See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/326128535

Seismic damage criteria for a steel liquid storage tank shell and its interaction
with demanded construction material

Article in Bulletin of the New Zealand Society for Earthquake Engineering · June 2018
DOI: 10.5459/bnzsee.51.2.70-84

CITATIONS READS

2 4,764

2 authors:

Morteza Moeini Mohammad Ali Goudarzi

International Institute of Earthquake Engineering and Seismology International Institute of Earthquake Engineering and Seismology
5 PUBLICATIONS 39 CITATIONS 44 PUBLICATIONS 579 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Morteza Moeini on 14 February 2020.

The user has requested enhancement of the downloaded file.

70
Bulletin of the New Zealand Society for Earthquake Engineering, Vol. 51, No. 2, June 2018

SEISMIC DAMAGE CRITERIA FOR A STEEL LIQUID

STORAGE TANK SHELL AND ITS INTERACTION WITH
DEMANDED CONSTRUCTION MATERIAL

Morteza Moeini1 and Mohammad Ali Goudarzi2

(Submitted July 2017; Reviewed October 2017; Accepted January 2018)

ABSTRACT
In this paper, the relation between the steel cylindrical tank geometry and the governing critical damage mode
of the tank shell is numerically determined for all practical ranges of liquid storage tanks (aspect ratio
H/D = 0.2 to 2). In addition, the interaction between the seismic intensity, soil type, acceptable seismic risk
and tank geometry along with the extra material demanded by the seismic loads is examined based on the
provisions of major codes. The importance of seismic factors on the economics of the design of a liquid tank
in zones with high seismic activity is comprehensively discussed. In this regard, an empirical relation to
estimate the steel volume required for specific seismic conditions and tank geometries is proposed based on
the results of analysis.
Keywords: Liquid storage tank, Failure mode, Shell buckling, Material consumption

INTRODUCTION the earthquake depend upon the rigidity of the foundation [4].
The effects of past earthquakes indicate that tank geometry in
Thin-walled cylindrical shells for liquid storage tanks having a
terms of the height-to-diameter ratio, (H⁄D), is a major factor
uniform or stepwise variable thickness are employed in many
in shell buckling. Broad tanks (H⁄D < 1) generally experienced
engineering fields. Vertical aboveground cylindrical steel
elephant-foot buckling (E), whereas slender tanks (H⁄D >1)
storage tanks are vital components of lifelines and industrial
demonstrated diamond-shaped (D) buckling [5, 6].
facilities and are usually constructed with a thin bottom plate, a
wall plate and a closed roof or open top. They are used to store Initial efforts to solve the shell buckling problem under axial
water, oil, fuel, petrochemical products and other fluids. compression led to the determination of linear bifurcation stress
(or classic elastic critical buckling stress σcl ) for a cylinder with
Dynamic loads caused by an earthquake can damage these
simply supported ends and uniform membrane (in-plan) pre-
storage tanks and cause dysfunction, environmental pollution
buckling stress distribution. The boundary conditions,
and fire. Damage observed in such tanks in the past earthquakes
eccentricities and non-uniformities in the applied load, types of
include overturning, shell buckling, roof damage, anchor bolt
support, geometric imperfections and residual stress prompted
and nozzle failure, sliding and uplifting [1]. In most previous
initial experiments that confirmed that real cylinders buckle at
earthquakes, shell buckling occurred in atmospheric vertical
much lower loads than that exhibited under classic buckling
steel tanks. The seismic design of thin cylindrical tanks to
stress [7].
prevent shell buckling is an important challenge for structural
engineers. Experimental studies by Niwa and Clough on the seismic
behaviour of cylindrical tall wine storage tanks on a shaking
The effect of the geometrical parameters and anchorage
table showed that D buckling occurred in tall tanks while E
condition of a tank are major factors affecting shell damage
buckling was more common in broad tanks. They concluded
which has been confirmed by observation after previous
that the critical buckling stress assumed in American standards
earthquakes. A review of the failure modes of steel tanks during
(API650 and AWWA-D100) for steel tank design are rather
earthquakes in Chile in 1985 (M = 7.8) and 2010 (M = 8.8)
conservative estimations of buckling strength [8]. This
showed that some tanks performed well during the 2010
conclusion was rejected by Jia and Ketter, but the need for
earthquake due to the use of mechanical anchors, but that most
revision of code provisions for E buckling was emphasized [9].
of the self-anchored tanks failed in the 1985 earthquake [2].
They concluded that E buckling may occur in anchored or
Damage to stainless steel cylindrical tanks used for wine
unanchored tanks and that anchorage did not prevent the shell
storage and fermentation caused by the South Napa
from buckling.
earthquake ( Mw = 6.0) included local buckling of tank walls,
anchorage failure and damage at the top of the tanks to the Rotter et al. carried out early systematic studies of elephant-foot
catwalk system. It was concluded that more research is required buckling and derived preliminary results [10-13] that led to the
to develop simplified nonlinear models for this type of tank development of a semi-empirical equation for estimation of E
design and to develop retrofitting methods [3]. buckling load. El-Bkaily and Peek presented an algorithm to
predict the location of the E buckling and the extent of bulging
Comparison of observed tank performance and that predicted
using the finite element method (FEM) [14]. The effects of ten
by published theoretical methods performed on two tanks with
seismic codes on different tanks was reviewed and compared
different aspect ratios after the 2008 Silakhor earthquake in Iran
by Jaiswal et al. [15]. Their study revealed significant
(ML = 6.1) produced results that were in good agreement. It was differences among the codes on the design seismic forces for
concluded that axial stresses on the tank shells uplifted during

1 Ph.D. Candidate, International Institute of Earthquake Engineering and Seismology (IIEES), No. 21, Arghavan St., North Dibajee, Farmanieh.
Tehran-IRAN P.O.Box: 19537-1445, Email: [email protected].
2 Corresponding Author, Associate Professor, International Institute of Earthquake Engineering and Seismology (IIEES), No. 21, Arghavan St., North
Dibajee, Farmanieh. Tehran-IRAN P.O.Box: 19537-14453, Email: [email protected].
 71

various types of tanks. Chen et al. compared resistance to E determined and agreed upon previously by researchers. The
buckling according to the standards of different countries for oil results of this agreement are reflected in the similar attitudes
tanks [16]. Virella et al numerically examined the critical adopted by prominent codes. However, there are important
horizontal peak ground acceleration which induced the unclear issues for those who want to use the outcomes of such
buckling of a set of anchored cylindrical tanks [17]. research as it relates to shell buckling. For example, the
interaction of various seismic factors with the tank geometry to
Elephant-foot buckling of thick cylindrical shells under allow optimization of the material used has not been addressed
transverse loading was studied numerically by Chonghou et al. in previous studies. In other words, the effects of seismic
[18]. Their research showed that moderately thick cylindrical parameters such as seismic zone, level of seismic intensity and
shells under transverse loading exhibited a D buckling mode soil-shell interaction on shell buckling have not been
while thick cylindrical shells exhibited an E buckling mode. systematically addressed by researchers.
This was significantly influenced by the radius to thickness
ratio and the material yield strength, rather than by the length to The present study discusses these issues on two levels. First, the
radius ratio and the axial force. Chen et al. examined the effect paper tries to provide a comprehensive view of the effects of
of openings on E buckling of large non-anchored welded steel tank geometry on various types of shell damage. In this regard,
oil tanks and utilized a numerical simulation method to analyse a range of tank dimensions and related parameters that cover all
the effect of size and position of openings on E buckling critical practical dimensions of steel liquid tanks have been considered.
loading [19]. A comprehensive parametric study also has been performed to
determine the geometrical boundaries of each shell damage
Yang et al. proposed an analytical formula for elastic-plastic mode. The resulting classifications can benefit researchers who
instability of large oil storage tanks with perfect shell walls work in this field to properly select their geometric domain.
[20]. They concluded that the interaction of high hydraulic and Designers also are provided with comprehensive insight about
axial compression causes changes in the material properties of the interaction of their preliminary geometries and possible
the tank wall which causes the buckling strength of the tank critical seismic damage modes. Second, the paper also focuses
wall to decrease rapidly. Elkashef et al. used nonlinear FEM to on the effects of seismic factors on the selection of the optimum
investigate the formation of E buckling in anchored tanks [21]. geometry for a tank. The interaction of cost-effective factors
They used the von Mises yield criterion to determine the such as the land required on which to construct a liquid tank and
location and most important parameters in the occurrence of E the material required with which to construct a tank to store a
buckling in vertical tanks. They concluded that the point at certain volume of liquid are quantitatively discussed. An
which the hoop stress reaches a maximum value was the most empirical relation is proposed herein based on a parametric
likely location for E buckling to occur in steel liquid tanks and study for preliminary estimation of the amount of steel material
estimated that this type of buckling typically occurs at 76.2 to required for various seismic conditions and tank geometries.
152.4 cm above the tank base.
Two recent review studies on shell buckling were carried out by EVALUATION OF SHELL BUCKLING
Zingoni [22] and Godoy [23]. Zingoni [22] studied the strength,
stability and vibration behaviour of liquid-containment shells One of the most important failure modes of cylindrical storage
along with the design of these facilities to withstand loading by tanks is shell buckling. Generally, shell buckling modes include
liquid and wind pressure, ground movement and thermal shear buckling and bending buckling. Bending buckling
effects. Godoy [23] studied the buckling problems of vertical, comprises diamond-shaped (D) and elephant-foot (E) buckling.
aboveground tanks under static or quasi-static loads caused by These buckling modes are related to the geometric parameters
uniform pressure, wind, settlement of the foundation and fire. of a tank such as the aspect ratio or height to diameter (H/D),
and the tank diameter to thickness (D/t). Shear buckling occurs
Because of the low radius-to-thickness ratio in the cylinders and for small H/D whereas the bending buckling occurs for large
because shell buckling is the major failure mode of steel H/D. Shear buckling is caused by shear force and brings about
cylindrical tanks, rehabilitation methods have also been many large diagonal wrinkles in the centre of a tank side wall.
considered in recent years. The application of fibre-reinforced This type of buckling has not been considered in this study. In
plastics (FRP) to strengthen the tank shells against E buckling D buckling, the cross-section in the buckling region bends
has been considered. An experimental study on the buckling inward. In E buckling, the buckling cross-section expands
behaviour of unstiffened and stiffened cylindrical shells under outward in a ring and the structural strength decreases [30].
quasi-axial loading showed an increase in buckling load as a
result of the plastic fibres [24]. The effects of adding FRP In addition to E and D buckling, secondary buckling (SB) also
composite material to the tank shell has been investigated using exists and is caused by external pressure and cavitation. Tanks
numerical and experimental methods. The results indicate the that have failed and lost their contents sometime display
usefulness of these fibres as retrofitting to prevent the substantial damage to the top half of their height, probably the
occurrence of E buckling and increasing the capacity of a steel result of the vacuum created by rapid loss of contents due to
cylindrical tank [25]. damage at the base (Figure 1). The current study focused mainly
on E and D buckling modes of shell damage as they have
In addition to analytical, numerical and experimental frequently been observed in steel liquid tanks during previous
procedures, probabilistic methods using fragility curves have earthquakes. E buckling has been explicitly considered in most
been recognized as useful alternatives to the deterministic code major codes, such as Eurocode 8 [31] and the New Zealand
prescription approach. Researchers believe that uncertainty guidelines [32, 33] and are implicitly accounted for in the API
related to structural performance versus excitation makes 650 formulation [34].
probabilistic seismic risk analysis one of the best methods for
measuring the seismic performance of a structural system. A Eurocode8 and the New Zealand guidelines suggest the formula
fragility curve reveals the probability of reaching or exceeding proposed by Rotter to compute the buckling capacity with
the limit state for a particular value of ground motion intensity. respect to elastic–plastic buckling [35]. These equations are:
Consequently, to build a fragility curve, a failure criterion (limit 𝑃𝑅 2 1 𝜌 + (𝐹𝑌 /50)
state) for the analysed structure is required. Numerous failure 𝐹𝑚𝑝 = 𝑓𝑐𝑙 [1 − ( ) ] (1 − 1.5
)[ ] (1)
criteria have been proposed in the literature. Increasing the 𝐹𝑌 𝑡 1.12 + 𝜌 𝜌+1
capacity against failure is the desired outcome of most design 𝜌 = 𝑅/(400𝑡) (2)
methods [26-29]. It is clear from the literature mentioned above 2
𝑡
𝑓𝑐𝑙 = 𝐸 𝑡/(𝑅 √3(1 − 𝜈 ))(𝑡/𝑅) ≅ 0.605𝐸 (3)
that the mechanism of shell buckling has been properly 𝑅
72

where R is the tank radius, t is the shell thickness, P is the total D

internal pressure, FY is the steel yielding stress, ν = 0.3 is the
Poisson's ratio, E is the Yang's modulus and fcl is the classic
elastic critical buckling stress produced by linear bifurcation
analysis.

th h H
W

Ph

(a)
Th Th

Figure 2: Structural shell design [37].

(b)
The American Petroleum Institute (API) has published
specifications for welded steel oil storage tanks. Seismic
loading was introduced to API 650 in the early 1980s with the
addition of Appendix E (Seismic Design of Storage Tanks).
Shell thickness in this code is calculated in a manner similar to
the aforementioned procedure called the ‘one-foot’ method,
which calculates the thicknesses required at design points 0.3 m
(1 ft) above the bottom of each shell course as:

(c) 𝑡𝑑 = 4.9𝐷(ℎ − 0.3)𝐺 / 𝑆𝑑 + 𝐶𝐴 (8)

Figure 1: (a) E buckling in 2010 Maule earthquake [5]); (b) 𝑡𝑡 = 4.9𝐷(ℎ − 0.3) / 𝑆𝑡 (9)
D buckling; (c) secondary buckling in 2012 Emilia
earthquake [6]. where 𝑡𝑑 and 𝑡𝑡 are the design shell thickness and hydrostatic
test shell thickness (mm), respectively, G is the design specific
INITIAL ESTIMATION OF STRUCTURAL gravity of the liquid (which should not be less than 1 because
DIMENSIONS of the water test), CA is the corrosion allowance (mm), Sd and
St are the allowable stress for the design condition and
A cylindrical shape with a diameter of approximately 3 to 126 hydrostatic test condition (MPa), respectively. The required
m is the most common form employed for a liquid storage tank shell thickness should be greater than the design shell thickness
[36]. The required capacity (Q) of a reservoir is determined by (including any corrosion allowance) or hydrostatic test shell
local requirements. For steel cylindrical tanks, the shell thickness, but the shell thickness should not be less than the
thickness is allowed to increase to 5 cm (2 in) and the maximum values presented in Table 1.
permissible height of the tank is determined based on the
allowable bearing capacity of the soil upon which the tank is Table 1: Minimum nominal shell thickness in API 650.
placed. Trial calculations are required to determine the best
dimensions for a tank. The following simple equations are Tank nominal Diameter Nominal Plate Thickness
generally used for initial estimations [37]: (m) (mm)
<15 5
𝐻𝑚𝑎𝑥 = 𝑃𝑆𝑜𝑖𝑙 /𝛾𝐿 (4) 15 to < 36 6
36 to 60 8
𝐷 = √(4 × 𝑄)/(𝜋 × 𝐻𝑚𝑎𝑥 ) (5) > 60 10

where Hmax is the maximum permissible height, PSoil is the SEISMIC REQUIREMENTS OF STORAGE TANKS
allowable soil pressure, γL is the specific weight of the FROM NZSEE 2009
contained liquid and D is the tank diameter. The shell plate is
made from one or more horizontal plates. The tank thickness is NZSEE published its recommendations for the seismic design
calculated at the bottom of each course as: of storage tanks (known as the ‘Red Book’) in 1986 [32]. This
guideline generally does not allow yielding of tank elements, or
𝑇ℎ = 𝛾𝐿 × ℎ × 𝐷 (6) a reduction of design forces for ductility; thus, it is somewhat
more conservative than API 650-E. The revised edition (called
𝑡ℎ = 𝛾𝐿 × ℎ × 𝐷 / 2𝑓𝑒 (7) the ‘Blue Book’) was presented in NZSEE 2009 [33]. This
document is used in the current study because it provides
where 𝑡ℎ is the shell thickness, 𝑇ℎ is the shell tension, ℎ is the reasonable procedures for determining seismic design actions
depth from the tank top, 𝑒 is joint efficiency and f is allowable in accordance with NZS 1170.5 [38]. In NZSEE 2009, more
unit stress. guidance is recommended to designers on the appropriate
 73

ductility factors depending on the tank configuration, Table 4: Site subsoil classes [38].
anchorage and critical failure mechanism. Prescriptive
Class Description Definition
guidance is given for combining the said modes and resolving
*UCS > 50 MPa & Vs30 > 1500 m/s & not
the actions into base shear and overturning moment. In current
A Strong Rock underlain by < 18MPa or Vs 600m/s
practice, a range of damping levels is used so that, for the materials.
convective sloshing mode, damping is assumed to be about 1 < UCS < 50 MPa & Vs30 > 360 m/s & not
0.5% and, for the impulsive mode, ξi =2% is common. The underlain by < 0.8 MPa or Vs 300 m/s
B Rock
beneficial effects of additional radiation damping from soil- materials, a surface layer no more than 3 m
structure interaction can also provide significantly higher depth (HW-CW rock/soil).
damping, This is calculated based on the tank properties and not class A, B or E, low amplitude natural
C Shallow Soil
subsoil shear wave velocity from the figures in NZSEE 2009. period ≤ 0.6s
not class A, B or E, low amplitude natural
Ductility factor 𝜇 is applied for reduction of seismic forces and Deep period > 0.6s
is calculated based on tank type and anchorage condition for the D or underlain by < 10 m soils with undrained
horizontal impulsive mode from the suggested table in NZSEE Soft Soil shear strength < 12.5 KPa, or < 10 m soils
2009. The value is 1.0 for the convective and vertical modes. SPT N < 6
> 10m soils with undrained shear strength <
Correction factor k f (μ .ξi ) has been provided for the NZS
12.5 KPa, or > 10m soils with SPT N < 6, or
1170.5 elastic site hazard spectrum (Figure 3) to account for the Very Soft
E > 10m soils with Vs ≤ 150m/s, or
ductility and level of damping included in the shell design Soil
> 10m combined depth of previous
(Table 2). Return period factor R u proposed by this code properties.
reflects a range of life-safety, property and environmental *UCS= Unconfined compressive strength
exposure values, plus the community significance derived
based on the consequences of failure. Risk is classified as Seismic Actions Based on NZSEE 2009
negligible, slight, moderate, serious or extreme based on the
importance level, life safety and hazard level of the tank In the New Zealand recommendations, the dynamic response of
contents, environmental exposure, national or community liquid-tank systems is analysed using the Veletsos model
significance and adjacent property value at direct risk from tank (1984) for rigid tanks and Haroun-Housner model (1982) for
failure (2007 cost index). The corresponding values for the flexible system. Based on this code, the horizontal seismic shear
return period based on the consequences of failure are presented acting at the base of a tank and the overturning moment can be
in Table 3. calculated using the following expressions:

Site Subsoil Class 𝑉𝑖 = 𝐶𝑑 (𝑇𝑖 )𝑚𝑖 𝑔 (10)

The site subsoil class according to NZS 1170.5 shall be 𝐶𝑑 (𝑇𝑖 ) = 𝐶(𝑇𝑖 )𝑘𝑓 (𝜇 . 𝜉𝑖 )𝑆𝑃 (11)
determined as being one of classes A to E according to the
𝐶(𝑇𝑖 ) = 𝐶ℎ (𝑇𝑖 )𝑍𝑅𝑢 𝑁(𝑇𝑖 , 𝐷) (12)
properties listed in Table 4.
𝑉𝑓 = 𝐶𝑑 (𝑇̌𝑓 )[𝑚𝑓 + 𝑚𝑤 + 𝑚𝑡 ]𝑔 (13)
𝑉1 = 𝐶𝑑 (𝑇1 )𝑚1 𝑔 (14)
𝑀𝑟 = 𝐶𝑑 (𝑇̌0 )[𝑚0 − 𝑚𝑓 ]𝑔ℎ0 (15)
𝑀𝑓 = 𝐶𝑑 (𝑇̌𝑓 )[𝑚𝑓 ℎ𝑓 + 𝑚𝑤 ℎ𝑤 + 𝑚𝑡 ℎ𝑡 ]𝑔 (16)
𝑀1 = 𝐶𝑑 (𝑇1 )𝑚1 𝑔ℎ1 (17)

where mi is the equivalent mass of a tank and its contents

responding to the particular mode of vibration, m0 is the rigid
tank impulsive mass (total impulsive mass for the flexible
tank), mr = m0 -mf is the rigid impulsive mass, mf is the flexible
impulsive mass, m1 is the first convective mass, mw and mt are
the wall and roof mass, respectively, T1 is the period of
Figure 3: Spectral shape factor [38]. vibration of the first convective mode of vibration and Ť 0 is the
period of the impulsive mode of vibration for the tank-
Table 2: Correction factor k f (μ .ξi) to NZS 1170 elastic site foundation system and for a rigid foundation (Ť 0 =0). Ť f is the
hazard spectrum for damping and ductility [38]. period of vibration of the first horizontal tank-liquid impulsive
mode, including the effects of foundation flexibility. For a rigid
k f (μ .ξi )
Ductility
ξi
tank Ť f = 0 and mf = 0. Ti is the period of vibration of the
𝝁 appropriate mode of response. Cd (Ti ) is the horizontal design
0.5% 1% 2% 5% 10% 15% 20% 30%
1.00 1.67 1.53 1.32 1.00 0.76 0.64 0.56 0.47 action coefficient for mode i, C (Ti ) is the ordinate of the elastic
1.25 1.08 1.04 0.96 0.82 0.67 0.58 0.52 0.44 site hazard spectrum for horizontal loading for the site subsoil
2.00 0.91 0.89 0.84 0.74 0.63 0.55 0.50 0.43 type and the relevant mode obtained from NZS 1170.5
according to tank importance level. Ch (Ti ) is the spectral shape
Table 3: Return period factors [38]. factor for the subsoil type site and relevant mode (Figure 3), Z
is the seismic zone hazard factor, N(Ti D) is the near fault
Consequence of Failure Return Period Factor (R u ) factor, μ is the displacement ductility factor for horizontal
impulsive modes and is taken as 1.0 for the convective and
Negligible 0.5 vertical modes and 1.25 for the impulsive mode.
Slight 0.5
Moderate 1.0
Serious 1.3
Extreme 1.8
74

Shell Buckling Criteria in NZSEE 2009 4. The density of the contained liquid is ρ = 1000 (kg/m3 ) and
the steel type A36M for which FY = 250 MPa has been
In NZSEE 2009, the allowable shell buckling stress is
considered. Based on these variables, a total of 4224 tanks have
calculated from the following equations. Stresses due to
been analysed in numerical analysis to cover all practical
hydrodynamic pressure or overturning moment plus the shell
dimensions. The assumed nominal roof plate thickness and
weight (fm ) shall not exceed the stress required to induce
slope are 𝑡𝑟 = 5 mm and 𝜃𝑟 = 50 , respectively. The effect of
buckling under membrane compression (Fm ) according to the tank wall flexibility has been considered in addition to the
following equations: impulsive period correction for the soil-structure interaction
𝑓𝑚 = (𝑊𝑠 / 𝐴) + (𝑀𝑂𝑇 / 𝜋𝑅2 𝑡) ≤ 𝐹𝑚 (18) (foundation flexibility) according to NZSEE 2009
recommendations.
Allowable E buckling (𝐹𝑚𝑝 ) can be calculated using Eq. (1) as
NUMERICAL ANALYSIS
suggested by Rotter and the allowable D buckling (𝐹𝑚𝑒 ) can be
calculated as: To evaluate the effects of various parameters on the critical
buckling mode of the considered tanks, a comprehensive
𝐹𝑚𝑒 = [0.19 + 0.81(𝑓𝑝 /𝑓𝑐𝑙 )] 𝑓𝑐𝑙 (19) program is provided based on the provisions of NZSEE 2009
for the design process. The cases introduced in the previous
𝑓𝑝 = 𝑓𝑐𝑙 √((1 − (1 − 𝑝̅ /5)2 (1 − (𝑓0 / 𝑓𝑐𝑙 )2 )) ≤ 𝑓𝑐𝑙 (20) section are analysed using the developed program according to
the algorithm presented in Figure 5. As the two main outcomes,
𝑓𝑜 = 𝑓𝑦 (1 − 𝜆2 / 4) → { 𝑤ℎ𝑒𝑟𝑒 (𝜆2 = 𝑓𝑦 / 𝜎̅𝑓𝑐𝑙 ) ≤ 2 } (21) the required shell thickness and the demanded steel volume to
𝑓𝑜 = 𝜎̅ 𝑓𝑐𝑙 → { 𝑤ℎ𝑒𝑟𝑒 (𝜆2 = 𝑓𝑦 / 𝜎̅𝑓𝑐𝑙 ) ≥ 2 } (22) build a tank are extracted from the results of these analyses.
2
𝜎̅ = 1 − 𝛹(𝛿 / 𝑡) [(1 + 𝛹 (𝛿 / 𝑡))1/2 − 1] , (𝛹 = 1.24) (23) Main Group 1 H/D=0.2
(H=2.4m)
𝛿/ 𝑡 = (0.06/𝑎)√𝑅/𝑡 (24) H/D=0.3
Main Group 2
H/D=0.4
𝑃̅ = 𝑃 𝑅 / 𝑡 𝑓𝑐𝑙 (25) (H=4.8m)
Z=0.4 H/D=0.5
Main Group 3 Extreme
Soil (A&B)
where the value of the parameter “a” for normal, quality and (H=7.2m) H/D=0.6
Z=0.3
very high quality construction is 1.00, 1.5 and 2.5, respectively. Main Group 4 Soil (C) Serious H/D=0.8
(H=9.6m) H/D=1.0
Simultaneous shaking in both a single horizontal direction and Z=0.2
Main Group 5 Soil (D) Moderate H/D=1.2
the vertical direction shall be considered by combining of the (H=12m) Z=0.1
components by the square root of the sum of the squares Slight H/D=1.5
Main Group 6
(SRSS). These equations for calculation of the stress levels (H=14.4m) H/D=1.8
should be corrected using the effect of internal hydrostatic (𝑃ℎ ), H/D=2.0
Main Group 7
impulsive (𝑃𝑖 ), convective (𝑃1 ) and vertical pressures (𝑃𝑉 ). For (H=16.8m)
elastic D buckling, the limiting condition will be when the
Main Group 8
internal pressure is at a minimum. This occurs when the vertical (H=19.2m)
acceleration is at a maximum in the upward direction, thus
reducing the hydrostatic pressure. All pressure should be Figure 4: Categories and parameters of all tanks.
calculated at the base of the tank using selected ductility facto 𝜇
as: GOVERNING BUCKLING DOMAIN BASED ON A
TANK GEOMETRY
𝑃 = 𝑃𝑚𝑖𝑛 = 𝑃ℎ + √(𝑃𝑖 )2 + (𝑃1 )2 − 𝑃𝑉 (26)
Essentially, three types of failure mode may occur in the shell
For E buckling, the limiting condition will be when the internal of a liquid tank: (1) Material yielding due to hydrodynamic
pressure is at a maximum as: hoop stress (H); (2) Elastic D buckling and; (3) Elasto-plastic E
buckling. The shell thickness should be thick enough to prevent
𝑃 = 𝑃𝑚𝑎𝑥 = 𝑃ℎ + √(𝑃𝑖 )2 + (𝑃1 )2 + (𝑃𝑉 )2 (27) these failure modes. For a specific tank, one of these buckling
modes may be critical depending upon tank dimensions and the
Determining material yielding caused by tensile hoop stresses governing seismic conditions. For researchers who study shell
is performed using the following equation and should be less buckling, it is important to select the proper range of tank
than material yield stress (Fy ): dimensions for their parametric studies. Likewise, tank
designers must choose an appropriate primary tank geometry
𝑓𝐻𝑜𝑜𝑝 = 𝑓ℎ + √(𝑓𝑖 )2 + (𝑓1 )2 + (𝑓𝑉 )2 ≤ 𝐹𝑦 (28) based on the governing seismic conditions. The goal of the
analyses presented here is to provide better insight about the
relation of tank geometry to dominate type of buckling.
PARAMETRIC STUDY OF SHELL BUCKLING
PHENOMENON Figures 6 to 8 show the results of analysis for selected tanks
having various risk factors (risk factor = 𝑍𝑅𝑢 ). In these
To provide a comprehensive assessment of the effects of tank figures, the distribution of buckling stress in the vertical
geometry on the various shell damage types, a range of tank direction of the tank shell are illustrated versus the required
dimensions and parameters has been considered here to include thickness of the tank shell. It can be seen that E buckling is
all practical ranges of liquid tanks. All selected tanks are generally the governing failure mode near the base level for all
assumed to be anchored and categorized according to the cases considered, as has been observed in field observations
effective parameters for shell buckling. These parameters during past earthquakes. Although the buckling stress is
include the height of liquid content (H), aspect ratio of a tank different for the different vertical levels of a tank shell, a critical
(H/D), soil type, zone seismicity and risk classification. Eight type of buckling mode can be found for each case.
values for tank height H =2.4 to 19.2 have been chosen and, for
each specific tank height, the tank aspect ratio ranges Note that although the soil type increases the required steel shell
from H⁄D = 0.2 to 2. Four seismic intensity levels as well as volume, it does not considerably affect the failure mode; thus,
three soil categories have been considered as shown in Figure the figures are presented only for tanks placed on soil D. This
critical failure mode has been extracted and presented in Table
 75

5 for all considered cases where H, E and D represent the Table 5, the failure modes in low-height tanks (H < 8 m) are
required shell thickness for critical failure modes of material limited to a high H/D value and high risk factors. In the other
yielding due to hoop stress, elephant-foot buckling and ranges, the structural thickness is usually adequate and there is
diamond shape buckling, respectively. Based on the results in no failure mode.

* The investigated tanks in current study are assumed to be anchored with non-ductile holding down bolts and according to the table
3.1 in NZSEE2009, ductility factor in theses tanks can be considered as 1.25. Therefore, this value was chosen for ductility factor in
this study. Steel volume predicted here may be reduced if ductile behaviour is expected.

Figure 5: Algorithm developed to evaluate the critical buckling mode.

76

Risk Factor = 0.70 Risk Factor = 0.54 Risk Factor = 0.40 Risk Factor = 0.20

H/D = 0.40
H/D = 1.00
H/D = 1.50
H/D = 2.00

*Required shell thickness for each failure mode: elephant-foot (E), diamond-shaped (D), hoop (H) and structural (STR)

Figure 6: Required shell thickness for each failure mode (H=19.2 m, Soil D).

Risk Factor = 0.70 Risk Factor = 0.54 Risk Factor = 0.40 Risk Factor = 0.20
H/D = 0.40
H/D = 1.00
H/D = 1.50
H/D = 2.00

*Required shell thickness for each failure modes: elephant-foot (E), diamond-shaped (D), hoop (H) and structural (STR)

Figure 7: Required shell thickness for each failure mode (H = 12 m, Soil D).
 77

Risk Factor = 0.70 Risk Factor = 0.54 Risk Factor = 0.40 Risk Factor = 0.20

H/D = 0.40
H/D = 1.00
H/D = 1.50
H/D = 2.00

*Required shell thickness for each failure modes: elephant-foot (E), diamond-shaped (D), hoop (H) and structural (STR)

Figure 8: Required shell thickness for each failure mode (H = 7.2 m, Soil D).

Table 5: Prominent failure modes in tank shell for all tanks.

Risk Factor
0.70 0.54 0.4 0.3 0.2
Height (m) Height (m) Height (m) Height (m) Height (m)
19.2
16.8
14.4
12.0

19.2
16.8
14.4
12.0

19.2
16.8
14.4
12.0

19.2
16.8
14.4
12.0

19.2
16.8
14.4
12.0
H/D

9.6
7.2
4.8

9.6
7.2
4.8

9.6
7.2
4.8

9.6
7.2
4.8

9.6
7.2
4.8
0.2 H H H H H H H H H H H H H - H H H H H - - - - - - - - - - - - - - - -
0.3 H H H H H H E H H H H H H - H H H H H - - - - - - - - - - - - - - - -
0.4 H H H H E E E H H H H E E - H H H H E - - - - - - - - - - - - - - - -
0.5 E E E E E E E E E E E E E - H E E E E - - - - - - - - - - - - - - - -
Soil A & B

0.6 E E E E E E E E E E E E E - E E E E E - - - - - - - - - - - - - - - -
0.8 E E E E E E E E E E E E E - E E E E E - - E E E E - - - - - - - - - -
1.0 E E E E E D D E E E E E E - E E E E E E - E E E E E - - - - - - - - -
1.2 E E E E D D D E E E E E D D E E E E E E - E E E E E - - - - - - - - -
1.5 E E D D D D D E E E D D D D E E E E D D - E E E E E D - E E E - - - -
1.8 E E D D D D D E E D D D D D E E D D D D D E E E D D D - E E E E - - -
2.0 E E D D D D D E E D D D D D E E D D D D D E E E D D D - E E E E - - -
0.2 H H H H H H H H H H H H H H H H H H H H - H - - - - - - - - - - - - -
0.3 H H H H H H E H H H H H H E H H H H H H - H - - - - - - - - - - - - -
0.4 H H H H E E E H H H H E E E H H H H E E - H - - - - - - - - - - - - -
0.5 E E E E E E E E E E E E E E E E E E E E - E E - - - - - - - - - - - -
0.6 E E E E E E E E E E E E E E E E E E E E - E E - - - - - - - - - - - -
Soil C

0.8 E E E E E E E E E E E E E E E E E E E E - E E E E - - - - - - - - - -
1.0 E E E E E D D E E E E E E D E E E E E E - E E E E E - - E E - - - - -
1.2 E E E E D D D E E E E D D D E E E E E D - E E E E E - - E E - - - - -
1.5 E E E D D D D E E E D D D D E E E D D D D E E E E D D - E E E E E - -
1.8 E E D D D D D E E D D D D D E E D D D D D E E E D D D - E E E E E - -
2.0 E E D D D D D E E D D D D D E E D D D D D E E D D D D - E E E E D - -
0.2 H H H H H H H H H H H H H H H H H H H H - H - - - - - - - - - - - - -
0.3 H H H H H H E H H H H H H E H H H H H H - H - - - - - - - - - - - - -
0.4 H H H H E E E H H H H E E E H H H H E E - H - - - - - - - - - - - - -
0.5 E E E E E E E E E E E E E E E E E E E E - E E E E - - - - - - - - - -
0.6 E E E E E E E E E E E E E E E E E E E E - E E E E - - - - - - - - - -
Soil D

0.8 E E E E E E E E E E E E E E E E E E E E - E E E E - - - - - - - - - -
1.0 E E E E E D D E E E E E D D E E E E E E - E E E E E - - E E - - - - -
1.2 E E E E E D D E E E E E D D E E E E E D - E E E E E - - E E - - - - -
1.5 E E E D D D D E E E D D D D E E E D D D D E E E E D D - E E E E E - -
1.8 E E E D D D D E E D D D D D E E D D D D D E E E D D D - E E E E E - -
2.0 E E E D D D D E E D D D D D E E D D D D D E E D D D D - E E E E D - -
(-) : Structural thickness provides satisfactory seismic safety (H): Hydrodynamic Hoop Stress
(E) : Elephant-Foot Buckling (D): Diamond Shape Buckling
78

Considering the dominant failure mode for each case, it appears structural steel is sufficient to resist seismic loads. Even for
that the critical buckling mode mainly depends on the tank slender tanks under low seismic intensity, the extra required
height (H) and aspect ratio (H/D). It can also be observed that seismic steel volume is not more than 15% for the high level of
E buckling is rarely the dominant mode in broad tanks accepted seismic risk. For a slight level of accepted seismic risk,
(H/D ≤ 0.4) and, thus, this type of failure mode can be ignored the extra required steel volume is not considerable (less than
for them. In fact, for broad tanks, the dominant failure mode of 10%) in high seismic intensity zones. However, for moderate,
a tank shell is essentially caused by the dynamic H stress. serious and extreme levels of accepted seismic risk, the extra
Practically speaking, E buckling can be expected in medium or steel volume demanded by seismic loads can reach 15%, 25%
slender tall tanks. Elastic D buckling was observed to be critical and 45% for broad tanks, respectively. For slender tanks, these
in slender tanks (H/D ≥ 1.2) of medium height (H ≤14). This values are even higher, at up to 40%, 60% and 100%,
type of buckling is not likely to happen for broad or medium respectively. It is clear that in places with high seismicity, using
tanks (H/D≤1.2) as well as slender tanks with large tank height a smaller height and lower aspect ratio can significantly reduce
(H≥14). the required material and related construction cost. Therefore, a
proper economic balance should be struck between the cost of
Field observations of past earthquakes have demonstrated a a higher tank diameter (more occupied land) and lower amount
similar trend in which E buckling is critical in medium tanks of material required to store a certain volume of liquid.
(H/D = 1) and D buckling in slender tanks (H/D = 2) [5, 6]. Past
observations are in agreement with the general trend of the When considering soil type effects, it should be mentioned that,
current study results. Note that, although looser soil reduces the regardless of tank geometry, the demanded steel volume for
critical buckling stress, the type of critical buckling mode does tanks built on soft soil (type D) can be up to 40% more than for
not change considerably in response to a change in soil type. those placed on rocky soil (types A and B). This value is smaller
Therefore, the effects of soil type on the dominant failure mode for lower levels of accepted seismic damage. Considering this
can be ignored. major difference, the use of different methods of enhancing soil
strength or employing different types of tank foundation is an
PARAMETRIC STUDY ON DEMANDED STEEL FOR economically reasonable alternative for tanks placed on soft
TANKS soil. In Figures 9 and 10, the extra steel volume demanded by
the seismic loads are shown for extreme (EX Steel Diff%),
Parameters such as welding, transportation and site conditions serious (SE Steel Diff%), moderate (MO Steel Diff%) and
can affect the total fabrication cost of a tank. The demanded slight (SL Steel Diff%) levels of seismic risk. As an example,
cost of each factor mainly depends on the regional conditions the list of input parameters used for specific tank geometries
where the tank will be placed. Most of these factors can be and seismic conditions is presented in Table 6.
estimated based on the material required for construction of the
tank. In the current study, the total quantity of demanded steel Table 6: Input parameters for a typical example tank.
is considered as the main component of the total cost. This
assumption may not always be the case, but does cover most Tank Geometric Parameters
practical situations.
Diameter 𝐷 = 15 𝑚
To acquire initial insight into the interaction of seismic effects
and the cost of a liquid tank, the volume of required material Height of Liquid 𝐻 = 12 𝑚
should be considered. The required shell thickness of the tanks Height to Diameter ratio 𝐻/𝐷 = 0.8
at various vertical levels were extracted in the previous section Shell initial thick. (mm) “Assumed”
and can be used to calculate the total volume of demanded steel Minimum shell thickness 6 𝑚𝑚
for each tank. Five important parameters should be considered
Bottom plate thick. (mm) 8 𝑚𝑚
to determine the extra material necessary to counter the effect
of seismic activity: (1) height of the tank; (2) tank aspect ratio; Roof plate thick. (mm) 6 𝑚𝑚
(3) soil type; (4) seismic intensity; and (5) level of accepted Roof slope 5%
seismic risk. Corrosion allowance (mm) 1 𝑚𝑚
Note that the steel volume required to resist hydrostatic loads Fluid Property
(structural steel) may be sufficient to resist seismic loads as Specific gravity of fluid Water (𝑆𝐺 = 1.0)
well. However, the presence of seismic loads demands more Fluid density 𝜌𝑤 = 1000 𝑘𝑔/𝑚3
steel than the required structural steel for most loads (seismic
steel) and can be determined by subtracting the structural steel Material
from total required steel. Knowing the seismic steel volume for A36 𝐹𝑌 = 250 𝑀𝑃𝑎
a comprehensive set of practical circumstances can give Density 𝜌𝑠 = 7850 𝑘𝑔/𝑚3
valuable feedback about construction costs which have not been
Poisson’s 's ratio 𝜈 = 0.3
discussed in previous studies. Moreover, a designer can select
the optimum dimension for the primary design mainly by Young's modulus 𝐸 = 200 𝐺𝑃𝑎
comparing the cost of the land and material required to build a Seismic Input Parameters
tank in high seismic zones. Evaluation of the demanded steel is Product Design Stress 𝑆𝑑 = 160 𝑀𝑃𝑎
necessary to determine if employing a particular seismic
Hydrostatic test Stress 𝑆𝑡 = 171 𝑀𝑃𝑎
reduction method is economically valuable.
Site subsoil class “C”
By way of example, the extra seismic steel for all considered Seismic zone hazard factor 𝑍 = 0.4
tanks and for the high and low levels of seismic intensity
Consequence of Failure “Extreme”
(Z = 0.4 and 0.1) are presented in Figures 9 and 10. It can be
seen that for broad tanks under low seismic intensity (𝑍 = 0.1), Distance to near fault 25 𝑘𝑚
an extra volume of seismic steel is not required and the
 79

Soil A&B Soil C Soil D

Figure 9: Extra steel volume demanded by seismic loads in high seismic zones (Z = 0.4).
80

Soil A&B Soil C Soil D

Figure 10: Extra steel volume demanded by seismic loads in high seismic zones (Z = 0.1).

ESTIMATION OF DEMANDED MATERIAL FOR capacity of the tank can be properly approximated as a linear
TANK CONSTRUCTION function. Figure 11 shows examples of such graphs for high
intensity seismic zones. As can be seen, for a specific tank
From the results of numerical analysis, it can be observed that capacity and aspect ratio using a higher tank height can
the variation in the total demanded steel for the tanks versus the
 81

substantially increase the demanded material. In all risk, a high seismic zone hazard factor (Z = 0.4) and soil types
circumstances, choosing the lower tank height for a specific A and B. The results were similar for other cases; thus, they
volume of contained liquid is economically preferable in high have not been presented here. The second-order polynomial
seismic zones. equations on the plots are the trend lines obtained by the Excel
The variation in required material volume versus tank capacity curve fitting function. As can be seen in Figure 12, the second-
is generally more useful for a specific aspect ratio (H/D). This order equation appears to properly predict the results of
variation can be adequately approximated using a second-order analysis; therefore, this function has been used to estimate the
function for all tanks as can be seen in Figure 12 for extreme total quantity of demanded steel.

Figure 11: Demanded steel volume versus tank capacity for various tank heights and seismic parameters (H/D = 0.25 - 2)
Shell Steel Volume (𝑚3 )
Shell Steel Volume (𝑚3 )
Shell Steel Volume (𝑚3 )

Tank Capacity (1000𝑚3 ) Tank Capacity (1000𝑚3 )

* VS=Required steel volume in tank shell (excluded the roof and floor steel) & VT=Tank Capacity

Figure 12: Required shell steel volume versus tank capacity for different H/D ratios (Z = 0.4; extreme; soil A&B.)
82

The general form of this relation in which the demanded steel Coefficients α, β and γ depend on the active seismic variables.
volume for building a tank is expressed as a second-order By considering six risk factors, the numerical value of 𝛼, 𝛽 and
function of tank capacity can be written as: 𝛾 can be extracted from the results of analysis and are presented
in Table 7 for all considered cases. Using this table, the total
𝑉𝑆 = 𝛼𝑉𝑇2 + 𝛽𝑉𝑇 + 𝛾 (29) demanded steel volume can be properly estimated in the initial
design stage. The optimum dimensions of a tank for storing of
In this relation, VT and VS are the total volume of stored liquid a specific volume of liquid then can be approximated based on
and the total demanded steel volume for construction of a the governing economic variables.
certain tank (excluded the roof and floor steel), respectively.
The units used for VS is in 𝑚3 and for VT is in 1000𝑚3 .

Table 7: Coefficients for calculating required shell steel volume

Risk Factor
0.70 0.54 0.40 0.30 0.20 0.13 (Structural)
H/D 𝜶 𝜷 𝜸 𝜶 𝜷 𝜸 𝜶 𝜷 𝜸 𝜶 𝜷 𝜸 𝜶 𝜷 𝜸 𝜶 𝜷 𝜸
0.1 0.003 0.92 -1.82 0.003 0.76 -1.25 0.002 0.70 -0.83 0.002 0.67 -0.66 0.002 0.69 -0.79 0.002 0.69 -0.79
0.2 0.004 1.15 -1.29 0.004 1.03 -0.96 0.003 0.98 -0.94 0.003 0.94 -0.66 0.003 0.94 -0.66 0.003 0.94 -0.66
0.3 0.011 1.28 -0.53 0.009 1.16 -0.46 0.008 1.06 -0.31 0.007 0.99 -0.12 0.007 0.99 -0.10 0.007 0.99 -0.99
0.4 0.018 1.36 -0.31 0.015 1.21 -0.14 0.014 1.08 0.01 0.010 1.10 -0.03 0.010 1.10 -0.03 0.010 1.10 -0.03
Soil A & B

0.5 0.028 1.37 -0.07 0.025 1.19 0.10 0.020 1.12 0.17 0.018 1.10 0.17 0.018 1.10 0.17 0.018 1.10 0.17
0.6 0.037 1.43 0.07 0.030 1.27 0.18 0.020 1.24 0.18 0.017 1.25 0.17 0.017 1.25 0.17 0.017 1.25 0.17
0.8 0.052 1.72 0.07 0.030 1.63 0.09 0.024 1.50 0.12 0.020 1.45 0.15 0.015 1.47 0.14 0.015 1.47 0.14
1.0 0.106 1.84 0.11 0.052 1.83 0.08 0.045 1.68 0.13 0.031 1.60 0.14 0.018 1.62 0.14 0.018 1.62 0.14
1.2 0.140 2.03 1.14 0.088 1.96 0.13 0.069 1.81 0.16 0.050 1.72 0.18 0.028 1.70 0.18 0.028 1.70 0.18
1.5 0.129 2.76 0.09 0.054 2.50 0.10 -0.063 2.45 0.10 -0.028 2.22 0.13 -0.136 2.26 0.13 -0.127 2.19 0.14
1.8 0.023 3.36 0.07 -0.020 2.95 0.09 -0.075 2.82 0.10 -0.124 2.65 0.11 -0.273 2.62 0.11 -0.350 2.66 0.12
2.0 0.080 3.61 0.08 0.090 3.17 0.09 -0.012 3.03 0.10 -0.182 2.85 0.11 -0.502 3.00 0.11 -0.567 2.97 0.11
0.1 0.004 0.91 -1.47 0.003 0.81 -1.30 0.003 0.72 -0.94 0.002 0.67 -0.75 0.002 0.69 -0.79 0.002 0.69 -0.79
0.2 0.005 1.23 -1.77 0.004 1.08 -1.27 0.003 0.98 -0.94 0.003 0.94 -0.66 0.003 0.94 -0.66 0.003 0.94 -0.66
0.3 0.013 1.39 -0.69 0.010 1.23 -0.58 0.010 1.08 -0.32 0.008 0.99 -0.12 0.007 0.99 -0.10 0.007 0.99 -0.99
0.4 0.022 1.44 -0.35 0.017 1.28 -0.24 0.015 1.15 -0.11 0.011 1.10 -0.03 0.010 1.10 -0.03 0.010 1.10 -0.03
0.5 0.033 1.46 -0.86 0.027 1.27 -0.01 0.023 1.17 0.08 0.018 1.10 0.16 0.018 1.10 0.17 0.018 1.10 0.17
Soil C

0.6 0.038 1.59 -0.06 0.028 1.42 0.07 0.026 1.27 0.16 0.017 1.25 0.17 0.017 1.35 0.17 0.017 1.25 0.17
0.8 0.054 1.80 0.03 0.041 1.63 0.08 0.032 1.51 0.12 0.013 1.51 0.13 0.015 1.47 0.14 0.015 1.47 0.14
1.0 0.110 1.95 0.10 0.079 1.79 0.10 0.046 1.73 0.11 0.048 1.58 0.15 0.018 1.62 0.14 0.018 1.62 0.14
1.2 0.148 2.14 0.12 0.110 1.94 0.13 0.081 1.85 0.15 0.072 1.69 0.19 0.028 1.70 0.17 0.028 1.70 0.18
1.5 0.170 2.81 0.09 0.092 2.57 0.10 -0.110 2.48 0.10 -0.038 2.33 0.12 -0.102 2.21 0.13 -0.127 2.19 0.14
1.8 0.120 3.48 0.07 0.014 3.13 0.09 0.148 3.00 0.08 -0.176 2.80 0.10 -0.257 2.63 0.12 -0.350 2.66 0.12
2.0 0.110 3.93 0.05 -0.102 3.50 0.08 -0.096 3.17 0.09 -0.228 3.04 0.10 -0.403 2.95 0.11 -0.567 2.97 0.11
0.1 0.004 0.91 -1.40 0.003 0.79 -1.35 0.003 0.71 -1.21 0.004 0.55 0.07 0.002 0.63 -0.76 0.002 0.69 -0.79
0.2 0.005 1.27 -1.73 0.004 1.10 -1.30 0.003 1.01 -1.09 0.003 0.94 -0.66 0.003 0.94 -0.66 0.003 0.94 -0.66
0.3 0.013 1.45 -0.86 0.011 1.25 -0.69 0.010 1.11 -0.46 0.008 0.98 -0.11 0.007 0.99 -0.10 0.007 0.99 -0.10
0.4 0.021 1.51 -0.40 0.017 1.30 -0.23 0.015 1.16 -0.09 0.013 1.06 0.04 0.010 1.10 -0.03 0.010 1.10 -0.03
0.5 0.035 1.44 -0.06 0.028 1.26 0.01 0.022 1.17 0.08 0.019 1.09 0.19 0.018 1.10 0.17 0.018 1.10 0.17
Soil D

0.6 0.039 1.6 -0.05 0.034 1.37 0.12 0.034 1.37 0.12 0.017 1.25 0.17 0.017 1.25 0.17 0.017 1.25 0.17
0.8 0.063 1.84 0.04 0.048 1.66 0.07 0.048 1.66 0.07 0.028 1.43 0.16 0.015 1.47 0.14 0.015 1.47 0.14
1.0 0.120 2.12 0.04 0.082 1.90 0.08 0.082 1.90 0.08 0.041 1.65 0.14 0.018 1.62 0.14 0.018 1.62 0.14
1.2 0.180 2.24 0.11 0.120 2.05 1.30 0.121 2.05 0.13 0.080 1.76 0.16 0.028 1.70 0.18 0.028 1.70 0.18
1.5 0.220 3.04 0.07 0.130 2.76 0.09 0.130 2.76 0.09 0.032 2.32 0.12 -0.073 2.26 0.13 0.127 2.19 0.14
1.8 0.240 3.65 0.05 0.045 3.37 0.07 0.045 3.37 0.07 -0.019 2.76 0.10 -0.196 2.69 0.11 0.350 2.66 0.12
2.0 0.170 4.23 0.04 0.080 3.67 0.08 0.081 3.67 0.08 -0.107 3.07 0.10 -0.260 2.91 0.11 0.567 2.96 0.11

CONCLUSIONS 3) For slender tanks (H/D ≥ 1.2), the diamond-shaped

buckling mode is the governing mode, except for tall tanks.
The present study qualitatively and quantitatively provides
For tall slender tanks (H ≥ 14), elephant-foot buckling
comprehensive insight into the interaction of seismic
should be typically expected.
parameters and shell buckling modes of a cylindrical steel
liquid storage tank. For this purpose, a range of tank dimensions 4) Although looser or softer soil reduces the buckling stress,
and related parameters covering all practical cases was the critical buckling mode in the shell of a tank does not
considered in a comprehensive parametric study. The range of alter in response to a change in soil type; therefore, the
tank geometries for each shell damage type was determined effects of soil type on the dominant failure mode can be
based on the available provisions. The interaction between the ignored.
governing shell buckling mode, tank geometry and various In the second part of the study, the effects of seismic parameters
seismic parameters was investigated and the following on the economics of a project as well as the selection of the
conclusions were drawn: optimum tank geometry for a specific seismic area have been
1) The critical buckling mode mainly depends on both the tank discussed and the following conclusions were drawn:
height (H) and the tank aspect ratio (H/D). 5) In low seismic intensity zones (Z = 0.1 and 0.2), there is no
2) For broad tanks (H/D ≤ 0.4), the dominant failure mode of need for the extra material demanded by seismic loads for
a tank shell is mainly caused by the dynamic hoop stress. broad tanks; the structural steel will be sufficient to resist
For medium tanks (0.4 < H/D ≤ 1.2), the governing seismic loads. Also, for slender tanks, the extra material
buckling mode of failure is mainly elephant-foot buckling. required by seismic loads is not considerable (less than
13%).
 83

6) In high seismic intensity zones (Z = 0.4), the extra steel 7 Teng JG and Rotter JM (2004). “Buckling of thin Metal
material demanded by seismic loads is not considerable Shells”. CRC Press, ISBN-13: 978- 0419241904, 520 pp.
(less than 10%) if the level of seismic risk is low or slight. 8 Niwa A and Clough RW (1982). “Buckling of cylindrical
However, if moderate, serious or extreme levels of seismic liquid‐storage tanks under earthquake loading”. Earthquake
risk exist, the extra steel volume demanded by seismic loads Engineering & Structural Dynamics, 10(1): 107-122.
can be up to 1%, 25% and 45% for broad tanks,
9 Jia Z and Ketter RL (1989). “Experimental Study of
respectively. For slender tanks, these values will increase
elephant Foot Bulge Instability of Thin-walled Metal
further up to 40%, 60% and 100%, respectively. Therefore,
Tanks”. Technical Report NCEER-89-0004, National
in high seismic intensity zones, choosing a smaller height
Center for Earthquake Engineering, State university of New
and lower aspect ratio for specific seismic circumstances
York at Buffalo.
can significantly reduce the required material and related
construction cost. An economical balance should be struck 10 Rotter JM (1985). “Local Inelastic Collapse of Pressurized
between the higher cost of the land required and the lower Thin Cylindrical Steel Shells under Axial Compression”.
seismically demanded material for storing a specific Research Report No. R502, School of Civil and Mining
volume of liquid. Engineering, University of Sydney, Australia, 26pp.
7) Regardless of tank geometry, the steel volume solely 11 Rotter JM (1987). “Bending theory of shells for bins and
demanded by seismic loads for a tank built on soft soil (type silos”. Transactions of Mechanical Engineering, Institution
D) is up to 43% more than for one placed on rocky soil of Engineers, Australia, 12(3): 147-159.
(types A and B). Therefore, employing various methods for 12 Rotter JM and Seide P (1987). “On the design of unstiffened
enhancing the soil strength or using the different types of cylindrical shells subject to axial load and internal
tank foundations to tackle the softness of soil can be well pressure”. Stability of Plate and Shell Structures,
compensated for by the cost of extra material demanded by Proceedings of An International Colloquium, Ghent,
earthquake loads in high seismic intensity zones. Belgium, 6-8 April 1987.
A useful empirical relation that covers all practical situations is 13 Rotter JM (1990). “Local collapse of axially compressed
proposed herein based on the results of analysis for preliminary pressurized thin steel cylinders”. Journal of Structural
estimation of the extra material demanded by earthquake Engineering, ASCE, 116(7): 1955-1970.
loading under different various seismic conditions and for 14 El-Bkaily M and Peek R (1998). “Plastic buckling of
different tank geometries. The demanded steel volume provided unanchored roofed tanks under dynamic loads”. Journal of
in this study is based on the assumption of ductility factor of engineering mechanics, 124(6): 648-657.
1.25 for anchored tanks with non-ductile holding down bolts. It
should be noticed that the steel volume predicted here may be 15 Jaiswal OR, Durgesh CR and Sudhir KJ (2007). “Review of
slightly reduced if ductile behaviour is somehow expected. seismic codes on liquid containing tanks”. Earthquake
Spectra, 23(1): 239–260.
ACKNOWLEDGEMENTS 16 Chen Z, Sun B, Yu C, Fang Z and Zeng M (2009).
“Comparison of the strength design and prevention method
This work has been supported by the International Institute of of elephant foot buckling among countries’ standards of oil
Earthquake Engineering and Seismology (IIEES) and Iranian tanks”. Proceedings of the ASME 2009 Pressure Vessels
National Science Foundation (INSF). The author wishes to and Piping Division Conference (PVP2009), Prague, Czech
express his gratitude to them for their supports. Republic, July 26-30, 2009, Vol. 1, 461–466.
17 Virella C, Godoy LA and Suárez LE (2006). “Dynamic
REFERENCES buckling of anchored steel tanks subjected to horizontal
1 Cooper TW (1997). “A study of the performance of earthquake excitation”. Journal of Constructional Steel
petroleum storage tanks during earthquakes, 1933-1995”. Research, 62(6): 521-53 pp.
Report NIST GCR 97-720, US National Institute of 18 Chonghou Z, Yansheng L and Goto Y (2008). “Plastic
Standards and Technology, 102 pp. buckling of cylindrical shells under transverse loading”.
2 Pineda P, Saragoni GR and Arze E (2012). “Performance of Tsinghua Science & Technology, 13(2), 202–210.
steel tanks in Chile 2010 and 1985 earthquakes”. 19 Chen Z, Yan S, Yu C, Cao G, Wang L and Yang L (2011).
Proceedings of the 7th International Conference on “Effect of opening on elephant’s foot buckling of large oil
Behavior of Steel Structures in Seismic Areas (Stessa’12), tanks”. Advanced Materials Research, 163(2011): 169-172.
9-11 January, Santiago, Chile, S09b-3, 337-342. 20 Yang L, Chen Z, Cao G, Yu C and Guo W (2013). “An
3 Erica F, Liu C and Varma AH (2016). “Investigation of analytical formula for elastic-plastic instability of large oil
Cylindrical Steel Tank Damage at Wineries during storage tanks.” International Journal of Pressure Vessels
Earthquakes: Lessons Learned and Mitigation and Piping, 101(2013): 72-80.
Opportunities”. Practice Periodical on Structural Design 21 Elkashef M, Rashed A and Haroun M (2014). “A study on
and Construction, 21(3): 04016004-1-11. the elephant-foot bulge formation of anchored steel tanks”.
4 Eshghi S and Razzaghi MS (2007). “Performance of Australian Journal of Structural Engineering, 15(3): 321-
cylindrical liquid storage tanks in Silakhor, Iran earthquake 335.
of March 31, 2006”. Bulletin of the New Zealand Society 22 Zingoni A (2015). “Liquid-containment shells of
for Earthquake Engineering, 40(4): 173-182. revolution: a review of recent studies on strength, stability
5 Erick G, Almazán J, Beltrán J, Herrera R and Sandoval V and dynamics”. Thin-Walled Structures, 87(2015): 102-
(2013). “Performance of stainless steel winery tanks during 114.
the 02/27/2010 Maule Earthquake”. Engineering Structures 23 Godoy LA (2016). “Buckling of vertical oil storage steel
56: 1402-1418. tanks: Review of static buckling studies”. Thin-Walled
6 Brunesi E, Nascimbene R, Pagani M and Beilic D (2014). Structures, 103(2016): 1-21.
“Seismic performance of storage steel tanks during the May 24 Yazdani M, Rahimi H, Khatibi AA and Hamzeh S (2009).
2012 Emilia, Italy, earthquakes”. Journal of Performance “An experimental investigation into the buckling of GFRP
of Constructed Facilities, 29(5): 04014137-1-9. stiffened shells under axial loading”. Scientific Research
and Essays, 4(9): 914-920.
 84

25 Vakili M and Showkati H (2015). “Experimental and 31 Standard, British (1998). “EN 1998-4, Eurocode 8: Design
Numerical Investigation of Elephant Foot Buckling and Provisions of Earthquake Resistance of Structures, Part 4:
Retrofitting of Cylindrical Shells by FRP”. The Journal of Silos, Tanks and Pipelines”. European Committee for
Composites for Construction, 20(4): 04015087-1-9. Standardization, Brussels, 83 pp.
26 Colombo I and Almazán JL (2015). “Seismic reliability of 32 Standards New Zealand (1986). “Seismic design of storage
continuously supported steel wine storage tanks retrofitted tanks, Recommendations of a NZSEE Study Group on
with energy dissipation devices”. Engineering Structures, Seismic Design of Storage Tanks”. (Red Book), Standards
98(2015): 201-211. New Zealand, Wellington, 180 pp.
27 Vathi M, Karamanos SA, Kapogiannis IA and Spiliopoulos 33 Standards New Zealand (2009). “Seismic design of storage
KV (2015). “Performance Criteria for Liquid Storage Tanks tanks, Recommendations of a NZSEE Study Group on
and piping systems subjected to seismic loading”. Journal Seismic Design of Storage Tanks”. (Blue Book), Standards
of Pressure Vessel Technology, 139(5): 051801. New Zealand, Wellington, 177 pp.
28 Paolacci F, Phan HN, Corritore D, Alessandri S, Bursi OS 34 American Petroleum Institute (2013). “Welded Steel Tanks
and Reza MS (2015). “Seismic fragility analysis of steel for Oil Storage”. API 650 Twelfth Edition, USA, 514pp.
storage tanks”. Proceedings of the 5th ECCOMAS Thematic 35 Rotter JM (2006). “Elephant’s foot buckling in pressurised
Conference on Computational Methods in Structural cylindrical shells”. © Ernst & Sohn Verlag für Architektur
Dynamics and Earthquake Engineering (COMPDYN’15), und technische Wissenschaften GmbH & Co. KG, Berlin ·
Crete Island, Greece, 25–27 May 2015. Stahlbau, 75: 742-747.
29 Bakalis K, Vamvatsikos D and Fragiadakis M (2015). 36 Petrowiki (2016), http://petrowiki.org/Oil_storage,
“Seismic Fragility Assessment of Steel Liquid Storage (Accessed 17/9/2016).
Tanks”. Proceedings of ASME 2015 Pressure Vessels and
Piping Conference (PVP), Boston, Massachusetts, USA. 37 Wozniak RS (1990). “Steel Tanks in Structural
July 19-23, 2015, PVP2015-45370. Engineering Handbook”. Gaylord, EH and Gaylord, Eds.,
3rd Edition, McGraw-Hill, NY, USA, 27-1:27-29.
30 Maekawa A (2012). “Recent Advances in Seismic Response
Analysis of Cylindrical Liquid Storage Tanks”. in 38 Standards New Zealand (2004). "NZS1170.5: Structural
Earthquake-Resistant Structures-Design, Assessment and Design Actions. Part 5: Earthquake Actions‐New Zealand”.
Rehabilitation, Chapter 12, InTech, 307-336. Standards New Zealand, Wellington, New Zealand, 76 pp.

View publication stats