Behavior of Concrete Liquid Containing Structures Subjected

to Seismic Loading

Kianoush, M.R., Tso, W.K., and Hamidi, M.

Synopsis:

The behavior of liquid containing structures (LCS) subjected to seismic excitations is

reviewed. The major parameters affecting the response of concrete circular tanks for LCS
are discussed. Existing codes and standards related to seismic design of LCS are reviewed.
With the aid of a design example, results of the various design standards are compared. The
effects of earthquake load on the behavior of reinforced concrete (RC) tanks are also
investigated through a detailed example.

Keywords: seismic, earthquake, design code, concrete, liquid-containing, water retaining,

circular tank
ACI member, M. Reza Kianoush is a Professor of Civil Engineering at Ryerson
University, Toronto, Canada. He received his Ph.D. from the University of Alberta in 1986.
He is a member of ACI Committee 350 as well as the Chair of Education Subcommittee.
His research interests include analytical modeling of reinforced concrete structures as well
as behavior of concrete liquid containing structures under seismic loading.

W.K. Tso is Professor Emeritus of Civil Engineering at McMaster University, Hamilton,

Canada. He obtained Masters and Ph.D. degrees from California Institute of Technology.

H. Hamidi is a Research Associate at Western University, London, Canada. He obtained

his Ph.D. from Sharif University of Technology, Tehran, Iran in 2000. His research
interests include analytical modeling and design of structures with a particular emphasis on
seismic performance.

INTRODUCTION

Liquid containing structures (LCS) for environmental engineering concrete (EEC)

structures are considered as lifeline facilities that require careful design and detailing. The
performance of LCS during earthquakes is more important than the economic values of the
tanks and their contents. It is important that utility facilities remain operational following
an earthquake to meet the emergency requirements such as firefighting water or meet the
public demands as a source of water supply. On the other hand, the containment of
hazardous waste is important to meet the public safety after a seismic event. For these
reasons, serviceability (crack and leakage control) becomes the prime design consideration
in most of these structures. Cracking in these structures should be limited to micro-cracks.

There has been considerable research carried out on the seismic behavior of steel circular
tanks. However, the behavior of concrete circular tanks for LCS during earthquakes is quite
different from that of steel circular tanks. The most common mode of failure in a circular steel
tank is by buckling of the wall near its base that is commonly referred to as "elephant’s foot"
buckling. The main cause of this type of failure is due to large overturning moments caused by
hydrodynamic effects. However, in concrete tanks, large hydrodynamic forces could develop
in the highly stressed regions of the tank which could lead to severe cracking and leakage
which may eventually lead to failure of the tank.

Fig. 1 (a) shows the model of a circular concrete tank under consideration. Under horizontal
seismic excitations, the hydrodynamic forces in the tank can be induced due to two effects.
First, the bottom portion of the liquid in the tank accelerates in unison with the walls of the
tank. This mass is shown as MI and is called the impulsive mass. This system of structural
response behaves in a high frequency (short period) mode of vibration. Second, the upper
portion of liquid in the tank behaves as if an oscillating mass, MC, flexibly connected to the
walls. This is known as the convective mass. This system of response behaves in a low
frequency (long period, T > 3.0 sec) mode of oscillation with damping of the order of 0.5%.
The impulsive component of the mass generates pressure and suction, which accelerates into
the fluid and away from the fluid, respectively. These effects will produce the force PI which
acts at a distance hI from the bottom of the tank and is evenly divided on the water side of the
walls as shown in Fig. 1(b). The convective pressures on the tank wall induce the resultant
force PC, which acts at a distance of hC from the tank bottom. Since the impulsive and
convective masses oscillate at different frequencies, it is reasonable to decouple the two
components and treat them separately.

The mathematical model described above was originally developed by Housner [1] that
assumes that the tank is rigid. This model was later modified by Veletsos [2] by assuming that
the tank-liquid system behaves as a single degree of freedom system. The analytical study by
Veletsos and Yang [3] showed that the impulsive forces in flexible tanks could be considerably
higher as compared with rigid tanks. However, the convective components were not affected
due to the flexibility of the tank.

The tank model described above is valid when the tank is relatively shallow (i.e. H / R ≤ 1.5 ).
However, for deep tanks (i.e. H / R > 1.5 ), a slightly different model needs to be adopted.
Housner [1] showed that for deep tanks, generation of liquid velocity relative to the tank is
essentially restricted to a height of 1.5 R from the free surface. The mathematical model for
these types of tanks is described by Epstein [4]. Most of the ground tanks built in Canada for
EEC structures are of shallow type. For this reason, the work described in this paper is
restricted to shallow tanks. However, minor modifications will make it applicable to deep
tanks.

In this paper, the major parameters that affect the response of concrete circular tanks for LCS
subjected to earthquake loading will be investigated. The existing documents related to seismic
design of LCS shall be reviewed. Based on such a review, recommendations shall be made on
the seismic design criteria for LCS.

Response of Tank-Liquid System

Tank parameters

Based on the Housner’s tank model described above, the impulsive and convective weights
(WI and WC) and the equivalent heights (hI and hC) are given by:

β 3.187
[1] WI = ( ) WL tanh ( )
3.187 β
0.846
[2] WC = ( ) WL tanh (β)
β
[3] h I = 0.375 H L
 cosh β - 1
[4] h C = (1 − )HL
β sinhβ

where β = 1.84 (HL / R) and WL is the total weight of liquid in the tank.

The impulsive and convective forces (PI and PC) produce bending moments in the tank wall
just above the base. However, to determine the overturning moment at a section just below the
base of the tank wall, the dynamic pressure distribution on the bottom of the tank should be
considered. For this purpose new equivalent heights must be used. These equivalent heights
are defined in the TID-7024 report [5].

Assuming that the tank wall is rigid, the maximum horizontal force exerted against the tank
due to the impulsive and convective weights of liquid are:

WI &&
[5] PI = ( ) Ug
g
W &&
[6] PC = ( C ) U
g

&& is the maximum acceleration of the

&& is the maximum ground acceleration and U
where U g
sloshing (convective) mass. However, if it is assumed that the tank wall is not rigid, the effect
of the tank flexibility needs to be considered in equation 5. For this purpose, the period of
vibration of the tank-liquid system has to be established.

Period of Vibration of Tank-Liquid System

Impulsive component

A tank wall can vibrate in several different modes if assumed that it behaves elastically.
Representative modes of vibration for a cantilever circular tank wall are shown in Fig. 2. The
axially symmetrical mode corresponding to breathing response (n=0) is exited by vertical
motion only and produces hoop stresses in the tank wall. Of interest here is the bending
response (n=1) of the tank, which is excited by horizontal motion. The frequency of this mode
is affected by the impulsive mass of the liquid and the geometric and material properties of the
tank itself.

The natural frequencies of circular tanks can be obtained using closed form solutions. Yang
[6] used Flugge’s shell theory to determine the natural modes of vibration of cylindrical steel
tanks. The effect of the contained liquid was approximated by assuming that a portion of the
mass of the liquid was attached to the shell. Based on height over radius ratio (i.e. H / R), the
behavior of cylindrical tanks was described in terms of four simple types of structural actions
that are summarized in Table 1.

For concrete circular tanks under consideration, i.e. H / R<1.5, the behavior will be mostly
similar to type 1 or 2 action. The fundamental natural frequencies for this mode of behavior
(type 2) as given by Yang [6] are:

1 E
[7] ω = Empty Tank
R 1− ν 2 ρS

1 1 E
[8] ω = Full Tank
R 1− ν 2
1 ε L ρL R ρS
1+
2 ε S ρS t

Where t is the wall thickness, E and ν are the modulus of elasticity and the Poisson’s ratio of
concrete respectively. ρL, ρS and εL, εS are the mass density and the virtual mass coefficients for
the liquid and the tank respectively [6].
A similar expression for the natural frequency of fluid-tank system has also been developed by
Haroun and Abou-Izzadin [7]. This is given in terms of non-dimensional parameter ω that is
based on a best-fit formula as follows:

1 E  t H 
[9] ω = 2π 2   ω
 H ρ L   R R 

[
[10] ω = 0.000335(H/R) 3 − 0.000021(H/R) 2 − 0.016361(H/R) + 0.065598 ]

Convective Component

As mentioned earlier, the flexibility of the tank wall has little or no effect on the convective
forces. However, to determine seismic forces, the frequency of the convective component of
the liquid has to be established.

The frequency of the mass-spring system shown in Fig. 1 is given by:

1.84 g tanh β
[11] ω 2 =
R
 Design of Tank-Liquid System

Review of existing codes and standards

Currently, there are few provisions in the North American Building Codes related to seismic
design of LCS. The Uniform Building Code, UBC [8] addresses seismic requirements for
liquid holding tanks in a very general manner. Recently the ACI Committee 350 [9] has
developed an independent code for seismic design of LCS for EEC structures. In addition, this
Committee has developed a separate document [10] in a mandatory form for seismic design of
LCS. Also, there are some standards that provide guidelines for determining seismic forces for
LCS. The U.S. Atomic Energy Commission Report, TID-7024 [5] and the NSF report [11] are
most widely used in practice. It is believed that tanks that have been designed based on these
two standards have performed well under earthquakes. In these two standards, there are no
specific statements related to concrete tanks. The only comprehensive code other than ACI
350 [9] that is related to concrete LCS is the New Zealand Standard, NZS 3106 [12].

In most current codes and standards, the determination of tank parameters is based on Eqs. 1 to
4 and 11. Table 2 shows a summary of a comparison of some of the current codes and
standards for determination of the seismic base shear. In this table, some of the symbols are
slightly adjusted for better comparison.

Both the TID 7024 report and the NSF report are based on the rigid tank assumption. The
TID-7024 method is based on average velocity response spectrum, which requires information
for the particular site. The NSF report is based on the provision of the ATC-3-06 [13]. Based
on this report, two alternative design approaches can be used, namely the working stress design
(WSD) method and the strength design (SD) method. The coefficients, CW and CS , control the
performance criteria of the tank for WSD and SD, respectively. For example for CW = 0.8,
shut down of the facility is acceptable up to 6 months and for CW = 2.4, continuous operation is
essential. In both reports, the total base shear is determined by simply adding the forces
obtained from the impulsive and the convective components.

The NZS [12] is also based on rigid tank assumption for circular tanks. The effect of tank
flexibility is considered only in rectangular tanks. A risk factor, R, is used which is based on
the performance criteria of the tank and the annual probability of exceedance (APE) of the
design earthquake. For example, R = 1.0 is for tanks which should remain functional in the
restoration period for a moderate earthquake based on APE of 0.01 and R = 2.0 is for tanks
containing highly valuable or hazardous contents based on APE of 0.0004. The impulsive and
convective components of forces are combined by the square root of the sum of the squares
(SRSS) rule. Since the two components oscillate at two different frequencies, the use of this
rule is quite valid. In the proposed ACI 350 method, the flexibility of the tank wall is
considered using the CI parameter. The basic equation for the seismic base shear is similar to
the NSZ standard. However, different structural system factors are used for impulsive (RWI)
and convective (RWC) components. RWI varies from 2.0 to 4.5. For example, anchored
flexible-base tanks and nonflexible-base tanks (fixed or hinged) are assigned RWI values of 4.5
and 2.75 respectively. The value of RWC is set to be 1.0. This is because the convective
component of the liquid has a high natural period of vibration (T > 3 seconds) in which the
system will respond in the displacement sensitive region of the response spectrum curve. In
this case, the maximum deformation will be the same as the deformation of the linear system
and therefore RWC = 1.0. The importance factor I = 1.0 is used for tanks which may undertake
repairable structural damage and I = 1.25 are for tanks that must remain usable with slight
structural damage. For tanks containing hazardous materials I is assigned a value of 1.5.

Design example

Consider an open top circular concrete fixed-base unburied tank having the following
dimensions:
R = 15 m, H = 6 m and t = 0.3 m.
The tank is assumed to be full of water of density = 1000 kg/m3. Concrete has a mass density
of 2400 kg/m3 and a modulus of elasticity of 2x1010 N/m2. Furthermore, it is assumed that
importance factor for the reservoir is I=1.0, the type of soil under the tank is rock (S=1.0) and
it is located in a high-risk seismic zone (Z=0.4).
The design forces according to four design standards described earlier are computed through
formulas in Table 2, and the design forces are summarized in Table 3. The comparison
between different design methods indicates that all of the methods give forces greater than that
of ACI-350 (they are 40 to 60% more conservative than ACI). It is generally accepted that
the reservoirs designed according to TID or NSF are considered as very conservative designs.

Earthquake Load Effects on RC Tanks

To practically understand the earthquake load effect on the RC circular tank walls, a
parametric study is conducted to study the effect of the ratio of earthquake load to the
hydrostatic load. This study was done using the ACI 350 Code (2001). Earthquake moment
MEQ, static moment MS, earthquake base shear VEQ, static base shear VS are calculated based
on the code procedure. Variables used in this study include the height, HL (3 to 10m) and the
thickness, tw (300 to 600mm) of the tank wall and the seismic zone factor Z. The radius of the
tank in this study is considered as a constant parameter equal to 20 meters.

The results of the study are shown in Figs. 3 to 5. Detailed comparison of the results leads to
the following observations:

1. The effect of the earthquake moment MEQ is more significant than that of earthquake
shear VEQ. Comparison of Figs. 3 and 4 shows that the ratio of MEQ/MS is consistently greater
than VEQ/VS for different tank walls. Therefore, the effect of the moment due to earthquake on
the structure is more critical than that of shear to the structure.

2. The earthquake effect is more critical for shallower tank walls (3 to 5m) than that of
taller tanks (6 to 10m). As shown in Figs. 3 and 4, MEQ/MS and VEQ/VS decrease as the height
of the wall increases. Therefore, it can be concluded that for a relatively shallow tank (3 to 5m)
the earthquake load effect is more significant than that of static load. This is because static base
shear, VS, increases with the parabolic of the water height and the static moment MS increases
with the cubic of the water height, while the earthquake base shear, VEQ, and moment, MEQ,
increase fairly linearly. Thus, for taller tanks, (e.g. wall height of 10m) the static shear and
moment, VS and MS, are the major parameters that affect the response of the tank.

3. While the earthquake load effect is significant for designing in seismic Zone 4 (with
Z=0.4), the earthquake effect also needs to be considered in designing in seismic Zone 2B
(with Z=0.2). However, as indicated in Fig. 5, the effect of the static moment is more critical
than that due to earthquake load effect.

4. The effect of tank wall thickness on the earthquake load is negligible. Figs. 3 and 4
show that the MEQ/MS and VEQ/VS are not so sensitive to wall thickness changes. The effect of
wall thickness becomes more significant only in very tall tanks.

Conclusion

The simplified approach, developed originally by Housner for analyzing liquid containing
structures is adopted by most of the design standards to estimate the design seismic forces.
Based on the usage and importance of the tanks and its content, these standards use greater or
smaller coefficients for the design. As the example in this paper showed, significant differences
between different standards exist.

A parametric study on the effects of the earthquake on the circular tanks showed that:
- In designing of the tank walls, earthquake induced bending moment is more critical
than shear.
- Earthquake load dominates the design of shallow water tanks, but for taller tanks, the
static load becomes the dominating factor.
- Earthquake load is not affected significantly by the variation of wall thickness.
- Although, in the seismic Zone 2B, the static load dominates the design of the tank, but
the seismic load cannot be disregarded, due to its effect as compared with static loads.

References

1. Housner, G.W. 1957. Dynamic pressures on accelerated fluid containers, Bulletin of the
Seismological Society of America, 47(1): 15-35.

2. Veletsos, A.S. 1974. Seismic effects in flexible storage tanks, proc. int. assoc. for
earthquakes eng., Rome, Italy, pp. 630-639.
3. Veletsos, A.S., and Yang, J.Y. 1977. Earthquake response of liquid storage tanks. Proc. of
2nd Annual ASCE Engineering Mechanics Specialty Conference, North Carolina State Univ.,
Raleigh, N.C., pp. 1-24.

4. Epstein, H.I. 1976. Seismic design of liquid storage tanks. ASCE Journal of Structural
Division, 102 (ST9): 1659-1673.

5. TID-7024, 1963. Nuclear reactors and earthquakes, Prepared by Lockhead Aircraft Corp.
and Holmes and Nurver, Inc. U.S. Atomic Energy Commission.

6. Yang, J.Y. 1976. Dynamic Behavior of Fluid-Tank System. Ph.D. Thesis, Rice University,
Texas.

7. Haroun, M.A. and Abou-Izzeddine, W. 1992. Parametric study of seismic soil-tank

interaction I: Horizontal excitation, ASCE Journal of Structural Engineering, 118(3): 783-797.

8. Uniform Building Code, 1995. International Conference of Building Officials. Whittier,

California.

9. ACI 350-01, 2001. Code requirements for environmental engineering concrete structures
and commentary, American Concrete Institute, Detroit, Michigan.

10. ACI 350.3/350.3R-01, 2001. Seismic design of liquid-containing concrete structures and
commentary. American Concrete Institute, Detroit, Michigan.

11. NSF Report, 1980. Earthquake design criteria for water supply and wastewater systems. A
National Science Foundation Report, Environmental Quality Systems Inc., Rockville,
Maryland.

12. New Zealand Standard, 1986. Code of practice for concrete structures for storage of
liquids. NZS 3106:1986, New Zealand.

13. ATC-3-06, 1978. Tentative provisions for the development of seismic regulations for
buildings, Applied Technology Council, National Bureau of Standards Special Publication -
510, National Science Foundation Publication 78-8.
Table 1. Effect of H/R Ratio on Structural Action

Type H/R Structural Action

1 < 0.2 Bending and shearing vibrations. Each vertical strip of tank wall
behaves as a cantilever beam in either bending or shearing
vibration
2 <1.0 Extentional vibration of a half circular arch. Each cross section
of the tank behaves as a circular arch with an angle of opening
of 180O fixed at both ends.
3 Medium H/R Shearing vibrations of a cantilever beam mode shape which is
values say similar to a half sine curve.
1.5
4 > 5.0 Bending vibration of a cantilever beam.
Table 2. Lateral force requirement for liquid containing structures.

TID - 7024 NSF (1980) NZS (1986) ACI 350-01

(1963) (2001)
&&
U 1.2CA a ZISCI
Base Shear 0
WI WI CIWI WI
impulsive (VI) g g R WI

Base Shear A1 ω 2 1.2C C1 CCWC ZISCC

WC WC WC
convective (VC) g 2R R WC

Base Shear VI + VC VI + VC VI 2 + VC 2 VI 2 + VC 2
Total

Numerical ω = circular --- --- ---

Coefficients frequency of
slosh liquid
&& = max. Aa = 0.05 to Z = 0.075 to
U 0 A0 = 0.4R
ground accl. 0.4 0.4
Zone Factor 0.3R
AV = 0.05 to 0.2R
A1 = max.
displacement 0.4
--- AV S 2.75
C1 = CI = A0 CI =
T 2/3
C
S
Seismic or
or
Response 1.25
CI =
Factor 2.5A V S 2.7A0 Ti2/3
C1 = CC =
T 4/3
TC1.4
C
6
CC =
TC2

--- C = Cs = 1.0
Importance 2.0
3.0 I = 1.0
Factor to
C = Cw = 0.8
1.6 R = 2.0 1.5
2.4 1.6
1.3
--- 1.0 RWI = 2.0
Structural R = 2.0 to
0.7
System 4.5
Factor
RWC = 1.0
Soil Factor --- S = 1.0 to 1.5 --- S = 1.0 to 2.0
Table 3. Comparison of Design Parameters
 Base Shear (kN)
Base moment Coefficients
(kN-m)
Impulsive Convective Total

TID-7024 5475 2083 7558 18840

(1963)
NSF (1980) 5257 1521 6779 17776 CW = 1.6
NZS (1986) 7121 2598 7580 19403 R = 1.3
ACI 350-01 4560 1372 4762 11621 RWI = 2.75
(2001)