You are on page 1of 5

Modeling of Soil-Structure Interaction as Finite

Element Using Using SAP2000

Chirag N. Patel Kinjalraj Sharma H. S. Patel


Ph.D. Candidate P.G. Student Associate Professor
Faculty of Engineering Applied Mechanics Department Applied Mechanics Department
Pacific Academy of Higher L. D. College of Engineering L. D. College of Engineering
Education and Research University Ahmedabad, Gujarat, India Ahmedabad, Gujarat, India
Udaipur, Rajasthan, India maxorgan@gmail.com dr.hspatel@yahoo.com
cnpatel.693@gmail.com

Abstract—Software tools are nowadays become very important In addition to the determination of static behaviour,
and essential for the structural problems where a lack of well- determination of the dynamic behavior of tanks which used for
established methods is available for analysis. Soil Structure many engineering fields is also important. Soil-structure and
Interaction (SSI) analysis is similar kind of problem where all the fluid-structure interaction are not important for ordinary
information about analysis is remains up to research level. In this structures as to be for tanks are effective for estimation of
paper, a finite element elevated reinforced concrete water tank is dynamic behaviour of elevated tanks subjected to dynamic
modeled and analysed considering SSI. For mounting soil effect loads. Thus it is necessary to consider the interaction effects
at the base of tank, widely used cone model and elastic solid correctly for elevated tanks subjected to dynamic loads like
modeling techniques has been selected. In first method,
earthquake. It is seen from literature survey that very few
equivalent spring model is used at the base of water tank and for
other soil is modeled as elastic solid body. The effectiveness of
studies have been carried out for elevated tanks considering
both methods is discussed as well as output results are compared interaction effects and related parameters. The basic difference
and contrast between modeling techniques using SAP2000 of fixed base analysis and soil-structure interaction analysis is
structural software. that; the nodal displacement of the structure are relative to
moving base in soil-structure interaction and base movement is
Keywords-elevated water tank; soil structure interaction; spring function of base material provided. There are few research has
support; finite element; SAP2000 program been done to obtain base movement by equivalent spring model
in which value of horizontal stiffness and rotational stiffness
I. INTRODUCTION are found out using either empirical relation and theoretical
Liquid storage tanks are very essential structures in water as approximation. The another approach is to model soil as solid
well as oil and gas industries. On the other hand a number of elastic material and providing a firm interface between super
cases of damage to tanks have been observed due to past structure and soil and then whole system is analysed together as
earthquakes. Concerning the importance of these systems, a single system. In this paper a comparison of seismic response
especially their seismic safety for avoiding the adverse of an elevated reinforced concrete water tank is compared
consequences such as fires, explosions and environment using both the method. Modeling by both the method is
pollution, better understanding of their seismic behavior still discussed and results are compared with fixed base analysis
seems necessary. There are several factors, which affect the results.
dynamic characteristics of a liquid storage tank. It is generally
recognized that the seismic response of a structure supported II. SOIL - STRUCTURE INTERACTION
on flexible ground may diverge considerably from that of the Soil-structure interaction between a structure and the
same structure supported on a non-deformable, rigid base. This supporting soil; is a phenomenon that influences many aspects
difference is due to the effect of the soil-structure interaction in the design of a structure: safety, serviceability (crack
(SSI). Over the last 40 years, the dynamic soil-structure propagation) and costs. Also, produce a significant effect on
interaction has attracted an interest among researchers and the response of the structure and the resulting localization of
engineers in the fields of structural dynamics, wave mechanics damage during seismic events. All structures are effected soil-
and soil dynamics over the world. The methods of their structure interaction with varying emphasis in the earthquake
investigations consist of experimental study and analysis excitation as known. Especially, for structures like elevated
research. The analysis methods are numerical simulation water tank with slenderness and heavy mass at the top, the soil
methods. Due to the underdevelopment of computer effects should be considered for different soil properties as
technology, analytical method was popular in the 1970’s. substructure method to consider soil effects [1]. The estimation
Along with the rapid progress in the art of computer science, of earthquake motions at the site of a structure is the most
now software simulations are widely used to the study on soil- important phase of the design or retrofit of a structure. It is
structure interaction. generally assumed that the elevated tanks are fixed with the
ground. So, attention is focused on the dynamic behavior of the elevated tank under seismic load of Fluid-Structure-interaction
fluid and/or supporting structure, but the effects of subsoil on problems is investigated by using added mass approach
the dynamic behavior of elevated tanks also equally important. suggested by Westergaard (1931).
Haroun and Ellaithy (1985) developed a model including an
analysis of a variety of elevated rigid tanks undergoing
translation and rotation; the model considers liquid sloshing
modes; and it assesses the effect of tank wall flexibility on the
earthquake response of elevated tanks. Resheidat and Sunna
(1986) investigated the behavior of rectangular elevated tank
during earthquakes considering soil-foundation-structure
interaction. They neglected the sloshing effects on the seismic
behavior of the elevated tanks. Haroun and Temraz (1992)
analyzed models of two-dimensional X-braced elevated tanks
supported on isolated footings to investigate the effects of
dynamic interaction between the tower and the supporting soil-
foundation system and they neglected the sloshing effects too.
As seen from studies mentioned above, very few studies have
been carried out on the soil-structure interaction effects for
elevated tanks. Generalised problem for dynamic fluid-elevated Figure 2. FEM model for fluid-strucutre-interaction added mass approach
tank-soil interaction can be seen from Fig.1. If the soil material
can be considered linear, the SAP2000 program, using the solid The general equation of motion for a system subjected to an
element, can calculate either the one-, two- or three- earthquake excitation can be written as,
dimensional free-field motions at the base of a structure.
𝑀𝑢̈ + 𝐶𝑢̇ + 𝐾𝑢 = −𝑀𝑢̈ 𝑔 (1)
In which M, C and K are mass, damping and stiffness
matrices, 𝑢̈ , 𝑢̇ and 𝑢 are the acceleration, velocity and
displacement respectively, and is the ground acceleration. In
the case of added mass approach the form of equation (1)
become as below:

𝑀∗ 𝑢̈ + 𝐶𝑢̇ + 𝐾𝑢 = −𝑀∗ 𝑢̈ 𝑔 (2)


In which M* is the new mass matrix after adding
hydrodynamic mass to the structural mass. While the damping
and stiffness matrices, are same as in equation (1).

IV. DISCRIPTION OF TANK MODEL


Figure 1. Generalised model for dynamic soil-elevated tank interaction
A reinforced concrete elevated tank is considered in seismic
Different type of soil properties has been considered for the analysis with frame supporting structure in which columns are
analysis as shown in Table I. connected by the circumferential beams at regular intervals.
The tank container is of the Intze type and filled with water to
TABLE I. DIFFERENT TYPE OF SOIL PROPERTIES
a density of 1000 kg/m3. More selected data are as described
Modulus of
Poisson Ratio
Weight per Unit in Table II. To evaluate the dynamic response of the elevated
Type of Soil Elasticity (E) Volume tanks, considering three cases; soft, medium and hard soil
(υ)
kN/m2 (kN/m3)
Hard 397649 0.3 19.8
were simulated using time history data of 6.9 magnitude
Medium 140631 0.3 18.5 earthquake with peak ground acceleration of 0.644 g as shown
Soft 33540 0.3 13.5 in Fig. 3.

TABLE II. STRUCTURAL DATA FOR FRAME TYPE STAGING


III. FLUID STRUCTURE INTERACTION Capacity of the tank 1000 m3
The water tank is modeled as two mass model as per Unit weight of concrete 25 kN/m3
GSDMA guidelines [2] and suggested by George Thickness of Top Dome 0.15 m
W. Housner (1963) [3]. The convective mass is added as lump Rise of Top Dome 2.2 m
mass with member having rotational restrains at node while Size of Top Ring Beam 0.35 m × 0.35 m
released in translation and the impulsive mass is added in the Diameter of tank 13.6 m
density of cylindrical wall as shown in Fig. 2. The analysis of Height of Cylindrical wall 6.8 m
Thickness of Cylindrical wall 0.33 m
Size of Middle Ring Beam 1.2 m × 0.6 m Therefore, Wolf and Meek (1992, 1993) have proposed a
Rise of Concal dome 2.35 m cone model, an example of the substructure method, for
Thickness of Conical shell 0.5 m evaluating the dynamic stiffness and the effective input motion
Rise of Bottom dome 1.6 m of a foundation on the ground as shown in Fig. 4. Wolf and
Thickness of Bottom dome shell 0.2 m Preisig (2003) adopted this method for the layered half-space.
Number of Columns 8
Compared to more rigorous numerical methods, this cone
Number of Bracings Level 3
model requires only a simple numerical manipulation within
reasonable accuracy (Wolf 1994, Takewaki 2003). In this
Size of Bottom Ring Beam 1 m × 1.2 m
method super structure is assumed to be supported on springs.
Distance between intermidiate bracing 4m
These springs represent the horizontal, rocking, vertical and
Height of Stsging above Foundation 16 m torsion stiffness of the soil. The stiffness of the spring depends
Diameter of Columns 0.75 m on the modulus of sub grade reaction of soil. The most popular
Size of Bracing 0.5 m × 0.5 m model is the Winkler's model. The basic limitations of Winkler
hypothesis is that this model cannot account for the dispersion
of the load over a gradually increasing influence area with
increase in depth. Springs are assumed independent, so the
effect of the externally applied load gets localized to the sub
grade only to the point of its application. This implies no
cohesive bond exists among the particles comprising soil
medium. The raft is model as thick plate and elastic modulus is
provided to raft material in such manner, it behaves as a rigid
body. The values for spring model are calculated as follows
considering freedom with corresponding apex ratio (opening
angle), wave-propagation velocity and distortion.

Figure 3. Generalised model for dynamic soil-elevated tank interaction

V. EQUIVALENT SPRING MODEL


There are different ways to consider the soil-structure
interaction. First of them is the modifying method that can be
constructed by modifying the fixed base solution of the
structural system (Veletsos and Meek 1974). This method has
been widely used in the studies (Aviles and Suarez 2001) and
the codes such as ATC-1978 (Veletsos etc. 1988), FEMA 368-
369 (2001), and Eurocode-8 (2003). To represent the property
of the elastic or viscoelastic half-space accurately, the spring Figure 4. Cones for various degrees of freedom (Wolf 1994).
and dashpots are required to depend on the frequency of
excitation (Wu and Smith 1995). The second one is the In SAP2000 software, water tank is modeled by providing
substructure method that can consider the frequency-dependent spring of calculated stiffness in different directions at the base,
or independent dynamic stiffness and the damping of the directly through joint spring and also restrain in the horizontal
soil/foundation system. If the frequency-dependent dynamic and rotational direction. As shown in Fig. 5, static stiffness
stiffness or the damping is required to be considered, the values of rigid circular foundation is considered and provided
governing equation for the structure-foundation system is in form of spring as per equation given in Table III; where, G:
expressed and solved in the frequency domain using Fourier or shear modulus, r0: radius of circular foundation, υ: Poisson
Laplace transformation (Wu and Smith 1995, Aviles and Perez- ratio.
Rocha 1998, Takewaki 2003). The third one is the direct
method. Here, finite element and boundary element methods or TABLE III. STATIC STIFFNESS VALUES OF RIGID CIRCULAR FOUNDATION
a mixture of these are used in the time or frequency domain
Vertical Horizontal Rocking Torsional
(Wolf and Song 1996a, 1996b, Wolf 2003). The most striking Stiffness
(Kv ) (KH ) (KR ) (KT )
feature in an unbounded soil is the radiation of energy towards
infinity, leading to so-called radiation damping even in a linear Foundation 8𝐺𝑟03
4𝐺𝑟0 8𝐺𝑟0 16𝐺𝑟03
with no
system. Mathematically, in a frequency-domain analysis, the embedment 1−𝜐 2−𝜐 3(1 − 𝜐) 3
dynamic stiffness relating the amplitudes of the displacements
to those of the interaction forces in the nodes of the structure-
soil interface of the unbounded soil is a complex function. This
occurs when the unbounded soil consists of a homogeneous
half-space (Wolf 2002).
Figure 5. Spring assignment to the base of SAP2000 model

VI. EQUIVALENT ELASTIC SOLID MODEL

Figure 7. Material property of elastic solid at the base of SAP2000 model

VII. RESULTS AND CONCLUSIONS


Roof and sloshing displacement are compared with respect
to time, for time history data of Loma Prieta. Graph
overlapping has been displayed over here. The Base reaction of
water tank is not directly attained in solid model as SAP2000
shows base reaction of support while we need reactions at the
base of tank. Hence the base reaction is obtained from column
shear and moment and only maximum values are compared
with spring model. Here all displacements are in meter and
base shear in kN and base moment in kNm.

Figure 6. Elastic solid to the base of SAP2000 model

In this method the whole soil mass is divided into small


elements connected at node points. Then theses elements may
assign the corresponding soil property. Disadvantage of this
method is high cost of analysis. In present study finite element
modeling has been done as elastic solid model as shown in Fig.
6. Properties for solid element has been applied based on
properties of different type of soil; especially modulus of
elasticity and poison ratio as shown in Fig. 7. In this type of
Figure 8. Comparision of roof displacement time history for soft soil
model it is assumed that soil has only finite extent in all
direction. Means infinite extent of soil is converted into finite
one. In present study the width and depth of soil is taken as 5
and 3 times of water tank base dimension relatively in the
corresponding direction. At the base of system fixed supports
are provided and at periphery vertical rollers are provided.
Here only elastic modulus of soil and Poisson’s ratio of
undrained condition is taken and soil is modeled as eight
nodded brick element with three degrees of freedom at each
nod. The size of the elements should be such that aspect ratio
remains between 1 - 1.5.

Figure 9. Comparision of roof displacement time history for medium soil


and analysis. Because running the analysis of fine mesh solid
model consumes lots of time and processer usage. But for
medium or soft soil it is preferable to perform solid modeling.
REFERENCES
[1] R. Livaoglu and A. Dogangun, “A Simple Seismic Analysis Procedure
for Fluid-Elevated Tank-Foundation/Soil Systems”, 6th International
Congress on Advances in Civil Engineering, October – 2004, pp. 570-
580.
[2] IITK-GSDMA Guidelines for Seismic Design of Liquid Storage Tanks
Provisions with commentary and explanatory examples.
[3] George W. Housner, “The dynamic behaviour of water tanks”, Bulletin
of the Seismological Society of America, February-1963, Vol-53, No.2,
Figure 10. Comparision of roof displacement time history for hard soil Pg. 381-387.
[4] Edward L. Wilson, “Three-Dimensional Static and Dynamic Analysis of
The graph shows that for hard soil, ther is not much Structures,”University of california at Berkely.
variance in pick value of both model but just a little bit increase [5] Wilson. E.L. and Habibullah. A. 2003, “Structural Analysis Program
in solid model compared to spring model. Same thing is SAP2000. “ User’s manual, Computers and Structures, Inc., Berkeley,
obsereved in medium soil for the roof displacement. In the soft Calif.
soil, sudden and very large variation occurs and that is quite
visible compared to medium and hard soil.

Figure 11. Comparision of base shear for different soil and base

Figure 12. Comparision of overturning moment for different soil and base

From above results, it is proved that elastic solid model


give lower response compared to spring model. Also, it has
been seen that soft soil strata for spring base gives much
change in the value of overturning moment. So, it can be useful
to say that method of modeling depends on soil configuration
as well as confinement of finite elements also. It means if a
hard stratum is observed beneath the structure to be analyzed, it
is favorable to use spring model as it saves time for modeling

You might also like