Civil Engineering Infrastructures Journal, 50(1): 135 – 149, June 2017

Print ISSN: 2322-2093; Online ISSN: 2423-6691

DOI: 10.7508/ceij.2017.01.008

Study on Fundamental Frequencies of Cylindrical Storage Tanks

Obtained from Codes and Finite Element Method

Yazdanian, M. * and Ghasemi, S.

Young Researchers and Elite Club, Ahvaz Branch, Islamic Azad University, Ahvaz, Iran.

Received: 18 Sep. 2016; Revised: 30 Dec. 2016; Accepted: 01 Jan. 2017

ABSTRACT: In the case of study on the seismic behavior of tanks, one of the most important
subjects is selection of convective and impulsive frequencies for storage tanks. These two
frequencies are defined by Housner. The major utility of these frequencies is for Rayleigh
damping which is a must for time history analysis. API 650, EUROCODE 8 have suggested
some analytical solutions for finding convective and impulsive frequencies and other
parameters. In addition, these frequencies can be obtained from modal analysis by finite
element software. In current study, these frequencies are obtained using Finite Element (FE)
software and performing modal analysis. In the next stage, these modes are compared with
analytical methods from standards. To compare tanks with different (H/R) ratios, all these
tanks are modeled with the same volume and weight. This study showed that there is a good
agreement between convective frequencies extracted from Finite Element method and those
from standards. Furthermore, an increase in liquid level led to a rise and dip in convective
and impulsive frequencies, respectively. In addition, results show that a 1.47 times increase
in the viscosity leads to a 17 percent decrease in the values of impulsive frequencies.

Keywords: Convective, Finite Element, Frequency, Impulsive, Tanks.

INTRODUCTION wall. Hence, the stability of the liquid storage

tanks under earthquake conditions must be
Liquid storage tanks are used widely in investigated cautiously (González, 2013).
petroleum industries, urban water storage Many damages of tanks have been observed
facilities, and nuclear water places. Many around the world. For example, During the
parameters are associated with the seismic Niigata earthquake severe tanks have been
behavior of liquid storage tanks and these damaged. This earthquake led to a
include many issues (Ghateh et al., 2010; widespread eruption in tanks. These tanks
Chen and Kianoush, 2005). Experience of burned for 15 days and it caused to break out
past earthquakes in different regions showed to other area (Korkmaz et al., 2011). A further
that these types of structures are vulnerable explanation is Kocaeli earthquake, along this
along the severe earthquakes. The earthquake that happened in northwest
performance of these structures is different Turkey many of tanks were destroyed and
from the other structures like bridges and most damages of tanks were observed at 19
buildings and these different is due to the km from the earthquake epicenter (Sezen et
effect of hydrodynamic pressure on the tank al., 2008). The first study on the dynamic
* Corresponding author E-mail: [email protected]

135
 Yazdanian, M. and Ghasemi, S.

behavior of tanks regarding liquid-structure flexibility was considered in the suggested

interaction was conducted by Housner method. Kianoush and Chen (2006)
(1963). He proposed a new method for rigid investigated the impact of horizontal and
tanks by using analytical technique. In vertical ground motion combination in
Housner method, hydrodynamic pressure is rectangular tanks in a 2D space. They
divided into convective and impulsive part. separated the impulsive and convective parts
The Housner’s method has been selected for of the tanks and they compared different
seismic analysis in most of the past codes and structural responses according to these
regulations and this method has been separated parts.
considered for basis of many standards. An Kianoush and Ghaemmaghami (2011)
expanded application of Housner’s method, investigated the effect of different frequency
in terms of an advanced design method, was contents of earthquakes on the dynamic
presented by Epstein (1976). However, response of two rectangular storage tanks
attention to interaction between the solid and using the FE model regarding soil-liquid-tank
fluid is not a matter just related to storage interaction. They focused on the both
tanks. Different subjects can be found convective and impulsive parts of the tanks
regarding this on the literature review and for this purpose they obtained structural
(Ardebili, 2013). responses of these two separate parts.
Many conducted research on storage tanks Ormeño et al. (2015) provided methods for
has been done with an emphasis on Finite seismic ground motion scaling in dynamic
Element method. Haroun and Tayel (1985) analysis of liquid storage tanks. They used
proposed a FE method for investigating the three different methods for doing several
seismic response of tanks using Housner analyses of earthquake responses. Ruiz et al.
model. Further research on the behavior of (2015) proposed an efficient computational
tanks under dynamic loads using FEM was method for seismic analysis of liquid storage
investigated by Ruiz and Gutiérrez (2015), tanks. The proposed model was based on the
Kazem and Mehrpouya (2012), Ghateh et al. theory of potential flow, in which the
(2015), Hosseinzadeh et al. (2013), Kianoush continuity equation is solved through the
and Ghaemmaghami (2011) and Moslemi Finite Element method. Yazdanian et al.
and Kianoush (2012). Livaoglu et al. (2011) (2016a) used FE method to evaluate the
investigated the dynamic behavior of the seismic behavior of two concrete rectangular
backfill on rectangular tanks through the FE tanks regarding fluid-structure interaction.
method and by considering both the liquid– They concluded that the structural responses
structure and soil–structure interaction effects obtained from time history analysis are more
as well as non-linear interaction effect of the accurate than those obtained from response
soil and contact nodes between the soil and spectrum analysis. Yazdanian et al. (2016b)
the wall. Ranjbar et al. (2013) used FE to used FE method to study on the dynamic
model an elevated water tank system. The behavior of cylindrical storage tanks. They
elevated tank was placed on a framed investigated different analysis to show how a
structure. They showed that the structural tank behaves under different analysis. They
responses are highly influenced by the concluded that the peak ground acceleration
earthquake characteristics. Chen and of an earthquake is highly important in the
Kianoush (2005) used a sequential method dynamic analyses of tanks. In addition, their
for estimating the hydrodynamic pressure method investigated different parameters in
along the walls of concrete rectangular tanks detail. Eskandari-Ghadi et al. (2013)
in a 2D space. Also, the parameter of wall proposed an analytical solution for 2D

136
 Civil Engineering Infrastructures Journal, 50(1): 135 – 149, June 2017

coupled in a cylinder. They used numerical axial modes of the tallest and medium height
inversion of Laplace transform to obtain the tanks. Maekawa et al. (2010) described the
stresses, displacements and temperature in results of vibration tests using a 1/10 reduced
terms of time. scale model of large-scale cylindrical water
There is also some researches on the storage tanks to clarify their dynamic
fundamental frequencies of tanks. However behavior under seismic excitation. Their
the majority of them just regarded vibration examination showed that the oval-type
analysis. Virella et al. (2006) studied on the vibration, which is a high-order vibration
fundamental modes of tank-liquid systems mode, occurring on the sidewall of the tanks
under horizontal motions. They found similar affects the distribution shape and magnitude
fundamental periods and mode shapes from of dynamic fluid pressure. Gnitko (2011)
the free vibration (modal) analyses and from used both Boundary Element and FE methods
the harmonic response analyses, using the to solve coupled problem. In their model
added mass formulation and the model with differential equations of transient problem
liquid acoustic finite elements. Nicolici and were solved numerically by Runge-Kutta
Bilegan (2013) used both computational fluid method of 4th and 5th order. Tedesco (1982)
dynamics analyses and the FE to obtain both presented an in-depth investigation regarding
convective and impulsive frequencies. They the vibrational characteristics and dynamic
concluded that convective frequency depends analysis of cylindrical tanks subject to a
on liquid height and other geometry horizontal component of earthquake ground
parameters, while impulsive frequency motion. They developed simple analytical
changes with the elastic properties of the tank expressions in the form of cubic polynomials
material. Jaiswal (2008) carried out an which accurately predicted frequencies and
experimental research on different scaled radial mode shapes corresponding to the
tank models, then, tanks excited by a shake fundamental mode of vibration.
table. To compare with numerical study, the The seismic behavior of the storage tanks
FE model of solid-fluid system was is highly dependent on their dynamic
developed. They observed a good agreement characteristics such as natural frequencies,
between the results obtained from FE and mode shapes and damping ratios. In
experimental method. Furthermore, it can be investigation of seismic behavior of tanks,
found some studies on the analytical one of the most crucial issues is how to find
approach to assess the response of storage the convective and impulsive frequencies in
tanks. Shahmardani et al. (2014) evaluated storage tanks (Bayraktar et al., 2010). For
the impacts of different fluid field models and finding convective and impulsive frequencies
the mass of equipment at the top of offshore and other parameters, API 650 (2008) and
structure which was simulated as lumped EURO 8 (2003) codes for steel tanks have
mass on the responses of offshore structures. suggested some analytical solutions.
They showed that different approximations Regarding concrete tanks, ACI (2006) is a
regarding the fluid field could largely change key code which present some formulation for
the value of maximum displacement these type of tanks accordingly.
evaluated by the models. Amiri and Sabbagh- Alternatively, these frequencies can obtain
Yazdi (2012) investigated the effect of roof through modal analysis using FE software. In
on dynamic characteristics of dome roof current study, these frequencies are extracted
tanks partially filled with liquid. They using FE software (2014) and performing
concluded that at low liquid levels, equal to modal analysis. Also these modes are
1.80 and 4.80 m, the tank roof does affect the

137
 Yazdanian, M. and Ghasemi, S.

compared with analytical methods from API {u} is acceleration vector; {f } is the applied
and EURO 8 standards for cylindrical tanks. load vector.
Considering the previous studies carried In this method, the step by step integration
out on this topic, it can be observed many is applied directly to achieve the solution for
studies on dynamic behavior of tanks but the original equations of motion of the
most of them have just considered the seismic system. For solving this equation, it is needed
behavior of the tanks and there is a limited to [C] . Damping matrix of the system can be
research on these significant frequencies. In resulted from sum of the stiffness and mass
addition, none of the previous studies has matrices of the system. In many researches,
investigated the impact of different similar classical damping system said as
parameters on vibration analysis of tanks. As Rayleigh damping is applied for both
a result, there is a shortcoming in the recent structural and fluid domains leading to a
references related to storage tanks. It is classical damping for the whole scheme.
worthwhile that in this paper a Alpha and Beta damping are applied to
comprehensive study has been carried out on indicate Rayleigh damping constants α and
these two frequencies, as well as the other β . The damping matrix within the fluid
different parameters. For this purpose, a
domain [C] contains two parts which are
parametric study has been conducted and
different parameters are investigated. To because of impulsive and convective
account for the effect of liquid density and components of the stored fluid. In fact
thickness, different liquid and thickness are without these, it is not possible to do an
taken into account. In addition, tanks with accurate time history analysis (Kianoush and
different liquid height are modeled to Ghaemmaghami, 2011). It should be noted
consider the impact of this parameter. that under free oscillations, the motion of free
Overall, in this paper, different parameters liquid surface decays due to damping forces
are vastly considered for cylindrical tanks created by viscous boundary layers. From this
with a focus on frequencies analysis of these point of view, evaluation of damping
types of structures. characteristic for a fluid–tank system needs
more considerations. However, due to lack of
MATERIALS AND METHODS sufficient data in this field, the classical
damping scheme is used in the Finite Element
Governing Equation of Dynamic Forces model. Considering impulsive and
Time history analysis can take all the convective parts of liquid domain, damping
nonlinear factors into the analysis, so it is matrix can be given as Eq. (2)
used to estimate the exact amount of (Ghaemmaghami, 2010).
structural response. The discretized
formulation for a multi degree of freedom
[C] = α[M ] + β[K ] (2)
system subjected to external dynamic forces
can be defined as Eq. (1) (Chopra, 2000). in which α and β are computed by Rayleigh
damping method. α is defined based on the
[M ]{u} + [C]{u } + [K ]{u} = {f } (1) natural frequency of the primary sloshing
mode and calculates for the damping because
in which [M ] is the mass matrix of the system; of sloshing on the free surface of the tank
[C] is the damping matrix of the system; [K ] is liquid. β is defined based on the primary
stiffness matrix of the system; {u} is frequency of the tank and simulates the
displacement vector; {u } is velocity vector; damping because of the impulsive part. They

138
 Civil Engineering Infrastructures Journal, 50(1): 135 – 149, June 2017

can be found according to the following As evident, these tanks have regular element
equation (Moslemi & Kianoush, 2012). shape. It should be mentioned that fluid
element at a boundary should not be attached
 ω − ωi  directly to the structural elements but should
α  2ω i ω j  j ξ 
β  = 2 2 −
1 1  i  (3) have separate coincident nodes that are
  ω j + ω i  ω  ξj
ω i    coupled only in the direction normal to the
 j

interface. The materials of the tank shell are

As offered by API 650, damping ratios of assumed isotropic and linearly elastic. This is
2% and 0.5% are specified for the impulsive also worthwhile to mention that in the bottom
and convective parts of steel tanks, of all the tanks, the liquid is coupled with the
respectively, while as offered by ACI 350.3- cylindrical body which makes the results
06 (2006), damping ratios of 5% and 0.5% are more and more accurate. In fact, this subject
specified for the impulsive and convective requires tremendous attention because of the
parts of concrete tanks. Two impulsive and fact that the cylindrical body of the tanks is
convective frequencies are achieved from the also fixed along its another area.
modal analysis in FE software according to The shell is modeled using shell element
maximum effective mass or they can be (Shell 63). This element has 6 degree of
obtained by analytical solutions. According freedoms at each node and it has three
to the above equations, one of the most translational and three rotational degrees of
important issues is obtaining of impulsive and freedoms. It has both the bending as well as
convective frequencies in time history the membrane capacity. It has four nodes and
analysis. This importance is due to the fact variable thickness can be input at the nodes,
that damping matrix is a crucial factor for giving a shell with smoothly varying
solution of multi degree of freedom systems thickness. The fluid inside the cylindrical
subjected to external dynamic forces. It tank is modeled using fluid element (Fluid
should be noted that this study only covers 80). This element is a three dimensional solid
anchored tanks, with an emphasis on the element having three translational degree of
behavior of the cylindrical tanks. As a result, freedom system at each node. This element
all the modeled tanks are ground-supported has the ability to model contained fluids
and the base of the tanks is assumed to be giving the result of hydrostatic pressure,
fixed. In Figure 1, convective and impulsive hydrodynamic pressure as well as fluid
masses are shown. Impulsive mass is in the structure interaction. The fluid elastic
lower height than convective mass. modulus of the element (in the input) should
be taken as the bulk modulus of the fluid.
Finite Element Method Modal analysis is done to obtain natural
In this paper, a FE software i.e. ANSYS frequencies of these structures.
(2014) as a general purpose computer code is Determination of these parameters in the first
utilized to perform modal analyses. Modal step can be very useful in interpreting the
analysis can find the various periods of a behavior of the tank. In addition, this analysis
structure according to naturally resonate by can be a starting point for other analysis, such
using the overall mass and stiffness. These as response spectrum or time history
periods of vibration can be used to assess analyses. Convective and impulsive modes
dynamic behavior of the structure. In Figure are integral modes which have the maximum
2 the FE model of one of the tanks is shown. effective mass and they should be taken into
account for the dynamic analysis.

139
 Yazdanian, M. and Ghasemi, S.

Fig. 1. Convective and impulsive mass

Fig. 2. FE modeling of tank

According to the maximum effective mass precise meshing was found to provide good
in modal analysis two modes of impulsive results for the purpose of the present study.
and convective are extracted and these modes Since many tanks are modeled in this study,
amount would be set in Eq. (3). In modal the details of tanks including size of meshes
analysis, Lanczos solver is used. In this type are not mentioned. However, the number of
of solver all the natural frequencies and mode elements for each tank is shown in Table 4. It
shapes are found within a specified range of must be pointed out that the studies of
frequencies. It should be noted that this Yazdanian et al. (2016b) and Kianoush and
system is the most effective and precise Ghaemmaghami (2011) showed that the
method to find the natural frequencies. Figure values of frequencies and other dynamic
3 shows both fundamental frequencies of tank responses do not depend extremely on the
no. 3. It can be seen from Figure 3(b) that the size and the numbers of elements after a
convective frequency is only related to the specified range of size and elements number.
movement of liquid, while in Figure 3(a) both However, to show the validity of presented
shell and liquid are moved with each other. FE method, a comparison between the results
Convergence studies were carried out and of Barton and Parker (1987) with the current
the FE meshes which have the regular and FE method is conducted. For this purpose,

140
 Civil Engineering Infrastructures Journal, 50(1): 135 – 149, June 2017

two fundamental frequencies of tank which obtained. The liquid considered was water,
were obtained by Barton and Parker (1987) for which, bulk modulus and mass density
are compared with the current FE method. It were taken as 2.18×109 N/m2 and 1000
is obvious that current FE method and Barton Kg/m3 respectively. For the glass, the elastic
and Parker results are in excellent agreement. modulus, mass density and Poisson’s ratio
This further verifies the validity of the were considered as 1.173×109 N/m2, 955 Kg/
proposed FE technique. In addition, further m3 and 0.35 respectively. It should be noted
validation can be found in the Yazdanian et that the thickness has no effect on the
al. (2016b) study. convective frequencies and it will be
approved in this study in the next sections.
Another significant validation of FE
model of cylindrical tanks that is considered API and EURO 8 Method
in this study is related to comparison of the API 650 is presented the following
proposed FE method with experimental equation for calculating the value of
method. For this purpose, the fundamental convective period.
sloshing (convective) frequency of two tanks
which are obtained previously using Tc = 1.8 K s D (4)
experimental method by Jaiswal et al. (2008) 0.578
Ks =
are compared with the proposed FE method. 3.68 H (5)
tanh( )
For this purpose, two tanks are modeled using D
the proposed FE method and then the
convective frequencies of the tanks are

(a) (b)
Fig. 3. Fundamental mode shapes of tanks; a) impulsive; b) convective

Table 1. Comparisons fundamental frequencies calculated with the present study and with Barton and Parker (1987)
Type of Study Impulsive Convective
Current Study 11.5 0.29
Barton and Parker Study 11.5 0.29

Table 2. Fundamental sloshing frequencies of liquid

Depth of Liquid (mm) Experimental FEM
50 2.07 2.05
100 2.3 2.27

141
 Yazdanian, M. and Ghasemi, S.

where Tc is the natural period of the RESULTS AND DISCUSION

convective mode (s), D is nominal tank
diameter (m) and H is maximum design Characteristics of the both selected tanks and
product level (m) and K s is sloshing factor. their containers in details are shown in Tables
According to EURO 8, the natural period of 3 and 4, respectively. For comparison and
convective response can be taken from investigation of the impact of height to radius
following equation which is suggested by (H/R) ratios, all these eight tanks are modeled
Malhotra et al. (2000): with the same volume and weight, while all
these tanks have different (H/R) ratios, with
Tc = Cc r (6) ranging from 0.3 to 3. These tanks are
selected for covering all the (H/R) ratios
which was presented by Malhotra et al.
in which Cc is a coefficient for obtaining
(2000). It is highly unlikely to find a real tank
convective periods and can be obtained from in a (H/R) ratio which is not covered in this
Figure 4 and r is radius of the tank. research. In addition, tank no. 1 is selected to
A simplified method was presented in the investigate the parameter of viscosity. For
API and EURO 8 to determine the impulsive this purpose, type of liquid is considered to be
period. This method was taken from Malhotra changed in three different liquids and other
et al. (2000): parameters are considered similar to each
other. Furthermore, in Table 4 the number of
1 Ci H ρ
Ti = ( )( )( ) both fluid and shell elements is shown.
2000 tu E (7)
In the first step, static analysis is carried
D out and in this stage the tanks were analyzed
based on their weight and the hydrostatic
in which, Ti is impulsive period, Ci is a pressure of internal fluid. This analysis can be
coefficient for determining of impulsive used in compound loading, and also it can be
period and can be obtained from Figure 4, tu used as a criterion for the evaluation of
is the wall thickness, ρ is density and E is constructed model. The generated hydrostatic
elasticity module. Furthermore, Malhotra et pressure causes annular tensile stresses along
al. (2000) suggested values of Ci and Cc as a the tank wall. The values of displacement and
hydrostatic pressure obtained from static
table which can be used instead of Figure 4.
analysis can be found in Table 5. It can be
observed that both structural responses are
increased as the ratio of H/R is increased.

Fig. 4. Impulsive and convective coefficients (Malhotra et al., 2000)

142
 Civil Engineering Infrastructures Journal, 50(1): 135 – 149, June 2017

Effect of Height to Radius (H/R) This was previously shown by Housner

In Table 6 other impulsive and convective (1963). According to Table 6 which was
parameters are shown. These parameters are suggested by Malhotra et al. (2000), the
height (hc and hi) and mass ratio (WC/WL and values of convective and impulsive mass and
Wi/WL) which convective and impulsive height can be obtained according to (H/R)
forces are exerted respectively. As it can be ratio. There can be found a height and mass
observed the height which convective force is for each ratio.
exerted is higher than the impulsive height.
Table 3. Material properties used in the analysis
Materials Density Poisson’s Young’s Modulus Bulk’s Modulus
Properties (kg/m3) Ratio (Gpa) (Gpa)
Steel 7850 0.3 210 -
Fluid (water) 1000 - - 2.07
Fluid (oil) 800 - - 1.62
Fluid (gasoline) 680 - - 1.03

Table 4. Characteristics of the selected tanks

Tank H R Thickness Number of Elements
H/R
No. (m) (m) (m) Fluid Shell
1 0.3 2.427 8.09 0.01 1728 624
2 0.5 3.414 6.828 0.01 1500 500
3 0.7 4.272 6.103 0.01 960 352
4 1 5.419 5.419 0.01 2100 580
5 1.5 7.101 4.734 0.01 1920 512
6 2 8.602 4.301 0.01 1920 512
7 2.5 9.98 3.993 0.01 2688 640
8 3 11.272 3.755 0.01 2880 672

Table 5. Output of static analysis

Tank No. H/R Hydrostatic Pressure (MPa) Displacement (m) × 10-3
1 0.3 0.021 0.70
2 0.5 0.030 0.76
3 0.7 0.038 0.78
4 1 0.049 0.82
5 1.5 0.066 0.85
6 2 0.079 0.86
7 2.5 0.094 0.88
8 3 0.100 0.89

Table 6. Comparison of impulsive and convective parameters

Convective Impulsive
Tank No. H/R Height Mass Height Mass
hc/HL hc (m) WC/WL WC (ton) hi/HL hi (m) Wi/WL Wi (ton)
1 0.3 0.521 1.264 0.824 412 0.4 0.971 0.176 88
2 0.5 0.543 1.854 0.7 350 0.4 1.366 0.3 150
3 0.7 0.571 2.439 0.586 293 0.401 1.71 0.414 207
4 1 0.616 3.338 0.452 226 0.419 2.27 0.548 274
5 1.5 0.69 4.899 0.314 157 0.439 3.117 0.686 343
6 2 0.751 6.46 0.237 118.5 0.448 3.854 0.763 381.5
7 2.5 0.794 7.924 0.19 95 0.452 4.51 0.81 405
8 3 0.825 9.299 0.158 79 0.453 5.106 0.842 421

143
 Yazdanian, M. and Ghasemi, S.

In this study, the values of these frequencies. A more detailed look reveals that
parameters are obtained using Malhotra table. this decrease was just below 2.4 times in the
It can be seen from the table that an increase case of impulsive frequencies. The reason
in the ratio of H/R, leads to an increase and why the values of impulsive frequencies are
decrease in the values of Wi and Wc, decreased is because of the fact that both
respectively. It must be pointed out that all the convective and impulsive frequencies of
tanks have the same volume and weight, tanks are dependent on each other. Both these
while the values of convective and impulsive frequencies differed with the liquid level and
weight are different. when H/R ratio is increased, the level of
It can be seen from the Figure 7 and Table liquid are increased in the tanks. Then, the
7 that an increase in the value of H/R leads to values of convective frequencies are
an increase in the values of convective increased and as a result the values of
frequencies in all three methods. The impulsive frequencies are decreased. In fact,
maximum values of convective frequencies they have opposite ratio with each other. This
are obtained for the tank with (H/R) = 3, while is because of the fact that both these
the highest values of impulsive frequencies is frequencies have a different effective mass
seen in the tank with (H/R) = 0.3. It must be and they are divided according to their
pointed out that 10 times increase in the H/R effective mass. It can be seen that convective
leads to well above double increase in frequencies are closer together in both FE
convective frequencies. It can be observed method and analytical methods than those
that an increase in the (H/R) ratio leads to a obtained for impulsive frequencies through
decrease in the values of impulsive these methods.

0.4 25
Convective Frequencies

Impulsive Frequencies

20
0.3
15
0.2 API
FEM 10 API & EURO8
0.1 EURO8 5 FEM
0 0
0 1 2 3 0 1 2 3
H/R H/R
(a) (b)
Fig. 5. Comparison of the effect of (H/R) on frequencies (Hz); a) convective; b) impulsive

Table 7. Comparing of impulsive and convective frequencies (effect of (H/R))

Tank Convective Frequencies (Hz) Impulsive Frequencies (Hz)
H/R
No. API EUROCODE8 FEM API EUROCODE8 FEM
1 0.3 0.169 0.168 0.159 22.621 22.621 20.841
2 0.5 0.222 0.22 0.213 20.987 20.987 19.76
3 0.7 0.255 0.253 0.247 19.814 19.814 18.567
4 1 0.285 0.283 0.279 18.062 18.062 17.059
5 1.5 0.311 0.310 0.307 15.478 15.478 14.691
6 2 0.327 0.326 0.321 13.081 13.081 12.444
7 2.5 0.340 0.338 0.335 11.077 11.077 10.619
8 3 0.351 0.349 0.345 9.437 9.437 9.062

144
 Civil Engineering Infrastructures Journal, 50(1): 135 – 149, June 2017

Furthermore, it is found that the highest values remained unchanged with changes of
values of convective frequencies are obtained density. It must be pointed out that the
under API method and the lowest are convective frequencies depend only on
obtained under FE method. However, the dimension of liquid and the other parameters
average difference among all these three such as weight do not play any role.
methods is nearly one percent. In addition, Conversely, impulsive frequencies are
impulsive frequencies obtained from API and function of parameters such as weight and
EURO 8 methods are higher than FE method. other conditions of tanks body. This is due to
the fact that impulsive frequencies are those
Effect of Viscosity which are related to both liquid and
hc and hi and mass ratio (WC/WL, Wi/WL ) cylindrical shell. In other words, liquid moves
which convective and impulsive forces are in unison with the tanks body in impulsive
exerted respectively and presented in Table 8. frequency, while convective frequency is one
As it can be observed the height exerted by that undergoes sloshing motion and does not
the convective force is more than the depend highly on tanks body and the overall
impulsive height. It can be seen from the table weight of tanks. The other key factor which
that a decrease in the amount of viscosity can be clearly seen in both Table 9 and Figure
leads to a corresponding decrease in the 8b is related to discrepancy in impulsive
values of Wi and Wc, while the amount of frequencies obtained from different methods.
viscosity has no effect on hi and hc responses. However, this is below 10 percent. The
Figure 8 and Table 9 present the values of integral factor which pose this discrepancy is
convective and impulsive frequencies. It can because of the fact that codes do not consider
be seen from the results that the highest bulk’s module while, in FE software this
values of convective frequencies are seen in factor is taken into account.
the API method. Also, the lowest values of
convective frequencies are obtained for FE Effect of Thickness
method. In addition, it can be clearly seen Since the thickness parameter has a great
from Figure 8 that viscosity has no effect on deal of importance in designing of cylindrical
the values of convective frequencies in all tanks, in this study a tank (no. 3) with radius
methods. The maximum values of impulsive of 6.103 and height of 4.272 is selected for
frequencies are obtained for API and EURO study on effect of thickness on frequencies.
8 method and the minimum values are seen in Shell thickness of tank no. 3 is assumed with
FE method. It can be concluded from Figure four different ratios (0.08, 0.01, 0.012, and
8 and Table 9 that the maximum values of 0.014). In Table 10, other impulsive and
impulsive frequencies are obtained in tanks convective parameters are shown. It can be
which are filled with gasoline and the seen from this table that the convective mass
minimum is obtained for a tank which is filled part of the liquid is higher than impulsive
with water. Therefore, the results show that a part.
decrease in the viscosity leads to a rise in the Table 11 and Figure 9 show that tanks with
values of impulsive frequencies. In other different thicknesses have the same
words, the values of impulsive frequencies convective frequencies and the values of
are decreased by increasing in the density thickness have no effect on convective
parameter, when the density increased by values, while an increase in the ratios of
1.47 times, the impulsive frequencies faced thicknesses lead to increase in the values of
17 percent reduction. It is more marked with impulsive frequencies. This is due to the fact
results for convective frequencies, since their that when the thickness of tanks increased the

145
 Yazdanian, M. and Ghasemi, S.

shell body of tanks which has to move with analytically method than impulsive
fluid part of tank shows more strength against frequencies. The frequency values of FE are
movement. As a result it is understandable less than API and EURO 8 codes in both
why the values of impulsive frequencies are convective and impulsive. The highest values
increased with the rise of thickness. The of convective frequencies are obtained for
highest values of impulsive frequencies are API code, while the values of impulsive
obtained for tank with thickness of 14 mm. It frequencies in both analytical methods (API
can be seen that convective frequencies are and EURO 8) are the same.
closer together in both FE method and
Table 8. Comparing of impulsive and convective parameters
Convective Impulsive
Liquid Type Height Mass Height Mass
hc/HL hc (m) WC/WL WC (ton) hi/HL hi (m) Wi/WL Wi (ton)
Water 0.521 1.264 0.824 412 0.4 0.971 0.176 88
Oil 0.521 1.264 0.824 349 0.4 0.971 0.176 61
Gasoline 0.521 1.264 0.824 300 0.4 0.971 0.176 52

Table 9. Comparing of impulsive and convective frequencies (effect of viscosity)

Convective Frequencies (Hz) Impulsive Frequencies (Hz)
Liquid Type
API EUROCODE8 FEM API EUROCODE8 FEM
Water 0.169 0.168 0.159 22.621 22.621 20.841
Oil 0.169 0.168 0.159 25.29 25.29 23.135
Gasoline 0.169 0.168 0.159 27.43 27.43 24.88

Table 10. Comparing of impulsive and convective parameters

Convective Impulsive
H/R Height Mass Height Mass
hc/HL hc (m) WC/WL WC (ton) hi/HL hi (m) Wi/WL Wi
0.7 2.439 0.586 0.586 291 0.401 1.71 0.414 205.6

0.17 30
ConvectiveFrequencies

Impulsive Frequencies

0.165 25

0.16 20
API & EURO8
API
0.155 FEM 15 FEM
EURO8
0.15 10
water oil gasoline water oil gasoline
(a) (b)
Fig. 6. Comparison of the effect of viscosity on frequencies (Hz), a) convective, b) impulsive

Table 11. Comparing of impulsive and convective frequencies

Convective Frequencies (Hz) Impulsive Frequencies (Hz)
Thickness (mm) H/R
API EUROCODE8 FEM API EUROCODE8 FEM
8 0.7 0.255 0.253 0.247 17.62 17.62 16.684
10 0.7 0.255 0.253 0.247 19.814 19.814 18.567
12 0.7 0.255 0.253 0.247 21.581 21.581 20.248
14 0.7 0.255 0.253 0.247 23.31 23.31 21.776

146
 Civil Engineering Infrastructures Journal, 50(1): 135 – 149, June 2017

0.26 25
Convective Frequencies

Impulsive Frequencies
20
0.25
15
API 10 API & EURO8
0.24
FEM 5 FEM
EURO8
0.23 0
8 10 12 14 8 10 12 14
Thickness (mm) Thickness (mm)
(a) (b)
Fig. 7. Effect of thickness on frequencies (Hz); a) convective; b) impulsive

CONCLUSIONS 3- This study showed that by increasing

the level of liquid, convective frequencies are
In this paper, two fundamental frequencies increased. Also impulsive frequencies are
are obtained using FE, API and EURO 8 decreased by increasing the level of liquid.
methods. To compare the tanks with different This was seen in FE, API and Euro8 methods.
(H/R) ratios, all these tanks are modeled with 4- This study showed that liquid viscosity
the same volume and weight. In addition, it has no effect on convective frequencies
should be noted that different (H/R) ratios because of all the convective responses
between 0.3 and 3 are chosen. Furthermore, remains unchanged. It was observed that a
to investigate the effect of liquid type, one of 1.47 times increase in the viscosity leads to a
the tanks is filled with three different liquids. 17 percent decrease in the value of impulsive
Four different thicknesses have been taken frequencies.
into account to investigate the effect of 5- Results of this study showed that the
thickness variation. Based on this study, the value of thickness has no effect on the values
following observations and conclusions can of convective frequencies. It was shown that
be drawn: an increase in the ratios of thicknesses leads
1- It is resulted that convective frequencies to increase in the value of impulsive
are so close together in FE, API and EURO 8 frequencies. To be more precise, an increase
methods. The discrepancy of convective of 1.5 times in values of thickness led to
frequencies between FEM and analytical values of impulsive frequencies rise by 30
codes is nearly one percent, while this percent.
discrepancy is more marked for impulsive
frequencies, with almost five percent average ACKNOWLEDGEMENTS
difference.
2- Moreover, the results showed that the The authors would like to express their
height which convective force is exerted is appreciation to the Young Researchers and
higher than the impulsive height. This was Elite Club, Ahvaz Branch, Islamic Azad
also shown by Housner (1963). It can be University for their financial support
resulted that an increase in liquid height leads provided to the authors and providing the
to an increase and decrease in the value of necessary facilities to conduct this research
impulsive and convective mass. Also, a project.
decrease in the amount of viscosity leads to a
decrease in the values of Wi and Wc. But the
amount of viscosity has no effect on hi and hc.

147
 Yazdanian, M. and Ghasemi, S.

REFERENCES Ghateh, R., Kianoush, M.R. and Pogorzelski, W.

(2015). “Seismic response factors of reinforced
ACI Committee 350.3-06. (2006). Seismic design of concrete pedestal in elevated water tanks”,
liquid-containing concrete structures (ACI 350.3- Engineering Structures, 87, 32-46.
06) and commentary (ACI 350.3R-06), Farmington Gnitko, V., Marchenko, U., Naumenko, V. and
Hills (MI, USA), American Concrete Institute. Strelnikova, E. (2011). “Forced vibrations of tanks
Amiri, M. and Sabbagh-Yazdi, S.R. (2012). “Influence partially filled with the liquid under seismic load”,
of roof on dynamic characteristics of dome roof In Proceedings of XXXIII Conference “Boundary
tanks partially filled with liquid”, Thin-Walled Elements and other mesh reduction methods”, WIT
Structures, 50(1), 56-67. Press, Transaction on Modeling and Simulation,
ANSYS, Inc. (2014). ANSYS release 15.0 pp. 285-296.
Documentation, USA, González, E., Almazán, J., Beltrán, J., Herrera, R. and
API Standard 650. (2008). Welded steel tanks for oil Sandoval, V. (2013). “Performance of stainless
storage, 11th Ed., American Petroleum Institute, steel winery tanks during the 02/27/2010 Maule
Washington, D.C. Earthquake”, Engineering Structures, 56, 1402-
Barton, D.C. and Parker, J.V. (1987). “Finite Element 1418.
analysis of the seismic response of anchored and Hosseinzadeh, N., Kazem, H., Ghahremannejad, M.,
unanchored liquid storage tanks”, Earthquake Ahmadi, E. and Kazem, N. (2013). “Comparison of
Engineering and Structural Dynamics, 15(3), 299- API650-2008 provisions with FEM analyses for
322. seismic assessment of existing steel oil storage
Bayraktar, A., Sevim, B., Altunışık, A. and Türker, T. tanks”, Journal of Loss Prevention in the Process
(2010). “Effect of the model updating on the Industries, 26(4), 666-675.
earthquake behavior of steel storage tanks”, Housner, G.W. (1963). “The dynamic behavior of
Journal of Constructional Steel Research, 66(3), water tanks”, Bulletin of the Seismological Society
462-469. of America, 53(2), 381-387.
Chen, J.Z. and Kianoush, M.R. (2005). “Seismic Haroun, M.A. and Tayel, M.A. (1985). “Response of
response of concrete rectangular tanks for liquid tanks to vertical seismic excitations”, Earthquake
containing structures”, Canadian Journal of Civil Engineering and Structural Dynamic, 13, 583-595.
Engineering, 32, 739-52. Hariri Ardebili, M.A., Mirzabozorg, H., Kolbadi, S.
Chopra, A.K. (2000). Dynamics of structures: Theory and Mahdi, S. (2013). “Are there any differences in
and applications to earthquake engineering, 2nd seismic performance evaluation criteria for
Ed., Prentice-Hall. concrete arch dams?”, Civil Engineering
Epstein, H.I. (1976). “Seismic design of liquid storage Infrastructures Journal, 46(2), 233-240.
tanks”, Journal of the Structural Division, ASCE, Jaiswal, O.R., Kulkarni, S. and Pathak, P. (2008). “A
102(9), 1659-1673. study on sloshing frequencies of fluid-tank
Eskandari-Ghadi, M., Rahimian, M., Mahmoodi, A., system”, In Proceedings of the 14th World
and Ardeshir-Behrestaghi, A. (2013). “Analytical Conference on Earthquake Engineering, pp. 12-17.
Solution for Two-Dimensional Coupled Kazem, H., and Mehrpouya, S. (2012). “Estimation of
Thermoelastodynamics in a Cylinder”, Civil sloshing wave height in broad cylindrical oil
Engineering Infrastructures Journal, 46(2), 107- storage tanks using numerical methods”, Journal of
123. Structural Engineering and Geo-Techniques, 2(1),
Eurocode 8. (2003). Design of structures for 55-59.
earthquake resistance, Part 1: General rules Kianoush, M.R. and Chen, J.Z. (2006). “Effect of
seismic action and general requirements for vertical acceleration on response of concrete
structures, Part 4: Silos, tanks and pipelines, rectangular liquid storage tanks”, Engineering
European Committee for Standardization. Structures, 28(5), 704-15.
Ghaemmaghami, A. (2010). “Dynamic time-history Kianoush, M.R. and Ghaemmaghami, A.R. (2011).
response of concrete rectangular liquid storage “The effect of earthquake frequency content on the
tanks”, Dissertation Abstracts International, seismic behavior of concrete rectangular liquid
72(08). tanks using the finite element method incorporating
Ghateh, R., Kianoush. M.R. and Pogorzelski, W. soil–structure interaction”, Engineering Structures,
(2015). “Seismic response factors of reinforced 33(7), 2186-2200.
concrete pedestal in elevated water tanks”, Korkmaz, K.A., Sari, A., and Carhoglu, A.I. (2011).
Engineering Structures, 87, 32-46. “Seismic risk assessment of storage tanks in
Turkish industrial facilities”, Journal of Loss

148
 Civil Engineering Infrastructures Journal, 50(1): 135 – 149, June 2017

Prevention in the Process Industries, 24(4), 314- in flexible tanks”, Nuclear Engineering and
320. Design, 258, 51-56.
Livaoglu, R., Cakir, T., Dogangun, A., and Aytekin, Yazdanian, M., Razavi, S.V. and Mashal, M. (2016a)
M. (2011). “Effects of backfill on seismic behavior “Seismic analysis of rectangular concrete tanks by
of rectangular tanks”, Ocean Engineering, 38(10), considering fluid and tank interaction”, Journal of
1161-1173. Solid Mechanics, 8(2), 435-445.
Maekawa, A., Shimizu, Y., Suzuki, M. and Fujita, K. Yazdanian, M., Razavi, S.V. and Mashal, M. (2016b).
(2010). “Vibration test of a 1/10 reduced scale “Study on the dynamic behavior of cylindrical steel
model of cylindrical water storage tank”, Journal liquid storage tanks using finite element method”,
of Pressure Vessel Technology, 132(5), 051801. Journal of Theoretical and Applied Vibration and
Malhotra, P.K., Wenk, T. and Wieland, M. (2000). Acoustics, 2(2), 144-165.
“Simple procedure for seismic analysis of liquid-
storage tanks”, Structural Engineering
International, 10(3), 197-201.
Moslemi, M. and Kianoush, M.R. (2012). “Parametric
study on dynamic behavior of cylindrical ground-
supported tanks”, Engineering Structures, 42, 214-
230.
Ormeño, M., Larkin, T. and Chouw, N. (2015).
“Evaluation of seismic ground motion scaling
procedures for linear time-history analysis of liquid
storage tanks”, Engineering Structures, 102, 266-
277.
Ranjbar, M.M., Bozorgmehrnia, S. and Madandoust,
R. (2013). “Seismic behavior evaluation of
concrete elevated water tanks”, Civil Engineering
Infrastructures Journal, 46(2), 175-188.
Ruiz, D.P. and Gutiérrez, S.G. (2015). “Finite Element
methodology for the evaluation of soil damping in
LNG tanks supported on homogeneous elastic half
space”, Bulletin of Earthquake Engineering, 13(3),
755-775.
Ruiz, R.O., Lopez-Garcia, D. and Taflanidis, A.A.
(2015). “An efficient computational procedure for
the dynamic analysis of liquid storage tanks”,
Engineering Structures, 85, 206-218.
Sezen. H., Livaoglu, R. and Dogangun, A. (2008).
“Dynamic analysis and seismic performance
evaluation of above-ground liquid containing
tanks”, Engineering Structures, 30, 794-803.
Shahmardani, M., Mirzapour, J. and Tariverdilo, S.
(2014). “Dynamic response of submerged vertical
cylinder with lumped mass under seismic
excitation”, International Journal of Engineering,
Transactions A: Basics, 27(10), 1547-1556.
Tedesco, J.W. (1982). “Vibrational characteristics and
seismic analysis of cylindrical liquid storage
tanks”, Dissertation Abstracts International Part
B: Science and Engineering [DISS. ABST. INT. PT.
B- SCI. & ENG.], 43(3).
Virella, J.C., Godoy, L.A. and Suárez, L.E. (2006).
“Fundamental modes of tank-liquid systems under
horizontal motions”, Engineering Structures,
28(10), 1450-1461.
Nicolici, S. and Bilegan, R.M. (2013). “Fluid structure
interaction modeling of liquid sloshing phenomena

149