Pile Foundations: Experimental Investigations, Analysis and Design
Pile Foundations: Experimental Investigations, Analysis and Design
Pile Foundations: Experimental Investigations, Analysis and Design
net/publication/280844461
CITATIONS READS
95 2,120
3 authors, including:
Some of the authors of this publication are also working on these related projects:
Piles under horizontal load / Energy piles / Disconnected piles under vertical load View project
All content following this page was uploaded by Alessandro Mandolini on 17 July 2015.
A. Mandolini
Department of Civil Engineering, 2nd University of Napoli, Italy
G. Russo & C. Viggiani
Department of Geotechnical Engineering, University of Napoli Federico II, Italy
ABSTRACT: Selected topics in the field of pile foundations are addressed. The effects of the installation technique on the bearing ca-
pacity and the load-settlement response of a single pile are discussed. The latter effect is shown to be less significant; a settlement
controlled design is thus less dependent on the technological factors. Monitoring of the installation parameters shows some potential
for controlling the pile response. The available experimental evidence on the behaviour of pile foundations under vertical loads (set-
tlement, load sharing, bearing capacity), by monitoring of full scale structures or by research experiments, is reviewed. Simple em-
pirical methods for a preliminary evaluation of the settlement are suggested. The (more limited) evidence about horizontal loading is
also reviewed and discussed. The methods for the analysis of pile foundations under vertical load are next reported. They may be con-
sidered satisfactory for engineering purposes, provided they are used paying due attention to the correspondence relations between
theories and reality. The criteria for an optimum design, achieving maximum economy while keeping satisfactory performances, are
different for different kinds of pile foundations (small groups, large rafts). Safety against a bearing capacity failure, average settle-
ment, differential settlement, moment and shear in the raft and cost are the quantities to be controlled. It is claimed that the conven-
tional capacity based approach, still prevailing in practice, is not suited to develop a proper design. Present codes and regulations, es-
sentially based on this approach, at the time being act as a restraint rather than a stimulus and need some revision.
RESUME: On présente une sélection de thèmes concernant le domaine des fondations profondes. On discute de l’effet du mode
d’installation sur la capacité portante et sur la réponse charge-tassement d’un pieu isolé. On montre que ce dernier effet est moins si-
gnificatif ; un dimensionnement fondé sur des critères de tassement est donc moins dépendant des facteurs technologiques. Le suivi
des paramètres d’installation représente une voie potentielle pour le contrôle de la réponse du pieu. Une revue des données expéri-
mentales sur le comportement des fondations sur pieux sous charges verticales (tassement, distribution de la charge, capacité portante)
est présentée à travers les mesures effectuées sur des ouvrages réels ou celles sur des expériences de recherche. On propose des mé-
thodes empiriques simples pour une première évaluation du tassement. Le cas (plus limité) du chargement horizontal est également
passé en revue et discuté. Les méthodes d’analyse des fondations sur pieux sous charge verticale sont ensuite discutées. Elles peuvent
être considérées comme satisfaisantes pour les besoins de l’ingénieur, pourvu qu’elles soient utilisées en faisant bien attention aux re-
lations de passage de la théorie à la réalité. Les critères pour un dimensionnement optimum, le plus économique tout en gardant la
meilleure performance, sont différents selon les différents types de fondations sur pieux (groupes à faible nombre de pieux, radiers de
grande dimension). Les paramètres à contrôler sont la sécurité vis à vis de la capacité portante limite, le tassement moyen, les tasse-
ments différentiels, les moments et cisaillements dans le radier, et enfin le coût. Le dimensionnement classique fondé sur la capacité
portante, qui prévaut encore dans la pratique, n’est pas adapté pour développer un dimensionnement approprié. Par conséquent les co-
des et règlements actuels, fondés essentiellement sur cette approche, représentent une restriction plutôt qu‘un stimulant et nécessitent
une certaine révision.
1 INTRODUCTION placement screw piles are about 60% of the total installed yearly
in Belgium (ten times than in the world market) while bored and
Piles have been used by mankind for foundation purposes since CFA piles reach more than 90% of the total in Italy (about two
prehistoric times; their behaviour, however, is far from com- times than in the world). Again in Italy, in recent years CFA
pletely clear and a substantial volume of research is being car- piles gained market against other bored piles, increasing from
ried out on the subject. The field is in evolution with continuous about 30% (Trevisani, 1992) to about 55% (Mandolini, 2004).
developments in the technologies, in the methods of analysis The regional practice in the different countries develops
and in the design approaches. In fact the design of piles is a along different paths under the push of the local market. Such a
rather complex matter which, although based on the theoretical situation brought the Belgian Geotechnical Society and the
concepts of soil mechanics, heavily relies on empiricism. This is European Regional Technical Committee (ERTC3) of the ISS-
an inevitable consequence of the marked variability of behav- MGE to organize an International Seminar on the design of axi-
iour of the piles, which is partly due to random factors but de- ally loaded piles (De Cock & Legrand, 1997) with the aim of
pends also on the effects of the installation techniques (De Beer, reviewing the practice in the European countries. Irrespective of
1988; Van Weele, 1988; Van Impe, 1991; Viggiani, 1989, the most widespread type of pile in each country, the contribu-
1993). tions to the Seminar confirmed that the common approach for
According to Van Impe (2003), bored and CFA piles account the design of a single pile is still based on semi-empirical rules,
for 50% of the world pile market, while the remaining is mainly sometimes calibrated against purposely performed load tests
covered by driven (42%) and screw (6%) piles. Summing up, (Van Impe et al., 1998).
the market is equally subdivided between displacement (driven, A number of comprehensive and authoritative reports on pile
jacked, screwed, etc.) and non-displacement piles (bored, con- foundations have been issued in recent years (Randolph, 1994,
tinuous flight augered, etc.). 2003; Poulos et al., 2001; Mandolini, 2003; Poulos, 2003); ac-
Different proportions may be found locally: for instance dis- cordingly, the present Report will not attempt a complete cover-
177
Mandolini, A. et al. Pile foundations: Experimental investigations, analysis and design. XVI ICSMGE, Osaka 2005. Millpress Vol. 1
Made available with permission from the Publisher: Millpress Science Publishers, Rotterdam, The Netherlands
age of the matter, but rather address some selected topics be- at a displacement of the pile head equal to 10%d, either directly
lieved to be timely and relevant. attained in the test or determined by hyperbolic extrapolation.
The results are summarized in table 1 in terms of the ratio QS/P.
Bored piles give the smallest value (QS on average 12 times
2 SINGLE PILE greater than the weight of the pile) and the larger scatter (COV
= 26%); driven piles give the largest value (73 times the weight
of the pile) and the smallest scatter (COV = 8%); CFA piles
2.1 Experimental evidence and investigations have an intermediate behaviour.
178
Made available with permission from the Publisher: Millpress Science Publishers, Rotterdam, The Netherlands
2,0 nique affects the axial stiffness of the piles much less than their
bearing capacity (Poulos, 1989; Viggiani, 1989, 1993;
a) Randolph, 1994; Van Impe, 1994). The data collected in table 1
1,5 and in figures 1 and 2 seem to support this view and confirm
that the initial stiffness of the piles depends primarily on the
small strain shear modulus of the soil (Mandolini, 1994;
K0 / K
1,0 VP > VPcrit, the removed volume is less than the displaced one
(net compression effect, similar to that of a driven pile); if VP <
VPcrit, the opposite is true (net decompression effect, similar to
0,5
that of a bored pile).
Viggiani (1989) found that, in order to satisfy condition (3)
0,0 whatever torque MT is available, a substantial vertical thrust is
BORED
DRIVEN
DRIVEN
CFA
CFA
CFA
CFA
CFA
CFA
CFA
179
Made available with permission from the Publisher: Millpress Science Publishers, Rotterdam, The Netherlands
0,75
0,85
0,95
250
500
100
150
150
250
350
10
20
30
10
20
20
40
50
50
0
0
0
pile n° 1
10
depth [m]
20
V Pcri dN
30
0,55
0,65
0,75
250
500
100
150
150
250
350
10
20
30
10
20
20
40
50
50
0
0
0
pile n° 2
10
depth [m]
20
V Pcri dN
30
0,75
0,85
0,95
250
500
100
150
150
250
350
10
20
20
40
50
50
10
20
30
0
0
0
pile n° 3
10
depth [m]
20
V Pcri dN
30
Figure 3. qC-profiles and parameters measured during the installation of three piles: n° 1 (total length LT = 25.5 m; embedded length L = 24 m; nominal
diameter dN = 0.8 m), n° 2 (LT = 24 m; L = 22.5 m; dN = 0.6 m) and n° 3 (LT = 25.1 m; L = 24.1 m; dN = 0.8 m).
concrete and the nominal volume is equal to 1.27⋅VC / but not for pile n° 1. Within the base formation of pozzolana,
(dN2VR); if it is above unity, the effect is a lateral compression on the contrary, VP < VPcrit; in that soil all the piles were thus
of the soil and hence a better behaviour of the pile, but also installed essentially by boring.
over-consumption of concrete and cost increase (d > dN). The results of the load tests are reported in figure 4 as
Three load tests to failure on trial instrumented CFA piles load-settlement curves (total load Q, shaft load, S and base
have been recently performed at a site were the subsoil condi- load P) and load distributions along the pile shaft. Some rele-
tions are relatively uniform in horizontal direction (Mandolini vant data are listed in table 2.
et al., 2002). From the ground surface downwards the follow-
ing soils are found: (a) topsoil, about 1 m thick; (b) alluvial Table 2: Results of load tests
soils of pyroclastic origin tightly interbedded with organic silt Pile dN L Qmax wmax Pmax Smax
layers, about 20 m thick; (c) base formation of pozzolana to n° (m) (m) (MN) (mm) (MN) (MN)
the maximum investigated depth (50 m). The groundwater ta- 1 0.8 24.0 4.08 75.6 1.55 2.81
ble fluctuates between 1.2 m and 1.6 m below ground surface. 2 0.6 22.5 3.26 81.9 0.89 2.59
Three CPT profiles are reported on the left side of figure 3. 3 0.8 24.1 5.30 22.8 1.36 3.94
The installation parameters of the test piles during the
penetration (rate of revolution n, rate of penetration VP and The transfer curves of the shear resistance along the pile
torque MT) and during the extraction of the screw (concrete shaft and of the pressure at the pile base are reported in figure
flow QC and retrieval rate VR) are also reported in figure 3. 5; the curves labelled “uncorrected” have been obtained refer-
Along most of the upper part of the pile shaft, crossing the al- ring to the nominal diameter dN of the piles, those labelled
luvial soils (from the ground surface to a depth of about 20 “corrected” refer to the actual diameter d = 1.13⋅(QC / VR)0.5
m), the condition VP � VPcrit is satisfied for pile n° 2 and n° 3 obtained by the installation data.
180
Made available with permission from the Publisher: Millpress Science Publishers, Rotterdam, The Netherlands
Being the subsoil rather uniform, the differences in behav- The unit shaft resistance qS and unit base resistance qB in
iour among the three piles are to be ascribed to differences in granular soils can be related to the values of the cone penetra-
the installation details. The low unit shaft resistance of pile n° tion resistance qC by the following expressions:
1 is related to a penetration rate slower than the critical value
(along the shaft VP / VPcrit averages 0.66 < 1, table 3) deter- q S = α S ⋅ q c ,S (4)
mining an overall net decompression effect on the surround-
ing soil. On the contrary, during the installation of piles 2 and qB = α B ⋅ qc ,B (5)
3 the rate of penetration was on average larger than before
(VP / VPcrit = 0.96 to 1.05, table 3), with a slight compression where: αS , αB are empirical coefficients; qC,S is the average
effect on the surrounding soil giving rise to larger unit shaft value of qC along the pile shaft down to a depth z = L – 4⋅d;
resistances. The transfer curves of the base resistance for all qC,B is the average value of qC between the depths (L – 4⋅d)
the piles, once corrected for the actual base diameter, are and (L + d). The values of αS and αB are listed in table 3 and
practically coincident being equal the conditions of penetra- plotted in figure 6 against the corresponding ratios between
tion of the auger. It may be noted that, in the absence of the actual penetration rate and the critical one.
monitoring of the installation parameters and hence without a In table 3 the values of the ratio VP / VPcrit averaged re-
correction of the diameter, the higher base pressure for pile n° spectively along the pile shaft down to a depth z = L – 4⋅d
3 would have been probably interpreted as due to random soil and between the depths (L – 4⋅d) and (L + d) are also re-
variability. ported.
5 5 5
10 10 10
depth, z [m]
15 15 15
20 20 20
25 25 25
30 30 30
Figure 4. Load tests results: load-settlement curve (above) and axial load distribution along total pile length (below).
181
Made available with permission from the Publisher: Millpress Science Publishers, Rotterdam, The Netherlands
0 0,03
0 20 40 60 80
S
settlement, w [mm]
0,02
pile n° 1
3,0 pile n° 3
0,01
base pressure, p [MPa]
1,0
_____ corrected 0,30
-------- uncorrected
0,0
0 20 40 60 80 0,20
settlement, w [mm]
B
Figure 5. Corrected and uncorrected load transfer curves. αB = 0,115 (V P / V Pcrit) + 0,153
0,10
It is widely accepted that a static load test to failure is the Test with kentledge
most reliable design method of a pile. As a matter of fact, In the case of a test with kentledge Poulos (2000a) claims that
most of the present insight into the behaviour of piles and the the stress arising in the subsoil from the weight of the
most significant advances in analysis and design have been kentledge tends to cause an increase of the shaft friction and
obtained by collecting and interpreting load tests data. end bearing pressure of the pile. As the load on the pile is in-
Until a few years ago the aim of a load test was essentially creased by jacking against the kentledge, the stress will re-
the determination of the bearing capacity, to be employed in a duce and some upward displacements tend to develop in the
capacity based design. Recently the attention is being soil, while the pile undergoes settlement. The pile head stiff-
182
Made available with permission from the Publisher: Millpress Science Publishers, Rotterdam, The Netherlands
ness is thus overestimated, while the pile ultimate capacity approached only when the two resistances have nearly the
may be relatively close to that of the ideal test. same value. On the other hand in the OCT a reaction system
Figure 8 reports the results of a parametric study on the is not needed; for this reason it may be a cheap alternative to
load settlement curves of a pile subjected to an Ideal Load tests with kentledge or reaction piles and anchors.
Test and to a test with kentledge. The curves have been ob-
Q [kN]
tained by nonlinear finite element elasto-plastic analyses as-
0 500 1000 1500 2000 2500
suming the soil to be a uniform sand with different values of
the friction angle �’ and test piles with d = 1 m and L/d = 10, 0
20 and 50. The test with kentledge overestimates the initial
stiffness of the pile, the more the higher the ratio L/d. On the
contrary, at relatively large displacements (w = 10%d) the 20
discrepancies decrease and eventually the value of the ulti-
ILT
mate capacity is practically unaffected by the influence of the
kentledge. Similar trends have been found for undrained 40
L/d = 10
w [mm]
clays.
load kentledge 60
fixed
point
jack support
level
test test 80
pile pile
100
= 23°
= 27°
= 31°
= 35°
a) ideal test b) test with kentledge
spreader beam
20
c) reaction piles anchor
ground anchors
ILT
40
w [mm]
test L/d = 20
tell pile
tales 60
Osterberg
load cell
d) Osterberg 80
c)Osterberg cell
=31°
=35°
=27°
Osterberg Cell
The Osterberg Cell Test (OCT, figure 7d) has been developed
80
commercially by Osterberg (1984). A special cell hosting one
or more hydraulic jacks is cast at or near the pile base; by ap-
plying pressure, the base is pushed downward while the shaft
is jacked upwards and provides the reaction. The test goes on 100
=31°
=23°
=27°
=35°
until either the base or the shaft reach the ultimate resistance.
This is a disadvantage of OCT, since only a lower bound of
the total bearing capacity may be determined; the full value is Figure 8. Load test with kentledge vs. Ideal Load Test.
183
Made available with permission from the Publisher: Millpress Science Publishers, Rotterdam, The Netherlands
L/dCOLT
= 20tp tionship of the base and the shaft can be considered sepa-
rately.
60 A parametric FEM analysis has been carried out on this
s/d = 4
topic (Recinto, 2004). The subsoil was assumed as a purely
frictional or a purely cohesive, elastic perfectly plastic mate-
80 rial. A comparison equivalent to that of figure 8 is reported in
figure 10. A substantial overestimation of the pile head stiff-
ness is again evident at low displacements, while a better
100 agreement occurs in the late stage of the test. The shortcom-
= 23°
= 27°
= 31°
= 35°
ing related to OCT is evident for piles with L/d = 50: the end
bearing capacity is many times larger than the shaft capacity,
preventing the OCT to explore the behaviour of the pile fur-
Q [kN] ther than a settlement w = 1.5%d.
0 2000 4000 6000 8000 Figure 11 summarizes the results of comparisons between
0 the ILT and other test setups in term of the ratio k = (K -
KILT)/KILT for frictional and cohesive material.
Summing up, the test setups that have been examined are
20 suitable for the determination of the bearing capacity; only
the OCT may have significant limitation in this respect. As
far as the load-settlement behaviour, and especially the initial
40 ILT
ILT
stiffness, is the main purpose of the test, substantial correc-
w [mm]
80 2.2 Analysis
= 27°
= 31°
= 35°
= 27°
= 31°
= 35°
piles (Jardine & Chow, 1996); centrifuge tests (de Nicola &
Randolph, 1993; Fioravante, 2002; Colombi, 2005). This
work has produced a better insight of the mechanisms of de-
velopment of side friction and base resistance; from a practi-
Figure 9. Load test with tension piles vs. Ideal Load Test. cal viewpoint, however, methods based on SPT (Meyerhof,
1956; Poulos, 1989; Decourt, 1995) and CPT (Poulos, 1989;
Osterberg (1995) and Schmertmann & Hayes (1997) give MELT, 1993; De Cock et al., 1999) provide simple and ade-
suggestions to derive from the results of OCT a load- quate estimates of the bearing capacity.
184
Made available with permission from the Publisher: Millpress Science Publishers, Rotterdam, The Netherlands
Detailed scrutiny of most recent results on the topic of 200 frictional material
bearing capacity of a single pile have been provided, among
others, by Poulos et al. (2001) and Jamiolkowski (2004). OCT
150
Q [kN] tension
k (%)
100 piles
0 500 1000 1500 2000 2500
0
50
kentledge
20 0
0 10 20 30 40 50
ILT L/d
40
w [mm]
L/d = 10
200 cohesive material
OCT
60
150
tension
k (%)
80 100 piles
50
100
= 35°
= 27°
= 31°
= 23°
kentledge
0
0 10 20 30 40 50
Q [kN] L/d
0 2000 4000 6000
Figure 11. Ratio between the initial stiffness as deduced by different
0 load test setup and by an Ideal Load Test.
= 27°
= 35°
Figure 10. Osterberg Cell Test vs. Ideal Load Test. Most foundations are subjected to some horizontal loads
185
Made available with permission from the Publisher: Millpress Science Publishers, Rotterdam, The Netherlands
which are generally smaller than the vertical ones, earth re- 3 PILE GROUPS AND PILED RAFTS, EXPERIMENTAL
taining structures being one of the exceptions. In the case of EVIDENCE
significant horizontal loads, raking piles have been also in-
stalled, providing horizontal resistance by means of the hori-
zontal component of the axial capacity. Nevertheless also ver- 3.1 Monitoring of full scale structures
tical piles can support horizontal loading; the present report
will be limited to this topic. In the early 1970’s several buildings supported on piled foun-
Starting from the simple and effective idea of considering dations were monitored in UK (Hooper, 1979). In the 90’s the
the pile as an elastic beam restrained by springs (Matlock & investigations carried out during the construction of several
Reese, 1960), widely known as p-y method, many tools for tall buildings in Germany, mainly in the Frankfurt area
the analysis have been subsequently developed. The p-y (Katzenbach et al., 2000), provided new stimulating data.
method is still widespread in practice; its main advantage is These and similar observations led to a deeper insight into the
the ability to easily incorporate variations of the soil stiffness mechanisms which govern the behaviour of piled foundations.
with depth. Non linear p-y curves were later introduced by The behaviour of full scale structures contrasts with the
Matlock (1970) and Reese et al. (1975). The p-y curves, ei- simplicity of research oriented experiments, either at labora-
ther linear or non linear, must be deduced by experiments and tory scale or in the field. The history of construction, the
cannot be easily transferred to different situations; accord- complexity of subsoil conditions and the interaction between
ingly, many experiments have been carried out to define the superstructure and the foundation make the back analysis
them, mainly as a function of the soil type. Some questions of the observed behaviour far from straightforward. On the
are still open, however, on the influence of the geometry of other hand, such a complexity constitutes a richness, and
the pile and the installation technique (O’Neill & Dunnavant, some times unexpected phenomena have been detected and
1984; Reese & Van Impe, 2001; Huang et al., 2001); for a highlighted.
comprehensive coverage reference may be made to Reese & The case history of the main pier of the cable stayed bridge
Van Impe (2001). over the river Garigliano (Southern Italy) will be reported in
Duncan et al. (1994) produced a series of solutions with some detail as an example, adding some new results to those
non linear p-y curves and derived simple equations to predict already published elsewhere.
the load - displacement relationship at the ground line, the The subsoil conditions at the site, reported in figure 12
maximum bending moment along the shaft and its depth of (Mandolini & Viggiani, 1992a), are characterized by a deep,
occurrence. Poulos et al. (2001) present in detail the proce- rather compressible silty clay deposit. The foundation of the
dure, known as the characteristic load method. main pier, resting on driven tubular steel piles, is represented
On the other hand Poulos (1971a) proposed the application in figure 13. Load tests to failure on instrumented piles and
of the boundary element method to the analysis of a vertical proof load tests on production piles were carried out. The
pile under horizontal load, modelling the soil as an elastic foundation was monitored during the construction and after-
continuum. Evangelista & Viggiani (1976) pointed out the wards, measuring settlement, load sharing between piles and
importance of a proper discretisation on the accuracy of the raft and load distribution among the piles.
solutions. The construction of the bridge started in October 1991 and
A similar approach was proposed by Banerjee & Davies the latest set of data has been recorded in October 2004, thir-
(1978) with the pile embedded into a non homogeneous soil. teen years later. The settlement is measured by means of pre-
Non linearity was introduced by Davies & Budhu (1986) cision levelling; 35 out of the 144 piles were equipped with
and Budhu & Davies (1987) using a cut-off procedure to limit load cells at the top to measure the load transmitted by the cap
the maximum value of the interaction force between the pile to the pile; furthermore, 8 pressure cells were installed at the
and the surrounding soil. interface between the cap and the soil. The load cells and
Randolph (1981) obtained solutions by FEM and summa- pressure cells were constructed on site using three sensing
rized the results into analytical expressions of the deflections units for each of them; a total of 129 vibrating wire load sens-
and rotations of the pile head as well as the maximum bend- ing units were used. Further details on the instruments and the
ing moment along the pile shaft. Observing that the displace- installation technique are reported by Mandolini et al. (1992)
ment and the bending moment along the pile shaft are usually and Russo & Viggiani (1995).
confined to an upper portion of the pile, he defined a critical In figure 14 the load history and the measured average set-
length LC. If L > LC, it is the critical length instead of the true tlement are reported; differential settlement was negligible
length to govern the behaviour of the pile. The ratio LC/d de- due to the very stiff pile cap. The net load is the total applied
pends mainly on the relative pile-soil stiffness; typically LC/d load minus the buoyancy, as deduced by piezometer readings.
< 10. An interesting consequence is that only a limited upper An accurate evaluation of the total pile load can be obtained
portion of the soil profile must be adequately characterized. by the measurements on 35 piles, with only minor extrapola-
Poulos (1982) reviewed some suggestions for the determi- tions. The total raft load as measured via 8 pressure cells was
nation of the soil properties relevant to the prediction of the almost negligible at all stages. It is possible, however, that the
response of piles under horizontal load both in clays and in pressure cells did not work properly since their installation.
sands; a significant scatter can be revealed. Suggestion for the The first load increments were due to the casting of the raft
evaluation of the undrained shear modulus and its degradation (October to November 1991) and of the pier (March to July
are given by Poulos et al. (2001). 1992); with the construction of the bridge deck the applied
In the continuum based approaches non linear analyses load increased rapidly to its maximum value. In the early
usually requires a limiting value of the pile-soil interaction. stage, when the raft was concreted, apparently almost the en-
The results by Broms (1964a, 1964b) are still widely used; tire net load was measured on piles. About four months later,
some later assessments (Kulhawy & Chen, 1993) have con- under constant applied load, the measured load on piles had
firmed their validity. vanished. The weight of the raft was actually supported by the
The maximum bending moment along the pile shaft, rather soil, even if not measured by the pressure cells; the apparent
than the head displacement, is probably the critical design is- pile load was an effect of the hydration heat of the concrete
sue. Randolph (1981) and Duncan et al. (1994) developed on the vibrating wire load sensors (Russo, 1996). Since the
simple but reliable analytical procedures for its evaluation, start of the installation of the bridge deck, in February 1993,
and showed that the maximum bending moment is much less the increments of the applied load match almost exactly the
sensitive than the deflection to the exact values of the stiff- corresponding increments of the observed total load on piles.
ness parameters of the soil. Except the weight of the raft, almost the entire weight of the
186
Made available with permission from the Publisher: Millpress Science Publishers, Rotterdam, The Netherlands
bridge was transmitted to the piles, sometimes with a minor city approach in which no reliance is given to the load trans-
delay. At the end of construction (March 1995) the settlement mitted by the raft to the soil.
was about 42 mm; in the following ten years it has progres- In figure 15 a plan view of the foundation with the location
sively increased to reach 52 mm in October 2004. of the 35 instrumented piles is reported. The behaviour of the
At the time being, 13 years after their installation, 127 of various piles can be grouped into four distinct categories, cor-
the 129 vibrating wire load sensors are still properly working; responding to four zones underneath the pile cap. In table 4
the last set of measurements confirms that the dead weight of the average values of the pile load for each of the selected ar-
the bridge is still resting almost entirely on the piles. This eas are reported, as a ratio to the mean value of all the piles.
finding is to be related to the subsoil properties and to the de- The values reported refer to three different stages: end of con-
sign of the foundation, based on a conventional bearing capa- struction, three years later and ten years later.
90 m
Garigliano
0
Sand
depth (m)
20
30
Clayey silt n.c.
40
50
0 5 10 15 20 0 2 4 6 8 10 0 20 40 60 80
qc (MPa) OCR Go (MPa)
Sand and gravel
Figure 12. Subsoil profile at the location of the main pier of the Garigliano bridge.
SECTION A-A
At the end of construction the measurements show a sig-
nificant edge effect, as it was to be expected under a stiff cap,
and some load concentration below the pier. Three years later
the load distribution was undergoing significant variations:
the load on the peripheral piles was decreasing, while that on
BORED PILES the piles below the pier was slightly increasing. Ten years
d=0.8 m; L=12 m later this trend is still confirmed. To the writers’ knowledge,
such a phenomenon had not been observed before; the ob-
served trend of variation suggests that the main factor is creep
DRIVEN PILES
PLAN VIEW of the reinforced concrete raft.
Figure 16 reports the values of the load on some typical
B
behaviour.
Table 4: Garigliano; load distribution among the piles vs. the time
Corner Edge Internal Piles under
25
piles piles piles the pier
End of con- 1.30 1.00 0.80 0.90
struction
3 years later 1.16 0.96 0.90 0.98
B
SECTION B-B
10.6 m
10 years later 1.10 0.93 0.94 1.03
Figure 13. Layout of the foundation of the main pier of the bridge.
187
Made available with permission from the Publisher: Millpress Science Publishers, Rotterdam, The Netherlands
applied load
120 0.000
pile load
80 20.000
settlement [mm]
load [MN]
60 30.000
40 40.000
settlement
20 50.000
raft pressure cells
0 60.000
Oct-91
Oct-92
Oct-93
Oct-94
Oct-95
Oct-96
Oct-97
Oct-98
Oct-99
Oct-00
Oct-01
Oct-02
Oct-03
Oct-04
Figure 14. Total applied load compared to observed load sharing and measured settlement for the foundation of the main pier of the Garigliano Bridge.
900
800 corner
bridge dec k
J an-94
J an-95
J an-96
J an-97
70 64
3.2.1 Settlement
9 Mandolini et al. (1997) and Mandolini & Viggiani (1997) col-
lected 22 well documented case histories of the settlement of
piled foundations. The data base has been increased by Vig-
giani (1998) to 42 cases. The collection of further evidence
corner brings now the total number of cases examined to 63; for all
of them, besides the settlement records, load test on single
edge piles and documentation on the subsoil and the construction
under the pier are available. The main features of the case histories collected
internal
are listed in table 5. A wide range of pile types (driven, bored,
CFA) assembled in a variety of geometrical configurations (4
� n � 6500; 2 � s/d � 8; 13 � L/d � 126) and regarding very
Figure 15. Plan view of the foundation with the location of the in- different soils (clayey to sandy soils, stratified, saturated or
strumented piles. not, etc.) are included.
The available measured settlement may be used as the ba-
188
Made available with permission from the Publisher: Millpress Science Publishers, Rotterdam, The Netherlands
w = RS wS = n RG wS (6)
0,6 RG,max = RS,max (eq. 7) / n
where wS is the settlement of a single pile under the average
RG,max
working load Q/n of the group (Q = total load applied to the
foundation; n = number of piles), RS is an amplification factor 0,4
named group settlement ratio, originally introduced by
Skempton et al. (1953) and representing the effects of the in-
teraction between piles, and RG = RS/n is the group reduction 0,2
factor. The settlement of the single pile wS is obtained by load
tests on single pile. The group settlement ratio RS has been
expressed by Skempton et al. (1953), Meyerhof (1959), Vesic 0,0
(1969) as a function of geometrical factors as the number n, 0,1 1,0 10,0 100,0
the spacing s and the slenderness L/d of the piles. aspect ratio, R [-]
On this empirical basis the following expressions for the
upper limit RS,max and the best estimate of RS, as a function of
the aspect ratio R = (ns/L)0.5 introduced by Randolph & Figure 17. Relationship between RG,max and R.
Clancy (1993), have been found:
wmax 0 ,50 1 1,0
RS ,max = = ⋅ 1 + ⋅n (7)
wS R 3R
N = 63
0,8
w
RS = = 0 ,29 ⋅ n ⋅ R −1,35 (8)
wS
0,6 RG = RS (eq. 8) / n
Some of the case histories include information on the
RG
189
Made available with permission from the Publisher: Millpress Science Publishers, Rotterdam, The Netherlands
190
Made available with permission from the Publisher: Millpress Science Publishers, Rotterdam, The Netherlands
In figure 20 the ratio between the settlement measured at observation interval (14 years) have been rather long.
the end of construction, weoc, and the settlement measured at Poulos (1993) claims that there are no theoretical solutions
the end of the observation period, wfobs, is plotted versus the available for the rate of consolidation of pile groups. Numeri-
ratio between the length of the piles L, and the width of the cal analyses of an impermeable block equivalent to the pile
pile group B; the data reported for L/B = 0 refer to raft foun- group indicate that the consolidation rate decreases with in-
dations. creasing the length to diameter ratio of the equivalent block.
In evaluating these data, it is to remind that the settlement This result implies that the rate of consolidation of a shallow
at the end of construction probably includes some consolida- foundation is faster than that of a pile group. Available ex-
tion settlement, and conversely the settlement at the end of perimental evidence does not confirm this trend; on the con-
the observation period is probably smaller than the true final trary, the time to the final settlement seems independent of
settlement. In any case, moving from raft to piled foundations the type of foundation.
the settlement ratio increases; for the same subsoil, the higher Further data with accurate long term settlement observa-
the ratio L/B the higher the settlement ratio. The only excep- tions are needed to confirm the outlined trends.
tion to this trend is the case of Torhaus; for this case the ap-
parent anomaly could be explained by the very fast construc- 3.2.2 Load sharing and distribution
tion (figure 21), compared to the other case histories.
A structure, its foundation and the surrounding ground inter-
1
act with each other whether or not the designers allow for this
5 interaction (Burland, 2004). The load sharing between the
0.8 9 4
8 piles as a group and the raft is a fundamental quantity in the
w eoc/w fobs
0.6 6 2
3 advanced design methods and in the new codes about piled
1 7 raft foundations, in order to make the right use of the coopera-
0.4 tion of the two elements. The load distribution among piles is
a more complex issue, being markedly affected by the natural
0.2 London Clay Frankfurt clay soil heterogeneity and the unavoidable pile variability (Evan-
0 gelista et al., 1977). Unfortunately, the bending moment and
shear in the raft are strictly depending upon such distribution
0 0.2 0.4 0.6 0.8 1
(Poulos et al., 1997).
L/B
The experimental evidence on soil-structure interaction, ei-
Figure 20. Ratio between settlement at the end of construction and ther by small scale tests or monitoring of full scale structures,
settlement at the end of the observation vs. L/B. is much less than that available for settlement. Some data on
load sharing between raft and piles and load distribution
among piles, however, are gradually accumulating. In contrast
100 to the 63 well documented case histories available on settle-
ment (table 5), after a careful review of the literature only 22
80
sufficiently well documented case histories of soil-structure
Time [months]
191
Made available with permission from the Publisher: Millpress Science Publishers, Rotterdam, The Netherlands
About the load sharing between the raft and the group of s/d
piles, the data reported in figure 23 come from only 11 out of Figure 24. Load shared by the raft vs. spacing
22 cases of table 7, and refer to foundations with piles more
or less uniformly spread underneath the whole area of the raft 100
(Ag/A > 0.83, where A is the area of the raft and Ag is the area
80
of the pile group). The simple geometrical parameter s/d
raft load [%]
plays a major role in load sharing; the higher the spacing the 60
higher the load taken by the raft.
40
100
20
80
0
raft load [%]
60 3 6 9 12 15
40 (s/d) / (A g /A)
20 Figure 25. Load shared by the raft vs. spacing divided by the area ra-
tio Ag/A.
0
0 3 6 9 12 3.2.3 Bearing capacity
s/d A piled raft foundation consists of three elements: the raft, the
piles and the subsoil. The load is equilibrated partly by the
Figure 23. Load shared by the raft vs. spacing. contact pressure between the raft and the soil and partly by
the piles.
In figure 24 the plot is extended to all the 22 cases re- At failure, the bearing capacity of an unpiled raft QR may
ported in table 7; the resulting relationship between the load be evaluated by the conventional bearing capacity theory
sharing and s/d is not as close as it was in figure 23. (Terzaghi, 1943; Brinch-Hansen, 1970; Vesic, 1973;
The added cases are generally characterized by piles con- Randolph et al., 2003). Collapse of the pile group may occur
192
Made available with permission from the Publisher: Millpress Science Publishers, Rotterdam, The Netherlands
(Q / cud2) x 10-3
i =1 QBF
where Qi,P is the bearing capacity of the i-th pile and η is a 8 7x7
group efficiency factor depending on pile layout and type and
soil type (Kezdi, 1957). Values for the efficiency η have been
suggested by Whitaker (1957), Vesic (1969), De Mello 4 5x5
(1969), Brand et al. (1972), O’Neill (1982), Briaud et al.
(1989).
When considering the bearing capacity QBF by failure of 3x3
the overall block of soil, it is generally assumed that the full 0
shear strength of the soil is mobilised on the vertical surfaces 1,5 2,0 2,5 3,0 3,5 4,0
of the block defined by the perimeter of the piles, as well as s/d
the bearing pressure at the base of the block. A suitable factor
of safety FS should be provided against both modes of failure, Figure 26. Experiments by Cooke (1986) on pile groups and piled
taking into account that the settlement needed to mobilize the rafts with L/d = 48.
base capacity of the block is of the order of 5% to 10% of its
width (Cooke, 1986). Since the end-bearing pressure qB in Cooke (1986) summarized the results of a broad laboratory
granular soils is much greater than the average skin friction q S investigation including load tests on model rafts, pile groups
(typically qB/qS ranges between 50 and 200), Fleming et al. and piled rafts, founded on remoulded London clay with an
(1992) claim that block failure may occur only when the base undrained shear strength cu ranging between 5 and 15 kPa.
area is many times smaller than the side area. Groups consist- The tested piles had a ratio L/d = 24 and 48 and were ar-
ing of closely spaced long piles are thus more likely to fail as ranged in 32, 52, 72 and 92 groups. Some results are shown in
a block than groups consisting of short piles at the same spac- figure 26 for the case L/d = 48.
ing. Such conclusion is consistent with the experimental data The experimental data for pile groups are in good agree-
collected by a number of researchers (e.g. Vesic, 1969; Liu et ment with the theoretical curves; the bearing capacity of the
al., 1985; Ekstrom, 1989; Phung, 1993). pile group QP is equal to QBF for s < scrit, while QP = n⋅QS for
De Mello (1969) summarized data for pile groups up to 92 s > scrit. On the other hand the values of the bearing capacity
in clay soils; the block mode of failure occurs for spacing of piled rafts QPR, at a measured settlement w = 10%B, fit the
smaller than 2 to 3 pile diameters. Similar results have been curves corresponding to block failure, for values of s/d both
reported by Cooke (1986). below and above scrit/d. By introducing the coefficient:
Taking into account all the above evidence Poulos (2000b)
suggested to estimate the vertical bearing capacity QPR of QPR
ζ PR = (13)
piled rafts as the smaller of the following values: QP
• the ultimate capacity QBF of the block containing the
piles, plus that of the portion of the raft outside the pe- the experimental results can be summarized as follows:
riphery of the pile group;
• the sum of the ultimate loads of the raft QR and of all the • for s/d < scrit/d, QPR ∼ QBF ∼ QP ⇒ ζPR ∼ 1
piles QP in the system: • for s/d > scrit/d, QPR ∼ QBF > QP ⇒ ζPR > 1�
QPR = QR + QP (11) The critical spacing ratio scrit/d is generally defined as that
value below which block failure occurs for pile groups. It
the latter having been proposed by Liu et al. (1985).
may be actually better viewed as that value above which the
The installation of the piles, however, may affect the soil
raft either transfers part of the load directly to the soil or en-
properties and consequently modify the performance of the
forces block failure for the pile group. Therefore, ζPR may be
raft in comparison with that of the unpiled raft. Moreover, it
assumed as a measure of the increase of bearing capacity due
is becoming more and more evident that the behaviour of the
to raft-soil contact.
piles belonging to a piled raft is affected not only by the inter-
Figure 27 shows that ζPR increases with increasing spacing
action among piles but also by the surcharge exerted by the
ratios and with decreasing number of piles; at s/d = 4, it is
raft. As a consequence Liu et al. (1994) and Borel (2001a)
about 1.7 for 32 and 1.25 for 92 piles. For the case L/d = 24 at
suggested to modify eq. (11) as follows:
s/d = 4, Cooke’s experiments give ζPR ∼ 2.1 for 32 and ζPR ∼
QPR = α R ⋅ QR + α P ⋅ QP (12) 1.9 for 92 piles.
All the experimental values of ζPR obtained by Cooke
where αR and αP are coefficients affecting the failure load of (1986) are reported in figure 28 as a function of the ratio
the raft and the pile group when combined in a piled raft. s/scrit, in the range s/scrit > 1. The value of ζPR increases with
The available experimental evidence is reviewed in the increasing s/scrit; with a conservative design approach, it could
following, to assess the likely values of the coefficients αR be assumed in clay soils that ζPR = s/scrit (broken line in figure
and αP and elucidate the factors affecting them. 28).
The curves reported in figure 26 represent the theoretical Sales (2000) reports a field investigation on the behaviour
failure load of a pile group, assuming either “block” failure or of piled foundations in the clay of Brasilia. Only the four load
“pile group” failure, as a function of the pile spacing. Below a tests carried out on undisturbed soil (table 8) are considered
critical value scrit/d of the spacing ratio (scrit/d increases from here. The settlements attained during the tests range between
about 2.5 for 32 piles to about 3.5 for 92 piles, bold line in fig- ∼ 20 mm for test II and ∼ 45 mm for test IV. In any case, the
ure 26) the failure load QBF corresponding to block failure is piled raft (test VI) attained a settlement not larger than 3%B,
the smallest one and block failure should thus occur; at s/d > significantly smaller than those reported by Cooke (1986).�
scrit/d, the failure load of pile group QP should apply. Comparing the results of single pile (test II) and 22 pile
193
Made available with permission from the Publisher: Millpress Science Publishers, Rotterdam, The Netherlands
group (test IV), an efficiency factor η=1 is derived; moreover, an extensive series of centrifuge tests in fine grained soils on
the sum of the ultimate loads for pile group (test IV) and raft single piles (at prototype scale, L = 9 and 18 m; d = 0,63 m),
(test I) is 450 kN, that means 12.5% greater than the observed unpiled square rafts (B = 9 and 18 m), pile groups (32 and 72
value for the piled raft (test VI, 400 kN). In terms of coeffi- piles, s/d = 4) and piled rafts, obtained by a number of combi-
cient ζPR the experiments yield a value of 400/300 = 1.33. nations of the above components. Five series of tests were
carried out to measure the response of the components up to
� 2,0 very large settlement (for single piles and pile groups, at least
80%d; for unpiled and piled rafts, not less than 10%B). Apart
a few cases, a punching type failure has been systematically
1,8 observed in the tests, i.e. the load continuously increases as
the settlement increases. The analysis of the experimental re-
3x3 sults is thus strongly affected by the displacement level as-
sumed to represent failure.
= QPR / QP
1,6
The values of the coefficient ζPR obtained for two series of
5x5 tests at final settlement are reported in figure 29 as a function
1,4 of the parameter RM defined as:
PR
7x7
n⋅s A
9x9 RM = ⋅ = R (14)
1,2 L Ag Ag
A
1,0 For piled rafts with piles spread below the whole raft,
1,5 2,0 2,5 3,0 3,5 4,0 Ag/A is close to unity and RM approaches the aspect ratio R as
defined by Randolph & Clancy (1993); on the contrary, for
spacing / diameter, s/d [-]
piles concentrated in some region of the raft (for instance in
Figure 27. Relationship between the increase of bearing capacity and the central zone), A > Ag and, consequently, RM > R.
the increase of spacing ratio. For piled rafts with A/Ag ∼ 1, ζPR at large settlement at-
tains values from ∼2 to ∼4. For A/Ag ∼ 4, ζPR significantly in-
2,5 creases, attaining a value of about 10. All these values of ζPR
correspond to values of αP and αR close to 1.�
Similar results have been obtained by centrifuge tests car-
ried out on model piled rafts in granular soils (LCPC, 1998).
= QPR / QP
2,0
12 Box A
PR
1,5 Box B
10
8
1,0
PR
At maximum settlement, the piles carried about 70% of the Figure 29. Relationship between ζPR and RM.
total load (280 kN), the raft the remaining 30% (120 kN). As-
suming these values as the final ones, eq. (12) gives: αR = Summing up the available experimental evidence, the as-
0.80 and αP = 0.93.� sumption that the failure load of a piled raft with pile spacing
Borel (2001a) reports a full scale load test on a capped pile above some critical value is equal to that of the pile group
in stiff clay. The cap is a concrete circular raft, 2 m diameter alone (αR = 0 and αP = 1 in eq. 12) appears overly conserva-
and 0.5 m thick; the closed-end steel displacement pile has d tive. The contribution of the raft to the ultimate capacity is
= 0.45 m, L = 12.2 m. Under the maximum applied load of always positive (ζPR > 1); a fraction αR of the ultimate value
2.25 MN a settlement of about 225 mm, i.e. ∼ 11% of the cap for the unpiled raft QR can be definitely considered. For set-
diameter, was attained. At this stage, QP/QPR = 59%; QR/QPR tlement of the order of some percent (say 5% to 10%) of B,
= 41%. By comparing the load carried by the pile (∼ 1330 the collected experimental evidence yields values for αR ap-
kN) with its failure load when isolated (QS ∼ 1200 kN), a proaching unity; being such a settlement too large for practi-
value αP ∼ 1.1 is found. In the same way, for the circular cap cal purposes, even for small rafts, later on (§ 4.1.5) the results
(QR = 1000 kN) a value αR ∼ 1.1 is derived. In terms of coef- of some numerical analyses will be reported allowing to ex-
ficient ζPR the experiment yields a value of 2250/1200 ∼ 1.9. plore which values of αR and ζPR may be expected at smaller
Conte (2003) and Conte et al. (2003) report the results of settlement.
194
Made available with permission from the Publisher: Millpress Science Publishers, Rotterdam, The Netherlands
Much alike piles under vertical loads, the response of a later- single pile
ally loaded pile group with relatively closely spaced piles is
quite different from that of a single pile, because of the inter-
action between piles through the surrounding soil, the rota- pile in the
tional restraint exerted by the cap connecting the piles at the
p
group
head, the additional resistance to lateral load provided by fric- a
tional resistance at the cap-soil interface, and passive resis- b
tance if the structure is totally or partially embedded. fmi=b/a
The experimental evidence is rather scanty, compared to
that available for vertical load, and is mainly related to the
first two items. Relatively small groups have been tested, y
typically small scale models and full scale foundations with 2
to 16 piles. In recent years the use of centrifuge allowed the Figure 31. P-y multipliers for group effects.
study of slightly larger groups (16 to 21 piles).
A comprehensive review of the experimental evidence is The experiments usually attain rather large displacements,
reported by Mokwa (1999). Valuable experimental investiga- being oriented to extreme events. The p-y multipliers are ob-
tions have been recently added (Remaud et al.,1998; Borel, tained at the ultimate displacement reached during the test,
2001a, 2001b; Rollins & Sparks, 2002; Ilyas et al., 2004; and generally they are considered independent of the dis-
Rollins et al., 2005). placement or the load level. Field test displacements of the
Prakash & Saran (1967), Alizadeh & Davisson (1970), single pile and the pile group ranging between 10% and 15%
Matlock et al. (1980), Schmidt (1981, 1985) conducted a of the pile diameter are rather usual. In centrifuge tests dis-
number of full scale and 1g model tests. In most cases only placements as high as 25% to 50% of the pile diameter are
the load displacement relationship for single free head piles typically attained.
and for both free and fixed head pile groups was recorded; the The collected observations (Brown et al., 1988; McVay et
experimental findings appeared generally compatible with the al., 1998; Rollins et al., 1998; Mokwa, 1999) seemed to sug-
framework of the analytical tools available at that time. gest that:
Later centrifuge (Barton, 1984) and full scale tests (Ochoa • the multipliers can be defined for rows orthogonal to the
& O’Neill, 1989) revealed that for a given horizontal load direction of the load vector, being the differences among
parallel to the columns of a group, calling trailing row the the piles in a row almost negligible;
first row of the group while the last one is the leading row • after the third leading row the same multiplier applies to
(figure 30), the leading piles carry more load than the trailing the other rows, except the trailing one;
ones even at load levels far from failure. • the multipliers are independent of the soil type, pile type
Selby & Poulos (1984) measured shears and bending mo- and load level, but depend essentially on the spacing;
ments in the leading piles larger than that in the central and • at a spacing above 6 to 8 diameters in the direction of the
trailing piles in 1g model tests; they called this effect “shield- load vector, and 4 diameters in the orthogonal direction,
ing”. Brown et al. (1988) observed the same effect in a full the interaction among piles is negligible and the multipli-
scale test on a 32 pile group and introduced the term “shadow- ers can be assumed equal to 1.
ing” to mean the phenomenon for which the soil resistance of A parameter frequently used to compare the response of sin-
a pile in a trailing row is reduced because of the presence of gle pile and pile groups under horizontal load is the group ef-
the leading pile ahead of it. ficiency:
A rather large amount of experiments have been carried
H
out in the last decade with the aim of deriving “general” rules
to adapt p-y curves to account for group effects. Brown et al. Ge = n (15)
(1988) introduced the concept of p-y multiplier, fm, a multi- HS
plier of the p values capable of stretching the p-y curve for the
single pile to account for the interaction among the piles in a where H is the total horizontal load applied to the group, n the
group (figure 31). The multipliers have obviously values in number of piles in the group and HS the horizontal load car-
the range 0 to 1. A major part of the latest experimental work ried by a single pile at the same horizontal displacement. It is
has been devoted to the determination of the p-y multipliers. worth noting that, when dealing with piles under vertical load,
The latest experiments are at field scale or in centrifuge; the the group effect is expressed through a multiplier RS of the
use of 1g small scale tests has been abandoned having recog- settlement at a given load; on the contrary for piles under
nized how misleading could be the obtained results in terms horizontal load, a reduction of the load per pile at a given dis-
of stiffness. placement is used. This reflects the fact that in the former
case the emphasis is on displacements, in the latter the main
design issue is the stress in piles.
leading trailing
The efficiency Ge can be easily expressed in terms of the
piles piles
p-y multipliers as follows:
m
∑ f mi
Ge = i =1 (16)
load vector m
where fmi is the multiplier of the i-th row while m is the num-
ber of rows in the group. Assuming constant values for the p-
y multipliers, irrespective of the load or the displacement
level, the efficiency of the group Ge is also constant.
In figures 32 and 33 some experimental values of the effi-
ciency Ge are plotted against the displacement normalised by
Figure 30. Leading and trailing piles for a given load vector. the diameter of the pile. The data reported in figure 32 were
obtained by field tests while those in figure 33 by centrifuge
195
Made available with permission from the Publisher: Millpress Science Publishers, Rotterdam, The Netherlands
tests. In both cases the experiments were carried out under The efficiency Ge is always below unity and decreases
free head conditions for both the single piles and the pile with increasing displacement (figures 32 and 33). In figure 34
groups. the efficiency is above unity, as expected, due to the rota-
The data reported in figure 34, on the contrary, were also tional restraint at the head of the piles in group, but again Ge
obtained by centrifuge tests but the single pile was tested un- significantly decreases with increasing displacement. Figures
der free-head conditions while the pile group had a rotational 33 and 34 show a dependence of the efficiency Ge on the size
restraint at the pile head. All the data reported in figure 32, 33 of the group: the larger the group size, the lower the effi-
and 34 refer to pile groups with a constant spacing s = 3d. ciency.
Being a widespread belief that the p-y multipliers can be
Brow n et al. 1987 assumed constant for each row and independent of the num-
100 ber of piles contained in the rows, the tests reported in figure
Brow n et al. 1988
95 34 were carried out just increasing the number of rows and
Rollins et al. 1998 keeping constant the number of piles in each row.
90 To compare the behaviour of pile groups under horizontal
85 Rollins et al. 2005 and vertical load, the definition of efficiency Ge can be easily
extended to pile groups under vertical load, just exchanging
80
the shear H with the axial head force Q on top of the piles. In
Ge [%]
100
100
1x2 groups
80 80
2x2 groups 60
60
Ge [%]
Ge [%]
40
40 3x3 & 4x4 groups
20
Ilyas et al. 2004 (nc & oc clays)
20
McVay et al. 1995 (loose & dense sand, 3x3 group)
0
Kotthaus & Jessberger 1994 (sand, 1x3 group)
0 0.015 0.03 0.045 0.06
0 w/d
0.00 0.10 0.20 0.30 0.40 0.50 0.60
y/d
Russo 1994 (144 piles)
Cooke et al. 1981 (351 piles)
Figure 33. Efficiency Ge vs displacement for pile groups of different Koizumi & Ito 1967 (3x3 field test)
size under horizontal load (centrifuge tests). O'Neill et al.1982 (2x2 field test)
O'Neill et al. 1982 (3x3 field test)
increasing
Figure 35. Efficiency Ge vs displacement for pile groups under verti-
group size
340 cal load.
McVay et al. 1998 (3x3 to 7x3 pile groups; sand:Dr = 55% )
300 McVay et al. 1998 (3x3 to 7x3 pile groups; sand:Dr = 36% ) On the contrary, under horizontal load the rather marked
decrease of the efficiency with the increase of the displace-
260 ment reveals a growing interactivity among the piles of the
group. The interaction mechanisms under vertical and hori-
Ge [%]
196
Made available with permission from the Publisher: Millpress Science Publishers, Rotterdam, The Netherlands
The field test of a 42 pile group in sand carried out by Ru- cap – soil friction is fully mobilised at a displacement lower
esta & Townsend (1997) revealed differences above 100% than that needed to mobilize the shear at the head of the piles.
between piles in the same row and in the same column. It is evident that further investigations on the effect of a cap in
McVay et al. (1998) performed centrifuge tests on groups of contact with the soil are badly needed.
variable size and found differences between two adjacent
piles in the same rows not always negligible and, sometimes, 200
comparable to the differences between two adjacent rows.
180
Similar results are reported by Ilyas et al. (2004).
Rollins et al. (2005) published the results of a field test on 160
a 32 group at spacing of 3.3d. Substantial differences in the
Mi/MSP [%]
load sharing among piles in the same row were observed. The 140
internal piles carried systematically the lowest load. The ratio 120
between the centre and the outer pile loads in the same row is
in the range 65% to 80%; the same range applies also for piles 100
belonging to different rows. Even if the lower interactivity 80
among piles placed orthogonal to the direction of the load
vector compared to that among piles aligned in the direction 60
of the vector is a widely accepted evidence, the above data 0.00 0.05 0.10 0.15 0.20 0.25
show that the effects in a group are not at all negligible. y/d
Being the bending moment in the piles more critical than Brown et al. 1987 - LR Brown et al. 1987 - MR
the pile head deflection when designing piled foundations un- Brown et al. 1987 - TR Brown et al. 1988 - LR
der horizontal load, it is also interesting to summarize the Brown et al. 1988 - MR Brown et al. 1988 - TR
available experimental evidence on this item. The experimen- Ruesta&Townsend 1997 - LR Ruesta&Townsend 1997 - MR
tal results are affected by some scatter, probably due to the Ruesta&Townsend 1997 - MR Ruesta&Townsend 1997 - TR
experimental difficulties and also to the detail of the rota- Rollins et al. 1998 - LR Rollins et al. 1998 - MR
tional restraint imposed at the pile head. Some general trends Rollins et al. 1998 - TR Rollins et al. 2005 - LR
can be however identified.
Data collected for the cases where both the single pile and Figure 36. Ratio between bending moments of piles in a group and
the pile group were tested with free head conditions are re- those in a single isolated pile vs. displacement.
ported in figure 36. The ratio between the maximum bending
moment in different piles in a group and that in a single iso-
lated pile is plotted against the displacement of the pile group. 4 PILE GROUPS AND PILED RAFTS; ANALYSIS
The ratio is evaluated at the same average load per pile. The
data are rather scattered but in the majority of the cases the
values of the moments in the piles belonging to the group are 4.1 Vertical loads
larger than the values in the single pile, the increase being a
growing function of the displacement. Larger moments occur
for the piles belonging to the Leading Row (LR) if compared 4.1.1 Model and reality
either to the middle (MR) or the trailing piles (TR). This plot
provides a valuable piece of information. It would be unsafe Modern engineering is characterised by a design performed in
to approach the prediction of the moment in a pile within a the framework of scientific theories; it is tightly linked to the
group calculating the bending moment with the average load methodological structure of science and could not come into
and the model of a single pile interacting with the soil. being without it. A scheme of the relationships between Sci-
This occurs for two reasons. The maximum bending mo- ence and Engineering is reported in figure 37 (Viggiani,
ment in a pile subjected to a horizontal load at the head de- 2001). A scientific theory, such as Euclidean geometry or
pends on: (i) the load and (ii) the depth needed for the pile to thermodynamics or theory of elasticity, is characterized by
develop a sufficient reaction into the surrounding soil; the two essential points:
deeper the soil reaction the higher the bending moment with • it does not deal with real objects, but with abstract entities
the same head load. In a pile group the applied load is not uni- specific to each theory: points, angles, segments; tempera-
formly shared among the piles, being the load distribution a ture and entropy; elastic half spaces;
function of their position within the group. The higher mo- • the structure of the theory is deductive. It consists of a
ments of the leading piles in figure 36 are thus partially due to small number of fundamental statements (axioms, or prin-
head loads higher than average. The interaction among the ciples, or postulates) involving the above entities, and a
piles, furthermore, develops a deeper reaction in the surround- universally accepted method to derive from them an end-
ing soil, compared to what observed for a single isolated pile. less number of consequences. All the problems that can be
This trend is more marked for the trailing piles than for the formulated within the framework of a theory can thus be
leading ones. This may well explain why in some cases dif- solved by demonstration and calculus, and there is a gen-
ferences in the observed maximum bending moments be- eral agreement on the solution among the scientists. In this
tween leading and trailing piles are not very large, even being sense, the truth of the scientific statements is warranted.
the head load on the leading piles much higher. However, the application of a theory in engineering depends
Kim & Sing (1979) tested both a free standing pile group on correspondence relationships between the abstract entities
and pile group with the cap in contact with the soil. The of the theory and real objects. Unlike the statements which
maximum bending moments observed in the piles were ini- are internal to the theory, these relationships have no absolute
tially similar in the two experiments, probably due to a negli- validity; they have to be checked by experiments and in any
gible mobilisation of the friction between the cap and the soil. case their validity is always limited.
At higher load level, however, the observed moments in the Moving now from the heavens of science to the ground of
latter experiment were less than half those of the free standing foundation engineering, a similar scheme can be applied to
group. Such an effect was likely due to the contribution of the the analysis and design of piled rafts. In recent years, a con-
friction between the cap and the soil, resulting in a decrease siderable research effort has been devoted to the procedures
of the loads transmitted to the pile head. On the other hand of analysis for the evaluation of the settlement of piled foun-
Horikoshi et al. (2002), in a centrifuge test, observed that the dations and the study of soil-structure interaction under verti-
197
Made available with permission from the Publisher: Millpress Science Publishers, Rotterdam, The Netherlands
cal loads. Generally soil, raft and piles have been modelled as involving the traditional subsoil investigation as well as the
elastic bodies, and their interaction analysed by numerical results of load test on single pile. The procedure is described
method, the most widely used being BEM (Poulos, 1968; in detail by Mandolini & Viggiani (1997) and Viggiani
Banerjee, 1970; Poulos & Davis, 1980; Banerjee & (1998), and is summarised in the following. The results of all
Butterfield, 1981; Basile, 1999). Different approximations the available site and laboratory investigations are first used
have been introduced to curb the computational resources to develop a model of the subsoil, in which the geometry is
needed or to deal with non linearity, such as the interaction adapted to a scheme of horizontal layering. The relative stiff-
factors (Poulos, 1968) and so called hybrid methods (Chow, ness of the layers is also evaluated, such an evaluation being
1987). FEM are increasingly used with suitable constitutive relatively easy on the basis of the results of laboratory tests,
models for the soil and the interfaces (Reul, 2000; Katzen- or site tests as CPT, SPT, DMT. The absolute values of the
bach et al., 1997; de Sanctis, 2001; Potts & Zdravkovi�, stiffness of the different layers are then fixed by fitting the
2001), and it can be foreseen that, with increasing computa- load settlement curve of the single pile (preferably obtained
tional resources, the use of FEM will further spread out. As a by a load test, or simulated with a suitable procedure) to the
result of this effort, entirely belonging to the realm of theory, results of an elastic analysis of the single pile based on the
quite a number of algorithms are by now available making the previously developed subsoil model. Once the subsoil model
analysis relatively simple and straightforward. is fixed and the stiffness of each layer is established, the same
model is used for the analysis of the piled foundation.
It is necessary to choose whether implementing a linear or
Correspondence relations non linear analysis. In the former case, the soil stiffness may
(modelling) be determined fitting the results of the elastic analysis either
to the initial tangent of the load–settlement curve of the single
REAL WORLD SCIENTIFIC THEORY pile (Linearly Elastic or LE analysis), or to a secant corre-
Observation and feedback
ABSTRACT ENTITIES sponding to the mean service load of the piles (Elastic analy-
ENGINEERING sis based on Secant modulus, or ES). This latter choice, ap-
Axioms, principles, parently the most reasonable one, is at present the most
Analysis
hypoteses widespread.
Design
construction
If non linearity is believed to be significant, then a step-
Theorems
manteinance wise linear incremental analysis (NL analysis) is performed,
of structures, Demonstration, calculus updating the stiffness matrix at each load step.
machineries and The three different procedures lead obviously to different
Exact and repeatable
services results, and in some cases the difference is significant. A
results
comparison with the experimental evidence clarifies the
Correspondence relations meaning of the different analyses and helps selecting the most
(design) suited one.
To this aim the code NAPRA (Russo, 1996, 1998a; Man-
dolini & Russo, 2005) has been employed in the back analysis
Figure 37. Relationships between Science and Engineering. of a number of case histories. Following Caputo & Viggiani
(1984) the overall behaviour of each pile is modelled in
Less attention has been paid to the development and vali- NAPRA by a non linear relation (e.g., a hyperbolic relation
dation of suitable correspondence relations. In the more fa- between load and settlement), while the interactions among
miliar terms of foundation engineering, this means: the sub- the pile and other elements are still assumed to be linear. In
soil model, the determination of parameters, the choice other words, the non linearity is concentrated at the pile–soil
between a linear elastic (either tangent or secant) or a non lin- interface.
ear analysis. Many Authors (among others: Poulos, 1972; A substantially similar approach has been suggested by
Caputo & Viggiani, 1984; Randolph, 1994; El Mossallamy & Randolph (1994) and El Mossallamy & Franke (1997). It may
Franke, 1997; Mandolini & Viggiani, 1992a, 1997; Viggiani, be shown that, in terms of settlement, this procedure is
2000) have addressed these topics, but their relevance seems equivalent to adding the non-linear component of the settle-
to be not yet widely appreciated. In facts, some views about ment of the single pile to the settlement of the group, obtained
the reliability of the analysis and the need for further devel- as in the LE analysis. In any case, the main conclusions that
opment of the procedures (Goossens & Van Impe, 1991; Pou- will be presented are essentially independent of the particular
los, 1993; Tejchman et al., 2001) can probably be corrected code employed, and focused on the correspondence relation-
by a proper consideration of these factors. ships.
In the opinion of the Authors, the available procedures of
analysis may be considered satisfactory for engineering pur- 4.1.3 Settlement prediction
poses provided they are properly applied, paying due atten-
tion to the correspondence relations. A comparison between the observed average settlement of 48
out of the 63 case histories listed in table 5 and the predictions
4.1.2 Evaluation of soil properties and implementation of obtained by NAPRA is reported in figure 38.
the analysis The majority of the analysed foundations had been de-
signed according to a conventional capacity based approach.
The elastic properties of the soil, to be used in the analysis, As a consequence, their safety factor under the working load
are difficult to evaluate because of the marked non linearity of is rather high, and a simple linear analysis may be expected to
the stress–strain relation and the influence of pile installation. be adequate for engineering purposes. Indeed the LE analysis,
Some Authors have suggested utilising to this aim the results based on the moduli back figured by the initial stiffness of the
of pile load tests (Poulos, 1972; Mandolini & Viggiani, 1997; load test on single piles, gives a rather satisfactory agreement
Mandolini et al., 1997). Such tests are often available for im- with the observed values in all these cases (fig. 38a, open
portant projects; should not this be the case, the load – settle- dots).
ment behaviour of a single pile can be simulated for instance There are, however, some cases in table 5 and in figure
by the transfer curves approach. 38a (full dots) referring to small pile groups constructed for
In order to reduce the uncertainties connected to the choice research purposes and submitted to a load level close to fail-
of the parameters, a standard procedure has been developed, ure (Brand et al., 1972; Briaud et al., 1989). For these cases
198
Made available with permission from the Publisher: Millpress Science Publishers, Rotterdam, The Netherlands
non linearity plays obviously a major role and hence LE and scattered than those on the average settlement. This may
analysis is less satisfactory, resulting in a substantial underes- be due, at least to some extent, to the unknown and variable
timation of the settlement. influence of the stiffness of the foundation and superstructure.
The NL analysis (figure 38b), which essentially consists in
adding the non-linear component of the settlement of the sin- 4.1.4 Prediction of load sharing and distribution
gle pile to the settlement of the group, obtained as in the LE
analysis, slightly improves the prediction of the average set- Among the 22 case histories listed in table 7, only 4 provide
tlement in all the cases where the LE analysis was already the elements (including a load test on single pile) to carry out
successful. In the cases where the non linearity plays a sig- a prediction by NAPRA and to perform an assessment similar
nificant role, NL analysis significantly improves the predic- to that reported in figure 38 for settlement. Two of them
tion. (Stonebridge Park, Cooke et al. 1981; Garigliano Bridge,
Russo, 1996 and present Report) are conventional piled foun-
1000 dations, designed following a capacity based approach assign-
100% ing all the load to the piles. The sodium hydroxide tanks n° 12
LE 20% and n° 14 in the Port of Napoli (Russo et al., 2004 and present
100 -20% Report), on the contrary, have piles acting as settlement re-
w c alc ulated [mm ]
-100% ducers.
The building at Stonebridge Park is founded on London
10 Clay, with 351 bored piles, 0.45 m in diameter and 13 m long,
connected by a raft in contact with the ground.
The load at the head of 8 piles and the pressure at 11
1 points beneath the raft have been measured; on this (rather
a) limited) basis Cooke et al. (1981) backfigured the load shar-
ing between piles and raft. In the early stage of construction
0,1 the raft carried above 40% of the total applied load; this per-
0,1 1 10 100 1000 centage decreased below 25% at the end of the observation.
The calculations carried out with NAPRA (Mandolini et
1000 al., 1997; Russo, 1998b; Viggiani, 1998) in undrained condi-
100% tions gave a load on the raft equal to about 20% of total; un-
NL 20% der drained conditions the load on the raft was almost negli-
100 -20% gible.
w c alc ulated [mm]
-100% For the pier of the bridge across the Garigliano, at the end
of the construction period the load taken by the piles was
10 about 78% of the net applied load. After 10 years this per-
centage increased to 87%. Considering only the bridge deck
weight and leaving apart the weight of the foundation raft
1 these percentages increase to 86% and 98% respectively. The
b)
calculations were carried out considering the raft already in
place acted upon by the bridge deck load (Mandolini et al.,
0,1 1997; Russo, 1998b; Viggiani, 1998). In undrained conditions
0,1 1 10 100 1000 89% and in drained conditions 100% of the load was carried
by the piles. The load distribution among the piles calculated
in undrained conditions by NAPRA is compared with that
1000 measured at the end of construction in the contours plot re-
100%
20% ported in figure 39. The agreement is rather satisfactory.
100 ES -20% Four steel tanks for the storage of sodium hydroxide have
w c alc ulated [mm]
been recently built in the area of the Port of Napoli. The sub-
-100%
soil of the area consists essentially of cohesionless deposits,
overlain by a cover of made ground; the water table is found
10
at a depth of 2.5 m below the ground surface. A typical soil
profile at the site of the tanks, including SPT blow counts and
2 CPT profiles is reported in figure 40.
1
The tank foundations are rather stiff reinforced concrete
raft, with CFA piles designed to act as settlement reducers.
c) The design issues, the type of piles used and the results of a
0,1
load test on a single pile are reported by Russo et al. (2004)
0,1 1 10 100 1000 and will be briefly recalled in § 5 below. Settlement and load
sharing among the raft and the piles were monitored during
w measur ed [mm] construction and first filling.
Figure 38. Comparison between predicted and measured settlement. The loading program was controlled by the storage and
supply needs of the sodium hydroxide, and thus neither a
The ES analysis (figure 38c), on the contrary, incorrectly complete filling of a single tank, nor a contemporary filling of
amplifies both the elastic and plastic components of the set- all the four tanks was performed. In figure 41 the time history
tlement of the single pile, and thus substantially overpredicts of the applied load and the average settlement of three tanks
the observed settlement. It is clear that the choice of perform- are reported, together with a sketch of the four foundations.
ing an elastic analysis on the basis of some secant modulus, The only load considered is the weight of the liquid filling the
the most widespread and apparently the most reasonable one, tanks, because the measurements of the settlement started af-
is in fact rather misleading. ter the construction of the rafts and the weight of the steel
The above comments also apply to the prediction of the tanks is almost negligible.
maximum differential settlement (Mandolini & Viggiani, The load distribution among the piles on a diameter of the
1997). The available data, however, are slightly more scanty tanks 12 and 14 is plotted in figure 42 at six typical dates. At
199
Made available with permission from the Publisher: Millpress Science Publishers, Rotterdam, The Netherlands
the beginning the piles carry part of the tank and raft weight, into account the contemporary partial re-filling of tank 12.
as it can be seen by the piles of the tank 14 which was still Analysis 1) calculated that 45% of the applied load was
unloaded. At the partial filling of the tank 12, the piles show supported by the piles of the tank 12, in substantial agreement
an edge effect, as it was to be expected under a stiff raft. It is with the observed percentage of 50% (figure 44). The pile
interesting to point out that the load on pile 39, belonging to load calculated by the analysis 2) for the tank 14 was about
the tank 14, decreases when the adjacent tank 12 is loaded 55% of the total, again in reasonable agreement with the ob-
(date 3). At the maximum applied load (date 4) both the foun- served percentage of 45%. The calculated load distribution
dations show an edge effect with a significant asymmetry among the piles located along a diameter of the tank 12 is
produced by the interaction between the two tanks. The sub- compared to the observed distribution in figure 45. The unex-
sequent unloading leaves in the piles of both tanks a residual pected observed asymmetry is larger than that revealed by the
load larger than that acting before the sequence of filling calculations, but the agreement is not that bad. Furthermore
started. the calculations reproduce the load decrease on the edge pile
In figure 43 the settlement along the same diameters at the n° 39 of the tank 14 during the loading of the tank 12.
selected dates shows how significant was the interaction in From the above comparisons, it seems that the limited
terms of rotations. available evidence is encouraging about our capacity of pre-
In figure 44 the total tanks load, including the weight of diction; further comparisons, and hence further observations,
the rafts and of the steel, is plotted against the time together are however badly needed.
with the total load acting on piles expressed as a percentage
of the tanks load. The data reported refer to the same dates se- 4.1.5 Bearing capacity
lected for figures 42 and 43, plus a further one at the begin-
ning, showing the load sharing recorded before filling started. On the basis of a broad numerical investigation, de Sanctis &
For both the tanks, the pile supported initially 30% to 40% of Mandolini (2003) developed a simple criterion to obtain the
the weight of the raft plus the steel tank. During filling the ultimate vertical load QPR of piled raft from the separate ca-
pile load increases up to about 50% of the total applied load. pacities of the unpiled raft QR and of the uncapped pile group
The unloading of the tanks produces a substantial change QP, as obtained by the conventional bearing capacity theory.
in the load supported by piles, whose percentage approaches The main parameters adopted for this study and some of
almost 100%. the results are listed in table 9. Once checked the validity of
For the tanks 12 and 14 two back-analyses have been car- conventional theories for evaluating the bearing capacities of
ried out to simulate two stages of the complex load-time his- an unpiled raft (QR) and of an uncapped pile group (QP), they
tory reported in figure 41: compared the sum of the ultimate load of single piles with
1. first filling of the tank 12 up to maximum load without those of pile groups, deriving from eq. 10 a value of the effi-
taking into account the partial intermediate unloading; ciency factor η equal to unity, in agreement with the findings
2. first filling of the tank 14 up to the maximum load, taking of Cooke (1986) for s/d ≥ 4.
Corner
800
Load [kN]
600
400 Center
200
1
2
3
4 9 10
5 7 8
5 6
3 4
1 2
a)
������
800
����������
600
������
400
200
1
2
3 9 10
4 7 8
5 5 6
4
2 3 b)
1
Figure 39. Load sharing among the piles of a quarter of the foundation of the main pier of the Garigliano bridge: a) calculated; b) measured.
200
Made available with permission from the Publisher: Millpress Science Publishers, Rotterdam, The Netherlands
NSPT qC [MPa]
0 10 20 30 40 50 60 0 10 20 30
10
20 CPT2
CPT1
25 m
Silty sand
z[m]
30 m
Figure 40. Soil profile with site investigations at the site of the tanks in the harbour of Napoli.
30
48
18 22 26
11 13 35
14 42 51
38 47
34
40 43 46 49 52
27
15 21 33 36 39 3.5 2 2 3
17 25
19 23 45
41 50
37
44
20 24 28
16
2 4 6 8 10
Lenght (m)
30 2 3 4
25 1 5
6
20
15
S 11
Q [MN]
10
S 14
5
S 12
0
27 May
05 Aug
19 Aug
11 Nov
25 Nov
08 Jul
22 Jul
10 Jun
24 Jun
14 Oct
28 Oct
02 Sep
16 Sep
30 Sep
09 Dec
0
5
S14
10
S11
w [mm]
15
20
S12
25
30
35
Figure 41. Applied load and average settlement for three tanks in the harbour of Napoli.
201
Made available with permission from the Publisher: Millpress Science Publishers, Rotterdam, The Netherlands
-200
pile 33 pile 36 pile 39 pile 40 pile 43 pile 46 pile 49 pile 52
0
200
400
pile load [kN]
4
4
600
4 4
800
4
4
4
1000
1 2 3 4 5 6
1200
4
1400
Figure 42. Load distribution among the piles on a diameter of the tanks 12 and 14.
10
w [mm]
15
20
25
1 2 3 4 5 6
30
35
Figure 43. Settlement along the diameters of the two adjacent tanks 14 and 12.
100 30000
pile 40
pile 43
pile 46
pile 49
80 24000
pile load [%]
60 18000 0
0.2
40 12000
0.4
20 Pile load [%] 6000
Total applied load 0.6
0 0
Qi/Qave
0.8
23-May
04-Jul
18-Jul
06-Jun
20-Jun
01-Aug
100 30000
Tank S14 1
80 24000
total applied load [kN]
1.2 Measured
pile load [%]
60 18000
1.4 Calculated
40 12000
1.6
20 Pile load (%) 6000
Total applied load
0 0 Figure 45. Load distribution among the piles along the diameter of
the tank 12.
23-May
04-Jul
18-Jul
06-Jun
20-Jun
01-Aug
Figure 44. Total applied load and pile load vs. the time.
202
Made available with permission from the Publisher: Millpress Science Publishers, Rotterdam, The Netherlands
Table 9: Parametric study Table 10: Results of the parametric studies (pref = 100 kPa)
Case B/d L/d n s/d Ag/A αR,10%d αR,25%d αR,10%B Case QR/prefd2 QG/prefd2 QPR/prefd2 Q/prefd2 FSR FSG FSPR ξPR
[%] [%] [%] [-] [-] [-] [-] [-] [-] [-] [-]
1 28 40 49 4 0,73 29 36 41 1 1235 4090 4909 633 1.95 6.46 7.76 0.92
2 28 40 9 4 0,08 41 62 95 2 1235 751 1828 633 1.95 1.19 2.89 0.92
3 28 40 9 8 0,33 41 66 100 3 1235 751 1891 633 1.95 1.19 2.99 0.95
4 28 20 49 4 0,73 23 31 41 4 1235 1364 2120 633 1.95 2.15 3.35 0.82
5 28 20 9 4 0,08 43 67 100 5 1235 251 1446 633 1.95 0.40 2.28 0.97
6 28 20 9 8 0,33 43 69 100 6 1235 251 1465 633 1.95 0.40 2.32 0.99
7 20 40 25 4 0,64 36 40 42 7 630 2087 2644 299 2.11 6.98 8.84 0.97
8 20 40 9 4 0,16 44 60 78 8 630 751 1284 299 2.11 2.51 4.29 0.93
9 20 40 9 8 0,64 45 67 97 9 630 751 1348 299 2.11 2.51 4.51 0.98
10 20 20 25 4 0,64 31 39 48 10 630 696 1088 299 2.11 2.33 3.64 0.82
11 20 20 9 4 0,16 44 62 87 11 630 251 761 299 2.11 0.84 2.54 0.86
12 20 20 9 8 0,64 45 68 98 12 630 251 809 299 2.11 0.84 2.70 0.92
13 12 20 9 4 0,44 40 53 66 13 227 251 408 101 2.26 2.49 4.06 0.86
14 12 40 9 4 0,44 43 52 61 14 227 751 976 101 2.26 7.47 9.71 1.00
Moreover they found systematically αP = 1 (eq. 12); hence 4.2 Horizontal loads�
derived different values of αR from the same equation at dif-
ferent displacements reached with the analysis (10%d and Poulos (1971b) first proposed the interaction factors method
25%d) or extrapolated by hyperbolic interpolation of the nu- to analyze pile groups under horizontal load, using the theory
merical data (up to 10%B). The three sets of values for αR are of elasticity to derive the interaction factor αij between a
listed in table 9 and plotted in figure 46 against the quantity loaded pile i and an unloaded adjacent pile j. The matrix of
(s/d) / (Ag/A) already introduced in § 3.2.2. As expected, αR interaction factors is symmetric and independent of the load
increases for increasing values of this quantity (see also Fig- level. Compared to the case of vertical load where αij is only
ure 25, working conditions) and for increasing displacement; a function of the spacing of the piles i and j, computational
moreover, it seems that some limiting value (s/d) / (Ag/A) ∼ complexity is slightly increased by the dependence of the in-
10 exists, above which no significant increase of αR occurs. teraction factor on the angle β between the load vector and the
line connecting the pile i and j.
100 In figure 47 (Poulos, 1971b) the variation of αij with the
angle � is reported; it attains a maximum at � = 0° and 180°
80
and a minimum at � = 90°.
0,4
60 0,35
[%]
0,3
R
40 0,25
w = 10%d 0,2
ij
20 w = 25%d 0,15
w = 10%B 0,1
β
0 0,05
0 20 40 60 0
(s/d) / (A G/A) [-] 0 15 30 45 60 75 90 105 120 135 150 165 180
β [°]
Figure 46. Relationship between αR and (s/d) / (AG/A).
Figure 47. Typical variation of αij with the angle β.
de Sanctis & Mandolini (2005a, 2005b) defined a coeffi-
cient:
Poulos & Davis (1980) presented a wide range of charts of
QPR FS PR interaction factors; Randolph & Poulos (1982) developed ana-
� PR = = (16)
QR + QP FS R + FS P lytical formulae based on the critical length, LC.
Banerjee & Davies (1978) and Davies & Budhu (1986)
representing the ratio between the ultimate load of a piled raft presented linear and non linear analyses based on boundary
as derived form the numerical analysis and that of the unpiled element method for piles embedded in non homogeneous soil
raft and of the uncapped pile group evaluated by conventional profiles.
theories. Under a given applied load Q on the piled raft, three Along a different research path, the spring model intro-
different factors of safety may be defined: that for the unpiled duced by Matlock & Reese (1960) has been extended to the
raft (FSR = QR/Q), for the pile group (FSP = QP/Q) and for the analysis of pile groups. The p-y method cannot actually ac-
piled raft (FSPR = QPR/Q). Their ratio is always equal to ξPR, count for the interaction through a continuum both along the
independently of the selected value for Q (or w). single pile and among piles in a group. In order to analyse a
In table 10 are reported the values of ξPR under a load Q pile group, the shape of the curves must thus be adapted tak-
corresponding to a settlement w = 3.5‰B for the unpiled raft, ing into account the type of soil and the geometry of the
that is typical for a piled raft under working conditions when group.
designed neglecting the contribution of the raft-soil contact The experimental findings presented in § 3.3 stimulate
(Cooke, 1986). As it can be seen, the sum of the factors of some comments.
safety of unpiled raft and uncapped pile group equals to • The widely diffused assumption that p-y multipliers fm are
within ± 20% the computed value of FSPR (0,82 ≤ ξPR ≤ 1,00). independent of displacement appears a reasonable choice
In other words, the safety factor of a piled raft is slightly if a particular displacement level is of concern, i.e. an
lower than the sum of the two safety factors of the unpiled equivalent secant approach is adopted. On the contrary, if
raft and the uncapped pile group. Such a result may be useful the objective of the analysis is an accurate prediction of
in design. the full load-displacement curve, the multipliers should be
203
Made available with permission from the Publisher: Millpress Science Publishers, Rotterdam, The Netherlands
better selected as decreasing functions of the displacement leads to unnecessarily small settlement and to a significant
level. cost increase. The requirement of satisfying some FS value
• The p-y curve method for the single pile and its extension for each pile individually is a further aggravation, because of
to the pile group is an empirical procedure; both the shape the non uniform distribution of the load among piles due to
of the curves and their multipliers can only be deduced by the stiffness of the cap and possibly of the superstructure (§
ad hoc experiments. Until observations on large size 3.2.2), and leads to overly conservative design. When it was
groups will not be available, the reliability of the method adopted by some UK Road Construction Units in the 1980’s,
cannot be assessed. Bearing in mind the marked influence the cost of the piled bridge foundations at least doubled (Bur-
of the size of the group on the efficiency under vertical land, 2004); as a consequence, it was quickly dropped and de-
loads, and the lack of experimental data on large pile signers reverted to applying a factor of safety to the bearing
group subjected to horizontal load, the use of the multipli- capacity of the group as a whole. In fact, if a corner pile ap-
ers obtained by experiments on small groups in the analy- proaches its full capacity its stiffness decreases and load re-
sis of large pile groups under horizontal load is question- distribution takes place to the adjacent piles. Burland (2004)
able and could be overly unconservative. comments that the analogous situation for a rigid footing is
The experience shows that the asymmetry caused by the shad- that the high edge stress causes local yield with stress redis-
owing effect is confined to the very edge of the group tribution towards the middle of the footing; it has never been
(McVay et al. 1998). The rows typically affected are just the suggested that local factors of safety should be applied to
leading, the second and the trailing one. Modelling the soil as such edge stress.
an elastic continuum and adopting boundary elements or in- A good design should be aimed to satisfying some opti-
teraction factors, such an asymmetry is not reproduced. Nev- mising criterion as, for instance, that “achieving maximum
ertheless these methods take into account the geometry and economy of the solution while keeping a satisfactory behav-
the size of the group. Is it preferable to push the use of em- iour” (Russo & Viggiani, 1998). It is in the relation between
pirical methods beyond the limits of the available evidence, or these two aspects that an optimum has to be found; it is sur-
to use rational methods based on an analysis of the interaction prising that most of the papers dealing with optimum design
through the continuum? In our opinion, the question is open. do not deal at all with cost of the solution and/or definition of
a satisfactory behaviour.
In order to clarify the interrelations between cost and per-
5 DESIGN formance, reference may be made to Figure 48 (de Sanctis et
al., 2002), where a quantity S defining the behaviour (abso-
Only the design under vertical loads is considered. lute or differential settlement, stress, distortion) is plotted
The first step in the design of a pile foundation is the se- against the cost of the solution.
lection of the pile type and installation method. The choice
should depend on the subsoil properties, but it is often influ-
enced by the local market and the regional practice.
Once the selection of the pile type has been made and the
installation method specified, the next step is the evaluation
2
of the bearing capacity of the single pile, that has been shown
to be tightly connected to the installation procedures (§ 2.1.1).
+
204
Made available with permission from the Publisher: Millpress Science Publishers, Rotterdam, The Netherlands
criterion. In the latter case, the most convenient solution is uniform load (figure 50) and expressed the results in terms of
again that corresponding to Smin. Any further increase of the the ratio ∆w/∆wR between the differential settlement of a
cost is useless, or even detrimental. piled raft and that of the corresponding unpiled raft. The ratio
An example of steady improvement of the performance is mainly affected by the parameter Ag/A, and attains a mini-
with increasing the number of piles (curve 1 in figure 48) is mum at a spacing ratio s/d = 3 irrespective of the number of
the redesign of the foundation of the main pier of the Garigli- piles. In figure 51 the results for two different values of raft-
ano bridge (Mandolini et al., 1997). The total load acting on soil relative stiffness KRS (Fraser & Wardle, 1976) and three
the foundation during construction was Q=113 MN (figure different ratios L/B ≤ 1 are reported. Larger values for KRS
14), approximately equal to the bearing capacity of the un- and L/B are considered unrealistic.
piled raft (QR = 112 MN). In a conventional capacity based
design, 144 piles were added in order to increase the bearing B = 45m
capacity. With an ultimate capacity of the single pile QS = 3 n = 9, 25, 49, 81, 121, 169, 225
MN, as deduced by load tests to failure, and a group effi- Ag
L/B = 0.4, 0.7, 1.0
ciency η = 0.7, the Italian regulations (no contribution of the
raft, FS ≥ 2,5) have been satisfied (Viggiani, 2001). As re- L/d = 18, 31.5, 45
ported in § 3.1, such a design resulted in a measured settle- s/d = 3, 4.5, 6, 7.5, 9, 12, 15
ment of 52 mm, while the actual load transmitted to the piles Krs = 0.01, 0.10
QP was about 87% Q.
Mandolini et al. (1997) back analysed this case history by
the computer code NAPRA, obtaining satisfactory agreement
both for settlement (figure 38) and for load distribution (fig-
ure 39). The same numerical model was then adopted to pre-
0.0 0.05 0.10 0.15 0.20 0.25 0.30 cu (MPa)
dict the behaviour of the foundation with a decreasing number
of piles uniformly spread below the raft. The results obtained
are reported in figure 49 in terms of settlement and load shar- 0.0 10 20 30 40 50 60 E (MPa)
ing, together with the prediction by Mandolini (2003), based 0.0
on the simple PDR method (Poulos, 2000b).
Leaving apart the extreme solutions with a very small 5.0
number of piles, for which the assumption of elastic raft-soil
10.0
interaction is strongly questionable, figure 49 shows that a
significant reduction of pile number (say from 144 to 72) is
15.0
possible without noticeable increase of the settlement. Since
the cost of the foundation is roughly proportional to the total 20.0
length of piles, in this case a trend of performance vs. costs
like curve 1 in figure 49 is occurring. There is a potential for 25.0
substantial savings without significant reduction of perform-
ance, provided such a design was allowed by the existing 30.0
regulations.
z (m)
1,00 Mandolini et al. (1997)
z > 200m → E = ∞, c u = ∞
PDR
0,75
Figure 50. Scheme for parametric study of large piled rafts.
w /wR
0,50
The same results are plotted in figure 52 against the total
pile length nL, which may be considered roughly proportional
0,25
to the cost of each solution. The curve is similar in shape to
curve 2 in figure 48, and a minimum does actually occur. Fur-
0,00 thermore, the longer the piles, the more economic is the solu-
0 40 80 120 tion; for a given total quantity of piles, a small number of
number of piles, n long piles is the most convenient choice. For instance, adopt-
ing 81 piles with L=31.5 m (L/B = 0.7; nL = 2552 m),
1,00 Mandolini et al. (1997) ∆w/∆wR reduces to 15% with KRS = 0.01, and to 2% with KRS
= 0.10. The same results may be obtained by using only 30
PDR piles with L = 45 m (L/B = 1; nL = 1350 m). Similar charts
0,75
have been recently proposed by Reul (2002), confirming that
the addition of a small number of relatively long piles in the
0,50 central zone of the foundation is very effective in reducing
R
205
Made available with permission from the Publisher: Millpress Science Publishers, Rotterdam, The Netherlands
ture consists of steel frames with reinforced concrete stiffen- a suitable factor of safety, and thus the primary reason to
ing cores. About 66% of the total load is transmitted via the add piles is to achieve a sufficient factor of safety. This
stiffening cores, while the remaining 34% is uniformly shared generally means that the width B of the raft amounts to a
by the steel columns. few metres, typically 5m < B < 15 m. In this range the
flexural stiffness of the raft may be made, and usually is,
KRS = 0,01
rather high and the differential settlement does not repre-
1,20 L/B = 0,4
s/d = 3
sent a major problem; the requisite for an optimum design
L/B = 0,7 is the limitation of mean settlement and, subordinately,
1,00
L/B = 1,0 bending moments and shears in the raft. The width of the
0,80 raft B is generally small in comparison with the length of
the piles L (say B/L < 1);
0,60 • large piled rafts, i.e. those in which the bearing capacity is
w/ w R
0,60 0,40
w/ w R
0,40 0,20
0,20 0,00
0,00 -0,20
0 5000 10000 15000
-0,20 nL [m]
0,00 0,20 0,40 0,60 0,80 1,00
A g/A 1,20 L/B = 0,4
KRS = 0,10
Figure 51. Influence of the ratio Ag/A on the maximum differential L/B = 0,7
1,00 s/d = 3
settlement of piled rafts (∆w) and unpiled rafts (∆wR). L/B = 1,0
0,80
The code NAPRA was used again in order to fit the ob-
served behaviour of the towers in terms of settlements. The 0,60
w/ w R
206
Made available with permission from the Publisher: Millpress Science Publishers, Rotterdam, The Netherlands
86.50 m
Y3
32.70 m
Y2
40.00 m 47.10 m
Tower U Tower A
benchmark for optical survey
Figure 53. Plan of the foundation and cross section of the towers.
42.25 m
1.00 m
1.60 m
2.40 m
207
Made available with permission from the Publisher: Millpress Science Publishers, Rotterdam, The Netherlands
distance across the slab [m] bearing capacity failure (FS � 5), but an average settlement of
160 to 180 mm was predicted. The computed settlement, and
0 10 20 30 40
the expected associated differential settlement, was consid-
0 ered incompatible with a safe operation of the tanks. The next
choice was a piled foundation. Following the Italian regula-
10 computed, piled raft tions, the conventional capacity based design ended in a total
of 128 piles for the four tanks, with an estimated settlement of
12 – 13 mm.
settlement [mm]
20
The solution finally adopted includes 52 piles instead of
128, with an estimated settlement of 14 to 21 mm. Monitoring
30
of the tanks at first filling showed a completely satisfactory
behaviour (figure 41).
40
50 alignment Y2 1000
computed, unpiled raft alignment Y3
60
Figure 55. Measured and computed settlement for actual piled raft
[3]
208
Made available with permission from the Publisher: Millpress Science Publishers, Rotterdam, The Netherlands
Lack of a proper graduation in designing piled foundation geotechnical parameters rather than the details of the analysis
becomes more and more evident for point 5, i.e. for large employed”.
piled foundations. In such cases, piles are necessary only to
reduce settlement (overall and differential) to some admissi-
ble value. Often the pile length L cannot exceed the raft width ACKNOWLEDGMENTS
B, so that it is practically impossible to reduce significantly
the absolute settlement, as shown above in figure 55 (de The Authors are deeply in debt to G. Price, whose skill and
Sanctis et al., 2002). In these cases piles should be located to availability over the years made possible most of the field
prevent differential settlement (figures 51 and 52). measurements of load sharing.
It is a pity that restrictions by codes and regulations hinder a Alizadeh, M. and Davisson, M.T. 1970. Lateral load tests on piles –
free selection of the most proper design approach, forcing to Arkansas River Project. Journal of the Soil Mechanics and
the adoption of a capacity based design whatever the actual Foundations Division, 96, SM 5, 1583 – 1604
design requirement is. This was not the case of the of Euro- Banerjee, P.K. 1970. A contribution to the study of axially loaded pile
foundations. PhD Thesis, University of Southampton
code n° 7 (EN 1997-1: 1994), provisionally issued more than Banerjee, P.K., 1978. Analysis of axially and laterally loaded pile
ten years ago. Quoting from § 7 - Pile foundations: “When the groups. In: Developments in Soil Mechanics, C.R.Scott ed., Ap-
piles are used to reduce the settlement of a raft, their resis- plied Science Publisher, 317-346
tance corresponding to the creep load may be used in analys- Banerjee, P.K. and Butterfield, R. 1981. Boundary element method in
ing the serviceability state of the structure.” engineering science. McGraw Hill
Inexplicably, the last version of Eurocode 7 (EN 1997-1: Banerjee, P.K. and Davies, T.G. 1978. The behaviour of axially and
2004), formally approved in March 2004, includes a different laterally loaded single piles embedded in nonhomogeneous soils.
statement: “The provisions of this Section should not be ap- Géotechnique, 28, No 3, 309 – 326
plied directly to the design of piles that are intended as set- Barton, Y.O. 1984. Response of pile groups to lateral loading in the
centrifuge. Proc. Symp. on Appl. of Centrifuge modelling to Geo-
tlement reducers, such as in some piled raft foundations.” tech. Design, Balkema, Rotterdam, The Netherlands
Piled rafts are thus not forbidden, and a somewhat Basile, F. 1999. Non linear analysis of pile groups. Proc. Inst. Civil
stretched interpretation of this statement allows any design Eng. Geotech. Eng., 137, No 2, 105-115
approach. Considering the rather bureaucratic aptitude of Borel, S. 2001a. Comportement et dimensionnement des foundations
many public authorities, however, the Authors are rather mixtes. Thèse de Docteur de ENPC, Spécialité Géotechnique,
sceptic on the practical possibilities of introducing significant 297 pp.
innovations in design. Again, the new regulation acts as a re- Borel, S. 2001b. Calcul des fondations mixtes semelle-pieux sous sol-
straint rather than a stimulus. It seems that a step backwards lecitation horizontale. Proc. XV Int. Conf. Soil Mech. Geot.. Eng.
has been taken, just in a period when significant advances in Istanbul, 2, 847-850
Brand, E.W., Muktabhant, C. and Taechathummaraka, A. 1972. Load
the understanding of the mechanisms and in the capacity of tests on small foundations in soft clays. Proc. Spec. Conf. on Per-
analysis have been registered. formance of Earth and Earth Supported Structures, Purdue Uni-
In the Authors’ opinion, there are a number of points that versity, 1, part 2, 903-928
can be considered “ready for use” in design: Breth, H. and Amann, P. 1974. Time settlement and settlement distri-
• increasing the number of piles is generally beneficial but bution with depth in Frankfurt clay. Proc. BGS Conf. Settlement
does not always produce the optimum solution. There is an of Structures, Cambridge, Pentech Press, 141-154
upper limit to the useful number of piles, beyond which Briaud, J.L., Tucker, L.M. and Ng, E. 1989. Axially loaded 5 piles
further increase is useless or even detrimental. Conven- group and single pile in sand. Proc. XII Int. Conf. Soil Mech.
tional design results generally in a number of piles beyond Found. Eng., Rio de Janeiro, 2, 1121-1124
Brignoli, E., Cestelli Guidi, M. and Ventura, P. 1997. Analisi speri-
this limit; mentale dei parametri di progetto delle strutture in c.a. di una ci-
• to control the average settlement, an optimum performance miniera. Unpublished Report, ISMES
is achieved by the use of piles with a length L larger than Brinch-Hansen, J. 1970. A revised and extended formula for bearing
the width B of the raft, uniformly spread below the whole capacity. Danish Geotechnical Institute, Bulletin n° 28, 5-11
raft area. This is practicable for small and medium rafts, Broms, B.B. 1964a. Lateral resistance of piles in cohesive soils.
but not for large ones. In this latter case, the average set- Journ. Soil Mech. and Found. Div., ASCE, 90, SM 2, 27-63
tlement is but slightly reduced by the addition of piles; Broms, B.B. 1964b. Lateral resistance of piles in cohesionless soils.
• to control the differential settlements, an optimum per- Journ. Soil Mech. and Found. Div., ASCE, 90, SM 3, 123-156
formance is achieved by a suitable location of a relatively Brown, D.A., Reese, L.C. and O’Neill, M.W. 1987. Cyclic lateral
loading of a large-scale pile group. Journal of Geotechnical En-
small number of piles, rather than using a larger number of gineering, 113, No 11, 1326 – 1343
piles uniformly spread or increasing the raft thickness. The Brown, D.A., Morrison, C. and Reese, L.C. 1988. Lateral load behav-
most suited location depends on the distribution of the ior of pile group in sand. Journal of Geotechnical Engineering,
load; for instance, in the case of uniform load over the raft, 114, No 11, 1261 – 1276
the piles are best concentrated in the central zone (Ag/A = Budhu, M. and Davies, T.G. 1987. Nonlinear analysis of laterally
0.20 to 0.45). Again, the longer the piles, the most effec- loaded piles in cohesionless soils. Canadian Geotechnical Jour-
tive they are in reducing the differential settlement; nal, 24, No 2, 289 – 296
• the thickness of the raft affects the bending moments and Burland, J.B. 1973. Shaft friction of piles in clay. A simple funda-
the differential settlement, but has little effect on the load mental approach. Proc. Ground Engineering, 6, N° 3, 30-42.
Burland, J.B. 2004. Interaction between structural and geotechnical
sharing between raft and piles and on the average settle- engineer. New Perspectives in the Design and Construction of
ment. Foundation Structures. Joint Structural Division Annual Semi-
As a final comment, the Authors wish to insist that, in their nar, Hong Kong
opinion, the available procedures of analysis may be consid- Butterfield, R. and Banerjee, P.K. 1971 The elastic analysis of com-
ered satisfactory for engineering purposes provided they are pressible piles and pile groups. Géotechnique, 21, No 1, 43-60
properly applied, paying due attention to the correspondence Caputo, V. and Viggiani, C. 1984. Pile foundation analysis: a simple
relations. The same opinion is expressed in a slightly different approach to non linearity effects. Rivista Italiana di Geotecnica,
form by Poulos et al. (2001) when they write that: “the key to 18, No 1, 32-51
successful prediction is more the ability to choose appropriate Caputo, V. and Viggiani, C. 1988. Some experiences with bored and
209
Made available with permission from the Publisher: Millpress Science Publishers, Rotterdam, The Netherlands
auger piles in Naples area. Deep Foundations on Bored and Au- tions for laterally loaded piles in elastic half-space. II Int. Conf.
ger Piles, Balkema, Rotterdam, 273-282 on Num. Meth. in Geom., Blacksburg, Virginia, 3, 1367 – 1370
Chaudhry, A.R. 1994. Static pile-soil-pile interaction in offshore pile Evangelista, A., Pellegrino, A. and Viggiani, C. 1977. Variability
groups. Ph.D. thesis, Brasenose College, University of Oxford among piles of the same foundation. IX Int. Conf. Soil Mech.
Chow, Y.K. 1987. Axial and lateral response of pile groups embed- Found. Eng., Tokyo
ded in nonhomogeneous soils. Int. Journ. for Num. and Anal. Fioravante, V. 2002. On the shaft friction modelling of non-
Meth. in Geomech. 11, No 6, 621-638 displacement piles in sand. Soils and Foundation, 42, No 2, 23-
Colombi, A. 2005. Physical modelling of an isolated pile in coarse 33.
grained soils. Ph. D. thesis, University of Ferrara, Italia. Fleming, W.G.K., Weltman, A.J., Randolph M.F. and Elson W.K.
Conte, G. 2003. Analisi Teorico-Sperimentale delle Fondazioni su 1992. Piling Engineering, 2nd edition, Blackie A&P
Pali. Report, Department of Civil Engineering, Second Univer- Fraser, R.A. and Wardle, L.J. 1976. Numerical analysis of rectangular
sity of Napoli, Italy rafts on layered foundations. Géotechnique, 26, No 4, 613-628
Conte, G., Mandolini, A. and Randolph, M.F. 2003. Centrifuge mod- Goossens D., Van Impe W.F. 1991. Long term settlements of a pile
elling to investigate the performance of piled rafts. Proc. of the group foundation in sand, overlying a clay layer. Proc. X Eur.
IV Int. Seminar on Deep Foundations on Bored and Auger Piles, Conf. Soil Mech. Found. Eng., Florence, vol. 1, 425-428
Millpress, Rotterdam, 359-366 Hansbo, S. 1984. Foundations on friction creep piles in soft clays.
Cooke, R.W. 1986. Piled raft foundations on stiff clays: a contribu- Proc. 1st Int. Conf. Case Histories Geotech. Eng., St. Louis, Mis-
tion to design philosophy. Géotechnique, 36, No 2, 169-203 souri, USA, 2, 913-922
Cooke, R.W., Bryden Smith, D.W., Gooch, M.N. and Sillet, D.F. Hight, D.W. and Green, P.A. 1976. The performance of a piled raft
1981. Some observations of the foundation loading and settle- foundation for a tall building in London. Proc. VI European
ment of a multi-storey building on a piled raft foundation in Lon- Conf. Soil. Mech. Found. Eng., Vienna, 2, 234-237
don clay. Proc. I.C.E., part 1, 70, 433-460. Holeyman, A. and Charue, N. 2003. International pile capacity pre-
Coyle, H.M. and Reese, L.C. 1966. Load transfer for axially loaded diction event at Limelette. Belgian Screw Pile Technology – de-
piles in clay. Journ. Soil Mech. and Found. Div., proc. ASCE, 92, sign and recent development, Maertens & Huybrechts eds.,
SM2, 1-26 Balkema, Rotterdam, 215-234
D’Appolonia, E., and Romualdi, J.P. 1963. Load transfer in end bear- Hooper, J.A. 1973. Observations on the behaviour of a piled raft
ing steel H-piles. Journ. Soil Mech. and Found. Div., proc. foundation on London clay. Proc. Inst. Civ. Eng. , part 2, 1973,
ASCE, 89, SM2, 1-25 55, 855-877.
Davies, T.G. and Budhu, M. 1986. Non-linear analysis of laterally Hooper, J.A. 1979. Review of behaviour of piled raft foundations.
loaded piles in heavily overconsolidated clays. Géotechnique, 36, CIRIA report n° 83.
No 4, 527 – 538 Hooper, J.A. and Wood, L.A. 1977. Comparative behaviour of raft
Day, R.W. 2000. Geotechnical Engineer’s Portable Handbook. Mc and piled foundations. Proc. IX Int. Conf. Soil Mech. And Found.
Graw Hill Portable Engineering, New York, 610+122 pp. Eng., Tokyo, Japan, 1, 545-550
De Beer, E. 1988. Different behaviour of bored and driven piles. Hooper, J.A. and Levy, J.F. 1981. Behaviour of a flexible piled foun-
Proc. Int. Geoth. Sem. on Deep Found. On Bored and Auger dations. Proc. 10th Int. Conf. Soil Mech. Found. Eng., Stockolm,
Piles, Ghent, Balkema ed., Rotterdam, 47 – 82 Sweden, 2, 735-740
Decourt, L., 1995. Prediction of load settlement relationships for Horikoshi, K., Matsumoto, T., Watanabe, T. and Fukuyama, H. 2002.
foundations on the basis of the SPT-T. Ciclo de Conferencias In- Performance of piled raft foundations subjected to seismic loads.
tern. Leonardo Zeevaert, UNAM, Mexico, 85-104 Foundation design codes and soil investigation in view of Inter-
De Cock, F., Gwidzala K., Holeyman A.., Legrand C., Lehane B., national harmonization and performance, Honyo et al. eds.,
Mandolini A. and Skov R. 1999. Survey report on the present- Swets & Zeitlinger, Lisse, 381-389
day design methods for axially loaded piles. European practice. Huang, A.B., Hsueh, C.K., O’Neill, M.W., Chern S. and Chen, C.
Addendum to Proc. XII European Conference Soil Mechanics 2001. Effects of construction on laterally loaded pile groups.
and Geotechnical Engineering. Amsterdam, the Netherlands. Journal of Geotechnical and Geoenvironmental Engineering,
De Cock, F. and Legrand, C. (ed.) 1997. Design of axially loaded 127, No 5, 385 – 397
piles – European Practice. Proc. ERTC3 Seminar, Brussels. Ilyas, T., Leung, C.F., Chow, Y.K. and Budi, S.S. 2004. Centrifuge
Balkema, Rotterdam model study of laterally loaded pile groups in clay. Journal of
De Mello, V.F.B. 1969. Foundations of buildings on clay. State of the Geotechnical and Geoenvironmental Engineering, 130, No 3,
Art Report, Proc. VII Int. Conf. Soil Mech. Found. Eng., Mexico 274 – 283
City, vol. 1, 49-136 ISC2 2004. International Prediction Event on the Behaviour of Bored,
de Nicola, A. and Randolph, M.F. 1993. Tensile and compressive CFA and Driven Piles in ISC’2 experimental site. Workshop, 2nd
shaft capacity of piles in sand. Journ. Soil Mech. and Geotech. Int. Conference on Site Characterization. Porto, Portugal, 19-22
Eng., proc. ASCE, 119, GT12, 1952-1973 september 2004
de Sanctis, L. 2001. Modellazione ed analisi delle fondazioni su pali. Jardine, R.J. and Chow. F. 1996. New design methods for offshore
Ph. D. Thesis, Università di Napoli, Federico II. piles. MTD Pub. 96/103, Marine Technology Directorate, Lon-
de Sanctis, L., Mandolini, A., Russo, G. and Viggiani, C. 2002. Some don, UK.
remarks on the optimum design of piled rafts. ASCE Geotech. Jamiolkowski, M.B. 2004. Axial load capacity of a bored pile in
Spec. Publ. 116, O’Neill & Townsend ed., Orlando, 405-425 coarse grained soils. Unpublished lecture, Turin Technical Uni-
de Sanctis L. and Mandolini A. 2003. On the ultimate vertical load of versity
piled rafts on soft clay soils. Proc. 4th Int. Geot. Seminar on Jendeby, L. 1986. Friction piled foundations in soft clay. A study of
Deep Foundations on Bored and Auger Piles, Ghent. Millpress, load transfer and settlements. Ph.D. Thesis, Dept. Geotech. Eng.
Rotterdam, 379-386 Chalmers University of Tech, Goteborg
de Sanctis, L. and Mandolini A. 2005a. A simple criterion to evaluate Joustra, K., De Sitter, W.R. and Den Ouden, N.W. 1977. Tall build-
the bearing capacity of a vertically loaded piled raft. Proc. Int. ing settlement and pile load measurements. Proc. IX Int. Conf.
Geotech. Conf. on Soil Structure Interaction: calculation meth- Soil Mech. Found. Eng., Tokyo
ods and engineering practice. Saint Petersburg, 26-28 may Katzenback, R., Arslan, U., Gutwald, J., Holzhäuser, J. and Quick, H.
de Sanctis, L. and Mandolini A. 2005b. Bearing capacity of piled 1997. Soil structure interaction of the 300 m high Commerzbank
supported rafts on soft clay soils. Journ. Soil Mech. and Geotech. tower in Frankfurt am Main. Measurements and numerical stud-
Eng., proc. ASCE, submitted for publication ies. Proc. XIV Int. Conf. Soil Mech. Found. Eng., Hamburg, 2,
Duncan, J.M., Evans, L.T. and Ooi, P.S.K. 1994. Lateral load analy- 1081-1084
sis of single piles and drilled shafts. Journal of the Geotechnical Katzenbach, R., Arslan, U. and Moormann, C. 2000. Piled raft foun-
Engineering Division, ASCE, 120, No 5, 1018 – 1033 dations projects in Germany. Design applications of raft founda-
Ekstrom, J. 1989. A field study of model pile behaviour in non- tions, Hemsley J.A. Editor, Thomas Telford, 323-392
cohesive soils. Ph. D. Thesis, Chalmers University of Technol- Katzenbach, R., and Moormann, C. 2001. Recommendations for the
ogy, Sweden. design and construction of piled rafts. Proc. XV Int. Conf. Soil
El Mossallamy, Y. and Franke, E. 1997. Piled rafts: numerical model- Mech. Found. Eng., Istanbul, 2, 927-930
ling to simulate the behaviour of piled raft foundations. Pub- Kenny, M.J., Guasti, S. and Zsak, P. 2003. Continuous flight auger
lished by the authors, Darmstadt, 182. boring in sandy soils. BGA Int. Conf. on Foundations: Innova-
Evangelista, A. and Viggiani, C. 1976. Accuracy of numerical solu- tions, observations, design and practice. Thomas Telford, Lon-
210
Made available with permission from the Publisher: Millpress Science Publishers, Rotterdam, The Netherlands
don, 434-441 MELT 1993 Regles techniques de conception et de calcul des fonda-
Kezdi, A. 1957. Bearing capacity of piles and pile groups. Proc. IV tions des ouvrages de genie civil. CCTG, Fascicule No 62, Titre
Int. Conf. Soil Mech. Found. Eng., London, 2, 46 V, Min. de l’equip. du lodge. et des transp., Paris.
Kim, J.B, and Sing, P.L. 1979. Pile cap soil interaction from full scale Meyerhof, G.G., 1956. Penetration tests and bearing capacity of co-
lateral load tests. Journ. Geotech. Eng. ASCE, 105, GT5, 643- hesionless soils. Journ. Soil Mech. and Found. Div., proc. ASCE,
653 82, SM1, 1-19
Kitiyodom, P., Matsumoto, T. and Kanefusa, N. 2004. Influence of Meyerhof, G.G. 1959. Compaction of sands and bearing capacity of
reaction piles on the behaviour of a test pile in static load testing. piles. Journ. Soil Mech. And Found. Div., Proc. ASCE, 85, SM6,
Canadian Geotechnical Journal, 41, 408 – 420 1-29
Koerner, R.M. and Partos, A. 1974. Settlement of building on pile Meyerhof, G.G., 1976. Bearing capacity and settlement of pile foun-
foundation in sand. Journ. Soil Mech. Found. Div., Proc. ASCE, dations. Journ. Geotech. Eng. Div., proc. ASCE, 102, GT3, 195-
SM6, 1-29 228
Koizumi, Y. and Ito, K. 1967. Field tests with regard to pile driving Mokwa, R.L. 1999. Investigation of the resistance of pile caps to lat-
and bearing capacity of piled foundations. Soils and Founda- eral loading. Ph. D. Thesis, Virginia Polytechnic Institute and
tions, 7, No 3, 149-202 State University.
Kotthaus, M. and Jessberger, H.L. 1994. Centrifuge model tests on Morrison, C. and Reese, L.C. 1986. A lateral load test of a full-scale
laterally loaded pile groups. XIII ICSMFE, New Delhi, 2, 639 – pile group in sand. Geotechnical Engineering Report GR86-1,
644 Geotechnical Engineering Center, University of Texas, Austin
Kraft, L.M., Ray, R.P. and Kagawa, T., 1981. Theoretical t-z curves. Morton, K. and Au, E. 1974. Settlement observations on 8 structures
Journ. Geotech. Eng., proc. ASCE, 107, GT11, 1543-1561 in London. Proc. BGS Conf. Settlement of Structures, Cam-
Kulhawy, F.H. and Chen, Y.J. 1993. A thirty year perspective on bridge, 1974, Pentech Press, 183-203
Broms’ lateral loading models, as applied to drilled shafts. Proc. O’Neill, M.W., Ghazzaly, O.I. and Ha, H.B. 1977. Analysis of Three-
Broms Symposium, Nanyang Univ. of Technology, Singapore. Dimensional Pile Groups with Non-Linear Soil Response and
LCPC, 1998. Chargement vertical de réseaux de micropieux en cen- Pile-Soil-Pile Interaction. Proc. 9th Annual OTC, Houston, Paper
trifugeuse. Rapport du Projet national FOREVER, 48 OTC 2838, 245-256
Liu, J.L., Yuan, Z.L. and Shang, P.K. 1985. Cap-pile-soil interaction O’Neill, M.W., Hawkins, R.A., and Mahar, L.J., 1982. Load Transfer
of bored pile groups. Proc. XI Int. Conf. Soil Mech. Found. Eng., mechanisms in piles and pile groups. Journ. Geotech. Eng.Div.,
San Francisco, 3, 1433-1436. proc. ASCE, 108, GT12, 1605-1623
Liu, J., Huang, Q., Li, H. and Hu, W.L. 1994. Experimental research O’Neill, M.W. and Dunnavant, T.W. 1984. A study of the effects of
on bearing behaviour of pile groups in soft soils. Proc. XIII Int. scale, velocity and cyclic degradability on laterally loaded single
Conf. Soil Mech. Found. Eng., New Delhi, 2, 535-538 piles in overconsolidated clay. Report UHCE 84-7, Department
McVay, M., Casper, R. and Shang, T.I 1995. Lateral response of of Civil Engineering, University of Houston, Texas, 368 p.
three-row groups in loose to dense sands at 3D and 5D pile spac- Ochoa, M. and O’Neill, M.W. 1989. Lateral pile interaction factors in
ing. Journal of Geotechnical Engineering, 121, No 5, 436 – 441 submerged sand. Journal of Geotechnical Engineering, proc.
McVay, M., Zhang, L., Molnit, T. and Lai, P. 1998. Centrifuge test- ASCE, 115, GT3, 359 – 378
ing of large laterally loaded pile groups in sands. Journal of Geo- Osterberg, J.O. 1984. A new simplified methods for load testing
technical and Geoenvironmental Engineering, 124, No 10, 1016 drilled shafts. Foundation Drilling, Association of Drilled Shafts
– 1026 Contractors, 23, No 6, 9-11
Mandolini, A. 1994. Cedimenti di fondazioni su pali. Ph. D. Thesis. Osterberg, J.O. 1995. The Osterberg cell for load testing drilled shafts
Department of Geotech. Eng. University of Napoli Federico II. and driven piles. FHWA-SA-94-035 Res. Report, 1-92
Mandolini, A. 2003. Design of piled rafts foundations: practice and Peiffer, H. and Van Impe, W.F. 1993. Evaluation of pile performance
development. Deep Foundations on Bored and Auger Piles, based on soil stress measurements. Field test. Proc. 2nd Int.
Millpress, Rotterdam, 59-80 Geot. Seminar on Deep Foundations on Bored and Auger Piles,
Mandolini, A. 2004. Load tests on piles – Italian practice and per- Ghent. Balkema, Rotterdam, 385-390
sepctive. Workshop: Quasi-static pile testing. Delft University of Phung, D.L. 1993. Footings with settlement-reducing piles in non-
Technocogy, october cohesive soil. Ph. D. Thesis, Dept. of Geot. Eng., Chalmers Uni-
Mandolini, A., Price, G., Viggiani, C. and Wardle, I.F. 1992. Moni- versity of Technology, Gothenburg
toring load sharing within a large pile cap foundation. Géotecni- Potts D.M., Zdravkovic L. 2001. Finite element analysis in geotech-
que et Informatique, Paris, 113-121 nical engineering. Vol. 2, application. Thomas Telford, London,
Mandolini, A. and Viggiani, C. 1992a. Settlement prediction for piled 428 pp.
foundations from loading test on single pile. Proc. Wroth Memo- Poulos, H.G. 1968. Analysis of the settlement of pile groups. Géo-
rial Symposium on Predictive Soil Mechanics, Oxford, 464-482 technique, 18, 449-471
Mandolini, A. and Viggiani, C. 1992b. Terreni ed opere di fondazione Poulos, H.G. 1971a. Behavior of laterally loaded piles: I – single
di un viadotto sul fiume Garigliano. Rivista Italiana di Geotecni- piles. Journal of the Soil Mechanics and Foundations Division,
ca, 26, No 2, 96-116 97, SM 5, 711 – 731
Mandolini, A. and Viggiani, C. 1997. Settlement of piled foundations. Poulos, H.G. 1971b. Behavior of laterally loaded piles: II – pile
Géotechnique, 47, No 3, 791-816 groups. Journal of the Soil Mechanics and Foundations Division,
Mandolini, A., Russo, G. and Viggiani, C. 1997. Pali per la riduzione 97, SM 5, 733 – 751
dei cedimenti. Proc. Conf. Geot. di Torino, XVI Ciclo Poulos, H.G. 1972. Load settlement prediction for piles and piers.
Mandolini, A., Russo, G. and Viggiani, C. 2002. Full scale loading Journal of the Soil Mechanics and Foundations Division, 98, SM
tests on instrumented CFA piles. Deep Foundation Congress, 9, 879 – 898
GeoInstitute of the ASCE, Orlando, Florida, 1088-1096. Poulos, H.G. 1982. Developments in the analysis of static and cyclic
Mandolini, A. and Russo, G. 2005. Napra user’s guide – in press lateral response of piles. 4th Int. Conf. on Num. Meth. in Geom.,
Marchetti, S. 1989. Alcuni recenti orientamenti nell’analisi dei pali di Edmonton, Canada, 3, 1117 – 1135
fondazione. Atti XVII Convegno Italiano di Geotecnica, Taormi- Poulos, H.G. 1989. Pile behaviour – theory and application. Géotech-
na, 2, 7-33 nique, 39, No 3, 365-415
Matlock, H. 1970. Correlations for design of laterally loaded piles in Poulos, H.G. 1993. Settlement prediction for bored pile groups. Deep
soft clay. Proc. II Annual Offshore Technology Conference, Foundations on Bored and Auger Piles, Balkema, Rotterdam,
Houston, Texas, 577 – 594 103-118
Matlock, H. and Reese, L.C. 1960. Generalized solutions for laterally Poulos, H.G. 2000a. Pile testing – From the designer’s viewpoint.
loaded piles. Journal of Soil Mechanics and Foundations Divi- Statnamic Loading Test ’98, Balkema, Rotterdam, 3-21
sion, ASCE, 86, SM5, 63 – 91 Poulos, H.G. 2000b. Practical design procedures for piled raft founda-
Matlock H., Ingram, W.B., Kelly, A.E. and Bogard, D. 1980. Field tions. Design applications of raft foundations, Hemsley J.A. Edi-
tests of lateral load behavior of pile groups in soft clay. Proc. XII tor, Thomas Telford, 425-467
Annual Offshore Technology Conference, Houston, Texas, 163 – Poulos, H.G. 2003. Deep foundations – Can further research assist
174 practice?. Proc. IV Seminar on Deep Foundations on Bored and
Mattes, N.S., and Poulos, H.G., 1969. Settlement of single com- Auger Piles, Millpress, Rotterdam, 45-55
pressible pile. Journ. Soil Mech. and Found. Div., proc. ASCE, Poulos, H.G. and Davis, E.H., 1968. The settlement behaviour of axi-
95, SM1, 189-207. ally loaded incompressible piles and piers. Géotechnique,18,
211
Made available with permission from the Publisher: Millpress Science Publishers, Rotterdam, The Netherlands
212
Made available with permission from the Publisher: Millpress Science Publishers, Rotterdam, The Netherlands
213