Fleming, W. G. K. (1992). GCotechnique 42, No.
3, 411-425
A new method for single pile settlement
W. G. K. FLEMING*
A method is presented for the analysis and predic-
tion of single pile behaviour under maintained
loading, based on the use of hyperbolic functions to
describe individual shaft and base performance.
When these functions are combined, and elastic
pile shortening is added by a relatively simple pro-
cedure, an accurate model is obtained. By a simple
method of linkage, which relies on the fact that a
hyperbolic function of the type described requires
only definition of its origin, its asymptote and
either its initial slope or a single point on the func-
tion, conventional ‘elastic’ soil parameters and ulti-
mate loads may be used to describe total
performance. By means of the changing slope of
such functions, this method also reflects well in the
increase of soil moduli at low strains. Examples
are given from back-analysis of some fully instru-
mented and other cast-in-place pile test results, to
demonstrate that good agreement with all recorded
features can be achieved using the model. Exten-
sive use has confirmed its validity for maintained
load tests in a wide range of soils. Provided that
piles have been made to settle sufficiently under
load, so that the latter part of each relationship is
well defined beyond the stage where shaft friction
is close to a constant value, all the main relevant
parameters can be determined with good accuracy
in back-analysis. The derived data may then be
used to predict behaviour of piles in similar cir-
cumstances on other sites or of piles of different
diameter in the same soils. Subject to the condi-
tions described in the Paper, the method has far-
reaching implications for design, construction and
testing techniques.
INTRODUCTION
In his Rankine Lecture, Poulos (1989) catalogued
the available methods for predicting pile per-
formance under load, ranging from simple to
complex methods using finite element solutions.
He drew attention to the versatility of some of the
more complex methods, but also demonstrated
that in the realm of pile performance prediction,
Discussion on this Paper closes 4 January 1993; for
further details see p. ii.
* Cementation Piling Foundations Ltd; Visiting Pro-
fessor, Department of the Built Environment, Queen’s
University, Belfast.
41
prediction and analysis
L’article prbente une methode pour analyser et
predire le comportement d’un pieu unique sous
chargement continu. Elle est basee sur I’emploi de
fonctions hyperboliques pour decrire les per-
formances du fiit isole et de la pointe. Lorsque ces
fonctions sont combinees et qu’on y ajoute le
raccourcissement elastique du pieu it en resulte un
modele precis. Par une methode tres simple de con-
nexion il est possible d’employer des parametres
elastiques conventionnels du sol et des charges
limites de rupture pour decrire les performances
totales. Cette methode reflete bien I’accroissement
des modules du sol avec basses contraintes. Des
examples sont present& pour dtmontrer que
l’emploi du modele s’accorde bien avec toutes les
donnees enregistrees. Son emploi frequent a confir-
me sa validite pour des essais a chargements con-
tinus pour une large gamme de SOB. Pourvu que les
pieux soient assez enfoncits sous chargement, on
trouve que tous les parametres principaux impor-
tants peuvent &tre determines avec une precision
satisfaisante par analyse retrospective. Alors il est
possible d’employer les donnees derivees pour pre-
dire le comportement des pieux sous des circon-
stances analogues a d’autres emplacements ou bien
de pieux de diametre different dans les memes SOB.
Cette methode a des implications d’une grande
portee pour les etudes, la construction et la tech-
nique des essais.
KEYWORDS: analysis; bearing capacity; field tests;
foundations; piles; settlement.
the result is only as good as the input informa-
tion. The sophisticated input data required are
not normally available from conventional site
investigation, and there would therefore seem to
be a place for a simpler approach that could
readily be correlated with site experience and
mainly used parameters that most geotechnical
engineers would recognize and understand.
Chin (1970, 1972, 1983) has made the method
of plotting the behaviour of both footings and
piles according to the hyperbolic method well-
known. This method has been widely adopted,
although it has not been linked with soil param-
eters, but rather used as a method for defining
ultimate loads.
412 FLEMING

Fellenius (1980) has discussed the Chin method approximately the same yield load. This is an
and other methods for defining ultimate loads; he interesting hypothesis, and appears plausible.
and others have drawn attention to the fact that In this Paper a means of analysis and fore-
the Chin method appears to overpredict. casting pile settlement based on the simple hyper-
However, there is little doubt that in most cases, bolic function is developed. It is first necessary to
according to the plotting method, linear functions consider the obvious criticisms of the use of
represent pile performance very well. Chin’s method in practice so that items that affect
The method is expressed by Chin (1970, 1972) performance and are not normally hyperbolic can
in the form A/P = mA + C,, where A is pile head be separated from the general soil functions.
settlement, P is applied load and C, is a constant. Two obvious features lead to the criticism that
Thus if A/P is plotted against an abscissa of A, a the method overpredicts ultimate load. First, by
linear plot is obtained and the inverse slope l/m the nature of the function, the slope of the plotted
gives an asymptotic limiting value of P. This, lines represents an asymptote in each case. Most
according to the evidence presented by Chin, is definitions of ultimate load are arbitrary, as Fell-
true of piles that carry most of their load by shaft enius (1980) shows, being based either on a settle-
friction, and also of footings and piles that carry ment related in some way to diameter or on
most of their load in end bearing. A typical geometrical manipulation. Most theoretically
relationship between pile head settlement A and satisfactory bearing capacity coefficients are
settlement divided by load A/P is shown in Fig. 1. based on soil mechanisms that would automati-
Many such relationships for piles are bilinear: cally imply asymptotic values. However, asymp-
it has been suggested by Chin & Vail and has totic load values will always exceed those
often been accepted that the first part (A) of the determined arbitrarily. The second distorting
relationship represents shaft friction while the influence is the elastic shortening of the pile body,
second part (B) represents total load. This cannot as can easily be demonstrated by making realistic
be strictly true because of the nature of hyper- estimates of shortening and removing this item
bolic functions, but it can easily be accepted that from the settlement before plotting the functions.
individually shaft and base performance are of It must also be borne in mind, that some
hyperbolic form. driven piles, in particular, show the characteristic
It is interesting to speculate as to why the of set-up, which means that after installation their
simple hyperbolic function should be important frictional capacity increases and on subsequent
in the matter of foundation settlement. Chin loading it declines at large strains. This may also
(1983) suggests that mobilization of stress in a soil be true of certain piles in soft sensitive alluvial
with increase of strain is a function of an increas- deposits, but there is little evidence of it in cast in
ing number of effective soil contacts rather than place piles in overconsolidated soils at least up to
of a general increase of intergranular stress on a movements of the order of 5% of pile diameter.
constant number of grain contacts. He suggests Within this range the stated hyperbolic function
that intergranular stress in a flocculated clay, for appears to hold true. Interestingly, Burland &
example, is virtually constant and independent of Twine (1988) suggest that residual strengths apply
the applied or effective stress. On this basis he along cast in place pile shaft surfaces in clay, and
derives a hyperbolic function for the stress-stain that under maintained loading conditions there is
relationship. It may be visualized that when a soil no decline in load following a peak value, this
is under compressive stress, the load is transmit- being a feature of a dynamic context, for example
ted by internal columnar grain structures and in CRP tests.
that as these reach limiting loads, more and more
columns begin to support load, each having
SETTLEMENT PREDICTION
Settlement and differential settlement are
perhaps the most important features in pile
Settlement/load A/P
design, and the problem is complicated by struc-
tural stiffness, pile load redistribution, construc-
A
tion techniques and group effects. Settlement
4
E control, however, receives the most attention and,
E if the performance of a single pile cannot be ade-
0, B
%
quately forecast, it poses something of a dilemma
co as many specifications include numbers with
“\ which it is difficult to comply without some
understanding of the mechanisms involved. For-
tunately, most specifications are not concerned
Fig. 1. Relationship of settlement and settlement/load with group settlements, although the calculation
 SINGLE PILE SETTLEMENT PREDICTION 413

methods based on elastic theory are of consider- give

able help in this context. The use of empirical
group load reduction factors is now generally dis- KSUSPS
credited, as they have no basis apart from geo- As = ~ (2)
us - Ps
metrical manipulation.
Cementation Piling Foundations Ltd has for Similarly, base performance can be expressed as
many years addressed the problem of pile settle-
ment by simple means, particularly for cases in
AB = ~
&a UBPB
which the design load of a pile exceeds the ulti-
mate shaft friction, using the methods set out by u, - P,
Fleming & Thorburn (1983). It would be useful,
where the load P, corresponds to a settlement
however, to develop a function that could charac-
AB. For a rigid pile, AB is movement at the pile
terize pile load/settlement behaviour, and the
head.
hyperbolic function offers a key to this. Given
that piles in general behave according to such a
function with respect to shaft friction and end
bearing, a method can be derived relatively Shaft friction and settlement
simply by relating the performance to a contin- There is substantial evidence that the settle-
uous function which can mostly be linked to con- ment of a pile shaft for a given load is a direct
ventional soil parameters. function of the diameter Ds (see for example the
finite element studies carried out by Randolph &
Wroth (1982)). Similarly, a considerable number
of studies seem to indicate that K, is an inverse
DEVELOPMENT OF BEHAVIOURAL MODEL function of U,, i.e. settlement for a given load
In Fig. 2, which represents a typical plot as decreases with increasing ultimate shaft load.
used by Chin in considering a truly rigid pile, the Thus, from
slope of A represents ultimate shaft friction and
the slope of B is the ultimate end bearing as Ms 4
defined by vertical asymptotes to the load/ KS = -
settlement relationships. Thus ultimate shaft fric- US
tion is given by
it is found that M, becomes a dimensionless flex-
ibility factor in the nature of an angular rotation,
AS and equation (1) can be rewritten as
Us=(As/Ps) - Ks
Ms Ds Ps
where As represents settlement of the shaft head As = ~
at any load P,, and Ks is the intercept on the us - Ps
horizontal axis. Equation (1) can be rearranged to
Ms is in fact the tangent slope at the origin of the
hyperbolic function representing shaft friction.
Randolph (1991) points out that M, is the
equivalent of jrsJ2G in the notation of Randolph
and Wroth (1978, 1982) where [ is ln(r,,,/rJ, rm is
the radius at which soil deflexions become van-
ishingly small, r, is the pile radius, 7s is the shear
stress at the pile surface and G is the soil shear
modulus. M, is also dimensionless in this nota-
tion. Because G/7, lies in the range 500-2000 in
the findings of Randolph & Wroth, M, would be
expected to have values in the range O~OOllO~OO4.

Base load and settlement

As far as base performance is concerned, the
settlement of a circular footing is commonly
expressed as
Individual shaft and base performance
AB = ; $ D,(l - v”)fi
Fig. 2. Individual shaft and base performance B
414 FLEMING

where E, is the modulus of the soil under the and the base load is available from equation (9)
footing, 4 is the applied base pressure, AB is the
base settlement, Ds is the diameter, v is Poisson’s P, =
4 E, ABU,
(13)
ratio and f, is a standard settlement reduction 0.6U, + D, E, A,
factor related to foundation depth. For increasing
These terms may be expressed more simply and
load on a given foundation this means a linear
handled in a general form by writing the expres-
relationship between load and settlement.
To evaluate the secant modulus E, from a real sion for total applied load at a given settlement
and inserting the total pile head settlement value
load/settlement relationship in a standard way, it
is usual to take its value at one quarter of the AT
ultimate stress in non-linear functions. Thus in
the case of piles equation (6) can be simplified to P, = -+f
aAT bAT
(14)
c + AT d + eA,
A = 0,6075qD, where a = U,, b = D,,EBUB, c= M,D,, d=
B (7)
El3 0,6U, and e = D, E,.
To solve for A, given any specific value of PT,
by attributing values of, say, v = 0.3 and fr =
equation (14) has to be rearranged in the form
0.85.
If at a load of U,/4, equations (3) and (7) are (eP, - ae - b)AT2 + (dP, + ecP,
set equal, the coefficient K, can be determined for
the point where the hyperbolic function and the -ad - bc)A, + cdPT = 0 (15)
linear elastic functions intersect. Thus If for convenience we let eP-, - ae - h =J dP,
+ ecP, - ad - bc = g and cdP, = h, this yields
the solution

A = --9 f J(Y2 - 4fh)

This value of K, can now be used to determine I- (16)
the whole of the hyperbolic function. Equation (3) v”
can therefore be rewritten as Only the positive resulting value of Ar is used.

0.6U, P,
AB = ELASTIC SHORTENING
D, J%(~, - P,)
The elastic shortening of a pile shaft under load
This allows an expression for the total load/ is clearly additional to settlement calculated by
settlement relationship to be formulated. Note the above method, and must depend on the rela-
that within a hyperbolic function of this type it is tive development of load transfer between the pile
necessary only to define the origin, the asymptote and soil along its length, as well as on any free
and one point (e.g. the E,, point) in order to length or near friction-free length at the pile head,
define the whole function. Of course, the secant and on the load being transferred at the pile base.
modulus value in such a function is highest at the To work out the elastic shortening accurately
origin and falls linearly with increasing load, to would require a considerable knowledge of the
zero at the asymptote; this accords with general load transfer flexibility M, along the shaft, and
experience of high E values at low strain. would involve an iterative method, whereby the
pile was divided into elements and compatibility
of strains was studied at given levels. This would
TOTAL SETTLEMENT OF A RIGID PILE make for a somewhat cumbersome procedure,
If a pile is purely rigid, then obviously the involving the complication of varying soil strata
loads taken by the shaft and base can be added to and thickness.
give a total load at any given settlement AT It is suggested that a simplified method can be
As = AH = Ar used: a study of some piles in which elastic short-
(10)
ening has been measured indicates the following
and the total load is method to be sufficiently accurate for most pur-
poses. The simplified method is indicated in Fig.
P, = P, + P, (11) 3, which considers shortening in three stages
The shaft load is available from equation (5), and (a) a free or low friction length extending to a
can be written distance L, from the pile head
(b) a length L, over which friction is transferred
P, =
USAS (c) the whole pile shortening as a column a.fter
(14 the ultimate shaft friction has been reached.
MsDs + As
 SINGLE PILE SETTLEMENT PREDICTION 415

can be expressed as

Friction-free or
A==-- 4 K, LrPr
low friction zone
n D,=E,
When the applied load P, exceeds the ultimate
shaft load Us, additional load causes shortening
Frictional load of the full length L, so that it may be treated
transfer length
simply as a column carrying the excess load, and
the shortening of L, becomes
friction transfer

A = 4 VT - U&F
3 (19)
MobWed base load t (pr - US) for p, > U, n

Fig. 3. Simplified method of calculating elastic short- As total elastic shortening AE is the sum of the
ening elemental shortenings being brought into play,
for loads P, up to the ultimate shaft load Us

The first of these elements is easily considered; A =ywo+wF)

E
the shortening A, is given by n Ds=&
4 LOP, and for
A, z-p (17)
n D,=E,
A$ n & CP&o LF) L, USC1
where Ec is Young’s modulus for the pile material
in compression.
(21)
The second stage represents the elastic short-
ening which takes place during load increase up By the combination of equations (16) and or
to the stage when ultimate shaft friction has been (21) the total settlement pile
mobilized. For uniform friction the elastic short- load the ultimate load may cal-
ening will, at maximum, be equivalent to that of a culated, including a good estimate of elastic
column of length 0.5Lr. However, as Ms and K, shortening.
are both important elements in determining the A computer program has written
early slope of the load/settlement relationship, tate rapid calculation, and given the
and friction development takes place more CEMSET. Help screens been established to
rapidly at the top of this section than at its base, give guidance in choice for
it seems preferable to use an effective column various pile soil.
length slightly lower than 0.5Lr. A series of
elastic shortening comparisons with the present
method have been carried out using the Ran- APPLICATION THE METHOD
dolph and Wroth method in the form of the Having accepted hyperbolic function
PIGLET computer program, based on an elastic closely the load/settlement behaviour
soil of uniform stiffness. These indicate an effec- of piles, the very simple, and
tive column length for this case of O-4Lr. Like- its importance lies in ability link func-
wise, for a uniformly increasing soil stiffness and tion sensibly
strength, from zero at the top of this length, full that should use the asymptotic
mobilization of friction would lead to an effective
column length of 0.67Lr. The elastic method sug- are confusing and
gests that in the early stages of loading an effec- also implies strongly that
tive length of 0.47L, is appropriate. For a typical application any factor load
London clay case, where strength increases lin- alone as a means of controlling deformation is
early from a finite value at the top of the section, crude and
an equivalent column length of 0.45Lr is a reas- The normalized for a wide
onable good approximation. from softest are likely
The effective column length appears to be used to very and soft rocks, are shown
between 70@80% of the distance from the top of Figs 4 5 both and
the friction transfer length to the centroid of the bearing show familiar
friction load transfer diagram. If the coefficient characteristics of pile load/settlement relation-
applied to the friction length to give the effective ships the rigid pile case. Note also that the
column length is denoted as K,, then shortening shape and the function
416 FLEMING

s2
i:
G?
5
54-
Cemset
% analysis: rigid pile
‘,
2
-$$6-
S
$
5 _
rSO

101 , , I I
0 20 40 60 80 100
Load applied/ultimate load: %

Fig. 4. Normal&xl plot of shaft friction settlement relationships

for a range of soils from soft to very stiff (Ms = 0+05-0+00!5)

imply that low strain moduli are always one-third
higher than the E,, value.
It will be observed that, for example, the soil
modulus below a pile base would be of the order
of say 50000 kN/mZ for a stiff overconsolidated
clay with an undrained shear strength approach-
ing 200 kN/m*, giving an E$q, ratio of -30,
whereas the shaft flexibility factor Ms would be
-0.002. A simple comparison of the related
curves shows that they are very different in char-
acter at this level of soil strength. This means
that, if in a pile test the pile has been pushed suffi-
ciently far to mobilize a reasonable part of the
end bearing curve, the equations may be used to
separate and back-calculate all the main param-
eters for the pile. As the base reaction stiffens and
end bearing becomes more ‘brittle’, there is a
remote possibility that the shaft and base charac-
teristics may become too similar to separate
mathematically in a reliable way. Only a very
small proportion of the total range of piles are
likely to be in this category.
If Figs 4 and 5 are used to judge pile per-
formance without reference to the formulae, care
should be taken that the scales are similar to
those used in the diagrams. The graphs, however,
are fully dimensionless and general given the con-
ditions attached to equation (7).

Load applied/ultimate load: %

Fig. 5. Normalized plot of end bearing/settlement relationships for a range

of soils from soft to very stiff: qu = ultimate base pressure (kN/m’); Es =
modulus of soil below base; E,/q, = 5-200; the latter valw corresponds to
soft rock materials
 SINGLE PILE SETTLEMENT PREDICTION 417

It should be stressed that good quality test data Effective column length factor K,
are required for accurate mathematical separation This factor converts the length L, to an effec-
during back-analysis. Tests carried out com- tive free column length. It is necessary first to find
mercially are often good enough to give reason- the centroid of friction transfer by calculation.
able indications of the various parameters, but The friction length down to the centroid should
any improvement in test quality leads to much be multiplied by a factor in the range 0.7-0.8. In
greater accuracy. Quality often suffers because of stiff overconsolidated clays, which increase in
an inability to hold loads constant in maintained stiffness with depth, K, is usually -0.45.
load tests, and because test procedures have long
and short load holding periods which give incon-
sistent degrees of consolidation or creep at the Shaftflexibilityfactor M,
various stages. It also appears that the longer This is found to vary from 0.004 in soft to firm
loads are held constant at any stage, the greater or relatively loose soils to -OX)005 in very stiff
are the errors in the measuring system as a result soils or soft rocks. As stated, it lies in the range
of factors such as temperature variation. For the that would be expected from Randolph and
purposes of accurate back-analysis, the settlement Wroth (1978), and decreases with increasing soil
at each load hold should be projected to infinite stiffness. In stiff overconsolidated clays, for
time before plotting the points to be used. It is example, it is found to be in the range O%I-
found that the results of continuous rate of pen- 0.002, although some variations are found, even
etration tests can be analysed approximately, but on a single site, which appear to be related to pile
the stiffness coefficients obtained are naturally type, construction practice, pile straightness and
higher than one obtains from maintained load possibly time-dependent construction processes.
tests. It is also clear that, as Burland & Twine
(1988) suggest, ultimate shaft load is increased
with subsequent decline if this test procedure is Modulus of soil beneath pile base E,
followed, so that rather distorted ultimate load Back-analysis shows this to be one of the most
results are obtained. interesting parameters of the method. It is obvi-
The method described has been used both as a ously related to the intrinsic soil properties, but it
predictive and an analytical tool for maintained is also highly construction dependent. There is a
load tests by Cementation Piling Foundations wide range of choices, depending on whether a
Ltd for three years. It has proved useful in assess- pile is driven or bored, and pile base condition is
ing whether or not piles under test will perform very important.
according to specification, and in discovering Overconsolidation has an important effect on
appropriate parameters to use in future designs. most soils. As site investigations as carried out at
In the analysis mode it is analogous to the signal present are more concerned in practice with
matching procedure now used in dynamic load strength than with deformation, this factor is not
testing, and it similarly requires a certain degree usually directly determinable. Instead, there are
of movement to acquire adequate data. Its advan- several attempts in the literature to establish
tage in comparison with dynamic signal matching stiffnesses by correlation with other soil proper-
lies in the fact that the dynamic procedure ties, for example by Meigh (1987), Burland &
involves many more parameters and at present Burbridge (1985) and Stroud (1989). These are
relies on an inferior bilinear model. The following helpful in regard to the factors that generally
comments on the various parameters may be influence stiffness, but data from a pile loading
helpful. test seem to be best, as they also incorporate the
construction factors. Indeed, it would seem highly
desirable to test piles to higher loads and greater
Diameters settlements than is done at present in order to
The diameters of the shaft and base are regard- establish all the parameters reliably.
ed as known items (Ds and D,J. Equivalent diam-
eters can be used for non-circular sections.
Concrete modulus Ec
In practical terms it seems highly desirable to
Length obtain the E, value directly from the material of
The overall length must be known. The com- the pile. A common figure for concrete piles at the
ponent L, is the free length or length through fill age of test is -30 x lo6 kN/m’, but with high
or soft alluvial deposits from the pile head. These strength mixes and excellent curing conditions in
soft soils rarely contribute significantly to bearing cast-in-place piles, values as high as 50 x lo6
capacity. The component L, is the pile length kN/m’ and infrequently higher seem to occur. A
transmitting load by shaft friction. short extensometer or set of extensometers in the
418 FLEMING

head region of a pile, but outside the zone of
stress concentration below the load application
level, seems an adequate answer to the problem.
An alternative might be to cast a short dummy
pile nearby, which could be extracted and tested
in a testing press concurrent with pile loading.
all the parameter determinations are greatly
diminished. Using the computer program, it is a
simple matter to investigate sensitivity in any par-
ticular case, and it can easily be appreciated that
sensitivity depends on the relative magnitude of
the parameters in individual situations.

Ultimate shaft load U, Examples

At present, conventional means of calculation A large and growing number of field test results
are used for forecasting the ultimate shaft load. have been examined by this method, and it is
However, back-analysis shows that in reality con- clear that with good data piles in a wide range of
ventional calculation is usually conservative but soil conditions follow the calculated form very
occasionally not so, possibly due to installation closely indeed. At present, most of the piles that
techniques that may alter the surrounding soil have been examined are of the cast in place type.
properties. This is particularly likely to happen in Where it is possible to find instrumented pile
the interglacial sands and silts with certain types tests, the data are usually good enough to
of bored pile. confirm that the base alone, the elastic shortening
and the pile as a whole can be modelled closely.
The following examples have been selected
Ultimate base load U, from the database of pile tests back-analysed by
The ultimate base load is also calculated by the method to illustrate its application in a range
conventional means for the purpose of prediction. of ground conditions and for cast in place piles of
Again, using the logical asymptotic definition it different types. It is a simple matter when the
seems from analysis that for deep bases in clay database is sufficiently large to use the method for
soils the N, factor is slightly higher than the con- prediction purposes, as the main parameters are
ventional value of 9. Installation method is of remarkably consistent with specific ground condi-
primary importance, particularly for short piles, tions and installation. The database currently
which rely heavily on end bearing. For conven- extends to some 200 cases. All the input data
tional bored piles in such circumstances the points used in the examples are taken directly
cleaning of bases is important, continuous flight from site records.
auger piles behave well given good construction
techniques, and driven piles obviously densify
cohesionless soils markedly in most cases. The Bored piles in stiflclay soils
stiffness of the soil in such circumstances may be Useful information can be found in Whitaker
increased by a factor of two or three for a driven & Cooke (1966) which deals with instrumented
pile, and is often even higher where the technique tests carried out at Wembley on both straight
of driving bulbs is used. Data on all the param- shafted rotary bored piles and under-reamed
eters are currently being collected for a wide piles. The paper provides information on the soil
range of pile types and ground conditions; it is conditions, and although the majntajned load
hoped to pubhsh the more important findings in data are given in detail over only part of the total
due course. load settlement curve, the ultimate base and shaft
loads are quite accurately known. Whitaker &
Cooke took a definition of failure as correspond-
Sensitivity ing to about 10% of pile base diameter, and the
From the equations and Figs 4 and 5 it will ultimate loads were determined by continuous
readily be appreciated that the most important penetration tests. The data are fuller in some
parameters in the early stages of any pile load set- cases than in others; the maintained load results
tlement relationship are the M, and Ec values. for two straight shafted piles have been chosen
Fortunately, in most cases these parameters have for illustration purposes.
very limited ranges and have only minor effects In each case the compatibility of the solution
on the ultimate shaft friction, end bearing and has been checked against the ultimate load given,
base soil stiffness moduli where movements are and the base/settlement relationship has been
large in back-analysis. The E, and U, values checked independently. These piles were not
have significantly different effects, and with sufti- made to settle sufficiently during the maintained
cient settlement data can be separated readily. load test to give a clear solution for the base from
The most important consideration is that if piles the overall settlement data, but, usefully, this is
are made to settle well beyond the stage where supplemented by records from each pile base,
shaft friction is fully mobilized, potential errors in allowing a full solution to be obtained. The solu-
 SINGLE PILE SETTLEMENT PREDICTION 419

x = inputdata \
D, = 0.775 m
D, = 0.775 m
Us= 1994kN
U,= 1002kN
L, = 1.4 m
L, = 10.8 m
M, = 0.0017
E, = 33125
E; = 1.95E + 07
Whltaker & Cooke (1966)
K, = 0.45
--- Base
--- - Elastic shorteni,q
I , I 1 I

40 60 120 160 200

Load t
Fig. 6. Comparison of results from the present study with those of Whit-
aker 81 Cooke (1966) : pile H

tion given in Figs 6 and 7 is entirely compatible Under-reamed pile in stijfclay

with all the information supplied. For pile H, It is difficult to find results for under-reamed
Whitaker & Cooke quoted an ultimate shaft load piles that have been made to settle significantly;
of 1960 kN and a base load of 770 kN. These again the work of Whitaker & Cooke (1966) at
figures correspond to asymptotic values of 1994 Wembley provides an interesting case. No satis-
and 1009 kN respectively. For pile N, ultimate factory and straightforward result for the behav-
shaft and base loads were given as 3070 and 870 iour of under-reamed bases at this site could be
kN; the analysis by this method corresponds to found by the matching program until the original
asymptotic values of 3100 and 1068 kN. The paper was studied more carefully. Pile P has been
analysis values correspond closely to the Whit- taken as an example. The under-reaming tool
aker and Cooke values, if 10% of diameter settle- produced a dome-shaped upper surface, and did
ment criterion is taken as defining ultimate load. not at the time conform with usual specification
The method represents well the performance of requirements that the side slope should make an
bored piles in stiff overconsolidated clay. angle of 60” or more with the under-ream floor.

x = lnpuldata
D, = 0.94 m
Db = 0.94 m
U, = 3100 kN
&,=1066kN

---- Elastic shorlemng

I I I I I I
20
0 60 120 180 240 300 360

Load t
Fig. 7. Comparison of results from the present study with those of Whit-
aker & Cooke (1966): pile N
420 FLEMING

Whitaker 8 Cooke (1966)

x = input data
D, = 1.854 m
I&,= 1.854m
U, = 700 kN
U, = 3940 kN
L, = 0.001 m
L, = 0.5 m
M, = 0.0015
Eb = 53819
E, = 9E + 10
K. = 0.45
----Base

=--G-
0 ’
I

100
Load: t
4 I
150
I I

200
I

Fig. 8. Under-reamed pile P at Wembley: base characteristic only

Hence the under-ream was cut, then the shaft was expose features of construction that might other-
deepened a little and a second cut was made. This wise go unnoticed.
allowed the composite under-ream to meet the
specification requirement, but inevitably produc-
ed a peripheral surface on which friction could Driven piles in dense sand
act. Once friction was allowed a part in base per- The examples given in Figs 10 and 11 are from
formance, the solution shown in Fig. 8 resulted de Beer, Lousberg, de Jonghe, Wallays & Carpen-
for pile base capacity; the consequent solution for tier (1979). A series of Franki piles, with and
the complete pile is shown in Fig. 9. In the total without enlarged bases, were driven through - 8 m
solution, the 700 kN of shaft resistance on the of very soft clay and peat to a penetration of just
base now appears in the shaft result, and the over 1 m in very dense sand. The enlarged based
overall solution is exactly compatible with that piles were subsequently extracted and measured,
given by Whitaker and Cooke, bearing in mind so the dimensions are fully known.
their criterion of approximately 10% of base Although four of these piles have been exam-
diameter for the ultimate condition. ined in detail, two have been chosen to exemplify
The method can thus represent well the case of the results. Fig. 10 shows the results of analysis
an under-reamed large diameter bored pile in all on pile V in the series, a straight-shafted pile. This
its aspects in stiff overconsolidated clay, and can pile was cast within a 406 mm steel tube: for the

-->
4
! x
Cemsolve
analvsis

input
h-1

;: = m
Db=

1
. U, 2905
U, = 3940 kN
:: L, = 0.5 m
d
Lf = 14.1 m \
M, = 0.0012
ED = 53819
E, = 1.6E + 07
K, = 0.45
---- Elastic shortening
161 I I I I I I I I
0 100 200 300 400
Load: t
Fig. 9. Under-reamed pile P at Wembley: total pile performance
 SINGLE PILE SETTLEMENT PREDICTION 421

20- Cemset
analysis
:
c: x = input data
D, = O+ZJ m Pile 5
5
E 40- & = 0.406 m
d U, = 140 kN
% U, = 1790 kN
d
L, = 0.2 m
L, = 1.33 m
60 - M, = 0.001
Eb = 416427
E, = 3.5E + 07
K. = 0.45
406 mm

YY

0 40 80 120 160 200

Load: t
Fig. 10. Result of pile load test on a driven cast-in-place pile at Kallo,
near Antwerp

purpose of calculating elastic shortening the steel piles with enlarged bases. It is of interest to note
area has been converted to equivalent concrete by that the stiffness of the base reaction is substan-
the modular ratio method. The ultimate shaft tially different: this is much more likely to be due
load was very small and could not be determined to construction technique than to natural varia-
accurately, but the base settlement characteristic tion in the founding layer.
clearly conforms with Fig. 5, the soil modulus E,
value being approximately 416000 kN/m’ and
the ratio E,/qu being 30. Bored piles in chalk
Pile 2 (Fig. 11) was driven through a slip-sleeve Figure 12 shows the results of a test on a pile in
arrangement and had an enlarged base. The fric- chalk at Norwich. This was an instrumented pile,
tion on this pile was effectively removed, and the for which data have kindly been provided by Ove
soil modulus below the base now appears as Arup. The chalk in this instance has standard
1000 000 kN/m’. Again the form of the result is as penetration test results of the order of N = 10,
indicated in Fig. 5 (l&/q, = 79). The method rep- and because the pile was instrumented by the
resents well the performance of driven piles in a Building Research Establishment, the base, total
dense sand both for straight-shafted piles and and elastic shortening characteristics were all

o*

20-
Cemset
: analysis Pile 2

c: 40 _ x = input data Drivmg tube

5 D, = 0.324 m
Db = 0639 m
E
U, = 0.01 kN
B
:: 60 - IJb = 2890 kN
5 L, = 9.26 m
L, = 0.4 m
M, = 0.001
=
80 - Eb = 5E
E, 1 000 000
+ 07 1
K. = 0.45 e=15mm

539 mm dia.
100 I I I , I
0 60 120 180 240 300
Load: t
Fig. 11. Result of pile load test on a driven cast-in-place pile at Kallo,
near Antwerp
422 FLEMING

--- Base
- --- Elastic shorlemg
I I I \ ,
160 240 320 400
Load. t

Fig. 12. Large diameter pile in soft chalk

measured. Fig. 12 shows the agreement of the ranging up to very hard, and that very good
solution with all the data. It will be observed that matching is possible in all cases.
although the chalk was generally of relatively
poor quality, the base soil modulus is 84000
kN/m’, and while this value may be due to Piles in silty conditions
harder layers in the soft chalk, there is evidence The results of two tests on piles constructed
that even soft chalks show relatively high using continuous flight augers at Shrewsbury in
modulus values. The details of this pile are in very complex silty conditions are shown in Figs
Twine & Grose (1989). 13 and 14. These piles are of interest because they
Again, the method represents well in detail the were made to settle a long way under load. Below
results of an instrumented large diameter pile in some 6 or 7 m of organic silt and clay there were
chalk. This chalk was weak according to the site layers of very silty sands, clayey silts and silty
investigation information, but the end bearing is clays. Results of Dutch cone tests varied violently
higher and the base stiffness is greater than might with friction ratios in the range 2-4. These results
have been expected. The elastic shortening is also and those from standard penetration tests imply
well represented. Note that there are several piles loose to at best medium-dense conditions, with
tested in chalk in the database, with chalk SPT N values increasing from 8 or 9 at the top to

OX__
x Shrewsbury TPl

Cemset
analysis
E 40
x = input data
-:5 D, = 0.75 m
b

ji , !f:

>
/
0 100 200 300 400 500 600
Load: t
Fig. 13. 750 mm continuous flight auger pile 27 m long, in Shrewsbury
 SINGLE PILE SETTLEMENT PREDICTION 423

OX

Shrewsbury. TP2

2oc
Cemset
.WlalySlS

x = input data
E 40- D, = 0.75 m
i:
& D, = 0.75 m
E U, = 2350 kN
4 t/b = 3709 kN
60-
% L,=6m
6 Lt = 26 m
M, = 0.00095
Eb = 327 686
80 - EC = 34E + 07
K, = 0.53

100 1 I I I 1
0 120 240 360 480 600
Load: t
Fig. 14. 750 mm continuous flight auger pile, 32 m long, in Shrewsbury

about 20-30 near the pile bases. Groundwater
stood near the piling surface level. Towards the
lower end of each pile conditions became a little
more sandy, but bands of silty clay and clayey silt
persisted. The area is well known for its difficult
piling conditions.
The results of test loading are of good quality
and show that in spite of the very mixed ground
conditions, end bearing is a more significant com-
ponent of capacity than might have been
expected. The full computer solutions are shown;
again good agreements with the hyperbolic equa-
tion forms are apparent.
These are long piles, and the Young’s modulus
for the concrete may not be exact. If the concrete
modulus is varied it is found to have only a very
minor influence on the resulting load distribution
between shaft and base, using the least squares
curve fitting method contained within the
analysis program for all the variable parameters.
The matching and the values of parameters are
good and stable, particularly in the case of test
pile 2, where the data quality is better.
Pile in weathered Mercia mudstone
Figure 15 shows the results of a test on a con-
tinuous flight auger pile founded in a weathered
Mercia mudstone in the Bristol area. Again the
pile was made to deflect sufficiently to give a
good fix on the various parameters. The soil was
layered with softer and harder bands in the
region of the toe of the pile, but it is evident that
the pile base behaved in accordance with effective

E, = 4.5E + 07
40
t K, = O-5

I I I , \ I
60 120 180 240 300
Load. t
Fig. 15. Pile founded in weathered Mercia mudstone
424 FLEMING
stress parameters, although for this ground condi-
tion the modulus of soil below the base is not as
high as might be expected. The ground consisted
of fill and soft peaty clays to a depth of 7.2 m.
Below this was soft clayey silt to just over 10 m
depth (N = 6). The pile then entered sand and
gravel (N = 49, and at a depth of 14.2 m encoun-
tered Mercia mudstone with SPT results up to
N = 120. Water stood at 2.4 m below ground.
This result is of interest because it appears to
be possible to achieve good results even in a vari-
able and layered founding material. The relatively
high concrete modulus is indicated by small set-
tlements which occurred in the early stages of
loading.
Summary of examples
The above examples show that using the
method, both the form of the load/settlement
relationship and the derived parameters provide a
very good representation of pile behaviour. Many
other cases have been examined in an extensive
range of soils, showing that the method is capable
of a wide range of application provided that tests
produce sufficient data for analysis.
CONCLUSIONS
The method described is straightforward and
its success depends mainly on three factors. These
are the accuracy of the model, the parameter
linkage and the elastic shortening calculation.
Chin’s method has been used for many years
and demonstrates the basis of the model well,
although imperfectly in that it deals with a func-
tion of an amalgam of base resistance and shaft
friction load and not with components separately,
nor does it take into account adequately elastic
shortening. Once these problems are dealt with,
the model is exceptionally good, as is demon-
strated in this Paper and elsewhere for cast in
place piles. No case in the large number back-
analysed does not conform to the model with a
high degree of accuracy. Studies have dealt with
piles carrying load mainly by shaft friction and
mainly by end bearing, and with a wide range of
combinations in many different ground condi-
tions. Some tests have involved direct measure-
ment of friction and end bearing; these also
indicate the reliability of the model.
The method linking the mode1 with conven-
tional soil parameters is mathematically straight-
forward. It recognizes that only the initial slope of
the described function, or a single point on it, has
to be defined in addition to the origin and the
asymptote. Where the single point method is
used, this is provided by the conventional secant
modulus as used in other soil techniques. The
linkage method fits comfortably with the known
increase of secant modulus at low strains.
The mode1 is intended to take elastic short-
ening into account, not to determine it with
absolute precision. It is adequate for the purposes
of prediction and analysis. In practice it can have
little effect on the ultimate states and in particular
on the modulus of the soil below the pile base,
because in general these determinations involve
much larger movements.
The method has been in use for three years,
and has led to a series of improvements in testing
practice, to be covered in a further paper. As
methods have been improved, the fit of higher
quality data with the model has also improved
noticeably. It should be stressed, however, that a
sensible analysis can only be carried out on piles
that have been made to settle sufficiently to mobi-
lize a reasonable part of the end bearing charac-
teristic. In mathematical terms the solution of
each set of data is simply that of a set of simulta-
neous equations, using screen graphics as an aid.
The method calls for an asymptotic definition
of ultimate loads as originally suggested by Ter-
zaghi (‘We identify the failure of the earth support
with the transition of the (load/settlement) curve
into a vertical tangent’). This removes the need
for multiple definitions, which are mainly strain-
related, for lack of any other means of consider-
ing deflexion constraints. This method, although
not primarily designed to do so, allows any other
definitions to be put in context simply by assign-
ing a limit to deflexion and calculating the corre-
sponding load.
In order to use the method correctly in back-
analysis, not only should test piles be moved sufh-
ciently to mobilize a significant part of the base
load, but settlement should be projected under
each load to infinite time in order to remove time
dependence from the result. An accurate tech-
nique has been developed for this purpose, and
will be described in a future paper. Current
testing practice therefore requires some recon-
sideration.
Back-analysis of tests shows many interesting
features; for example, if a pile base is resting on
debris, the effect becomes obvious. It can also
show where the load-bearing soil around a pile
shaft is severely disturbed by a boring process, as
this leads to significant reductions in expected
friction.
The method implies that to obtain maximum
information from any preliminary test on a site,
the pile may be constructed to found in the same
ground as intended working piles by the same
technique, but made to have a smaller diameter.
This would mean for a given load and cost, the
pile could be made to settle substantially further,
allowing good determinations of all the relevant
 SINGLE PILE SETTLEMENT PREDICTION 425

parameters. Once the parameters are known, it is settlement of pile base under applied load
necessary only to insert the changes of pile dia- total elastic shortening of pile
meter into the equations to discover the likely settlement of pile shaft under applied load
performance of contract piles, and whether or not total settlement of rigid pile under applied
they will comply with specification. load P
Internal pile instrumentation is expensive and
components of elastic shortening of pile
there are many cases, such as continuous flight
piles, in ln (r,/r,)
auger which full load-recording Poisson’s ratio
equipment cannot at present be inserted to posi- shear stress at pile surface
tions where it would be useful. Subject to the con-
ditions stated, this method would appear to offer
valid results at the reasonably low cost of suffi-
cient load application. Direct measurement of REFERENCES
concrete elastic properties would be a useful and Burland, J. B. & Burbridge, M. C. (1985). Settlement of
fairly straightforward addition to the system. foundations on sand and gravel. Proc. Instn Ciu.
The method has been shown to have many Engrs, Part 1, 85, Dec., 132551381.
Burland, J. B. & Twine, D. (1988). The shaft friction of
consequences. The load/deformation performance
bored piles in terms of effective strength. Deep foun-
of piles is not a matter of random behaviour.
dations on bored and augered piles. Rotterdam:
Balkema.
Chin, F. K. (1970). Estimation of the ultimate load of
piles from tests not carried to failure. Proc. 2nd SE
Asian Conf: Soil Engng, Singapore, 81-92.
NOTATION Chin, F. K. (1972). The inverse slope as a prediction of
43 diameter of pile base ultimate bearing capacity of piles. Proc. 3rd SE
4 diameter of pile shaft Asian Conf. Soil Engng, Hong Kong, 83-91.
43 deformation secant modulus for soil Chin, F. K. (1983). Bilateral plate bearing tests. Pro-
beneath pile base at 25% of ultimate stress ceedings of international symposium on in situ testing,
EC Young’s modulus of pile concrete Paris 2, 29933.
E, Young’s modulus for any pile material Chin, F. K. & Vail, A. J. (1973). Proc. 8th Int. Conf: Soil
KE effective column length of shaft transferring Mech., Moscow 2, Part 1.47-52.
friction, divided by L, De Beer, E., Lousberg, E., De Jonghe, A., Wallays, M. &
KS,& intercepts on settlement/load axis when Carpentier, R. (1979). Prediction of the bearing
settlement is plotted against capacity of displacement piles, penetrating into a
settlement/load very dense sand layer. Proc. 7th Eur. Conf: Soil
upper length of a pile carrying no load or Mech., London 3, 51-59.
low loads by friction Fellenius, B. (1980). The analysis of results from routine
length of a pile transferring load to the soil pile tests. Ground Engng 6, Sept., 19-31.
by friction Fleming, W. G. K. & Thorburn, S. (1983). Recent piling
G shear modulus of soil advances. In Piling and ground treatment, pp. l-16.
Ms flexibility factor representing movement of London : Thomas Telford.
a pile relative to the soil when transferring Meigh, A. C. (1987). Cone penetration testina-methods
load by friction (dimensionless) and interpretation. London: CIRIA/Butterworth.
N standard penetration test result Poulos, H. G. (1989). Pile behaviour-theorv and annli-
I &

P load applied at pile head cation. Gtotech&ue 39, No. 3.

PLl load applied at pile base
Ps load applied to pile, carried by friction
P* load (P, + PJ applied at pile head
us ultimate shaft friction load
UB ultimate pile base load
a, b, c, d,
e,J g compound parameters
CI constant (Chin model)
C” undrained shear strength of clay
f, depth factor related to depth of foundation
below ground
m slope of line relating settlement to
settlement/load (Chin model)
4 stress due to applied load at pile base
rc pile radius (Randolph model)
rm radius at which soil deflexions become
vanishingly small (Randolph model)
adhesion factor
: settlement
Randolph, M. F. (1991). Private communication.
Randolph, M. F. & Wroth, C. P. (1978). Analysis of
deformation of vertically loaded piles. J. Geotech.
Engng Div. Am. Sot. Ciu. Engrs 104, GT 12, 1465-
1488.
Randolph, M. F. & Wroth, C. P. (1982). Recent devel-
opments in understanding the axial capacity of piles
in clay. Ground Engng 15, No. 7.
Stroud, M. A. (1989). The standard penetration test-its
application and interpretation. In Penetration
testing, pp. 29-49. London: Thomas Telford.
Terzaghi, K. von (1944). Theoretical soil mechanics, 2nd
edn, p. 118. New York: Wiley.
Twine, D. & Grose, W. (1989). Discussion on founda-
tion. In Chalk, p. 417. London: Thomas Telford.
Whitaker, T. & Cooke. R. W. (1966). . An investigation of
the shaft and base resistances of large bored piles in
London Clay. In Large bored piles, pp. 7-49.
London: Institution of Civil Engineers.