Professional Documents
Culture Documents
Othman A. Karim
Professor, Department of Civil and Structural Engineering and Director of
Centre for Information Technology, University Kebangsaan Malaysia, Bangi,
Selangor, Malaysia
e-mail: oak@vlsi.eng.ukm.my
Anuar Kasa
Senior Lecturer, Department of Civil and Structural Engineering, University
Kebangsaan Malaysia, Bangi, Selangor, Malaysia
e-mail: anuar@vlsi.eng.ukm.my
ABSTRACT
This paper is concerned with the soil and water behaviour of elevated concrete water tank
under seismic load. An artificial seismic excitation has been generated according to Gasparini
and Vanmarcke approach, at the bedrock, and then consideration of the seismic excitation
based on one dimension nonlinear local site has been carried out. Seven cases are chosen to
make comparisons with direct nonlinear dynamic analysis, mechanical models with and
without soil structure interaction (SSI) for single degree of freedom (SDOF), two degree of
freedom (2DOF), and finite elements method (FEM) models. The analysis is based on
superposition modal dynamic analysis. SSI and fluid structure interaction (FSI) have been
accounted using direct approach and added mass approach respectively. The results show that
a significant effects obtained in shear force, overturning moment and axial force at the base of
elevated tank.
KEYWORDS:
INTRODUCTION
Elevated concrete water tanks are mainly used for water supply and fire protection. One of
the major problems that may lead to failure of these structures is earthquakes. Therefore the
analysis of elevated tank must be carefully performed, so that safety can be assured when
earthquake occurs and the tanks remain functional even after earthquake. The irregular shape of
- 387 -
388
the elevated water tanks for which most of the mass confluent in the upper part of the tank makes
it more sensitive to any dynamic load, especially due to an earthquake.
Elevated water tank can be simulated based on SDOF, 2DOF or FEM, which governed by
one mode, two modes, or more respectively. It is widely recognized that these analysis are not
always the appropriate approach for simulating response of structures subjected to seismic
excitation. The estimation of damages made using this approach is normally poor (Hamburger,
1996). The response of elevated water tank when dynamic effects are considered is deeply
dependent upon the soil deformability and liquid characteristics (Somnath, et al., 2004;
Livaoglua, and Dogangu, 2006; Livaoglua and Dogangu, 2007; Halil et al., 2008). Therefore
interaction between tank foundation and liquid should be accounted in the analysis of these
structures. Previous studies on the seismic behavior of elevated water tanks were only focusing
on the linear seismic response, therefore in this paper investigation on the behaviors of elevated
water tanks based on nonlinear dynamic analysis is presented.
Due to the lack of real seismic excitations recorded at the site that need to be considered in
this study, an artificial seismic excitation has been generated which is compatible with local
response spectrum at the bedrock. The local site effects are estimated based one nonlinear one
dimensional approach.
SEISMIC EXCITATION
The whole main steps to generate artificial earthquake and account the local site effects can
be described as in Figure 1. It is started from artificial earthquake generation at the bedrock which
is compatible with local response spectrum. The generated earthquake has to be used to
accounting the local site effects to predict the ground motion and ground response spectrum.
389
and Vanmarcke, 1976) is adopted. The procedure is based on the fact that any periodic function
can be expanded into a series of sinusoidal waves (Housner, 1955).
() =
sin( + )
(1)
where is the amplitude and is the phase angle of the contributing sinusoidal, the
amplitudes are related to power spectral density function () in the following way:
=
( ).
(2)
the relationship between () of seismic excitation and response spectrum can be expressed as
following
1
( )
1
4
( )
.
/
.
()
(3)
where rS.P is the peak factor; SV is the target velocity of the response spectrum; is the circular
frequency of the i contributing sinusoid; is the fictitious time-dependent damping factor for
duration t . The definitions of these parameters can be found in Gasparini and Vanmarcke (1976).
To simulate the transient character of real earthquakes, the steady-state motions are
multiplied by a deterministic envelope function (). The artificial motion () becomes:
() = () = ()
sin( + )
(4)
There are three different envelope intensity functions available such as trapezoidal,
exponential, and compound (Gasparini and Vanmarcke, 1976). The procedure then artificially
raises or lowers the generated peak acceleration to match exactly the target peak acceleration that
has been computed by using seismic hazard analysis
In this study the response spectrum is obtained from seismic hazard analysis by other
researchers (Azlan et al., 2005). Figure 2 shows the target and calculated response spectrum at the
bed rock, and artificial seismic excitation compatible with its target response spectrum.
390
0.5
1.5
1
0.5
0
-0.5
-1
-1.5
-2
Acceleration
Acceleration(g)
(g)
0.4
0.3
Target
Calculated
Generated
0.2
0.1
0
0
4
Period (Sec)
10
10
15
20
Period (Sec)
Figure 2: (a) Target and calculated response spectrum at the bed rock, (b) Artificial
seismic excitation compatible with target response spectrum
391
1
0.8
0.6
0.4
0.2
0
G/Gmax
G/Gmax
0.0001
0.01
Shear Strain (%)
1
0.8
0.6
0.4
0.2
0
0.0001
Clay
0.01
1
Shear Strain (%)
Sand
G/Gmax
1
0.8
0.6
0.4
0.2
0
0.0001
0.001
0.01
0.1
Gravel
Acceleration (g)
10
15
20
Time (sec)
392
Figure 6: Period (s) Vs Response Spectrum Acceleration (g) at surface (Site Class: D)
DYNAMIC ANALYSIS
The general equation of motion for a system subjected to an earthquake excitation can be
written as,
[M]a + [C]a + [K]a = F(t)
(5)
where M, C and K are the mass matrix, damping matrix, and stiffness matrix respectively. F(t) is
the seismic excitation, and time-dependent vectors , and are accelerations, velocities and
displacements, respectively. Solution of Equation (5) can be divided as: direct integration which
no transformation of Equation (5) to another form, and mode superposition, where some
assumption is used to simplify the solution of Equation (5).
f = [M]a
+ (1 + )[C]a
[C]a + (1 + )[K]a
[K]a
(6a)
393
a
a
= a + t [(1 )a + a
= a + a t + t [(1 )a + 2a
(6b)
]/2
(6c)
where , and are free parameters which govern the stability and numerical dissipation of the
algorithm. In this paper the matrix C in Equation (5), is expressed as a linear combination of the
mass and stiffness matrix.
[C] = A[M] + B[K]
(7)
where A and B are constant, and the above equation can be solved through model superposition.
Such assumption leads to a constant damping ratio for the entire system which includes soil,
foundation and tank. This assumption is not usually pertinent to soils, where damping depends
strongly on the development of share strains and may vary abruptly over the geometrical domain
of integration (Wolf, 1985). Therefore for this reason, and when the shear strains are lower than
the threshold value , a variable damping solution can be adopted, in which the Rayleigh
assumption holds good.
Single DOF
SDOF idealization of tanks is normally carried out in 2DOF idealization, but in some seismic
codes it is suggested that the tanks can be analyzed as a SDOF system as shown in Figure 7. The
ACI 371R-98 suggests that the single lumped mass model should be used when the water load is
80% or more of the total gravity load (ACI 371R-98, 1995). In this case the fundamental period
of the elevated water tank with SDOF can be obtained by Equation (8)
= 2
(8)
where is the single lumped mass weight and is the ground acceleration. It should be noted
that according to ACI 371R, is consisted of the total mass weight of container, water, and 2/3
of the concrete support wall, whilst is the lateral stiffness of the staging
394
2DOF Idealization
Normally most of the elevated tanks are not completely filled with liquid. Hence the seismic
behaviors of elevated water storage tanks subjected to earthquakes are characterized by two
predominant modes of vibration. The first mode is related to the mass that rigidly moves together
with the tank structure (impulsive mass) and the other mode corresponds to the liquid sloshing
(convective mass) (Housner, 1963).
Seismic response of elevated tanks depends on complex FSI that may result in global
overturning moments and base shear induced by horizontal inertial forces as shown in Figure 8.
1.84
(9)
395
where is the radius of tank, is depth of water in the container and is the mass of
convective water, is the height of and is the height of . Eurocode-8 (EC-8) suggested
simple procedure by recommended design values for the cylindrical tanks. For any tank shapes
other than circular and rectangular (like truncated or conical shape), the values of and shall
correspond to that of an equivalent circular tank of the same.
Table 1: Recommended design values for the first impulsive and convective modes of
vibration as a function of the tank height-to-radius ratio (h/R) (EC-8, 2003)
/
0.3
/
0.176
/
0.824
/
0.40
/
0.521
/
2.640
/
3.414
0.5
0.300
0.700
0.40
0.543
1.460
1.517
0.7
0.414
0.586
0.40
0.571
1.009
1.011
1.0
0.548
0.452
0.42
0.616
0.721
0.785
1.5
0.686
0.314
0.44
0.690
0.555
0.734
2.0
0.763
0.237
0.45
0.751
0.500
0.764
2.5
0.810
0.190
0.45
0.794
0.480
0.796
3.0
0.420
0.158
0.45
0.825
0.472
0.825
(10)
3(2 )
(11)
where , are the coefficients of dimensionless, is the poissons ratio, is the shear
modulus of the soil and is the radius of the circular foundation.
The flexible base period is evaluated from Veletsos and Meek (1974),
= 1+
(1 +
)
(12)
where .is the fundamental period of a fixed-base structure, is the stiffness of a staging based
on fixed-base and is the effective height that is taken as 70% of the height to the level where the
gravity load is effectively concentrated as shown in Figure 7.
396
T
T
(13)
where represents the contribution of the radiation damping, and the soil damping as shown in
Figure 10 and shall be determined by averaging the values obtained from solid lines and the dash
lines.
FEM Analysis
FEM idealization is usually used to analyze structures with complex or irregular geometry or
non-uniform loading such as elevated water tank. The model is usually constructed using 3dimensional (3D) FEM. The seismic input for a 3D model includes three orthogonal components
of seismic excitation, two horizontal and the vertical. These seismic excitations can be applied
along the principal axes of the structure.
The dynamic interaction with the foundation introduces flexibility at the base of the tank
model and could provide additional damping mechanisms through material and radiation
damping. The tank also interacts with the retained water through hydrodynamic pressures at the
structure-water interface. This interaction is coupled in the sense that motions of the tank generate
hydrodynamic pressures that affect deformations (or motions) of the tank, which in turn influence
the hydrodynamic pressures.
397
398
[M]a + [C]a + [k]a = F(t)
(14)
Figure 11: The FEM models for FSI added mass approach system.
Westergaard Model
In his study, Westergaard approach is chosen to analyze the FSI of the elevated tank. This
method was originally developed for the dams but it can be applied to other hydraulic structure
under earthquake loads i.e. water tank (Livaoglua and Dogangu, 2006). The convective mass ( )
is obtained according to EC-8 (Table 1) technique and being added to the tanks walls according to
Westergaard approach as shown in Figure 12 using Equation (15),
7
( )
8
(15)
(a)
(b)
Figure 12: (a) Westergaard Added Mass Concept,(b) Normal and Cartesian
399
In the case of Intze tank where the walls having sloped and curved contact surface, the
Equation 15 should be compatible with the tank shape by assuming the pressure is still expressed
by Westergaard's original parabolic shape. But the fact that the orientation of the pressure is
normal to the face of the structure and its magnitude is proportional to the total normal
acceleration at the recognized point. In general, the orientation of pressures in a 3-D surface
varies from point to point; and if it is expressed in Cartesian coordinate components, it would
produce added-mass terms associated with all three orthogonal axes. Following this description
the generalized Westergaard added mass at any point i on the face of a 3-D structure is expressed
by Kuo, (1982),
= =
(16)
where: is the tributary area associated with node ; is the normal direction cosine
( , , ) as shown in Figure 12(b) and is Westergaard pressure coefficient.
CASE STUDY
Description of Existing Elevated Concrete Tank
The elevated tank has a capacity of 250 m3 with the top of water level at about 17.8 m above
ground. The tank is spherical in shape, 8.6 m in diameter and 7.85 m in height at its centre. The
support consists of 6 vertical circular columns and the columns are connected by the
circumferential beams at regular intervals, at 4,8,12 and 16 m as shown in the Figure 13.
400
401
conditions by trial and error to evaluate reasonable results. The seismic loads are applied at
critical direction as shown in the Figure 14, to evaluate the seismic response at the base of
columns
Brace beam
Column
II
I
I) Critical direction for shear force in column.
II) Critical direction for axial force in column.
Numerical Analysis
Two methods of seismic analysis are employed for elevated tank. The first method is the
eigenvalue natural frequency analysis solves using spectral response methods. For the analysis,
the response spectrum for 5% of critical damping gives a ZPA of 0.231g Figure 6. The analyses
are solved using the complete quadratic combination (CQC) method in all cases except SDOF to
take account of the correlation between modes of similar frequencies.
The second method is a full time stepping implicit dynamics solution of elevated tank using
acceleration time histories as input to the base supports of the model. The acceleration time
history was taken from Figure 2 it has a zero period acceleration (ZPA) of 0.149g.
Several linear integration dynamic analyses are made to investigate the performance of an
elevated tank through the implicit time stepping FEM dynamic analysis with assumption that the
soils do not liquefy. The investigation focuses upon how the responses vary with: (1) the size of
time step (); (2) the density of the soil mesh (). In this study Parameters obtained from direct
integration linear analysis ( = 0.02) and appropriate mesh size is used also for nonlinear
analysis.
402
Case-2
Fixed
(Sec)
1.038
(Sec)
-
SSI
(KK)
92.4
(KN.m)
1772.23
(Sec)
1.085
(Sec)
-
(KK)
91.41
(KN.m)
1753.20
(KK)
78.50
(KN.m)
1516.79
Case-4
Fixed
(Sec)
0.900
(Sec)
3.140
SSI
(KK)
81.84
(KN.m)
1581.48
(Sec)
0.941
(Sec)
3.140
403
Case-6
Fixed
(Sec)
0.98
(Sec)
3.140
SSI
(KK)
75.15
(KN.m)
1452.15
(Sec)
1.02
(Sec)
3.140
(KK)
71.80
(KN.m)
1387.57
The results of , , and observed a good agreement between 2DOF and FEM
models in case of fixed and SSI effects. Whilst in the case of SDOF for fixed and SSI observed a
significant variation compared with FE. The results above need to be examined with non linear
direct integration which is more realistic and accurate method.
Integration Dynamic
In these cases the linear and nonlinear results of , and , at base are obtained as flowing.
Table 5: Results of direct integration
Analysis
Linear
Case-7
Nonlinear
Case-8
(KN)
70.77
(KN.m)
1364.44
67.45
1315.70
To estimate the values of forces ( and ) created due to the seismic approach, the
deviations of the all results obtained before, case-1 to 6 and case-7 with case-8.
V
Vol.
16 [2011], Bund.. D
404
40
Deviation(%)
35
30
Oveturning
O
M
Moment
25
Shear Force
20
15
10
5
0
1
Cases
Figure 15
5: Percentagee of deviationns for shear foorce and overrturning mom
ment comparedd
to nonlineaar dynamic annalysis
A can be seen
As
n from figuree 15, the deviiations annd values calculated byy the nonlineaar
analysis carried ou
ut with FEM is gradually less than the ones calculatted accordingg to the otherrs.
In thee case of the deviatioons of SDOF in the case of fixed andd SSI effects are 37%, annd
35.52% respectiveely. Whilst in case of 2D
DOF (fixed +SSI) the deviations 21..33 and 16.338
similaarly in the caase of FEM (ssuperpositionn) models the deviations 6.5 and 5.92. Same scenariio
also with
w with minor deferennt
CON
NCLUSIO
ONS
Frrom this papeer it can be cooncluded that::Iddealizing the tank
t
based onn SDOF still inapplicable
i
a may remaain as overesttimate analysis
and
whichh is economiccally inapplicaable.
T simplified
The
d procedure thhat can be uttilized for evaaluating the dynamic
d
charracteristics annd
the dyynamic respo
onse of elevatted tank is moore adequate by using 2DOF, and econnomically maay
appliccable. Furthermore, analyysis with 2DO
OF proceduree needs less computationaal efforts thaan
FEM..
The results of FEM based on
T
o first mode (Impulsive mode)
m
are reveeals a very goood estimationn.
Whilsst in the casee of linear fuull dynamic analysis
a
the results
r
show excellent coonvergence buut
computationally more
m
expensive
A
ACKNOW
WLEDGE
EMENTS
S
T author wo
The
ould like to accknowledge my
m supervisorr Dr Siti Aminnah and co-suupervisors Proof
Othm
man Karim and
d Mr Anuar Kasa
K
for guidaance and assisstance.
405
REFERENCES
1. Hamburger, R. O. (1996) Implementing performance based seismic design in structural
engineering practice. Proceedings of 11th World Conference on Earthquake Engineering,
Acapuco, Mexico, Paper no. 2121, Elsevier Science Ltd
2. Livaoglua R., and Dogangu, A. (2006) Simplified seismic analysis procedures for
elevated tanks considering fluid-structure-soil interaction. Journal of Fluids and
Structures, 22: 421439.
3. Livaoglua R., and Dogangu, A. (2007) Effect of foundation embedment on seismic
behavior of elevated tanks considering fluidstructure-soil interaction. 27 855863
4. Halil Sezen, Ramazan Livaoglu, and Adem Dogangun (2008) Dynamic analysis and
seismic performance evaluation of above-ground liquid-containing tanks Engineering
Structures 30: 794803
5. Ramazan Livaoglu and Adem Dogangun (2007) Seismic behavior of cylindrical elevated
tanks with a frane supporting system on various subsoil, Indian Journal of Engineering
&Materials sciences 14, 133-145
6. Somnath, Dutta, Aparna Mandal, Sekhar Chandra Dutta (2004) Soilstructure interaction
in dynamic behaviour of elevated tanks with alternate frame staging configurations 277
825853
7. Azlan, Adnan,. Hendriyawan, and Masyhur, Irsyam (2005) Selection and Development
of Appropriate Attenuation Relationship for Peninsular Malaysia, Proceeding Malaysian
Science and Technology Congress, Mid Valley, Kuala Lumpur.
8. Gasparini, D. A., and Vanmarcke, E. H. (1976) Simulated earthquake motions compatible
with prescribed response spectra. Evaluation of Seismic Safety of Buildings Report No.
2, Department of Civil Engineering, Massachusetts Institute of Technology, USA
9. Housner, G.W. (1963) Dynamic behaviour of water tanks, Bulletin of the Seismological
Society of the America. 53:381387
10. Housner, G. W. (1955) Properties of strong ground motion earthquakes, Bulletin of
Seismological Society of America. 45:187-218.
11. Kuo, J.S.H. (1982) Fluid-structure interactions: Added mass computation for
incompressible fluid, UCB/EERC-82/09 Report, University of California, Berkeley,
12. LUSAS FEA 14.1, User Manual (2007) Version 14.1 United Kingdom 16, FEA. Ltd.
LUSAS Element Reference Manual, Version 14.1 United Kingdom.
13. Chatzigogos C. T., Pecker, A., and Salenon, J. (2007) Seismic bearing capacity of a
circular footing on a heterogeneous cohesive soil. Soils and Foundations, 4:783-787.
14. Wolf, J. P, Song, C. (1996) Finite-Element Modeling of Unbounded, Wiley, England
15. Wolf, J. P. (1985) Dynamic Soil-Structure Interaction, Prentice-Hall, Englewood
Cliffs,NJ.
16. Westergaard, H.M. (1931) Water pressures on dams during earthquakes, Proceedings of
the ASCE, 57: 1303.
17. http://www.lusas.com/products/options/dynamic.html
406
18. ACI 371R-98 (American Concrete Institute), (1995) Guide to the Analysis Design and
Construction of Concrete-pedestal Water Tower. ACI 371R
19. Eurocode-8 (2003) Design of structures for earthquake resistancePart 1.1: General
rulesSeismic action and general requirements for structures-Part 4: Silos, tanks and
pipelines. European Committee for Standardization, Final PT Draft
20. Veletsos, A. S., and Meek, J. W. (1974) Dynamic behaviour of building-foundation
systems. Engineering and Structural Dynamics, 3(2), 121-138
21. Jennings and Bielak (1973) P.C. Jennings and J. Bielak, Dynamics of building soil
interaction, Bull. Seismol. Soc. Am. 63:948
22. Wilson, E.L., and Khalvati, M. (1983) Finite elements for the dynamic analysis of fluidsolid systems. International Journal of Numerical Methods in Engineering 19, 1657
1668.
23. Donea, J., Guliani, S., and Halleux, J.P. (1982) An arbitrary LagrangianEulerian Finite
Element method for transient dynamic fluidstructure interaction. Computer Methods in
Applied Mechanics and Engineering 33, 689723.
2011 ejge