Professional Documents
Culture Documents
d e S t S
t i f
bb
f
bb
)] ( [
2
1
)] ( [
(1)
When the excavation is taken into account, the time domain dynamic stiffness
matrix )] ( [ t S
g
bb
of the excavated soil can also be acquired by a formula similar to
(1), with the superscripts being changed to g.
The expression of the dynamic stiffness matrix in the frequency domain
( )] ( [
g
bb
S ) can be determined using several different methods, including the
boundary element method,
[1]
the infinite element method,
[8]
and the Greens
Function.
[3]
f g
b
f
bb
f
b
S u
g
bb
g
b
S u
e
e
bb
S
Fig. 4 Reference Soil System
7
With the dynamic stiffness given, the basic equation of motion in the time
domain can be formulated as:
'
'
'
'
1
]
1
t
g
b
g
bb
t
t
b
g
bb b
s
t
b
t
s
bb bs
sb ss
d u t S
d u t S t P
t P
t u
t u
M M
M M
0
0
} ) ( )]{ ( [
} 0 {
} ) ( )]{ ( [
} 0 {
} ) ( {
} ) ( {
} ) ( {
} ) ( {
] [ ] [
] [ ] [
& &
& &
(2)
In the equation, ) (t u
t
denotes the vector of the total displacement;
) (t u
g
b
represents the so-called scattering motion; [] is the Mass matrix and {} is
the vector of the nonlinear internal forces of the system consisted of the structure
and the adjacent soil. The interaction forces acting on the boundary nodes are
equal to the convolution integral of the dynamic stiffness matrix )] ( [ t S
g
bb
and the
displacement relative to the ground.
d u u t S t R
g
b
t
b
t
g
bb b
}) ) ( { } ) ( )]({ ( [ } ) ( {
0
(3)
Material damping of the soil is very important to the analysis of the whole
dynamic soil-structure interaction system. It has been discovered that different
damping models used in the numerical soil-structure interaction models usually
bring very different results.
[7]
Since identification of the soil damping is still a
difficult problem, linear visco-elastic damping models, such as the Voigt Model
(Fig 5a) and the Three-Parameter Kelvin model, (Fig. 5b) are generally adopted.
In these two models, material damping is introduced by using a complex
material constant E
*
.
(a) Viogt Model (b) Three-Parameter Kelvin Model
Fig. 5 Linear Visco-Elastic Damping Models
8
In the Voigt Model:
) 2 1 (
*
i E E + (4)
In the Three-Parameter Kelvin Model:
i
i
E E
+
+
1
2 1
*
(5)
In both models, E is the Youngs medullas of the unbounded soil, and is the
damping ratio. Material damping of the soil is not introduced into the soil-
structure interaction system implicitly, but by replacing the Youngs medullas E
in the dynamic soil stiffness matrix with E
*
.
Another important factor that affects the analysis of this numerical model is
the constitutive model of the soil adjacent to the structure. In the 1980s, when
this model was established, the nonlinear elastic soil constitutive laws were
generally adopted.
Future Prospects and Recommendations
Because of the reasons stated ahead, future numerical models of dynamic soil
structure interaction are likely to continue to be concentrated in the time domain.
And direct method may become more attractive to researchers and engineers
because it can consider the nonlinearity of the unbounded soil as long as
sufficiently accurate transmitting boundaries can be developed.
[4]
However,
resulting transmitting boundaries may not be as simple as attaching certain
masses, springs and dashpots directly to the interaction horizon and may come
up with sophisticated coupling both in time and space.
[10]
The substructure
method, on the other hand, with its clear concept and rigorous formulation, will
also continue to develop.
Although current research are focused on such places as the interaction
horizon, the unbounded soil and so on, future research may pay more attention
to the soil adjacent to the structure, since it is this region that is in consistent
dynamic contact with the vibrating structure. Such behavior of the soil as its
9
separation from the structure foundation may change the behavior of the whole
system drastically. To address this problem, new forms of finite elements have
been developed to model the soil-structure interface, such as the Goodman
Contact Element and the Thin Layer Element. It can be foreseen that more work
will be done in this area.
One other important phenomenon associated with soil-structure interaction is
the liquefaction of the soil adjacent to the structure. On the one hand, it is very
hard to judge when and where exactly soil liquefaction will occur during soil-
structure interaction in earthquakes. On the other hand, once the soil liquefies, it
will become so soft that the supporting force of the structure will be totally lost,
which is a very difficult situation to model. There has been some research done
to address the problem of pure soil liquefaction, but little work has been done to
introduce this issue into the numerical model of dynamic soil-structure
interaction. Apparently, there should be and will be more research carried out in
this field.
In one word, since numerical modeling of dynamic soil structure is still at its
developing stage, future models will consider more affecting factors. And only
this way, can more accurate results be achieved.
Conclusion
Dynamic soil-structure interaction is an important part that should be
included in the seismic analysis when the structures are stiff and the soil is soft.
Numerical models using the substructure method are very rigorously formulated
but may not be able to consider some of the nonlinear properties of the system,
whereas the models using the direct methods can take the nonlinearities of the
unbounded soil into consideration but approximate transmitting boundaries
have to be included. In both methods, damping from the soil is very important.
Developments of both methods are now concentrated in the time domain and
may continue to be in the future. More study should be carried out to understand
10
and model the phenomena of soil-structure separation and soil liquefaction
better. Currently, if nonlinearity of the far field soil is not an important issue in
the analysis, the substructure method is more practical; otherwise, the direct
method is more desirable.
References:
[1] M. Nuray Aydinoglu, Development of Analytical Techniques in Soil-structure
Interaction, Developments in Dynamic Soil-Structure Interaction, Kluwer
Academic Publishers, 1992
[2] lysmer, J., Kuhlemeyer, R.L. Finite Model for Infinite Media, Journal of
Engineering Mechanics Division, ASCE, V 95, pp. 377-392, 1969
[3] Wolf, J. P. and Oberbhuber, P., Non-linear Soil-structure Interaction Analysis
Using Greens Function of Soil in the Time Domain, Earthquake Engineering and
Structural Dynamics, v 13 n 2, pp. 213-223 , Mar-Apr 1985
[4] Wolf, J. P., Soil-structure Interaction in Time Domain, Prentice-Hall,
Englewood Cliffs, N.J., 1988
[5] Wolf, J.P. a Comparison of Time-Domain Transmitting Boundaries, Earthquake
Engineering and Structural Dynamics, V 14, pp.655-673, 1986
[6] Bernal, Dionisio and Youssef, Akram, A Hybrid Time Frequency Domain
Formaulation for Nonlinear Soil-structure Interaction, Earthquake Engineering and
Structural Dynamics, v27, pp.673-685, 1998
[7] Yazdchi, M., Khalili, N. and Valliappan, S. Dynamic Soil-structure Interaction
Analysis Via Coupled Finite-Element-Boundary-Element Method, Soil Dynamics and
Earthquake Engineering, v18, pp.499-517, 1999
[8] Doo-Kie Kim and Chung-Bang Yun, Time Domain Soil-structure Interaction
Analysis in Two-Dimensional Medium Based on Analytical Frequency-dependent
Infinite Elements, International Journal of Numerical Methods in Engineering, v
47, pp.1241-1261, 2000
11
[9] Xiong Zhang, J.L. Wegner and J.B. Haddow, Three Dimensional Dynamic Soil-
structure Interaction Analysis in the Time Domain, Earthquake Engineering and
Structural Dynamics, v 28, pp. 1501-1524, 1999
[10] Wolf, J. P. Consistent Lumped-Parameter Models for Unbounded Soil: Frequency
Independent Stiffness, Damping and Mass Matrix, Earthquake Engineering and
Structural Dynamics, v 20, pp. 33-42, 1991