You are on page 1of 18

Soil Dynamics and Earthquake Engineering 141 (2021) 106491

Contents lists available at ScienceDirect

Soil Dynamics and Earthquake Engineering


journal homepage: http://www.elsevier.com/locate/soildyn

Liquefied natural gas storage tank simplified mechanical model and seismic
response analysis
Dongyu Luo a, b, Chunguang Liu a, b, Jiangang Sun c, *, Lifu Cui c, Zhen Wang c
a
Institute of Earthquake Engineering, Dalian University of Technology, Dalian, China
b
School of Hydraulic Engineering, Dalian University of Technology, Dalian, China
c
College of Civil Engineering, Dalian Minzu University, Dalian, China

A R T I C L E I N F O A B S T R A C T

Keywords: In this study, a scale model of a liquefied natural gas (LNG) storage tank was designed to carry out a shaking table
LNG storage Tank test, the main position of insulation layer on tank and the site effect were investigated under different ground
Insulation layer motions. The test results show that there is an obvious interaction between the insulation layer and the liquid-
Liquid-solid coupling
solid coupling position of the storage tank, and the seismic response of the tank was related to the frequency of
Simplified mechanical model
Site effect
the ground motions, the site effect should be considered when analysing the seismic response of liquid-solid
coupling structures. Based on the test results, a simplified mechanical model for the seismic design of LNG
storage tanks that considered the insulation layer was constructed, and the simplified mechanical model was
verified via a numerical simulation method. Moreover, based on the results of the 16 × 104 m3 LNG storage tank
simplified mechanical model, the base shear and bending moment were reduced by 20%–25% on average, the
acceleration and displacement of the inner tank wall were reduced by 40% on average after filling the insulation.
In the seismic design of large scale LNG storage tanks, considering the damping effect of insulation is helpful to
save design capital, and ignoring the damping effect of insulation will make the design conservative.

1. Introduction deduced a simplified mechanical model of full-capacity LNG storage


tanks. Numerical simulations have been used by many researchers to
16 × 104 m3 full containment liquefied natural gas (LNG) storage analyse the seismic responses of storage tanks; topics covered have
tanks require a special structure to store ultra-low-temperature media. included the pile-soil-tank interaction [12], the seismic responses of the
They consist of a steel inner tank, which is the main container, a pre- inner and outer tanks [13], a seismic vulnerability of the tank [14], the
stressed reinforced concrete outer tank can be considered as a second­ effect of near-fault ground motion on a tank [15], the sloshing effect of
ary vessel if the failure of the inner tank, and resilient 0.3 m thick felt the liquid [16], etc. Using the shaking table test, researchers have ana­
and expanded perlite powder with a thickness of 0.7 m is arranged in lysed the seismic performance of steel storage tank or rectangular stor­
between the inner and outer tanks as the insulation layer [1–3]. A high age tank [17–19]. However, as it is difficult to perform the shaking table
seismic capacity of such storage tanks is required because of the leakage test on LNG tanks owing their special structure, only the inner tank shell
of LNG may lead to explosion or fire; therefore, carrying out seismic was modeled in the mock-up [20]. Hence, there are limited shaking
analyses of these storage tanks is crucial [4]. table test results available for LNG storage tanks.
Many assumptions and simplifications were made to deduce a me­ The previous research mentioned above can provide strong support
chanical model of LNG storage tanks. For vertical storage tanks, Housner for the design and construction of LNG storage tanks, but the role of the
[5] used the two-lumped-mass model with the assumptions that the tank insulation was neglected in the aforementioned analyses. Over time, the
wall was absolutely rigid and that the liquid behaved as an ideal fluid. insulation becomes dense [21]; therefore, the coupling effect that the
Veletsos et al. [6,7] and Haroun et al. [8,9] later modified the model by insulation provides between the inner and outer tanks cannot be
using the three-lumped-mass model after taking into consideration the neglected in the event of an earthquake. By analysing the explosion of an
elastic deformation of the tank wall. Subsequently, Zhang et al. [10,11] LNG tank, it was determined that the annular insulation transfers load
extended the vertical storage tank model to LNG storage tanks and between the tank walls, enhancing their individual strengths [22].

* Corresponding author.
E-mail addresses: sjg728@163.com, sjg@dlnu.edu.cn (J. Sun).

https://doi.org/10.1016/j.soildyn.2020.106491
Received 26 March 2020; Received in revised form 4 August 2020; Accepted 1 November 2020
Available online 12 November 2020
0267-7261/© 2020 Elsevier Ltd. All rights reserved.
D. Luo et al. Soil Dynamics and Earthquake Engineering 141 (2021) 106491

Zhang et al. [23,24] considered the insulation when conducting their The time similarity ratio is given by St = tp /tm . The period similarity
seismic analysis of LNG storage tanks using numerical simulations. Be­ ratios of the impulse and sloshing components of the liquid are
sides, relative researches indicated that the seismic response of storage considerably different [29], the sloshing period and liquid-solid
tank is the largest at the full tank, and the effect of insulation on storage coupling period of the model tank differ substantially from those of
tank is not related to ground motion intensity [25,26]. In this study, the the prototype tank owing to the excessive geometric similarity ratio.
seismic response of full level tank under different ground motions was According to the GB50341 code [30], the sloshing period of the proto­
analysed, the main position of the tank affected by the insulation under type tank and scale model were 9.58 s and 1.90 s, respectively, and the
the earthquake action was explored, which provided the basis for the liquid-solid coupling period of the prototype tank and scale model were
simplified mechanical model of the LNG storage tanks. Based on the test 0.50 s and 0.02 s, respectively. The design similarity ratio of test model
results, a simplified mechanical model for the seismic design of LNG considering liquid-solid coupling interaction was given in Ref. [31],
storage tanks was constructed, which considered the insulation, and its which is more suitable for liquid-solid coupling structures. In this study,
rationality was verified via a numerical simulation method. Moreover, the time similarity ratio was 5.2 if according to the similarity relation­
the influence of site effect on seismic response of storage tank was also ship given in Ref. [31]. The minimum time step that the shaking table
analysed. could output was 0.024 s, if the seismic records were scaled on the basis
of the time similarity ratio of 5.2, the loading range of the shaking table
2. Shaking table test investigation would be exceeded. Thus, the time similarity ratio was adjusted [29,32].
The time step of the original ground motion used in this study was 0.1s.
2.1. Test design description In order to compress the ground motion to the maximum extent, the
output time step of seismic wave was set to 0.024s, thus, the time sim­
The unidirectional horizontal-displacement earthquake shaking ilarity ratio was 4.17, which is close to the time similarity ratio ac­
table at Dalian Minzu University was used. It was a servo hydraulic cording to Ref. [31].
system and the specific parameters were as follows: the table size was The shaking table used in this study vibrates in the north-south di­
3.00 m × 3.00 m; the displacement limit ±80 mm; the maximum load rection, the measuring points of the model tank are shown in Fig. 2, the
50 t; and the frequency ranged from 0.1 Hz to 50 Hz. The scaled test installation of test model and sensors are shown in Fig. 3. For the
model was based on a 16 × 104 m3 prototype LNG storage tank, a insulation layer, only the expanded perlite was used; it is usually used in
structural diagram of the tank is shown in Fig. 1. practical engineering as an infill. Each layer was filled and tamped by
The scale design idea and method of the test model can refer to vibration. The inner and outer tanks were sealed with a plastic film for
Ref. [25–28]. Here, the test design can be summarized as follows: The waterproofing as the weight of expanded perlite can reach 4–9 times its
whole test model floated on the shaking table, and the inner tank was original weight after the absorption of water [33].
designed to be anchored on the bottom plate. The side length of the Site can be divided into four categories according to the equivalent
shaking table was 3 m, the maximum outside diameter of the prototype shear wave velocity of soil [34]. In this study, one ground motion was
cylindrical tank was 82 m. According to the proportion of shaking table selected as the seismic load in each of the four types of sites. The records
and prototype tank, the geometric similarity ratio was set as 27.3, from name and the site classification are show in Table 1. After Fourier
which the geometry size of the tank was determined, while the wall transform of ground motions, the predominant frequencies of these four
thickness of the test model designed according to the similarity ratio original seismic waves range from 0.78 Hz to 3.88 Hz. According to the
makes the tank wall too thin to be processed, hence, the tank wall similarity ratio given in Section 2.1, the predominant frequencies of the
thickness was appropriately enlarged. According to the geometrical compressed seismic waves range from 3.23 Hz to 16.71 Hz. Table 1 is the
similarity ratio, the model tank was made with a diameter of 3 m, inner detailed information of the compressed seismic records. Fig. 4 (a) shows
and outer tank heights of 1.31 m and 1.43 m, respectively, the design the acceleration time history of original seismic wave, the peak accel­
liquid level of full tank was 1.24 m, the thickness of the outer tank was eration of the compressed seismic wave is adjusted to 3 m/s2 in this test.
0.03 m. According to the processing conditions of the test model, the Fig. 4 (b) shows the displacement waves, which are used as the inputs in
wall thickness of the inner tank was 0.001 m. In addition, if the space the shaking table test.
between the inner and outer tanks was designed according to the geo­
metric similarity ratio, this space should be 0.037 m. However, in order 2.2. Discussions
to install the sensors, the final spacing similarity ratio was taken as Sl =

5, this expansion of the spacing would have amplified the influence of 2.2.1. Effect analysis of insulation
the insulation. In this way, the increase of tank wall thickness can just The results in Figs. 5 and 6 show that the acceleration from bottom to
make up for this defect. Moreover, water was used in the place of LNG. top of tank walls decrease after the insulation is filled, the average
damping rate of the acceleration of the inner and outer tank walls are
13.45%–33.67% and 5.36%–20.34%, respectively. Acceleration is an
important index in seismic design of storage tank, which can affect the
base shear and bending moment. If the tank can be simplified as multiple
lumped mass point, the base shear and bending moment can be
expressed by Eqs. (1) and (2), according to the acceleration of the tank, it
can be indirectly concluded that the base shear and bending moment
will be reduced after filling the insulation.
Q(t) = m1 ẍ1 (t) + ...mn ẍn (t) (1)

M(t) = m1 ẍ1 (t)h1 + ... + mn ẍn (t)hn (2)


When there is no insulation, the inner tank wall receives pressure
from water; after the insulation is filled, the inner tank wall receives the
passive pressure from the insulation. The relationship between the
passive pressure of the insulation on the inner tank wall and radial
displacement of the tank wall is shown in Eq. (3) [35]. Where, Ep is the
Fig. 1. Structural diagram of prototype tank. Elastic modulus of the insulation, dp is the width of the insulation, w is

2
D. Luo et al. Soil Dynamics and Earthquake Engineering 141 (2021) 106491

Fig. 2. Measuring points setting of test model.

Fig. 3. Arrangement of transducers in the model tank.

Table 1
Information of compressed seismic records.
Seismic record name Site classification Duration [s] Predominate frequency [Hz] Peak displacement [mm] Time step [s]

Golden Gate Prak I 9.552 16.71 − 14.867 0.024


El Centro II 12.888 9.00 12.159
Taft III 13.056 6.82 − 13.727
Pasadena IV 18.456 3.23 10.194

3
D. Luo et al. Soil Dynamics and Earthquake Engineering 141 (2021) 106491

Fig. 4. Original acceleration wave and compressed displacement wave.

Fig. 7. Relative displacements of double layer vessel.


Fig. 5. Acceleration response of inner tank wall.

Fig. 8. Passive pressure of insulation.

Fig. 6. Acceleration response of outer tank wall.

the radial displacement of the insulation, Pp is the passive pressure. The


test results in Fig. 7 show that the maximum deformation of the inner
tank wall occurs in the medial position (liquid-solid coupling position)
under most test cases, and the insulation can effectively control the
deformation at this position. Based on Eq. (3), the passive pressure of the
insulation on the inner tank wall can be obtained, the value of the
passive pressure is marked in Fig. 7, the results show that the maximum
passive pressure is at the liquid-solid coupling position. As shows in
Fig. 8, the direction of the passive pressure is opposite to that of the
hydrodynamic pressure, so the resultant force on the tank wall becomes
smaller, which has a positive influence on the seismic performance of
the internal vessel. This result has a guiding significance for the Fig. 9. Displacement of the outer tank wall.

4
D. Luo et al. Soil Dynamics and Earthquake Engineering 141 (2021) 106491

derivation of simplified mechanical model of LNG storage tank. More­ the wave heights increase when the tank is subjected to the Golden Gate
over, Fig. 9 shows that the deformation of the upper part of the outer Park and Pasadena waves, while the sloshing wave heights decrease
tank increases after the insulation is filled, which indicates that the under the action of El Centro and Taft waves. Even if the same seismic
insulation can squeeze the top of the outer tank wall, resulting in a slight wave is applied to the shaking table at each loading, there will be
increase in displacement. slightly different ground motion outputs from the table, which affects
the sloshing wave heights. The results in Figs. 5–9 show that the seismic
Ep
PP,E
P (z) = w(θ, z, t) (3) response of two layer tanks are larger under the Golden Gate Park and El
dp
Centro waves, but the results in Fig. 15 show that the largest wave height
The peak values of the compressive strain of inner and outer tank is under Taft wave. Radnić et al. [37] indicated that the size of the tank is
walls are analysed in this test. Axial stress is an important parameter in a main factor affecting the initial free oscillation period of the system.
the tank design [36]. The relationship between axial strain and bending When the tank is exposed to a base excitation whose period is close to
moment can be expressed by Eq. (4). Where, M is the bending moment, the first period of the system, resonant motion of the water in the tank
εa is the axial strain, E is the Elastic modulus of tank wall, R and hs are occurs, and this resonance may have been the cause of the violent
the radius and thickness of the tank wall, respectively. According to the sloshing and the liquid overflow. Furthermore, the ground displacement
data in Fig. 10, the axial strains reduce after the insulation is filled. of the seismic waves also should be considered in analysis. It can be seen
Based on Eq. (4), it can be concluded that the overturning moment from Fig. 4 (b) that there is large displacement of Taft wave in the later
acting on the inner tank wall decreases after filling the insulation. part of the time history, which is equivalent to applying a displacement
Fig. 11 shows that the axial strain collected by most strain gauges of the pulse to the sloshing liquid, it will promote the sloshing of liquid and
outer tank increase. In the event of an earthquake, the impact of the cause more severe sloshing, hence, Fig. 16 shows that Taft wave excites
liquid on the inner tank wall will squeeze the insulation, which gradu­ larger sloshing at 12s than 9s. During an earthquake, the impact force of
ally becomes denser during the vibrations. This increases the stiffness of the violent sloshing of the liquid on the tank wall is huge, and the
the insulation, causing it to squeeze the outer tank wall, hence, the axial leakage of LNG is very dangerous because it may cause accidents such as
strain of outer tank tends to increase. explosions and fires. Therefore, the anti-sloshing design of storage tanks
should be further studied.
M = εa Eπ R2 hs (4)
The relationship between hoop strain and hydrodynamic pressure 2.2.2. Site effect analysis
can be expressed by Eq. (5). Where, P is the hydrodynamic pressure, εh is Figs. 5 and 6 show that the acceleration of two layer tanks reduce
the hoop strain. In Fig. 12, the hoop strain is usually the largest in the after considering the influence of the insulation, the acceleration
intermediate section of the tank wall, which shows that the hydrody­ response of the inner and outer tank walls are largest under El Centro
namic pressure of the test tank is largest at the height of liquid-solid wave and Golden Gate Park wave, respectively. The overall trend can be
coupling component. According to Eq. (5), the hydrodynamic pressure considered that both the acceleration of the inner tank and outer tank
decreases after filling the insulation in most test cases, which illustrates are greater when the tanks encounter to the earthquake action with
that the effect of insulation on the pressure of tank wall cannot be higher predominant frequency in this test. It can be seen from Fig. 5 that
ignored. Fig. 13 shows that the hoop strains of the outer tank wall are the reduction percentage of acceleration response of the inner tank is
amplified under most test cases, especially in the middle position of the less than 15% under the action of Taft wave, while the damping rate is
outer tank. Besides, the results in Figs. 12 and 13 show that the hoop greater under the action of the other three seismic waves, among which
strain is larger in the middle position of these two layer tank walls, the damping rate under the action of El Centro wave is the largest. The
indicating that the liquid-solid coupling effect is obvious. Fig. 14 shows outer tank also shows the same reaction trend, Fig. 6 shows the damping
the hoop strain of the middle position of the inner and outer tanks in rate of the outer tank, is can also be concluded that the damping rate is
each direction. The results indicate that after the insulation is filled, the smallest under the Taft wave. The above results confirm that Golden
hoop strain at the liquid-solid coupling position of the inner tank de­ Gate Park wave and El Centro wave stimulate greater acceleration, and
creases, while the hoop strain at this position of the outer tank increases. their damping ratio is also larger. Besides, Figs. 5 and 6 shows that the
Insulation will relieve the pressure of the liquid on the inner tank wall, reduction in the acceleration amplification factor of the inner tank is
but at the same time, it will be squeezed by the inner tank wall and cause greater than the outer tank under most seismic actions except for Golden
greater pressure on the outer tank wall. This reflects the coupling effect Gate Park wave.
between the double layer tank and the insulation at the liquid-solid In terms of displacement under different ground motions, Fig. 7
coupling position. Combined with the tank wall displacement shows that the relative displacement varies greatly under different
response, it can be concluded that the effect of the insulation on the ground motions when the insulation is not filled, while displacement
liquid-solid coupling position of the storage tank should not be ignored. values caused by different ground motions are similar after the insu­
lation is filled. In general, the displacement is larger when the tank is
εh Ehs
P= (5) encounter to the Golden Gate Park and El Centro waves. The above re­
R sults show that the deformation of these two layer tanks is larger under
Fig. 15 is the time history of the liquid sloshing. With the insulation, the earthquake action with higher predominant frequency. Fig. 9 shows

Fig. 10. Axial strains collected by SA16-SA39 strain gauges on inner tank wall.

5
D. Luo et al. Soil Dynamics and Earthquake Engineering 141 (2021) 106491

Fig. 11. Axial strains collected by SA1-SA15 strain gauges on outer tank wall.

Fig. 12. Hoop strains collected by SH16-SH39 strain gauges on the inner tank wall.

Fig. 13. Hoop strains collected by SH1-SH15 strain gauges on the outer tank wall.

Fig. 14. Hoop strain in each direction with and without insulation.

that the deformation of the upper part of the outer tank wall increases under different ground motions are as follows: In this test, the liquid-
obviously after filling the insulation, especially under the action of the solid coupling component has a great influence on the storage tank,
seismic waves with low predominant frequency. the liquid-solid coupling period of the model tank is 0.02 s, the pre­
The reasons for the difference in seismic response of the test model dominant period of the Golden Gate Park and El Centro waves are close

6
D. Luo et al. Soil Dynamics and Earthquake Engineering 141 (2021) 106491

Fig. 15. Sloshing wave heights under the four different seismic waves.

Fig. 16. Liquid sloshing observed under the Taft wave.

to the liquid-solid coupling period, if the natural vibration period of the seismic records.
tank is close to the predominant period of ground motion, a greater
response of the structure will be stimulated. Therefore, the seismic 3. Simplified mechanical model of LNG storage tanks
response of the test model under Golden Gate Park and El Centro waves
are larger than other seismic waves. The outer tank of the 16 × 104 m3 LNG storage tank is composed of a
Based on the test results, the insulation has a great inhibitory effect sidewall and dome, the sidewall is regarded as a shear cantilever beam
on the seismic response of the middle position of the tank wall. In the in the simplification. Fig. 17 shows the simplified model of the sidewall,
simplification of liquid, the intermediate position is considered as the where mb and hb are the equivalent mass and height, respectively, and cb
liquid-solid coupling component, which vibrates with the intermediate are the stiffness and damping coefficients, respectively. The dome was
section of the inner tank wall [38–40]; therefore, the insulation can be connected to the sidewall and had no contact with the insulation;
simplified as a spring-damper that connects the sidewall and the therefore, it could be regarded as a concentrated mass acting on the top
liquid-solid coupling component. Besides, the seismic performance of of the sidewall, where is the total mass, the stiffness and damping co­
the tank will be greatly affected by the predominant frequency of ground efficients areaes, respectively. The simplified model of the outer tank is
motions, the number of ground motions selected in the test is limited. shown in Fig. 18.
Hence, in the following study, the influence of ground motion frequency The lateral displacement of the outer tank is given by Eq. (6):
on the seismic response of storage tanks was further analysed using more

7
D. Luo et al. Soil Dynamics and Earthquake Engineering 141 (2021) 106491

( )
1 1 1 1 1 1
= + + + + 2 (12)
fot2 fF2 fS2 fR2 2
fcF fcS

where fF is the engineering frequency of the bending cantilever beam, fS


is the engineering frequency of the shear cantilever beam, fR is the en­
gineering frequency of the cantilever beam based on a series of inde­
pendent ring assumptions, fcF is the engineering frequency of a bending
cantilever beam with a concentrated mass at the end of the rod, and fcS is
the engineering frequency of a shear cantilever beam with a concen­
trated mass at the end of the rod. The calculation of the frequency co­
efficients is given by Eqs. (13)–(17) [10].
Fig. 17. Simplified model of the sidewall. √̅̅̅̅̅
0.2Dc Ec
fF = 2 (13)
L ρc
√̅̅̅̅̅
1 Ec
fS = √̅̅̅̅̅̅̅̅̅̅̅̅̅ (14)
8L 1 + υc ρc
√̅̅̅̅̅
1 Ec
fR = √̅̅̅̅̅̅̅̅̅̅̅̅̅ (15)
Dc π 1 − υ2c ρc
√̅̅̅̅̅̅̅
1 KcF
fcF = (16)
2π md
( )3
Dc
Fig. 18. Simplified model of the outer tank. kcF = 3πEc tc (16a)
2L
(πz) √̅̅̅̅̅̅
x(z, t) = sin ϕ(t) (6) 1 kcS
2h fcS = (17)
2π md
Assuming that the mass of a unit height of the sidewall is m, the shear
force and bending moment under the horizontal motions of an earth­ π Dc
quake can be calculated by Eqs. (7) and (8), respectively. The equivalent kcS = Ec tc (17a)
4(1 + υc ) L
mass and height of the sidewall are obtained from the two following
equations: where Dc is the medium diameter of the outer tank, L is the height of the
∫h ( πz) ( ) outer tank wall, md is the dome mass, tc is the thickness of the outer tank
πhb
m sin dz = mb sin (7) wall, and Ec , vc and ρc are the elastic modulus, the Poisson’s ratio, and
2h 2h
0
the density of the concrete, respectively.
∫ h ( πz) ( ) Then, the total stiffness of the outer tank can be calculated by Eq.
π hb
m sin zdz = mb sin hb (8) (18), ωot can be calculated from fot .
0 2h 2h
k = mωot 2 (18)
The stiffness of the sidewall is calculated according to the method of
the equivalent stiffness calculation of a shear wall with small opening. Finally, the stiffness of the dome can be calculated according to Eq.
Eq. (9) is the wall displacement under a concentrated load, and the (19), where the damping coefficient of the dome is calculated according
reciprocal of the displacement under a unit load is the equivalent stiff­ to Rayleigh damping.
ness coefficient kb (H = hb ).
k × kb
( ) k* = (19)
PH 3 3μEI kb − k
u = 1.2 × 1+ 2
(9)
3EI GAH
The liquid can be simplified as three components, the total velocity
potential of the liquid can be obtained from the sum of the velocity
where EI is the bending rigidity, GA is the shearing rigidity, H is the
potentials of these three components, as shown in Eq. (20), and the
height of the wall, and μ is the form factor, which was taken as 2 (μ = 2)
hydrodynamic pressure acting on the tank wall can be calculated by Eq.
here.
(21). The base shear acting on the inner tank wall can be expressed by
Based on Eq. (9), the circular frequency of sidewall is given by Eq.
Eq. (22). During an earthquake, the extrusion of liquid on the inner tank
(10):
wall will produce a force on the insulation. At the moment, the insu­
√̅̅̅̅̅̅̅̅
1 lation serves as a passive pressure on the inner tank wall, which is
ωb = (10)
umb opposite to the action direction of the hydrodynamic pressure. Hence,
the resultant force acting on the inner tank wall can be expressed by Eq.
The damping coefficient of the sidewall can be calculated indirectly
(23).
from Eq. (11).
Φ(r, θ, z, t) = φ1 + φ2 + φ3 (20)
cb = 2ξb mb ωb (11)
The natural frequency of the outer tank can be calculated by Eq. (12) P(r, θ, z, t) = − ρL
∂Φ
(21)
[10], ∂t

8
D. Luo et al. Soil Dynamics and Earthquake Engineering 141 (2021) 106491

∫ HW ∫ 2π
Q(t) = RcosθPdθdz (22) where T and V are the kinetic energy and potential energy of the system,
0 0 respectively, and Wnc is the work done by non-conservative forces.
∫ ∫ ∫ ∫ π
According to the simplified mechanical model in Fig. 19 (b), T, V and
Hw 2π Hw

Q (t) = R cos θPdθdz − 2
2
R cos θPp dθdz (23) Wnc can be expressed as Eqs. 27–29:
0 0 0 − π2 ( )2 ( )2
1 1
T = m* ẋ* (t) + ẋg (t) + mc ẋc (t) + ẋg (t) +
where Φ is the total velocity potential of the liquid, ϕ1 , ϕ2 and ϕ3 are the 2 2
(27)
velocity potential of the three components of the liquid. ρL is the density 1
( )2
1
( )2
1
of the liquid. HW is the height of the liquid. Pp is the passive pressure
2
mi ẋi (t) + ẋg (t) + mb ẋb (t) + ẋg (t) + m0 ẋg (t)
2 2 2
caused by insulation.
Based on the test results, the insulation has a great interaction with 1 1 1 1
the middle position of the tank wall, hence, the passive pressure of V = kc xc (t)2 + ki xi (t)2 + kb xb (t)2 + kz (xi (t) − xb (t))2
2 2 2 2
insulation on inner tank wall is simplified as a spring-damper that 1
+ k* (x* (t) − xb (t))2 (28)
connects the sidewall and the liquid-solid coupling component. Fig. 19 2
(a) shows the simplified mechanical model without the insulation,
which is currently commonly used [10], Fig. 19(b) shows the modified W(nc = − cc ẋc (t)δx
) c (t) − ci ẋi (t)δxi (t) − (cb ẋb (t)δxb (t)−)
simplified mechanical model in which the insulation is considered. (29)
cz ẋi (t) − ẋb (t) δ(xi (t) − xb (t)) − c* ẋ* (t) − ẋb (t) δ(x* (t) − xb (t))
Where, mc , kc and cc are the mass, stiffness and damping coefficients of
the sloshing component, respectively; mi , ki and ci are the mass, stiffness Based on Eqs. 27–29, Eq. (30) is obtained. The base shear, bending
and damping coefficient of the liquid–solid coupling component, moment and sloshing wave height are shown in Eqs. 31–33,
respectively. m0 is the mass of the rigid pulse component. respectively.
Based on the formula above, the liquid-solid coupling lumped mass ⎧ ⎫
point is larger in mass and oscillates at a smaller frequency than the ⎡ * ⎤⎪ * ⎪
m 0 0 0 ⎪ ⎪ ẍ (t) ⎪
⎪ ⎪

sidewall; thus, the it can be considered as a large mass block moving in ⎢ 0 mb 0 0 ⎥
⎨ ⎬
⎢ ⎥ ẍb (t)
low-frequency motion. Therefore, the liquid-solid coupling lumped mass ⎣ 0 0 mc 0 ⎦⎪ ⎪ ẍ (t) ⎪
⎪ c ⎪ ⎪
point plays an integral role in the interactions between the two layer 0 0 0 mi ⎪ ⎩ ẍi (t) ⎪⎭
tanks and the insulation. The damping caused by the insulation can be ⎧ ⎫
⎡ * ⎤⎪ * ⎪
calculated using Eq. (24), and the stiffness coefficient of the insulation is c − c* 0 0 ⎪

⎪ ẋ (t) ⎪


⎨ ⎬
kz , it can be calculated using Eq. (25) [41]. ⎢ − c* c* + cb + cz 0
⎢ − cz ⎥
⎥ ẋb (t)
+⎣ ⎦
cz = 2ξmi ωi (24)
0 0 cc 0 ⎪ ẋc (t) ⎪

⎪ ⎪

0 − cz 0 ci + cz ⎪ ⎩ ẋi (t) ⎪

[ ( )]2 ⎡ * ⎤ ⎧ ⎫
∫∫
ksp sin 2hπz
dA k − k* 0 0 ⎪
⎪ x* (t) ⎪⎪
* *
⎢ − k k + kb + kz 0 ⎨ ⎬
kz = (25) − kz ⎥
⎥ xb (t)
[ ( )]2 +⎢ ⎣ 0 ⎦
0 kc 0 ⎪
⎪ x (t) ⎪
sin π2hhi ⎩ c ⎪ ⎭
0 − kz 0 ki + kz xi (t)
⎧ *⎫

⎪ m ⎪
E E ⎨ ⎪ ⎬
ksp = = √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ (25a) =−
mb
ẍg (t) (30)
Δ m ⎪
R0 − (Ri sin θ)2 − Ri cos θ
2 ⎪

⎩ c⎪ ⎭
mi
where ξ is the damping ratio of the insulation, Eurocode 8 [42] rec­ [ ] [ ]
ommends that the damping ratio of granular material is equal to 10%. mi Q = − m* ẍg (t) + ẍ* (t) − mb ẍg (t) + ẍb (t) − m0 ẍg (t)
and ωi are the mass and circular frequency of the liquid-solid coupling [ ] [ ] (31)
component, respectively, ksp is the elastic stiffness of the insulation, dA is − mi ẍg (t) + ẍi (t) − mc ẍg (t) + ẍc (t)
the micro element area, E is the elastic modulus of the insulation, Δ is the
spacing between the outer and inner tanks, R0 is the radius of the outer [ ] [ ]
tank, and Ri is the radius of the inner tank. M = − m* h ẍg (t) + ẍ* (t) − mb hb ẍg (t) + ẍb (t) − m0 h0 ẍg (t)
The seismic mechanical formula of LNG storage tank can be deduced [ ] [ ] (32)
according to the Hamilton principle which is introduced in Eq. (26), − mi hi ẍg (t) + ẍi (t) − mc hc ẍg (t) + ẍc (t)
∫ t2 ∫ t2
δ (T − V)dt + δWnc dt = 0 (26)
t1 t1

Fig. 19. Simplified mechanical model of LNG storage tanks.

9
D. Luo et al. Soil Dynamics and Earthquake Engineering 141 (2021) 106491

[ ]
Table 2
ẍg (t) + ẍc (t) Elements of LNG storage tank.
hs = 0.837R (33)
g Structural component Element Structural component Element

Inner tank Shell Insulation 3-D solid


4. Seismic response analysis of simplified mechanical model Outer tank 3-D solid Suspended ceiling Shell
Liquid 3-D fluid
4.1. Model validation
4.17. It can be concluded from Figs. 28 and 29 that the acceleration of
Reference [26] performed a numerical simulation analysis on 16 ×
the inner and outer tank walls decrease after the insulation is filled;
10 m3 LNG storage tanks via ADINA, referring to this method, the
4
especially in the middle position of the tank, which is in a good agree­
numerical simulation of the tank is carried out to verify the simplified
ment with the results of the simplified model.
mechanical model. Fig. 20 and Table 2 display the finite element model
The numerical simulation results in Fig. 30 show that the deforma­
and the element selection, respectively. Twenty ground motions are
tion of the outer tank wall under Qianan seismic wave decreases from
selected in this study, and the results are compared with those of the
4.29 mm to 3.98 mm, the displacements of the dome and sidewall in the
simplified model to verify its validity. Adjusting the peak ground ac­
simplified mechanical model decrease from 4.6 mm to 3.7 mm and 4.2
celeration (PGA) of the seismic motions to 4 m/s2, its response spectrum
mm–3.4 mm, respectively. The results of these two methods show that
curves are shown in Fig. 21.
the insulation layer can not effectively control the deformation of outer
The results in Figs. 22 and 23 show that the peak values of the base
tank wall. Fig. 31 shows that the insulation layer can well control the
shear and bending moment calculated by the two simplified mechanical
displacement in the middle position of the inner tank wall, which also
models are close, and the time history waveforms are similar.
confirms that it is reasonable to simplify the insulation as a spring­
Comparing the results of the simplified mechanical model 2 and nu­
–damper system connected between the sidewall and the liquid–solid
merical simulation, the base shear and bending moment are relatively
coupling component in the simplified model. The simulation result in­
close. It can be seen from Figs. 24 and 25 that the time-history wave­
dicates that the displacement of the inner tank decreases by more than
forms of the base shear and bending moment calculated by the two
70% on average after the insulation is filled, while the displacement of
methods are similar, and the peak values are relatively close. Moreover,
the liquid–solid coupling component in the simplified mechanical model
Fig. 26 shows that the peak values of the sloshing wave height calculated
reduces by 40%. Therefore, the insulation in the simulation model can
by the simplified model are smaller than the simulation results. Refer­
greatly limit the deformation of the inner tank, which is closer to the
ence [10] mentioned that the simplification of the liquid only consid­
actual situation. Generally speaking, the deformation of inner tank wall
ered the first-order sloshing, while the numerical simulation has the
calculated by the simplified model is conservative.
superposition of a multi-stage array; hence, the calculated wave height is
Fig. 32 displays the average damping rate of seismic response ob­
small. According to Fig. 27, the waveforms of these two methods are
tained via the simplified mechanical model and numerical simulation
similar.
under twenty seismic motions. The results demonstrate the following:
In the simplified model, the outer tank, dome, and liquid–solid
The damping rate of the seismic response calculated by the simplified
coupling components were simplified as the lumped mass point, the
mechanical model are larger than those by the numerical simulation
peak values of the acceleration and displacement of the points are
except for the displacement of the inner tank and the acceleration of the
analysed in this study. In the numerical simulation, the maximum ac­
outer tank. On the whole, the results of the two methods show that: after
celeration and displacement on the two layer tanks are analysed.
the 16 × 104 m3 LNG storage tank is filled with insulation, the base shear
Therefore, the acceleration and displacement of tank walls calculated by
and bending moment reduced by 20%–25%, and the acceleration of the
the two methods can not be directly compared. Here, the influence trend
outer and inner tank walls reduced by 10% and 35% on average,
of the insulation on the tank walls is compared and analysed.
respectively. The displacement of the inner tank wall reduced by more
From the numerical simulation results in Fig. 28, the maximum ac­
than 40%, but the displacement of the outer tank wall is amplified.
celeration of outer tank wall under Fsd seismic wave decreases from
Although the results obtained via the simplified mechanical model and
6.497 to 6.261 after filling the insulation, and the simplified mechanical
the numerical simulation method are different, they are close in nu­
model results show that the accelerations of the dome and sidewall
merical value and reflect the same trend.
decrease from 8.62 to 7.24 and 7.55 to 6.48, respectively. Besides,
Based on the above results, it can be considered that the base shear
Fig. 29 shows that the maximum acceleration of inner tank wall under
and bending moment calculated from numerical simulation and
artificial seismic wave 2 decreases from 10.27 to 7.379, and the accel­
simplified model are close, and the damping effect of insulation on
eration of the liquid-solid coupling component decreases from 7.56 to

Fig. 20. Numerical simulation model of LNG storage tank.

10
D. Luo et al. Soil Dynamics and Earthquake Engineering 141 (2021) 106491

Fig. 21. Acceleration response spectrum curve.

Fig. 22. Comparison of seismic response peak values between model 1 and
model 2. Fig. 24. Comparison of base shear and bending moment peak values between
model 2 and numerical simulation.

Fig. 23. Comparison of seismic response time history between model 1 and
model 2.

seismic response is consistent, hence, the model proposed in this paper is


Fig. 25. Comparison of base shear and bending moment time history between
verified to be reasonable. model 2 and numerical simulation.

4.2. Seismic response analysis of 16 × 104 m3 LNG storage tank The results in Tables 3–6 and Figs. 33 and 34 show that the base
shear and bending moment decrease after considering the insulation, for
The simplified mechanical model proposed in this study is used to which the damping rate is approximately 15% under the ground motions
calculate the seismic response of LNG storage tank, the results are shown in class I site condition and more than 25% under the other three types of
in Tables 3–14.

11
D. Luo et al. Soil Dynamics and Earthquake Engineering 141 (2021) 106491

the 16 × 104 m3 LNG storage tank, the sloshing period is approximately


10 s, which is a long period; thus, the sloshing will be very violent if the
tank subjected to seismic action with a long predominant period.
Reference [34] indicates that the type I site is relatively hard, and the
natural vibration period is short, which is beneficial to the seismic
resistance of long-period buildings. Type IV site belongs to a soft soil
site, which is unfavourable to the seismic resistance of long-period
buildings. Therefore, the class IV site ground motion causes a higher
wave height. In practical applications, violent sloshing can easily cause
buckling of the tank wall, and the structures attach to the upper part of
the tank could also be damaged. Moreover, the values of base shear and
bending moment calculated by M2WI are larger than those of M1. This
indicates that it is more conservative to use the results of M2WI as the
seismic design basis of the storage tank.
It can be seen from Tables 7–10 and Fig. 35 that the insulation
provides a good control action on the acceleration of the liquid-solid
coupling lumped mass point, but there is only a slight control action
on the sidewall and dome, and an increased acceleration of the sidewall
Fig. 26. Comparison of sloshing wave height peak values between model 2 and is observed under the class IV site seismic waves. The response trend is
numerical simulation. consistent with the results obtained from numerical simulation.
Considering the thin-walled structure used for the inner tank of the 16 ×
104 m3 LNG storage tank, which is less rigidity than the outer tank wall.
Hence, the impact or extrusion of liquid will easily lead to deformation
or even buckling of the inner tank wall under the action of earthquake,
the insulation layer will slow down this effect. Besides, the insulation is
lightweight with a low stiffness [43], it has little influence on the outer
tank.
The results in Tables 11–14 and Fig. 36 show that the displacement
of the liquid-solid coupling lumped mass point has a significant decrease
after considering the insulation, but there is only a slight control action
on the displacement of the sidewall and dome, even has an amplification
effect when the insulation is considered. These results are also similar to
the experimental test results and numerical simulation results. However,
the amplification ranges of tank wall displacement caused by insulation
layer is less than 1 mm, which has little impact on the safety of the tank.
Moreover, the displacement of the sidewall and dome decrease with the
increase of the site type, while the displacement of the liquid-solid
coupling lumped mass point increases with the increase of the site
type. The reason for these phenomena is the nature frequency of the
sidewall and dome is longer than that of inner tank, hence, the defor­
Fig. 27. Comparison of sloshing wave height time history between model 2 and mation of the outer tank is smaller under the seismic action with lower
numerical simulation. predominant frequency.
To sum up, the insulation layer plays a positive role in the seismic
site conditions. Moreover, there is no damping effect on the sloshing design of LNG storage tanks. In the seismic design of LNG storage tank,
wave height. In addition, the results in Tables 3–6 demonstrate that the considering the damping effect of insulation is helpful to save design
mean values of the base shear, bending moment, and sloshing wave capital, and ignoring the damping effect of insulation will make the
height increase with an increase in the site type. The results summarize design conservative, thus ensuring the safety of storage tanks to a
that the mean value of the sloshing wave height is smallest and largest greater extent. Moreover, the inner tank of large scale long period
under class I and class IV sites ground motion, respectively. This implies storage tank is relatively unsafe under class IV ground motion, while the
that the sloshing wave height is greatly affected by the type of site. For outer tank is relatively unsafe under class I ground motion. Since the

Fig. 28. Acceleration of outer tank wall under Fsd seismic wave.

12
D. Luo et al. Soil Dynamics and Earthquake Engineering 141 (2021) 106491

Fig. 29. Acceleration of inner tank wall under Artificial seismic wave 2.

Fig. 30. Displacement of outer tank wall under Qianan seismic wave.

Fig. 31. Displacement of inner tank wall under RH2 seismic wave.

Fig. 32. Average damping rate of the seismic responses.

13
D. Luo et al. Soil Dynamics and Earthquake Engineering 141 (2021) 106491

Table 3
Seismic response under class I site ground motion.
Seismic responses Working conditions Golden Gate Park Qianan Cmp_Cape Fsd_Santa Artificial seismic wave 1 Mean value
8
Base shear/10 N M1 2.07 2.17 1.28 1.61 2.74 1.97
M2WI 2.29 2.46 1.79 1.90 2.92 2.27
M2I 2.14 1.69 1.56 1.76 2.69 1.97
R1/% 6.55 31.30 12.85 7.37 7.88 13.19
Bending moment/109N⋅m M1 7.02 8.79 4.30 4.68 8.93 6.74
M2WI 6.37 7.62 4.11 4.76 7.99 6.17
M2I 5.45 4.53 3.56 4.35 6.62 4.90
R1/% 14.12 40.55 13.38 8.61 17.15 18.76
Sloshing wave height/m M1 0.31 0.07 0.15 0.55 0.51 0.32
M2WI 0.31 0.07 0.15 0.55 0.51 0.32
M2I 0.31 0.07 0.15 0.55 0.51 0.32
R1/% 0 0 0 0 0 0.00

Note: M1 refers to the Model 1; M2WI refers to the Model 2 without insulation; M2I refers to the Model 2 with insulation; R1 refers to the damping rate of the model 2
after the insulation is filled.

Table 4
Seismic response under class II site ground motion.
Seismic responses Working conditions El Centro RH2 TangShan Tar_Tarzana Artificial seismic wave 2 Mean value

Base shear/108N M1 5.67 2.01 5.41 2.05 3.09 3.65


M2WI 5.88 2.59 5.84 2.43 3.59 4.07
M2I 4.22 1.92 3.65 1.74 2.57 2.82
R1/% 28.23 25.87 37.50 28.40 28.41 29.68
Bending moment/109N⋅m M1 15.77 9.15 12.08 5.07 8.78 10.17
M2WI 15.95 7.09 14.34 5.91 9.14 10.49
M2I 10.58 5.01 8.75 4.66 6.33 7.07
R1/% 33.67 29.34 38.98 21.15 30.74 30.78
Sloshing wave height/m M1 0.76 0.21 0.62 0.21 0.51 0.46
M2WI 0.76 0.21 0.62 0.21 0.51 0.46
M2I 0.76 0.21 0.62 0.21 0.51 0.46
R1/% 0 0 0 0 0 0.00

Table 5
Seismic response under class III site ground motion.
Seismic responses Working conditions Taft PEL CPC_ TOPANGA EMC_ FAIRVIEW Artificial seismic wave 3 Mean value

Base shear/108N M1 3.27 1.62 4.22 2.62 4.57 3.26


M2WI 3.41 2.42 4.56 3.33 5.03 3.75
M2I 2.97 1.71 3.85 2.43 3.41 2.87
R1/% 12.90 29.34 15.57 27.03 32.21 23.41
Bending moment/109N⋅m M1 9.45 10.49 11.67 9.61 13.77 11.00
M2WI 9.08 6.40 11.83 9.26 13.20 9.95
M2I 7.82 4.44 9.87 6.68 8.76 7.51
R1/% 13.88 30.63 16.57 27.86 33.64 24.52
Sloshing wave height/m M1 0.58 0.33 0.25 0.17 1.27 0.52
M2WI 0.58 0.33 0.25 0.17 1.27 0.52
M2I 0.58 0.33 0.25 0.17 1.27 0.52
R1/% 0 0 0 0 0 0.00

Table 6
Seismic response under class IV site ground motion.
Seismic responses Working conditions Pasadena TRI TH4 TianJin Artificial seismic wave 4 Mean value

Base shear/108N M1 6.92 2.41 5.48 2.84 5.23 4.58


M2WI 6.99 3.62 5.88 3.07 5.52 5.02
M2I 4.03 3.27 4.22 2.92 2.93 3.47
R1/% 42.35 9.67 28.23 4.89 46.92 26.41
Bending moment/109N⋅m M1 18.59 7.27 14.97 8.08 15.60 12.90
M2WI 18.66 9.41 15.95 7.94 14.44 13.28
M2I 10.13 8.64 10.60 7.04 7.16 8.71
R1/% 45.71 8.18 33.54 11.34 50.42 29.84
Sloshing wave height/m M1 1.72 1.11 0.75 1.35 1.25 1.24
M2WI 1.72 1.11 0.75 1.35 1.25 1.24
M2I 1.72 1.11 0.75 1.35 1.25 1.24
R1/% 0 0 0 0 0 0.00

14
D. Luo et al. Soil Dynamics and Earthquake Engineering 141 (2021) 106491

Table 7
Peak values of the acceleration under class I site ground motion.
Seismic responses Working conditions Golden Gate Park Qianan Cmp_Cape Fsd_Santa Artificial seismic wave 1 Mean value
− 2
Acceleration of m*/m⋅s M2WI 9.60 8.29 6.60 8.62 6.91 8.00
M2I 8.03 6.39 5.58 7.24 6.56 6.76
R1/% 16.35 22.92 15.45 16.01 5.07 15.16
2
Acceleration of mb/m⋅s− M2WI 8.90 7.61 6.04 7.55 6.60 7.34
M2I 7.59 5.88 5.20 6.48 6.45 6.32
R1/% 14.72 22.73 13.91 14.17 2.27 13.56
2
Acceleration of mi/m⋅s− M2WI 2.63 4.24 2.22 3.35 7.45 3.98
M2I 1.85 2.44 1.25 1.74 4.73 2.40
R1/% 29.66 42.45 43.69 48.06 36.51 40.07

Table 8
Peak values of the acceleration under class II site ground motion.
Seismic responses Working conditions El Centro RH2 TangShan Tar_Tarzana Artificial seismic wave 2 Mean value
2
Acceleration of m*/m⋅s− M2WI 5.67 9.26 4.21 6.37 7.47 6.60
M2I 5.91 6.99 4.15 5.83 6.72 5.92
R1/% − 4.23 24.51 1.43 8.48 10.04 8.05
2
Acceleration of mb/m⋅s− M2WI 5.37 8.62 4.18 6.09 7.12 6.28
M2I 5.57 6.58 4.13 5.59 6.43 5.66
R1/% − 3.72 23.67 1.20 8.21 9.69 7.81
2
Acceleration of mi/m⋅s− M2WI 13.47 3.39 13.66 5.70 7.56 8.76
M2I 8.12 2.11 6.44 3.58 4.17 4.88
R1/% 39.72 37.76 52.86 37.19 44.84 42.47

Table 9
Peak values of the acceleration under class III site ground motion.
Seismic responses Working conditions Taft PEL CPC_ TOPANGA EMC_ FAIRVIEW Artificial seismic wave 3 Mean value
2
Acceleration of m*/m⋅s− M2WI 8.00 10.13 8.01 12.29 6.22 8.93
M2I 6.84 7.25 7.15 11.06 5.20 7.50
R1/% 14.50 28.43 10.74 10.01 16.40 16.02
2
Acceleration of mb/m⋅s− M2WI 7.67 9.34 7.50 11.51 5.89 8.38
M2I 6.60 6.70 6.92 11.11 5.96 7.46
R1/% 13.95 28.27 7.73 3.48 − 1.19 10.45
2
Acceleration of mi/m⋅s− M2WI 6.66 4.13 10.09 4.60 11.19 7.33
M2I 5.32 2.85 8.44 3.04 5.79 5.09
R1/% 20.12 30.99 16.35 33.91 48.26 29.93

Table 10
Peak values of the acceleration under class IV site ground motion.
Seismic responses Working conditions Pasadena TRI TH4 TianJin Artificial seismic wave 4 Mean value
2
Acceleration of m*/m⋅s− M2WI 4.76 11.24 5.57 3.98 5.01 6.11
M2I 4.65 9.59 5.91 3.95 5.07 5.83
R1/% 2.31 14.68 − 6.10 0.75 − 1.20 2.09
2
Acceleration of mb/m⋅s− M2WI 4.69 10.61 5.37 3.97 4.93 5.91
M2I 4.61 9.09 5.57 3.94 4.98 5.64
R1/% 1.71 14.33 − 3.72 0.76 − 1.01 2.41
2
Acceleration of mi/m⋅s− M2WI 18.48 5.18 13.47 7.71 11.62 11.29
M2I 9.10 5.02 8.12 5.13 5.58 6.59
R1/% 50.76 3.09 39.72 33.46 51.98 35.80

Table 11
Peak values of the displacement under class I site ground motion.
Seismic responses Working conditions Golden Gate Park Qianan Cmp_Cape Fsd_Santa Artificial seismic wave 1 Mean value

Displacement of m*/mm M2WI 5.9 4.6 4.0 5.1 4.3 4.8


M2I 4.8 3.7 3.9 4.7 4.0 4.2
R/% 18.64 19.57 2.50 7.84 6.98 11.11
Displacement of mb/mm M2WI 5.4 4.2 3.7 4.7 3.9 4.4
M2I 4.5 3.4 3.3 4.3 3.7 3.8
R/% 16.67 19.05 10.81 8.51 5.13 12.03
Displacement of mi/mm M2WI 20.5 32.7 17.4 26.0 57.9 30.9
M2I 14.2 16.7 8.4 12.7 35.3 17.5
R/% 30.73 48.93 51.72 51.15 39.03 44.31

15
D. Luo et al. Soil Dynamics and Earthquake Engineering 141 (2021) 106491

Table 12
Peak values of the displacement under class II site ground motion.
Seismic responses Working conditions El Centro RH2 TangShan Tar_Tarzana Artificial seismic wave 2 Mean value

Displacement of m*/mm M2WI 3.3 5.1 2.3 3.7 4.4 3.8


M2I 4.2 4.1 2.5 3.7 4.2 3.7
R/% − 27.27 19.61 − 8.70 0.00 4.55 − 2.36
Displacement of mb/mm M2WI 1.1 4.7 2.2 3.4 4.0 3.1
M2I 1.4 3.8 2.3 3.5 3.9 3.0
R/% − 27.27 19.15 − 4.55 − 2.94 2.50 − 2.62
Displacement of mi/mm M2WI 104.8 26.0 105.7 44.3 58.6 67.9
M2I 59.8 15.1 48.9 25.6 30.6 36.0
R/% 42.94 41.92 53.74 42.21 47.78 45.72

Table 13
Peak values of the displacement under class III site ground motion.
Seismic responses Working conditions Taft REL Cpc_ Topanga EMC_ FAIRVIEW Artificial seismic wave 3 Mean value

Displacement of m*/mm M2WI 4.7 5.6 4.8 7.2 3.6 5.2


M2I 4.4 4.2 4.6 6.8 3.5 4.7
R/% 6.38 25.00 4.17 5.56 2.78 8.78
Displacement of mb/mm M2WI 4.3 5.1 4.4 6.7 3.3 4.8
M2I 4.2 3.9 4.3 6.3 3.3 4.4
R/% 2.33 23.53 2.27 5.97 0.00 6.82
Displacement of mi/mm M2WI 51.9 31.7 78.6 36.2 86.9 57.1
M2I 39.3 13.9 63.3 20.7 43.7 36.2
R/% 24.28 56.15 19.47 42.82 49.71 38.49

Table 14
Peak values of the displacement under class IV site ground motion.
Seismic responses Working conditions Pasadena TRI TH4 TianJin Artificial seismic wave 4 Mean value

Displacement of m*/mm M2WI 2.7 6.2 3.3 2.2 2.8 3.4


M2I 3.5 5.5 3.6 2.5 3.1 3.6
R/% − 29.63 11.29 − 9.09 − 13.64 − 10.71 − 10.36
Displacement of mb/mm M2WI 2.5 5.7 3.1 2.0 2.6 3.2
M2I 3.3 5.1 3.2 2.2 3.1 3.4
R/% − 32.00 10.53 − 3.23 − 10.00 − 19.23 − 10.79
Displacement of mi/mm M2WI 143.1 49.9 104.8 60.3 114.0 94.4
M2I 68.5 38.3 60.1 38.5 87.5 58.6
R/% 52.13 23.25 42.65 36.15 23.25 35.49

Fig. 33. Time history of base shear. Fig. 34. Time history of bending moment.

inner tank is the main container for storing LNG, it is suggested that the 5. Conclusions
construction site of the storage tank should be located on class I or II sites
to ensure its safety. Based on the shaking table test results, a simplified mechanical
model of an LNG storage tank is proposed and verified by a numerical
simulation method. The conclusions are as follows:

16
D. Luo et al. Soil Dynamics and Earthquake Engineering 141 (2021) 106491

Fig. 35. Tine history of acceleration.

Fig. 36. Tine history of displacement.

(1) The test results reviled that the passive pressure of the insulation Jiangang Sun: Funding acquisition, Writing - original draft. Lifu Cui:
on the inner tank wall is the largest in the middle position of the Writing - review & editing. Zhen Wang: Writing - review & editing.
tank wall, which shows that there is a strong interaction between
the insulation layer and the liquid-solid coupling position of the
storage tank. The hydrodynamic pressure decreased after filling Declaration of competing interest
the insulation, which shows that the effect of insulation on tank
wall pressure should not be ignored. The authors declare that they have no known competing financial
(2) The simplified mechanical model, numerical simulation method, interests or personal relationships that could have appeared to influence
and shaking table test revealed the same seismic response trends. the work reported in this paper.
The base shear and bending moment calculated by both the
simplified mechanical model and the numerical simulation Acknowledgments
method are similar, the damping effect of insulation on the
seismic response of tanks is consistent, which verifies the validity This research work was supported by National Natural Science
of the mechanical model proposed in this paper. Foundation of China (51878124). The authors also would like to thank
(3) Based on the results of the simplified mechanical model of a 16 × Dr. Yuan Lyu from Dalian Maritime University for his help in the
104 m3 LNG storage tank, the base shear and bending moment research process of this project.
were reduced by 20%–25% on average, moreover, the accelera­
tion and displacement of the inner tank wall were reduced by Appendix A. Supplementary data
40% on average after filling the insulation. In the seismic design
of LNG storage tank, considering the damping effect of insulation Supplementary data to this article can be found online at https://doi.
is helpful to save design capital, and ignoring the damping effect org/10.1016/j.soildyn.2020.106491.
of insulation will make the design conservative, thus ensuring the
safety of storage tanks to a greater extent.
References
(4) The seismic response of test model is larger under the action of
ground motion with a higher predominate frequency, the seismic [1] Buris OS, di Filippo R, La Salandra V, Pedot M, Reza MS. Probabilistic seismic
response of 16 × 104 m3 LNG storage tank is larger under the analysis of an LNG subplant. J. Loss Prevent Proc. 2018;53:45–60. https://doi.org/
action of earthquake with a lower predominant frequency. In the 10.1016/j.jlp.2017.10.009.
[2] Roetzer J, Douglas H, Maurer H. Hazard and safety investigations for LNG-tanks.
seismic response analysis of the large scale long period tank, the LNG Journal 2005;15(6):72–89.
attention should be paid to the site effect on tank. [3] Peng Q, Wu H, Wang D, He Y, Chen H. Numerical simulation of aircraft crash on
large-scale LNG storage tank. Eng Fail Anal 2019;96:60–79. https://doi.org/
10.1016/j.engfailanal.2018.10.002.
CRediT authorship contribution statement [4] Fan H, Xu J, Wu S, Shi G, Guan W. LNG bunkering pontoons on inland waters in
China. Nat Gas Ind B 2018;5(2):148–55. https://doi.org/10.1016/j.
Dongyu Luo: Conceptualization, Formal analysis, Data curation, ngib.2018.04.001.
[5] Housner GW. Dynamic pressure on accelerated fluid containers. Bull Seismol Soc
Methodology, Writing - original draft, Writing - review & editing. Am 1957;47(1):15–35.
Chunguang Liu: Software, Formal analysis, Writing - review & editing. [6] Veletsos AS. Seismic effects in flexible liquid storage tanks. Rome, Italy: The 5th
World Conference on Earthquake Engineering; 1974.

17
D. Luo et al. Soil Dynamics and Earthquake Engineering 141 (2021) 106491

[7] Veletsos AS, Tang Y, Tang HT. Dynamic response of flexibly supported liquid- [25] Luo D, Sun J, Liu C, Cui L, Wang Z. Damping effect of insulation on a liquefied
storage tanks. J Struct Eng 1992;118(1):264–83. https://doi.org/10.1061/(ASCE) natural gas tank based on shaking table tests. J Earthq Eng 2020:1–21. https://doi.
0733-9445(1992)118:1(264). org/10.1080/13632469.2020.1774443.
[8] Haroun MA. Vibration studies and tests of liquid storage tanks. Earthq Eng Struct [26] Luo D, Sun J, Liu C, Cui L, Wang Z. Study on shock mitigation of insulation of LNG
Dynam 1983;11(2):179–206. https://doi.org/10.1002/eqe.4290110204. storage tank. J Vib Eng 2020;33(5):885–900.
[9] Haroun MA, Ellaithy HM. Model for flexible tanks undergoing rocking. J Eng Mech [27] Sun J, Yuan C, Hao J. Experimental research on rubber base-isolation tank.
1985;111(1):143–57. J Harbin Inst Technol 2005;37(6):806–10.
[10] Zhang R, Weng D, Ren X. Seismic analysis of a LNG storage tank isolated by a [28] De Angelis M, Giannini R, Paolacci F. Experimental investigation on the seismic
multiple friction pendulum system. Earthq Eng Eng Vib 2011;10(2):253–62. response of a steel liquid storage tank equipped with floating roof by shaking table
https://doi.org/10.1007/s11803-011-0063-3. tests. Earthq Eng Struct Dynam 2010;39(4):377–96. https://doi.org/10.1002/
[11] Zheng J, Sun J, Cui L, Wang X, Hao J. Seismic response of LNG storage tank with eqe.945.
base isolation considering pile-soil. Earthq Eng Eng Vib 2014;34:223–32. https:// [29] Cheng X, Jing W, Du Y, Bao C, Wu Z. Study on shock mitigation of concrete
doi.org/10.13197/j.eeev.2014.02.223.zhengjh.030. 02. rectangular liquid storage structure with sliding shock insulator and limiting
[12] Zhang C, Shan T, Fu Z, Xiao L. A large LNG tank technology system “CGTank®”of devices based on shaking table test. China Civ Eng J 2018;51(12):120–32. https://
CNOOC and its engineering application. Nat Gas Ind B 2015;2(6):530–4. https:// doi.org/10.15951/j.tmgcxb.2018.12.010.
doi.org/10.1016/j.ngib.2016.03.002. [30] Gb 50341. Code for design of vertical cylindrical welded steel oil tanks. Ministry of
[13] Martí J, Crespo M, Martínez F. Seismic isolation of liquefied natural gas tanks: a housing and urban-rural development of the people’s Republic of China. Beijing:
comparative assessment. Seismic Isolation and Protection Systems 2010;1(1): China Planning Press; 2014.
125–40. https://doi.org/10.2140/siaps.2010.1.125. [31] Xiong J. Research on the shaking table test model of PCS storage tank of AP1000
[14] Pedot M, di Filippo R, Bursi O S. Seismic vulnerability analysis of a liquefied nuclear power engineering. In: International conference on nuclear engineering,
natural gas subplant. ASME 2018 pressure vessels and piping conference. Prague, 57808. American Society of Mechanical Engineers; 2017. https://doi.org/
Czech Republic. 10.1115/ICONE25-67480.
[15] Uckan E, Umut Ö, Sisman FN, Karimzadeh S, Askan A. Seismic response of base [32] Zhang R, Weng D, Ge Q. Shaking table experiment on a steel storage tank with
isolated liquid storage tanks to real and simulated near fault pulse type ground multiple friction pendulum bearings. Struct Eng Mech 2014;52(5):875–87. https://
motions. Soil Dynam Earthq Eng 2018:58–68. https://doi.org/10.1016/j. doi.org/10.12989/sem.2014.52.5.875.
soildyn.2018.04.030. [33] Bahadori A. Thermal insulation handbook for the oil, gas, and petrochemical
[16] Lee DH, Kim MH, Kwon SH, Kim JW, Lee YB. A parametric sensitivity study on LNG industries. Houston: Gulf Professional Publishing; 2014.
tank sloshing loads by numerical simulations. Ocean Eng 2007;34(1):3–9. https:// [34] Gb50011. Code for seismic design of buildings. In: Ministry of housing and urban-
doi.org/10.1016/j.oceaneng.2006.03.014. rural development of the people’s Republic of China. Beijng, China: China
[17] Clough DP. Experimental evaluation of seismic design methods for broad Architecture & Building Press; 2010.
cylindrical tanks. Ph.D. Dissertation. Berkeley, California, America: University of [35] Shigapov R, Kovalchuk O. Investigation of seismic behaviour of low-temperature
California; 1977. tanks taking into account loose perlite insulation. In: IOP conference series:
[18] Hashemi S, Aghashiri MH. Seismic responses of base-isolated flexible rectangular materials science and engineering, 661. IOP Publishing; 2019, 012015.
fluid containers under horizontal ground motion. Soil Dynam Earthq Eng 2017: [36] Ormeño M, Larkin T, Chouw N. Experimental study of the effect of a flexible base
159–68. https://doi.org/10.1016/j.soildyn.2017.05.010. on the seismic response of a liquid storage tank. Thin-Walled Struct 2019;139:
[19] Compagnoni ME, Curadelli O, Ambrosini D. Experimental study on the seismic 334–46. https://doi.org/10.1016/j.tws.2019.03.013.
response of liquid storage tanks with sliding concave bearings. J. Loss Prevent Proc. [37] Radnić J, Grgić N, Kusić MS. Shake table testing of an open rectangular water tank
2018;55:1–9. https://doi.org/10.1016/j.jlp.2018.05.009. with water sloshing. J Fluid Struct 2018;81:97–115. https://doi.org/10.1016/j.
[20] Castellano MG, Infanti S, Dumoulin C. Shaking table tests on a liquefied natural gas jfluidstructs.2018.04.020.
storage tank mock-up seismically protected with elastomeric isolators and steel [38] Housner GW. The dynamic behavior of water tanks. Bull Seismol Soc Am 1963;53
hysteretic torsional dampers. Proc. of 12th WCEE 2000 [Paper No. 2082]. (2):381–7.
[21] Plannerer HN, Romero DM. Mechanical and physical properties of perlite from the [39] Malhotra P. Practical nonlinear seismic analysis of tanks. Earthq Spectra 2000;16
geophysical array for small-scale explosive experiments socorro. 1995. New (2):473–92. https://doi.org/10.1193/1.1586122.
Mexico. [40] Sobhan MS, Rofooei FR, Attari NK. Buckling behavior of the anchored steel tanks
[22] Thompson VK. Elastic buckling of double walled cylindrical storage tanks- under horizontal and vertical ground motions using static pushover and
numerical and analytical estimates. Comput Struct 1986;23(3):451–5. https://doi. incremental dynamic analyses. Thin-Walled Struct 2017;112:173–83. https://doi.
org/10.1016/0045-7949(86)90235-X. org/10.1016/j.tws.2016.12.022.
[23] Zhang R and Weng D. Assessment of the seismic effect of insulation on extra-large [41] Lin S. VCFPB base isolation and prestressed tendon construction research for LNG
cryogenic liquid natural gas storage tanks. J. Loss Prevent Proc. 2014;30:9–16. storage tank. Ph.D. Dissertation. Beijing: Beijing University of Technology; 2016.
https://doi.org/10.1016/j.jlp.2014.04.003. [42] Eurocode 8 EN 198-4. Design of structures for earthquake resistance - Part 4: silos,
[24] Chen S, Ma Z, Ren J, Liu F, Li Y, Ren Y. Analysis on the influence of thermal tanks, and pipelines. Brussels, Belgium: European committee for Standardization;
insulation layer on the dynamical characteristic of large-scale LNG storage tank. 2006.
Cryogenics&Superconductivity. 2015;43:36–42. 04. [43] Papa E, Medri V, Murri AN, Laghi L, De AG, Bandini S, Landi E. Characterization of
alkali bonded expanded perlite. Construct Build Mater 2018;191:1139–47. https://
doi.org/10.1016/j.conbuildmat.2018.10.086.

18

You might also like