Reduce (leak) the pressure in the pump
in decrements and take readings of the pressure and dial gages values
at each level of reduced load as obtained. It is important that the jack
piston does not reverse its direction of travel, that is, the pressure must
not be increased even if the desired pressure or load level is missed.
The first two decrements are to be small in order to enable the influence of the piston friction, if any, to be determined. Then the unloading
continues in some four or five larger decrements until only a small load
still is on the pile head, which is then unloaded in two small decrements. Before removing the gages, a final reading is taken after the
pile has been under zero load for about five minutes.
7

2.4 Reporting of Results

The results of a static loading test must be presented in a report
conforming to the applicable requirements of the ASTM standards.
More specifically, the immediate test results should be provided in a
table showing pertinent pile identification and the times for start and
finish of the test, and, for each load increment, the load cell readings
and load cell loads, the jack pressure readings and jack loads, move
ments measured for each dial gage and averaged head movements,
and other recorded data.
The load-movement readings should be presented in a diagram in
the first quadrant with the load on a linear scale on the ordinate (vertical
axis) and the movement on a linear scale on the abscissa (horizontal
axis).
To facilitate the interpretation of the test results, the diagram scales
should be selected so that the line representing the calculated elastic
line of the pile (the "column line") will be inclined at an angle of about
20 degrees to the load axis. The slope of the elastic line is computed
from the following expression:
(2.1)
where
=
Q =
L=
A=
E=

calculated elastic shortening

applied load
pile length
pile cross sectional area
elastic modulus of the pile material

The calculation is best performed inserting all parameters in base

units: Q in N (lb), L in m (inch), A in m2 (in2), and E in Pa (psi), which
gives d in N/m (lb/inch). Then, division by 1,000,000 (2,000) gives d
in KN/mm (tons/inch) and the elastic line is simple to draw.
Often the time-movement curve is of interest. Also this curve should
be drawn in the first quadrant. Plot time in a linear scale on the abscissa
and movement in a linear scale on the ordinate.
3. SAFETY CONSIDERATIONS
Safety should be foremost in mind when performing a static loading
test. First of all, the safety of the persons present at the test, and, then,
the safety of the test itself.
3.1 Safety to Persons
There are numerous accidents occurring at static tests, which have
caused serious injury and death to persons. Common for them all
8

is that with some foresight and precautions they could have been
avoided.
The immediate and obvious detail to consider is the stack of separate items placed between the pile head and the main reaction beam
consisting of a jack, a load cell, a swivel plate, and some spacing
plates. It is impossible to ensure that the pile is perfectly perpendicular
to the main beam, that the jack, load cell, swivel plate, and spacers are
placed absolutely concentric and in perfect alignment with each other
and the pile, or that the geometric center of the system coincides with
the force center. The stack of individual parts is usually a good deal
less stable than it appears and its parts can easily fall. Wearing a hard
hat and toe reinforced shoes is advisable although such precaution
does not replace care because they do not provide much protection
from a falling steel plate. Consider also that the system is subjected to
loads which can amount to several hundred tons, which builds large
energy into the system that is released should a plate slip out of the
stack or the pile fail suddenly, which can hurl the items around injuring
the bystanders. The stack must be retained by a cage protecting the
persons positioned near the pile, and/or all parts be secured by a wire
or rope that will catch them should they fall.
Other safety concerns rest with the arrangement of load on the
reaction platform. A good rule-of-thumb is not to build the platform
load higher than its width. The foundation of the loaded platform must
be safe. Many testing failures are preceded by, even originate in, a
shift of a platform foundation.
When reaction is provided by anchors, make sure that should one
anchor fail, the others must not act as a suddenly released slingshot
sending beams and material swinging through the air.
If at all possible, all gages should be read from a distance to eliminate the need for going in under a test platform or close to a test set
up. Use binoculars or a camera tele lens.
It is a good approach to rope off the immediate test site and proclaim
the area off limit to everyone not actively participating in the performance of the test. This goes for uninvolved curious onlookers, as well
as for the involved ones such as the client and the owner. All persons
involved should be on alert for strange noise and movements of the
entire system. The person in charge should not become so absorbed
by the task of collecting data to forget occasionally to walk around and
visually inspect the test set-up for signs of distress concentrating on
the following questions:
1. Is everything plumb? In all directions?
2. Is there enough weight on the frame?
3. Is the jack in line with the pile?
9

4. Has the dial gage stems moved sideways indicating lateral instability
5. Are there any signs of leak of hydraulic oil from jack or pump?
If the answer to any of the questions is "NO", you must take the time
to investigate what the cause is and you may have to abort the test
and rectify the situation. Do not hesitate about being assertive. An
unsafe situation is not to be taken lightly and the need for safety must
not be underestimated.
3.2 Safety to the Test
Also little things that do not harm any person may jeopardize the
test. For instance, if the reference beam is not protected from sunshine,
the movement readings may be wrong. Or, if a reference beam is
disturbed (someone puts his foot up), the data may be spoiled from the
dial gages connected to that beam. Then, if the reference beams have
been interconnected, the disturbance of one beam may offset the position of also the second beam and all data may be lost. Note, stiffening
the reference beams by connecting them is a violation of the ASTM
recommendations.
Also for reasons of safety to the test, it is a good idea to rope off the
test area and make it off-limit for everyone not directly involved.
3.3 Points of Warning
Performing a static pile loading test can be a risky process. The test
arrangement must be designed and built by persons having experience
from this type of work. Below is offered a checklist for reference to the
danger points to consider before starting the test.
1. Check that the intended maximum load is smaller than the structural strength of the pile by a safe margin.
2. Check that the maximum test load is smaller than about 90
percent of the jack capacity.
3. Check that the maximum test load is obtained at a jack pressure
of about 80 percent of the maximum capacity of the pressure
gage (manometer).
4. Check that the reaction load available is about 20 percent larger
than the maximum test load.
5. Check that the package between the pile head and the main test
beam (jack, load cell, swivel plate, and spacer plates) are secured
in a cage or otherwise prevented from falling to the ground
should they become loose during the test.
6. Previously used test beams should be inspected to ensure that
their strength has not been reduced by cutting or corrosion.
7. Cut all unnecessary temporary welds within the reaction system
before starting the test.
10

Fig. 4.1 Illustration of the

conceived failure load's
dependency on the
draughting scale

8. Do not allow welding or torch cutting close to tension steel such

as high strength thread bars. The heat may weaken the steel
and create a dangerous situation.
9. Ensure that all personnel wears a hard hat.
10. Note that hydraulic valves or connections must not be tightened
or otherwise adjusted while pressure is on. Jets of hydraulic
fluid can cause considerable injury; whipping hydraulic hoses
can kill.
4. INTERPRETATION OF FAILURE LOAD
For a pile which is stronger than the soil, a failure load by plunging is
reached when rapid movement occurs under sustained or slightly increased load. However, this definition is inadequate, because plunging
requires large movements and the ultimate load reached is often less a
function of the capacity of the pile-soil system and more a function of the
man-pump system.
A common definition of failure load has been the load for which the
pile head movement exceeds a certain value, usually 10% of the diameter
of the pile. This definition does not consider the elastic shortening of the
pile, which can be substantial for long piles, while it is negligible for short
piles. In reality, a movement limit relates only to the allowable movement
allowed by the superstructure to be supported by the pile, and not to the
capacity of the pile.
Sometimes, the failure value is defined as load value at the intersection
of two straight lines, approximating an initial pseudo-elastic portion of the
load-movement curve and a final pseudo-plastic portion. This definition
results in interpreted failure loads, which depend greatly on judgement
and, above all, on the scale of the graph. Change the scales and the
failure value changes also, as illustrated in the load-movement diagram
presented in Fig. 4.1. A loading test is influenced by many occurrences,
but the draughting manner should not be one of these.
Without a proper definition, interpretation becomes a meaningless
venture. To be useful, a definition of failure load must be based on some
mathematical rule and generate a repeatable value that is independent
11

Fig. 4.2 Load-Movement diagram

from a C.R.P. test

of scale relations and the judgment call of the individual interpreter.

Furthermore, it has to consider shape of the load-movement curve or, if
not, it must consider the length of the pile (which the shape of the curve
indirectly does).
The results given as a load-movement curve in Fig. 4.2 will be used to
present nine different definitions of failure. The example is taken from a
test on a 12-inch (300 mm) diameter concrete pile installed through 60
ft (18 metre) of sensitive clay, 10 ft (3 metre) of clay silt and 6 ft (2 metre)
of silt. The pile was tested six weeks after driving. Method of testing was
the C.R.P.-method. The pile started to plunge when test load reached 200
tons, but at the maximum load of 206 tons the load necessary to achieve
the movement was still increasing.
In Fig. 4.3 is applied a methodoffset limitproposed by Davisson
(1972), also referenced by Peck et al. (1974). Davisson's limit load is
defined as the load corresponding to the movement which exceeds the
elastic compression of the pile by a value of 0.15 inch (4 mm) plus a
factor equal to the diameter of the pile divided by 120. For the 12-inch
diameter example pile, the value is 0.25 inch (6 mm). The Davisson limit
was developed in conjunction with wave equation analysis of driven piles
and dynamic measurements. It has gained a widespread use in phase
with the increasing popularity of this method of analysis. It is primarily
intended for test results from driven piles tested according to quick
methods.
Fig. 4.4 gives a method proposed by Chin (1970; 1971) for piles in
applying the general work by Kondner (1963). To apply the Chin-Kondner
method, divide each movement value with its corresponding load value
and plot the resulting value against the movement. As shown, after some
12

Fig. 4.3 Davisson's offset limit

method

Fig. 4.4 Chin-Kondner's method

initial variation, the plotted values fall on straight line. The inverse slope
of this line is the Chin failure load.
Generally speaking, two points will determine a line and third point on
the same line confirms the line. However, beware of this statement when
using Chin's method. It is very easy to arrive at a false Chin value if applied
too early in the test. Normally, the correct straight line does not start to
13

Fig. 4.5 DeBeer's method

Fig. 4.6 Brinch-Hansen's 90%

criterion

materialize until the test load has passed the Davisson limit. As a rule, the
Chin Failure load is about 20% to 40% greater than the Davisson limit.
When this is not a case, it is advisable to take a closer look at all the test
data.
The Chin method is applicable on both quick and slow tests, provided
constant time increments are used. The ASTM "standard method" is
therefore usually not applicable. Also, the number of monitored values
are too few in the "standard test"; the interesting development could well
appear between the seventh and eighth load increments and be lost.
Fig. 4.5 presents a method proposed by DeBeer (1967) and DeBeer
and Wallays (1972), where the load movement values are plotted in a
double logarithmic diagram. When the values fall on two approximately
straight lines, the intersection of these defines the failure value. DeBeer's
method was proposed for slow tests.
Fig. 4.6 illustrates a method proposed by BrinchHansen (1963), who
defines failure as the load that gives twice the movement of the pile head
14

Fig. 4.7 Brinch-Hansen's method

(the 80% criterion)

as obtained for 90% for that load. This method, also known as the 90%criterion, has gained widespread use in the Scandinavian countries
(Swedish Pile Commission, 1970).
BrinchHansen (1963) also proposed an 80%-criterion defining the failure load as the load that gives four times the movement of the pile head
as obtained for 80% of that load. The 80%-criterion failure load can be
estimated by extrapolation from the load-movement curve directly, which
gives about 210 tons. The failure load according to the BrinchHansen
80-percent criterion can also be more accurately determined in a plot
which is very similar to that of the Chin-Kondner plot. Fig. 4.7 shows this
plot for the test data from the example test, where the square root of
each movement value is divided with its corresponding load value and
the resulting value is plotted against the movement.
The following simple relations can be derived for computing the ultimate failure, Qu, according to the BrinchHansen 80%-criterion:
(4.1)

(4.2)
Where
Qu
Au
C1
C2

=
=
=
=

failure load
movement at failure
slope of the straight line
y-intercept of the straight line

When using the BrinchHansen 80%-criterion, it is important to check

that the point 0.80 Q u /0.25 Au indeed lies on or near the measured loadmovement curve.
In the example case, Qu is 211 tons, which agrees well with the value
determined from the load-movement curve, directly.
Notice that both the BrinchHansen's 80%-criterion and the Chin
method allow the later part of the curve to be plotted according to a
15

Fig. 4.8 Mazurkiewicz's method

mathematical relation, and, which is often very tempting, they make an

"exact" extrapolation of the curve possible. That is, it is easy to fool
oneself and believe that the extrapolated part of the curve is as true as
the measured.
In Fig. 4.8, the method by Mazurkiewicz (1972) is illustrated. First, a
series of equally spaced lines parallel to the load axis are arbitrarily
chosen and drawn to intersect with the load-movement curve. Then, from
each intersection, a line is drawn parallel to the movement axis, toward
and crossing the load axis. At the point of intersection with the load axis
of each such line, a 45 line is drawn to intersect with the line above.
These intersections fall, approximately, on a straight line which own intersection with the load axis defines the failure load. Mazurkiewicz' method
is also, understandably, called "the method of multiple intersections".
When drawing the line through the intersections, some disturbing freedom of choice is usually found.
Fig. 4.9 illustrates a simple definition by Nordlund (1966) and Fuller
and Hoy (1970). The failure load is equal to the test load for where the
load-movement curve is sloping 0.05 inch per ton.
Fig. 4.9 also shows a development of the above definition proposed
by Butler and Hoy (1977) defining the failure load as the load at the
intersection of the tangent sloping 0.05 inch/ton, and the tangent to the
initial straight portion of the curve, or to a line that is parallel to the
rebound portion of the curve. As the latter portion is more or less parallel
to the elastic line (compare Fig. 4.3), Fellenius (1980) suggests that the
intersection be that of a tangent parallel to the elastic line, instead.
The Nordlund/Fuller and Hoy method penalizes the long pile, because
the elastic movements for a long pile are larger, as opposed to a short
pile; the slope of 0.05 inch/ton occurs sooner for a longer pile. The
Butler and Hoy development takes the elastic deformations into account,
substantially offsetting the length effect.
16

Fig. 4.9 Nordlund/Fuller and

Hoy's method

Fig. 4.10 Vander Veen's method

Fig. 4.10 shows the construction of the failure load as proposed by
Vander Veen (1953). A value of the failure load, Qult/ is chosen and values
calculated from ln(1 - Q/Qult), are plotted against the movement. When
the plot becomes a straight line, the correct Qult has been chosen. The
Vander Veen method was proposed long before pocket calculators were
available. Without using those, however, its application is very time consuming.
17

Fig. 4.11 Compilation of failure

criteria

In Fig. 4.11, the above determined nine values are plotted together.
As shown, the offset limit of 181 tons is the lowest and the Chin value of
235 tons is the highest. The other seven values are near the maximum
test load of 207 tons.
It is difficult to make a rational choice of the best criterion to use,
because the preferred criterion depends heavily on one's past experience
and conception of what constitutes failure. One of the main reasons for
having a strict criterion is, after all, to enable a set of compatible reference
cases to be established. The author prefers to use, not one, but four of
the criteria. The preferred criteria are the Davisson limit load, the BrinchHansen 80%-criterion, the Chin-Kondner failure load and the Butler and
Hoy failure load with the proposed modification. In case of an engineering
report, the preference and experience of the receiver of the report may
result in the use of also other methods. Naturally, whatever one's preferred mathematical criterion, the failure load or pile capacity value intended for use in design of a pile foundation must not be higher than the
maximum load applied to the pile in the test. A safety factor applied to
an extrapolated "capacity" is not reliable.
The Davisson limit is chosen because it has the tremendous merit of
allowing the engineer, when proof testing a pile for a certain allowable
load, to determine in advance the maximum allowable movement for this
load with consideration of the length and size of the pile. Thus, as proposed by Fellenius (1975), contract specifications can be drawn up including an acceptance criterion for piles proof tested according to quick
testing methods. The specifications can simply call for a test to at least
twice the design load, as usual, and declare that at a test load equal to a
factor, F, times the design load, the movement shall be smaller than the
elastic column compression of the pile, plus 0.15 inch, plus a value equal
to the diameter divided by 120. The factor F should be chosen according
to circumstances in each case. The usual range is 1.8 through 2.2.
18

The BrinchHansen 80%-criterion is chosen because it usually gives a

Qu-value, which is close to what one subjectively accepts as the true
ultimate failure value. The value is smaller than the Chin value. However,
the criterion is more sensitive to inaccuracies of the test data than is the
Chin criterion.
The Chin-Kondner method is chosen, because it allows a continuous
check on the test, if a plot is made as the test proceeds, and a prediction
of the maximum load that will be applied during the test. Sudden kinks
or slope changes in the Chin line indicate that something is amiss with
either the pile or with the test arrangement (Chin, 1978). The Chin value
has the additional advantage of being less sensitive to imprecisions of
the load and movement values.
The Butler and Hoy method is chosen primarily because of its resemblance to the offset-limit method. In some cases, a Davisson limit load
can be obtained without the interpreter being willing to accept intuitively
that the pile has reached failure. (In such cases, the Chin value will be
much higher than the Davisson limit). Further, as the Butler and Hoy slope
of 0.05 inch/ton is not approached unless failure is imminent, absence
of a Butler and Hoy failure in addition to a high Chin value indicates that
the particular Davisson value is imprecise. The reasons for the latter
can be wrongly chosen values of pile elastic modulus or pile length, or
imprecise or erroneous values of load or movement.
5. FACTOR OF SAFETY AND ACCEPTANCE CRITERIA
5.1 Factor of Safety
The most common purpose of a static loading test is to determine
the capacity of a pile or that the pile has an at-least capacity. The
capacity is related to the desired safe load on the pile, the allowable
load, by a factor of safety, which is the ratio between the capacity
determined in the test and the allowable load. In so-called factored
design, a "resistance factor" is applied to the capacity and a "load
factor" is applied to the load. In Europe, the latter approach is called
"partial factor of safety approach" and is today the dominant approach.
The factor of safety is not a singular value applicable at all times.
The value to use depends on the desired freedom from danger, loss,
and unacceptable consequence of failure, and on the level of knowledge and control of the aspects influencing the variation of capacity
at the site. Not least important are the method used to determine or
define the ultimate load from the test results and how representative
the test is for the site. For piled foundations, practice has developed a
range of factors to apply, as follows.
For example, in a testing programme early in the design work, using
piles which are not necessarily the same type, size, or length as will
be used for the final project, the safety factor applied could be 2.5. In
19

the case of testing during a final design phase, when testing the under
conditions more representative for the project, the safety factor could
be reduced to 2.2. Then, when testing is carried out on the actual pile
used for the project and installed by the actual piling contractor for
purpose of verifying the final design, the factor commonly applied is
2.0. Well into the project, when testing is carried out for proof testing
purpose and conditions are favorable, the factor may be further reduced and become 1.8. Reduction of the safety factor may also be
warranted when limited variability is confirmed by means of combining
the design with detailed site investigation and control procedures of
high quality. One must also consider the number of tests performed
and the scatter of the test results between tests. Not to forget the
assurance gained by means of incorporating dynamic methods for
controlling hammer performance and for capacity determination
alongside the static procedures.
The value of the factor of safety to apply depends, as mentioned, on
the method used to determine it. A conservative method, such as the
Davisson offset load, warrants a smaller factor than a method such as
the BrinchHansen 80%-criterion. It is good practice to apply more than
one method for defining the capacity and to apply to each method its
own factor of safety letting the smallest allowable load govern the
design. As mentioned earlier, it is not good practice to extrapolate the
test results to a capacity larger than the maximum test load and apply
a factor of safety to the extrapolated value. That is to say, a factor-ofsafety approach should not be used with capacity determined from the
Chin method.
In a design geared toward determining the load distribution along
a pile, the location of the neutral plane, and the settlement of the piled
foundation, the factor of safety may not be the governing aspect. The
design may then be completed with a factor of safety that is larger
than the mentioned values, as well as, in some cases, smaller. The
more important the project, the more information that becomes available, and the more detailed and representative the analysis of the pile
behaviorfor which a static test is only a part of the overall design
effortthe more weight the settlement analysis gets and the less important the factor of safety becomes.
5.2 Acceptance Criteria
Proof testing piles is carried out less for determining capacity (ultimate resistance; failure load) and more for determining an at-least
capacity. The maximum test load is normally only twice the intended
allowable load. In older days, the acceptance criterion for the test was
simply that the movement at the maximum load must not be larger
than a specified value, and that after unloading, the net movement
20

must not be larger than a specified value. Usually the testing method
was the so-called standard ASTM method. For short piles, which demonstrate small 'elastic' compression for the applied load, this was normally an economical and practical albeit somewhat liberal criterion,
while for long piles, it was often uneconomically conservative. Also,
the "maximum-and-net-movement criterion" came into practice when
loads were much smaller than the current loads and when most structures were less sensitive to differential movements. Apart from only
using one point on the curve neglecting the information provided by
the load-movement behavior of the tested pile, the "maximum-and-netmovement criterion" includes the misconception that the movement
acceptable for the structure and the long-term movement of the pile
cap has anything directly to do with the load-movement behavior of
the tested single pile.
If movement of the pile cap is critical to the design, the design must
include a proper settlement analysis of the pile group and the static
pile test may have to include instrumentation of the piles. If not, then
a simple factor of safety approach is sufficient as based on the shape
of the load-movement curve and the capacity determined from the
static test. From reasons of practical engineering and contractual aspects, the acceptance criterion should be based on the combination
that the offset limit and the failure load should not be reached before
test loads of, say, 1.8 through 2.2 and 2.0 through 2.5 times the allowable load, respectively.
6. INSTRUMENTATION OF THE PILE
6.1 Introduction
In the routine static loading test, measurements are taken at the pile
head only and it is impossible to estimate with any worthwhile accuracy the mobilized toe resistance from load-movement data obtained
at the pile head. That is, the pile-head load-movement data essentially
only tell the total capacity of the pile giving very little to aid an interpretation of the load distribution in the test pile. Yet, in most tests, after
having determined the total capacity, one may be equally concerned
over what portion of the capacity is obtained at the pile toe or over the
lower portion of the pile, where is the neutral plane located, what is
the shaft resistance in a specific soil layer, etc. The costs and efforts
involved in addressing these questions vary with the specific conditions and degree of accuracy required. However, already a minimal
and low-cost instrumentation effort may give a considerable boost to
the value of a static test.
In brief, instrumentation of the pile refers to instrumentation down
the pile which is extra to the routine instrumentation at the pile head
for pile head movement, applied load, and jack pressure.
21

Instrumentation consists of a wide array of efforts from the simple

telltale rod inserted to the pile toe over installing a multi-telltale or an
electrical strain gage system all through to the incorporation of separate load cells. The topic is huge and the scope of these guidelines
precludes providing details of the various instrumentation systems.
Therefore, the views presented in this chapter are limited to those
necessary to familiarize the reader with the different aspects involved
in instrumenting a pile. In an actual case, it is necessary to make
reference to more comprehensive texts, such as "Geotechnical Instrumentation for Monitoring Field Performance" by Dunnicliff (1988),
which gives extensive background to instrumentation of piles.
6.2 Telltale Instrumentation
The static loading test can be substantially enhanced by placing
telltales in the pile. A telltale is a rod (or wire) which lower end is
connected to the pile, usually at the toe, but which stands free from
the pile along its overall length by means of a guide-pipe arrangement.
By attaching a dial gage at the upper end of the rod and measuring
the change of distance between the rod top and the pile head, the
shortening of the pile during the test is monitored. The telltale rod tells
a tale: that of the movement of its lower end and, therefore, of the
movement of the pile at the location of the lower telltale end in relation
to the pile head position. The absolute movement of the pile toe is
obtained as the measured pile shortening subtracted from the movement of the pile head.
With use of some foresight and planning, telltales can be installed
rather easily and cheaply in all types of piles. Suggestions for simple
telltale arrangements are included in the ASTM D1143 standard with
reference to arrangement for telltale rods in pipe piles, steel H-piles,
and wood piles. Naturally, a telltale can also be installed in precast
prestressed concrete piles if they are equipped with a guide pipe cast
in the pile in the precast yard. Alternatively, outside placing of guide
pipes can be used. Instead of a stiff rod, a telltale can also consist of
a stretched wire. Telltales can be installed singly or as "multiple telltale". For details, see Dunnicliff (1988).
Fig. 6.1 presents an example of test results from a static loading test
on a precast concrete pile. A guide-pipe for a telltale had been cast in
the pile allowing a telltale to be inserted to the pile toe after the driving
to monitor the compression (shortening) of the pile.
The difference between the pile head movement and the movement
of the telltale end is the shortening (or, in uplift testing, the lengthening)
of the pile between the pile head and the location of telltale end.
The shortening value can be transferred to a value of strain over the
particular length of the pile by dividing the value with the length. By
22

Fig. 6.1 The load-movement

diagram of a pile
equipped with a telltale
to the toe of the pile.

multiplying the strain with the modulus of elasticity (that is, applying
Hooke's Law), the average stress in the pile over the telltale length is
obtained. By multiplying the stress with the cross sectional area of the
pile, the average load in the pile is obtained.
In the case of a constant unit shaft resistance, the average load is
equal to the load in the middle of the pileor middle of the telltale
length. In the case of a linearly increasing unit shaft resistance, the
average load is equal to the load at a level located somewhere between
the midpoint and the upper third point. Obviously, knowledge of the
distribution of the shaft resistance is essential for the evaluation of the
load distribution.
The mathematical formula is as follows:
(6.1)
where
Q = average load
A = cross sectional pile area
E = elastic modulus
L = shortening (lengthening) of the pile
L = pile length
Having several telltales in a pile results in several values of average
load and an improvement of the representativeness of the load distribu
tion evaluated from the measurements. Having two telltales results in
23

three average values of load; the third one being obtained from the
difference in compression measured over the distance between the
two telltale ends connected to the pile. Correspondingly, having three
telltales results in six load values, etc. There is a practical limit, because
from primarily practical considerations of accuracy, it is not worthwhile
to have telltale lengths and distances shorter than about 5 to 8 metre
(15 to 25 feet).
When using telltales, the accuracy of the compression measurements must be better than the accuracy usually accepted for movement measurements.
The nominal precision of measurements of movement using dial
gages is usually only 0.025 mm (0.001 inch). The actual accuracy of
the values is, of course, smaller than the precision. At best when using
mechanical gages, the error is about 0.1 mm (0.005 inch) or larger. On
special occasions, dial gages with a ten times finer reading precision
are used, the ten times finer gages will have a smaller error, but not a
ten times smaller.
It is necessary to have dial gages with stems that are long enough
to allow the telltale records to be taken during the entire test without
having to reset the gages or to shim them, because otherwise errors
are introduced which will destroy the value of the records.
A telltale rod must not be subjected to forces along its length or be
let to snake and move about. Therefore, it is usually installed in a sleeve
or a guide pipe. To minimize friction, the outside of the rod is well
greased and/or the annulus between the rod and the sleeve is filled
with lubricating oil. The exception being telltales which are made up
of very heavy duty pipes capable of standing free inside a pipe pile
and where low accuracy is accepted.
Theoretically, it would seem as if it does not matter if one references
the upper end of the telltale to the measuring beam, in which case one
measures movement, or to the pile head, in which case one measures
shortening. By simply subtracting the telltale measurement from the
pile head movement, one obtains the other value. However, in practice,
one should always measure the shortening directly, that is, reference
the telltale to the pile head, because shortening data used to determine
strain and stress, require an order of magnitude or better accuracy
than movement data do. And any reading, be it from the pile-head gage
or the telltale gage, is obtained with some inaccuracy. Having to take
the difference between two readings to get the shortening value, increases the inaccuracy in the shortening value as opposed to measuring it directly. Therefore, shortening, requiring the higher level of accuracy, should be measured directly and telltale dial gages should be
installed to measure between the telltale upper end and the pile head.
An additional reason is that a tilting of the pile head will result in
greater error for a telltale measuring movement (reading against the
24

reference beam) as opposed to the telltale reading compression directly (referenced to the pile head).
Apart from the obvious that results of an analysis of telltale measurements depend foremost on the accuracy of the measurements, the
analysis introduces the modulus of the pile material and the results
depend also on how accurately the modulus is known. Steel has a
constant modulus and steel piles are suitable for telltale instrumentation. In contrast, the modulus of concrete is not constant over the stress
range considered in a static loading test. Therefore, telltale measurements in concrete piles and concreted pipe piles are difficult to analyze. As mentioned, strain evaluated from telltale data is obtained from
readings of two telltales, whereas strain is obtained directly from strain
gages. For these reasons, apart from when telltales are placed at the
toe of a pile, telltales in concrete piles should not be used as the
primary gage for determining load.
For evaluating load in the tested pile, the accuracy of the measurements must be very high. This means that the mechanical type dial
gages, even those with high precision gradation, are not suitable. Linear voltage displacement transducers, LVDTs, are preferred.
In fact, when planning a static loading test and considering the
inclusion of telltales, it is recommended that the telltales be limited to
one to the toe and one back-up placed, say 5 metre (15 feet) above
the toe. To obtain data which are useful for a detail analysis of load
distribution, the rest of the instrumentation for measuring strain should
be electrical strain gages.
The primary purpose of telltale instrumentation of a test pile is to
determine movement and in particular the movement of the pile toe.
For any pile, where the elastic shortening or lengthening of the pile is
difficult to calculate from pile material data and geometry with sufficient accuracy when determining the movement of the pile toe, a telltale to the pile toe should be installed. This means that most tests on
piles of embedment length exceeding about 15 m (50 ft) will benefit
from having a telltale installed to the toe of the pile. A single toe telltale
is easily installed and its costs are insignificant in relation to the overall
costs of the test, as well as to the benefits derived from the measurement.
6.3 Strain Gage Instrumentation
To determine load at a point in a pile with some accuracy, necessitates strain gages. Strain gages can be electrical resistance gages or
vibrating wire gages. In very cursory principle, the strain gage reacts
by changing its resistance or frequency in response to a shortening or
lengthening and the response is picked up by a read-out instrument
and calibrated to strain. As mentioned earlier, the strain can be trans25

ferred to stress and/or to load. Dunnicliff (1988) presents a number of

aspects related to using strain gage instrumented piles.
As opposed to telltale data, load values obtained by means of strain
gages are not average loads over a length of a pile, but the load acting
at the location of the gages. Furthermore, the strain gage will provide
strain data which are an order of magnitude or better than obtainable
with the best telltale system. However, the accuracy of a strain-gage
determined load value still relies on the accuracy of the elastic modulus
of the pile material which is used with the strain data. In other words,
the strain gage uses the pile as a part of the load determination.
Most strain-gages have a tendency to drift and, therefore, strain
gages may not be very accurate for measurements stretching over a
longer period of time.
It is not possible to have a telltale in a pile before it is installed in
the ground. Therefore, telltales are "zeroed" to the conditions of strain
and load existing in the pile at the time of the installation of the telltale.
In contrast, some strain gages can be installed before driving a
premanufactured pile and be theoretically "zeroed" to the stress and
strain conditions in the pile immediately after the driving or even before
it was driven. However, the driving stresses usually cause the zero
value to drift, and normally a new reference reading under zero conditions must be taken before every static loading. For this reason, the
load changes induced in a pile between its installation and a test are
normally lost and the data interpreted without consideration of such
effects. Chapter 8 discusses methods of overcoming these difficulties.
Strain gage instrumentation costs more than telltale instrumentation, strain gages must be installed by well trained technicians, and
they are sensitive to mechanical damage and moisture. While a toe
telltale should be incorporated in almost every static test, strain gages
belong to special projects with specific questions to address in the
test.
6.4 Load-Cells
Sometimes, it is necessary to very accurately determine the load not
just at the pile head but also down the pile. In particular when long
term stability is desired, special load cells must be designed to fit
the pile that are insensitive to time effects, moisture changes, and
properties of the pile material. Such load cells are expensive and not
readily available, but they do exist. Details on them, however, lie outside the scope of this publication.
7. DETERMINATION OF ELASTIC MODULUS FOR USE WITH
STRAIN DATA
7.1 Basic Principles of Stress-Strain Analysis
In a static pile loading test where the pile is instrumented with strain
gages or telltales, the gages serve to determine the axial strain induced
26

in the pile by the applied load. The strain data are used to evaluate the
load distribution in the pile according to Hooke's Law, that is, the
stress-strain relation expressed by Eq. 7.1
(7.1)
where
=
E=
L =
L=
e =

stress
modulus of pile material
change of length (telltale length)
length (telltale length)
strain

Fig. 7.1 shows a typical stress-strain diagram of data from an instru

mented loading test on a pile with constant elastic modulus. The line
with "data" points that is curved near the origin and becomes linear
toward higher strains, the upper line, indicates "measured" data. The
line which is straight from the origin, the lower line, is the theoretical
elastic line for a column with equal properties to that of the pile. The
difference between the lines is, of course, due to shaft resistance act
ing on the pile in the loading test.
All shaft resistance has been overcome in the test, when the "mea
sured" curve becomes parallel to the "theoretical". When evaluating
the results from a loading test, finding this point is desirable, although,
in practice, its location is often difficult to determine.
However, by plotting the tangent modulus of the "measured" curve,
the point becomes easily discernible. The tangent modulus is the slope
of the curve and it is plotted as the increment of load divided by the
increment of strain plotted against the strain. The tangent modulus plot
of the stress-strain lines is shown in Fig. 7.1B. The tangent modulus,
or, more correctly termed, the chord modulus initially reduces with
increasing strain to become constant at a certain amount of strain.
This occurs when all the shaft resistance has been overcome and the
constant value is equal to the pile modulus.
Often, the exact modulus of the test pile is not known. Then, the
tangent modulus plot becomes a valuable aid in determining the modu
lus, which then is used in the calculations to determine the distribution
of the load in the pile.
7.2 Actual Test Results
Fig. 7.2 presents actual test results from a static loading test on a
steel pile. The pile was equipped with two telltales, one upper and one
to the toe of the pile. The figure shows the applied load at the pile head
plotted against measured strain (i.e., shortening divided by telltale
27

Fig. 7.1 Typical data from an instrumented static pile

loading test on a pile with a constant
modulus. A. Stress-strain diagram of the pile
head (upper curve) and of the corresponding
free standing column. B. Plot of tangent
modulus against strain. (After Fellenius,
1989).
length) for the upper and lower telltales and for the difference between
the telltales, i.e., the strain along the bottom portion of the pile.
It is very difficult to obtain anything quantitative from the diagram
in Fig. 7.2A. However, when studying the diagram in Fig. 7.2B showing
the tangent modulus plot, it can easily be determined that the curve
for the upper telltale indicates that a constant modulus (a horizontal,
straight line portion) develops at a value of 0.3 millistrain, which oc28

Fig. 7.2 Load-strain and modulus diagrams. A. Loadstrain diagram for two telltale lengths and for
the difference between the two telltale
lengths. B. Tangent modulus diagram for the
two telltale lengths and for the difference
between the two telltale lengths. (After
Fellenius, 1989)
curred when the applied load was 1,070 KN (120 tons). For the lower
telltale, a constant modulus is indicated for a strain of 0.8 millistrain
occurring when the applied load was 2,450 KN (275 tons). Finally, the
curve for the telltale difference (bottom portion of the pile) indicates a
constant modulus at a strain of 0.5 millistrain at the applied load of
2,490 KN (280 tons).
The analysis of the tangent moduli for a range of applied load of
2,518 KN to 2,670 KN (283 to 300 tons) indicates a modulus for the
upper, lower, and bottom portion telltale lengths, of 2.776, 2.785, and
2.847 MN/strain (312, 313, and 320 ton/millistrain). The agreement
between the upper and lower telltale values is excellent. It is not surprising that the value for the lower portion is slightly off as any inaccu29

racy in the telltale readings would be exaggerated when taking the

difference of them.
Thus, the evaluation indicates that the tangent modulus of the pile
cross section is equal to 2.78 MN/strain (312 ton/millistrain). By inserting this value into the conventional relation LOAD = AREA times
MODULUS times STRAIN with the cross sectional area equal to 128.5
cm 2 (19.9 in2), an "elastic" modulus of 214 GPa (31,000 ksi) is obtained, which compares well with the usually assumed value of 210
GPa, when considering the accuracy of, in particular, the values of
cross sectional area of the pile and of the guidepipes.
The analysis becomes a little bit more difficult when evaluating
strain data from other than steel piles, i.e., concrete piles or concretefilled pipe piles. Contrary to common belief, a concrete column does
not exhibit a linear stress-strain relation when loaded. That is, the
Young's modulus of concrete reduces with the applied load. Fig. 7.3A
illustrates an assumed stress-strain curve of a concrete column (lower
line) having a stress dependent modulus. It has been assumed that the
line is a second degree curve and that the final slope of the line is 30%
of the initial slope. This reduction of the slope, that is, the modulus, is
extreme, and has been chosen for reasons of instructional clarity. (An
example of an actual case will be given later).
The upper curve in Fig. 7.3A, the line with the data points, shows
the same column taken as a pile subjected to shaft resistance. As in
the case of the pile with the constant modulus illustrated in Fig. 7.1,
as soon as all the shaft resistance has been overcome, the two lines
are parallel. Due to the curving of the lines, it is very difficult to tell
when this occurs, however.
In Fig. 7.3B, the tangent modulus of the column line is plotted
against the strain (solid line). Because the stress-strain relation for the
column has been assumed to follow a second degree equation, the
tangent modulus is a straight line, and, as the modulus is not constant
but reducing, the line slopes downward with increasing strain. The line
with the "data" points is the tangent modulus line for the pile. It becomes parallel with that of the column after the shaft resistance has
been overcome. As shown, it plots slightly below the column line.
Extrapolating the pile modulus line to the y-axis and integrating it,
would result in a "restored column curve" located marginally below
the true column stress-strain curve in Fig. 7.3A.
7.3 Mathematical Relations
Mathematically, the lines and curves are expressed, as follows:
The equation for the tangent modulus line is:
(7.2)
30

Fig. 7.3 Typical data from an instrumented static pile

loading test on a concrete pile with a
modulus reducing with increasing stress. A.
Stress-strain diagram of the pile head (upper
curve) and of the corresponding free standing
column. B. Plot of tangent modulus against
strain. (After Fellenius, 1989).
where
Et = the tangent modulus
= induced stress
= induced strain
A = slope of the tangent modulus line
B = Y-intercept (initial tangent modulus)
Integrating the tangent modulus line results in the following equation
31

Fig. 7.4 Load-strain and modulus

diagrams from a static
loading test on a
prestressed concrete
pile. A. Load-strain
diagram for two telltale
lengths and for the
difference between the
two telltale lengths. B.
Tangent modulus
diagram for the two
telltale lengths and for
the difference between
the two lengths. (After
Fellenius, 1989)

for the column stress-strain relation:

(7.3)
And the stress in the pile for an induced strain:
= Es

(7.4)

where
Es = the secant modulus
and
(7.5)
7.4 Example from a Pile with a Non-Constant Modulus
The tangent modulus method of evaluation applied to piles of nonconstant "elastic" modulus is illustrated in Fig. 7.4 by the results from
a static loading test on a precast prestressed concrete pile equipped
with several telltales. Data from two telltales have been chosen for the
illustration: Telltale 7 at a depth in the pile of 38.6 m (126.64 feet) and
Telltale 9 at a depth of 50.2 m (164.62 feet), where the maximum
movements relative to the pile head measured for the telltale points
were 11.3 mm (0.44 inch) and 24.9 mm (0.98 inch), respectively. These
telltales were chosen for reasons of ensuring that all or most of the
shaft resistance over the telltale lengths had been overcome at the
maximum load, which is not the case for the lowest telltale lengths.
32

Fig. 7.4A shows the applied loads plotted against the measured
strains over the two telltale lengths and over the difference between
the two telltales. It is obvious that the lines are curved. An immediate
question when seeing such curves is: "are they curved because the
shaft resistance is not yet fully overcome, or because the modulus is
reducing with increasing load, or both?"
The answer to the question is given in Fig. 7.4B, showing the tangent
modulus plot of the data. The tangent modulus lines are becoming
straight at larger strains, which, indeed, indicates a second degree
curve for the stress-strain relation for where shaft resistance is not a
factor.
The tangent modulus lines are not only used to evaluate at what
applied load the shaft resistance along the pile was fully mobilized, but
also for determining the secant modulus of the pile material (needed for
calculation of the load in the pile according to Eqs. 6.1 and 7.1).
Linear regression of the data points making up the straight portion
of the three lines may be used to provide the equation of the modulus
lines, that is, to determine the constants A and B in Eq. 7.2. Regression
of the modulus line for Telltale 7 results in that the constants A and B
are equal to2.14 KN/millistrain and 4.877 KN/millistrain, respec
tively (in English units: 0.240 ton/millistrain and 548.2 ton/millistrain, respectively). The linear regression correlation coefficient is
0.9980.
Applying Eqs. 1 through 4, results in the following values of initial
and final tangent moduli, and final secant modulus:

Initial
and

Final
Secant

Et

= 37.8 GPa (5,480 ksi)

Et = 16.2 GPa (2,340 ksi)

Es = 27.0 GPa (3,900 ksi)

Inserting the values of A and B into Eq. 2 gives the average load in
the pile over the length of a telltale as a function of induced strain:
Q = - 1 . 0 7 2 + 4880 (KN)
In English units the relation becomes:
Q = - 0 . 1 2 2 + 548 (tons)
Naturally, the tangent modulus method is not restricted to the analy
sis of telltale data, but are as easily applicable to strain gage data. (In
fact, a considerable improvement of the accuracy of the load determi
nation is obtained by using strain gages directly in lieu of telltales).
One of the most immediately noticed benefits of the tangent modu
lus method is that inaccuracies in the data become readily apparent.
For example illustrating this aspect, see Fellenius (1989).
By means of the tangent modulus analysis, strain measurements
can be analyzed to determine accuracy, to establish at what applied
33

load the shaft resistance is fully mobilized, and to evaluate what value
of the secant modulus to use for determining the load distribution in
the pile in the data reduction effort following the tangent modulus
analysis. A data reduction should consider factors such as the residual
strain in the pile, as well as variation between individual gages.
8. INTERPRETATION AND EVALUATION OF INSTRUMENTATION
DATA
8.1 Basic Analysis of Telltale Data
When analyzing data from a telltale instrumented loading test, the
toe resistance of the pile can be estimated from the values of average
load calculated according to Eq. 6.1 from the telltale measured shortening of the pile. Building on the assumption of constant unit shaft
resistance acting along the full length of the pile (the telltale length),
the following relations can be derived (Fellenius 1980):
(6.1)
Rt = 2Q ave - Qh

(8.1)

Rs = Qh - Rt

(8.2)

Where
Qave =
A=
AL =
L=
Rt =
Qh =
Rs =

average load in the pile

cross sectional area of the pile
measured compression of the pile
pile or telltale length
toe resistance
load applied to the pile head
shaft resistance

The data for analysis should be chosen from when the applied load,
Qh, is closest to the failure load obtained from an analysis of the pile
head load-movement data (Chapter 4).
Instead of assuming constant unit shaft resistance, it is assumed
that the unit shaft resistance increases linearly (triangular distribution),
the relation (Eq. 8.1) for the toe resistance becomes:
Rt = 30 ave - 20 h
Where
Rt = toe resistance
Qave = average load in the pile
Qh = load applied to the pile head
34

(8.3)

Fig. 8.1 Load-movement diagram for shaft and toe

resistances assuming that the unit shaft
resistance is either constant (rectangular)
or linearly increasing (triangular).
By inserting test data into the equations, the toe and shaft resistances can be placed in between the two extremes of unit shaft resistanceconstant and linearly increasing, respectively. Fig. 8.2 shows
a plot of the resulting ranges of resistance for the example given in
Fig. 8.1.
8.2 Leonards and Lowell's Method of Analysis of Telltale Data
Leonards and Lovell (1978) presented an analysis method for determining the load distribution in a pile instrumented with one telltale,
where only the relative distribution of unit shaft resistance needs to
be known. Alternatively, the ranges of the relative distribution are
known and an upper and lower boundary type analysis is performed.
The shaft resistance does not need to be uniform, but can be of any
irregular distribution. The Leonards-Lovell method of analysis builds
on a few basic definitions as illustrated in Fig. 8.2 for a pile of a
length, L, subjected to a load at the pile head, Qh. The applied load has
mobilized a shaft resistance, Rs, and a toe resistance, Rt The middle
35

Fig. 8.2 Basic concepts of the Leonards-Lovell

method.
diagram shows the distribution of load in the pile, Qz, from the pile
head to the toe, and to the right is shown a diagram of the unit shaft
resistance, rs, acting along the pile.
If the pile was a free standing column, there would be no shaft
resistance and the toe resistance would be equal to the applied load.
In the test, the applied load causes a compression of the pile, L,
which is measured by means of a telltale to the pile toe. The compres
sion can also be calculated by means of the following relation:
(8.4)
When the pile has no shaft resistance, that is, acts as a column, the
expression for the compression becomes:
(8.5)
Of course, the compression for a column can always be calculated,
whereas the compression for a pile is a function of an unknown amount
of shaft resistance. The compression is measured in the test, however,
and it is useful to define a ratio between the measured compression
of the pile and the compression of an equivalent column, as follows:
(8.6)
The compression of the pile obtained from shaft resistance only is
Ls. It cannot be measured, but it can be calculated, as follows:
(8.7)
36

The corresponding compression for a column subjected to a load

equal to the shaft resistance cannot be measured, either, but it also
can be calculated, as follows:
(8.8)
A ratio between the two calculated shaft-compression values is defined, as follows:
(8.9)
As shown by Leonards and Lovell (1978), the ratio is equal to the
relative distance from the top of the embedded portion of the pile to
the centroid of the area of the unit shaft resistance distribution (length
to centroid over embedment length considered). Thus, for a constant
unit shaft resistance (rectangular distribution), C is equal to 0.5. For a
linearly increasing unit shaft resistance (triangular distribution), C is
equal to 0.67.
The ratio, C, is determined from a known relative distribution of shaft
resistance obtained from borehole data and other information. It is
not necessary to know actual values, only the general shape of the
distribution.
Also a third ratio is defined in the Leonards-Lovell method: the ratio
between the toe resistance, Rt, and the applied load, Qh as follows:
(8.10)
Leonards and Lovell (1978) show that the alpha-ratio can also be a
function of C and C, as follows:
(8.11)
The Leonards-Lovell method consists of determining the appropriate
C-ratio from the soil profile data and the C'-ratio from the calculated
column compression and the measured actual compression. Then, the
portion of the applied load that reaches the pile toe as toe resistance
is obtained from Eq. 8.11.
8.3 Example of a Leonards and Loveil's Analysis
A static loading test has been performed on a steel pile and the
applied load closest to the ultimate load is 230 ton. At this load, the
measured compression of the pile was 1.061 inch. The length of the
pile (tell-tale length, rather) is 90 feet and the cross sectional area is
37

14 in2. What values of shaft and toe resistances were mobilized in the
test?
Inserting the data in Eq. 8.5, a value of the column compression is
obtained, as follows:

Then, Eq. 8.6 gives the C'-ratio:

Assume that the shaft resistance is constant along the pile, which
means that the C-ratio is equal to 0.5. Then, Eq. 8.11 gives the -ratio,
as follows:

and Eq. 8.10 gives the toe resistance:

Rt = a Qh = 0.734 230 = 169 tons
and the shaft resistance is:
Rs = Qh - Rt = 230 - 169 = 61 tons
A linearly increasing shaft resistance would have given a C-ratio
equal to 0.667, instead, and the a-ratio, as follows:

and Eq. 8.10 the toe resistance:

Rt = a Qh = 0.601 230 = 138 tons
and the shaft resistance is:
Rs = Qh - Rt = 230 - 138 = 92 tons
Thus, depending on whether the unit shaft resistance is constant or
increases linearly, the mobilized toe resistance is 169 or 138 tons,
respectively.
The example data could just as well have been analyzed using the
simple relations expressed in Eq. 6.1 and Eqs. 8.1 through 8.3. How
ever, not if the distribution of shaft resistance had been in a soil for
which more complicated distributions had been valid.
Furthermore, the simple relations do not lend themselves toward
analyzing the data from several telltales in a pile, but the Leonards38

Lovell method does. For instance, to analyze the compression measured between two points in a pile as obtained from taking the difference between two telltales measurements, the C-ratio for the about
trapezoidal distribution of shaft resistance between the points is easily
estimated and the analysis rapidly performed.
Having more than two telltales in a pile, will provide a means for
matching load distributions calculated from compressions over different lengths. Obviously, while the calculations are simple, having more
than two telltales in a pile will then necessitate carrying out the calculations by means of a computer. Matching the results means trying out
which C-ratio that will give the same load distribution along the pile
for all compression data over all telltale lengths (Lee and Fellenius,
1989).
9. INFLUENCE OF RESIDUAL COMPRESSION
9.1 Residual Compression in a Leonards-Lovell Analysis
Residual compression is compression induced in the pile from reconsolidation of the soil around the pile after the installation, or compression induced in the pile due to negative skin friction occurring
before the commencement of the test. In most analyses, residual compression is ignored by assuming that all telltale readings show zero at
the start of the test and only considering the compressions imposed
and measured during the test. However, as this can introduce large
errors into the evaluation of load transfer, the analysis should be extended to include residual compression or, at least, to investigate the
consequence of a potential residual compression.
The significance of the effect of residual compression can be demonstrated by adding a small value to the measured shortening of the
pile before proceeding with the analysis. For the example presented
above, a residual compression of a mere 0.1 inch included in the
analysis, results in the following:
The "new" measured compression becomes 1.061 + 0.100 =
1.161 and the "new" C'-value becomes:

The C-values, 0.500 and 0.667, are unchanged and, therefore, the
"new" a-values are 0.906 and 0.859 resulting in "new" values of toe
resistance, Rt, of 208 and 198 tons, as opposed to the earlier values
of 169 and 138 tons, respectively. In this case, obviously, determining
the correct amount of residual compression is more important than
determining the exact distribution of unit shaft resistance. Or to rephrase, without pinning down both influences, it becomes very difficult
to correctly separate shaft and toe resistances.
39

However, by having more than one telltale, it is possible for a given

distribution of unit shaft resistance, to determine the range of residual
compression potentially present in the test pile. The approach is illustrated in Fig. 9.1 showing the load distribution calculated from the test
data with and without consideration of residual compression in a pile
equipped with two telltales. One telltale is inserted to the toe of the
pile and the other one stops at a point above the toe. The two solid load
distribution curves indicate the load distributions for the two telltales
calculated for a given distribution of unit shaft resistance. The dashed
curve indicates the distribution after correcting both the compression
values for residual compression.
The correction is performed by adding an initial compression value
to, for example, the compression measured by the lower telltale, and
calculating what initial compression value the upper telltale must have
for the curves to match. In an actual analysis, a range of possible
residual compression values for the lower telltale are given as input to
a computer program. The computer then provides a matching range
of residual compression values for the upper telltale which would
cause the two curves to match (Lee and Fellenius, 1989). The lower
range of residual compression of the full length of the pile is the one
that gives also residual compression in the upper length of the pile.
Mathematically, the matching can arrive at negative values, but the toe
resistance must not be negative. An additional condition is that the
toe resistance must not become larger than the applied load. Some
judgement will assist in narrowing the range. Then, the analysis is

Fig. 9.1 The principle of

estimating residual
compression by
matching load
distributions determined
from two telltales. Solid
lines: load distribution
unadjusted for residual
strain. Dashed line: after
adjustment causing the
two distributions to
become equal.

40

Fig. 9.2 The principle of estimating residual compression

by means of comparing compression strain
measured from a push test with elongation
strain measured in a pull test on the same pile.
repeated with a different assumption of distribution of unit shaft resistance. The intended end result of the analysis is to arrive at a reasonably reliable range of toe resistance mobilized in the test.
9.2 Determining Residual Compression from Push-Pull Tests in
Combination
Hunter and Davisson (1969) demonstrated the usefulness of combining a compression test (push test) with a tension test (pull test)
where the pile was instrumented with a toe telltale. In the analysis of
the test data from the pull test, a tension resistance appeared to occur
at the pile toe, which, of course, is not possible. When a condition of
zero toe resistance was imposed, a value of residual compression in
the pile resulted. Adding this value to the measurements of the compression test, an amended load distribution was determined.
For the case of strain measurements, the combination of push and
pull measurements can be particularly useful. In particular for the case
of strain measurements at locations in the pile other than at the pile
toe. Fig. 9.2 attempts to illustrate in a typical diagrama strain cycle
of load versus strain the usefulness of combining a push test with a
pull test. The diagram indicates a push test performed before the pull
test where the strain was recorded continuously during and between
the tests.
The strain cycle consists of a curve in push commencing at an origin.
Point 0', which is offset from a true origin, Point 0, by the amount of
41

the residual strain in the pile. When the pile is unloaded, the curve
returns to an increased (in this case) locked-in strain, Point A', on the
abscissa. The pull test commences at this point. Then, in unloading to
zero load from the maximum pull load, the curve returns to a negative
locked-in strain, Point B', at a distance, B, from Point A'.
The position of the true origin, Point 0, in Fig. 9.2 is unknown.
However, Point 0 must lie between Points O' and B'. Since the initial
residual strain (the strain at the start of the push test) is equal to the
distance between Points O and O', its magnitude now ranges from its
minimum value of zero, when Points O' and O are equal, and its maximum, which is the distance between Points O' and B' By plotting the
test data according to Fig. 9.2, the range of potential residual strain
can be determined, and therefore, also the range of actual load in the
pile with consideration to the residual load.
In an actual case, the range may be small or large. The push-pull
combination will not determine the actual value, only the range of
residual strain influencing the test data. However, if the analysis demonstrates a presence of only small residual strain, the load distribution
in the pile can be evaluated with some confidence. If large, one is
prevented from drawing false conclusions from the test data.
Having accepted the expense in terms of money and time of instrumenting a pile for a static loading test, the relatively small extra expenditure of adding a pull test to the programme would be well spent.
10. REFERENCES
AMERICAN SOCIETY FOR TESTING AND MATERIALS, ASTM. Annual
Book of Standards, Section 4, Construction, Vol. 04.08 Soil and Rock,
Building Stones, and Geotextiles.
BRINCH-HANSEN, J., 1963. Discussion on hyperbolic stress-strain response. Cohesive soils. American Society of Civil Engineers, ASCE. Journal for Soil Mechanics and Foundation Engineering, Vol. 89, SM4, pp.
241-242.
BUTLER, H. D. and HOY, H. E., 1977. Users manual for the Texas quickload method for foundation load testing. Federal Highway Administration,
Office of Development, Washington, 59 p.
CHIN, F. K., 1970. Estimation of the ultimate load of piles not carried
to failure. Proceedings of the 2nd Southeast Asian Conference on Soil
Engineering, pp. 81-90.
CHIN, F. K., 1971. Discussion on pile test. Arkansas River project. American Society of Civil Engineers, ASCE Journal for Soil Mechanics and
Foundation Engineering, Vol. 97, SM6, pp. 930-932.
CHIN, F. K., 1978. Diagnosis of pile condition. Lecture at 6th Southeast
Asian Conference on Soil Engineering, Bangkok, 1977, Geological Engineering, Vol. 9, pp. 85-104.
42

DeBEER, E. E. and WALLAYS M., 1972. Franki piles with overexpanded

bases. La Technique des Travaux. No 333, 48 p.
DAVISSON, M. T., 1972. High capacity piles. Proceedings, Soil Mechanics Lecture Series on Innovations in Foundation Construction, American
Society of Civil Engineers, ASCE, Illinois Section, Chicago, March 22,
1972, pp. 81-112.
FELLENIUS, B. H., 1975. Test loading of piles. Methods, interpretation,
and new proof testing procedure. American Society of Civil Engineers,
ASCE Journal of Geotechnical Engineering, Vol. 101, GT9, pp. 855-869.
FELLENIUS, B. H. and SAMSON, L, 1976. Testing of drivability of concrete piles and disturbance to sensitive clay. Canadian Geotechnical Journal, Vol. 13, No. 2, pp. 139-160.
FELLENIUS, B. H., 1980. The analysis of results from routine pile loading
tests. Ground Engineering, Foundation Publishing Ltd., Vol. 13, No. 6, pp.
19-31.
FELLENIUS, B. H., 1984. Ignorance is blissand that is why we sleep so
well. Geotechnical News Magazine, Canadian Geotechnical Society and
the United States National Society for Soil Mechanics and Foundation
Engineering, Vol. 2, No. 4, pp. 14-15.
FULLER, R. M. and HOY, H. E., 1979. Pile load tests including quick load
test method, conventional methods, and interpretations, Highway Research Board, HRB 333, pp 78-86.
HOUSEL, W. A., 1966. Pile load capacity. Estimates and test results.
American Society of Civil Engineers, ASCE Journal for Soil Mechanics
and Foundation Engineering, Vol. 92, SM4, pp 1-30.
HUNTER, A. H. and DAVISSON, M. T., 1969. Measurements of pile load
transfer. American Society for Testing and Materials, ASTM Symposium
on Performance of Deep Foundations. Special Technical Publication, STP
444, pp. 106-117.
LEE, S. Q. S. and FELLENIUS B. H., 1989. The analysis of telltale measurements. Ottawa.
KONDNER, R. L., 1963. Hyperbolic stress-strain response. Cohesive soils.
American Society of Civil Engineers. ASCE Journal for Soil Mechanics
and Foundation Engineering, Vol. 89, SM1, pp. 115-143.
LEONARDS, G. A. and LOVELL, D., 1978. Interpretation of load test on
high capacity driven piles. American Society for Testing and Materials,
ASTM, Symposium on Behavior of Deep Foundations, R. Lundgren Editor,
Special Technical Publication, STP 670, pp. 335-357.
MAZURKIEWICZ, B. K., 1972. Test loading of piles according to Polish
regulations. Royal Swedish Academy of Engineering Sciences, Commission on Pile Research, Report No. 35, Stockholm, 20 p.
43

MOHAN, D., JAIN, G. S. and JAIN, M. P., 1967. A new approach to load
tests. Geotechnique, Vol. 17, pp. 274-283.
NEW YORK DOT, 1974. Static load test manual. New York Department
of Transportation, Soil Mechanics Bureau, Soil Control Procedure SCP4/
74, 35 p.
NORDLUND, R. L, 1966. Personal communication as referenced by Housel (1966), p. 467.
PECK, R. B., HANSON, W. E., and THORNBURN, T., 1974. Foundation
Engineering. Second Edition, John Wiley & Sons, New York, p. 215.
SWEDISH PILE COMMISSION, 1970. Recommendations for pile driving
tests and routine test loading of piles. Royal Swedish Academy of Engineering Sciences, Commission on Pile Research, Report No. 11, Stockholm, 35 p.
VANDER VEEN, C, 1953. The bearing capacity of a pile. Proceedings
of the 3rd International Conference of Soil Mechanics and Foundation
Engineering, ISSMFE, Zurich, Vol. 2, pp. 84-90.
WHITAKER, T., 1957. Experiments with model piles in groups. Geotechnique, Vol. 7, No. 4, pp. 147-167.
WHITAKER, T. and COOKE, R. W., 1961. A new approach to pile testing.
Proceedings of the 4th International Conference of Soil Mechanics and
Foundation Engineering, ISSMFE, Paris, Vol. 2, pp. 171-176.
WHITAKER, T., 1963. The constant rate of penetration test for the determination of the bearing capacity of a pile. Proceedings of the Institute of
Civil Engineers, ICE., Vol. 26, London, pp. 119-123.

44