Professional Documents
Culture Documents
Wu1989 PDF
Wu1989 PDF
IN P I L E DRIVING ANALYSIS
By A. K. H. Wu, 1 R. L. Kuhlemeyer,2 and C. W. S. To 3
Downloaded from ascelibrary.org by University Of Ulster AT on 06/01/15. Copyright ASCE. For personal use only; all rights reserved.
INTRODUCTION
In recent years, extensive use has been made of the one-dimensional wave
equation method for the analysis and design of practical piling problems in
both onshore and offshore structures. Isaacs (1931) is believed to be the first
to recognize that the transmission of energy from hammer to pile toe would
not occur immediately during impact, and wave action occurred during pile
installation. Smith (1960) capped his earlier work by suggesting a numerical
solution of the wave equation for complicated pile driving problems and
idealized a pile-soil system as a lumped mass and spring model. Fig. 1 shows
an actual pile-soil system and the idealization of the various components in
the system. The pile is divided into a number of segments and internal springs
are inserted between the segments to account for the pile stiffness. The weights
of the segments denote the weight distribution of the pile. The soil medium
is assumed to be weightless. Each segment is attached to a spring and a
dashpot system, which represents the soil stiffness and damping resistance,
respectively. At the bottom segment, there is an extra system to account for
the toe resistance.
Fig. 2 shows the load-deformation characteristic of soil used in the Smith
wave equation analysis. The soil is assumed to be elastic, perfectly plastic.
For the soil adjacent to the pile shaft, the load-deformation is represented
by path OABCDEFG. (For the soil at the pile toe, the load deformation path
is OABCF because only compressive loading occurs around the pile toe, but
this part of the problem is not considered in this paper.) Each pile segment
in the idealized system can have its own soil mode, which consists of a
'Grad. Res. Student, Dept. of Civ. Engrg., Univ. of Calgary, 2500 University
Dr., N.W., Calgary, Alberta, Canada T2N 1N4.
2
Assoc. Prof., Dept. of Civ. Engrg., Univ. of Calgary, Calgary, Alberta, Canada.
3
Assoc. Prof., Dept. of Mech. Engrg., Univ. of Western Ontario, London, On-
tario, Canada.
Note. Discussion open until February 1, 1990. Separate discussions should be sub-
mitted for the individual papers in this symposium. To extend the closing date one
month, a written request must be filed with the ASCE Manager of Journals. The
manuscript for this paper was submitted for review and possible publication on March
18, 1988. This paper is part of the Journal of GeotechnicalEngineering, Vol. 115,
No. 9, September, 1989. ©ASCE, ISSN 0733-9410/89/0009-1285/$1.00 + $.15
per page. Paper No. 23886.
1285
s
Downloaded from ascelibrary.org by University Of Ulster AT on 06/01/15. Copyright ASCE. For personal use only; all rights reserved.
INTERNAL 1 J l
SPRING
±
K =— (1)
Q
in which Ru = ultimate soil resistance; and Q = soil quake (maximum elastic
deformation). The ultimate soil resistance for each pile segment is assigned
a fraction of the total pile resistance obtained from a static pile load test.
1286
1287
The pile shaft-soil system can be idealized as a simple finite element model
based on two assumptions.
1. The pile is assumed to be infinitely long, circular, and rigid such that end
effects are ignored. Hence, the vertical axisymmetric displacement of the soil is
independent of depth and the problem is one of plane strain such that a thin layer
is adequate to show the response of the system.
2. The radial displacement of the soil is negligible relative to the vertical dis-
placement. (This assumption restricts the wave energy to propagate only as a
shear wave in the radial direction.)
For convenience, these two assumptions are called assumption 1 and as-
sumption 2 hereafter. Both assumptions have been shown to give accurate
results for elastic pile, steady-state vibration problems (Kuhlemeyer 1979;
Novak 1977). The resulting model, shown in Fig. 3, is an axisymmetrical
pile shaft-soil system, with one pile element at the center and a series of
soil elements that extend outward to the boundary. The pile element is ap-
proximately rigid with a shear modulus much larger than the soil. Each node
of the soil element is connected to a dashpot which accounts for the soil
material damping. According to assumption 1, vertical nodal displacements
along OA will be identical to those along O'B and, therefore, the response
of the system can be represented by those nodes along either OA or O'B.
Each nodal point has only one degree of freedom.
In order to simulate the effect of slipping between the pile and the soil,
an axisymmetric interface element is inserted at the pile-soil interface. The
interface element has zero thickness and allows a relative displacement at
the interface when the shear stress of the element exceeds a defined yield
limit. The existence of the interface element will not affect the compatibility
of the finite element model. That is, the response of a model with an in-
1288
_(L o /
SOIL DASHPOT ENERGY
ABSORBING
BOUNDARY
1
M = - [. (8)
1289
{e} = (9)
Downloaded from ascelibrary.org by University Of Ulster AT on 06/01/15. Copyright ASCE. For personal use only; all rights reserved.
GH-n 1 -1
[k] = (rj - rf) (11)
R2
-1 1
in which H — the model thickness.
The stress vector can be written in terms of nodal displacement a, and aj
as
{cr}=[£>]{e} = - [ - l (12)
R
Note that the element has only two degrees of freedom and the displacement
variation is linear; thus, the element stress is constant throughout the ele-
ment. That means Eq. 12 represents the shear stress for any point within
the element.
The consistent mass matrix, which accounts for the actual mass distri-
bution throughout the system, is
in which p = the mass density. Since [w] = [N]{a}, the shape function [N]
can be found by Eq. 8. The element mass matrix can be expressed as
rj-2rjn
+8r/? - 3rt +2r/, 3 - r\
TtpH
(14)
r) - 2r]r, 3rJ - Sr]n
+2rff - rf +6rjrf - rf
Simplified Interface Element
The interface element, also known as a slip element or joint element, sim-
ulates the relative movement at the pile-soil interface. The interface element
1290
Damping Matrix
The internal dissipation of energy in soil is often considered to be hys-
teretic damping, which is represented by the total area within the loop of
the damping force versus displacement curve for each loading cycle. Hys-
teretic damping is found to depend on the soil strain (Kim and Novak 1981;
Seed and Idriss 1970). The equation of motion with hysteretic damping may
be written in complex notation as
[MM + (1 + OQ[K]{a} = {P}eiM (18)
in which {a}, {a} = acceleration and displacement vectors, respectively; £ =
damping ratio; i = imaginary unit; to = exciting frequency; and t = time.
Assuming {a} = {X}"", where X is a real number that represents the dis-
placement amplitude, Eq. 18 can be rewritten as
[M]{d) + [CM + [K]{a} = {P}eiM (19)
and
[C]=-[K] (20)
CO
in which {a} = velocity vector. The damping ratio is highly strain dependent;
the general relation of shear strain amplitude and damping ratio follows that
proposed by Seed and Idriss (1970). Eq. 20 is only suitable for the damping
formulation in the entire system. In order to consider the damping variation
1291
in which [c] and [k] = element damping and stiffness matrices. The element
Downloaded from ascelibrary.org by University Of Ulster AT on 06/01/15. Copyright ASCE. For personal use only; all rights reserved.
damping ratio £ can be found according to the individual element strain from
the damping ratio versus shear strain amplitude curve given by Seed and
Idriss (1970). The damping matrix for the entire system is then obtained by
the general assemblage procedure. Since the stiffness matrix is tri-diagonal
for the present model, the assemblage damping matrix will thus be a tri-
diagonal matrix. It should be emphasized that neither Eq. 20 nor Eq. 21 are
usual relationships for expressing material damping in geotechnical prob-
lems; most investigators prefer to use the Rayleigh damping (proportional
damping) expression, which depends on the natural frequency of the model.
However, the finite element model presented in this study does not have a
natural frequency because the energy absorbing boundary can be placed at
any distance from the pile and actually makes the system infinitely large.
Therefore, Eq. 21, which depends on the exciting frequency, is employed
in this study.
pV, (22)
H?{a0)\
in which jzr = shear stress at boundary; p = mass density of boundary ma-
terial; Vs = shear wave velocity of boundary material; a„ = frequency ratio,
oyr/Vs; //o2)(a„), Hf\a0) = Hankel functions of second kind of zero order
and first order, respectively; / = imaginary unit; and a = velocity at bound-
ary.
Eq. 22 shows that the shear force at the boundary for steady state vibration
is a complex number. However, this force can be separated as a spring force
and a damping force which can then be introduced into the stiffness and
damping matrices in the equation of motion. In other words, the shear wave
absorber is actually a spring and a dashpot being added to the boundary at
a distance r from the pile center. The detailed method of force separation
is well known and is presented in detail by Wu (1985).
DEFORMATION
1294
with the yield stress in the finite element slip element. (Note that in wave
equation practice, a rather arbitrary value of the soil quake is assumed and
the total ultimate pile resistance is rather arbitrarily divided into shaft and
toe resistance, where the total resistance is obtained from a static load test.)
Hence, it was concluded that the Smith model soil spring stiffness and quake
values must partially include dynamic effects.
In order to determine the spring stiffness that includes some of the dy-
namic effects, a dynamic finite element analysis was completed with zero
soil damping (the damping would be included later). In reality, a pile driving
force pulse has a very large magnitude but very short duration. To incor-
porate this force pulse into the direct integration solution, the force pulse
was approximated by an equivalent triangular function symmetrical about an
axis through the peak. Since the formulation of the energy-absorbing bound-
ary is based on a steady-state solution, the applied loading must be sinu-
soidal. Fourier series were employed to resolve the equivalent triangular
function into a set of harmonic functions. However, it was noted that only
the odd-term harmonic functions contributed to the summation of the series
and the amplitude of the first harmonic function was nine times that of the
third harmonic function; therefore, the hammer force pulse was approxi-
mated closely by the first harmonic of a symmetrical triangular function.
The duration (end of the triangular function) was assumed to be in the range
of 0.01 to 0.03 s. The correlation of the finite element model with the Smith
model was performed in two steps.
1. Find the pile response from the dynamic finite element analysis.
2. Carry out an analysis with the Smith model by adjusting the spring stiffness
until the first pile displacement peak is similar to that given in step 1. The ob-
tained spring stiffness is referred to as the dynamic spring stiffness Kdyn.
Results obtained in the correlation show that the dynamic spring stiffness
values depend on the shear modulus G, model thickness H, pile radius r0
(r„ between 0.15 and 0.2 m), and force pulse duration. Fig. 7 shows the
variation of the ratio Kdyn/GH with the shear modulus at the different pulse
durations and pile radii.
The correlation of the yield limits is based on two assumptions.
1295
begin in the two analyses. Referring to Figs. 5 and 6, the correlation can be
done by equating the ultimate resistance Ru of the Smith model with the cor-
responding shear force induced at the pile-soil interface in the finite element
model at the instant when the interface element begins to slip. This shear force
is calculated as the product of the shear strength Cu and the shear area A of the
pile
Ru = KdynQ = C„A : (23)
and Q can be expressed as
2Cuirr„H
Q = — (24)
1. Find the pile displacement from the finite element analysis by including
both material and radiation damping.
2. From Fig. 7, choose a spring stiffness that corresponds to the shear mod-
ulus G used in step 1.
3. Use Eq. 24 to calculate the soil quake.
4. Use the information obtained from steps 2 and 3, and carry out analyses
on the Smith model by adjusting the damping parameter J' until the first pile
displacement peak is similar to that given in step 1. (In conventional analysis,
the final displacement, or set, will be of more interest. However, the modified
Smith's equation, Eq. 4, is not generally valid when the pile displacement passes
the first peak and displaces with a negative penetration velocity. Therefore, for
a consistent comparison, the first pile displacement peak was chosen as the cri-
terion.) The obtained J' represents the damping value for the soil type used in
the analysis.
This method uses the finite element results to evaluate the soil damping
values of the Smith model (i.e., the damping parameter in wave equation
analysis). Step 1 involves calculating the pile displacement from the finite
element analysis by taking into account material and radiation damping. Steps
2 and 3 determine the required stiffness parameters for the analysis using
the Smith model (these parameters do not involve the damping effect). Step
4 is a reasonable method of introducing the damping effect into the analysis
(the first displacement peak obtained from the analysis matches with that
from the finite element analysis).
RESULTS
1296
(c)
C = 200kPa
.. . " I
°'0 1 ' . . . I I I I'I I I l'l I I I 1° =t~r-
0 1 2 3 4 5
VELOCITY U/s)
FIG. 8. J' versus Pile Velocity: (a) G = 10 MPa; (b) G = 40 MPa; (c) G = 100
MPa (Using Smith's Equation)
1. The damping parameter J' is not a constant but varies with the pile pen-
etration velocity. The damping parameter decreases as the pile velocity increases.
The general shape of these curves agrees quite well with what was found in
some references (Cayle and Gibson 1970; Coyle et al. 1972; Litkouhi and Poskitt
1980).
1297
2.0 -
if 1.8-
Downloaded from ascelibrary.org by University Of Ulster AT on 06/01/15. Copyright ASCE. For personal use only; all rights reserved.
1.6 -
FIG. 9. p/py versus Yielding Zone Radius for G = 100 MPa, c„ = 200 kPa
2. For a given shear modulus, the damping parameter varies in a lower range
as the shear strength increases. That means that for a given shear modulus, the
damping parameter decreases with increasing soil shear strength.
3. For soils with the same shear strength, the one with the lower shear mod-
ulus is slightly more sensitive to damping.
1 .2 -i 1
0.8 -
~> 0.4 -
°-° I MT-| I I i
0.008 0.012 0.016 0.020 0.024 0-028 0.032
DURATION (s)
FIG. 10. J' versus Duration for G = 40 MPa, C„ = 150 kPa, P/Py = 1.6
1298
is applied to the system for a longer period of time and, thus, the total
loading increases. As a result, increasing the duration decreases the value
of damping parameter.
The curve also shows that the damping parameter varies in a least extent
between 0.02 to 0.03 s, because rate of change of duration is higher when
the duration is in the lower range of value. Therefore, the damping parameter
is found to vary more between the durations 0.01 and 0.02 s.
o. i
o. i
140kPa
200kPa
0.0
0. 1
Cu = 200kPa (c)
1 „ .
I ' ' ' ' I ' ' ' ' I ' '"
1 2 3
VELOCITr (m/a)
FIG. 11. J' versus pile velocity: (a) G = 10 MPa; (b) G = 40 MPa; (c) G = 100
MPa (Using Smith's Modified Equation)
1299
The analyses were carried out (using the aforementioned data) by raising
the velocity to power 0.18, which was suggested to be an acceptable A' value
for clay soil (Coyle and Gibson 1970; Litkouhi and Poskitt 1980). Figs.
Downloaded from ascelibrary.org by University Of Ulster AT on 06/01/15. Copyright ASCE. For personal use only; all rights reserved.
11 (a)-(c) show the variation of the damping parameter with pile penetration
velocity. In Figs. 11(b) and 11(c), the damping parameter is found to be
reasonably constant, while in Fig. 11(a), the damping parameter varies over
the range of pile velocity. These results reveal that the modified Smith's
equation reduces the variation of the damping parameter to a smaller range.
For soil with higher shear modulus, the damping parameter obtained by the
modified Smith's equation is reasonably constant. However, for soil with
low shear modulus, the damping parameter still varies with the pile pene-
tration velocity.
CONCLUSION
1. The damping parameter, J', in Smith's original equation varies with both
the pile velocity and the force pulse duration.
2. The damping effect is less significant for soil with high shear modulus.
3. For soils with the same shear strength, damping is more significant for the
one with lower shear modulus.
4. The radius of the soil yielding zone due to shaft movement during pile
driving is less than about 1.5 pile diameters.
5. The damping parameter determined from the modified Smith's equation is
constant only for soil with high shear modulus.
ACKNOWLEDGMENTS
The writers are thankful to the University of Calgary for providing the
computing facilities as well as financial assistance to the first writer.
APPENDIX I. REFERENCES
Aitken, A. C. (1937). "The evaluation of the latent roots and latent vectors of a
matrix." Proc, Royal Soc. of Edinburgh, 57, Edinburgh, U.K., 269-304.
1300
a„ = frequency ratio;
A = shear area;
{«'},{«},{a} = acceleration, velocity, and displacement vectors;
bub2,bi,bi = unknown coefficients;
[B] = strain-displacement transformation matrix;
C„ = soil shear strength;
[C] = system damping matrix;
[c] = element damping matrix;
D = total soil displacement from original position;
D' = total plastic deformation;
[D] = elasticity matrix;
G = shear modulus;
Gp = plastic slope;
H = model thickness;
//o 2) ,// ( i 2) = Hankel function of second kind of zero order and first or-
der;
/ = imaginary unit;
J, J' = damping parameters for pile toe and pile shaft;
K = spring stiffness in Smith's model;
1301
1302