Académique Documents
Professionnel Documents
Culture Documents
Simple Method For Dynamic Stiffness and Damping of Floating Pile Groups PDF
Simple Method For Dynamic Stiffness and Damping of Floating Pile Groups PDF
4, 557-574
A simple analytical solution is developed for com- Une solution analytique t&s simple est dbvelopp6e
puting the dynamic impedances of floating rigidly- pour calculer les imp&lances dynamiques de
capped pile groups with due consideration to groupes de pieux flottants $ t&s rigides dans le
pilesoikpile interaction. The method introduces cadre de i Iinteraction pieux/sol/pieux. La m&h-
some sound physical approximations and considers ode introduit des approximations physiques val-
the interference of cylindrical wave fields originat- ables et prend en considbration Iinterference de
ing along each pile shaft and spreading radially champs dondes cylindriques qui ont leur origine le
outward. Axial, lateral, and rocking oscillations of long de chaque fiit de pieux et se propagent radi-
rigidly-capped pile groups are studied parametri- alement vers lext6rieur. Larticle Ctudie sous
cally. Results are presented for the dynamic laspect parametrique les sollicitations axiales,
stiffness and damping of the whole group, and for laterales et oscillants de groupes de pieux P
the distribution of dynamic loads amongst the indi- casques rigides. Des r&sultats sont p&e&s pour
vidual piles. The predictions of the simple method la rigiditC dynamique et Iamortissement du groupe
for vertical and rocking oscillations compare entier et pour la distribution des charges dyna-
extremely well with rigorous numerical solutions, miques entre les pieux individuels. Les prtiictions
thereby offering a valuable insight into the nature par cette methode t&s simple des oscillations verti-
of pile-so&pile interaction. It is demonstrated and tales et basculantes supportent extr&mement bien
explained how the dynamic efficiency may far la comparaison avec celles des solutions num(?r-
exceed unity at certain resonant frequencies due to iques rigoureuses, ce qui donne un aperqu valable
destructive wave interference. The proposed de la nature de linteraction pieux-sol-pieux. On
method can be readily applied by engineers already dCmontre et explique comment le groupe dyna-
familiar with the use of static interaction factors in mique repr6sentant Ieffet utile peut dbpasser de
the design of pile groups. beaucoup luniti! $ certaines frbquences de rbon-
ante $ cause de linterfbrence destructive des ondes.
La m6thode propo&e peut &tre facilement appli-
qu&e par des ingCnieurs qui connaissent dbjzi bien
KEYWORDS: analysis; dynamics; foundations; piles; Iemploi de facteurs statiques dinteraction dans la
stiffness; waves construction de groupes de pieux.
557
558 DOBRY AND GAZETAS
of each individual pile in the 1975), as well as by using simple but physically
group sound approximations (e.g. Randolph & Wroth,
r. pile radius 1979; Scott, 1981).
time Static interaction factors do not provide useful
S axis-to-axis distance between information on the steady-state dynamic response
two neighbouring piles of pile groups, except perhaps at very low fre-
v, shear-wave velocity quencies of oscillation. Early numerical studies by
VLa Lysmers analog wave Wolf and Von Arx (1978) and Nogami (1979)
velocity = 3.4 V$[n(l - v)] showed that the dynamic group efficiency of a
4 w, e horizontal displacement, ver- cluster of piles exhibits a strongly oscillatory
tical displacement, and angle behaviour and may attain values well above
of rotation of individual pile unity-a result verified in subsequent studies by
and of pile group Nogami (1980 and 1983; Kaynia & Kausel(1982a
dynamic interaction factor and b); Tyson & Kausel (1983); Sheta & Novak
for vertically (axially) oscil- (1982); El Sharnouby & Novak (1985); Novak
lating piles (1984); Waas & Hartmann (1981 and 1984);
&h dynamic interaction factor Kagawa (1983); Roesset (1984); Sen, Davies &
for horizontally (laterally) Banerjee (1985), and others.
oscillating piles The available formulations differ from one
v, B Poissons ratio and damping another in the simplifications introduced when
ratio of the soil modelling this complicated boundary value
1 shear wave length: 1= VJf problem. They are all of a numerical nature and
w circular frequency of oscilla- they invariably involve discretizing each pile and
tion = 2lrf the supporting soil; hence, application of even the
most simplified of them may entail a substantial
Subscripts computational effort. Also, in some cases these
z, x, rx response parameters in verti- methods rely on proprietary computer codes.
cal (axial), horizontal (lateral) In this Paper a very simple analytical solution
and rotational (rocking) to the problem of dynamic pile-soil-pile inter-
oscillation, respectively action in uniform soil is developed, and para-
metric results for the stiffness and damping of
Superscripts groups of floating piles are presented.
S, G differentiation between single The advantages of the proposed simple method
pile and pile group response for pile groups can be summarized as follows.
(a) The procedure is simple enough to be taught
in a course on soil dynamics, and can be
INTRODUCTION
easily understood and applied by the engi-
At static working loads, the displacement of a pile
neer, even without the help of a computer.
increases if this pile is located within the deforma-
(b) For a wide range of material parameters, pile
tion field of a neighbouring pile. As a result, the
separation distances, and frequencies of oscil-
overall displacement uG of the group of piles is
lation, the results of the method are in excel-
greater than the individual displacement us which
lent accord with rigorous solutions.
each pile would experience were it left alone to
(c) The method can be applied to all modes of
carry the average load. The static group efficiency
oscillation while retaining its simplicity.
uSluG is thus always below unity, and it tends to
(6) The procedure can be extended to handle pile
decline when the distance between piles is short-
installation effects in a simplified but physi-
ened or when the number of piles in the group
cally sound way.
increases.
Rational analyses of pile group displacements However, some limitations of the proposed
were pioneered by Poulos (1968, 1971), who method must also be noted. First, it tends to
introduced the concept of interaction factors and overestimate the peak values of both stiffness and
showed that pile group effects can be assessed by damping for pile groups embedded in (relatively)
superimposing the effects of only two piles. In the very stiff soil, i.e. soil whose average effective
last 15 years, interaction factors for each degree Youngs modulus E, is greater than about E,/300,
of freedom of the pile head have been obtained by where E, is the pile modulus. Fortunately, this is
recourse to integral equation-based methods hardly a severe restriction on the range of appli-
(Poulos & Mattes, 1971; Butterfield & Banerjee, cability of the method: soft clays, medium clays,
1971; Poulos & Davis, 1980) and finite element and loose sands will typically have much lower
formulations (Naylor & Hooper, 1975; Ottaviani, effective moduli than E,/300, and in these soils
DYNAMIC STIFFNESS AND DAMPING OF FLOATING PILE GROUPS 559
\t
.-....
T
z
Homogeneous halfspace
,= . r- .,.,_.. . .
0.40, [j = 0.05 ,.. (. .,. . ..
. . ., . .; ;.. .;:.. .:,
:. . Homogeneous layer ,
.. ;.; ,p = 0.025 _ . ,
Bedrock
the proposed method works very well. Moreover, priate selection of the effective E,,.) (Randolph,
the method might also overpredict the 1981; Gazetas & Dobry, 1984; Fleming,
interference-related peaks of large pile groups Weltman, Randolph & Elson, 1985.) Fig. 1
(n > 16); a way of overcoming this limitation is sketches the two specific soil profiles for which
suggested in this Paper. parametric results are presented herein. However,
the method is quite general and can be directly
applied to any uniform profile for which the
PROBLEM DEFINITION dynamic impedance of the single pile has been
The problem studied in this Paper is that of a previously computed or measured.
group of floating cylindrical piles embedded in a For each particular harmonic excitation of fre-
uniform stratum or halfspace and subjected to an quency w, the complex-valued dynamic imped-
arbitrary harmonic excitation at the top. The ance XG of the pile group is defined as the ratio
applied dynamic forces are transmitted onto each of total excitation (vertical force FG; horizontal
pile through the rigid cap. Vertical, horizontal, force HG; or rocking moment MC) over the corre-
and rocking oscillations are considered. sponding motion of the rigid cap (vertical dis-
The soil is modelled as a linear hysteretic placement wG; horizontal displacement uG; or
material of Youngs modulus E,, Poissons ratio rotation eG). For example, for vertical excitation
v, material damping ratio 8, and shear velocity
.IX?~=I?~+~~,,C,~=F~/~~
I z (1)
V,. Each pile is a solid cylinder of length L and
diameter d, made of a linear elastic material of and similarly for XXG and XrXG, where a, =
Youngs modulus Ep . (The results are also applic- wd/Vs is a dimensionless frequency in the equa-
able to hollow-cylinder shapes, to metal pipes tion. In all cases i = ,,I- 1. The terms RaG and
filled with concrete, or even to H-piles, by appro- C,G (where a = z, x, or rx) are interpreted as
560 DOBRY AND GAZETAS
equivalent spring and dashpot coefftcients at It is worth drawing an analogy with scattering
the head of the pile group; they are both func- of sound waves by a rigid cylinder of diameter d:
tions of the circular frequency of excitation for wavelengths 1 greater than the perimeter nd,
w = 2rrj In this Paper KaG and C,G_are contrast- the net pressure acting on the cylinder (per unit
ed to the corresponding values of KaS and C, of length) has only a negligible phase lag behind the
the single pile, evaluated at the same frequency w. pressure which the incident wave would induce at
the centre of the cylinder, if the latter were not
there (Morse & Ingard, 1968). In our case A 2
2nd and hence the approximation is valid with a
FUNDAMENTAL ASSUMPTIONS OF THE good degree of accuracy.
METHOD At the frequency range of interest, as the active
Poulos (1968, 1971) superposition procedure, pile p undergoes oscillations in, say, the vertical
originally developed for statically loaded pile direction, cylindrical waves are assumed to
groups, is also valid for the dynamic problem- emanate from the pile perimeter along the pile
an assumption employed frequently in other for- length and to propagate radially outward in the
mulations. Kaynia & Kausel (1982a and b); horizontal direction. This assumption is remi-
Sanchez-Salinero (1983); and Roesset (1984) have niscent of the shearing of concentric cylinders
demonstrated that the results of this approx- around statically loaded piles and pile groups
imation are, indeed, in very good agreement with assumed by Randolph & Wroth (1978 and 1979)
more rigorous dynamic solutions. Therefore, the and is also somewhat similar to the Winkler
response of the pile group can be obtained from assumption introduced by Novak (1974) and
interaction factors derived from the study of only extensively used in dynamic analyses of pile
two piles at a time. In essence, it is assumed that groups. However, the simple method presented
the presence of the rest of the piles does not here goes a step further; the cylindrical waves are
appreciably affect the interplay between the two assumed to emanate simultaneously from all
piles being considered. This means that, when points along the pile length and hence, for a
computing the influence of pile p on pile q the homogeneous deposit, to spread out in-phase and
intermediate piles are considered to be form a cylindrical wavefront having a common
transparent-an assumption not far from reality axis with that of the generating pile (Fig. 2(a and
in view of the fact that the wavelengths (1) of b)). For an observer at the ground surface, the
practical interest propagation and interference of these cylindrical
wavefronts would appear as Fig. 2(a). (This is
L > 6d (2)
reminiscent of the water waves that would be
are too large for the waves propagating in the soil created by a very light, rigid structure floating on
to see the cylindrical pile. However, when the a pond, composed of a horizontal platform on
number of piles in a group is very large and the columns, and excited by steady-state vertical
pile spacing relatively short, the interaction vibration.)
between two distant piles in the group will The variation of wave amplitude with depth
unavoidably be reduced due to scattering of along the cylindrical front arriving to the passive
waves and the corresponding shadows formed pile is assumed to be analogous with the ampli-
by the piles in-between. tude variation of the active single pile, as depicted
To derive the influence of the motion of pile p in Fig. 2(b). For steady-state harmonic oscillation,
(active) on pile q (passive) we can replace pile q the assumption of waves originating simulta-
by its axis, neglecting its dimensions. Sanchez- neously along the whole pile length seems to be
Salinero (1983) and Roesset (1984) have shown quite realistic in many cases of practical interest.
that this is a realistic approximation, except Specifically
perhaps at very high frequencies and close pile
spacings. Other researchers have also taken (a) for axially loaded piles in the frequency range
advantage of this assumption (e.g. Nogami, 1979; covered in this Paper, waves propagate down
Sheta & Novak, 1982). Equation (2) may again the pile at a nearly infinite apparent phase
provide the clue to the success of this assumption; velocity, while the motion diminishes almost
the deformations which waves emanating from exponentially with depth; in other words, all
pile p induce at each point along the contour of points along the pile shaft move in phase at
pile q are all nearly in-phase, due to the relatively the excitation frequency. (The apparent veloc-
small dimensions of the pile (d < J/6). Therefore, ity in the pile would be essentially infinite if
the average deformation of passive pile q at a the surrounding soil provided no damping
given depth is approximately equal to the defor- and the dimensionless frequency factor cod/K
mation of the middle point, i.e. at the axis of this were less than a cut-off frequency factor,
pile. which is always beyond the frequency range
DYNAMIC STIFFNESS AND DAMPING OF FLOATING PILE GROUPS 561
(b)
Fig. 2. (a) analogy between the- cyliodrical wave assumption for group of piles in
soil aad cylindrical water waves; (b) distribution of displacement amplitudes along
the shaft of an oscillating (active) pile and of a neighbouring (passive) pile are
assumed to be of the same shape; (c) pile-head deformation and reactions during
rocking; (d) assumed apparent velocities of waves emanating from a laterally oscil-
lating pile
of interest here (i.e. greater than l).) (Wolf, DEVELOPMENT OF DYNAMIC INTERACTION
1985.) FUNCTIONS
(b) for lateral loading, waves propagate down the Axial (vertical) vibration
pile at a finite phase velocity (Wolf, 1985) so For a homogeneous deposit (as in Fig. l(a)),
that simultaneous arrival can be tried only as only two pieces of information are needed as
a first approximation. input to the method
In the interest of simplicity, the basic solution (a) the S-wave velocity V, and the hysteretic
for cylindrical waves is simplified by considering damping ratio fl of the soil.
the large-argument asymptotic approximation (b) the complex-valued dynamic impedance
(Morse & Ingard, 1968). Here, one works with ,X,(w) of the single pile at the particular fre-
simple exponentials instead of Hankel functions; quency (or frequencies) of interest.
this will allow solutions for a small number of
piles to be achieved with the help of only a pocket The axial rigidity and the slenderness of the
calculator. It has been found in sensitivity studies pile enter the analysis only through their effect on
that the improvement resulting from the use of the impedance of the single pile. Also, the
the exact cylindrical wave solution is negligibly complex-valued dynamic impedance for a single
small and would hardly rectify any potential limi- pile is defined as
tations of the method. X,(w) = Kzs(w) + ia, C,(w) = FS/wS (3)
In addition to these fundamental assumptions,
some other assumptions are made when studying and can be obtained by any of several available
the horizontal and rocking modes of vibration. numerical methods (see Sheta & Novak, 1982, for
562 DOBRY AND GAZETAS
a list of related publications) or, potentially, from This simple expression is all that is needed to
full-scale dynamic pile load tests in the field. compute the vertical response of any group of
Consider two identical piles, p and q, separated piles, once the single pile impedance Xps is avail-
by a distance S between axes. The effect of the able.
vibration of pile p on the response of pile q can
be conveniently expressed through the dynamic
interaction factor a,, which is a function of fre- Rocking
quency Rocking of piles fixed to a rigid cap induces
a, = a,(w) = wqplwqq both axial and rotational deformations (Fig. 2(c)).
Based on the evidence available from the static
additional disnlacement of solutions (Poulos & Davis, 1980) and on indirect
pile q caused by pile p evidence for the dynamic problem, it is assumed
= displacement of pile q under
(4) that no interaction takes place due to the rota-
own dynamic load tional deformation of each pile. Such deformation
(under zero lateral head displacement) is felt only
The dynamic displacement field around a a few diameters down from the pile head, and
vibrating pile p is described by the following produces a rapidly decaying stress field around
asymptotic cylindrical wave expression (see the pile. Hence the neighbouring piles fall outside
Morse & Ingard, 1968) each others zone of influence for this effect.
Therefore, although both axial and rotational
w(r) z A -
;r
exp (-pm/V,) exp
[- 91
rw t - 7
(
(5)
deformations of pile q are considered in pile q
due to its own load, only the influence of motions
coming from pile p on the axial deformations of
pile q is considered.
where r = horizontal distance from the axis of
pile p and A = A(z) = an amplitude constant. The
other three factors in equation (5) imply that
Lateral vibration
(a) the amplitude of motion decays in proportion For laterally oscillating piles the interaction
to r-II2 times the hysteretic damping depen- factor ah depends, for a given frequency w, not
dent factor exp (-/km/V,) only on the distance S but also on the angle 6
(b) the phase lag of the motion at a particular between the line of the two piles and the direction
location within the soil is only a function of of the horizontal applied force. However, it is suf-
the radial distance r. ficient to compute a,, only for 0 and 90 angles,
and then use
At the axis of the neighbouring pile q, located
at a distance r = S, equation (5) provides ah(P) N ah(Oo)cos2 8 + a,(90) sin 0 (9)
W qp E w(r = S)
to obtain very good estimates for any arbitrary
angle 0 (Poulos, 1971; Kaynia & Kausel, 1982a
= A i exp ( -/?wS) exp [iw(t - S/V,)] (6) and b; Sanchez-Salinero, 1983). Approximate esti-
JS
mates of a,,(P) and ah(90) are obtained with the
It is assumed that this wqp is also approximately help of Fig. 2(d). Based on evidence provided by
the same in the periphery of pile q. To obtain the Gazetas & Dobry (1984) the 90-pile q is affected
displacement wqq of pile q under its own dynamic essentially only by S-waves which emanate from
load, equation (5) is used once more to provide active pile p and which have a phase velocity V,.
the approximate expression However, the 0-pile q is affected by compression-
extension waves coming from p and propagating
wqq v A i exp (iwt) with an apparent phase velocity which is approx-
Jr0 imately equal to the so-called Lysmers analog
in which r,, = d/2 is the radius of the pile. Equa- velocity, V,, = 3.4 VJ[lr(l - v)] (see Dobry &
tion (7) recognizes that there is no time lag Gazetas, 1986). Thus as a first approximation
between the axis and perimeter of the pile vibrat-
ing under its own load. a,(90) 2: a, (10)
By dividing equation (5) (for r = S) by equation and
(7) we obtain the approximate expression for the
S -l/Z
0
interaction function
ah(V) = ah0 = r, exp ( - PwSlK.)
- I,2
exp (-@S/V,) exp (-ioS/K) (8)
x exp ( - id/V,,) (11)
DYNAMIC STIFFNESS AND DAMPING OF FLOATING PILE GROUPS 563
FIRST APPLICATION: AXIAL IMPEDANCE OF transmitted by each pile differ. Let F, be the
RIGIDLY-CAPPED PILE GROUPS amplitude of the load carried by the corner piles,
Referring to Fig. l(b), the following examples of F, that of the centre pile, and F, = F, that of the
application of the method are offered. four remaining edge piles (see Fig. l(b)). The dis-
placement of the centre pile 2 is
= -$
I
[l + Za,(S) + a,(SJ2)] (14) = $j Cl - a,(S)+ a,(S)- a,(S$)l
-G
= UllU + al2 + al3 + ~1
K,
4KzS 1
= 3 { 1 + cLh0(S)+ a,(S) + f
_f,G=- 4ptl
0
UG
-2 I
10
r _
1 + a,(S) + O+#J2) + ClJS) + O%,,(SJ2)
8-
(23)
6-
In similar fashion, although computationally
more involved, one can derive the lateral imped-
C,G
4 K,
ances of larger pile groups.
4-
v,
Sld
-s=s
d
-G
K,
9 KzS
old wd
v, v,
Fig. 4. Vertical dynamic stiffness and damping group factors as a function of frequency: comparison of proposed simple
method with rigorous solution of Kaynia & Kausel (1982) for a group of 3 x 3 fixed-head piles in a homogeneous
halfspace, &,/ES = 1000, L/d = 15
Fig. 3). Fig. 6 plots the distribution of forces constant. To explain the causes of the similarity,
among the piles of a 3 x 3 group as a function of refer to equation (2). At such close spacings
frequency. For vertical response, the wavelength 1 (S = 24 the relevant wavelengths II are greater
equals 2nVJw and therefore the frequency factor than S by a factor of at least 3. Hence the soil
wdJVs = 24(1/d). Figs 7 and 8 correspond to a mass between piles tends to vibrate in phase with
pile group in a soil stratum over bedrock. the piles and so the pile groupsoil system
Several important trends can be seen from Figs responds as a block.
3-8. Groups with more amply spaced piles (S/d = 5
For very close spacings (S/d = 2) the stiffness or 10) exhibit a more complicated behaviour with
and damping group factors exhibit a smooth the stiffness and damping curves having peaks
variation with frequency, with the pile group and valleys which depend on the size of the group
behaving very much like an isolated embedded and the spacing of the piles.
foundation (Gazetas, Dobry & Tassoulas, 1985). Initially, at low enough frequencies, the group
Indeed, while the damping coefficients remain stiffnesses invariably decrease with frequency, as
essentially constant in the frequency range the relevant wavelengths are again large enough
studied, the dynamic group stiffnesses decrease compared to pile spacing (1 > 3s) and the soil
steadily with increasing frequency, achieving between the piles moves in phase with the piles
negative values at higher frequency factors. Such (response as a block). However, beyond a certain
a behaviour is reminiscent of the simple l-degree limiting value of the frequency, wave interference
of freedom oscillator having mass m, spring con- phenomena start dominating the response of the
stant K and dashpot coefficient C, the effective soil-pile group system. This limiting frequency is
dynamic stiffness of which, K = K - ma=, essentially independent of group size but
decreases parabolically with frequency while C is decreases with increasing pile spacing; it is
566 DOBRY AND GAZETAS
S/d
0 10
A 5 1 Prowxed srnple method a
C,G
16 K,
I
0.5 OOk
rud
v,
Fig. 5. Vertical dynamic stiffness and damping group factors as a function of frequency: comparison of propused simple
method with rigorous solution of Kaynia 81 Kausel (1982) for a group of 4 x 4 fixed-bead piles in a homogeneous
halfspace, E,/E, = 1000, L/d = 15
roughly equal to wd/K = 0.40 for S/d = 5, and to static and low-frequency excitation the corner
cod/K = 0.15 for S/d = 10. piles carry the largest portion of the applied
When the frequency exceeds that limiting load-their force amplitude IF, 1 varies from
value, pile-t-pile interaction leads to stiffness about 1.1 to 1.5 times the average applied force
group factors that may far exceed both unity and as the spacing decreases from 10 to 2 diameters.
the single-pile factor. The predominant peaks By contrast, at the same low frequencies the
occur in each case at the frequency at which the central pile carries only 0.7 to 0.15 times the
corresponding wavelength 1 is about two times average load, respectively. However, this picture
the pile spacing S. The explanation is straightfor- changes at higher frequencies as cylindrical waves
ward: when S = I/2/2 or, in general, S = I(* + n) emanating from other piles affect the response,
with n = 0, 1, 2 . . . . cylindrical waves originating producing undulations in the individual pile-load
with a certain phase from a pile p arrive at the curves. The central pile is particularly sensitive to
neighbouring pile q in exactly opposite phase, such wave interferences, since the waves emitted
thereby inducing a displacement wqP which is by the surrounding eight piles hit this pile with
negative compared to the displacement wqq due similar phases, due to the identical distances, S
to this piles own load. Hence, the dynamic inter- and SJ2, of the four edge and four corner piles
action factor a, is negative (equation 4) and a from the centre pile. Because of this, their effects
larger force must be applied onto pile q to enforce (favourable or unfavourable) add up (constructive
a certain displacement amplitude. or destructive interference). At wavelengths I in
The distribution of forces among the piles of the range
the 3 x 3 group, shown in Fig. 6, is very sensitive
to variations in frequency and pile spacing. For 2S6152J2S (24)
DYNAMIC STIFFNESS AND DAMPING OF FLOATING PILE GROUPS 567
0 0 0
a 0 a
Pile 1 (corner)
PIIC! 2 (centre)
Fig. 6. Variation of the distribution of axial force amplitudes carried by the corner and
centre piles in a 3 x 3 fixed-head pile group, as a function of vibration frequency and pile
spacing; EJE, = 1000, L/d = 15, vertical loading: comparison of proposed simple method
with rigorous solution of Kaynia & Kausel(l982)
these waves tend to lift the centre pile upward corner piles drops down to nearly 60% of the
when its own load pushes downward; as a conse- average load-a rather dramatic reversal of the
quence, only by increasing its share of the total static situation.
applied load can this pile be forced to follow the In Figs 3-8 the predictions of the proposed
uniform displacement of the rigid cap. It can be simple analytical method compare extremely well
seen in Fig. 6 that in the frequency ranges corre- with the rigorous numerical solutions; qualitat-
sponding to equation (24), 1F, 1of the centre pile ively as well as quantitatively. In particular, even
may rise to nearly 2 times the average load detailed trends arising from pile-soil-pile inter-
(regardless of pile spacing) while the share of the action were successfully predicted. The fact that
[i = 0.025
Lid = 37.5
H/L = 2
6-
1.5
4 h
1.0
_G
K,
C,G 4-
2 K, 2 K,
05
2-
S/d
x E Nogamls apprownat~on
(1963)
-- 2 I
0 10
A 5 \ Proposed sample method
Bedrock 0 2)
, 1
0.5 1 00 0.5 1
cod cod
v, v,
Fig. 7. Vertical dynamic stiffness and damping group factors as a function of frequency: comparison of proposed simple
method with Nogamis (1983) approximate solution for a group of two slender fixed-head piles in a homogeneous
stratum, I&/E, = 00
2 Lid = 37.5
H/L = 2
S/d = 5
I = 0.4
/i = O-025
1c
Bedrock
1
-
C,G
K, z Nogm (1963)
4 K,
KZG
l Proposed simple
0 1 method
1 I 1 t
(
0.5 1 0.5 1
,r,d wd
VS v,
Fig. 8. Vertical stiffness and damping for a group of 2 x 2 slender fixed-head piles in a stratum: the proposed method is
in accord with Nognmis approximation, although with infinitely rigid piles the latter solution leads to larger resonant
peaks
DYNAMIC STIFFNESS AND DAMPING OF FLOATING PILE GROUPS 569
c, 3_ ---__
s/a
- 10
Rigorous solutfon
of
-5
Kayma - Kausel (1982)
-- 2
0 10
a 5 Proposed ample method
0 2
Fig. 10. Rocking stiffness and damping group factors as a function of fre-
quency: comparison of proposed simple method with the rigorous solution of
Kaynia 81 Kausel (1982) for a group of 3 x 3 fixed-head piles in a homoge-
neous halfspace, E&E, = 1000, L/d = 15
in the group grows, is probably due to the simpli- modifies the wave interference phenomena con-
fied wave field assumed in Fig. 2(d) and to the trolling the pile group response in ways not fully
finite apparent phase velocity with which flexural accounted for by the simple model.
waves propagate down the laterally loaded pile. Nonetheless, even in this case the simple
This renders the assumption of simultaneous method proposed predicts the distribution of
emission of the waves from the various points shear forces at the head of each pile in the group
along the pile shaft somewhat inaccurate for hori- with remarkable accuracy for all frequencies, pile
zontal fixed-head pile vibration. Hence, simulta- spacings and group sizes considered (see Fig. 13).
neous arrival of these waves at the shaft of a (Although Fig. 13 presents only some of the com-
neighbouring pile does not reflect reality in an parisons of lateral force distributions obtained by
accurate way. As a result, the arriving waves have the Authors, the excellent agreement shown is
(slightly) different phases along this pile, and this typical of all results.) The trends in Fig. 13 are
DYNAMIC STIFFNESS AND DAMPING OF FLOATING PILE GROUPS 571
- 5) Rigorous
SOlUtlOn of
_ 2 j Kaym - Kausel (1982)
wd
v,.
Fig. 11. Lateral (swaying) stiffness and damping group factors as a function of
frequency: comparison of proposed simple method with rigorous solution of
Kaynia & Kausel (1982) for a group of 2 x 2 fixed-head piles in a homoge-
neous balfspace, E,/E, = 1000, L/d = 15
qualitatively similar to those of Fig. 6 for axial pile groups. The method requires as input the
vibration. related dynamic impedance of the single pile and
three soil parameters: V, , v and /I.
For a wide range of pile separation distances,
CONCLUSION frequencies of oscillation, and number of piles in
A very simple method which captures the the group, the results of the method are in excel-
essential physics of the problem has been devel- lent accord with those of far more sophisticated
oped for estimating the dynamic impedance func- solutions. Potential limitations of the procedure
tions of pile groups. The approach leads to simple are also pointed out and ways of overcoming
analytical expressions, and it can be easily under- them are suggested.
stood and applied by engineers familiar with Even though the method has been developed
using static interaction factors in the design of and applied in this Paper only to the case of a
DOBRY AND GAZETAS
S/d
- 10
*I
Rigorous soluton of
- 5 Kaynla - Kausel (1982 a b)
-
-G
-!L
9 KxS
0
3
c,
9 K,
I I
0.5 1 3
wd cud
v, v,
Fig. 12. Lateral (swaying) stilfness and damping group factors as a function of frequency: comparison of proposed
simple method with rigorous solution of Kaynia & Kausel(l982) for a group of 3 x 3 fixed-head piles in a homogeneous
halfspace, E,/E, = 1000, L/d = 15
homogeneous linear soil layer or halfspace, it From Fig. 3, for a, = cod/V, = 0.3, the single pile values
could in principle be extended to handle in are
approximate but physically sound ways, both
4&S/4K,S = ~,SjK,S = 1.16 (27)
inhomogeneous deposits and pile installation
effects. It is anticipated that such material heter- 4Czs/4KzS = CzS/KzS = 3.0 (28)
ogeneities and nonlinearities would tend to
reduce the substantial interference peaks noticed i.e.
in the stiffness and damping curves of pile groups
Xzs = 1,16K,s + (3Kzs)(0.3)i (29)
embedded in homogeneous, linear soil deposits.
X, = (1.16 + 0.9i)K,s (30)
XG=KRSiia
I z 0 CsI (26) OS/V, = @d/V&?/d) = (10x0.3) = 3. (33)
DYNAMIC STIFFNESS AND DAMPING OF FLOATING PILE GROUPS 513
Pile 1 (corner)
r Pile 2 (centre)
Fig. 13. Variation of the distribution of shear force amplitudes carried by the corner
and centre piles in a 3 x 3 fixed-head pile group, as a function of lateral vibration
freqwncy and pile spacing, EP/E, 3: 1000, L/d = 15, vertical loading: comparison of
proposed simple method with ngoroos solution of Kayaia & Kausel(l982)
which would provide an additional data point for Novak, M. (1984). Evaluation of dynamic experiments
S/d = 10 and wd/< = 0.3 in both plots of Fig. 3. on a pile group. J. Geotech. Engng Div. Am. Sot. Cio.
Engrs 110,738-756.
Ottaviani, M. (1975). Three-dimensional finite element
REFERENCES analysis of vertically loaded pile groups. Ghotech-
Banerjee, P. K. (1978). Analysis of axially and laterally nique 25, No. 2, 159-174.
loaded uile grouts. Deuelonments in soil mechanics. Poulos, H. G. (1968). Analysis of the settlement of pile
(ed. C. k. %ottj London:. Applied Science Publi- groups. Gt?oiechnique l& No. 4,449-471.
shers. Poulos, H. G. (1971). Behavior of laterally-loaded piles
Butterfield. R. & Baneriee,
_ P. K. (1971). The elastic II: pile groups. J. Soil Mech. Fdns Div. Am. Sot. Ciu.
analysis of compressible piles and pile groups. GCo- En&s 91, Sti5,733-751.
technioue 21. No. 1.43-60. Poulos. H. G. & Mattes, N. S. (1971). Settlement and
Davies, ?. G., Sen, R. & Banerjee, P. K. (1985). load distribution analysis oi pile groups. Austral.
Dynamic behavior of pile groups in inhomogeneous Geomech. J. Gl, No. 1, 18-28.
soil. J. Geotech. Engng Div. Am. Sot. Cio. Engrs 111, Poulos, H. G. & Davis, E. H. (1980). Pile foundation
13651379. analysis and design. New York: Wiley.
Dobry, R. & Gazetas, G. (1986). Dynamic response of Randolph, M. F. & Wroth, C. P. (1978). Analysis of
arbitrarily shaped foundations. J. Geotech. Engng deformation of vertically loaded piles. J. Geotech.
Div. Am. sot. &. Engrs 112, 109-135. Engng Div. Am. Sot. Cio.~Engrs 104~1465-1488.
El Shamouby, B. & Novak, M. (1985). Static and low- Randolph, M. F. & Wroth, C. P. (1979). An analysis of
frequency response of pile groups. Can. Geotech. J. the vertical deformation of pile groups. Gtotechnique
22,79-94. 29, No. 4,423439.
Fleming, W. G. K., Weltman, A. J., Randolph, M. F. & Randolph, M. F. (1981). The response of flexible piles to
Elson, W. K. (1985). Piling engineering, pp. 188-198. lateral loading. Gtotechnique 31, No. 2,247-259.
Glasgow: Surrey University Press. Roesset, J. M. (1984). Dynamic stiffness of pile groups.
Gazetas. G. & Dobrv. R. (1984). Horizontal response of Pile foundations. New York: ASCE.
pilesin layered s&ls. J: Geotech. Engng Div. Am. Sot. Sanchez-Salinero, I. (1983). Dynamic stiffness of pile
Ciu. Engrs 110, 2&40. groups: approximate solutions. Geotechnical Engin-
Gazetas. G., Dobrv, R. & Tassoulas, J. L. (1985). Verti- eering Report GR83-5, University of Texas at
cal response of-arbitrarily shaped embedded founda- Austin.
tions. J. Geotech. Engng Div. Am. Sot. Ciu. Engrs Scott, R. F. (1981). Foundation Analysis. Englewood
111,75&770. Cliffs: Prentice-Hall.
Kagawa, T. (1983). Dynamic lateral pile-group effects. J. Sen, R., Davies, T. G. & Banerjee, P. K. (1985).
Geotech. Emma Div. Am. Sot. Cio. Enars 109. 1267- Dynamic analysis of piles and pile groups embedded
1285. - _ in- homogenedus so&. Earthquake engineering and
Kaynia, A. M. & Kausel, E. (1982a). Dynamic behav- struct. Dyn. 13, 53-65.
iour of pile groups. 2nd Inter. Co& Numer. Meth. Sheta, M. & Novak, M. (1982). Vertical vibration of pile
Offshore Piling, Austin, Texas, 509-532. groups. J. Geotech. Engng Div. Am. Sot. Ciu. Engrs
Kaynia, A. M. & Kausel, E. (1982b). Dynamic stiflness 108,57&590.
and seismic response of pile groups. Research Report Tyson, T. R. & Kausel, E. (1983). Dynamic analysis of
R82-03, Massachusetts Institute of Technology. axisymmetric pile groups. Research Report R8347,
Morse, P. M. & Ingard, K. U. (1968). Theoretical ocous- Massachusetts Institute of Technology.
tics. New York: McGraw-Hill. Varadan, V. K., Varadan, V. V. & Pao, Y.-H. (1978).
Naylor, D. J. & Hooper, J. A. (1975). An effective stress Multiple scattering of elastic waves by cylinders of
finite element analysis to predict the short and long- arbitrary cross-section I. J. Acoust. Sot. America. 63,
term behaviour of a pile-raft foundation on London 131&1319.
clay. Symp. Settlement of Structures, Cambridge. Waas, G. & Hartmann, H. G. (1981). Analysis of pile
Nogami, T. (1979). Dynamic group effect of multiple foundations under dynamic loads. Co@ on Struct.
piles under vertical vibration. Proc. Engng Mech. Mech. in Reactor Technol., Paris.
Div. ASCE Specialty Con& Austin, Texas, 75&754. Waas, G. & Hartmann, H. G. (1984). Seismic analysis of
Nogami, T. (1980). Dynamic stiffness and damping of pile foundations including soil-pile-soil interaction.
pile groups in inhomogeneous soil. Dynamic 8th World Conf: Earthauake Enana 5.55-62.
response of pile foundation. (eds M. ONeill and R. Wolf, J. P. & Van Arx, 6. A. (19?8j. Impedance func-
Dobry). ASCE Special Technical Publication. New tion of a group of vertical piles. Proc. ASCE Spe-
York: ASCE. cialty Conf Soil Dyn. Earthquake Engng 2,
Nogami, T. (1983). Dynamic group effect in axial 1024-1041.
responses of grouped piles. J. Geotech. Engng Dio. Wolf, J. P. (1985). Dynamic soil-structure interaction, pp.
Am. Sot. Cio. Engrs 109,225-243. 166172. Englewood Cliffs: Prentice-Hall.
Novak, M. (1974). Dynamic stiffness and damping of
piles. Can. Geotech. J. 11, No. 4.