Computers and Geotechnics 70 (2015) 50–59

Contents lists available at ScienceDirect

Computers and Geotechnics

journal homepage: www.elsevier.com/locate/compgeo

Research Paper

Vertical response of a thin-walled pipe pile embedded in viscoelastic soil

to a transient point load with application to low-strain integrity testing
Changjie Zheng a,b, Hanlong Liu c, George P. Kouretzis b, Scott W. Sloan b, Xuanming Ding c,⇑
a
Key Laboratory of Ministry of Education for Geomechanics and Embankment Engineering, College of Civil and Transportation Engineering, Hohai University, Nanjing 210098, China
b
ARC Centre of Excellence for Geotechnical Science and Engineering, Faculty of Engineering and Built Environment, The University of Newcastle, Callaghan, NSW 2308, Australia
c
Key Laboratory of New Technology for Construction of Cities in Mountain Area, College of Civil Engineering, Chongqing University, Chongqing 400045, China

a r t i c l e i n f o a b s t r a c t

Article history: This paper presents an analytical method to compute the dynamic response of a thin-walled pipe pile due
Received 23 March 2015 to a vertical transient point load acting on its head. Inspired from challenges faced during the interpre-
Received in revised form 13 July 2015 tation of low-strain integrity tests on pipe piles, the proposed method moves beyond the widely used
Accepted 30 July 2015
one-dimensional wave theory to consider the asymmetric nature of the problem, and stress wave prop-
Available online 8 August 2015
agation along both the vertical and circumferential directions. Coupling of pipe pile–viscoelastic soil
vibration is considered via modeling the outer and inner soil as a series of infinitesimally thin layers in
Keywords:
perfect contact with the pile, and their low-strain properties are directly introduced in the solution.
Thin-walled pipe pile
Wave propagation
The methodology is validated against numerical results, before discussing the mechanisms governing
Low-strain integrity testing the dynamic response of the pipe pile–soil system to the impact load, with emphasis on the vertical
Pile dynamics velocity measured at a hypothetical receiver placed on the pile head. Additional results from a parametric
analysis are used to provide insights on the accurate estimation of the arrival time of the receiving wave,
and the optimal location of the receiver.
Ó 2015 Elsevier Ltd. All rights reserved.

1. Introduction embedment depth of deep foundations. Wang et al. [5] investi-

gated the vertical dynamic response of an inhomogeneous vis-
Low strain integrity testing techniques have been used exten- coelastic pile embedded in layered soil. In more recent studies Ni
sively for the assessment of the quality of foundation piles. The test et al. [6] proposed a new numerical signal processing method to
consists of applying an impact load on the pile head with a explore the time–frequency component of recordings from tests
hand-held hammer. An accelerometer placed on the pile head cap- on drilled piles with high slenderness ratio. Later, Ni et al. [7]
tures the vertical vibrations due to the hammer impact in the form adopted the continuous wavelet method with time–frequency dis-
of a velocity time history, which is used to determine the as-built tribution to enhance the interpretation potential of both numerical
properties of the pile. The theory behind the interpretation of simulations and experimental measurements.
low-strain pile test results is based on the analysis of the axial Most studies in the literature, including the ones mentioned
response of a cylinder embedded into the soil to a transient point above, focused on the common case of solid piles. The above theo-
load, which is the subject of a number of studies in the literature. retical studies for solid piles are all based on one-dimensional (1D)
For example, Smith [1] first applied the wave equation method for wave theory, which implies that wave propagation satisfies the
pile driving analysis to determine the response of a discrete plane-section assumption; that is, the impact waves propagate
hammer-pile-soil model. Goble et al. [2] developed the Wave along the vertical direction only. For small diameter solid piles,
Equation Analysis Program (WEAP87) in the 1970s, newer versions the plane-section assumption is reasonable, and 1D wave theory
of which are now applied widely to predict the drivability of piles suffices for the description of the problem. However, the
and process field test data. Rausche et al. [3] proposed a simple soil plane-section assumption is not valid for large diameter
resistance model and deduced a quasi closed-form solution for dri- thin-walled pipe piles which are also commonly used in practice
ven piles subjected to impact loads. Morgano [4] discussed the use [8–11], because the stress field produced by a transient point load
of non-destructive methods for the determination of the is no longer axisymmetric. In addition to that, the stress waves
generated by the impact load propagate not only in the vertical
but also in the circumferential direction. Gazis [12,13] was the first
⇑ Corresponding author.
to propose a three-dimensional solution for wave propagation in
E-mail address: dxmhhu@163.com (X. Ding).

http://dx.doi.org/10.1016/j.compgeo.2015.07.016
0266-352X/Ó 2015 Elsevier Ltd. All rights reserved.
 C. Zheng et al. / Computers and Geotechnics 70 (2015) 50–59 51

an infinite hollow cylinder. Although Gazis’ solutions contribute to Possible pile defects considered include two types: an anomaly in
the understanding of three-dimensional wave propagation in a the pile shape, which results in a local variation of the area of the
hollow cylinder, they cannot be directly applied to the pipe pile cross section, and a local degradation of the pipe pile material
problem, as piles feature a finite length. With that in mind Ding properties. The first type of defect corresponds to bulging/necking,
et al.[14–16] derived the frequency- and time-domain solutions and can be introduced in the solution by locally varying the thick-
for a pipe pile of finite length, while simulating the resistance of ness of the pipe pile wall hk while keeping r0 constant. The second
the soil along the pile shaft by means of a Winkler or Voigt model. type corresponds to a weak material zone due to e.g. poor concrete
However, one drawback of Winkler or Voigt models is that they compaction, and can be introduced in the analysis by locally reduc-
cannot capture coupling of vibrations between the pile and its sur- ing the Young’s modulus of the pipe pile material Epk . Furthermore,
rounding soil. Furthermore, the calibration of the spring and dash- certain assumptions are introduced to simplify the solution [17]:
pot parameters is cumbersome, even though they have a
prominent effect on the vibration response of the soil-pile system. (1) The outer and inner soil are modeled as a series of homoge-
The purpose of this paper is to tackle this deficiency of existing neous infinitesimally thin layers. Furthermore, it is reason-
methods. We present a formulation to describe the vertical able to neglect the gradient of the vertical stresses of the
response of a thin-walled pipe pile to a transient point load, by thin soil layers [18].
considering coupling of pipe pile and the outer and inner viscoelas- (2) The pile is elastic and is in perfect contact with the outer and
tic soil vibrations along the shaft of the pipe pile. The cyclic inner soil, which is a reasonable assumption given the low
response of the viscoelastic soil can be accounted for straightfor- amplitude of the excitation.
wardly via its low-strain shear modulus and damping ratio, which (3) Only vertical displacements of the pile-soil system are con-
can be estimated from laboratory tests. For the reasons discussed sidered, and the vertical displacement of the thin-walled
above, wave propagation in the thin-walled pipe pile is considered pipe pile is assumed invariant in the radial direction.
in both the vertical and circumferential directions. The frequency (4) The soil-pile system is at rest at t = 0. This constitutes the
domain analytical solutions are derived based on the assumption initial conditions of the problem.
of perfect contact between the pipe pile and the soil, with the latter (5) The base resistance of the pipe pile can be introduced in the
modeled as a series of infinitesimally thin layers. Accordingly, the solution via an elastic spring, of stiffness kp (Fig. 2).
time domain results are obtained by the means of inverse Fourier (6) The vertical transient point load pðh; t Þ is simulated as a
transformation. Results from the proposed model are verified by half-sine pulse [19], described by Eq. (1) and shown in
comparison with numerical analyses. Finally we discuss the mech- Fig. 3. The coordinates of the point of load application are
anisms governing vibration propagation in pipe piles and investi- ðz; hÞ ¼ ð0; 0Þ.
gate the influence of the key parameters on the vertical
vibrations of the pile head via arithmetic applications, which lead
p 
pðh; t Þ ¼ P0 sin t dðhÞHðT  t Þ ð1Þ
to certain important findings of practical interest. T
where P0 is the amplitude of the impact force; T is the
2. Conceptual model and basic assumptions half-period; dðÞ is a Dirac function; and HðÞ is a Heaviside
step function.
The following equations are formulated on the cylindrical coor-
dinate system shown in Fig. 1. The conceptual model is presented 3. Governing equations
in Fig. 2. The pile is discretized into three segments, with the
Young’s modulus and wall thickness of each segment being Epk Ignoring the stress gradient in the vertical direction, the
and hk , respectively, where k = 1, 2, 3. The elevation of the bottom dynamic equilibrium equations of the outer soil can be expressed
of each segment is Hk . The mean radius of the pipe pile is r0 . as:

p ( ,t)

r
r0

E1 h1
H1
outer inner
H2 soil soil thin-walled
pipe pile
H3 h2
E2

E3 h3

kp

Fig. 1. Cylindrical coordinate system originating at the center of the pipe pile head. Fig. 2. Conceptual model of the pipe pile–soil system.
52 C. Zheng et al. / Computers and Geotechnics 70 (2015) 50–59

!
@2 @ @2
l1 þ þ U 1 ¼ q1 x2 U 1 ð7Þ
@r2 r@r r 2 @h2

P0 where U 1 ðr; h; xÞ is the Fourier transform of u1 ðr; h; t Þ.

Assuming U 1 ¼ R1 ðr ÞU1 ðhÞ, Eq. (7) can be expressed as:
R001 ðr Þ l1 R01 ðr Þ l1 U001 ðhÞ
l1 þ þ ¼ q1 x2 ð8Þ
R1 ðr Þ r R1 ðr Þ r 2 U1 ðhÞ
As points ðr; hÞ and ðr; h þ 2pÞ coincide, we have
U1 ðhÞ ¼ U1 ðh þ 2pÞ ð9Þ
So the general solution for U1 ðhÞ can be obtained as:
U1 ðhÞ ¼ A1 cos ðmhÞ þ B1 sin ðmhÞ ð10Þ
0 T t
where m ¼ 0; 1; 2;    ; A1 and B1 are undetermined coefficients.
Fig. 3. Half-sine impact force acting on the pipe pile head. Owing to the geometry of the problem, it is known that
U 1 ðr; h; xÞ is essentially an even function of h, i.e.,
@ 2 u1 @u1 @ 2 u1 @ 2 u1 U 1 ðr; h; xÞ ¼ U 1 ðr; h; xÞ ð11Þ
l1 2
þ l1 þ l1 2
u 1 ¼ q1 2 ð2Þ
@r r@r r2 @h @t
Hence,
where u1 ðr; h; t Þ is the vertical displacement of the outer soil;
pffiffiffiffiffiffiffi U1 ðhÞ ¼ A1 cos ðmhÞ ð12Þ
l1 ¼ G1 ð1 þ n1 iÞ, in which i ¼ 1 and G1 and n1 are the
low-strain shear modulus and hysteretic damping ratio of the outer Substituting Eq. (12) into Eq. (8), we obtain:
soil; and q1 is the mass density of the outer soil.
R1 ðr Þ ¼ C 1 K m ðq1 r Þ þ D1 Im ðq1 r Þ ð13Þ
Accordingly, the dynamic equilibrium equation of the inner soil
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
can be described as: where q1 ¼ q1 x2 =l1 ; Im ðÞ and K m ðÞ are the first and second
kinds of modified Bessel functions of order m, respectively; and C 1
@ 2 u2 @u2 @ 2 u2 @ 2 u2
l2 þ l2 þ l2 ¼ q2 2 ð3Þ and D1 are undetermined coefficients.
@r 2 r@r 2 2
r @h @t The displacement of the outer soil at an infinite distance from
where u2 ðr; h; tÞ is the vertical displacement of the inner soil; the pipe pile r ! 1 is zero, thus
l2 ¼ G2 ð1 þ n2 iÞ in which G2 and n2 are the low-strain shear modu- D1 ¼ 0 ð14Þ
lus and hysteretic damping ratio of the inner soil; and q2 is the mass
density of the inner soil. The vertical displacement of the outer soil can now be expressed
The dynamic equilibrium equation of the k-th segment of the in the following form:
thin-walled pipe pile can be established while considering the X
1

equilibrium of forces along its z-direction, as: U1 ¼ Am K m ðq1 rÞ cos ðmhÞ ð15Þ
m¼0
Epk @ 2 upk ðz; h; tÞ Epk @ 2 upk ðz; h; tÞ f 1k r 1k þ f 2k r2k where Am is a series of undetermined coefficients.
þ   2 
1  mpk
2 @z 2
2 1 þ mpk r 0 @h2 hk r 0 Accordingly the frictional resistance developing at the interface
@ 2 upk ðz; h; tÞ between the k-th pile segment and the outer soil can be calculated
¼ qpk ð4Þ as:
@t2
@U 1 X1
where upk ðz; h; t Þ is the vertical displacement of the k-th pile seg- F 1k ¼ l1 j ¼ l1 Am g1km cos ðmhÞ ð16Þ
ment; r 1k ¼ r 0 þ hk =2 and r2k ¼ r0  hk =2 are the outer and inner @r r¼r1k m¼0
radii of the k-th pile segment; Epk ; qpk and mpk are the Young’s mod-
where F 1k ðh; xÞ is the Fourier transform of f 1k ðh; tÞ, and
ulus, mass density and Poisson’s ratio of the k-th pile segment, (
respectively; and f 1k ðz; h; t Þ and f 2k ðz; h; t Þ are the frictional resis- q1 K 1 ðq1 r 1k Þ m¼0
tances developing at the outer and inner interfaces of the k-th seg- g1km ¼ q1 ð17Þ
2
½K mþ1 ðq1 r 1k Þ þ K m1 ðq1 r1k Þ m > 0
ment, respectively.

4. Solutions of the governing equations 4.2. Solution of the governing equation of the inner soil

The Fourier transform is used to solve the governing equations. The solution for the inner soil is derived by a similar procedure
The one-sided Fourier transform of a function f ðtÞ is defined as: to that for the outer soil. Considering that the displacement of the
Z inner soil at r ¼ 0 is bounded, the solutions for the inner soil can be
1
FtðxÞ ¼ eixt f ðt Þdt ð5Þ obtained as:
0 X
1

Under the zero initial conditions mentioned in the assumptions, the

U2 ¼ Bm Im ðq2 rÞ cos ðmhÞ ð18Þ
m¼0
Fourier transform takes the following form:
 X1
F 00 ðxÞ ¼ x2 FðxÞ ð6Þ @U 2 
F 2k ¼ l2  ¼ l2 Bm g2km cos ðmhÞ ð19Þ
@r r¼r2k m¼0

4.1. Solution of the governing equation of the outer soil where U 2 ðr; h; xÞ and F 2k ðh; xÞ are the Fourier transforms of
u2 ðr; h; tÞ and f 2k ðh; tÞ, respectively and Bm is a series of undeter-
Performing the Fourier transform on Eq. (2) yields: mined coefficients. In addition,
 C. Zheng et al. / Computers and Geotechnics 70 (2015) 50–59 53

sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
q2 x2 where
q2 ¼  ð20Þ vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
" #ffi
l2 u
u1  m2pk Epk m2 r 1k l1 g1km Im ðqr 2k Þ þ r2k l2 g2km K m ðqr 1k Þ
bkm ¼ t    qpk x þ 2

( Epk 2 1 þ mpk r20 hk r0 K m ðqr 1k ÞIm ðqr 2k Þ

q2 I1 ðq2 r 2k Þ m¼0 ð32Þ
g2km ¼ q2 ð21Þ
2
½Imþ1 ðq2 r2k Þ þ Im1 ðq2 r 2k Þ m > 0
and C km and Dkm are the coefficients determined from the boundary
conditions at the head and base of the pile.
4.3. Solution of the governing equation of the thin-walled pipe pile Substituting Eq. (31) into Eq. (25), we can obtain the displace-
ment of the pile:
Performing the Fourier transform on Eq. (4) yields:
X
1
 
2 2
U pk ¼ C km ebkm z þ Dkm ebkm z cos ðmhÞ ð33Þ
Epk @ U pk Epk @ U pk m¼0
þ   þ qpk x2 U pk
1  m2pk @z2 2 1 þ mpk r20 @h2
F 1k r1k þ F 2k r2k The boundary conditions at the head and base of the thin-walled
¼ ð22Þ pipe pile can be expressed as:
hk r 0

where U pk ðz; h; xÞ is the Fourier transform of upk ðz; h; t Þ. Ep1 @U p1  PðxÞdðhÞ
¼ ð34Þ
Substituting Eqs. (16) and (19) into Eq. (22) yields: 1  m2p1 @z z¼0 h1 r 0

Epk @ 2 U pk Epk @ 2 U pk !
þ   2  qpk x2 U pk 
1  mpk @z
2 2
2 1 þ mpk r0 @h2 Ep3 ph3 r 0 @U p3 
þ kp U p3  ¼0 ð35Þ
1  m2p3 @z 
r 1k l1 X1
r 2k l2 X1
z¼H3
¼ Am g1km cos ðmhÞ þ Bm g2km cos ðmhÞ ð23Þ  
hk r 0 m¼0 hk r 0 m¼0
where PðxÞ is the Fourier transform of P 0 sin pT t HðT  tÞ and kp is
Assuming U pk ¼ Z pk ðzÞUpk ðhÞ, similarly with the above, we the stiffness coefficient at the base of the pipe pile. Militano and
obtain: Rajapakse [20] proposed to estimate the stiffness kp of a solid pile
from the static stiffness of a rigid disk resting on a half-space. It is
Upk ðhÞ ¼ cos ðmhÞ ð24Þ therefore reasonable to assume that the stiffness kp of a pipe pile
is equal to the stiffness of a ring foundation on a half-space [21]:
Thus U pk can be expressed as:
4l1 r 13
X
1 kp ¼ a ð36Þ
U pk ¼ Z pk cos ðmhÞ ð25Þ 1  v s1
m¼0
where ms1 is the Poisson’s ratio of the outer soil and a is a coefficient
Substituting Eq. (25) into Eq. (23) yields: defined by Veletsos and Tang [21] which depends on the ratio of the
inner radius to the outer radius of the pipe pile.
!
Epk @ 2 Z pk Epk m2 Substituting Eq. (31) into Eq. (34) yields:
2
    qpk x Z pk
1  m2pk @z2 2 1 þ mpk r 20 Ep1 X 1
PðxÞdðhÞ
b ðC 1m  D1m Þ cos ðmhÞ ¼  ð37Þ
r 1k l1 Am g1km þ r 2k l2 Bm g2km 1  m2p1 m¼0 1m h1 r0
¼ ð26Þ
hk r 0
The eigenfunction cos ðmhÞ (m – 0) has the following orthogonality:
From the requirement for continuity of displacements of the Z 2p 
0 ðm – nÞ
pipe pile and the outer and inner soil: cos ðmhÞ cos ðnhÞdh ¼ ð38Þ
0 p ðm ¼ nÞ
U 1 jr¼r1 ¼ U 2 jr¼r2 ¼ U pk ð27Þ
Multiplying both sides of Eq. (37) by cos ðmhÞ and subsequently
We can estimate the series of coefficients Am and Bm by substi- integrating over the interval ½0; 2p, we obtain:
tuting Eqs. (15), (18) and (23) into Eq. (27): C 1m  D1m ¼ nm PðxÞ ð39Þ
Z pk where
Am ¼ ð28Þ
K m ðq1 r 1 Þ 8
> m2 1
< 2ph rp1 b E m¼0
1 0 1m p1
Z pk nm ¼ ð40Þ
Bm ¼ ð29Þ >
: m2p1 1
I m ðq2 r 2 Þ ph1 b1m r0 Ep1 m>0

Substituting now Eqs. (28) and (29) into Eq. (26), we obtain: Substituting Eq. (31) into Eq. (35) yields:
! C 3m ¼ vm D3m ð41Þ
Epk @ 2 Z pk E m2
  pk   qpk x2 Z pk
1  m2pk @z2 2 1 þ mpk r 20 where
 
r 1k l1 g1km Im ðqr2k Þ þ r 2k l2 g2km K m ðqr1k Þ Epk ph3 r 0 b3m  kp 1  m2p3
¼ Z pk ð30Þ  e2b3m H3
hk r 0 K m ðqr1k ÞIm ðqr 2k Þ vm ¼  ð42Þ
Epk ph3 r 0 b3m þ kp 1  m2p3
The solution for Z pk results as:
The requirement for continuity of the resultant force and dis-
Z pk ¼ C km ebm z þ Dkm ebm z ð31Þ placement between consecutive segments at z ¼ H1 and z ¼ H2 is
expressed as:
54 C. Zheng et al. / Computers and Geotechnics 70 (2015) 50–59

8
< C 1m eb1m H1 þ D1m eb1m H1 ¼ C 2m eb2m H1 þ D2m eb2m H1 To obtain the response in the time domain, we employ an
E h b   E hb   inverse Fourier transform as:
: p111m2 1m C 1m eb1m H1  D1m eb1m H1 ¼ p212m2 2m C 2m eb2m H1  D2m eb2m H1
p1 p2
Z þ1
1
ð43Þ upk ðz; h; tÞ ¼ U pk ðz; h; xÞeixt dx ð58Þ
2p 1
8
< C 2m eb2m H2 þ D2m eb2m H2 ¼ C 3m eb3m H2 þ D3m eb3m H2 Z
E h b   E hb   þ1
1
: p212m2 2m C 2m eb2m H2  D2m eb2m H2 ¼ p313m2 3m C 3m eb3m H2  D3m eb3m H2 v pk ðz; h; tÞ ¼ V pk ðz; h; xÞeixt dx ð59Þ
p2 p3 2p 1
ð44Þ
where v pk is the time domain velocity response of the k-th pile
Eqs. (43) and (44) can be re-written in the following matrix segment.
form: The general solution of a defective pile degenerates to that of an
  intact pile by setting Ep1 ¼ Ep2 ¼ Ep3 ; mp1 ¼ mp2 ¼ mp3 ; qp1 ¼
C 1m C 2m
½T 1 ðH1 Þ ¼ ½T 2 ðH1 Þ ð45Þ qp2 ¼ qp3 and h1 ¼ h2 ¼ h3 . In this special case the displacement
D1m D2m
response of the pile head can be obtained from:
 
C 2m C 3m  
½T 2 ðH2 Þ ¼ ½T 3 ðH2 Þ ð46Þ X
1
vm þ 1 nm PðxÞ
D2m D3m Up ¼ cos ðmhÞ ð60Þ
m¼0
vm  1
where
" # If we set m ¼ 0, the proposed solution is further reduced to the
eb1m H1 eb1m H1 classical 1D solution, which will be used for comparison purposes
½T 1 ðH1 Þ ¼ Ep1 h1 b1m b H Ep1 h1 b1m b H ð47Þ
1m2
e 1m 1  1m2 e 1m 1 in the following discussion. In that case Eq. (60) becomes:
p1 p1
 
" # X
1
vm þ 1 nm PðxÞ
eb2m H1 eb2m H1 Up ¼ ð61Þ
½T 2 ðH1 Þ ¼ Ep2 h2 b2m Ep2 h2 b2m ð48Þ vm  1
1m2p2
eb2m H1  1m2p2
eb2m H1 m¼0

" #
eb2m H2 eb2m H2 5. Validation of the solution and discussion
½T 2 ðH2 Þ ¼ Ep2 h2 b2m Ep2 h2 b2m ð49Þ
1m2p2
eb2m H2  1m2p2
eb2m H2
In this section we demonstrate the validity of the derived solu-
" # tion via comparison with numerical results, and examine the vibra-
eb3m H2 eb3m H2
½T 3 ðH2 Þ ¼ Ep3 h3 b3m b H Ep3 h3 b3m b H ð50Þ tion response of a thin-walled pipe pile to impact point loads via a
1m2
e 3m 2  1m2 e 3m 2
p3 p3 series of arithmetical examples. The discussion is based on plots
presenting the velocity response time histories at the top of the
Eqs. (45) and (46) yields:
  pile. In order to accentuate the vibration characteristics of the pipe
C 3m w11 w12 C 1m pile, relatively low values of the shear moduli of the soils are con-
¼ ð51Þ
D3m w21 w22 D1m sidered in the examples. Unless otherwise mentioned, the follow-
ing problem parameters are used, corresponding to a typical pipe
where pile in soft soil: H = 10 m; r1 = 0.5 m; r2 = 0.4 m; G1 ¼ G2 =
w11 w12 1 MPa; n1 = n2 = 0.05; Ep = 25 GPa; q1 = q1 = 1800 kg/m3; qp =
¼ ½T 3 ðH2 Þ1 ½T 2 ðH2 Þ½T 2 ðH1 Þ1 ½T 1 ðH1 Þ ð52Þ 2500 kg/m3; mp = 0.15; P 0 = 1 kN; T = 0.001 s.
w21 w22
Combining Eqs. (41) and (51) yields:
5.1. Verification of the proposed solution
C 1m ¼ dm D1m ð53Þ

where dm ¼ vwm w22vww12 . We can now substitute Eq. (53) into Eq. (39) to First we compare the results of the proposed solution against
11 m 21
3D dynamic finite element analysis performed with the software
obtain the values of C 1m and D1m :
ABAQUS. Eight-node, isoparametric finite elements are used to
model the thin-walled pipe pile and viscoelastic soil. 3D infinite
dm nm PðxÞ
C 1m ¼ ð54Þ elements (type CIN3D8) that act as ‘‘quiet’’ boundaries are used
dm  1 to reduce the extents of the mesh while preventing wave reflec-
tions. Mass and stiffness Rayleigh damping with parameters a =
nm PðxÞ 0.5196 and b = 0.0036 is used to approximate the hysteretic damp-
D1m ¼ ð55Þ ing of the soil considered in the analytical solution. The mesh size
dm  1
is sufficiently small in the area of interest, to capture the propaga-
The remaining coefficients can then be determined by Eqs. (45)
tion of high-frequency stress waves (Fig. 4). The comparison in
and (46). Substituting into Eq. (33), we can obtain the vertical dis-
terms of the velocity response at a polar angle h = 90° from the
placement of the pile at any position. The displacement response of
point of application of the load is depicted in Fig. 5. Observe that
the pile head, which is of interest here, is expressed as:
the analytical solution matches well the numerical FEM results
X
1
ðdm þ 1Þnm P ðxÞ (Fig. 5a), although the latter are richer in high frequency content
U p1 ¼ cos ðmhÞ ð56Þ due to mesh discretization of the pile-soil system. This phe-
m¼0
dm  1
nomenon has been observed and discussed by Liao and Roesset
The velocity of the pile in frequency domain V pk ðz; h; xÞ is [22] and Fei et al. [23]. If we apply a low-pass filter with cut-off fre-
obtained through differentiation of the displacement as: quency of 2000 Hz [22,23] the unrealistic high frequency content
disappears, as shown in Fig. 5b. In that case the analytical solution
V pk ðz; h; xÞ ¼ ixU pk ðz; h; xÞ ð57Þ is practically identical to the filtered FEM results.
 C. Zheng et al. / Computers and Geotechnics 70 (2015) 50–59 55

vibration mode (m = 0) is axisymmetric, while higher modes (m >

0) are asymmetric. Fig. 7 depicts the velocity response of the first
asymmetric vibration mode v 1 (m = 1) at different polar angles
along the circumference of the pipe pile. Note that vibration ampli-
tudes of the first mode are remarkably high at polar angles h = 0°,
45°, 135° and 180°. This suggests that high-frequency interference
is mainly attributed to the first asymmetric vibration mode v 1 . This
is further supported by the fact that the velocity of the first mode
of asymmetric vibration is zero at a polar angle h = 90°, as it is a
function of cos(h). Another direct consequence of this is that the
phases at polar angles h = 0° and 45° points are opposite to those
at polar angles h = 135° and 180° (Fig. 7). Attenuating amplitudes
of the interference also follow the cosine pattern, and are lower
at h = 45° and 135° by 0.707 times compared to h = 0° and 180°.
Figs. 8–10 present velocity time histories attributed to the
axisymmetric and main asymmetric vibration modes at polar
angles h = 0°, 45° and 90°. As expected, the contribution of the
vibration mode v 1 to the velocity response is much more pro-
nounced, compared to higher asymmetric modes. This further sug-
gests that the vibration mode v 1 is the main source of the
high-frequency interference. Note that v 2 at h = 45° and v 1 and
v 3 at h = 90° are all zero as they are functions of cos(mh). The
high-frequency interference at h = 90° mainly results from the sec-
ond mode of vibration v 2 (Fig. 10). Since the relative amplitude is
rather trivial, placing the receiving point at 90° from the point of
impact will provide a signal with minimum interferences. The time
histories plotted on Figs. 8–10 also suggest that the pipe pile vibra-
tion modes m > 0 have little influence on the reflected wave.
Fig. 4. The 3D FEM mesh used for the numerical simulation of the problem.

0.6
5.2. Insights on the 3D vibration response of the thin-walled pipe pile
(a) Analytical solution
0.5 Raw FEM results
Next, by analyzing the vibration response at different points 0.4
along the perimeter of the pipe pile, we will attempt to investigate
0.3
Velocity (mm/s)

the propagation of impact vibrations through the pipe pile–soil

system and assess implications when interpreting low-strain 0.2
integrity tests. Fig. 6 presents the velocity time histories at the
0.1
head of the pile, obtained at polar angles h = 0°, 45°, 90°, 135°
and 180°. Notice that the response is highly non-uniform along 0.0
the cross-section. As expected, the peak velocity value is measured
-0.1
exactly at the point where the load is applied (h = 0°). The incident
wave propagates along the circumferential direction, as depicted -0.2
by the latency in the arrival times, and its amplitude varies along -0.3
the cross-section. This time lag of the incident wave crests should 0 1 2 3 4 5 6 7 8
be properly considered when using low-strain testing to determine Time (ms)
the length of the pile, which is calculated from the time difference
between the incident and reflected waves.
(b) 0.6
It is also found that some high-frequency secondary interfer- Analytical solution
0.5
ence follow the incident wave, whose amplitude attenuates with Filtered FEM results
time until the reflected wave arrives. The amplitude of these oscil- 0.4
lations is a minimum at h = 90°, compared to the rest of the control
0.3
Velocity (mm/s)

points along the cross-section, where they reach clear peaks. These
oscillations are attributed to the flexural vibration of the 0.2
cross-section of the pile, which is axisymmetric with respect to h 0.1
= 90°. As we move our hypothetical receiver away from h = 90°,
the amplitude of the oscillations increases. This high-frequency 0.0
interference may compromise the interpretation of measurements -0.1
for the assessment of the integrity of a pile. Therefore, during
-0.2
low-strain testing for the purpose of assessing the integrity of
the pile, the receiving point should be placed at 90° from the point -0.3
of hammer impact. 0 1 2 3 4 5 6 7 8
Insight on the origin of high-frequency interferences can be Time (ms)
obtained by considering the vibration modes of the pipe pile–soil
Fig. 5. Comparison of velocity time histories obtained by the proposed solution and
system. The dynamic response of the thin-walled pipe pile results from FEM analysis (h = 90°): (a) raw FEM results; (b) filtered FEM results with cut-
from the superposition of m vibration modes. The fundamental off frequency of 2000 Hz.
56 C. Zheng et al. / Computers and Geotechnics 70 (2015) 50–59

2.5 =0 0.6
2.0 =45
0.4
=90 =0
1.5 Incident wave crest =135 0.2
1.0 Reflected wave crest =180
Velocity (mm/s)

Velocity (mm/s)
0.5 0.0

0.0 -0.2
-0.5 v0
-0.4
-1.0 v1
High-frequency interferences -0.6 v2
-1.5
-0.8 v3
-2.0
-2.5 -1.0
0 1 2 3 4 5 6 7 8 0 1 2 3 4 5 6 7 8
Time (ms) Time (ms)

Fig. 6. Velocity time histories at the pile head at different points along its cross- Fig. 8. Contribution of the main vibration modes to the velocity response at h = 0°.
section defined through the polar angle h.

0.5 v0
1.0
=0 0.4 =45 v1
=45 0.3 v3
=135
0.2
0.5 =180
Velocity (mm/s)
0.1
Velocity (mm/s)

0.0

0.0 -0.1
-0.2
-0.3
-0.5 -0.4
-0.5
-0.6
-1.0 0 1 2 3 4 5 6 7 8
0 1 2 3 4 5 6 7 8 Time (ms)
Time (ms)
Fig. 9. Contribution of the main vibration modes to the velocity response at h = 45°.
Fig. 7. Variation of the velocity response of fundamental vibration mode v 1 (m = 1)
at different points along the cross-section, defined through the polar angle h.

0.5
v0
6. Effect of the pipe pile and soil properties on the velocity 0.4 =90 v2
response v4
0.3

In low-strain integrity testing, the pile length can be simply 0.2

Velocity (mm/s)

estimated from the measured time difference in the arrival time 0.1
between the incident and reflected waves:
0.0

C p Dt -0.1
H¼ ð62Þ
2 -0.2

-0.3
where C p is the compressional wave propagation velocity of the pile
material; and Dt is the time difference between the incident and -0.4
0 1 2 3 4 5 6 7 8
reflected waves. As mentioned earlier, there is also a time lag due
Time (ms)
to waves traveling along the circumferential direction of
thin-walled pipe piles, which can affect the estimation of Dt from Fig. 10. Contribution of the main vibration modes to the velocity response at h =
velocity recordings. Figs. 11 and 12 present the influence of 90°.
Poisson’s ratio and Young’s modulus of the pile material on the time
difference between the incident and reflected waves, as a function
of the polar angle measured from the point of impact h = 0°. It is piles, which feature a higher compressional wave propagation
clear that the time lag increases as we move the hypothetical recei- velocity.
ver away from the point of impact, caused by the delay of waves Fig. 13 depicts the influence of the Young’s modulus of the pipe
traveling along the circumferential direction. This time lag is a func- pile material on the velocity response at h = 90°, where the ampli-
tion of the pipe pile material properties, which is expected as the tude of high-frequency interferences reaches a minimum;
wave propagation velocity depends of the properties of the mate- although for less stiff pipe piles, this interference still has signifi-
rial. Therefore this phenomenon will be less prominent in stiffer cant amplitudes. Indeed, the less stiff pipe pile materials result in
 C. Zheng et al. / Computers and Geotechnics 70 (2015) 50–59 57

6.6 0.6
p
=0.15 r1=0.5m
6.5
p
=0.2 0.5 r1=0.6m
6.4 p
=0.25 0.4 r1=0.7m
Time difference (ms)

6.3
0.3

Velocity (mm/s)
6.2
0.2
6.1
0.1
6.0
5.9 0.0

5.8 -0.1
5.7 -0.2
5.6
0 45 90 135 180 225 270 315 360 -0.3
0 1 2 3 4 5 6 7 8
Angle ( ) Time (ms)
Fig. 11. Influence of the Poisson’s ratio of the pile material on the time difference Fig. 14. Influence of the outer radius of the pile on the velocity response at polar
between the incident and reflected waves for different points along the cross- angle h = 90°.
section.

remarkably as the Young’s modulus increases. A clear shift is also

noticed on the arrival time of the reflected wave, due to the
11
=10GPa
p
increased wave propagation velocity of the material of the pipe
=20GPa
p
pile.
10
=30GPa
p
Fig. 14 presents the influence of the outer radius of the pile on
the velocity response at h = 90° while keeping the thickness of
Time difference (ms)

9 the pile wall, h constant. As the outer radius increases, the

high-frequency interference increases significantly, due to the
8 increased flexural vibration of the pile cross-section. It is also noted
that the arrival time of the incident wave increases, while that of
7 the reflected wave does not vary significantly. Furthermore, the
peak amplitudes of the incident and reflected waves decrease
6 slightly.
Fig. 15 presents the influence of the thickness of the pile wall on
the velocity response at h = 90° while keeping the outer radius, r 1
5
0 45 90 135 180 225 270 315 360 constant. As expected from the previous discussion, the
Angle ( ) high-frequency interference increases with decreasing thickness
of the pile. Further to that, as the thickness of the pile decreases,
Fig. 12. Influence of the Young’s modulus of the pile material on the time difference the peak value of the incident wave increases, while that of the
between the incident and reflected waves for different points along the cross-
reflected wave decreases.
section.
Finally, we investigate the influence of the shear moduli of the
outer and inner soil on the velocity response at polar angle h = 90°
1.2 (Fig. 16). Notice that the shear moduli of the soils have negligible
Ep=10GPa influence on the high-frequency interference. However, the ampli-
1.0 Ep=20GPa tude of the reflected wave decreases as the shear moduli of the
0.8 Ep=30GPa soils increase. This is due to the fact that the stiffer the soil, the
0.6
Velocity (mm/s)

0.4
1.0
0.2 h=0.1m
h=0.075m
0.0 0.8
h=0.05m
-0.2
0.6
-0.4
Velocity (mm/s)

-0.6 0.4
-0.8
0 2 4 6 8 10 12 0.2
Time (ms)
0.0
Fig. 13. Influence of the Young’s modulus of the pile material on the velocity
response at polar angle h = 90°. -0.2

-0.4
lower flexural stiffness and thus increased flexural vibrations of the 0 1 2 3 4 5 6 7 8
Time (ms)
cross-section of the pile, and consequently, higher levels of inter-
ference. Additionally, as a result of the increased impedance of Fig. 15. Influence of the thickness of the pile on the velocity response at polar angle
the pipe pile, the peak amplitude of the incident wave decreases h = 90°.
58 C. Zheng et al. / Computers and Geotechnics 70 (2015) 50–59

0.6 2.5
G1=1MPa, G2=1MPa Present solution ( =0
2.0
0.5 G1=1MPa, G2=2MPa 1D solution
G1=2MPa, G2=1MPa
Velocity response (mm/s)

1.5
0.4
1.0

Velocity (mm/s)
0.3
0.5
0.2
0.0
0.1
-0.5
0.0
-1.0
-0.1 -1.5

-0.2 -2.0
0 1 2 3 4 5 6 7 8 0 1 2 3 4 5 6 7 8
Time (ms) Time (ms)

Fig. 16. Influence of the shear moduli of the outer and inner soil on the velocity Fig. 18. Comparison between the present solution at h = 0° and the 1D solution.
response at polar angle h = 90°.

0.6
Present solution ( =90
more resistance is provided to the pile vibrations, and therefore the 0.5
1D solution
amount of wave energy dissipated towards the soil increases. It is
0.4
also found that the influence of the shear modulus of the outer soil
is larger than that of the inner soil. This suggests that the portion of Velocity (mm/s) 0.3
resistance contributed by the outer soil is higher than that by the 0.2
inner soil.
0.1

0.0

7. Application: Low-strain testing of a defective pipe pile -0.1

-0.2
As mentioned earlier, the proposed method can be employed
for the interpretation of low-strain tests in defective pipe piles. -0.3
0 1 2 3 4 5 6 7 8
To demonstrate this, we consider a 10 m-long pipe pile of thickness Time (ms)
0.1 m, which exhibits a neck at the depth between 3 and 4 m,
where the thickness of its cross-section is locally reduced to Fig. 19. Comparison between the present solution at h = 90° and the 1D solution.
0.05 m (Fig. 17). The remaining problem parameters are the same
as in section 5. In Figs. 18 and 19 we compare results obtained with 8. Concluding remarks
the proposed solution at h = 0° and 90° with results obtained with
the 1D solution. Observe in Fig. 18 that the defect cannot be accu- We presented an analytical method to calculate the dynamic
rately distinguished from the high-frequency interferences at h = response of a thin-walled pipe pile to a vertical transient point load
0°. Even the wave reflected at the pile toe is practically indistin- applied during a low-strain integrity test, which considers coupling
guishable, which suggests that velocity time histories obtained at of the pile and viscoelastic soil vibrations. Our formulation of the
h = 0° are of little value for determining the integrity and as-built problem allows for stress wave propagation along both the vertical
length of the pipe pile. On the contrary, the 3D and 1D solutions and circumferential directions. This enables quantification of the
are almost identical at the optimal position h = 90° of the hypothet- latency of the recorded incident wave, which chiefly depends on
ical receiver (Fig. 19), from where both the defect and pile length the properties of the pile material. Taking account of this time
can be deduced. lag during the interpretation of low-strain integrity tests will
improve the accuracy of their results. We also showed that
high-frequency interferences, attributed to the first asymmetric
mode of vibration of the cross-section, may complicate the inter-
pretation of velocity measurements. This effect is more prominent
in flexible piles, where the arrival time of the reflected wave can-
h1=0.1m not be easily identified. To reduce it, the receiver (accelerometer)
H 1=3m should be placed at a polar angle of h = 90° relatively to the point
H 2=4m h2=0.05m of impact, where the amplitude of these oscillations is minimum.
Good agreement with finite element analysis results suggests that
the proposed computational method can be used as a benchmark-
ing tool for facilitating the interpretation of low-strain integrity
h3=0.1m tests for thin-walled pipe piles.

Acknowledgements
H 3=10m
This work was supported by the 111 project (Grant No. B13024)
Fig. 17. Pipe pile with a necking defect considered in the application. and the National Natural Science Foundation of China (Grant Nos.
 C. Zheng et al. / Computers and Geotechnics 70 (2015) 50–59 59

51378177 and 51420105013). The first author would like to [11] Xu XT, Liu HL, Lehane BM. Pipe pile installation effects in soft clay. Geotech Eng
2006;159(4):285–96.
acknowledge the support from the ARC Centre of Excellence for
[12] Gazis DC. Three-dimensional investigation of the propagation of waves in
Geotechnical Science and Engineering. hollow circular cylinders: I. Analytical foundation. J Acoust Soc Am
1959;31(5):568–73.
References [13] Gazis DC. Three-dimensional investigation of the propagation of waves in
hollow circular cylinders: II. Numerical results. J. Acoust Soc Am
1959;31(5):573–8.
[1] Smith EAL. Pile driving analysis by the wave equation. J Soil Mech Found Div [14] Ding XM, Liu HL, Liu JY, Chen YM. Wave propagation in a pipe pile for low-
1960;86(4):35–64. strain integrity testing. J Eng Mech ASCE 2011;117(9):598–609.
[2] Goble GG, Likins GE, Rausche F. Bearing capacity of piles from dynamic [15] Ding XM, Liu HL, Zhang B. High-frequency interface in low strain integrity
measurements. Final Rep. Dept of Civil Engineering, Case Western Reserve testing of large-diameter pipe pile. Sci China Technol Sci 2011;54:420–30.
Univ, Cleveland; 1975. p. 41–5. [16] Ding XM, Liu HL, Kong GQ, Zheng CJ. Time-domain analysis of velocity waves
[3] Rausch F, Moses F, Goble GC. Soil resistance prediction from pile dynamics. J in a pipe pile due to a transient point load. Comput Geotech 2014;58:101–16.
Soil Mech Found Div 1972;98(9):917–37. [17] Zheng CJ, Kouretzis GP, Ding XM, Liu HL, Poulos HG. Three-dimensional effects
[4] Morgano CM. Determining embedment depths of deep foundations using non- in low strain integrity testing of piles: analytical solution. Can Geotech J 2015.
destructive methods. In: Proceedings of the 5th international conference on http://dx.doi.org/10.1139/cgj-2015-0231.
the application of stress-wave theory to piles, Orlando, FL; 1996. p. 734–47. [18] Novak M, Nogami T, Aboul-Ella F. Dynamic soil reactions for plane strain case. J
[5] Wang KH, Wu WB, Zhang ZQ, Chin JL. Vertical dynamic response of an Eng Mech Div ASCE 1978;104(4):953–9.
inhomogeneous viscoelastic pile. Comput Geotech 2010;37(4):536–44. [19] Chow YK, Phoon KK, Chow WF, Wong KY. Low strain integrity testing of piles:
[6] Ni SH, Lehmann L, Charng JJ. Low-strain integrity testing of drilled piles with three-dimensional effects. J Geotech Geoenviron Eng 2003;129(11):1057–62.
high slenderness ratio. Comput Geotech 2006;33(6–7):283–93. [20] Militano G, Rajapakse RKND. Dynamic response of a pile in a multi-layered soil
[7] Ni SH, Lo KF, Lehmann L, Huang YH. Time–frequency analyses of pile-integrity to transient torsional and axial loading. Geotechnique 1999;49(1):91–109.
testing using wavelet transform. Comput Geotech 2008;35(4):600–7. [21] Veletsos AS, Tang Y. Vertical vibration of ring foundations. Earthq Eng Struct
[8] Liu HL, Fei K, Xu XT. Development and application of the large-diameter driven Dynam 1987;15(1):1–21.
cast-in-place concrete thin-wall pipe pile. In: Proc 16th Int Conf on Soil Mechanics [22] Liao ST, Roesset JM. Dynamic response of intact piles to impulse loads. Int J
and Geotechnical Engineering. Amsterdam, Netherlands: IOS Press; 2005. Numer Anal Meth Geomech 1997;21(4):255–75.
[9] Liu HL, Chu J, Deng A. Use of large-diameter cast–in situ concrete pipe piles for [23] Fei K, Liu HL, Zhang T. Three dimensional effects in low strain integrity test of
embankment over soft clay. Can Geotech J 2009;46(8):915–27. PCC pile. Rock Soil Mech 2007;28(6):1095–102.
[10] Liu HL, Ng CWW, Fei K. Performance of a geogrid-reinforced and pile-
supported highway embankment over soft clays: a case study. J Geotech
Geoenviron Eng ASCE 2007;133(12):1483–93.