Proceedings of the Twenty-first (2011) International Offshore and Polar Engineering Conference

Maui, Hawaii, USA, June 19-24, 2011

Copyright 2011 by the International Society of Offshore and Polar Engineers (ISOPE)
ISBN 978-1-880653-96-8 (Set); ISSN 1098-6189 (Set); www.isope.org

Characteristics of Dynamic Response of Mark III LNG Containment Subjected to Idealized Triangular
Sloshing Impact
Mi-Ji Yoo(1),Sung-Je Lee(1), Sung-Chan Kim(3), Jang-Hyun Lee(2) and In-Sik Nho(4)
(1)
Dept. of Naval Architecture and Ocean Engineering, Inha University Graduate School, Incheon, Korea
(2)
Dept. of Naval Architecture and Ocean Engineering, Inha University, Incheon, Korea
(3)
Dept. of Ship and Ocean System, Inha Technical College, Incheon, Korea
(4)
Dept. of Naval Architecture and Ocean Engineering, Chungnam National University, Korea

ABSTRACT carriers presently operating with good experiences. The strength

Dynamic response of LNG cargo containment subjected to sloshing
impact is very complex because of high variability of sloshing load
and insulation materials. During the structural strength assessment of
the cargo containment in response to the sloshing impact, it requires
careful investigations to the structural behavior of the containment.
In this study, dynamic characteristics of Mark III type containment
subjected to the sloshing impact are investigated with the variation of
parameters in the pattern of sloshing load that are assumed from the
pressure history. Characteristics of dynamic structural responses to
various sloshing impacts are investigated in detail based on the results
of implicit-transient finite element analysis. The dynamic behavior of
containment are analyzed for the variation of duration time and
skewness, and the peak pressure which represents the sloshing load
assumed from the model test or computational fluid dynamic analysis.
An important observation is that although the maximum pressure and
the impulse are similar, the dynamic response of the containment
shows significant variations if the duration time and rising time is
different.
KEY WORDS: Sloshing, Impact Pressure, Duration time, Skewness,
Impulse, Transient Finite Element Analysis
INTRODUCTION
One of the important loads on the LNG tank cargo structure is caused
by sloshing. Sloshing is an impact load of an internal fluid motion in a
partially filled tank, which becomes critical when the LNG carrier is
excited with a frequency similar to the natural frequency of LNG flow.
In recent years, the size of LNG carrier cargo hold has been enlarged
because of the market requirement. This tendency can lead the higher
sloshing impact load than ever. Both strength assessment procedure of
LNG cargo containment and sloshing load estimation has been required
in order to support the guidance during the LNG CCS design.
Generally, two major approaches are provided for the sloshing
assessment and LNG cargo containment design procedure by shipping
registers. The first one is comparative approach, and the other one is
absolute approach. The comparative method is used for the standard
size tank or conventional designs that are similar to proven LNG
assessment of new design LNG carrier is to be compared with the
previously designed LNG carriers. However, this approach is not
reliable when a major design parameter changes significantly beyond
the proven design range. In this case, the absolute approach may be
selected.
Many attempts such as numerical, analytical, and model test
approaches have been tried to develop the sloshing assessment by
shipyards, shipping registers, and research scholars (Ito et al., 2008;
Maguire et al., 2009; Graczyk and Moan 2009). Nevertheless, it is very
difficult to expect the sloshing impact and structural response in
realistic way because of the complexity of the sloshing, uncertainties
relating to model test, scaling laws, local effects, and fluid-structure
interactions. The most realistic analysis might be FSI (Fluid-Structure
Interaction) analysis. Although the FSI analysis is expected to
investigate the hydro-elastic effect and fluid motion in the LNG CCS,
this approach still has some drawbacks. Due to limited knowledge of
hydro-elastic analysis, it is difficult to obtain accurate solutions through
the FSI analysis. The FSI method also require massive amount of
computation time. Therefore, current shipyard and shipping registers
practice for sloshing assessment for new LNG carrier is the
combination of model tests (Kuo et al., 2009; Lloyd Register, 2009),
stochastic post processing and the simplified absolute approach.
Pressure time histories are obtained from small scale model tests.
Thereafter, the measured pressure time histories are scaled into full
scale of real LNG cargo hold. The FE analysis is directly performed
with scaled sloshing load to assess the structural safety of LNG CCS
(ABS, 2006; Lloyd Register, 2009). The simplified absolute procedure
or direct assessment approach associated with model test can be applied
to new design or partial filling operations at this moment. Once the
numerical analysis is properly validated, numerical simulation based on
direct approach is likely to become the preferred tool for LNG sloshing
assessment, either in conjunction with model tests or eventually
independently (Zheng et al., 2010).
To compensate the uncertainties and variability in the absolute
assessment approach, optional levels of design procedures are
introduced in current guidance (Lloyds Register, 2009; ABS, 2006).
One of procedures is directed by linear dynamic structural analysis
performed by finite element analysis. The linear dynamic analysis
procedure is relatively simple for practical design guidance, but it

177
requires a linear transient structural response for triangular pressure
time history obtained by applying the scale law to the measured
pressure time history of model tests (Nho et al., 2010; Kuo et al., 2009;
Graczyk et al., 2008). Real time histories, however, are complex, non-
triangular shape. The time histories are characterized by the triangular
patterns. In order to use the time history for the linear transient analysis,
the structural response to triangular pressure has been recommended by
ABS (ABS, 2006; Kim et al., 2010).
This paper has performed several parametric studies in order to
investigate the dynamic response of structures subjected to sloshing
impact simplified by triangular pressure. Most of studies are focused on
the variation of duration time, peak pressure and skewness of the
triangular pressures. Transient response of the insulation panel is
analyzed based on the assumption that the triangular pressures are
already obtained through the model test and scale law processing. We
have studied the structural response for distributions of the peak
pressure magnitude, duration, skewness (ratio of rise time and duration
time), and ratio of duration and natural period of structures. As the
patterns of the triangular pressure vary, a difference between the
structural responses in the linear dynamic analysis is of main interest.
FINITE ELEMENT ANALYSIS MODEL OF
INSULATION SYSTEM
Insulation Model
In this study, Mark-III type insulation system is considered. The
insulation panel of Mark-III type CCS is composed of corrugated
membrane, plywood, R-PUF(Reinforced Polyurethane Foam), triplex
and mastic as shown in Fig. 1. It also consists of two layers of R-PUF
divided by triplex material and it is attached to the hull by mastics. The
panels are covered by a thin stainless steel membrane of 1.2 mm
thickness that has corrugations to absorb the sloshing loads and thermal
deformations. Inside the membrane panel, two-layered R-PUF is joined
by triplex while the triplex is inserted between the polyurethane foam
to prevent the gas leakage. Since the main interest of this study is to
investigate the dynamic response of R-PUF, plywood, and the highly
loaded regions around the mastic supports, insulation system without
triplex and membrane is considered for the finite element analysis.
Hence the primary membrane and triplex is ignored in the finite
element modeling.
Fig. 1 Schematic view of Mark III insulation panel
Material Properties
The material properties of each component are listed in Table 1. The
properties are selected from the design guidance suggested by ABS
(ABS, 2006). Although the effects of temperature dependent material
properties, material property non-linearity and strain rate should be
included in the structural analysis, R-PUF and mastic is assumed be
isotropic elasticity, and the plywood is assumed to be orthotropic
elasticity at room temperature, respectively. The plywood may be
modeled as a linear material with orthotropic properties.
178
Table 1. Material Properties of Mark III
Plywood R-PUF Mastic
Density (kg/m3) 710 120 1,500
Potssons Ratio 0.17 0.18 0.3
E1=E2=8,900
Youngs Modulus (MPa) 84 2,880
E3=520
G12=3,803
Shear Modulus (MPa) - -
G13=G23=196
Triangular pressures
The rising and decaying range of the triangular pressure was assumed
simply by the two lines as shown in Fig. 2. More accurate results can
be obtained for a selected pressure time histories approximated by a
trapezoid or realistic pressure than by a triangular function. This issue
was investigated by Graczyk and Moan (2006) in more detail. The
dynamic structural response can be characterized by the linear transient
dynamic equation, which is dependent on the load history and the
structural response is a function of the triangular load history, duration
time and ratio of the loading duration time to a natural period of the
structure. The applied impact load on the top surface of the upper
plywood layer is the triangular impact this is characterized in terms of
the skewness parameter of the impulse pressure.
Fig. 2 Triangular pressure impulse
Finite Element Model
Important layers of the insulation panel with respect to the dynamic
response are considered to the finite element model: two layers of
plywood with two layers of R-PUF between triplex. R-PUF and
mastics are modeled as isotropic materials, while plywood plate has
orthotropic properties. Fig. 3 shows the finite element models of 2-
dimensional (2D) and 3-dimensional (3D) shape simplified from Fig. 1.
The size of finite element model was determined to avoid the massive
amount of computation time. Also, the model dimensions and mesh
size are determined based on the patch test. The thickness of plywood,
R-PUF, and mastic is selected from the real insulation model. Since
this study is focused on the global dynamic response of CCS, the
corrugated membrane and triplex are excluded as mentioned in the
previous chapter. The horizontal length and depth of the panel in 2D
FEA is 1,250mm and 316 mm, respectively. The dimensions of 3D
FEA model are 300 mm x 500 mm x 316 mm, corresponding to an
assumed span of girders and stiffeners in two perpendicular directions,
respectively.
The elements type is determined based on the design guidance
suggested by ABS (2006) and Lloyd registers (2009). All layers in 2D
FEA and 3D FEA are modeled by 4-node plane strain elements and 8-
node hexahedral solid elements with full integration, respectively.
Through the mesh patch test, the size of element at critical region is
determined as 5mm x 5mm. To reduce the computation time, the mesh
size of non-critical area is coarsely modeled. 14,750 elements and
15,190 nodes are generated for 2D model, 68,880 elements and 75,515
nodes are used for 3D model. Implicit finite element analysis code
MSC.Nastran is used for FEA.
Fig. 3 2D and 3D finite element model simplified from Mark-III
insulation panel
Loading and boundary condition are applied to the FE model based on
the ABS guidance (ABS, 2006) as shown in Fig. 4. In the 2D model,
the bottoms of all mastics in the containment system are constrained in
all two directions. Two sides of the structure are constrained in the
normal directions. In the 3D model, the bottoms of all mastics are
constrained in all three directions. And x plane and y plane are
constrained at each normal direction. Applied load on the top surface of
the plywood is the triangular pressure with time history. Stresses and
displacements from linear transient FE analyses are obtained for the
simplified model. The time history of structural response at critical
locations is to be determined from the linear transient stress FE analysis.
Parametric studies are to be carried out to obtain the stress within the
defined range of pressure duration, skewness, and ratio to the natural
period of the simplified insulation panel model.
Fig. 4 Displacement boundary conditions and loading conditions
NUMERICAL RESUTLS
In order to investigate the dynamic response of insulation system
subjected to the triangular impulse, parametric numerical simulation
were performed within the pre-defined range of pressure duration,
skewness, and ratio to the natural period of the insulation panel
structure. In the stress FE analysis performed by implicit scheme of
MSC.Nastran, the time increment was small enough to get the reliable
solution. Output results such as stress, strain and displacement
components were obtained.
179
Effect of duration time on the stress
The failure load versus duration time is presented by sloshing
assessment guidance (ABS, 2006) as shown in Fig. 5. The duration
time of the guidance is longer than the practical range. However, in a
very short duration time, the drastic changes of stress components
could be taken place. This study assumes the duration shorter than
3.5mili sec. as listed in Table 2. Sloshing pressure time histories were
modeled by a triangular function of which skewness was fixed at unit
value. Deviations in the pressure duration time from a triangular shape
may have an important effect on the structural response when the
duration time is close to one of the natural frequencies of the structure
(Graczyk and Moan, 2006). Attention was also given to investigate the
response of FE models when the duration time is close to the natural
period of structure. The first order natural frequency was obtained by
0.0015 sec.
Fig. 5 Failure load ( Pfailure ) from the failure model at the critical
locations (ABS, 2006)
Table 2. Variations of duration time at unit skewness
Parameter Value
Duration Time 0.25~3.5 mili sec (Increment : 0.25)
t 10-6 (s)
Through the static FE analysis, the critical location where the
maximum stress takes place was defined. The critical location was
found at the connection between lower plywood and mastic. Since the
symmetry condition is applied to the 2D FE model, the same stress
takes place at mastics. Therefore, attention was given to the stress of
the mastic located in center of FE model. In the 3D FEA, maximum
value of stress components was found at different locations.
Thereafter, linear dynamic analysis was performed under the triangular
pressures. Fig 6 shows the maximum value of Horizontal,, Vertical and
Horizontal Vertical, which was obtained by static constant pressure, in the
lower plywood and von-Mises stress in the mastic at each duration time.
Stress components of 2D and 3D FE model showed the similar pattern
as shown in Fig. 6. Typically, the stress of 2D FE model is almost 20%
higher than that of 3D FE model. All the maximum stress components
were found when the duration time approaches the natural frequency.
The features of stress variation for the ratio of duration time (DT) to the
natural period (T) can be summarized as follow:
DT < T/3: It was very difficult to calculate the stresses only by
transient analysis since the duration time was too short to be
solved by the simple dynamic equation. It should have been
calculated by the formulation that could consider the shock wave
and strain rate dependent material properties.
T/3 < DT < T: The maximum stress was higher by 88% than that
of static analysis. Therefore, the duration time range of this case is
very important for strength assessment of insulation system.
 T<DT<2T: The maximum stress was 35% higher than that of
static analysis. The maximum stress was reduced as the duration
time was getting increased.
DT>2T: As the duration time was larger than two times of the
natural period, the maximum stress was getting slightly decreased.
It was naturally observed that the maximum stress was found when the
duration time approached the natural period of structure.
The critical pressure (Pcr) was also evaluated by Eq. 1. Allowable
strength ( a ) was selected from ABS design guidance (ABS, 2006). In
order to evaluate the meaningful parameter, the maximum stress at
duration time 0.0015 that gave the highest stress level was used to
assess the critical pressure. The estimated value of critical pressure is
summarized in Table 3. The compressive stress and shear stress at the
plywood are very low.
a
Pcr * P, a allowable stress (1)

Table 3. Critical pressure which exceeds the allowable strength

Critical Pressure
Stress Component
2D (kPa) 3D (kPa)
Plywood_Horizontal Tension 1,380 1,910
Plywood_Horizontal Compression 1,380 1,900
Plywood_Vertical Tension 220 230
Plywood_Vertical Compression 2,220 2,330
Plywood_Shearing 250 380
Mastic_Von-Mises 790 1,090

Fig. 6 Peak stress of critical points obtained by 2D and 3D FE model
subjected to the pressure of unit skewness
Fig. 7 shows the time history of xx, as the duration time varies. The
features of response can be summarized as follow:
DTT: After the pressure impulse was applied to the structure, no
damping effect was found in stress response. It was found that the
maximum tensile stress and compressive stress had almost same
magnitude.
T<DT<2T: After the pressure was activated, the amplitude of
stress was gradually decreased and disappeared. It was because
that the some damping effect was acted on the structure.
DT>2T: Although the impact pressure was activated, the
maximum stress is immediately disappeared.
It was very peculiar that the stress was drastically decreased when the
duration time was two times larger than the natural period.
Effect of different skewness under the same impulse
The skewness is defined by rate of rising time to decaying time, as
depicted in Eq. 2.
Although the insulation system is loaded by the triangular pressure with
same magnitude of peak pressure, the duration time, and impulse, the
different dynamic structural response can be taken place under the
different skewness.
Parametric studies on the variations in the skewness were conducted by
FE model under the same duration time, same peak pressure, and same
amount of impulse.
2 * RiseTime
S (2)
Duration
The skewness is assumed to vary from 0.2 to 2.0. When the skewness is
less than 1, it is a typical pattern of hydrodynamic impact of an
incompressible fluid. When gas is trapped between fluid and thee tank
wall during the impact, it is observed that the skewness is close to unit
value. If the skewness is larger than 1, it represents the non-
hydrodynamic impact due to gas trapping and escaping (ABS, 2006;
Lloyd Register 2009). Parametric cases are listed in Table 4.

Table 4. Parameter of analysis for the variation of skewness

Parameter Value
Duration Time 0.25~3.5 mili sec (Increment : 0.25)
t 10-6 (s)
S 0.2/0.5/1.0/1.5/2.0 (5 cases)

Fig. 8 shows the time history of xx in the plywood as the skewness

varies from 0.2 to 1.9. Peak pressure of 10kPa is applied to the FE
model, and duration time is 0.0015 sec that is same with the natural
period of structure. The feature of stress history can be summarized as
follow:
Fig. 7 History of xx for the pressures with different duration times As the skewness increases, the peak stress appears later. It is

180
 because that the peak pressure is activated later when the
skewness is increased.
If the skewness increases from 0.2 to 1.0, the maximum peak
stress is getting slightly decreased.
However, if the skewness is less than unit value then the local
maximum stress is getting decreased. When the skewness is larger
than unit value (1.0 ~ 1.9), the stress oscillates with same peak
value repeatedly.
In order to investigate the effect of skewness in detail, we varied the
duration at the same skewness as shown in Figs. 10~12. It shows the
history of peak stress under the different skewness and duration time
whilst the amount of impulse is same.
If the pressure of skewness 0.2, which has very short rising time,
is imposed, then the stress oscillates without damping effect.
When the skewness is 1.0 and the duration time is the double the
natural frequency, the stress is activated only during the duration
time and it disappears after the duration time.
Since the skewness 1.9 has very long rising time, the peak stress
occurs after the duration time. It is found that the magnitude of the
maximum stress is lower than that of other skewness.
It is peculiar that the magnitude of the peak stress is found when
the skewness is smaller than unit value even though several
studies insisted that the peak stress is magnified when the
skewness approaches unit value. This phenomenon should be
investigated in detail through experiments in the future.

Fig. 8 History of xx versus skewness (DT = Natural period)

Fig. 9 shows the history of xx in the plywood as the skewness varies.

The peak pressure is assumed to be 10 kPa, and duration time is 0.003
second that is the double natural period of structure.

As the skewness increases from 0.2 to 0.5, the peak stress appears
later. The stress is oscillated without damping effect. Although the
compressive pressure is applied, the stress oscillates from
compression to tension stress. After the duration time, very short
oscillation period is found.
If the skewness is around 1.0, the pressure force is effective during
only duration time and the stress is rapidly disappeared after the Fig. 10 History of xx with respect to duration time at S=0.2
duration time.
When the skewness is larger than 1.5, the stress responses very
slowly. However, very short oscillation of the stress is found after
duration time.

Fig. 11 History of xx with respect to duration time at S=1.0

Fig. 9 History of xx versus skewness (DT= 2 x Natural Period)

181

Fig. 12 History of xx with respect to duration time at S=1.9
CONCLUSIONS
In order to understand the qualitative response of Mark III type LNG
CCS, 2D and 3D FE models are conducted. We have performed
parametric studies in order to investigate the dynamic response of
structures subjected to sloshing impact simplified by triangular pressure.
Parametric studies are focused on the variation of duration time, peak
pressure and skewness of the triangular pressures. Transient response
of the insulation panel is analyzed by the transient structural analysis.
The history of peak stress is calculated by the FEA for variations of the
magnitude of peak pressure, duration time, skewness, and ratio of
duration and natural period of structures. Main works and conclusion
can be summarized as follow:
The stress history of LNG CCS under the idealized triangular
pressure is investigated by 2D and 3D FE model.
The history of peak stress for variations of the magnitude of peak
pressure, duration time, skewness, and ratio of duration and
natural period of structures is discussed.
Although the impact pressure had same amount of impulse, the
stress history showed different pattern due to the skewness.
When the duration time was the natural period of the insulation
structure, the peak stress oscillates regardless of the skewness.
When the skewness was very small, the peak stress was larger
than that of unit skewness.
The main interest of this study is the investigation of qualitative
dynamic response of the insulation panel under the triangular pressure.
Even though several parametric studies have been performed to
understand the feature of response under triangular pressure, it is
necessary to investigate the sensitivities or response surface of each
parameter on the stress.
ACKNOWLEDGEMENTS
This paper is a part of Development of Large Test Facility &
Infrastructure for Sloshing in LNG Cargo Containment System - Part
182
2: Development of Procedural method and Structural Safety
Assessment for LNG CC Design supported by the Ministry of
Knowledge and Economy, Korea. Financial support through the
Industrial Strategic Technology Development Program (Grant No.
10033791) is greatly acknowledged.
REFERENCES
ABS (2006). Guidance Note on Assessment of Membrane-type LNG
Containment Systems under Sloshing Loads.
Arswendt, A, Liasjoe, O, and Moan, T (2010). Comparative Study of FE
Models of LNG Containment System No.96, Proc 20th (2010) Int
Offshore and Polar Eng Conf, Beijing, China, ISOPE, Vol 3, pp 62-68
Brosset, L, Mravak, Z, Kaminski, M, Collins, S, and Finigan, T (2009).
Overview of Sloshel Project, Proc 19th (2009) Int Offshore and
Polar Eng Conf, Osaka, Japan, ISOPE, Vol 3, pp 115-124
Graczyk, M, Moan, T, and Rognebakke, O (2006). "Probabilistic analysis
of characteristic pressure for LNG tanks," J Offshore Mechanics and
Arctic Engineering-Transactions of the ASME, Vol 128, No 2, pp 133-
144.
Graczyk, M, and Moan, T (2008). A Probabblilistic Asssessment of
Design Sloshing Pressure Time Histories in LNG Tanks, J Ocean
Eng, Vol 35, pp 834-855.
Graczyk, M, and Moan, T (2009). Structural Response to Sloshing
Excitation in Membrane LNG Tank, Proc 28th (2009) Int Conf on
Ocean, Offshore and Arctic Eng, Honolulu, Hawaii, USA, Vol. 2,
pp 489-498
Ito, H, Suh, YS, Chun, SE, Satish Kumar, YV, Ha, MK, Park, JJ, Yu, HC,
and Wang, B (2008). A Direct Assessment Approach for Structural
Strength Evaluation of Cargo Containment System under Sloshing
Inside LNGC Tans based on Fluid Structure Interaction, Proc 27th
Int Conf on Offshore Mech and Arctic Eng, Estoril, Portugal,
Vol 5, pp 835-845
Kim, SY, Kim, YH, Ryu, MC, and Kim, YS (2010). Prescreening and
Statistical Evaluation of Sloshing-Induced Impact Pressure, Proc
Society of Naval Architects of Korea, pp 514-523
Kuo, JF, Campbell, RB, Ding, Z, Hoie, SM, Rinehart, AJ, Sandstorm, RE,
Yung, TW, Greer, MN, and Danaczko, MA (2009). LNG Tank
Sloshing Assessment Methodology-The New Generation, Proc 19th
(2009) Int Offshore and Polar Eng Conf, Osaka, Japan, ISOPE,
Vol 3, pp 1-12
Lloyd (2009). Ship Right - Sloshing Assessment Guidance Document
for Membrane Tank Operations.
Maguire, JR, Whiteworth, S, Oguibe, CN, Radosavljevic, D, and Carden,
EP (2009). Sloshing Dynamics-Numerical Simulations in Support of
the Sloshel Project, Proc 19th (2009) Int Offshore and Polar Eng
Conf, Osaka, Japan, ISOPE, Vol 3, pp 153-160
Nho, IS, Lee, JM, and Yeom, CW(2010). A Fundamental Study for
Time History Modeling of Fluid Impact Pressure, J Society of Naval
Architects of Korea, Vol 47, No 3, pp 244-249
Wang, B, and Shin, Y (2009). Full-Scale Sloshing Impact Test and
Coupled Fluid-Structure FE Modeling of LNG Containment System,
Proc 19th (2009) Int Offshore and Polar Eng Conf, Osaka, Japan,
ISOPE, Vol 3, pp 135-142
Zheng, X, Maguire, JR, and Radosavljevic, D (2010). Validation of
Numerical Tools for LNG Sloshing Assessment, Proc 20th(2010) Int
Offshore and Polar Eng Conf, Beijing, China, ISOPE, Vol 3, pp 119-
128