You are on page 1of 16

INTERNATIONAL JOURNAL FOR NUMERICAL METHODS IN ENGINEERING

Int. J. Numer. Meth. Engng. 46, 713}728 (1999)


VIBRATION ISOLATION BY A ROW OF PILES USING
A 3-D FREQUENCY DOMAIN BEM
S. E. KATTIS, D. POLYZOS AND D. E. BESKOS*
Department of Mechanical and Aeronautical Engineering, ;niversity of Patras, GR-26500 Patras, Greece
Department of Civil Engineering, ;niversity of Patras, GR-26500 Patras, Greece
SUMMARY
The problem of vibration isolation by a row of piles is numerically solved in a three-dimensional context by
an advanced frequency domain boundary element method (BEM). Both the piles and the soil are modelled
by boundary elements and coupled together through equilibrium and compatibility at their interfaces.
Linear elastic or viscoelastic material behaviour is assumed for both the piles and the soil. The piles can be
tubular or solid and have circular or square cross-section. The vibration source is a vertical harmonically
varying with time force and the row of piles acts as a wave barrier in a passive way. The boundary element
method is "rst validated for accuracy by solving two three-dimensional wave di!raction problems dealing
with spheres and trenches as scatterers for which there are analytical and highly accurate numerical
solutions available in the literature. Numerical examples dealing with passive vibration isolation by a row of
piles are then solved and the screening e!ectiveness of these wave barriers is assessed and compared against
that of trenches. Copyright 1999 John Wiley & Sons, Ltd.
KEY WORDS: boundary element method; vibration isolation; pile wave barrier; screening wave e!ectiveness; wave
di!raction
1. INTRODUCTION
Vibration isolation of structures from ground transmitted waves generated by machines, tra$c or
blasting is accomplished actively or passively by various types of barriers. These barriers di!ract
the surface waves radiated from the vibration source and su$ciently reduce their amplitude.
Open or in"lled trenches, sheet piles, buried concrete plates or a row of solid or tubular piles are
typical types of wave barriers.
The classical books of Barkan [1] and Richart et al. [2] describe practical empirical and
experimental methods, respectively, for achieving e!ective wave screening by trenches or sheet
piles. However, vibration isolation problems can only be solved e$ciently and with a high degree
of accuracy by numerical methods, such as the Finite Element Method (FEM) and especially the
Boundary Element Method (BEM). The FEM has been used, e.g. by Haupt [3], Segol et al. [4],
May and Bolt [5] and Yang and Hung [6] for two-dimensional cases of vibration isolation by
* Correspondence to: D. E. Beskos, Department of Civil Engineering, School of Engineering, University of Patras,
GR-26500 Patras, Greece. E-mail: d.e.beskos@upatras.gr
Contract/grant sponsor: Greek General Secretariat for Research and Technology; Contract/grant number: 91 EA 386
CCC 0029-5981/99/290713}16$17.50 Received 1 June 1998
Copyright 1999 John Wiley & Sons, Ltd. Revised 8 February 1999
trenches, usually in connection with special non-re#ecting boundaries. The BEM is the ideal
method for this class of problems since it requires only a surface discretization and satis"es the
radiation conditions at in"nity automatically. It has been used, e.g. by Emad and Manolis [7],
Beskos et al. [8, 9], Leung et al. [10, 11], Beskos and Vardoulakis [12], Von Estor! and Antes
[13], Ahmad and Al-Hussaini [14], Al-Hussaini and Ahmad [15], Stamos et al. [16] and Mateo
and Alarcon [17] to successfully solve in the frequency or time domain two-dimensional
vibration isolation problems involving open or in"lled trenches in homogeneous or layered soils.
Three-dimensional cases of vibration isolation by open or in"lled trenches have also been solved
by the frequency domain BEM, e.g. by Beskos and Vardoulakis [18], Banerjee et al. [19],
Dasgupta et al. [20], Klein et al. [21, 22], Ahmad et al. [23] and Al-Hussaini and Ahmad [24].
The review articles of Beskos [25, 26] on the BEM as applied to dynamic problems contain
a complete bibliography on vibration isolation up to the year 1996.
Use of trenches as wave barriers is restricted to cases involving small to medium trench depth
because of soil instability and water table level problems. When the transmitted waves, primarily
Rayleigh surface waves, have long wavelengths, trenches cannot be used as e!ective wave barriers.
Thus, when the required barrier depth is very large, a series of piles is the solution. Woods et al.
[27] and Liao and Sangrey [28] were the "rst to study experimentally the problem of the
screening e!ectiveness of piles and provide some design guidelines. The problem of vibration
isolation by a row of piles is a truly three-dimensional multiple wave di!raction problem di$cult
to solve even by numerical methods. Aviles and Sanchez-Sesma [29, 30] and Baroomand and
Kaynia [31] have presented approximate analytical solutions of the problem, which, however, in
view of the various simplifying assumptions involved and the complexity of the problem, seem
only qualitatively correct.
In this work the problem of vibration isolation by a row of piles is solved for the "rst time by
a highly accurate and e$cient BEM in the frequency domain in a three-dimensional context. The
BEM is used to model both the piles and the soil, which are both assumed to be linearly elastic or
viscoelastic materials. The soil and pile domains are coupled together through equilibrium and
compatibility at their interfaces. The dynamic behaviour of foundation piles has been studied in
the past by using the BEM for the soil and the FEM for the piles treated as one-dimensional beam
structures [32]. However, this pile modelling and the imperfect matching between soil and piles
do not provide accurate results, while the symmetry advantage of the FEM equations is
destroyed when BEM and FEM are coupled together. Use of the BEM for both the soil and the
piles ensures perfect matching of the two kinds of domains and results of high accuracy. The piles
can be tubular or solid and have circular or square cross-sections. The vibration source is
a vertical concentrated force harmonically varying with time, while the row of piles acts as a wave
barrier in a passive way.
The BEM program used in this work employs various types of boundary elements (here
eight-noded quadratic quadrilateral boundary elements are used) continuous or partially discon-
tinuous and treats singular integrations directly [33] in a highly accurate manner. In addition the
program has symmetry, antisymmetry and substructuring capabilities, while it can be interfaced
to commercial pre- and post-processors. The full space fundamental solution is used in the
formulation, which requires a free soil surface discretization. However, this discretization is
restricted to a "nite part around the area of interest and is needed anyhow in vibration isolation
analysis where surface displacements are needed everywhere inside the area of interest.
The computer program is "rst tested for accuracy by comparing its results pertaining to two
three-dimensional wave di!raction problems involving spheres and trenches as scatterers against
714 S. E. KATTIS, D. POLYZOS AND D. E. BESKOS
Copyright 1999 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Engng. 46, 713}728 (1999)
highly accurate results available in the literature. Numerical examples dealing with passive
vibration isolation by a row of piles are then solved and the screening e!ectiveness of these wave
barriers is assessed and compared against that of trenches.
Preliminary reports on the above program and some early results of vibration isolation by pile
problems have been presented by the authors in conferences [34, 35]. This paper presents
a complete and much extended account of both the BEM program and the vibration isolation by
piles analysis and is based on the doctoral dissertation of Kattis [36].
2. ADVANCED 3-D FREQUENCY DOMAIN BEM COMPUTER CODE
The governing equations of motion in isotropic and homogeneous linear elastodynamics with
zero body forces expressed in terms of displacements have the following form in the frequency
domain [32]:
ju
GHH
#(z#j) u
HHG
#jcu
G
"0 (1)
where u
G
"u
G
(x, c) is the amplitude of the displacement vector, x is a point of the elastic body
with volume <, c is the circular frequency of vibration, j is the mass density of the body, z and
j are the Lame' elastic constants, summation is assumed over repeated indices and i, j"1, 2, 3.
Viscoelastic material behaviour can be easily taken into account by simply replacing z and j by
their complex counterparts z(1#i2[) and j(1#i2[), respectively, where 2[ is the viscoelastic
damping coe$cient [32]. In a well-posed problem, equation (1) is accompanied by appropriate
boundary conditions involving displacements u
G
and tractions t
G
on the boundary S"S
S
S
R
of
the body, which read
u
H
(x, c)"uN
G
(x, c), x3S
S
t
G
(x, c) "t
N
G
(x, c), x3S
R
(2)
where the overbars denote prescribed values.
Following the reciprocal theorem or the method of weighted residuals, one can achieve an
integral representation of the solution of equation (1) in the form [32]
c
GH
(x) u
H
(x)"

1
u*
GH
(x, y) t
H
(y) dS(y)!

1
t*
GH
(x, y) u
H
(y) dS(y) (3)
where u*
GH
and t*
GH
are the fundamental displacement and traction tensors, respectively, x and y are
the "eld and source points, respectively, and the jump tensor or free term c
GH
(x)"0 for x , <S,
c
GH
(x)"o
GH
for x3<and c
GH
(x)"(

)o
GH
for x3S and being smooth or c
GH
(x) is a function of the local
geometry at x when the surface at that point is not smooth [37]. When x3S equation (3) becomes
a boundary integral equation to be solved numerically by the BEM in terms of the unknown
vectors u
G
and t
G
on the boundary for given boundary data. When xPy on S or r"x!yP0,
tensors u*
GH
and t*
GH
become singular. Thus for three-dimensions one has [32]
u*
GH
"O(1/r), t*
GH
"O(1/r) (4)
For the solution of equation (3) when x3S, the surface S is discretized into a "nite number of
boundary elements and through a collocation process one receives the matrix equation
[;] t"[] u (5)
VIBRATION ISOLATION BY A ROW OF PILES 715
Copyright 1999 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Engng. 46, 713}728 (1999)
where [;] and [] are in#uence matrices with entries integrals over boundary elements with
integrands involving the fundamentals tensors u*
GH
and t*
GH
, respectively, while t and u denote
the vectors of the nodal boundary tractions and displacements, respectively. Employment of the
boundary conditions (2) in equation (5) and subsequent rearrangement to separate known from
unknown boundary nodal values results in the equation
[A] x"b (6)
to be solved for the unknown nodal values of displacements and tractions constituting the vector
x in terms of the known quantities constituting the vector b. When a problem involves more
than two elastic bodies in full contact, as it is the present soil}pile system, equation (5) is written
for every body in partitioned form and the resulting equations are coupled together through
equilibrium and compatibility at their interface [32].
Interior displacements and stresses as well as stresses on the boundary can also be determined.
The latter ones are evaluated highly accurately in a direct manner from the expressions for the
interior stresses by employing Huber's [38] algorithm to deal with the hypersingular integrals
involved. Thus, the classical approximate way for evaluating boundary stresses through the use of
boundary tractions and tangential displacement derivatives [32] is not needed any more. Of
course, in the present vibration isolation problems there is no need of computing either interior
displacements and stresses or boundary stresses.
An advanced computer code in Fortran 90 has been constructed for the accurate solution of
three-dimensional linear elastic systems in the frequency domain [34, 36] by following the
aforementioned BEM procedure. The main features of this code, characterized by high accuracy
and e$ciency, are the following:
(i) Use of linear or quadratic, triangular or quadrilateral boundary elements. These elements
can be continuous or discontinuous (totally or partially). Partially discontinuous elements
[39] are used to accommodate edges, corners, interfaces and discontinuities in the
boundary conditions.
(ii) Evaluation of singular and hypersingular integrals is done by the direct very accurate and
e$cient algorithm of Guiggiani [33] and Huber et al. [38], respectively. Thus, there is no
need to employ enclosing elements which have to be used when the evaluation of singular
integrals is done by using the rigid-body motion concept [26]. The evaluation of regular
integrals is done by Gauss quadrature and variable integration order which depends on
the distance r. In order to evaluate nearly singular integrals with accuracy, a method based
on Bu [40] is used.
(iii) Substructuring as well as symmetry and antisymmetry capabilities, which drastically
reduce the size of a large-order problem.
(iv) E$cient algorithm for solving systems of linear equations involving dense non-symmetric
complex matrices either in-core [41] or out-of-core [42].
(v) Interior and boundary displacements and stresses are accurately determined.
(vi) Interfaces to commercial pre- and post-processors for the convenience of the user.
In some respects the present computer code is similar to the code of Klein et al. [21, 22], apart
from the facts that here: (i) Geometric and boundary data discontinuities are treated in a more
general manner by discontinuous elements [39] (elements with collocation points inside rather
than on the element edges) and not by the employment of the free term c
GH
[37]. Thus, one can
716 S. E. KATTIS, D. POLYZOS AND D. E. BESKOS
Copyright 1999 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Engng. 46, 713}728 (1999)
treat corners, edges interfaces and discontinuities in the boundary conditions in a easy way
without serious numerical problems. (ii) Interior displacements and stresses as well as boundary
stresses can be calculated in addition to boundary displacements and tractions. The boundary
stresses are calculated using a direct highly accurate method [38] and (iii) More than one region
can be treated. The coupling between the di!erent regions is accomplished through equilibrium
and compatibility at their interfaces. More details about the present computer code can be found
in the doctoral dissertation of Kattis [36].
3. VALIDATION OF THE DYNAMIC BEM CODE
The frequency domain BEM code developed in the previous section for the dynamic analysis of
complex three-dimensional elastic or viscoelastic structural systems is validated in this section by
checking its accuracy on the basis of two problems for which analytical and highly accurate
numerical results are available in the literature.
Consider "rst a spherical cavity or elastic inclusion of radius : in the in"nite elastic
space di!racting an incident compressional P harmonic wave. The elastic space medium
is characterized by a P wave propagation velocity c

, a Poisson's ratio v and a mass density j,


while the elastic inclusion by corresponding quantities c

, v and j. For a wave di!rac-


tion problem equation (3) is augmented in its right-hand side by the term u
G
, indicating
the amplitude of the incident free wave "eld [36], while u
H
and t
H
denote the amplitudes of the total
displacement and traction "elds connected with the corresponding scattered ones u
G
and t
H
by
the relations u
H
"u
H
#u
H
and t
H
"t
H
#t
H
, with t
H
obtained from u
H
through Hookes'
law [36].
Figure 1. Discretization of one-quarter of the sphere surface for the example of the wave scattering by a sphere
VIBRATION ISOLATION BY A ROW OF PILES 717
Copyright 1999 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Engng. 46, 713}728 (1999)
Figure 2. Amplitude of radial displacements versus the scattering angle for the example of the wave scattering by a sphere
Figure 3. Amplitude of tangential displacements versus the scattering angle for the example of the wave scattering by
a sphere
For the solution of this problem by the present BEM code the discretization of

(due to
symmetry) of the spherical surface of the scatterer involving 14 quadratic quadrilateral eight-
noded boundary elements as shown in Figure 1 was used in the analysis. Figures 2 and 3 show the
amplitude of the radial and tangential displacement, respectively, versus the di!raction angle 0 for
k

:"c:/c

"1)3, v"v"0)25, j/j"1)0 and for the cases of a cavity, a soft inclusion
(c

/c

"0)5) and a hard inclusion (c

/c

"2)0) as obtained for the mesh of Figure 1 by the


present BEM as well as by analytic means [43]. In all cases the agreement between the numerical
and the analytical results is excellent.
718 S. E. KATTIS, D. POLYZOS AND D. E. BESKOS
Copyright 1999 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Engng. 46, 713}728 (1999)
Figure 4. Vertical section and top view of the active vibration isolation system
Consider next a rigid square foundation of side w bonded on an elastic half-space and
experiencing vertical vibrations under the in#uence of a vertical harmonically varying with time
load of magnitude P

"1 KN and operational frequency 50 Hz as shown in Figure 4. The motion


of the foundation generates waves (mainly Rayleigh waves) in the half-space which have to be
reduced in amplitude to mitigate the disturbance to nearby structures. To this e!ect the
foundation is surrounded by an open rectangular trench of width b and depth h at a distance r
from the foundation centre as shown in Figure 4. The material properties of the elastic half-space
soil medium are shear modulus G"132 MPa, Poisson's ratio v"0)25, mass density
j"17)5 Kg/m and damping coe$cient ["6 per cent. For these data one can obtain a Rayleigh
wavelength z
0
"5 m. Thus, the geometry of this active vibration isolation system in non-
dimensional terms reads ="w/z
0
"0)2, B"b/z
0
"0)06, H"h/z
0
"0)5 and R"r/z
0
"0)4.
Due to the symmetry of the problem, only

of it is considered in the analysis and thus the soil


discretization using 224 eight-noded quadratic quadrilateral boundary elements (827 functional
nodes) is shown in Figure 5. The free soil surface discretization extents to a distance approxi-
mately 4z
0
from the source of vibration. In the computations the foundation is replaced by just
the vertical harmonic load as it was found in Kattis [36] that its omission does not practically
VIBRATION ISOLATION BY A ROW OF PILES 719
Copyright 1999 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Engng. 46, 713}728 (1999)
Figure 5. Discretization of one-quarter of the active vibration isolation system
Figure 6. Amplitude reduction factor versus distance across the diagonal OA for the active vibration isolation system
a!ect the vibration results and in order to be compatible with the analysis of Klein et al. [21, 22],
who have solved the same problem and produced highly accurate results.
Figure 6 depicts the amplitude reduction factor AR versus the non-dimensional distance x/z
0
from the source of disturbance along the direction of the diagonal OA of Figure 4 as computed by
the present method as well as by that of Klein et al. [21, 22] and using the same mesh in both
cases. The factor AR is de"ned as the ratio of the vertical displacement component of the soil
surface in the presence of the trench over that displacement in the absence of the trench. The
present results are in excellent agreement with those of Klein et al. [21, 22]. This was expected in
view of the fact that the codes of the two methods are similar. It is obvious that for the data of the
problem, this active vibration isolation system is clearly e!ective, at least along the direction of
the diagonal OA, as AR)0)25 outside the trench barrier [2].
720 S. E. KATTIS, D. POLYZOS AND D. E. BESKOS
Copyright 1999 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Engng. 46, 713}728 (1999)
Both these examples as well as additional ones in [36] serve to illustate the BEM code
previously described and demonstrate its high accuracy.
4. VIBRATION ISOLATION BY A ROW OF PILES
As it was mentioned in the introduction, when the ground transmitted waves have a large
wavelength, which would require unrealistic trench depths (larger than 0)8z
0
) for screening
e!ectiveness, the only practical type of wave barriers is that of a row of closely spaced piles. In this
section the wave screening e!ectiveness of a row of piles is studied numerically by using the BEM
computer code described in Section 2. The piles may be tubular or solid and may have square or
circular cross-sections. Tubular piles are modelled as long cylindrical cavities in soil for reasons of
simplicity, while the solid ones can be of concrete or any other material exhibiting linear elastic or
viscoelastic behaviour.
Piles have some distinct advantages over trenches as wave barriers. These are: their ability to be
driven very deeply into the soil, the well-known technology for foundation piles and the
possibility of any desirable arrangement to create wave barriers. Their screening e!ectiveness,
however, as it will be shown in the following, is not as good, in general, as that of trenches. The
modelling of the vibration isolation by piles problem is also more complicated and consequently,
its solution is more di$cult than the one for the corresponding trenches problem. Indeed the
vibration isolation by piles problem is a multiple wave di!raction problem of a truly three-
dimensional character which requires many more substructures and a much larger number of
degrees of freedom than the corresponding trench problem. A typical isolation by a row of eight
concrete piles problem requires six times the computer solution time than the corresponding
equivalent concrete trench problem. This is probably the reason why the problem has not
been solved before, even with the BEM which is ideally suited for this type of problemand in spite
of the fact that it does not present serious theoretical di$culties, at least in principle. Here
use is made of the BEM code described in the previous two sections to solve the problem and
through parametric studies assess the screening e!ectiveness of piles and reach some practical
conclusions.
In the following, the passive vibration isolation problem by a row of eight concrete piles of
circular cross-section is solved in detail for illustration purposes. Results of parametric studies
involving rows of six and eight open or in"lled piles of square or circular cross-section are
subsequently presented which help to understand the behaviour of the system with respect to its
vibration isolation e!ectiveness.
Consider the row of eight concrete piles of circular cross-section shown in Figure 7. Every pile
has a sectional diameter d and length h, while the surface to surface spacing between two
successive piles is s (Figure 7). The distance of the centre of the row of piles from the vibrational
source, i.e. a vertical load harmonically varying with time with magnitude P

"1 KN and
operational frequency 50 Hz, is r as depicted in Figure 7. The soil is assumed to be linear
viscoelastic with a shear modulus G

"132 MPa, Poisson's ratio v

"0)25, mass density


j

"17)5 Kg/m and damping coe$cient [

"5 per cent, while the concrete piles are assumed


also to behave linearly viscoelastically with G

"34)29 G

, v

"v

"0)25, j

"1)37 j

and [

"
5)0 [

. The Rayleigh wavelength for the soil is computed to be z


0
"5 m and the normalized by
z
0
geometrical data of the problem read R"r/z
0
"1)5, D"d/z
0
"0)2, H"h/z
0
"1)0 and
S"s/z
0
"0)1.
VIBRATION ISOLATION BY A ROW OF PILES 721
Copyright 1999 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Engng. 46, 713}728 (1999)
Figure 7. Cross-section and top view of the pile barrier vibration isolation system
The solution of the problem is obtained as the superposition of a symmetric and an antisym-
metric problem for each of which only

of the problem requires discretization. This discretization


is shown in Figure 8 and involves "ve substructures (four for the piles and one for the soil) and 330
eight-noded quadratic quadrilateral boundary elements (1285 functional nodes). The size of the
elements corresponds to approximately "ve elements per Rayleigh wave length, at least in the
area close to piles. The free soil surface is extended to a distance of 3z
0
from the centre of the pile
barrier system. The meshes constructed on the basis of these rules were also found to provide
satisfactory results by means of convergence studies involving coarse and re"ned meshes not
shown here for lack of space. The "ve substructures are coupled together by enforcing equilibrium
and compatibility at their interfaces in an automatic manner by the code. Figure 9 depicts the
variation of the factor AR on the whole soil surface for the above passive isolation by eight
concrete piles problem. The average surface amplitude reduction factor behind the trench is
722 S. E. KATTIS, D. POLYZOS AND D. E. BESKOS
Copyright 1999 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Engng. 46, 713}728 (1999)
Figure 8. Discretization of one-fourth of the pile barrier vibration isolation system
Figure 9. Contour of the amplitude reduction factor for the pile barrier vibration isolation system (eight concrete circular
cross-section piles)
de"ned as AR"(1/A) j

ARdA, where A is the area behind the trench enclosed by the semicircle
with a radius l/2 (one-half the length of the row of piles) and AR is the amplitude reduction factor
de"ned previously. In this problem it was found that AR"0)712.
In the following, additional numerical results are presented for other cases of passive vibration
isolation with variations in the material of the piles and their cross-section. Thus, Figure 10 shows
the variation of factor AR on the whole soil surface for the case of eight &open' piles of circular
cross-section. The BEM mesh of this case involved 234 eight-noded quadratic quadrilateral
VIBRATION ISOLATION BY A ROW OF PILES 723
Copyright 1999 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Engng. 46, 713}728 (1999)
Figure 10. Contour of the amplitude reduction factor for the pile barrier vibration isolation system (eight open circular
cross-section piles)
Figure 11. Contour of the amplitude reduction factor for the pile barrier vibration isolation system (eight open square
cross-section piles)
boundary elements (865 functional nodes) for

of the system and the AR was found to be 0)812.


Figure 11 portrays AR for the case of eight &open' piles of square cross-section with side D"0)2,
while Figure 12 depicts corresponding things for the case of eight concrete piles of square cross-
section. The discretizations for these two last cases involved 206 and 274 boundary elements,
724 S. E. KATTIS, D. POLYZOS AND D. E. BESKOS
Copyright 1999 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Engng. 46, 713}728 (1999)
Figure 12. Contour of the amplitude reduction factor for the pile barrier vibration isolation system (eight concrete square
cross-section piles)
Table I. Average surface amplitude reduction factor (AR) for the various pile
barrier vibration isolation systems
Average surface
Number of Type of Type of amplitude reduction
piles pile material pile cross-section
factor (AR)
6 Open Circular 0)926
6 Concrete Circular 0)755
6 Open Square 0)868
6 Concrete Square 0)721
8 Open Circular 0)812
8 Concrete Circular 0)712
8 Open Square 0)698
8 Concrete Square 0)675
10 Open Circular 0)648
10 Concrete Circular 0)624
10 Open Square 0)642
10 Concrete Square 0)620
respectively, while their AR factors were found to be 0)698 and 0)675, respectively. Details
concerning cases involving six or ten &open' or concrete piles of circular or square cross-section
can be found in [36]. Table I provides values of AR for all the aforementioned cases of passive
vibration isolation by a row of six, eight or ten piles. It is observed that AR decreases with the
number of piles or equivalently with the decrease in the spacing between piles. It is also observed
VIBRATION ISOLATION BY A ROW OF PILES 725
Copyright 1999 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Engng. 46, 713}728 (1999)
Table II. Average surface amplitude reduction
factor (AR) for trench barrier vibration isolation
systems
Average surface amplitude
Type of trench
reduction factor (AR)
Open 0)172
Concrete 0)521
that &open' piles are more e!ective than concrete piles and circular cross-section piles have
a similar behaviour to the square cross-section ones.
In an e!ort to compare the screening e!ectiveness of a row of piles against that of a trench with
the same width, depth and length as the row of piles, the BEM code presented in the two previous
sections was used to solve for the screening e!ectiveness of such open and concrete "lled trenches
on the basis of the passive vibration isolation problem just discussed in connection with the piles.
The results of this study are summarized in Table II, which gives values of AR for open or
concrete "lled trenches. A comparison between Tables I and II reveals that the screening
e!ectiveness of a row of piles is much smaller than the corresponding one of trenches.
5. CONCLUSIONS
On the basis of the previous developments and additional parametric studies in [36] not
presented here for lack of space one can conclude the following:
1. An advanced-frequency domain BEM code has been developed for the dynamic analysis of
complex three-dimensional elastic structural systems.
2. The BEM analysis of a row of piles as wave barriers in vibration isolation problems is
analogous to that of trenches with the only di!erence being the size of the problem which is
always much larger for piles than for trenches.
3. The general wave screening behaviour of piles is similar to that of trenches with the latter
being always more e!ective than piles as wave barriers. However, piles are the only vibration
isolation solution for cases involving large Rayleigh wavelengths.
4. Length, depth and width of a row of piles in#uence its vibration isolation e!ectiveness in
a manner analogous to trenches.
5. The most important parameter in#uencing the screening e!ectiveness of a row of piles is
their spacing. A row of piles starts being e!ective when the spacing S is small.
6. The cross-sectional shape of the piles plays no signi"cant role in the e!ectiveness of a row of
piles.
ACKNOWLEDGEMENTS
The authors are grateful to the Greek General Secretarial for Research and Technology for
supporting this work under contract 91 EA 386 (PENED Programme).
726 S. E. KATTIS, D. POLYZOS AND D. E. BESKOS
Copyright 1999 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Engng. 46, 713}728 (1999)
REFERENCES
1. Barkan DD. Dynamics of Bases and Foundations. McGraw-Hill: New York, 1962.
2. Richart Jr. FE, Hall JR, Woods RD. <ibrations of Soils and Foundations. Prentice-Hall: Englewood Cli!s, NJ, 1970.
3. Haupt WA. Isolation of vibrations by concrete core walls. In Proceedings of the Ninth International Conference on Soil
Mechanics and Foundation Engineering, vol. 2. Tokyo, Japan, 1977; 251}256.
4. Segol G, Lee PCY, Abel JF. Amplitude reduction of surface waves by trenches. Journal of Engineering Mechanics
1978; 104:621}641.
5. May TW, Bolt BA. The e!ectiveness of trenches in reducing seismic motion. Earthquake Engineering and Structural
Dynamics 1982; 10:195}210.
6. Yang YB, Hung HH. A parametric study of wave barriers for reduction of train induced vibrations. International
Journal for Numerical Methods in Engineering 1997; 40:3729}3747.
7. Emad K, Manolis GD. Shallow trenches and propagation of surface waves. Journal of Engineering Mechanics 1985;
111:279}282.
8. Beskos DE, Dasgupta B, Vardoulakis IG. Vibration isolation of machine foundations. In <ibration Problems in
Geotechnical Engineering, Gazetas G, Selig ET (eds). ASCE: New York, 1985; 138}151.
9. Beskos DE, Dasgupta B, Vardoulakis IG. Vibration isolation using open or "lled trenches, Part 1: 2-D homogeneous
soil. Computational Mechanics 1986; 1:43}63.
10. Leung KL, Beskos DE, Vardoulakis IG. Vibration isolation using open or "lled trenches, Part 3: 2-D non-
homogeneous soil. Computational Mechanics 1990; 7:137}148.
11. Leung KL, Vardoulakis IG, Beskos DE, Tassoulas JL. Vibration isolation by trenches in continuously non-
homogeneous soil by the BEM. Soil Dynamics and Earthquake Engineering 1991; 10:172}179.
12. Beskos DE, Vardoulakis IG. Vibration isolation by trenches in homogenous and layered soils. In Structural
Dynamics2E>ROD>N+90, Kratzig WB et al. (eds). A.A. Balkema: Rotterdam, 1991; 693}700.
13. Von Estor! O, Antes H. Vibration isolation analyses in the case of transient excitations by a coupled BE/FE
procedure. In Computational Mechanics '91, Atluri SN, Beskos DE, Jones R, Yagawa G (eds). ICES Publications:
Atlanta, 1991; 1148}1153.
14. Ahmad S, Al-Hussaini TM. Simpli"ed design for vibration screening by open and in-"lled trenches. Journal of
Geotechnical Engineering 1991; 117:67}88.
15. Al-Hussaini TM, Ahmad S. Design of wave barriers for reduction of horizontal ground vibration. Journal of
Geotechnical Engineering 1991; 117:616}636.
16. Stamos AA, Von Estor! O, Antes H, Beskos DE. Vibration isolation in road-tunnel tra$c systems. International
Journal of Engineering Analysis and Design 1994; 1:109}121.
17. Mateo J, Alarcon E. On the use of trenches and walls on the control of ground transmitted railway vibrations.
In Boundary Element Method X<I, Brebbia CA (ed.). Computational Mechanics Publications: Southampton, 1994;
547}554.
18. Beskos DE, Vardoulakis IG. Vibration isolation of structures from surface waves. In Computational Mechanics +86,
Atluri SN, Yagawa G (eds). Springer: Berlin, 1986; vi 263}vi 267.
19. Banerjee PK, Ahmad S, Chen K. Advanced application of BEM to wave barriers in multi-layered three-dimensional
soil media. Earthquake Engineering and Structural Dynamics 1988; 16:1041}1060.
20. Dasgupta B, Beskos DE, Vardoulakis IG. Vibration isolation using open or "lled trenches, Part 2: 3-D homogeneous
soil. Computational Mechanics 1990; 6:129}142.
21. Klein R, Antes H, Le Houedec D. Vibration isolation by open trenches: three dimensional studies using the BEM.
In Advances in Simulation and Interaction echniques, Papadrakakis M, Topping BHV (eds). Civil-Comp. Ltd:
Edinburgh, 1994; 13}22.
22. Klein R, Antes H, Le Houedec D. E$cient 3D modelling of vibration isolation by open trenches. Computers and
Structures 1997; 64:809}817.
23. Ahmad S, Al-Hussaini TM, Fishman KL. Investigation on active isolation of machine foundations by open trenches.
Journal of Geotechnical Engineering 1996; 122:454}461.
24. Al-Hussaini TM, Ahmad S. Active isolation of machine foundations by in-"lled trench barriers. Journal of Geotechni-
cal Engineering 1996; 122:288}294.
25. Beskos DE. Boundary element methods in dynamic analysis. Applied Mechanics Reviews 1987; 40:1}23.
26. Beskos DE. Boundary element methods in dynamic analysis: Part II (1986}1996). Applied Mechanics Reviews 1997;
50:149}197.
27. Woods RD, Barnett NE, Sagesset R. Holography*a new tool for soil dynamics. Journal of Geotechnical Engineering
1974; 100:1231}1247.
28. Liao S, Sangrey DA. Use of piles as isolation barriers. Journal of Geotechnical Engineering 1978; 104:1139}1152.
29. Avilles J, Sanchez-Sesma FJ. Foundation isolation from vibration using piles as barriers. Journal of Engineering
Mechanics 1988; 114:1854}1870.
30. Avilles J, Sanchez-Sesma FJ. Piles as barriers for elastic waves. Journal of Geotechnical Engineering 1983; 109:
1133}1146.
VIBRATION ISOLATION BY A ROW OF PILES 727
Copyright 1999 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Engng. 46, 713}728 (1999)
31. Baroomand B, Kaynia AM. Vibration isolation by an array of piles. In Soil Dynamics and Earthquake Engineering <.
Computational Mechanics Publications: Southampton, 1991; 683}691.
32. Manolis GD, Beskos DE. Boundary Element Methods in Elastodynamics. Unwin-Hyman, Chapman and Hall:
London, 1988.
33. Guiggiani M. Computing principal value integrals in 3D BEM for time-harmonic elastodynamics*a direct
approach. Communications in Applied Numerical Methods 1992; 8:141}149.
34. Kattis SE, Polyzos D, Beskos DE. Dynamic response analysis of 3-D elastic systems by an improved freq-
uency domain BEM. In Boundary Element Method X<I, Brebbia CA (ed.). Computational Mechanics Publications:
Southampton, 1994; 521}528.
35. Kattis SE, Polyzos D, Beskos DE. Structural vibration isolation by rows of piles. In Soil Dynamics and Earthquake
Engineering <II, Cakmak AS, Brebbia CA (eds). Computational Mechanics Publications: Southampton, 1995;
509}516.
36. Kattis SE. Numerical solution of problems of vibration isolation of structures fromsurface elastic waves. Ph.D. hesis,
Department of Mechanical and Aeronautical Engineering, University of Patras, Patras, Greece, 1997 (in Greek).
37. Mantic V. A new formula for the C-matrix in the Somigliana identity. Journal of Elasticity 1993; 33:191}201.
38. Huber O, Lang A, Kuhn G. Evaluation of the stress tensor in 3D elastostatics by direct solving of hypersingular
integrals. Computational Mechanics 1993; 12:39}50.
39. do Rego Silva JJ. Acoustic and Elastic =ave Scattering ;sing Boundary Elements. Computational Mechanics
Publications: Southampton, UK, 1994.
40. Bu S. In"nite boundary elements for the dynamic analysis of machine foundations. International Journal for
Numerical Methods in Engineering 1997; 40:3901}3917.
41. IMSL, IMS Math/ibrary ;ser's Manual. Version 3.0, Visual Numerics Inc, Houston, Texas, 1994.
42. Cai RY, Zeng ZJ, Chen F. Partitioning techniques for the boundary element method. Computers and Structures 1991;
38:537}545.
43. Pao YH, Mow CC. Di+raction of Elastic =aves and Dynamic Stress Concentrations. Crane Russak: New York, 1973.
728 S. E. KATTIS, D. POLYZOS AND D. E. BESKOS
Copyright 1999 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Engng. 46, 713}728 (1999)

You might also like