Professional Documents
Culture Documents
Ufa 1998
Ufa 1998
_
,
2
1
It is of interest to the current findings that setting the exponent n to unity, gives a linear-fractional relationship for the
deformation.
He refers to the form of the relationship as "linear-fractional" which is probably a more appropriate term because it
more correctly describes a function in which the secant from the origin falls linearly with the proportion of the
distance the function has travelled from its start point towards the asymptote. The term "hyperbolic" could imply that
the relation is of the form sinh, cosh or tanh in mathematicians terminology and this is not what is intended. The
mathematical functions bear similarity but are not identical in form.
This observation led to the development of a double linear-fractional analysis method (M. England) which has been
tested in literally thousands of cases and which routinely can track pile behaviour for periods of 24 hours or more. The
method has been given the name TIMESET.
0 1 2 3 4 5 6 7 8
0
.1
.2
.3
.4
.5
.6
.7
.8
.9
1
1.1
1.2
1.3
TIME [Hours]
Time/Disp [hours/mm]
TIME/DISPLACEMENT vs TIME DIAGRAM
D
R
I
V
E
N
C
A
S
T
I
N
-
S
I
T
U
I
N
C
H
A
L
K
Constant load = 3052 kN
Time offset = 39.86 hours
Disp offset = 21.24 mm
Figure 5 Typical Chin plot of relative displacement/time
7
Current research on these functions is proving very interesting and enlightening and shows promise of having a
profound influence on the way we think about soil consolidation and creep behaviour.
Once accurate tracking of real pile behaviour becomes routinely possible, then the final settlement of a pile under any
given load can be fixed with sensible reliability. The asymptote of the time/deformation curve represents final long
term settlement for each given load and when all applied loads are similarly treated, the unique long term load-
settlement characteristic of the pile is displayed, as shown in Figure 6 for a site where piles were founded in chalk.
0 400 800 1200 1600 2000 2400 2800 3200 3600
32
28
24
20
16
12
8
4
0
LOAD [ kN ]
Displacement [mm]
LOAD DISPLACEMENT DIAGRAM LOAD DISPLACEMENT DIAGRAM
D
R
I
V
E
N
C
A
S
T
I
N
-
S
I
T
U
I
N
C
H
A
L
K
Figure 6
It has not been possible to track pile behaviour in practice for more than about 7 days under field conditions for two
reasons. Firstly, the movements become very small, and secondly, the disturbance caused by site traffic and
temperature changes assumes proportionately larger importance. Nevertheless, presented data on the long term
settlement of whole buildings is available and can be tracked by the same method, as shown in Figure 7 (Monadnock
Block, Chicago; Skempton et al). The similarity between the recorded data and the best fit linear-fractional curve
demonstrates that in this case even a single linear-fractional function can be applied reasonably well throughout the
recorded period of about 50 years.
8
The parameters which form the basis of the time/deformation tracking system under constant load are shown with the
basic equation in the Appendix at the end of this paper.
6.2.1 IDEALISED DISPLACEMENT-TIME BEHAVIOUR OF PILES
A simplified model of the reactions of the pile/soil elements to an induced loading at the pile head is shown in Figure-
8. The particular aspect of interest is the response in time of each element to a change of load; the development of the
reaction of each element is described and discussed in the following paragraphs.
As a result of an application of a constant axial compressive load (P) at the pile head, it can be deduced that the pile
head displacement will instantaneously exhibit elastic shortening of the pile length corresponding to the portion of pile
shaft where little or no friction exists (L
o
). The immediate change in pile head displacement will be directly
proportional to the elastic response of the friction free length and the change in load applied.
Application of a load increment to the top of a pile, will apply the same increment of force along the pile until some
resistance is encountered. Initial resistance would normally be the top of the layer in which friction exists. The short
term response of the friction zone would generally be to exhibit a higher strength than its long term value, as a
consequence, practically none of the load increase is transferred immediately to the base. As the shaft resistance
decreases, any of the load applied which will not finally be carried in friction, would be transferred to the base.
If P
s
. f(t) is assumed to describe the shaft resistance as mobilised in time, where P
s
represents the long term shaft
friction, we can conclude that the time function f(t) must approach unity for large values of time (t), to give the long
term value of P
s
.
0 12 24 36 48 60 72 84
0
2.4
4.8
7.2
9.6
12
14.4
16.8
19.2
21.6
24
TIME [Years]
Displacement [cm]
RELATIVE DISPLACEMENT TIME DIAGRAM
M
O
N
A
D
N
O
C
K
B
L
O
C
K
S
K
E
M
P
T
O
N
e
t
a
l
(
1
9
5
5
)
Time offset = 5 years
Disp offset = 32.02 cm
Proj rel disp = 28.93 cm
Proj tot disp = 60.95 cm
Figure 7 Settlement-time recordings on Monadnock Block, Chicago; Skempton
et al (1955)
9
In terms of load distribution, the load that may be
expected to be applied at the base of the pile can now
be expressed as P - P
s
. f(t) or F . f(t), i.e. the load
transferred to the base must be whatever remains of
the total P applied which is not carried in friction and
in time will rise to an asymptotic value according to
the rate at which the excess load carried by the shaft
decreases.
If P
b
. f(t) represents the base resistance as mobilised
in time under constant load, where f(t) characterises
the individual base time component, then, if friction
exists along the length of the pile shaft, the load
transferred to the base is not constant and is a
function of the way in which the shaft resistance is
mobilised in time and therefore the response of the
base alone will be governed by two independent time
functions.
Since the load applied to the base, is governed by f(t),
it follows that the reaction from the base to the load
applied can be represented by F. f(t). P
b
. f(t). The
combination of these two functions is manifest in the
displacement-time behaviour of the base. It may be
reasoned that the time constant associated with the mobilisation of the shaft, directly affects the rate at which the base
reaction is mobilised.
The above holds true whether the load applied is less than or exceeds the ultimate skin friction. It may be reasoned
that for loads less than the ultimate skin friction, a further time function may be needed to characterise the slower
development of friction along the pile length until it finds its long term equilibrium and distribution along the pile
length. It may be expected that the time function may be different according to the load applied. In contrast, for loads
applied which are above the ultimate skin friction, the time function characterising the transfer the load through to the
pile base may be expected to be constant. In practice, previous loading history distorts the time function exhibited.
Elastic shortening is dependent on the distribution of the applied load between the shaft and the base. Since the load
distribution is dependent only on the development of the shaft resistance, the influence of time on the elastic
shortening must follow the time component associated with the skin friction, with the final long-term elastic
deformation of the full pile length (L
o
+ L
f
) occurring once the skin friction is fully mobilised.
Since the pile base deflection is linked to the pile head by just the elastic shortening, it may be reasoned that the
displacement-time response of any part of the pile to a new load must exhibit two time functions simultaneously. A
particular case is the displacement-time response of the pile head, often available for measurement.
Two important conclusions may be drawn:
1. The load applied to the base is a function of how the shaft resistance develops in time.
2. The displacement-time response of any part of a pile will exhibit the time constants associated with the
development of skin friction and end bearing.
Thus far, no particular shape or response characteristic has been associated with the time functions. If the development
of skin friction is considered, in the short term f(t) needs to give an enhanced value for frictional resistance, and
therefore the time function must start at a value greater than unity and tend towards an asymptote as time progresses.
This would be consistent with the Whitaker & Cooke (1966) data.
Figure-8 Components determining time effects
10
A further point worthy of note is that the load applied need not be greater than the total shaft friction for the
elementary model described to be applicable. Each element of friction, associated with either each soil stratum or part
thereof, must also suffer the same phenomenon of gradually reducing its enhanced resistance and progressively
allowing a larger portion of the applied load to be transferred to greater depths. It can therefore be deduced that there
are practically always some elements contributing to the shaft friction which are fully mobilised, these are most
prevalent at the top of the pile. In practice, if the load applied is lower than the ultimate skin friction, the time
constant for the development of the shaft becomes longer; and since the application of load to the base is affected by
the shaft time function, the base time function also takes longer to develop and reach a condition of equilibrium.
While enhanced friction is to be expected in the short term, it might not be obvious that this is manifest in the typical
gradual reduction of settlement rate normally experienced after application of a different load to a pile head.
To illustrate this point a function,
assumed to represent the displacement-
time characteristic is displayed in Figure
6-9. It is represented as a linear fractional
(hyperbolic) function of the form
r
b
b
W t
T t
........... (1)
The function tends towards a total
displacement, W
b
, of 4 mm as shown; the
time scale is represented over
18 divisions, each representing a 1 hour
interval.
This displacement-time function
(equation 1) can be characterised by
quoting the asymptote and a point on the curve which may be arbitrarily chosen. If the time taken to get to 50% of the
asymptotic value is selected T
b
, in this case this is reached in 1 hour.
The differential of the function (1), represents the decreasing velocity in time, which in the illustration above is
indistinguishable from zero after approximately 10 hours. The function can be written as
( )
( ) ( )
r b b b
b
b b
b t
t T W W t
T t
W T
T t
+
+
2 2
.......................................(2)
A further differentiation of equation 2 can be performed. The result, shown in equation 3, is displayed graphically in
Figure 6-9, where the change of acceleration in time is represented by:
( )
2
2 3
2 r b b
b t
W T
T t
+
......................................................(3)
This varying deceleration (deceleration is indicated by the minus sign) is directly proportional to a change in force,
since force = mass x acceleration. Since the mass of the pile (considered with or without a boundary soil layer)
remains constant, the change in acceleration may be attributable directly to a change in the forces involved in the
pile/soil system. Conveniently therefore, only the displacement-time function need be addressed when trying to model
the behaviour of a pile in time, and attempting to deal with the specific changes of forces in the pile-soil system with
time, can be circumvented.
The magnitude of the deceleration is a function of the difference between applied and reactive force. It is therefore
clear the importance of true constant application of load to reveal how reactive forces develop in time.
Figure 6-9 Displacement, velocity and acceleration variation
in time
11
The following summarises the main issues which have been deduced by the authors from the displacement/time
analyses:
1. The time-displacement model appears to match measured field data from pile loading tests with high accuracy.
The model is a continuous function exhibiting no transitions from one particular mechanism to another. The
time-displacement behaviour for the period typically recorded must be governed by a single or mechanism.
2. The characteristic behaviour of time-displacement under constant load for piles founded in clays and sands and
gravels appear to be so similar that any differences are imperceptible.
This similarity of behaviour of materials of vastly different permeabilities would lead to the conclusion that
permeability cannot alone govern the time-displacement.
Since conventional consolidation theories are generally based on excess pore water pressure dissipation, it may
be concluded that this water expulsion effect only plays a secondary role and it is cncluded the mechanism is
creep.
3. The Timeset model employs two distinct time functions and the pile/soil interaction problem has two elements
which are likely to contribute in different ways to the displacement-time behaviour.
While it is difficult to obtain evidence of the single functions independently unless one of the contributing
functions is effectively constant, each of the Timeset components can be consistently equated to mobilisation of
the shaft and base in time.
It is therefore apparent that the skin friction is not a passive component as might be assumed. As, for every load
increment beyond the ultimate skin friction, it is apparent that the skin friction carries the additional load
temporarily and then, in time, returns to its ultimate capacity.
A mechanism which explains the short term, enhanced skin friction capacity, is sought.
A Viscous effect can be considered: however, the resistance typical of viscous drag is generally a function
of velocity. If the load applied is less than the ultimate capacity of the pile, end bearing is likely to be
partially mobilised and the displacement-time, and pile displacement continues and therefore so also
would the viscous effect.
Dilation has been explored and it appears to contain the range of necessary behaviour characteristics
being sought. It also provides an insight into a mechanism of rupture of skin friction, evident in some tests
on piles in a cohesive stratum with little end bearing.
6.3 ANALYSIS OF LOAD-DISPLACEMENT BEHAVIOUR
6.4 Load-settlement relationships
It had been suggested by several authors, including Professor Chin in the early 1970's, that both time/deformation and
load/deformation for piles followed hyperbolic (linear-fractional) laws. This observation had been made with regard
to the deformation of other structural members, going back into the 1930's and possibly earlier. Professor Southwell
in a lecture to the Royal Society in 1933 had drawn attention to the fact that steel struts behaved according to such
laws and several other workers in following years began to look for situations where the same rules might apply.
Many papers have appeared in geotechnics demonstrating that foundations have very nearly followed such laws but
usually not quite, and this led to doubts on the part of those who sought to use the simple single hyperbolic methods
for prediction or analysis.
12
It has only recently been recognised that there is a very good reason for the non-compliance of many piles with such a
function. Piles in general exhibit shaft friction, end bearing, and elastic shortening. It would therefore be perverse of
the soil to behave in terms of a single linear-fractional function as pointed out by Fleming (1992) and a twin function
accounting for the two main aspects of behaviour, allied to a simple elastic shortening model, has therefore to be
applied. The results are remarkable and no pile has been found in over a thousand cases which fails to comply with
the discussed formulations in respect to time-deformation and load-deformation, unless the pile has suffered some
physical damage, disturbance or there has been some error in the testing process.
The parameters required to fully represent the load/settlement behaviour of a pile are given in the Appendix. A full
account of the development of the equations is given in the above reference and the method has been given the name
CEMSET". A version of the program used for back analysis is called CEMSOLVE".
6.5 The pile performance analysis process
An example analysis using the CEMSOLVE method is shown in Figure 10.
0 400 800 1200 1600 2000 2400 2800 3200 3600
32
28
24
20
16
12
8
4
0
LOAD [ kN ]
Displacement [mm]
LOAD DISPLACEMENT DIAGRAM
D
R
I
V
E
N
C
A
S
T
I
N
-
S
I
T
U
I
N
C
H
A
L
K
CEMSOLVE
ANALYSIS
X = Input Data
Ds = .435 m
Db = .435 m
Us = 1172 kN
Ub = 3021 kN
Lo = 5 m
Lf = 12 m
Ms = .001
Eb = 381169
Ec = 3E+07
Ke = .5
Base Elastic shortening
Figure 10 CEMSOLVE analysis
These models mean firstly that a long term load/settlement curve can be drawn from relatively short observation of
pile deformation under each constant load. Secondly, provided all work is carried out to high standards and that the
observed settlement has been sufficient to mobilise a reasonable proportion of the pile base resistance, then a back
analysis can be carried out to separate the main parameters.
Clearly if one had a full load/deformation curve one could form a set of simultaneous equations and solve for all the
variables. However, this would not be very sensible because some parameters may be important while others are not
and could be subject to large proportional errors if determined in this way. The procedure for analysis which has been
developed is therefore based on screen curve fitting and numerical optimisation, recognising that many of the
parameters are either closely bounded or known.
13
For example, if one considers the parameter list for the CEMSET program as shown in the Appendix, the items D
s
,
D
b
, L
o
, L
f
are geometrical dimensions and M
s
, E
c
, K
e
are mostly closely known (The first four items are known from on
site measurements, the Site Investigation and the installation records). E
c
, the pile material modulus is generally
known or can be measured easily. K
e
, representing the position of the centroid of friction in relation to the friction
length, is usually an insensitive number provided it is roughly correct.
M
s
is a dimensionless number and has been found to lie normally within the limits 0.001 and 0.0015. This applies
throughout the whole range of experience in soils of all stiffness and it is suspected that it really is a constant in
naturally occurring soils which could only be determined with high accuracy by a series of instrumented pile tests in
different soils. Its constancy probably explains why trials indicate that it is not necessary to use iterative methods to
discover the exact value of the parameter K
e
in most analyses. This leaves three important items which have
potentially wider ranges, though not unlimited. It is usually best to insert known median figures for all the other
parameters and then solve for the last three on screen. Semi automatic optimisation to match the model with the data
is built into the computer program for analysis but allowance is made for engineer intervention when necessary.
Having arrived at a solution it is then possible to carry out a sensitivity study in order to gauge the reliability and
accuracy of the solution.
Occasionally one finds anomalous pile behaviour which does not comply with any reasonable choice of parameter.
This usually enables one to search for and identify the anomalous item. Some examples of anomalous behaviour are
briefly indicated in Section 7 of this paper.
A consequence of this kind of analysis is that it demonstrates that instrumentation of piles and the recording of say
shaft loads and base loads, cannot reliably be carried out without the aid of a satisfactory behavioural model. The
ultimate base load is not just the last reading which a load cell showed.
7. NON LINEAR ANALYSIS CONSEQUENCES
7.1 General understanding
It is not until one works with good non-linear behavioural models that one comes to appreciate their benefits. There
are a number of features of pile behaviour which rapidly come to notice.
(a) Ultimate load definition
It will be recognised that the asymptotic definition of ultimate load is the only practical one that works. It was first
proposed by Terzaghi and many attempts have been made to supplant it with strain related definitions over the last
fifty years. When dealing with piles in a wide range of diameters and soils of varying stiffness, it is clear that neither
fixed settlement numbers nor "percentage of diameter values" can give consistent results.
The theory of plasticity on which most bearing capacity factors are based has in any case no strain related base and
implies an asymptotic failure definition.
(b) Validity of plotted points.
It is common to unload and re-load piles in the course of a test. It is noticeable that on return to a load which has
previously been applied, the projected deformation is not the same as that obtained from the first visit to that load. If
all the other sequentially determined points are also plotted, it will be observed that the point which is inconsistent is
that referring to the re-load. This is due to the second approach to that load being by a different stress path which has
been conditioned largely by shaft stress reversal.
14
(c) Recovery from applied test loads.
Recovery from any test load is determined by the reversal or part reversal of the shaft friction and by the pile base
reaction. There appears to be a close relationship between recovery and the familiar process of hysteresis in other
materials (paper in preparation - M. England). Using a hysteresis linear-fractional model reasonable predictions of
pile recovery from load appear to be possible.
(d) The uniqueness of solutions.
It quickly becomes evident when solutions are attempted that if a pile has not been made to settle enough to mobilise
say 20 to 30% of the available base load, then unique solutions for this fraction of the load will be difficult to find and
will be very dependent on the precise accuracy of the records.
As a generality, reasonably good solutions for conventional piles carrying load mainly by friction can be arrived at
where the settlements are perhaps 15 to 20 mm. Where the pile rests on sand, then the load will usually be large in
end bearing and perhaps 20 to 40 mm settlement will be required in order to distinguish between the effects which are
separately due to base load resistance and base stiffness.
It is always possible and desirable to test the sensitivity of any solution against the credible range of parameters which
might possibly fit.
(e) Parametric observations.
It is found that there is a consistency of behaviour which is related both to the ground conditions and to the
construction method.
It will be noted that because of the way in which the data are processed, all results represent long term behaviour and
the deduced parameters have the same characteristic. This means that one can classify certain geological conditions as
likely to produce particular values of strength and stiffness.
Observation implies that the ratio of stiffness to strength is probably one of the most demonstrative ratios in the whole
of soil mechanics.
It also becomes apparent from many analyses that for very low strength and low stiffness soils, deformations required
to reach ultimate loads can be very large. On the other hand deformations in high strength high stiffness rocks can be
tiny and ultimate loads very high. In the former case large or very large factors of safety or partial factors would be
necessary to control movement of a supported structure within strict limits, while in the latter case one could have a
near unity factor of safety allied with negligible movement. This is not in accord with common factoring systems but
nevertheless needs to be considered.
7.2 Construction improvement.
The method makes clear that the history of pile installation and of the construction of adjacent piles can have
significant effects.
An important item which it has elucidated is the practice of base (and possibly shaft) grouting of piles. It shows clearly
that the process is akin to prestressing in general. A consequence is that it is unprofitable to grout the base of a pile
where the pile is shallow and there is little friction. Likewise grouting the base of a pile where the available base load
is small and the shaft friction is large is also unhelpful. By the use of the method it becomes possible to predict the
effects of the base grouting technique and reasonably to plan the grout pressures that may best be used.
Techniques of construction can be examined in regard to the variance of pile performance and there is much to be
learned about machine performance, operator variability, and the general nature of the construction process.
15
7.3 Partial Factors and performance forecasting.
One of the big disappointments of Geotechnics at present is the Partial Factoring system being proposed for the
Eurocode EC7.
A partial factoring system is one which allows the uncertainty of a desired quantity to be derived from the probability
of error in the parameters which govern its accurate representation. No other factors - e.g. to perform some external
control system are appropriate. If the model in use has a consistent and known offset then it should be corrected. If
this is not possible then extension of parametric ranges may be acceptable but it is not desirable.
The linear-fractional model is exceptionally good and needs no method factor provided it is accepted that the user's
interest is in long term behaviour of the foundation.
When using the linear-fractional method for forecasting, the main difficulties arise from the fact that there are
available many ways of arriving at pile design quantities. For example there are many different bearing capacity
factors for pile design. Most countries seem to use the factors given by Berezantsev for granular soils and a limited
study shows that these are reasonably good but are on the conservative side.
The question of soil stiffnesses is also one which poses difficulty because there are few items of Site Investigation
equipment which can measure in-situ values. It might be thought that the conventional pressuremeter would give
adequate numbers but its measurements are made in the horizontal rather than the vertical direction that pile design
requires. In any event the operative stiffness numbers are strongly influenced by the installation of a pile. It has been
found best to date to determine stiffnesses from records of pile tests on other nearby or similar condition jobs, provided
they have used the same type of pile.
Once the parameters shown in the list of the Appendix have been decided by the best and most reputable means
available, the next step is to assess each for the probability of variation. This gives a set of partial factors which are
used to modify the original probable values in such a way that the prescribed error will increase the resulting
settlement value. Examples are shown in Figure 9 and 10 for cases in which piles are respectively carrying most load
by shaft friction in stiff clay and by end bearing in sand. The curves shown represent probable and worst credible
load/deformation behaviour.
0 500 1000 1500 2000 2500 3000
40
35
30
25
20
15
10
5
0
LOAD [ kN ]
Displacement [mm]
LOAD DISPLACEMENT DIAGRAM
P
A
R
T
I
A
L
F
A
C
T
O
R
E
D
D
E
S
I
G
N
F
O
R
S
T
I
F
F
C
L
A
Y
PARTIAL
FACTORS
1.05
1.05
1.50
1.40
1.20
1.02
1.20
1.50
1.10
1.20
Factored Un-Factored (tabled values)
CEMSET
ANALYSIS
Ds = .6 m
Db = .6 m
Us = 2138 kN
Ub = 509 kN
Lo = 2 m
Lf = 18 m
Ms = .00125
Eb = 50000
Ec = 3E+07
Ke = .45
Figure 11 Design for stiff clay
0 500 1000 1500 2000 2500 3000 3500
40
35
30
25
20
15
10
5
0
LOAD [ kN ]
Displacement [mm]
LOAD DISPLACEMENT DIAGRAM
P
A
R
T
I
A
L
F
A
C
T
O
R
E
D
D
E
S
I
G
N
:
G
L
A
C
I
A
L
S
A
N
D
PARTIAL
FACTORS
1.05
1.05
1.50
1.70
1.10
1.02
1.20
1.50
1.10
1.20
Factored Un-Factored (tabled values)
CEMSET
ANALYSIS
Ds = .6 m
Db = .6 m
Us = 1622 kN
Ub = 4180 kN
Lo = 4 m
Lf = 12 m
Ms = .00125
Eb = 124000
Ec = 3E+07
Ke = .5
Figure 12 Design for glacial sand
Having derived the worst credible load-deformation relationship it is then necessary to consider the corresponding
settlement which the related structure can be allowed to suffer. This depends on the structure type and potential frame
rotations between support points. The acceptable design load can then be fixed, bearing in mind that the load itself
may require a partial factor, perhaps because of structural tolerances, live load variations and other considerations.
16
This method of Partial Factoring has been in use now for 5 years within Kvaerner Cementation Foundations and we
would not wish to take the backward step of using the methods of EC7. It is very simple to use, asks the right
questions and appeals to our young engineers who appreciate its simple logic. It also has a continuing refining effect
on the general process of design.
7.4 TEST OPTIMISATION
It will be appreciated that to carry out high quality tests it is necessary to have careful regard to both
time/displacement and load/displacement behaviour.
Final settlement under a specific load cannot be determined without obtaining sufficient time/displacement data to
project the answer. If load stages are restricted in duration to half and hour or 1 hour, it will often be found that only
say 1/3 to 1/2 of the final settlement will have occurred in such a period and final settlement will be difficult to define.
This generally means that in maintained load tests, loads need to be held for several hours.
Various guidelines have been suggested for deciding the length of load holding periods (e.g. specific pile head
settlement rates) but the answer really is to obtain sufficient data to give a good and consistent fix on the final
settlement. Using computer monitoring techniques this is not difficult, and a method for exploiting the benefit of using
normalised time constants allows further reduction of the load durations.
It should be realised also that where instruments are installed within piles (for research purposes or otherwise) the
same considerations apply and that short term readings have a distinct possibility of being misleading because load
distribution has to adjust with time.
8. IDENTIFICATION OF ANOMALIES
As one might expect anomalies arise from time to time in analysis. These are almost always traceable to the conditions
under which the particular pile was installed. Some examples are shown below for illustration of typical cases.
(1) Poor base construction on a bored pile.
The example shown in Figure 11 is one of a relatively short pile in stiff clay. There is no doubt that the base load of
the pile should have been over 200 kN, but the pile moved downward abruptly and there was no indication of any
response from its base.
17
This pile was bored by a rotary drill and it is
speculated that on the final journey to unload
excavated spoil, some fell back into the base of the
bore. The moral is that short piles are particularly
prone to this sort of accident and that either special
base cleaning tools should be employed or the pile
should have been made longer to allow for the risks
associated with ensuring a properly constructed base.
In this case the pile specified working load was
360 kN and the requirement was for an overall Factor
of Safety of 2.
It might be thought that it is nearly impossible to find
a deficient pile base on a driven preformed pile, yet
such cases are to be found. They are traceable to
heave of clay soils overlying a sandy or rocky
founding layer. The effect is to unseat the pile by a
small amount and hence the potential performance is,
at least initially, without end bearing. Heave is
particularly likely to cause such problems for
displacement piles in stiff clay and it is sometimes
necessary in design to give a pile sufficient penetration into the underlying founding stratum so that it is effectively
tied down.
(2) Hammer damage to a driven pile.
It is often not recognised that the driving of precast concrete piles requires careful consideration, especially in clay
soils or where a hard and resistant soil layer overlies one which is much softer.
It will be appreciated that end resistance to hammer impact in sand layers can be very high but if the pile suddenly
passes from the sand into a soft clay layer underneath and if the hammer drop is not smartly curtailed, then large
tensile reflections travel upward in the pile and can break it into a series of relatively short cracked sections.
Figure 14 shows a case of a precast pile driven into clay. The driving resistance was high and the pile length was 20
metres. However, quite stringent sets were being expected and a five tonne hammer was being used at about 0.5 m
drop.
0 100 200 300 400 500 600 700 800 900 1000 1100 1200
22
20
18
16
14
12
10
8
6
4
2
0
LOAD [ kN ]
Displacement [mm]
LOAD DISPLACEMENT DIAGRAM
O
V
E
R
-
D
R
I
V
E
N
P
R
E
C
A
S
T
P
I
L
E
I
N
C
L
A
Y
CEMSOLVE
ANALYSIS
X = Input Data
Xs = .27 m
Xb = .27 m
Us = 1128 kN
Ub = 1 kN
Lo = 10 m
Lf = 10 m
Ms = .0015
Eb = 50000
Ec = 1.24E+07
Ke = .5
Figure 14
0 100 200 300 400 500 600
22
20
18
16
14
12
10
8
6
4
2
0
LOAD [ kN ]
Displacement [mm]
LOAD DISPLACEMENT DIAGRAM
B
P
i
l
e
-
S
t
i
f
f
C
l
a
y
-
P
o
o
r
B
a
s
e
CEMSOLVE
ANALYSIS
X = Input Data
Ds = .5 m
Db = .5 m
Us = 550 kN
Ub = 0 kN
Lo = 1 m
Lf = 9 m
Ms = .001
Eb = 1
Ec = 3.1E+07
Ke = .5
Figure 13
18
When the Static Maintained Load test was carried out the result was a load/settlement relationship as shown in the
Figure 14 which might not have raised queries except that the movements are substantially greater than expected.
Analysis revealed that something was wrong. The only variable change in the analysis which could account for the
whole of the behaviour lay in the modulus of the material of the pile.
It was concluded that the pile had been severely over-driven and had been extensively cracked along its length, thus
changing the elastic compression behaviour.
Examination of tests in similar conditions originating from other sources have shown the same problem and this is an
area which we usually view with great caution.
(3) Rock anomalies.
The under performance of driven piles in mudstone rocks has been commented on by a number of authors in journals
over a period of years. It has been found that much the same feature occurs in blocky chalk rocks. An interesting
aspect of this, when subjected to analysis by the Cemsolve method, is that shaft friction appears to be higher than
expected and end bearing to be lower. Figure 15 shows the case of a driven precast pile which is only 5.3 m long and
which was driven by a hydraulic hammer with a set of about 30 mm per 10 blows and a 0.5m hammer drop. The pile
was founded 1.25 metres into mudstone and the overlying soils were of low strength. The net effect is a relatively low
end bearing value and a higher than anticipated shaft friction. The mudstone has a blocky structure with the blocks
being assessed by coring as smaller than the pile width. The inter-block weathered material is soft clay.
0 100 200 300 400 500 600 700 800 900
28
24
20
16
12
8
4
0
LOAD [ kN ]
Displacement [mm]
LOAD DISPLACEMENT DIAGRAM
S
h
o
r
t
P
r
e
c
a
s
t
P
i
l
e
-
o
n
B
l
o
c
k
y
M
u
d
s
t
o
n
e
CEMSOLVE
ANALYSIS
X = Input Data
Xs = .27 m
Xb = .27 m
Us = 360 kN
Ub = 500 kN
Lo = 4 m
Lf = 1.25 m
Ms = .0015
Eb = 200000
Ec = 2.5E+07
Ke = .6
Figure 15
It is believed that in this situation the blocks are pushed downward and effectively become an extension to the pile. As
they are pushed downward they develop friction in the same manner as the shaft of the pile. The end bearing is rather
19
developed by the pressure necessary to push the clayey material from between the blocks. Thus it is possible for the
rock to mimic the pile behaviour. Figure 16 illustrates the mechanism.
Figure 16 Blocky rock behaviour mechanism
(4) Inadequate design method or data.
A feature of the large number of analyses to date is that whereas many of the methods which we use are reasonably
reliable others frequently do not conform with expectation. Particularly it has been found that many of the sandy or
stony clays which are often treated as clays, in practice behave much more like sands. This is true of many, though not
all, of the "boulder clays" of the United Kingdom. Often these same soils show stiffness values (E
b
) which are also
very like sands.
Further, in regard to differences between sands and clays, the time constants used in the TIMESET analyses, are
frequently found to be not nearly as large as classical theory would lead one to expect. The time taken to carry out a
good quality test in sand is in practice not much different to that required in say a stiff clay.
(5) Shaft friction loss due to drilling disturbance.
It has been found that in sands, excessive rotation relative to penetration of continuous flight augers can cause
loosening of the soil and a noticeable loss of shaft friction. Additionally Dutch Cone Penetration tests have verified the
loosening effect.
These observations led to an attempt to model the digging mechanism used for this type of pile. The model has been
published (Fleming, 1995) and has been successfully applied in practice as a means of assessing the effects of the
various dimensional and other parameters.
In the United Kingdom and in many other countries common practice is to calculate shaft friction in sand on the basis
that unit shaft friction is reasonably represented by
f K su s v tan
20
where K
s
an earth pressure coefficient, v is the mean effective vertical pressure in the sand layer and is the angle of
friction between the pile and the soil which is usually taken as equal to the angle of internal friction of the sand in
the case of a continuous flight auger pile.
It has been found that the value of K
s
can fall from a normal value of say 0.7 to about 0.3 depending on the control
exercised over the digging process, though it is of course also a function of the initial ground condition, the auger
diameter, its pitch and other matters.
9. CONCLUSION
The basis of much of our foundation design rests on the results of pile loading tests. In spite of many attempts to find
cheaper or simpler ways of interpreting pile behaviour, such as through dynamic and similar tests, there is no doubt
that the static Maintained Load Test remains the main source of direct and interpretable information. All tests which
do not account for rate effects in terms the overall framework of time - load - deformation give misleading results and
in the end only cause confusion.
The objective of this paper has been to show that there is a simple way of carrying out and interpreting tests and that it
depends on carrying out the work to high standards under computer control and not trying to hurry natural soil
behaviour. The linear-fractional function has been shown to be the most satisfactory for interpretive purposes and this
is proved now by a large amount of high grade evidence.
The effect of adopting a systematic computer based control and analysis system is far reaching and it opens a door to
understanding of real pile behaviour which has been hidden to date. Among the consequences one of the great
opportunities is to cast away crude factoring systems which have existed to date and which seem to be entering into the
Eurocode EC7.
It is believed that many of the findings which have resulted from the development of computer controlled testing and
analysis systems allow insights into the design and construction processes, as this paper seeks to illustrate, and will
eventually lead to improvement of the whole basis of foundation work.
The cycle of Design - Installation Monitoring - Testing - Analysis - Design improvement is now a routine and we are
beginning to look like other industries where this cyclic improvement (Figure 15) is routine.
FIGURE 15
DESIGN
INSTALLATION
DETAIL
ANALYSIS TEST
21
APPENDIX:
Extracts from published papers
Note: The definition of all ultimate states is asymptotic
TIMESET
Relative settlement at timet
Wt
T t
W t
T t
r
s
s
b
b
:
+
+
+
W
s
asymptotic value for shaft related deformation
W
b
asymptotic value for base related deformation
T
s
mobilization time for half shaft related deformation
T
b
mobilization time for half base related deformation
t is time elapsed from application of load
CEMSET/CEMSOLVE
D
s
Diameter pile shaft
D
b
Diameter pile base
U
s
Ultimate pile shaft load
U
b
Ultimate pile base load
L
o
Length pile not frictional
L
f
Length pile with friction
M
s
Shaft/soil flexibility factor
E
b
Secant modulus base(25% u load)
E
c
Pile material elastic modulus
K
e
Equivalent column length/L
f
Applied shaft load at any load stage = P
s
Applied base load at any load stage = P
b
Total applied load at any stage = P
T
Total settlement corresponding to Load P
T
=
t
a = U
s
b = D
b
E
b
U
b
c = M
s
D
s
d = 0.6U
b
e = D
b
E
b
f = e P
T
- a e - b
g = d P
T
+ e c P
T
- a d - b c
h = c d P
T
( )
t
g g f h
f
2
4
2
using positive root.
Total elastic shortening
e
is additive to settlement
For applied loads (P
T
) up to U
s
( )
e
T o e f
s c
P L K L
D E
+ 4
2
_
,
1
]
1
4 1
1
2