On the Effectiveness of Two

Isolation Systems for the

Seismic Protection of
Fabrizio Paolacci
Elevated Tanks
Mem. ASME
This paper deals with the effectiveness of two isolation systems for the seismic protection
Department of Engineering,
of elevated steel storage tanks. In particular, the performance of high damping rubber
Roma Tre University,
Rome 00033, Italy
bearings (HDRB) and friction pendulum isolators (FPS) has been analyzed. As case
study, an emblematic example of elevated tanks collapsed during the Koaceli Earthquake
in 1999 at Habas pharmaceutics plant in Turkey is considered. A time-history analysis
conducted using lumped mass models demonstrates the high demand in terms of base
shear required to the support columns and their inevitable collapse due to the insufficient
shear strength. A proper design of HDRB and FPS isolator according to the EN1998 and
a complete nonlinear analysis of the isolated tanks proved the high effectiveness of both
isolation systems in reducing the response of the case tank. Actually, the stability condi-
tions imposed by the code and a reduced level of convective base shear obtained with the
second isolation typology suggests the use of FPS isolators rather than HDRB devices.
[DOI: 10.1115/1.4029590]

1 Introduction action on the tank two isolation systems are designed: HDRB and
FPS. Their behavior will be analyzed and compared in order to
The seismic response of elevated tanks on steel or concrete sup-
show the effectiveness of each isolation system, identifying the
port structures has been widely investigated in the past. For a slen-
main strengths and defects.
der support configuration, this kind of tanks has a natural filtering
capacity against seismic actions. In this case, the isolation system
does not represent a proper solution for the mitigation of the struc- 2 Dynamics of Elevated Tanks Upon Seismic Isolators
tural response; indeed, dissipative bracings could represent a more
The dynamics of cylindrical tanks subjected to a base motion
effective technique, as demonstrated in literature by several
has been extensively studied by several authors. Starting from the
authors [1,2]. On the contrary, in case of stiff support the positive
earliest work of Housner [10], the hydrodynamic pressure induced
filtering effect could be limited. Moreover, in case of elevated
by the liquid on the tank wall due to the base motion has been
tanks on reinforced concrete columns, the high shear stiffness of
determined, taking into account the deformability of the tank
the support may induce premature shear failure in the columns, as
wall; see for example Refs. [1113].
shown in recent seismic events. For example, during Itzmit earth-
In brief, the liquid mass can be imagined subdivided in two
quake (1999) in Turkey, a series of elevated storage tanks contain-
parts: an impulsive component, which follows the base motion
ing liquefied oxygen were seriously damaged or collapsed [3].
and the deformability of the tank wall, and a convective compo-
This is a clear case in which base isolation technique could repre-
nent, whose oscillations cause superficial waves of different fre-
sent an effective solution for the seismic response mitigation [4].
quency and a very low percentage of excited mass (4%) mostly
Typical isolation systems are often based on Lead Rubber or
related to the higher vibration modes; moreover, while in the slen-
HDRB [5], and spherical sliding bearings like FPS with single or
der tanks most of the liquid moves rigidly with the tank in the
multiple surfaces [6,7].
broad tanks most of liquid oscillates in the convective mode.
Despite the fact the seismic isolation of elevated tanks has been
Under the hypothesis of rigid tank, the impulsive and convec-
widely investigated in the past, only few applications have been
tive part of hydrodynamic pressure can be easily evaluated taking
proposed in Refs. [8] and [9]. This suggested more investigations
into account the effects of the ground acceleration and the relative
toward a better comprehension of the problem, especially in real
acceleration of the tanks with respect to ground [14]. When the
cases.
tank is placed upon support like RC columns or other type of sup-
Based on the above-depicted framework, this paper addresses
ports the dynamics is also influenced by the rotation at the tank
the problem of elevated tanks with particular attention paid on
base that has to be properly included in the equation of motions.
steel storage tanks placed on short RC column. The dynamic prob-
Nevertheless, this effect can be neglected when the lateral dis-
lem of elevated tanks subjected to seismic action is here briefly
placements of the tank base are small.
described and formalized. The base isolation of a real liquid natu-
On the contrary, the part, which depends on the deformability
ral gas (LNG) steel tank supported by reinforced concrete col-
of the tank wall, can be determined solving a fluidstructure inter-
umns is presented and discussed. The analyzed tank collapsed
action problem, whose solution depends on the geometrical and
during 1999 Itzmit earthquake for a premature shear failure of the
mechanical characteristics of the tank: radius R, liquid level H,
columns. First, the response of fixed base configuration is ana-
thickness s, liquid density q, and elastic modulus of steel E. The
lyzed, which will show the insufficient safety level against shear
problem can be uncoupled in infinite vibration modes, of which
failure of the columns. Subsequently, in order to reduce seismic
only few of them have a significant mass. Thus, the impulsive
mass is distributed among the first vibration modes of the wall
Contributed by the Pressure Vessel and Piping Division of ASME for publication
in the JOURNAL OF PRESSURE VESSEL TECHNOLOGY. Manuscript received July 4, 2014;
[14].
final manuscript received January 3, 2015; published online February 12, 2015. On the basis of the above observations, it can be drawn that the
Assoc. Editor: Chong-Shien Tsai. study of the hydrodynamic pressure in tanks subjected to a seismic

Journal of Pressure Vessel Technology Copyright V

C 2015 by ASME JUNE 2015, Vol. 137 / 031801-1

Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/ on 02/17/2015 Terms of Use: http://asme.org/terms

 Fig. 1 Equivalent springmass model of elevated tanks: (a)
general and (b) broad tanks
base motion can be easily performed using the simple model
shown in Fig. 1, in which the liquid mass is lumped and subdi-
vided in three components: rigid, impulsive, and convective
masses termed mi, mik (mass of kth mode of the wall vibration),
mck (mass of kth convective mode). The impulsive and convective
masses are connected to the tank wall by springs, respectively, of
stiffness kik and kck. The total pressure is given by adding the
effects of the mass mi subjected to the base motion acceleration,
of the masses mik subjected to the acceleration of the wall relative
to the bottom of the tank, and of the masses mck subjected to the
absolute acceleration.
In case of broad tanks, the model of Fig. 1(a) can be updated by
the simplest model shown in Fig. 1(b). In fact, the contribution of
the vibration modes of higher order is negligible and the entire
impulsive mass is practically equal to the mass of the first vibration
mode; moreover, because the distributions of the impulsive pres-
sure, with and without wall deformability, are almost coincident,
the effects of the impulsive action are simply taken into account by
the response in terms of absolute acceleration of a simple oscillator
of mass mi and stiffness ki. Neglecting the higher convective modes
effect, the model becomes a simple three degrees of freedom
(DOF) model (Fig. 2). The frequencies of the convective and im-
pulsive modes are generally very different (tenths of a second
against tens of seconds). This justifies the usual choice of neglect-
ing the interaction between these two phenomena.
Fig. 2 Lumped mass 3DOF model for nonisolated elevated
tanks
031801-2 / Vol. 137, JUNE 2015
Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/ on 02/17/2015 Terms of Use: http://asme.org/terms
Fig. 3 Lumped mass 4DOF model for isolated elevated tanks
The idea of seismic protection of tanks through base isolation
technique is not new. Starting to 1990 many works on this subject
have been done [15,16]. Unfortunately, few practical applications
have been realized [17] and a limited number of experimental
activities have been performed [14,18].
Based on the above observations a possible dynamic model of
elevated tanks upon base isolation is shown in Fig. 3.
The vibration period of the impulsive component of the pres-
sure generally falls in the maximum amplification field of the
response spectrum, whereas the convective period Tc is usually
very high and thus associated with a low amplification factor.
This implies a high effectiveness of the base isolation system,
which can reduce highly the base shear due to the impulsive pres-
sure component. Neglecting the influence of the lateral deforma-
tion of wall and support, the period of the isolated structure is
approximately given by
r
m i ms m b
Tiso  2p (1)
kiso
in which mi is the impulsive part of the liquid mass, ms and mb
are, respectively, wall and base tank masses, and kiso is the elastic
stiffness of the isolators.
Often the support structure is composed of short reinforced con-
crete columns and thus with a limited lateral deformability. Con-
sequently, Eq. (1) can be considered applicable in most of the
cases. In addition, because the elevated tanks are also very slen-
der, the convective motion can be very limited whose contribution
to the lateral pressure on the wall is also limited. The negative
effect of the sloshing is related only to the superficial motion,
because either the height of the wave can exceed the upper limit,
causing overtopping phenomenon, or the floating roof motion
could cause a breaking of the gaskets and the leakage of danger-
ous vapors of inflammable substances. Unfortunately, the base
isolation does not modify this phenomenon.
3 Description of the Case Study
As case study, an emblematic example of elevated tanks col-
lapsed during the Koaceli earthquake in 1999 at Habas pharma-
ceutics plant in Turkey has been considered [3,19].
The two damaged tanks on the left in Fig. 4 contained liquefied
oxygen, while the undamaged tank on the right contained lique-
fied nitrogen. Habas plant representatives on site reported that the
liquefied oxygen tanks were 85% full and the liquefied nitrogen
tank was about 25% full immediately before the earthquake. Each
tank consisted of two concentric stainless steel shells, one with an
Transactions of the ASME
 Fig. 4 Storage tanks of liquid oxygen at Habas plant after the Fig. 6 Section in the vertical plan of the tank
strong event of Itzmit (1999) (courtesy: The Karl V. Steinbrugge
slide and photograph collection world earthquakes and earth-
quake engineering). 4 Analysis of the Nonisolated Tank
4.1 Dynamic Characterization. In order to characterize
dynamically the tank, the modal analysis of the nonisolated case
outside diameter of 14.6 m, and the other with an outside diameter has been performed. This allows to identify the relative impor-
of 12.8 m (Fig. 5). The gap between the inner and outer shells was tance of the components of the liquid motion (impulsive, convec-
filled with perlite, which is a form of natural glass (foam) and a tive, and fluidstructure).
lightweight insulating material. At this end, masses and stiffness of Fig. 2 have been identified
The clear height of the tanks between the bottom slab and top (Table 1). In addition to impulsive and convective masses, the
stainless cover of the tanks is about 12 m (Fig. 6). Thus, the vol- mass of RC tank base, tributary mass of columns, and the mass of
ume of the tanks is approximately 1500 m3. All tanks were sup- the tank wall has been also considered, whose summation is indi-
ported on a 14.6 -m diameter, 1.07-mthick reinforced concrete cated in Table 1 as base.
slab that was in turn supported by sixteen 500-mm diameter rein- The convective and impulsive masses have been calculated
forced concrete columns. according to the prescriptions of EC8 part 4 [20]
Each column was 2.5 m in height and reinforced with sixteen
16-mm diameter longitudinal bars and 8-mm diameter ties spaced s
 
at approximately 100 mm on center. According to Ref. [3] the g H
xc 1:841 tanh 1:841 (2)
concrete used for the columns and the bottom slab was of class R R
C30/37, whereas the steel bars had a yielding strength of s
430 MPa. The density of the oxygen is 11.50 kN/m3. C Es
xi  (3)
H qR

While the expression of xc derives exactly from the previous

equations, the expression of xi is approximated, q is the fluid den-
sity, s is the thickness of the tank wall, and E is the young modu-
lus of the steel; the coefficient C depends on the ratio H/R and, for
usual values, varies between 0.17 and 0.13.
According to literature results [14], the damping ratio of impul-
sive and convective components of the motion have been chosen,
respectively, equal to 2% and 0.5%.
In Table 2, the results of the modal analysis obtained by using
the model of Fig. 2 are reported in terms of vibration periods and
participating mass of the first three periods. These results

Table 1 Dynamic parameters of the tank

Mass

Impulsive Convective Base

Mass (t) 1062 428 510

Damping ratio (n) 2.0 0.5 5.0
Period (T) 0.49 3.70 0.32
Fig. 5 Plan view of the tank

Journal of Pressure Vessel Technology JUNE 2015, Vol. 137 / 031801-3

Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/ on 02/17/2015 Terms of Use: http://asme.org/terms

 Table 2 Dynamic properties of nonisolated tank Table 4 Maximum base shear componentsnonisolated tank

Mode 1 2 3 Accelerogram Impulsive (kN) Convective (kN) Base (kN) Total (kN)

T (s) 3.57 0.56 0.16 Yarimca330 10892 982 1390 13264

MPM (%) 22 64 14 Yarimca060 13542 1753 2502 17796
Faith090 18062 100 2018 20079
Duzce270 10309 117 1596 11789
Table 3 Set of natural records used for the T-H analysis Duzce180 18573 541 2601 21715
Cekmece090 16534 291 3620 19863
Accelerogram Vs30 (m/s) Distance (km) PGA (g) Cekmece000 13011 91 3383 16303

Yarimca330 297 5 0.349

Yarimca060 297 5 0.268 1999 Koaceli earthquake and listed in Table 3. They have been
Faith090 228 55 0.159 selected according to the following hazard parameters: (a) Soil
Duzce270 276 15 0.358 type C, (b) S-waves velocity between 180 m/s e and 360 m/s, (c)
Duzce180 276 15 0.312 distance d from the fault: 10 < d < 30 km, and (d)
Cekmece090 346 67 0.133 0.20 g < PGA < 0.35 g.
Cekmece000 346 67 0.179 For example, the unscaled record of Yarimca 330 is depicted in
Fig. 7. It shows a typical behavior of a near-fault earthquake and a
significant length (arias intensity is up to 99%) particularly
highlight the practical independency of the convective motion of elevated.
the fluid (mode 1) with respect to the impulsive motion. Period The response spectra of all unscaled records are shown in Fig. 8
and modal participating mass ratio of the second mode, where the together with the elastic spectrum at ultimate limit state condition
motion of impulsive mass in mainly involved, are instead slightly [21]. The latter has been established based on the class of the
greater than the quantities indicated in Table 2; this is due to the structure (class IV) and consequently using an importance factor
presence of the base support. The increasing of the period can cI 1.6. Considering a PGA 0.22 g and the soil factor S 1.15,
have a positive effect when the ordinate of the acceleration spec- the PGA is equal to 0.4 g.
trum decreases as the period elongates, with a consequent reduc- In order to respect the compatibility conditions of the accelero-
tion of the inertia forces. grams with the design spectrum, each record has been first scaled
to minimize the mean square error and to obtain, at the fundamen-
tal period, a spectral ordinate equal to the design value.
4.2 Time-History Analysis. For the evaluation of the seismic For each accelerogram, a time-history analysis has been per-
response of the case study, time-history analyses were conducted formed using SAP2000 # software [22]. For each mass of Fig. 2,
using the simplified model of Fig. 2, subjected to a series of natu-
ral records selected among accelerograms recorded during the

Fig. 9 Nonisolated case: time-history of base shear compo-

Fig. 7 Accelerogram record of Yamarica (330 deg North) nents for Yarimca 330 record

Fig. 8 Response spectra of the seven unscaled accelerograms

031801-4 / Vol. 137, JUNE 2015 Transactions of the ASME

Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/ on 02/17/2015 Terms of Use: http://asme.org/terms

 Fig. 10 Average wall pressuresnonisolated case (a) total, (b) impulsive, and (c) convective
the maximum absolute acceleration has been obtained along with
the corresponding inertia forces, namely the base shear values
indicated in Table 4 for each accelerogram of Table 3. For exam-
ple, in Fig. 9, the response in terms of base shear forces to one of
the scaled accelerograms (Yamarica 330) is shown; in particular,
the total base shear and the relevant impulsive and convective
components are depicted. It can be noticed that:
the motion of the convective mass is practically independent
of the remaining masses as already confirmed by the modal
analysis;
in the steady-state condition the total base shear coincides
with the convective base shear, whose oscillations appear
less damped;
the impulsive mass undergoes higher accelerations than the
accelerations of the tank base, showing that the fluidstruc-
ture effect cannot be neglected;
the impulsive base shear is predominant with respect to the
convective one (see Table 4).
In addition to the base shear components, the distribution on
the tank wall of the impulsive, convective, and total fluid pressure
have also been obtained. They are expressed as function of the
maximum acceleration of impulsive and convective masses, as
suggested in Appendix A of EN1998:4 [20]. For example, for the
impulsive component the pressure reads
pi e; f; t Ci e; fq H cos h Ag t (4) using the model of Fig. 3, where Kiso 10,000 kN/m and
where Ci is a function of the normalized radius e r/R and height
f z/H, q is the density of the liquid, h is the angular coordinate
of the point in a cylindrical coordinate system, and Ag(t) is the
acceleration. A similar expression can be used for the convective
component. The pressure distribution calculated using the mean
value of the maximum accelerations is depicted in Fig. 10 for an
angle h 0 deg.
Using the maximum base shear, it has been possible to check
support columns and wall against failure. In particular, according
to the Turkish code [23], the maximum shear strength of the col-
umns is about 330 kN with respect to a maximum shear action of
about 1100 kN. This demonstrates the high vulnerability of these
tanks as dramatically shown during the 1999 Koaceli earthquake.
5 Analysis of Isolated Case: A Comparison Between
FPS and HDRB Isolators
The results of Sec. 4 suggested the use of a proper mitigation
technique as the base isolation. Given that the few applications
found in literature used either HDRB or friction pendulum, in
what follows their effectiveness in reducing the seismic response
of the case study of Fig. 4 is studied and compared.
In order to better understand the effectiveness of base isolation
systems in reducing the seismic response of elevated tanks, a
Journal of Pressure Vessel Technology
Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/ on 02/17/2015 Terms of Use: http://asme.org/terms
parametric analysis has been performed varying the parameters of
the models of Fig. 3, the base isolation period Tiso, the slenderness
c H/R pand the vibration period of the support structure
Ts  2p mtot =Ks The main results are presented in Ref. [24],
which confirm that both impulsive and convective components of
base shear, the minimum response is obtained for c 1, a part
from Tiso, even though the best results are obtained for Tiso 2.5/
3 s, which corresponds to a reduction of about 80% of impulsive
base shear and a slight increasing of the convective component.
It was also noticed that the isolation systems can be extremely
effective only for the impulsive components of the liquid motion
with a reduction of the base shear up to 80% with respect to the
nonisolated condition, whereas the convective motion is nega-
tively influenced with an increasing of base shear and liquid
height as already demonstrated in Ref. [14].
5.1 Seismic Response of Tank Isolated With HDRB. The
HDRB isolators were designed for an isolation period Tiso 2.5 s
and a damping ratio n 10%. Consequently, using Eq. (1) the
stiffness of the single device is K1iso 625 kN/m. Assuming a
transversal modulus G 0.6 MPa, each isolator has a diameter
D 450 mm and a total thickness of elastomeric layers
te 153 mm, the latter determined assuming a design shear defor-
mation c 100% and a design lateral displacement equal to
250 mm.
The modal analysis of the isolated tank has been performed
Miso 9.8 ton. The results are shown in Table 5. From these
results it can be drawn that:
The fundamental period of the tank, where the impulsive
motion was mainly involved, increased from 0.56 s to 2.34 s,
with a consequent reduction of the absolute accelerations.
The obtained period is not exactly equal to Tiso given the
dynamic interaction between the motion components.
The participating mass of the convective motion increases for
the high deformability of the isolators, for which it could
results in a possible interaction between convective and iso-
lated motion.
The convective period remains practically unchanged, dem-
onstrating how the isolation system does not affect the slosh-
ing motion.
The seismic analysis of the isolated tank has been performed
using the same accelerograms of Table 4. Anyhow, the
Table 5 Dynamic parameters of isolated tank (HDRB)
Mode 1 2 3
T (s) 4.10 2.34 0.28
MPM (%) 50% 49% 1
JUNE 2015, Vol. 137 / 031801-5
 Using a different elastic modulus (e.g., G 1.4 MPa), the
deformability can be reduced, but in any case it remains extremely
high. Consequently, in Sec. 5.2 a different type of isolators has
been proposed and investigated.

5.2 Seismic Response of Tank Isolated With FPS and

Comparison With HDRB Isolators. In this section, the seismic
response of the tanks isolated with FPS bearings is investigated.
For structures where the isolated mass is the entire weight, this
Fig. 11 Isolated case: time-history of base shear components typology represents an interesting solution because of the inde-
for Yamarica recordisolated case-HDRB
pendency of the response from the mass itself. In fact, in this case
the period can be calculated according to Eq. (5)
Table 6 Maximum base shear componentsisolated
tankHDRB r v v
M u M u 1
Tp 2p 2pu   2pu (5)
u
u  
Accelerogram Impulsive (kN) Convective (kN) Base (kN) Total (kN) Ke 1 l t 1 l

t
gM g
R X R X
Yarimca330 1795 1868 15 4498
Yarimca060 3040 2447 24 6767
Faith090 370 249 3 779 In case where the participating mass is a fraction of the total
Duzce270 1840 458 20 2454 mass, this independency is no longer valid. As a matter of fact, in
Duzce180 1402 1823 12 3880 case of storage tanks, the period is indicated in Eq. (6), where the
Cekmece090 695 118 7 1203 dependency of the mass is clearly shown. The greater the convec-
Cekmece000 605 124 5 1013 tive mass the lower the vibration period Tp. For example, in broad
tanks the sensibility of the response from the mass is higher than
in slender tanks, for which the convective mass is usually a small
scaling procedure has been performed considering as reference part of the total mass of the fluid
period Tiso. A new time-history analysis has been performed.
Figure 11 shows once again the response in terms of base shear r v
u
components. Analyzing the time histories it can be noticed that: M u Mimp Mss Miso
Tp 2p 2pu   (6)
the base and impulsive mass accelerations are very similar, Ke t 1 l
showing that the structure behaves rigidly during the motion; gMtot
R X
the convective motion is again independent from the rest of
the motion;
the convective shear force assumes values similar to those of In the analyzed case study, the convective/impulsive mass ratio
the nonisolated case, whereas the impulsive force decreases is about 50%, therefore the sensitivity of the response from the
up to 80%. mass cannot be considered totally negligible.
Considering a curvature radius R 3175 mm, a friction coefficient
This is clearly shown in Table 6 where the maximum force of of 3% and a maximum design displacement d 350 mm an equiva-
each mass is reported for each record. The mean total base shear lent period Teq 3 s and a damping ratio n 13.5% are obtained.
is about 2900 kN, which shows a reduction of about 80% with To perform time-history analysis, the simplified nonlinear
respect to the nonisolated case. model for FPS isolators already implemented in SAP2000 has been
The average pressure components acting on the wall are shown in used, which does not take into account the dependency of the
Fig. 12. The maximum impulsive pressure is now reduced to 9 kN/ response of the isolator from the vertical load variation. In the
m2 from about 90 kN/m2. This means that in all columns the shear present application, the friction coefficient of the isolators has
force demand is highly reduced, obtaining now values well under been assumed as constant. The effect of its variability with the
the shear strength limit. Similarly, the convective pressure increases, velocity will be investigated in further analysis.
remaining, in any case, confined within a range of low values. Figure 13 shows, for example, the response of the isolated tank
Finally, all necessary checks according to Eurocode 8 have to the Yamarica 330 record. From the results it can be drawn that:
been performed. Given that the average value of the maximum
displacement of HDRB isolators is 0.28 m and considering a The base and impulsive mass accelerations are very similar,
safety coefficient c 1.2, [21], the displacement limit of the devi- showing that, as in the case of HDRB isolators, the structure
ces corresponding to a shear deformation of 200% is attained. behaves rigidly during the motion.

Fig. 12 Average pressures on the wallisolated caseHDRB: (a) total, (b) impulsive, and (c)
convective

031801-6 / Vol. 137, JUNE 2015 Transactions of the ASME

Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/ on 02/17/2015 Terms of Use: http://asme.org/terms

 Fig. 13 Isolated case: time-history of base shear components
for Yamarica recordisolated case-FPS

Table 7 Maximum base shear componentsisolated tankFPS

isolators

Accelerogram Imp (kN) Conv (kN) Base (kN) Total (kN)

Yarimca330 1256 982 524 2772

Yarimca060 2201 1997 831 5045
Faith090 693 22 73 790
Duzce270 1182 69 117 1370
Duzce180 1232 681 492 2414
Cekmece090 857 207 90 973
Cekmece000 813 66 140 1021

The impulsive mass acceleration is extremely reduced with

respect to the nonisolated case, showing the high effective-
ness of the FPS isolation system as well.
The amplitude of the convective base shear seems to be
reduced with respect to the HDRB isolators. This is probably
due to a higher damping ratio (13.5%).
The maximum total base shear is now reduced of about 88%.
This is justifiable by the increasing of both equivalent period
and damping ratio.

Fig. 16 Membrane tensions: (a) HDRB and (b) FPS

In Table 7, the maximum base shear is reported together with

Fig. 14 Cyclic response of an FPS isolatorDuzce270 record
Fig. 15 Average pressures on the wallisolated caseFPS
isolators
the maximum values of convective and impulsive shear forces. In
this case an average value of base shear of 2055 kN is obtained.
This value is a little bit lower than the one obtained with HRDB
isolators.
To show the high dissipation capability of the FPS bearings
activated during the motion, the cyclic response for Yamarica330
record is shown in Fig. 14.
As for HDRB isolators the level of pressure acting on the wall
during the earthquake is extremely reduced, as depicted in
Fig. 15, where the average value of impulsive and convective
components along the height are shown.
The average value of the maximum displacements of FPS isola-
tors is about 290 mm that is compatible with most of the commer-
cial FPS bearings. This value is similar to the displacement
obtained for HDRB isolators, with the difference that now it is
fully compatible with the FPS isolation systems.
The stress level in the wall in both the cases is extremely lim-
ited. In particular, according to the EC8 part 4 [19], the stresses
corresponding to the elephant foot buckling and elastic buckling
condition are, respectively, 75 MPa and 104 MPa, whereas the
maximum stress obtained from the T-H analysis is about

Journal of Pressure Vessel Technology JUNE 2015, Vol. 137 / 031801-7

Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/ on 02/17/2015 Terms of Use: http://asme.org/terms

 20.8 MPa and 17.4 MPa for HDRB and FPS isolators, respec-
tively. The distribution of the vertical stress in the wall in both the
cases is shown in Fig. 16.
Finally, the maximum vertical displacement due to the sloshing
phenomenon is equal to 1.79 and 1.13 m for HDRB and FPS isola-
tors, respectively. This is fully compatible with the limit of 1.80 m
imposed by the geometry. The lower value obtained using FPS
isolators shows again their good performance.
6 Conclusions
In this paper, the effectiveness of two types of isolation devices
in reducing the seismic response of elevated tanks has been inves-
tigated. In particular, HDRB and FPS bearings have been
analyzed.
An emblematic example of elevated tanks collapsed during the
Koaceli earthquake in 1999 at Habas pharmaceutics plant in Tur-
key has been considered as case study: a group of three elevated
tanks, containing liquefied oxygen and liquefied nitrogen. The two
tanks with stored liquefied oxygen collapsed for shear failure of
the reinforced concrete short columns that sustained the tank. The
time-history analysis conducted using a lumped mass model dem-
onstrated the high demand in terms of base shear required to the
support columns and their inevitable collapse due to the insuffi-
cient shear strength.
Consequently, the design of the recalled isolation systems was
conducted and the response of the systems was investigated
though time-history analysis using mass lumped models properly
modified to account for the presence of the isolation system. In
particular, for HDRB an equivalent model was adopted, whereas
the FPS bearing where modeled with a simplified nonlinear model
in which the influence of the vertical load on the later stiffness
was neglected.
The capability of both the isolation systems in reducing both
base shear and stress level in the wall has been clearly demon-
strated. In particular the reduction of about 80% and 88% of the
total and impulsive base shear has shown the high effectiveness of
both the isolators. The use of HDRB was limited by the maximum
displacement required by the seismic action that instead is fully
compatible with the FPS isolators. The reduction of the stress
level eliminates any possible buckling phenomenon in the wall.
Finally, given the lower value of vertical sloshing displace-
ments of the liquid obtained using FPS isolator, it is suggested the
adoption of sliding bearings rather than HDRB in seismic isola-
tion of elevated tanks.
Acknowledgment
This work has been partially funded by the Italian RELUIS con-
sortium within the executive research program 20102013
research thrust twoSpecial Structures. Any opinions, findings
and conclusions or recommendations expressed in this paper are
those of the authors and do not necessarily reflect those of
RELUIS.
Nomenclature
Cc damping coefficient of the convective mode
Ci damping coefficient the impulsive mode
Cs damping coefficient the support system
E longitudinal elastic modulus of the wall
g gravity acceleration
G transversal elastic modulus of elastomer
H liquid height
kck stiffness of the kth convective mode
kik stiffness of the kth mode of the wall vibrations
kiso stiffness of the isolation system
mi impulsive mass
Mb mass of the tank base
Ms mass of the support system
031801-8 / Vol. 137, JUNE 2015
Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/ on 02/17/2015 Terms of Use: http://asme.org/terms
Mw mass of the tank wall
mck mass of kth convective mode
mik mass of kth mode of the wall vibrations
Miso mass of the isolation system
PGA peak ground acceleration
R tank radius
Tc vibration period of convective motion
Ti vibration period of impulsive motion
Tiso vibration period of the isolated structure
n damping ratio
q liquid density
References
[1] Paolacci, F., Giannini, R., and De Angelis, M., 2013, Seismic Response Miti-
gation of Chemical Plant Components by Passive Control Systems, J. Loss
Prev. Process Ind., 26(5), pp. 879948.
[2] Drosos, J. C., Tsinopoulos, S. V., and Karabalis, D. L., 2005, Seismic
Response of Spherical Liquid Storage Tanks With a Dissipative Bracing Sys-
tem, Proceedings of the 5th GRACM International Congress on Computational
Mechanics, Limassol, Cyprus, June 29July 1, pp. 313319.
[3] Sezen, H., Lvao glu, R., and Do
gangun, A., 2008, Dynamic Analysis and Seis-
mic Performance Evaluation of Above-Ground Liquid-Containing Tanks, Eng.
Struct., 30(3), pp. 794803.
[4] Shrimali, M. K., and Jangid, R. S., 2003, Earthquake Response of Isolated Ele-
vated Liquid Storage Steel Tanks, J. Constr. Steel Res., 59(10), pp.
12671288.
[5] Abali, E., and Uckan, E., 2010, Parametric Analysis of Liquid Storage Tanks
Base Isolated by Curved Surface Sliding Bearings, Soil Dynamics and Earth-
quake Engineering, 30(12), JanuaryFebruary, pp. 2131.
[6] Tsai, C. S., Chiang, T. C., and Chen, B. J., 2005, Experimental Evaluation for
Piecewise Exact Solution of Seismic Responses of Spherical Sliding Type Iso-
lated Structures, Earthquake Eng. Struct. Dyn., 34(9), pp. 10271046.
[7] Tsai, C. S., Chiang, T. C., and Chen, B. J., 2006, Component and Shaking
Table Tests for Full Scale Multiple Friction Pendulum System, Earthquake
Eng. Struct. Dyn., 35(13), pp. 16531675.
[8] Marioni, A., 1998, The Use of High Damping Rubber Bearings for the Protec-
tion of the Structures From the Seismic Risk, Jornadas Portuguesas de Engen-
haria de Estruturas, LNEC, Lisboa, pp. 2528.
[9] Santagelo, A., Scibilia, N., and Stadarelli, R., 2007, Seismic Isolation of a
Tanks at Priolo Gargallo (in Italian), Proceedings of the Giornate AICAP
2007, Salerno, Italy, Oct. 46, pp. 543550.
[10] Housner, G. W., 1963, The Dynamic Behaviour of Water Tanks, Bull. Seis-
mol. Soc. Am., 53, pp. 381387.
[11] Fischer, D., 1979, Dynamic Fluid Effects in Liquid-Filled Flexible Cylindrical
Tanks, Earthquake Eng. Struct. Dyn., 7(6), pp. 587601.
[12] Haroun, M. A., and Hausner, G. W., 1981, Earthquake Response of Deforma-
ble Liquid Storage Tanks, ASME J. Appl. Mech., 48(2), pp. 411418.
[13] Veletsos, A. S., and Tang, Y., 1987, Rocking Response of Liquid Storage
Tanks, J. Eng. Mech., 113(11), pp. 17741792.
[14] De Angelis, M., Giannini, R., and Paolacci, F., 2010, Experimental Investiga-
tion on the Seismic Response of a Steel Liquid Storage Tank Equipped With
Floating Roof by Shaking Table Tests, Earthquake Eng. Struct. Dyn., 39(4),
pp. 377396.
[15] Wang, Y., Teng, M., and Chung, K., 2001, Seismic Isolation of Rigid Cylin-
drical Tanks Using Friction Pendulum Bearings, Earthquake Eng. Struct.
Dyn., 30(7), pp. 10831099.
[16] Shrimali, M. K., and Jangid, R. S., 2002, Non-Linear Seismic Response of
Base-Isolated Liquid Storage Tanks to Bi-Directional Excitation, Nucl. Eng.
Des., 217(12), pp. 120.
[17] Tajirian, F. F., 1998, Base Isolation Design for Civil Components and Civil
Structures, Proceedings of the Structural Engineers World Congress, San Fran-
cisco, CA, pp. 233244.
[18] Calugaru, V., and Mahin, S. A., 2009, Experimental and Analytical Studies of
Fixed Base and Seismically Isolated Liquid Storage Tanks, Proceedings of the
3rd International Conference on Advances in Experimental Structural Engineer-
ing, San Francisco, CA, Oct. 16, pp. 112.
[19] Paolacci, F., 2014, On the Effectiveness of Two Isolation Systems for the
Seismic Protection of Elevated Tanks, ASME Paper No. PVP2014-28563.
[20] EN 1998-4, 2006, Eurocode 8: Design of Structures for Earthquake
ResistancePart 4: Silos, Tanks and Pipeline, EN 1998-4, Brussels, Belgium.
[21] CEN; 2006, Eurocode 8: Design of Structures for Earthquake Resistance
Design of Structures for Earthquake Resistance, Part 4: Silos, Tanks and Pipe-
line. European Committee for Standardisation, Brussels, Belgium.
[22] CEN; 2014, Eurocode 8: Design of Structures for Earthquake Resistance, Part
1: General Rules, Seismic Actions and Rules for Buildings. European Commit-
tee for Standardisation, Brussels. Belgium.
[23] Turkish Seismic Code, 2007, Specification for Structures to be Built in Disas-
ter Areas, Ministry of Public Works and Settlement Government of Republic
of Turkey.
[24] Paolacci, F., Giannini, R., Uckan, E., Akbas, E., and Corritore, D., 2014,
Seismic Response Mitigation of Elevated Tanks by HDRB and FPS Isolation
Systems, Proceedings of the Second European Conference on Earthquake
Engineering and Seismology, 2ECEES, Istanbul, Turkey, Aug. 2529,
(CDROM).
Transactions of the ASME