Dynamic experiments with small pile foundations

MILOSNOVAKA N D RONALDF. GRIGG

Can. Geotech. J. Downloaded from www.nrcresearchpress.com by UNIVERSITY OF ARIZONA LIBRARY on 09/06/13

Frtculty of Engineering Science, University of Western Ontario, London, Ont., Crtnndrc N6A 3K7
Received November 12, 1975
Accepted June 14, 1976

Dynamic experiments with small pile foundations were conducted in the field. Individual piles
and a group of piles were tested. The experimental data were compared with theoretical predic-
tions made on the basis of a theory described previously by Novak. It was found that the theory
predicts the general characteristics of the response very well. Resonant amplitudes and natural
frequencies can be predicted reasonably well when a shear modulus of the soil derived from a
static test of a pile is used and when a correction for the effect of pile group (interaction) is made.

Des essais dynamiques sur des fondations de pieux modeles ont CtC rCalisCs sur le terrain. Des
pieux isolCs et un groupe de pieux ont Cte soumis B essais. Les resultats experimentaux ont it15
compares aux previsions theoriques faites a partir d'une thiorie proposCe par Novak. On a
constate que la thCorie permet une trbs bonne prevision des caractkristiques gCnerales de la
rkponse. Les amplitudes de rksonnance et les frequences naturelles peuvent ktre prevues
raisonnablement bien lorsque le module de cisaillement du sol diduit de i'essai statique d'un pieu
est utilisi et lorsqu'on applique une correction pour tenir compte des effets de groupe.
[ ~ r a d u i t - p alar revue]
Can. Geotech. J., 13,372 (1976)
For personal use only.

Introduction comparison with the theory are described in

Foundations and structures supported by this paper.
piles are often exposed to dynamic loads such The experiments were conducted with pile
as those produced by wind, earthquakes, or foundations that were small enough to be in-
machines. The response of the structures to expensive, but large enough to be acceptably
these loads depends, to a high degree, on the representative of real foundations such as, e.g.
stiffness and damping that the piles can pro- machine foundations. The goal was to find out
vide. These characteristics of the pile are a if the theory could predict the behaviour of the
product of the interaction between the soil and pile foundations used in the experiments.
the pile. A theory was presented recently (No- Thus, there was no reason to make the test
vak 1974a, 1976) that makes it possible to foundations physically similar to any particular
evaluate the pile stiffness and damping for any prototype. This would be very difficult to
particular situation. The main assumptions of achieve anyway because the stiffness of soil
this theory are as follows: the soil is perfectly depends on confining pressure.
bonded to the pile, the soil behaves in a linear Particular attention was paid to harmonic
manner, and the soil properties are invariable excitation which represents a fundamental
with depth. Furthermore, only horizontally case. The stationary response to such a load
propagating waves were considered; this is defines the impedancc functions as well as the
equivalent to the assumption of plane strain or stiffness and damping constants of the soil-pile
to the assumption of infinitesimally thin layers. system. These constants are useful in design of
The latter assumption was found to yield very pile-supported footings and structures.
good results for slender piles and higher fre- The vibrations excited were small and this
quencies (Nogami and Novak 1976). The was sufficient because the theory to be ex-
validity of the theory was also confirmed by a amined is linear. Nonlinearity of the response
finite element solution by Kuhlemeyer ( 1976). is known to appear with large displacements.
Nevertheless, the assumptions made may devi- However, the virtues of the linear theory are
ate from reality and so it is desirable to verify that it offers an insight into the problem, it can
experimentally the applicability of the theory. be quite adequate for machine foundations,
A series of dynamic field experiments and their frozen soils, and other cases and finally, it may
 NOVAK AND GRIGG 373

be adjusted for large amplitudes by means of depends on the slenderness ratio l/ro, where
equivalent linearization. 1 = length of the pile and yo = radius of the
pile, and the stiffness of the soil. The condi-
Experiments tions of the site suggested a maximum pile
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by UNIVERSITY OF ARIZONA LIBRARY on 09/06/13

The Site and Soil Properties diameter of about 4.5 in. (1 1.5 cm). With
A suitable site was found on the campus of respect to the ease of installation standard
the University of Western Ontario, close to black steel pipes were chosen with diameters
the Thames River. It is a layer of fine silty of 3.5 and 2.4 in. (9 and 6 cm). The following
sand approximately 6 to 7 ft (1.8 to 2.1 m) test situations were examined: ( a ) a single
thick underlain by gravel and till. 3.5 in. (9 cm) diameter pile, ( b ) a single 2.4
The soil properties were determined by in in. ( 6 cm) diameter pile, (c) a group of four
situ and laboratory tests. The in sifir density 2.4 in. (6 cm) diameter piles, and ( d ) a group
was determined by means of the Rubber Bal- of four 2.4 in. (6 cm) diameter piles with soil
loon Method in an excavation. Verifications packed beneath the pile cap. The properties of
of the soil profile and soil samples were ob- the piles used are given in Table 2.
tained through the use of a hand augered test The piles were hydraulically driven into the
hole. soil until they reached the solid layer and by
The shear wave velocity of the soil was dropping a 140 Ib (62 kg) hammer 30 in. (76
established by means of the steady state vibra- cm). The piles used in the group were cut
tion technique. The ground motion was excited short and did not reach the solid layer.
by the piles and the results were clearer and T o examine the effect of the pile length,
more consistent than is usual with soil alone. several stiffness and damping parameters were
For personal use only.

The depth of the hard layer was evident from calculated from formulas given in Novak
the variation of the shear wave velocity with (1974a) as a function of slenderness. These
frequency and was also verified by a seismic parameters are shown in Fig. 1. It can be seen
test. The soil properties useful in dynamic that for the slenderness used, the stiffness and
analysis are givcn in Table 1. More details on
the shear wave velocities measured are given in
the Appendix.
Test Piles
The size of the piles used was limited,
among other considerations, by the force and
frequency range of the oscillator (exciter)
available and by the requirement that thc mo-
tion of the pile tips be kept as small as pos-
sible. The satisfaction of the latter condition is
facilitated by the theoretical finding that it is
only the upper part of the pile that actively
participates in the motion and determines the
pile stiffness and damping. This active length

,I 1, 1 ,
TABLE
1. Properties of soil (fine silty sand)

Mean bulk density

Mean dry bulk density
Mean moisture content
= 111.91b/ft3
= 101.7 1b/ft3
= 10.0%
1,L' , , ,
Coefficient of energy absorption = 0.14 OO 20 40
I/r,
60 80
PILE S L E N D E R N E S S
Shear wave velocity from steady
state vibration technique V, = 573 ft/s
Shear wave velocity from static FIG.1. Stiffness and damping parameters of hollow
steel test piles as functions of slenderness ratio.
test of pile, 2.4 in. diameter V,, = 147 ft/s
Shear wave velocity from static (f7.1-2 = rotational stiffness and damping, fo,i-s= cross
test of pile, 3.5 in. diameter V,, = 108 ft/s
stiffness and damping, fir,,-:, = vertical stiffness and
damping; no = 0.3).
 CAN. GEOTECH. J. VOL. 13, 1976

TABLE
2. Properties of piles

Diameter Area I Length l

Test (in.) (ft2) (ft4) (in.) [/yo
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by UNIVERSITY OF ARIZONA LIBRARY on 09/06/13

Single pile 3.5 0.0155 0.000147 88.5 50.6

Single pile 2.4 0.0074 0.000032 92.5 77.9
Group of four
(each pile) 2.4 0.0074 0.000032 82.0 69.1

damping parameters are largely independent of

the pile length with the exception of the verti-
cal direction.
In order to connect the pile head to the
steel test body, heavy steel flanges were welded
to the piles about 2-3 in. (5-7.6 cm) above
ground level. The group of piles was provided
with a rigid reinforced concrete cap to achieve
full group action.
Test Body
In order to get well pronounced resonant
peaks in the frequency range of the exciting
For personal use only.

mechanism a steel test body was bolted to the

pile flanges or the pile cap (Fig. 2 ) . The test
body was comprised of steel plates each weigh-
ing 125 Ib (57 kg). The number of the plates
was varied from 6 to 20. The steel plates had
machined contact surfaces to prevent slippage
between the plates. (The test body was origin-
ally designed by the first author and W. Lane
for dynamic tests of soils.)
As can be seen from Fig. 2, the body is
provided with oscillator mounts and pickup
mounts. The properties of the complete test
bodies actually representing rigid footings, are FIG.2(a) Test body bolted to single pile (vertical
given in Table 3. excitation). ( b ) Steel and concrete test body bolted
to group of piles (horizontal excitation).
Driving Mechanism
A vertical or horizontal excitation force was
IRD Model 544 electrodynamic pickups, a
produced by means of a Lazan mechanical
vibration meter Model 306 IRD, a dual beam
oscillator. Its excitation force is proportional
storage oscilloscope, a two channel Brush re-
to the square of the frequency and the maxi-
corder, and a stroboscope for accurate mea-
mum eccentric moment is 18 Ib-in. (0.21 kg-m).
surement of speed. The measured amplitudes
The oscillator was driven by a 220 V three
were corrected according to the frequency
phase motor equipped with a Kopp variator.
characteristics of the pickups and the calibra-
The variator yields stable operating speeds of
tion factors for the vibration meter.
380 to 3750 rev/min (6.3 to 62.5 Hz). The
For vertical excitation, displacements were
variator is connected to the oscillator via a
measured using two pickups mounted sym-
flexible shaft.
metrically. Horizontal displacements were mea-
Measuring Apparatus sured using two pickups, one on each side of
The measuring apparatus consisted of three the test body, at the height of its center of
 NOVAK AND GRIGG 375

TABLE
3. Properties of test footings

Height of Height of
Mass moment
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by UNIVERSITY OF ARIZONA LIBRARY on 09/06/13

Mass moment centroid excitation

of inertia of inertia above base above C.G.
Mass I*., =c
Footing (slugs) (slug-ft2) (ft)
6 plates 0.610
10 plates 0.742
20 plates 1.O25
20 plates
and concrete
pile cap 1.168

gravity. The rocking component of the motion

was established from vertical displacements
measured on the axis of the test body.
Test Procedures
Steady-state response to harmonic excita-
tion was measured after a suitable size of the
test body was found. The amplitudes were
For personal use only.

measured at different frequencies, always after

the motion became stationary, and with a few
excitation intensities. In addition, free vibration
tests were conducted.
Also, static load tests were conducted on
each of the two single piles after the dynamic
tests had been completed. A static horizontal
load was applied to the pile head and the
corresponding displacement measured.
Test Results
Static Tests
Curves of the horizontal load vs. deflection ( J
for both piles tested are shown in Fig. 3. The 0 400 800 1200

two different curves measured for the 2.4 in. FORCE ( I b )

(6 cm) diameter pile were established in two FIG.3. Static displacement of pile head vs. horizon-
directions differing by 90". tal load.
From these static tests an apparent shear
modulus of the soil can be established by com- static test is significantly lower than that estab-
paring the measured displacements with those lished from the wave propagation test. In other
predicted theoretically. Such a comparison is words, the theory significantly under-estimates
facilitated by fair linearity of the relationships the static response if the shear modulus estab-
shown in Fig. 3 for displacements smaller than lished from wave propagation is used. This
about 0.1 in. (2.5 mm). difference seems to be due to two main factors
Using Fig. 3 of the theoretical static solution which are the separation of the pile from the
in Poulos (1972) and assuming Poisson's ratio soil in the region of tension and the diminishing
v = 0.25, an iterative procedure yields the of the shear modulus with decreasing depth.
apparent shear modulus and wave velocity Both of these factors, indicated in Fig. 4,
given in Table 1 and denoted Vst. depend on confining pressure which is very
The shear wave velocity obtained from the small in the upper region of the soil layer. It
 CAN. GEOTECH J. VOL. 13, 1976

The response curves are shown either in dimen-

sional or dimensionless form. The dimension-
less amplitudes for vertical and horizontal
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by UNIVERSITY OF ARIZONA LIBRARY on 09/06/13

vibrations are

respectively, and for rotation (rocking) around

the centre of gravity (CG)

in which w o = vertical amplitude due to verti-

FIG. 4. Separation of pile and variation of shear cal excitation, uo, I),) = horizontal and rotation
nlodulus with depth. components of the coupled motion due to
horizontal excitation, m, I$ = mass and mo-
is the top layer, which is only a few pile ment of inertia of the footing about CG, m,,
diameters thick, that contributes most to the e = unbalanced mass and its eccentricity re-
horizontal stiffness of the pile. It is not sur- spectively. The true amplitudes did not exceed
prising therefore that the apparent static stiff- 0.025 in. (0.5 mm) and thus were of the order
ness of the soil, which actually reflects the common in machine foundations. (All ampli-
stiffness of the uppermost layer, is much tudes given in this paper are single amplitudes.)
For personal use only.

smaller than that obtained from wave propaga- If the vibrating system were perfectly linear
tion which reaches the deeper regions of soil. all the dimensionless curves shown in each
The difference between V , and V,, would figure should be identical. The amount by
probably be much smaller if V,,. were deter- which they differ, indicates the degree of non-
mined for the surface layer of the soil or if the linearity. The nonlinearity is demonstrated by
confining pressure were greater as it is in a real shifting of the resonant frequency and variation
pile foundation. in the resonant amplitudes. The nonlinearity,
The effect of pile separation indicated in though distinct, is not very strong. It could be
Fig. 4a, depends on the magnitude of the analyzed using a technique described by Novak
applied load and manifests itself by the non- (1971 ). However, for the purpose of this
linearity of the response. However, no strong study it is sufficient to use average natural
nonlinearity was observed in the experiments frequencies and resonant amplitudes. These
and this seems to indicate that the reduction are listed in Table 4.
of the shear modulus towards the surface was In this table, subscript 0 refers to the natural
a more important factor than the separation of frequency for vertical vibration, subscripts 1
the pile. In any case, it may be assumed that and 2 refer to the first or second resonance
the shear modulus obtained from the static test with horizontal excitation. The magnitudes not
incorporates in an approximate way the re- given either lay beyond the frequency range
duced stiffness of the surface layer of the soil of the equipment or were not obtained reliably
and the lack of bond between the soil and the because of malfunctioning of the oscillator.
pile. The group of four piles was tested for the
horizontal excitation in two directions. The
Dynamic Tests force acted in the direction of either the shorter
A few typical sets of response curves mea-
or the larger distance x , between the piles. An
sured with vertical or horizontal excitation are
example of the horizontal response of the pile
plotted in Figs. 5 to 8. Each figure shows a
group is shown in Fig. 7. As expected, mar-
set of response curves obtained with different
kedly different response was observed for the
intensity of excitation (excitation moment
two horizontal directions tested. The difference
m,e)l under otherwise unchanged conditions.
in resonant frequencies f l and f 2 can be seen
'The excitation moment is given in the figures as from comparison of the cases 3 and 4 in Table
m,ge, lb-in. 4. The vertical amplitude is not shown because
 NOVAK AND GRIGG
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by UNIVERSITY OF ARIZONA LIBRARY on 09/06/13

FREQUENCY (CPS)

A
SYMBOL LB-IN.
For personal use only.

FREQUENCY (CPS)
FIG.5. Experimental vertical response of vertically loaded pile ( 1 pile 3.5 in. diameter, mass
of footing = 86.41 slugs). ( a ) dimensional, ( b ) dimensionless.

the resonant peak lay above the frequency in Novak (1974a, 1976) with footing, pile and
range of the oscillator and could not be soil data corresponding to the experiments and
reached. given above. For horizontal excitation two
Finally, the group of four piles was tested shear wave velocities were used: VTV,derived
with cohesive soil compacted into the space from wave propagation, and Vst, obtained from
beneath the pile cap so that full contact was the static tests of the pile.
secured between the test footing and the soil. For vertical excitation two tip conditions
The response curves measured under these were assumed: A fixed tip and a relaxed tip.
otherwise identical conditions are shown in The degree of the relaxation of the tip can be
Fig. 8. The resonant frequencies measured are described by the shear wave velocity of the
given in Table 4. (The excitation load is not soil layer lying below the tip, V,. In addition
given in Fig. 8 because it consisted of both to the relaxation of the tip a reduction of the
forces and moments in this case). shear wave velocity appeared desirable but this
reduction was much smaller than that needed
Comparison of Theory With Experiments for the horizontal excitation. Good results were
Theoretical Response obtained with V, = 2 to 3 V , and V, = 1 to
Theoretical response curves were computed 2 v,.
by means of the general expressions derived An example of the theoretical response to
 CAN. GEOTECH. J. VOL. 13, 1976

0 40 SYMBOL LB-IN.

fl>
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by UNIVERSITY OF ARIZONA LIBRARY on 09/06/13

3
a
30
W
0
3
k
_I
n
2 20
(0
ln
W
_I
Z
0
y lo
W
E
0

0
0 10 20 30 40 50 60
FREOUENCY (CPS)

SYMBOL LB-IN
For personal use only.

o 0.855
A 0.598
0 0.427
o 0.171

FREQUENCY (CPS)

FIG.6. Experimental response of horizontaIIy loaded pile (1 pile 2.4 in. diameter, mass of
footing = 3 1.06 slugs). ( a ) horizontal component (sliding), ( b ) rocking component.

vertical as well as horizontal loads is plotted ing constants of individual piles assumed to
in Fig. 9. Comparison of the individual the- act in isolation. However, if the distance be-
oretical and experimental response curves is tween the piles is small, the stiffness and damp-
shown in Figs. 10 to 12. The theoretical nat- ing of each pile in a group are reduced because
ural frequencies and resonant amplitudes are the displacement of one pile contributes also
compared with the experimenta1 values in to the displacements of the other piles. No
Table 4. dynamic solution to this interaction of piles is
available at present but some insight can be
Eflect of Pile Group obtained from the approximate static solution
The total stiffness and damping constants of Poulos (1971). The static solution may
of the pile group used in the analysis were be used for guidance because the behaviour of
established as a sum of the stiffness and damp- a pile group should be similar in both static
 NOVAK AND GRIGG

fl SYMBOL L8-IN
2.558
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by UNIVERSITY OF ARIZONA LIBRARY on 09/06/13

0 10 20 30 40 50 60
FREQUENCY (CPS)
For personal use only.

FREQUENCY (CPS )

FIG.7. Experimental response of pile group to horizontal load (4 piles 2.4 in. diameter, mass
of footing = 103.3 slugs). ( a ) horizontal component (sliding), (b) rocking component.

and dynamic cases. The reason for this sim- where a , = the interaction factor describing the
ilarity is that damping is the imaginary part of contribution of the rth pile to the displacement
the complex dynamic stiffness and the dimen- of the reference pile; al = 1 and other factors
sionless frequency a, = r, w / V , , controlling the are smaller than unity and decrease with the
dynamic response, is very low. distance between the piles.
The relationship between the displacement Using [3] the effect of pile interaction upon
u~ of a group of n piles and the displacement the response of the pile group as shown in
ul of a single pile carrying the same load as a Fig. 11 can be examined. The first resonance
pile in the group, can be written as peak, which is of primary importance, is dom-
inated by the first vibration mode and consists
mainly of horizontal translation as can be seen
from Figs. 7a and 7b. Interaction factors CY, for
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by UNIVERSITY OF ARIZONA LIBRARY on 09/06/13

TABLE
4. Natural frequencies and resonant amplitudes

Dimensionless amplitude
Natural frequency
Piles Footing C P (Hz)
~ Vert Horizontal Rocking
diameter m
Case (in) 4 Values fo fi fi Aw A, 1 Auz A* I Aw
For personal use only.

Experimental
Theoretical V,
1 1 pile Theoretical V,,
2.4 Experimental
Theoretical V,
Theoretical V,,
Experimental 4.36
Theoretical V, 8.17
1 pile Theoretical V,, 4 . OO*
2 3.5 Experimental -
Theoretical V , 4.92
Theoretical V,, 9.20
4 piles Experimental
3 2.4 Theoretical V .
x, = 9 Theoretical V,,
4 piles Experimental
4 2.4 Theoretical V,"
x, = 13.5 Theoretical V,,
4 piles
2.4 Experimental
5 x, = 13.5 Theoretical V,"
+ soil Theoretical V,,
*Calculated with relaxed tip and V,, < V, < V,.
 NOVAK A N D GRIGG 38 1
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by UNIVERSITY OF ARIZONA LIBRARY on 09/06/13

where k, f l = the stiffness and the first natural

frequency respectively, calculated without the
pile interaction effect, i.e. with C a,. = 1. Equa-
tion 4 gives the adjusted frequency

f" =m=
14'7 12.3 cps

Similarly, the adjusted damping ratio in the

first mode is

in which CG, c = adjusted and unadjusted

damping coefficients respectively and D = the
400 600 800 1000 1200 1400 1600
unadjusted damping ratio.
FREQUENCY ( R P M I Then, the adjusted dimensionless resonant
For personal use only.

FIG. 8. Influence of soil beneath pile cap upon amplitude of sliding is approximately
experimental response of pile group ( 4 piles 2.4 in.
diameter, mass of footing = 103.3 slugs).

3
0
'Z
26
I O ~

SLIDING
1 \,VERTICAL
where A = the first resonant amplitude ob-
tained without the interaction effect. Equation
6 yields
AG = 9.13 J145= 11.0
2 The first resonant peak approximately adjusted
for the effect of the pile interaction is shown
in dotted lines in Fig. 1 1.
In other vibration modes or with free-head
piles the interaction effect can be expected to
be less pronounced.
The above approach based on summation of
FREQUENCY (CPS) a,. is approximate because it implies that all

FIG.9. Theoretical response of pile group to verti- piles carry the same load.
cal load and horizontal load ( 4 piles 2.4 in. diameter, Discussion
mass of footing = 103.3 slugs, V , = V , , = 147 ft/s,
fixed tip). A number of observations can be made on
the basis of the comparison of the theoretical
sliding fixed-head piles can be obtained from with the experimental results.
Fig. 3 in Poulos (1971). For the conditions of It can be seen that the theory predicts the
the experiment a value of a,. = 1.45 was general character of the response very well.
found using shear wave velocity V,,. The piled foundations respond to changes in
Stiffness of the pile group kG is inversely stiffness, number, and distance between piles
proportional to displacement (flexibility) and and in the footing mass in the manner that the
hence k , = k/C a,.. Therefore, the adjusted theory predicts. In particular, the following
natural frequency of the group is approx- trends predicted theoretically agree with the
imately experiments:
 CAN. GEOTECH. J. VOL. 13, 1976
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by UNIVERSITY OF ARIZONA LIBRARY on 09/06/13
For personal use only.

FIG.10. Comparison of theoretical and experimental response ( 1 pile 2.4 in. diameter, mass
of footing = 31.06 slugs; horizontal excitation). (a) sliding component, ( b ) rocking component.

For the single piles subjected to coupled and this in turn increases the natural frequen-
motion, the first resonant peak is larger than cies and reduces the resonant amplitudes.
the second for both sliding and rocking com- The quantitative comparison of the theoret-
ponents. ical predictions with experimental data can be
For the pile group subjected to coupled summarized as follows.
motion, the first resonant peak is dominated by The theory significantly over-estimates the
sliding and the second peak by rocking. The natural frequencies and, in some cases, also the
larger the distance from the center line of the resonant amplitudes if the shear wave velocity
footing to the pile, the greater are the natural derived from wave propagation in deeper
frequencies and the smaller the amplitudes. layers of soil is used. This over-estimation is
This implies that the vertical stiffness should primarily due to the reduction of the shear
not be neglected when considering horizontal modulus with decreasing depth. This reduction
loads as is often done. of the shear modulus can be accounted for,
The effect of soil beneath the pile cap is to approximately, by using the apparent shear
increase both the stiffness and the damping modulus derived from the static test in the
 NOVAK AND GRIGG
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by UNIVERSITY OF ARIZONA LIBRARY on 09/06/13

0 j I
0 1",?.3 20 30 40 50 60
FREOUENCY ICPSI

FIG.11. Comparison of theoretical and experimental response (group of 4 piles 2.4 in. diam-
eter, mass of footing = 103.3 slugs, horizontal excitation). Sliding component.

bringing the theoretical resonant frequencies

For personal use only.

and amplitudes much closer to the experi-

mental values. This can be seen from Table 4
where the results obtained with the relaxed
tips are denoted by an asterisk (see also Fig.
1 2 ) . A reduction of shear modulus appears
desirable even for the vertical case, but the
correction required is much smaller than that
needed in the horizontal case. This seems log-
ical as the relaxed tip transfers the motion into
the deeper layers of soil where the confining
pressure and hence also the stiffness of the soil
is higher.
With the soil acting beneath the footing base
FIG.12. Comparison of theoretical and experimen-
tal vertical response (One pile 3.5 in. diameter, mass (case 5 in Table 4, Fig. 8 ) , it is probably
of footing 86.41 slugs, relaxed tip). desirable to reduce the soil reactions calculated
for the base because there must be an inter-
dynamic analysis. With the shear modulus action between the base and the piles similar
from a static test, the natural frequencies are to the interaction between a group of piles.
predicted remarkably well as can be seen from (The theoretical soil reaction acting on the
Table 4. For a pile group, this agreement is base and used in the analysis was calculated
further improved by an approximate correction from formulas given in Novak 1974b).
for the pile interaction (Fig. 1 1 ). A good part of the difficulty met when com-
The vertical resonant frequencies and ampli- paring the theory with the experiments stems
tudes are significantly over-estimated as long from the very large difference between the
as the pile tips are assumed to be fixed. This shear velocities V,, and V,,. It may be that for
difficulty cannot be circumvented by consider- full scale conditions this difference will be
ing a reduced shear modulus of soil because smaller and the theory may work generally
such a reduction lowers both stiffness and better. In the model experiments there is no
damping. The relaxation of the pile tips re- overburden and the confining pressure is ex-
duces the stiffness and increases the damping tremely small because only a very shallow layer
 384 CAN. GEOTECH. J. VOL. 13, 1976

of soil a few pile diameters deep effectively Canada. The assistance of John Howell and
participates in the motion. In the full scale, Gary Lusk is gratefully acknowledged.
this effective layer is deeper because the pile
diameter is greater and the confining pressure KUHLEMEYER, R. L. 1976. Static and dynamic laterally
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by UNIVERSITY OF ARIZONA LIBRARY on 09/06/13

loaded piles. Res. Rep. No. CE76-9, Dep. Civ. Eng.,

is larger; also, there may be some pressure Univ. Calgary, Calgary, Alta., 48 p.
transmitted by the cap. The reduction of V,v NOVAK,M., 1971. Data reduction from nonlinear response
required may be smaller if the soil is cohesive. curves. ASCE, J. Eng. Mech. Div. 97, pp. 1187-1204.
1 9 7 4 ~ Dynamic
. stiffness and damping of piles.
Conclusions Can. Geotech. J . 11(4), pp. 574598.
19740. Effect of soil on structural response to wind
Model experiments conducted in the field and earthquake. Int. J. Earthquake Eng. Stlx~ct.Dyn.
with single piles and a pile group suggest the 3(1), pp. 79-96.
following conclusions concerning the potential 1976. Vertical vibration of floating piles. Res. Rep.
BLWT-SSI-76, Fac. Eng. Sci., Univ. Western Ontario,
of the theory presented by Novak (1974~1, London, Ont., 23 p.
1976). NOGAMI, T. and NOVAK, M. 1976. Soil-pile interaction in
The theory is able to predict all the qualita- vertical vibration. Int. J . Earthquake Eng. Struct. Dyn.
tive features of response of piled footings to 4, pp. 277-293.
dynamic loads. The relative magnitudes of the P o u ~ o s ,H. G. 1968. Analysis of the settlement of pile
groups. Geotechnique, 18, pp. 449-471.
first and second peaks of the coupled response 1971. Behavior of laterally loaded piles: 11-pile
to horizontal load and the effect of vertical groups. ASCE, J. Soil Mech. Found. Div. 97(SM5),
stiffness and damping of the piles on this re- pp. 733-75 1.
sponse were particularly well predicted. 1972. Behavior of laterally loaded piles 111. Sock-
eted piles. ASCE, J. Soil Mech. Found. Div. 98(SM4),
The quantitative prediction of the response
For personal use only.

pp. 341-360.
depends very much on the shear modulus of RICHART, F. E., HALL,J. R., and WOODS,R. D. 1970.
soil used in the analysis. With a shear modulus Vibrations of soils and foundations. Prentice-Hall, Inc.,
characteristic of the deeper layers of soil res- New York, N.Y.
onant frequencies and amplitudes can be sig-
nificantly over-estimated. Notations
Better results can be obtained with lower A = area of pile cross section
values of the shear modulus characteristic of = din~ensionlessamplitude of foot-
Ai,j
the uppermost layer of soil. For small displace- ing in direction i and mode j
ments and horizontal excitation, a suitable AG = dimensionless amplitude adjusted
value of this shear modulus can be derived for the effect of pile interaction
from a static test of the pile. This approach a, = r,w/V, = dimensionless frequency
also approximately accounts for the separation c = coefficient of damping
of the pile from the soil. CG = coefficient of damping adjusted
With the shear modulus from the static test, for the effect of pile interaction
natural frequencies are predicted very well and D = damping ratio of footing
the prediction of resonant amplitudes is in DG = damping ratio of footing adjusted
most cases also improved for horizontal ex- for the effect of pile interaction
citation. For pile groups, a further improve- e = eccentricity of rotating mass
ment can be achieved by the correction for the f~ = natural frequency adjusted for the
effect of pile interaction. This correction can effect of pile interaction
be based on the static analysis by Poulos f = natural frequency
(1968, 1971). G = shear modulus of soil
For vertical excitation, the pile tip should I = moment of inertia of pile cross
be considered as relaxed unless it rests on bed- section
rock or the pile is extremely long. Also, a = mass moment of inertia of footing
smaller reduction of the shear modulus appears k = stiffness constant
sufficient for the vertical response. k~ = stiffness constant adjusted for the
effect of pile interaction
Acknowledgements I = length of pile
This study was supported by a research m = mass of footing
grant from the National Research Council of me = unbalanced rotating mass
 NOVAK AND GRIGG 385

ro = radius of pile TABLE5. Measured shear wave velocity

uo = amplitude of horizontal transla-
tion Frequency Wavelength Shear wave velocity
vti = shear wave velocity of soil below f (Hz) L(ft) v," = L x f(ft/s)
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by UNIVERSITY OF ARIZONA LIBRARY on 09/06/13

pile tip 47 12.0 564

V, = ,h.%. . = shear wave velocity of soil 40.0 14.4 576
< t = apparent shear wave velocity de-
rived from static test of pile
33.3
27.7
17.9
20.9
596
558
20.0 35.5 710
vw = shear wave velocity of soil derived
from wave propagation
Wo = amplitude of vertical translation Richart et al. 1970). The wave lengths mea-
Xr = horizontal distance of pile from sured at different frequencies and the shear
center of gravity wave velocities calculated are given in Table 5.
2, = height of excitation above cen- The average velocity calculated from the
troid first four values is 573 ft/s (175 m/s). This
"r = interaction factor of pile in group value was used as V,,. in the paper. The last
P = mass density of soil value of 710 corresponds to the deeper layer.
40 = amplitude of rocking The wave velocities established for different
w = circular frequency of excitation frequencies are related to surface waves that
propagate to the depth proportional to the
Appendix. Experimental Values of Shear wave length. The exciter available could not
Wave Velocity reach the high frequencies needed to sample
For personal use only.

The velocity of shear waves in the soil was the upper foot or two of the soil. The introduc-
established by means of the steady state vibra- tion of values V,, attempts to describe the soil
tion technique. (The method is described in stiffness in the uppermost layer.