Soil Dynamics and Earthquake Engineering 69 (2015) 182195

Contents lists available at ScienceDirect

Soil Dynamics and Earthquake Engineering

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

Response of circular exible foundations subjected to horizontal

and rocking motions
Shi-Shuenn Chen, Jr-Gang Hou n
Department of Civil and Construction Engineering, National Taiwan University of Science and Technology, Taipei 106, Taiwan

art ic l e i nf o

a b s t r a c t

Article history:
Received 12 June 2014
Received in revised form
11 October 2014
Accepted 25 October 2014

A methodology using modal analysis is developed to evaluate dynamic vertical displacements of a

circular exible foundation resting on soil media subjected to horizontal and rocking motions. The
inuence of the soil reaction forces on the foundation is considered by introducing modal impedance
functions, which can be determined by an efcient procedure with ring elements. The displacements of
the foundation can then be easily solved by modal superposition. Parametric studies for modal responses
of the exible foundation indicate that the coupled response of the foundation is signicantly inuenced
by relative stiffness among the foundation and the soil medium, vibration frequency range, foundation
mass, and boundary contact conditions. The welded boundary condition should be considered to predict
the coupling response while the relaxed boundary condition may be used to predict approximately the
vertical displacements. As a foundation with a relative stiffness ratio more than three, it is found that the
foundation can be considered as rigid to calculate coupling displacements. For a slightly exible
foundation, considerations of three modes are sufcient enough to obtain accurate foundation
responses. Moreover, at low frequencies, the coupling effect due to higher mode can be neglected.
& 2014 Published by Elsevier Ltd.

Keywords:
Flexible foundations
Modal analysis
Soilstructure interaction

1. Introduction
Foundation vibration analysis plays a key role in solving soil
structure interaction problems, and has been investigated widely
since 1930s. In the early stage, the foundation was often assumed
as rigid and this assumption has been also considered in most of
recent studies [14]. After 1970s, effects of the foundation exibility are increasingly recognized and studied by many researchers [512]. In those studies, the considered soil conditions involve
elastic half-space, layered soil, and poro-elastic medium; the
considered dynamic loadings are vertical or rocking force; the
interface between the foundation and the underlying soil medium
is always assumed as frictionless.
The past studies have found that the rocking and horizontal
vibrations of the rigid foundation are coupled. Triantafyllidis and
Prange [13] studied the full coupling at the interface between the rigid
foundation and soil medium. Not only rocking forces but also
horizontal force can induce the vertical displacement. However, the
coupled response of exible foundations subjected to horizontal
loadings is hardly investigated. Furthermore, exible foundations are
multiple-degree-of-freedom (MDOF) systems, dynamic behaviors of
which are complicated and seriously affected by the position, the load

Corresponding author. Tel.: 886 2 26625858x58712; fax: 886 2 26622814.

E-mail address: e1716@eandc.com.tw (J.-G. Hou).

http://dx.doi.org/10.1016/j.soildyn.2014.10.022
0267-7261/& 2014 Published by Elsevier Ltd.

distribution, and the foundation stiffness. Because the modal analysis

is widely used in the engineering and features in reducing complexity
of the MDOF system and amount of computational work, the modal
method is suitable in this study. Recently, Chen and Hou [14] have
investigated the vertical vibration of the exible foundation by a
modal method, and their results indicated that three modes are
sufcient in the majority of analyses.
The primary objective of this paper is therefore to present a
modal concept method to evaluate and analyze the dynamic
vertical displacement of a exible circular foundation laying on
the soil medium subjected to harmonic horizontal forces and
antisymmetric rocking forces, as shown in Fig. 1. Welded contact
is considered between the foundation and the supporting soil. The
foundation plate behaves according to the classical thin plate
theory. In addition, the exible behavior of the foundation is only
considered in out of plane motion (i.e. vertical direction), and the
in-plane rigidity of the foundation is assumed as rigid.
In this study, the modal equations of motion for a circular plate are
rstly established. An efcient procedure is also introduced to
calculate modal impedance functions to consider the interaction effect
between the foundation and the underlying soil. The dynamic
responses of the exible foundation thus computed by the proposed
method are then compared with existing solutions and numerical
results by a computer program. Moreover, parametric studies are
presented to evaluate the inuence of primary parameters on the
foundation responses.

S.-S. Chen, J.-G. Hou / Soil Dynamics and Earthquake Engineering 69 (2015) 182195

183

2.2. Foundation deection and contact stress in terms of modal

displacements

2. Formulation of interaction problem

2.1. Equations of motion
Considering a circular exible foundation subjected to vertical
and horizontal loads as shown in Fig. 1, the equations of harmonic
motions with frequency in the z and x directions can be
expressed, respectively, by the following equations [15]:
F z : D4 wr;  mp 2 wr; pz r;  qz r;

1a


F x :  mp R2 2 P x  Q x

1b

Ep hp

121  vp 2

where R, Ep, hp, p and mp are radius, Young's modulus, thickness,

Poisson's ratio and the mass per unit area of the foundation plate,
respectively. w and are the vertical and horizontal displacements
of the foundation plate, respectively; pz and qz are the vertical load
per unit area of the foundation plate and the vertical contact
stress, respectively; Px and Qx are the total horizontal load and the
horizontal soil reaction, respectively; r and are the radial and
circumferential direction, respectively.
In addition, the vertical load pz is an antisymmetric rocking
force (about the y-axis), and can be expressed as follows:
pz r; pz r cos

where pz r is the vertical load amplitude. Also, the vertical, radial,

circumferential contact stresses (qz, qr, q) and the horizontal soil
reaction Qx can be presented as follows:
8
9 8
9
>
< qz r; >
= >
< qz r cos >
=
q r;  q r sin
for 0 rr rR
4
>
>
>
: q r; ; : q r cos >
;
r
r
Qx

R
0

wr; Y i i r;

6a

i0

where 4 2 2 and 2 is the Laplacian operator. D is the exural

rigidity of the foundation plate and dened as follows:
D

The deection of a circular foundation plate in z-direction can

be represented by a series of free vibration modes for a circular
plate as follows [5]:



r qr r q r dr

where qz r, qr r, and q r are the vertical, radial, and circumferential contact stress amplitudes, respectively.

Fig. 1. A schematic of a exible circular foundation resting on soil media.

jjY i jjeii i r;

6b

i0

where Yi, jjY i jj, and i are the modal displacement, the modal
amplitude, and the modal phase angle, respectively; i (i0, 1,.., N)
denotes the mode shape of a circular plate for the ith mode.
Since the foundation behaves according to the classic thin plate
theory and foundation deections are antisymmetric about the yaxis, free vibration mode shapes of a circular foundation can be
expressed as [15]

r

r 
bi I 1 i
cos
R
R

i r; i r cos ai J 1 i

where J1 is the Bessel function of the rst kind of order 1, I1 is the

modied Bessel function of the rst kind of order 1. The values of
ai, bi and i determine the mode shape and are solved from the
boundary conditions. For a surface foundation, the moments and
shear forces at the edge of the foundation plate are zero, and the
coefcients ai, bi and i can be determined as follows [15]:

i J 1 i  1  p J 2 i i J 1 i  2i J 2 i  1  p J 2 i

i I 1 i  1  p I 2 i i I 1 i 2i I2 i  1  p I2 i

8a

bi i J 1 i  1  p J 2 i

ai i I 1 i  1  p I 2 i

8b

The modes are orthogonal and normalized as

( 2
R Z 2
R i j
r i r; j r; dr d
0
iaj
0
0

Considering p 0.25, the mode shapes are shown in Fig. 2. The

mode shape i for i40 denotes the exible deformation in higher
modes, and 0 is the rigid body mode for rocking vibration and can be

Fig. 2. Free vibration mode shapes of a circular plate for rocking vibration.

184

S.-S. Chen, J.-G. Hou / Soil Dynamics and Earthquake Engineering 69 (2015) 182195

expressed as follows:
2r
0 r; 0 r cos cos
R

10

Based on the orthogonal property of modes, the modal displacements can be obtained as follows:
R 2 R R
r i r; wr; dr d
11
Yi 0 0
R2
For the rigid body mode, substituting Eq. (10) into Eq. (11) gives
R
Y0 0
2

12

where 0 is the rocking angle of the rigid body mode, as shown in

R 2 R R 2
r cos wr; dr d
13
0 0 0
R4 =4
Similarly, the contact stress components shown in Eq. (4) are
related to the displacement of the foundation and can also be
expanded in terms of modal displacements as follows:
N

qz r Y j qzj r qzx r

14a

j0
N

qr r Y j qrj r qrx r

14b

j0
N

q r Y j qj r qx r

14c

j0

where qzj r, qrj r, and qj r are the vertical, radial, and circumferential contact stress distributions corresponding to the modal
shape j, respectively; qzx r, qrx r, and qx r are the vertical,
radial, and circumferential contact stress distributions corresponding to a unit horizontal displacement in the x direction,
respectively.
2.3. Equations of motion in terms of modal coordinates

The relative stiffness ratio among the foundation and the soil
medium , the mass ratio bz, and the dimensionless frequency a0
are dened as follows:

20

Gs R3

bz

a0

mp R2

21

22

s R3

Vs

where Gs, Vs, and s are the shear modulus, the shear wave
velocity, and mass density of the soil medium, respectively.
Substituting Eqs. (20)(22) into Eqs. (16) and (1b) gives



4
Gs R i  bz a0 2 Y i P i  Q i ; i 0; 1; 2; ; N
23a
Gs R bz a20 P x Q x

23b

Eq. (23) can be also expressed in a matrix form as follows:

"
#
"
#!
I
f0g
 f0g
2
a
Gs R
b
z 0
1
f0gT
0
f0gT
) ( )

 (
P
f
g
Qz
fY z g
z


24

Px
Qx

where {Yz}, {Pz} and {Qz} are modal displacement, modal load, and
modal soil reaction vectors, respectively; [I] is the identity matrix;
[] is the characteristic matrix, as shown in
2
3
0
0

0
6 0 1 4
7
6
7
25
 6
7

0 5
4
4
0
0 N
The modal soil reaction vector {Qz} can be represented in terms
of the modal displacement vector {Yz} and horizontal displacement by substituting Eqs. (4) and (14a) into Eq. (18) as follows:
N

The modal coordinate is introduced into the equations of

motion by substituting Eqs. (6a) and (7) into Eq. (1a) as follows:
!
4
N

i0

i
R

 mp 2 Y i i r; pz r;  qz r;

15

For obtaining decoupled modal expressions, the orthogonal

property shown in Eq. (9) is used to rewrite Eq. (15) as follows:
!
4
D

 mp 2 Y i R2 P i  Q i ; i 0; 1; 2; ; N
R4

16

where Pi is the modal load and Qi is the modal soil reaction, as

shown in
Z R Z 2
r i r; pz r; d dr
17
Pi
0

Z
Qi

R
0

Z 2
0

Q i K ij Y j K ix ; i 0; 1; 2; ; N

26

j0

where Kij and Kix are the modal impedance and the modal
coupling impedance, respectively, as shown in
Z R
K ij Gs Rkij
r i rqzj rdr; i; j 0; 1; 2; ; N
27a
0

K ix Gs Rkix

r i rqzx rdr;

i 0; 1; 2; ; N

27b

Similarly, the horizontal soil reaction Qx can be rewritten by

substituting Eqs. (14b) and (14c) into Eq. (5) as follows:
N

Q x K xj Y j K xx

28

j0

r i r; qz r; d dr

18

For the modal load P0 (corresponding to the rocking rigid body

mode), substituting Eq. (10) into Eq. (17) gives
Z Z
2 R 2 2
2
19
P0
r cos pz r; d dr M
R 0 0
R
where M represents the total moments about the y-axes. It can be
seen from Eq. (19) that P0 is proportional to the moment M.

where Kxj and Kxx are the modal coupling impedance and horizontal impedance, respectively, as shown in

K xj Gs Rkxj

K xx Gs Rkxx


r qrj r qj r dr; j 0; 1; 2; ; N

29a



r qrx r qx r dr

29b

R
0

S.-S. Chen, J.-G. Hou / Soil Dynamics and Earthquake Engineering 69 (2015) 182195

Substituting Eqs. (26) and (28) into Eq. (24) gives

"
#
"
#
I
f0g
 f0g
2
a
Gs R

b
z 0
1
f0gT
0
f0gT
2

31
(
)

kzx
kzz 
fP z g
fY z g
5A
T

4 
Px

kxx
kxz

3.1. Soil displacements due to contact stresses

30

where [kzz], {kxz} and kxx are the dimensionless modal impedance
matrix, the dimensionless modal coupling impedance vector and
the dimensionless horizontal impedance, respectively.
For rigid foundations, Eq. (30) can be simplied as


Gs R k00  bz a0 2 Y 0 Gs Rk0x P 0
31a


Gs Rkx0 Y 0 Gs R kxx  bz a0 2 P x

31b

Substituting Eqs. (12), (19), (21), and (22) into Eq. (31) gives
!
K 00 R2
K 0x R
 I 0 2 0
M
32a
2
4



K x0 R
0 K xx  mp R2 w2 P x
2

32b

where I0 (mpR4)/4 is mass moment of inertia of the circular

foundation with respect to a horizontal axis through CG. Eq. (32)
shows the equation of motion for a rigid circular foundation
subjected to a rocking moment and a horizontal force. Note that
the modal impedance K00 represents the required modal load P0 to
produce a unit soil displacement with the mode shape 0, which
is proportional to the rocking impedance of the rigid foundation KR
as follows:
KR

K 00 R2
4

33

In a similar way, the modal coupling impedance Kx0 or K0x is

proportional to coupling impedance of the rigid foundation KC as
follows:
KC

K x0 R
2

185

34

Hence, Eq. (30) can be used in foundation vibration analysis.

Derivation for the modal impedance matrix will be presented
in Section 3. When the modal impedance matrix is known, the
modal displacement can be solved by Eq. (30), and then the
vertical deection of the foundation can be obtained by back
substituting the modal displacement in Eq. (6a). A brief owchart
is shown in Appendix A to summarize the derivation of equations
of motion in modal coordinates.

3. Formulation of modal impedance matrix

Solving the modal impedance matrix relies on solving the
contact stress corresponding to the modal displacement as shown
in Eq. (27) and (29). For the circular foundation problem, the
contact stress is often solved by the ring method, which discretizes
the contact area into several ring elements and uses Green's
function to calculate the inuence coefcient for each ring
element. The inuence coefcient fml is generally dened as the
displacement in the mth ring induced by a unit force acting on the
lth ring. To solve these inuence coefcients, several calculations
of integrals are required. Symmetries of the inuence coefcient
matrix are effectively utilized to reduce the computational work.

Kausel and Roesset [16] have derived stiffness matrices for wave
propagation problems in layered media. By using these stiffness
matrices, the exibility of the soil can be obtained, and then soil
displacements due to stresses can be solved by a transformation
between the spatial and the wave number domain [17]. Consider a
general state of stress on the surface of a soil deposit as follows:
(
) 9
8
cos n
>
>
>
>
>
>
q
r
>
>
r
>
>
>
>

sin
n
>
8
9 >
>
>
>
>
(
)
>
>
q
r;

>
>
>
r
=
<
= <
 sin n >
q r; q r
35

cos
n
>
>
>
>
>
: q r; ; >
>
>
>
>
(
)
z
>
>
>
>
cos n
>
>
>
>
>
> q r
>
>
z
>
>
;
:
sin n

Then surface soil displacements induced by these stresses can

be expressed as follows:
(
) 9
8
cos n
>
>
>
>
>
> ur
>
>
>
>
>
>
sin n
>
>
8
9 >
>
>
(
)>
>
>
ur;

>
>
>
=
<
= <
 sin n >
vr; vr
36

cos
n
>
>
>
>
: wr; >
; >
>
>
>
>
(
)
>
>
>
>
cos n
>
>
>
>
>
>
wr
>
>
>
>
;
:
sin n
where u, v, and w are radial, circumferential, and vertical displacements, respectively, as shown in

Z 1
Z R
q r q r
J n  1 rdr
ur

r r
2
0
0


F uu F vv
F uu  F vv
J n  1 r 
J n 1 r d

2
2

Z 1
Z R
q r  q r
J n 1 rdr

r r
2
0
0


F uu F vv
F uu  F vv
J n 1 r 
J n  1 r d

2
2

Z 1
Z R

F uw 
J n  1 r  J n 1 r d


rqz rJ n rdr
2
0
0
37a
Z

vr
0

r
0

qr r q r
J n  1 rdr
2


F uu F vv
F uu  F vv
J n  1 r
J n 1 r d

2
2

Z 1
Z R
q r  q r
J n 1 rdr

r n
2
0
0


F uu F vv
F uu  F vv
J n 1 r 
J n  1 r d
 
2
2

Z 1
Z R

F uw 
J n  1 r J n 1 r d


rqz rJ n rdr
2
0
0
37b

qr r q r
J n  1 rdr F wu J n rd
2
0
0

Z 1
Z R
qr r  q r
J n 1 rdr F wu J n rd

r
2
0
0

Z 1
Z R

rqz rJ n rdr F ww J n rd
Z

wr 

37c

186

S.-S. Chen, J.-G. Hou / Soil Dynamics and Earthquake Engineering 69 (2015) 182195

where is a wave number indicating the number of radians per

unit distance (i.e., 2/wavelength); Fuu, Fvv, Fww, and Fwu are the
exibility corresponding to the surface degrees of freedom of a soil
system. The exibility Fij denotes the component i (iu, v, w) of
surface displacements induced by a unit j-component (j u, v, w)
surface force. Hence, the stiffness matrix of the soil system can be
obtained by assembling the stiffness matrices of soil layer components and the half-space. Then, the exibility of a soil system can
be obtained by calculating the inverse of the stiffness matrix of the
soil system. For a half-space system, the exibility can be calculated as follows:
"
# "
#1
F uu F uw
Suu Suw
1
and F vv
38

F uw F ww
Suw Sww
Svv
in which Sww, Suu, and Suw are elements of the stiffness matrix for
the half-space system in the wave number domain.
By using Eq. (37), contact stress components can be determined by
considering consistent boundary conditions. To determine the modal
impedances Kij and Kxj for the rocking vibration case (n 1), the
foundation is applied by a unit modal displacement for the jth mode
without horizontal and other modal displacements. Thus, the soil
displacements need to satisfy the following:
8
9 8
9
0 >
>
< ur >
= >
<
=
0
vr
39
>
: wr >
; >
: r >
;
j

Fig. 4. Comparison of vertical displacement proles of a exible foundation on a

half-space subjected to horizontal force.

As for the modal impedances Kix and Kxx of the foundation, they
represent the forces generated by a unit horizontal displacement
without other modal displacements. The displacements need to
satisfy the following:
8
9 8 9
>
< ur >
= >
<1>
=
vr 1
>
: wr >
; >
: >
;
0

40

3.2. Efcient ring assumptions

For determining the contact stress distribution, the ring
method is used and the contact area is divided into Nr discrete
rings. Moreover, the stress distribution is assumed as follows:
9
8 N
r
rn  1 >
>
>
>

A
>
>

>
>
r n >
8
9 >
>
>
1
>
>
>
>
q
r

q
r
>
>

n1 >
=
< Nr
< r
= >
r
B n 2
qr r  q r
r >
>
>
>
>
:
; >
>
>1
qz r
>
>
>
n >
Nr
>
>
r
>
>
>
> C
>
:
n1 >
r ;
1
Fig. 3. Rocking and horizontal impedance coefcients for a rigid circular foundation on a half-space.

41



where r  2r r r r r 2r, r NRr , r  12 r; A , B , and C are
undetermined ring stress coefcients.

S.-S. Chen, J.-G. Hou / Soil Dynamics and Earthquake Engineering 69 (2015) 182195

Substituting Eq. (41) into Eq. (37) gives

Nr

02

Nr

Nr

20

12

10

C f m f m ; ; m 1; 2; :::N r
1

Nr

02

20

vr m A f m f m B f m  f m

ur m A f m f m B f m  f m
1
Nr

187

Nr

42  1

10

12

C  f m  f m ; ; m 1; 2; :::N r
1

Fig. 5. The modal impedance functions for the rst three modes used in the parametric study.

42  2

188

S.-S. Chen, J.-G. Hou / Soil Dynamics and Earthquake Engineering 69 (2015) 182195
Nr

Nr

01

Nr

21

wr m A  f m B f m C f m ; ; m 1; 2; :::N r
1

21
f m

r  2r

42  3
When
follows:

Written in a matrix form as

2

02

f   f 
6 0
6 f  f 02 
4
01
 f 

20

f   f 
2

20

 f  f 
21

f 

38
9 8 9
>
< u >
=
< fAg >
= >
7
10
12
f   f  7
5> fBg > > fvg >
: fCg ; : fwg ;
1
f 
12

10

f  f 

42  4
where
0
f m

r  2r


r
F uu k F vv k
dk
J
krdr
k
J
kr

m
n

1
n

1
4
r n

r 2r n

43  1
Z

f m

f m

r 2r
r j  2r

!
r
J
krdr
kJ n kr m F ww kdk
n
r n 1

r 2r n 1
r  2r

43  2

01

10

r 2r

43  4

!
r
F uw k
dk
J
krdr
kJ n  1 kr m
n
2
r n 1

43  5

10

01

20

02

21

12

44
By using the relation in Eq. (44), the numerical procedure can
be simplied and be more efcient. For example, Nr 15 induces
2025 numerical integrations; however, by using the symmetric
relation integrations can be reduced to 1035 and about half of the
computation time can be saved. Finally, by considering Eqs. (39),
(40), and (42) we have 3Nr unknowns and 3Nr equations, and then
ring stress coefcient vectors {A}, {B}, and {C} can be solved. If the
contact horizontal stress is neglected (i.e. relaxed contact), Eq. (424) can be simplied as follows:
45

After contact stresses are obtained, the modal impedance

matrix can be solved by Eqs. (27) and (29). A owchart is also
given in Appendix A to summarize the calculation steps for the
modal impedances. Moreover, a numerical example is presented in
Appendix B to illustrate the application of the proposed method.


r
F uu k  F vv k
dk
J n  1 krdr k J n 1 kr m
n
4
r

r 2r n
r  2r

!
r
F wu k
dk
J
krdr
kJ n kr m
2
r n n  1

r 2r n 1

f m f m ; f m f m ; f m f m ; f m f m ; f m f m ; f m f m


rn 2
F uu k F vv k
dk
J
krdr
k
J
kr

m
n1
n1
4
r n 2

r 2r n
r  2r

f m

02
f m

43  9

r is small, symmetric relations can be established as

f fCg fwg

43  3
f m

!
r
F wu k
dk
J n 1 krdr kJ n kr m
2
r n 2

r 2r n 2

43  6
20

f m

1
0


Z r r n 2
2 r
F uu k  F vv k
dk
J
krdr
k
J
kr

m
n

1
n

1
n2
4
r  2r r
43  7

12

f m

1
0

r 2r n 1
r 2r

r
F uw k
dk
J n krdr kJ n 1 kr m
2
r n 1

43  8

Fig. 6. Inuence of the relative stiffness ratio and load types on the normalized
static average amplitude for exible foundations on a half-space.

Fig. 7. Inuence of the relative stiffness on the normalized static modal amplitude
for exible foundations on a half-space.

S.-S. Chen, J.-G. Hou / Soil Dynamics and Earthquake Engineering 69 (2015) 182195

4. Verication of the proposed method

To verify the proposed method described in the preceding
sections, existing solutions and a computer program SASSI developed
by Lysmer et al. [18] in 1981 for SSI analysis are used for comparison.
Before validation of the proposed method, the convergence of the
solution with respect to the number of rings Nr is checked rst by
comparing impedance functions for a rigid circular footing supported
on an elastic half-space. In this paper, Nr 15 is used to check the
convergence. Also, the number of modes considered has inuence on
the convergence of the solution. The required number of modes is
dependent on the stiffness of the foundation, the modal load value in
the higher mode (i.e., Pi when i40), and the foundation mass. Chen
and Hou [14] found that numerically stable and converged solutions
for vertical displacement of exible circular foundations with
Z0:003 can be obtained in the frequency range 0 ra0 r 6 for
eight modes. Hence, eight modes are considered for calculating
modal impedance in this paper. Besides, Poisson's ratio of the
foundation plate is set to 0.25 for all numerical results.
Fig. 3 shows a comparison of rocking and horizontal impedance
coefcients taken from Wolf and Darbre [19] with those obtained
by an efcient ring method of the present study for a rigid circular
foundation resting on an elastic half-space. Wolf and Darbre [19]
used the denition of the impedance of the form
K K S k ia0 c

46

189

where KS is the static stiffness, and k and c are impedance

coefcients. It can be seen from Fig. 3 that good agreement is
obtained between the two solutions. Hence, sufcient convergence can be achieved with Nr 15.
Fig. 4 shows a comparison of results for the normalized
displacement prole of a circular exible foundation on a halfspace subjected to horizontal force at four different dimensionless
frequencies. In this case, the parameters of the foundation-soil
system are set as follows: (1) the relative stiffness ratio 0.003;
(2) the mass ratio bz 0; (3) Poisson's ratio of the soil vs 0.33;
(4) horizontal force Px 1. In the present method, displacements
are solved by considering eight modes and 15 rings to calculate
modal impedance matrix. To obtain the numerical solutions by
SASSI [18], the exible foundation is modeled by 600 plate
elements with 331 interaction nodes. Comparisons of results
presented in Fig. 4 indicate that a good agreement is observed
between the two solutions. Therefore, the proposed method is
veried by those independent comparisons.

5. Parametric study
In this section, foundation vibration analysis is performed for a
circular exible foundation on an elastic half-space to evaluate the
inuence of various parameters on the modal response by the
proposed method. For foundation vibration analysis, horizontal

Fig. 8. Modal magnication factors in the rigid body mode for circular foundations resting on a half-space.

190

S.-S. Chen, J.-G. Hou / Soil Dynamics and Earthquake Engineering 69 (2015) 182195

force with P x 1 and rocking force with modal load value P 0 1

are considered. P 0 1 means that vertical force distributes as
pz 2r= R cos . In addition, the effect of the boundarycontact
condition is also investigated by considering two set modal
impedance functions. The modal impedance functions used in this
study for the rst three modes are shown in Fig. 5, and Poisson's
ratio of the half-space is 0.33. In Fig. 5, the results obtained by
welded contact condition are compared with those obtained by
relaxed contact condition. It is noted that the coupling impedance
functions for rocking and horizontal motion (i.e. Kix or Kxi) only
exist in welded contact condition.
In addition, some dimensionless response parameters are
introduced. For foundation analysis, the vertical displacement is
rewritten from Eq. (6b) as follows:
"
#
wr;

Ai M i eii i r; jjY Rg;st jj

where A jjst =Rg;st jj is the normalized static horizontal amplitude, M jj=st jj is the dynamic magnication factor, jjst jj is the
static horizontal amplitude, jjRg;st jj is the static horizontal

47

i0

where Ai jjY i;st =Y Rg;st jj is the normalized static modal amplitude

and relates to the participation of modal components,
M i jjY i =Y i;st jj is the dynamic modal magnication factor, jjY i;st jj
is the static modal amplitude, and jjY Rg;st jj is the static amplitude
of rigid foundations.
Similarly, the horizontal displacement is expressed as follows:

A M ei jjRg;st jj

48

Fig. 10. Dynamic magnication factor for circular exible foundations resting on a
half-space subjected to horizontal force.

Fig. 9. Modal magnication factor in the rst mode for circular exible foundations resting on a half-space.

S.-S. Chen, J.-G. Hou / Soil Dynamics and Earthquake Engineering 69 (2015) 182195

amplitude of rigid foundations, and is the phase angle of the

horizontal displacement.
The inuence of the relative stiffness ratio on the static
displacement is illustrated in Figs. 6 and 7. In Fig. 6, A0 and A
equal to one when the foundation is rigid ( 100). For foundations subjected to horizontal force (Px 1), A0 approaches one as
increases to 10 while A equal to one and is not inuenced by .
For foundations subjected to rocking force (P0 1), A0 and A
approach one as increases to 10. It means that the horizontal
displacement of the foundation is also affected by the relative
stiffness ratio even if the horizontal stiffness of the foundation is
rigid. It is also noted that A due to rocking forces (P0 1) and A0
due to horizontal forces (Px 1) have the same values. In Fig. 7, An
for n 17 equal to zero when the foundation is rigid ( 100). For
foundations subjected to horizontal and rocking force, A1, A2, and
A3 reduce to zero as relative stiffness ratio increases to 3, 0.3, and
0.03, respectively. It indicates that the foundation can be assumed
as rigid at 3, two modes is sufcient to obtain accurate
displacements at 0.3, and three modes is enough at 0.03.
Hence, if the modal impedance functions for the rst three modes
are given, approximate solutions for the dynamic displacement of
the exible circular foundation with 0.03 subjected to horizontal force can be easily obtained. In addition, the dotted results in
Figs. 6 and 7 show the effect of the relaxed contact condition. The
more exible the foundation, the more inuence the contact
condition on the static vertical displacement.
Fig. 8 illustrates the inuence of the relative stiffness ratio
and mass ratio bz on the dynamic modal magnication factor M0. It
can be seen from Fig. 8 that the relative stiffness ratio has
inuence on the magnication factor M0 in the frequency range
a0 41. As the relative stiffness ratio decreases or the mass ratio
increases, the peak value of magnication factor M0 increases.
However, the relative stiffness ratio has no effect for 3 on the
magnication factor M0.
Fig. 9 illustrates the inuence of the relative stiffness ratio
and mass ratio bz on the modal magnication factor M1, which is
the rst mode of exible deformation modes. By comparing
Figs. 8 and 9, it can be found that the relative stiffness ratio and
mass ratio have much more inuence on M1 than on M0. The peak
value of M0 occurs primarily in the frequency range 1 oa0 o 2
while the location of the peak value of M1 is primarily in the
frequency range 3 oa0 o5 and strongly affected by the mass ratio.
In addition, the dotted results in Figs. 8 and 9 indicate that the
contact condition has more inuence on the high frequency range.
This phenomenon can be expected by observing modal impedance
functions shown in Fig. 5, which shows that the two set functions
are especially different in the high frequency range.
The inuence of the relative stiffness ratio and mass ratio bz
on the horizontal magnication factor M can be seen in Figs. 8(b),
(d) and 10. Fig. 8(b) and (d) shows that the relative stiffness ratio
and mass ratio have inuence on the magnication factor M in
the frequency range a0 41 when the foundation is subjected to
rocking force. It is noted that M due to rocking force P0 1 is the
same as M0 due to horizontal force Px 1. This phenomenon can be

191

explained by symmetric relation for the modal coupling impedance functions, i.e. Kox Kxo. In addition, Fig. 10 shows that only
the mass ratio has inuence on the horizontal displacement when
the foundation is subjected to horizontal force.

6. Conclusions
A method using modal analysis concept is developed to obtain
the dynamic vertical displacements of a circular exible foundation on soil media subjected to horizontal and rocking motion. The
vertical displacements of the foundation and contact stresses are
expressed by a set of modal coordinates corresponding to free
vibration mode shapes of a circular plate, and the interaction
effects between the foundation and the underlying soil are
represented using modal impedance functions. An efcient procedure with ring elements is developed to compute modal
impedance functions. The advantage of the proposed method
features reducing the computational work signicantly to analyze
the dynamic response of exible foundations. In addition, the
accuracy of the proposed method has been veried by existing
solutions and the computer program SASSI for soilstructure
interaction analysis.
Parametric analyses are performed for modal responses of a
circular exible foundation on an elastic half-space subjected to
foundation vibrations. The results indicate that the coupling
vertical displacements due to horizontal force or the coupling
horizontal displacements due to rocking force of the foundation
are signicantly inuenced by relative stiffness ratio among the
foundation and the soil medium, vibration frequency range,
foundation mass, and the boundarycontact condition.
The relaxed boundary condition may be used to predict
approximately the vertical displacement due to rocking force.
However, the welded boundary condition should be considered
to predict the coupling response. For a foundation with relative
stiffness ratio 3, the foundation can be considered as rigid to
calculate coupling displacement. For a slightly exible foundation
with 0.03 o3, three modes are considered sufcient to obtain
accurate coupling displacements of the foundation. For the frequency a0 o1, the coupling effect due to higher mode (i.e. exible
deformation mode) can be neglected.

Acknowledgments
The authors are grateful for the sponsor of the National Science
Council of Taiwan, Republic of China.

Appendix A. Summary of owchart

See Appendix Figs. A1 and A2

192

S.-S. Chen, J.-G. Hou / Soil Dynamics and Earthquake Engineering 69 (2015) 182195

Fig. A1. Summary owchart for derivation of equations of motion in modal coordinate.

S.-S. Chen, J.-G. Hou / Soil Dynamics and Earthquake Engineering 69 (2015) 182195

193

Fig. A2. Summary owchart for calculation of modal impedance functions.

Appendix B. Numerical example

Consider a circular exible plate resting on a uniform halfspace. The engineering properties of the foundation-soil system
are listed in Table B1. The exible plate is subjected to a 90 kN
horizontal force and a 270 kNm rocking moment induced by a
vertical force with linear distribution.
The following questions are to be solved by the proposed
method:

(a) Determine the static responses of the plate by using the rigid
assumption.
(b) Determine the static responses of the plate by using the modal
impedance in Fig. 5.
(c) Estimate the static responses of the plate by using dimensionless parameter (An).
(d) Suppose that the external force harmonically vibrates with a
frequency 20 Hz. Evaluate the dynamic amplitude of the plate
by using the modal magnication factor (Mn).

194

S.-S. Chen, J.-G. Hou / Soil Dynamics and Earthquake Engineering 69 (2015) 182195

Considering the static state (a0 0) and two rocking modes in

Eq. (30) give

Table B1
Engineering properties of the foundation system.
Circular plate

Soil

p 0.25
p 2.40 t/m3
Ep 21,525,562 kN/m2
R 5 m
hp 1 m

k00
6
Gs R4 k01
k0x

s 0.33
s 2 t/m3
Vs 300 m/s
Gs sV2s 180,000 kN/m2

k10

1 k11
4

k1x

9 8 9
38
kx0 > Y 0;st > > P 0 >
<
= < =
7
kx1 5 Y 1;st P 1
>
: >
; >
:P >
;
kxx
st
x

B  5

According to Fig. 5, the modal impedance can be used as

follows:
2

16  1
1 6 16   0:4
9004
16  0:048

For Question (a)

According to Eq. (19), P 0 2RM 108 kN.
Considering the static state (a0 0) for a rigid foundation, Eq.
(31) can be rewritten as
"

K S;00

K S;0X

K S;X0

K S;XX

#(

Y Rg;st

Rg;st

P0

)
B  1

Px

For obtaining the static response of the circular rigid foundation, existing solutions listed below for the static stiffness of the
foundation are used
K S;R 4GS R3 ; K S;C 0:382GS R2 ; K S;xx 4:8GS R

B  2

where KS,R, KS,C, and KS,xx are rocking stiffness, coupling stiffness,
and horizontal stiffness, respectively, for circular rigid foundation
on a half-space with s 0.33.
According to Eq. (33) and Eq. (34)
K S;00

4K S;R
R2

16GS R; K S;x0

2K S;C
0:764GS R
R

16   0:4
4

0:267  4:52 16  1:97

16  0:048

16  0:03 7
5

16  0:03

4:8

8
9 8
9
>
< Y 0;st >
= >
< 108 >
=
0
 Y 1;st
>
>
: >
; >
:
;
90
st
8
9
8
9
6:03 >
>
>
< Y 0;st >
=
<
=
1
0:33
) Y 1;st
mm
>
>
>
>
900
: ;
:
;
17:82
st

Thus


2r
0:33
r
r
wst r;  6:03
900  5 cos 900  3:833J 1 4:525  0:075I 1 4:525 cos

wst r 5;

11:37
5:53
mm; wst r 2;
mm
900
900

Comparing with responses of rigid foundations, the signicant

displacement difference occurs in inner location (r 2) while
minor displacement difference occurs in outer location (r 5).

B  3

Substituting Eqs. (B-2) and (B-3) into Eq. (B-1) leads to

) 

( Y

16
0:764
108
Rg;st

GS R
B  4
Rg;st
0:764
4:8
90

For Question (c)

According to Figs. 6 and 7, the normalized static modal
amplitudes (A0 and A1) can be approximated as
A0  1:020for P 0 1 and A0  0:980for P H 1

Therefore, the static response of circular rigid foundation can

be obtained as follows:
(
)


Y Rg;st
0:06298
 0:01002
1

Rg;st
180; 000  5  0:001  0:01002
0:20993




5:90
108
1
mm

900 17:81
90
wRg;st r; Y Rg;st 0 r; Y Rg;st
wRg;st r 5;

2r
11:8 r
cos
cos mm
R
900 5

11:8
4:72
mm; wRg;st r 2;
mm
900
900

A1  0:045for P 0 1 and A0  0:035for P H 1

Besides, from the Solution (a) we know

For P 0 1; Y Rg;st

0:06298
 0:01002
mm; For P H 1; Y Rg;st
mm
900
900

Therefore, the static response of circular exible foundation can

be approximated as follows:
Y 0;st P 0  A0;P0  Y Rg;st P 0 1 P H  A0;PH  Y Rg;st P H 1 6:05
900 mm

For Question (b)

Y 1;st P 0  A1;P0  Y Rg;st P 0 1 P H  A1;P H  Y Rg;st P H 1

From Eq. (20), the relative stiffness ratio can be calculated as

follows:

D
3

Gs R

3
E p hp
3
2 G
sR
p

121 

0:267

0:275
mm
900

The above responses approximately equal to those solved from

Question (b).

S.-S. Chen, J.-G. Hou / Soil Dynamics and Earthquake Engineering 69 (2015) 182195

For Question (d)

From Eqs. (21) and (22), the mass ratio and dimensionless
frequency can be calculated as follows:

bz

mp R2

s R 3

hP p R2

s R3

0:754; a0

R
Vs

2 f R
2:1
Vs

According to Figs. 8 and 9, the modal magnication factors (M0

and M1) can be approximated by using interpolation (between
bz 0 and 1) as
M 0  1:53for P 0 1 and M 0  1:89for P H 1
M 1  1:86for P 0 1 and M 1  2:8for P H 1
Therefore, the dynamic amplitude of circular exible foundation can be approximated as follows:
Y 0 P 0  M0;P 0  A0;P 0  Y Rg;st P 0 1 P H  M 0;P H  A0;P H  Y Rg;st P H 1
0:06298
0:01002 8:61
 90  1:89  0:98 

mm
108  1:53  1:02 
900
900
900
Y 1 P 0  M1;P 0  A1;P 0  Y Rg;st P 0 1 P H  M 1;P H  A1;P H  Y Rg;st P H 1
0:06298
0:01002 0:463
 90  2:8  0:035 

mm
108  1:86  0:045 
900
900
900

The dynamic magnication effect can be evaluated as follows:

Y0
8:61
1:42;

Y 0;st 6:05

Y1
0:463
1:68

Y 1;st 0:275

References
[1] Chen SS, Shi JY. A simplied model for coupled horizontal and rocking
vibrations of embedded foundations. Soil Dyn Earthq Eng 2013;48(5):20919.

195

[2] Chen SS, Shi JY. A response-based simplied model for vertical vibrations of
embedded foundations. Soil Dyn Earthq Eng 2011;31(56):77384.
[3] Chen SS, Shi JY. Simplied model for torsional foundation vibrations. Soil Dyn
Earthq Eng 2007;27(3):2508.
[4] Chen SS, Shi JY. Simplied model for vertical vibrations of surface foundations.
J Geotech Geoenviron Eng ASCE 2006;132(5):6515.
[5] Oien MA. Steady motion of a plate on an elastic half space. ASME J Appl Mech
1973;40:47884.
[6] Savidis SA, Richter T. Dynamic response of elastic plates on the surface of the
half-space. Int J Numer Anal Methods Geomech 1979;3(3):24554.
[7] Iguchi M, Luco JE. Dynamic response of exible rectangular foundations on an
elastic half-space. Earthq Eng Struct Dyn 1981;9(3):23949.
[8] Whittaker WL, Christiano P. Dynamic response of plate on elastic half-space.
(EM1)J Eng Mech Div 1982;108:13354.
[9] Rajapakse RKND. Dynamic response of elastic plates on viscoelastic half space.
J Eng Mech 1989;115(9):186781.
[10] Gucunski N, Peek R. Vertical vibrations of circular exible foundations on
layered media. Soil Dyn Earthq Eng 1993;12(3):18392.
[11] Gucunski N. Rocking response of exible circular foundations on layered
media. Soil Dyn Earthq Eng 1996;15(8):48597.
[12] Chen SL, Chen LZ, Pan E. Dynamic response of a exible plate on saturated soil
layer. Soil Dyn Earthq Eng 2006;26(67):63747.
[13] Triantafyllidis T, Prange B. Rigid circular foundation: dynamic effects of
coupling to the half-space. Soil Dyn Earthq Eng 1988;7(1):4052.
[14] Chen SS, Hou JG. Modal analysis of circular exible foundations under vertical
vibration. Soil Dyn Earthq Eng 2009;29(5):898908.
[15] Leissa AW. Vibration of plates. Washington: Scientic and Technical Information Division, National Aeronautics and Space Administration; 1969.
[16] Kausel E, Roesset JM. Stiffness matrices for layered soils. Bull Seismol Soc Am
1981;71(6):174361.
[17] Wolf JP. Dynamic soilstructure interaction. Englewood Cliffs, NJ: PrenticeHall; 1985.
[18] Lysmer J, Tabatabaie M, Tajirian F, Vahdani S, Ostadan F. SASSI, a system for
analysis of soilstructure interaction. (UCB/GT/81-02). Berkeley: University of
California; 1981.
[19] Wolf JP, Darbre GR. Dynamic-stiffness matrix of surface foundation on layered
halfspace based on stiffness-matrix approach. In: Proceedings of the summary
report of IAEA specialists' meeting on gas-cooled reactor seismic design
problems and solutions, IAEA IWGGCR/6 UC-77, General Atomic Company,
San Diego, California; 1983.