You are on page 1of 17

Article

Study on Impact Process of a Large LNG Tank Container


for Trains
Zhiqiang Wang 1, Caifu Qian 1,* and Wei Li 2

1 College of Mechanical and Electrical Engineering, Beijing University of Chemical Technology,


Beijing 100029, China
2 Beijing Tianhai Cryogenic Equipment Co., Ltd., Beijing 101109, China

* Correspondence: qiancf@mail.buct.edu.cn

Abstract: In this paper, the impact process of a large LNG tank container for trains was studied by
performing experiments and numerical simulations. Impact force with induced stress and defor-
mation on the container especially on the frame was investigated and LNG sloshing inside the con-
tainer was simulated. Experimental results show that for the initial velocity of 6.1 km/h, the maxi-
mum compressive stress is −366.3 MPa occurring on the longitudinal beam near the impact side
corner fittings. The impact force produced by the transport vehicle is influenced by both the initial
clearance and initial velocity, i.e., its maximum value increases with the clearance or velocity, which
in turn directly affects the LNG impact force on the head, the tank container axial acceleration at the
mass center and the frame deformation and stress distribution. The largest average pressure
brought on by the LNG impact force is 8.83% of the design pressure, the inner vessel should be
designed with a thickness allowance. When the initial velocity is 8 km/h, the ratio of the maximum
LNG impact force to the static inertia force at each clearance is less than 0.23, which means that the
calculation method of LNG static inertia force is conservative. In addition, the maximum axial ac-
celeration of the tank container can reach 63 m/s 2, greater than 4g inertial acceleration specified in
the container design standard, meaning if assessed by the impact, the specifications of the standard
are not conservative.

Keywords: tank container; rail impact test; liquid sloshing; numerical simulation; stress distribution

Citation: Wang, Z.; Qian, C.; Li, W.


Study on Impact Process of a Large
1. Introduction
LNG Tank Container for Trains.
Appl. Sci. 2023, 13, 1351. Tank containers may frequently be impacted during railroad transit due to the oper-
https://doi.org/10.3390/app13031351 ation of the train connection. Under impacts, the tank container’s motion state changes,
causing slosh of the internal liquid, which in turn will have an effect on the tank container.
Academic Editor: Nicola Bosso
In extreme situations, the impact would cause significant deformation or possibly struc-
Received: 16 December 2022 tural strength breakdown, leading to container leakage or even catastrophic disaster.
Revised: 10 January 2023 The effect and sloshing of liquid-filled containers has been extensively researched. In
Accepted: 14 January 2023 order to describe the hydrodynamic damping provided by vertical and horizontal baffles
Published: 19 January 2023 in partially filled rectangular liquid tanks, Goudarzi et al. [1] conducted analytical solu-
tion approach and experimental investigations. In order to study the nonlinear behavior
and damping characteristics of liquid sloshing, Akyildiz et al. [2] used a numerical algo-
Copyright: © 2023 by the authors. rithm based on the volume of fluid (VOF) method to simulate pressure variations and
Licensee MDPI, Basel, Switzerland. three-dimensional effects on liquid sloshing loads in a moving partially filled rectangular
This article is an open access article tank. Anghileri et al. [3] studied the drop of liquid carrying water tank using ALE and
distributed under the terms and other algorithms, and they discovered that ALE algorithm may qualitatively represent the
conditions of the Creative Commons
test findings. Zhang et al. [4] examined the differences of various methods by setting up
Attribution (CC BY) license
a numerical simulation of the collision between liquid-filled containers and container
(https://creativecommons.org/license
ships and discovered that the ALE method is more rational and reduces computation
s/by/4.0/).
time. Aquelet et al. [5] investigated the working medium sloshing in the oil tank after it

Appl. Sci. 2023, 13, 1351. https://doi.org/10.3390/app13031351 www.mdpi.com/journal/applsci


Appl. Sci. 2023, 13, 1351 2 of 17

impacted the ground using the ALE method and contrasted the effects of adding an anti-
wave plate. In their study of the nonlinear oscillation in the container, Nasar et al. [6] dis-
covered that the excitation frequency has an impact on wave height. Liu et al. [7] studied
the liquid sloshing in the rectangular tank, and then verified the numerical simulation
results with experimental methods. In order to explore the liquid sloshing during braking,
Kang et al. [8] employed the VOF model. They also examined the stress on the tank con-
tainer under various liquid filling ratios K and various wave plate areas. The work of nu-
merical simulations regarding the dynamic behavior of the container under impact load-
ing was given by Cao et al. [9]. Ibrahim [10] discussed the nonlinear liquid sloshing dy-
namics under sway and rotational excitations for various container geometries, together
with the dimensionless factors influencing the construction of small-scale models, and ex-
amined the liquid sloshing evaluation of liquefied natural gas tanks. The critical baffle
height, at which liquid never reaches the tank’s ceiling at any point, was discovered by
Jung et al. [11] after they numerically examined the impact of the vertical baffle height on
the sloshing of liquid in a laterally moving 3D tank. Through a new approach based on
the coupling strategy of smoothed finite element method (SFEM) and an enhanced version
of smoothed particle hydrodynamics (SPH) delivering superior accuracy, Zhang et al. [12]
computationally evaluated the sloshing mitigation utilizing elastic baffles. The numerical
study by Sauer [13] compared two different numerical approaches that are both imple-
mented in a research hydrocode: a pure Lagrangian discretization with Finite Elements
(FE) and element erosion, and a coupled adaptive FE/SPH discretization. The study sim-
ulated projectile impacts on fluid-filled containers. In a partially filled laminated compo-
site container, the non-linear motion of the liquid free surface caused by sloshing and the
accompanying coupling caused by the fluid-structure interaction effect were the main top-
ics of Pal et al. [14]’s study. A straightforward equivalent mass-spring model developed
by Reed et al. [15] predicts the typical pulse duration and pressure distribution on the
vessel walls throughout the impact test. By extending the semi-analytical scaled boundary
finite element method (SBFEM), which makes use of the benefits of both the boundary
element (BEM) and finite element methods, Wang et al. [16] developed a method to study
the effect of various baffles on liquid oscillations in partially filled rigid toroidal tanks
(FEM). Tiernan et al. [17] studied the statically and dynamically finite element modeling
of the tank container for road and rail environments. A new modular tank and frame were
developed and constructed using the FEA results. In order to explore the non-linear dy-
namic behavior of partially filled tank vehicles under large-amplitude liquid sloshing, Dai
et al. [18] introduced a novel methodology. Analysis and comparison of the non-linear
dynamic behaviors of the tank vehicle subjected to liquid sloshing and the excitations pro-
duced by uneven roads was carried out. Li et al. [19] devised a numerical technique for
estimating the impact pressure based on the contact algorithm on the backdrop grid and
extended the material point method (MPM) to handle the dynamic behavior of liquid
sloshing in a moving container. To explore liquid sloshing with low liquid depth, Wu et
al. [20] carried out a series of tests encompassing the lowest three natural frequencies of
rolling coupled pitching. Kim [21] simulated the sloshing flow in two-dimensional and
three-dimensional liquid containers using the finite difference method. Experimental, an-
alytical and numerical studies of the perforation process were conducted by Bendarma et
al. [22] to investigate the ballistic characteristics of aluminum alloys at different tempera-
tures and dynamic loading. The results show that the energy absorbed during perforation
is quasiconstant for the studied range of velocities (up to 121 m/s) and using conical
shaped projectile, the average value was 26 J at room temperature and decreased to the
average of 18 J at 300 ˚C. Kpenyigba et al. [23] analyzed the process of perforation of steel
sheet based on different projectile nose shape, finding the ballistic limit and failure modes
for each kind of projectile. In addition, a non-linear increase in the ballistic limit with
thickness of the plate was noticed and a decrease in the numbers of petals with the conical
projectile angle was observed. Bendarma et al. [24] studied tensile tests and the process of
perforation of aluminum alloy AW5005 sheets based on experimental, analytical and
Appl. Sci. 2023, 13, 1351 3 of 17

numerical methods. In addition, different failure criteria were discussed, coupling numer-
ical and experimental analyses for a wide range of strain rates.
It was found from above review that few studies were addressed on the impact of
large LNG tank containers for trains. However, in engineering, the capacity to endure
railway impact has been considered as a design and quality control criterion for tank con-
tainers. In this paper, impact process of a large LNG tank container for trains was studied
by performing experiments and numerical simulations. Impact force with induced stress
and deformation on the container especially on the frame was investigated and LNG
sloshing inside the container was simulated.

2. Experiment Measurement
2.1. Structure of the Studied Tank Container
As shown in Figure 1, the inner and outer vessels, eight support rings, and frame
make up the majority of the tank container. Eight support rings link the inner vessel
(which holds the LNG) with the outer vessel, and the jacket is vacuumed to stop heat
transmission. The frame’s dimensions are 12192 mm × 2438 mm × 2591 mm (L × W × H),
and the volume is 45.75 m3 with a 2425 mm outer vessel diameter. Table 1 lists the main
design parameters of the tank container.

Figure 1. Schematic diagram of the tank container.

Table 1. Main design parameters of the tank container.

Item Value Item Value


Specified filling rate 90% Material of the 8 support rings GFRP
Design pressure of the inner vessel 0.6 MPa Material of the inner vessel S30408 (solution treatment)
Design temperature of the inner vessel −196 °C Material of the outer vessel 16MnDR (normalizing)
Design pressure of the jacket −0.1 MPa Material of the frame Q450NQR1 (normalizing)
Design temperature of the jacket 50 °C Corrosion allowance 0

2.2. Impact Strength Test


The impact strength test was conducted in accordance with the Chinese standard
TB/T1335-1996 Strength design and test certification specification for railway vehicle to
determine whether the frame meets with the strength requirements. Water, foam, and
rigid fixed supports were used as the filling medium, where the foam and rigid fixed sup-
ports were used to guarantee that the center of gravity of water was consistent with that
of LNG of the same mass. The impact vehicle with MT-2 buffers slid down the hill and
then hit the stationary, non-braking transport vehicle with the tank container for the test
and MT-2 buffers. The impact vehicle was driven three times at each of the following ve-
locities: 4 km/h, 5 km/h, 6 km/h, 7 km/h, and 8 km/h. The twist lock fixed to the transport
vehicle then collided with the corner fittings of the tank container after the impact vehicle
collided with the transport vehicle. The two vehicles eventually traveled in the direction
of the rail after being linked by buffers. The impact test scheme and the field test are shown
in Figure 2.
Appl. Sci. 2023, 13, 1351 4 of 17

Figure 2. The impact test: (a) scheme; (b) field tests.

To measure stress, several strain gauges were installed on the tank container, partic-
ularly on the frame. When the impact vehicle’s initial velocity is 6.1 km/h, it is discovered
that measuring point A28 on the longitudinal beam, near the corner fitting of the impact
side, as shown in Figure 3a, has the largest compressive stress, with a maximum value of
−366.3 MPa and a direction parallel to the impact direction. The variation in the stress at
A28 over time is seen in Figure 3b. It can be seen that the impact has a significant effect on
the frame, which lead to a peak compressive stress at A28. The impact vehicle’s collision
effect is continually weakened as a result of the continuous energy consumption of MT-2
buffers, and the stress fluctuation at A28 decreases gradually.

Figure 3. Measuring point A28: (a) location of A28; (b) the variation of stress at A28 over time.

3. Numerical Simulations
The finite element method was employed to simulate the above impact test using the
explicit dynamic software LS-DYNA (MPP, R10.1.0).

3.1. Basic Algorithm


For Eulerian formulation, the Eulerian mesh node changes its position during one
computational time step (mesh deforms) because of the loading. After the time step the
analysis stops and the following two approximations are performed:
(1) Mesh smoothing: all the nodes of the Eulerian mesh, that have been displaced due
to loading, are, moved to their original position;
(2) Advection: the internal variables (stresses, flow fields, velocity field) for all the
nodes that have been moved are recomputed (interpolated) so that they have the same
spatial distribution as before the mesh smoothing. In this way the mesh smoothing does
not affect the internal variable distribution.
The described procedure is being repeated for each time step of the analysis. The ALE
solving procedure is similar to Eulerian procedure. The only difference is the mesh
smoothing. In the Eulerian formulation the nodes are moved back to their original posi-
tions, while in the ALE formulation the positions of the moved nodes are calculated ac-
cording to the average distance to the neighboring nodes, meaning that the Eulerian mesh
is movable and deformable.
Appl. Sci. 2023, 13, 1351 5 of 17

The Null model algorithm was used to explain the fluid’s flow properties, while the
control equations of the ALE were used to represent the fluid. The related Gruesien equa-
tion of state is further attached for resolution. The following is the Gruesien equation of
state:

0
0C 2  1 + (1 −
a 
) −  2 
p=  2 2 
+ ( 0 + a  ) E
2 (1)
 2 3 
1 − ( S1 − 1) − S2  + 1 − S3 (  + 1)2 
 
where C is the intercept of the  s (shock wave velocity)-  p (particle velocity) curve (in
velocity units); E is internal energy per unit volume; S1, S2, and S3 are the unitless coeffi-
cients of the slope of the curve;  0 is the unitless Gruneisen coefficient; a is the unitless,
first order volume correction to  0 ; and,

= −1 (2)
0

In general, the compressibility of fluid can be ignored, that is,   0 is constant.


Therefore, Equation (1) is simplified as:

p =  0E (3)

where the relationship of  0 with volume expansion coefficient  , specific heat capac-
ity cV , density  and isothermal compression coefficient kT is:


0 = (4)
 cV kT
The fluid structure interaction was realized by the penalty function method. Accord-
ing to the relative displacement, node forces are applied to the structure and the fluid, as
shown in Figure 4, when the fluid particle passes through the structural element. Volume
Fraction Initialization (VFI) was used to establish the boundary of the initial fluid material.

Figure 4. Lagrange-Eluerian coupling diagram.


Appl. Sci. 2023, 13, 1351 6 of 17

3.2. Material Model


Figure 5 shows the nominal stress-nominal strain curve for the Q450NQR1, 16MnDR,
and S30408 materials.

Figure 5. Nominal stress-nominal strain curve: (a) Q450NQR1 [25]; (b) 16MnDR [26]; (c) S30408 [27].

S = (1 +  ) (5)

e = ln(1 +  ) (6)

where  is nominal stress;  is nominal strain; S is true stress; e is true strain. The true
stress-strain curve is obtained by transforming the nominal stress-strain curve using
Equations 5 and 6. For the yielding phase, the discontinuity of the yielding process is elim-
inated by selecting specific data points after the transformation [26].
Table 2 lists more structural material properties. Table 3 lists the material properties
of LNG with data referencing Refprop, a specialist program for cryogenic physical prop-
erties. The force-displacement curve was given to simulate the mechanical performance
of the buffer using the experimental data [28]. The MT-2 buffer is characterized by a non-
linear increase in force with increasing compression distance during loading, and very
little during unloading, as shown in Figure 6.

Table 2. More structural material properties.

Item Density (t/mm3) Elastic Modulus (MPa) Poisson’s Ratio


Impact vehicle 0.115 × 10−4 0.201×106 0.3
Transport vehicle 0.295 × 10−8 0.201×106 0.3
Twist lock 0.785 × 10−8 0.201×106 0.3
Corner fitting 0.785 × 10−8 0.201×106 0.3
GFRP 0.185 × 10−8 0.720×105 0.26

Table 3. Material properties of LNG.

Item Value Item Value


Temperature (°C) −161.87 cV (J/kg·k) 2056.3
Pressure (MPa) 0.1 kT (1/Pa) 2.22 × 10−9
Density (kg/m3) 460  (1/K) 0.00346
Sound velocity (m/s) 1341.3 0 1.648
Viscosity (MPa·s) 0.118 × 10−9
Appl. Sci. 2023, 13, 1351 7 of 17

Figure 6. The force-displacement curve of MT-2.

3.3. Modeling and Meshing


The impact vehicle and the transport vehicle were simplified to rigid cubes. The twist
lock was fixed to the transport vehicle. The interface between LNG and containers was
realized by the penalty function method. The geometry was modeled using LS-DYNA
and Ansys software as shown in Figure 7. Table 4 lists the mass of each model component.

Figure 7. Geometry of the impact vehicle and the transport vehicle created in LS-DYNA and Ansys.

Table 4. Mass of each model component.

Item Impact Vehicle Transport Vehicle Tank Container LNG (90%)


Mass(t) 92.0 23.0 12.9 18.9

The buffer was meshed using Spring elements (Combi165, 2 nodes), while other com-
ponents were meshed using solid elements (Solid164, 8 nodes) with sweep method. The
ELFORM = −1 and ELFORM = 11 algorithms were assigned to solids and fluids element,
respectively. Figure 8 shows the mesh model. There are 2516583 nodes and 1,954,231 ele-
ments that make up the tank container model. The fluid domain contains a total of 577,304
elements.
Appl. Sci. 2023, 13, 1351 8 of 17

Figure 8. Mesh model: (a) structure, (b) ALE element (partial).

3.4. Loading and Boundary Condition


The weight of the tank container and the LNG, as well as the internal and external
pressure (both 0.1 MPa), are all included in the load. Boundary conditions were set as
follows. (1) The impact and transport vehicle traveled along the rail only, constraining the
other directional translational degrees of freedom. (2) Friction contact was used with a
friction coefficient of 0.15 between the corner fitting and the transport vehicle as well as
between the support rings and vessels. The corner fitting and twist lock utilized automatic
surface-to-surface contact. The common node method was used for the other structures.
(3) The twist lock only established contact with the impact side corner fittings. The mo-
ment when the transport vehicle and the impact vehicle just collided is set to zero moment.
Different initial velocities were applied to the impact vehicle, and there are differences in
the impact direction initial clearance between the corner fitting and twist lock.

3.5. Model Verification


Take the initial velocity of 8 km/h and initial clearance of 6 mm as an example to
verify the independence of load step and mesh, as shown in Figures 9 and 10. Under dif-
ferent load steps, the maximum impact force at A (shown in Figure 10) and the force of
LNG on the impact side head changed by 1.6% and 3.0%, respectively, as shown in Figure
9. Taking into account the mass change, the final load step was decided to be 1.89 × 10−7.
In Figure 10, when the number of meshes was doubled, the maximum impact force at A
changed by 2%, the axial stress at C (i.e., A28) changed by 4.2%, and the change of maxi-
mum Mises stress at other locations was within 0.2%. Additionally, the maximum LNG
force changed by 4.6% when the number of fluid meshes in the contact region between
the fluid domain and the tank container (part of the entire fluid domain) was refined from
91,936 to 728,832. After the verification, both the final load step and mesh were decided.

Figure 9. Load step independence verification: (a) impact force of the transport vehicle on the tank
container, (b) impact force of the LNG on the impact side head.
Appl. Sci. 2023, 13, 1351 9 of 17

Figure 10. Mesh independence verification: (a) frame mesh, (b) fluid mesh in the contact region.

Only the inner vessel finite element model was built and filled with 50% of the volume
of LNG, constraining all of the translational degrees of freedom of the outer surface of the
inner vessel, in order to verify the reasonableness of the fluid-solid coupling contact parame-
ters. When the theoretical value of the axial force of LNG on the head is compared to the sim-
ulation, the error is only discovered to be 2.2%. After the verification, the contact parameters
were decided. The following formula was used to calculate the theoretical value:

F  =  pds (7)

where p is hydrostatic pressure; s is area.


The finite element model’s A28 and A07 locations are shown in Figure 11. Figure 12
shows the comparison between the experimental value and the simulated value of A28 and
A07. It is found that the simulated value well agrees with the experimental one in stress vari-
ation trend, indicating that the finite element model is reasonable.

(a) (b)
Figure 11. Location of A28 and A07 in the finite element model: (a) A28; (b) A07.

Figure 12. Comparison between experimental and simulated values: (a) A28 with the impact vehicle
velocity of 6 km/h; (b) A07 with the impact vehicle velocity of 8 km/h.
Appl. Sci. 2023, 13, 1351 10 of 17

4. Results and Discussion


In this part, further simulation results were analyzed and discussed. The direction of
the force involved in this section was along the impact direction, that is, along the rail
transportation direction. The force was set to be positive in the direction of impact and
negative in the opposite direction.

4.1. Impact Force


There are two kinds of impact forces: the first is the force of the twist lock on the
corner fitting (F1), and the second is the force of the LNG on the inner vessel (F2). Figure
13 shows the variation in F1 over time at different initial velocities and initial clearance.
As can be seen, when the initial velocity is 8 km/h, the twist lock repeatedly impacts the
corner fitting at each initial clearance, producing several peak forces, with the positive
peak values being significantly greater than the negative peak values. The maximum of F1
in direct relation to the initial clearance, increasing from 5.46 × 105 N at 0 mm to 9.41 × 105
N at 8 mm. However, its appearance time gradually moves from the second peak to the
first peak. In addition, when the initial clearance is 8 mm, the magnitude and number of
the peak of F1 vary at different initial velocities. The F1 significantly affects the axial stress
at A07, which is shown in Figure 14 that when the force increases or decreases, the stress
changes in magnitude accordingly.

Figure 13. The variation in F1 over time: (a) at different initial clearance; (b) at different initial veloc-
ities.

Figure 14. The variation of axial stress at A07: (a) at different initial clearance; (b) relationship with
the F1.

Figure 15 shows the relationship between the F1 and the relative location between the
twist lock and the corner fitting. When the relative locations are between −8 and 0 mm,
separation occurs, and when they are outside of that range, collision happens.
Appl. Sci. 2023, 13, 1351 11 of 17

Figure 15. Relationship between the F1 and the relative location between the twist lock and corner
fittings.

As can be seen from Figure 16, when the initial velocity is 8 km/h, the different initial
clearance led to different velocities when the transport vehicle just collides with the corner
fitting, which in turn affects the different magnitudes of the equal velocities and ultimately
causes the different magnitudes of the first peak of F1.

Figure 16. Relationship between the F1 and the velocity of the transport vehicle and tank container
(a) at 2 mm initial clearance; (b) at 6 mm initial clearance.

When the initial velocity is 8 km/h, Figure 17 shows the variation of the LNG impact
force with time on the impact side head (back head) at different initial clearance. It can be
seen that the back head is repeatedly impacted by the LNG at each initial clearance, re-
sulting in multiple peak forces (such as A–D), and the maximum increases with the in-
crease in initial clearance from −1.12×105 N at 0 mm to −2.10×105 N at 8 mm. In fact, the
variation in LNG impact force is brought on by the F1. However, although the F1 changes
sharply between positive and negative, the LNG impact force must remain negative. As
in magnitude, the former is much larger than the latter. Figure 17d shows the variation in
LNG free surface at different peak impact force times (split along the longitudinal sym-
metry plane). LNG separates from the front head and upwells along the back head as a
result of inertia. LNG starts to surge along the axial direction of the tank container and
upwell along the front head when it reaches the top of the inner vessel.
Appl. Sci. 2023, 13, 1351 12 of 17

Figure 17. The variation of LNG impact force: (a) at 0 and 2 mm initial clearance; (b) at 4 and 8 mm
initial clearance; (c) relationship with the F1; (d) LNG free surface at peak impact force times.

When the initial velocity is 8 km/h, Figure 18 shows the variation of the tank con-
tainer axial acceleration at the mass center over time at different initial clearance. The
maximum acceleration in direct relation to the initial clearance, from 35.4 m/s 2 at 0 mm to
63 m/s2 at 8 mm. It is noted that the maximum acceleration is greater than 4g inertial ac-
celeration specified in the Chinese standard NB/T 47059-2017 Tank containers for refrig-
erated liquefied gas for the design of tank containers, which means if assessed by the im-
pact, the specifications of the standard are not conservative. In addition, as observed at
locations A, B, and C in the Figure 18b, the acceleration has started to decrease before the
peak value of F1 as F2 gradually approaches to the peak value.

Figure 18. The variation of the tank container axial acceleration at the mass center over time: (a) at
different initial clearance; (b) relationship with the F1, F2 and friction force between corner fittings
and the transport vehicle.
Appl. Sci. 2023, 13, 1351 13 of 17

The average pressure pav brought on by the LNG impact force on the back head is:

Fmax
pav = (8)
 r2
where Fmax is the maximum LNG impact force; r is the inner radius of the inner vessel.
The static inertial force F  is:
F  = ma (9)

where m is LNG mass; a is the inertial acceleration. F  is calculated as 1.42 × 104 N when
the LNG filling percentage is 90%. Table 5 lists the results of the calculation of each varia-
ble at initial velocity of 8 km/h.

Table 5. The results of the calculation of each variable (8 km/h).

Clearance Acceleration Fmax Fmax


Fmax(N) pav (MPa) F  (N)
(mm) (m/s2) F F 
0 35.4 1.12 × 105 0.028 6.69 × 105 0.167 7.887
2 43.9 1.83 × 105 0.046 8.29 × 105 0.221 12.887
4 50.5 1.90 × 105 0.048 9.54 × 105 0.199 13.380
6 56.9 2.05 × 105 0.052 1.07 × 106 0.192 14.437
8 63.1 2.10 × 105 0.053 1.19 × 106 0.177 14.789

The data show that the largest average pressure is 0.053 MPa, or 8.83% of the 0.6 MPa
design pressure. The inner vessel should be designed with a thickness allowance to ac-
count for the influence of LNG impact force. When the initial velocity is 8 km/h, the max-
imum LNG impact force on the back head is much more than the hydrostatic force, and
the ratio of the maximum to the static inertia force at each clearance is less than 0.23, which
means that the calculation method of LNG static inertia force is conservative.

4.2. Influence of LNG Impact Force on Back Head


Figure 19 shows the variation of the back head maximum relative axial displacement
over time (relative to the tank container axial displacement at the mass center) at the initial
velocity of 8 km/h and the initial clearance of 6 mm. The maximum absolute value, 2.19
mm, occurs at the first peak moment of the impact force, and there is a one-to-one rela-
tionship between the peak impact force and the peak displacement, as does the maximum
Mises stress in Figure 20. The maximum value, 50.9 MPa, occurs at the first peak moment
of the impact force.

Figure 19. The variation of the back head maximum relative axial displacement over time: (a) rela-
tionship with the LNG impact force; (b) location of three peak displacements.
Appl. Sci. 2023, 13, 1351 14 of 17

Figure 20. The variation of the maximum Mises stress at the back head over time: (a) relationship
with the LNG impact force; (b) location of three peak stresses.

4.3. Influence of Impact force F1 on Frame


Figure 21 shows the variation of the maximum Mises stress of the frame over time. It
is clear that the maximum Mises stress has been significantly affected as a result of the
impact force F1. A, B, C, and D are the four stages that it could be divided into.
1. At stage A, there is no F1. Gravity has a significant role in determining the frame
Mises stress, which is in the elastic range.
2. Multiple impacts in stage B produce multiple stress peaks exceeding the yield
strength (450 MPa). Peak value 1 is produced by the F1 reaching its first peak value,
and as can be seen in the Figure 22, there are many locations with high stress. With
the exception of the chamfer, where residual stress causes high stress and a change
in the stress state from compression to tension, as shown in Figure 23, the stress in
other locations is low at peak value 2 in compared to the first peak. Comparable to
peak values 1 and 2, peak value 3 and 4 have similar characteristics.

Figure 21. The variation of the maximum Mises stress of the frame over time.

Figure 22. Stress distribution at peak value 1.


Appl. Sci. 2023, 13, 1351 15 of 17

Figure 23. Stress distribution at peak value 2: (a) location, (b) change of stress state in chamfer.

Figure 24 shows the distribution at the peak value 5, which is produced by the negative
peak impact force. The negative impact force intensifies the tensile stress state at the chamfer.
3. Although stage C do not have F1, the maximum Mises stress is larger than the maximum
Mises stress at stage A because of the residual stress at the chamfer. Stage D’s F1 is low
and has little effect on the stress distribution.

Figure 24. Stress distribution at peak value 5: (a) location; (b) change of stress state in chamfer.

Figure 25 shows the axial displacement of the frame and outer vessel at the first peak of
F1. It is clear that the frame maximum axial displacement occurs at the corner fitting on the
impact side and then gradually decreases in both the longitudinal and transverse directions.
The relative deformation of the A-A1, B-B1, and C-C1 varies in accordance with the F1 alternat-
ing positive and negative variations. At the first peak of F1, there is a 3.39 mm maximum rela-
tive deformation.

Figure 25. The axial displacement of the frame and outer vessel: (a) at the first peak of F1, (b) rela-
tionship with the F1.
Appl. Sci. 2023, 13, 1351 16 of 17

5. Conclusions
By performing the experiments and numerical simulations, impact process of a large
LNG tank container for trains was investigated in this study. Conclusions are obtained as
follows:
1. The impact process of a large LNG tank container for trains was conducted experi-
mentally and stresses on the container especially on the frame were measured with
time. For the initial velocity of 6.1 km/h, the maximum compressive stress is −366.3
MPa occurring on the longitudinal beam near the impact side corner fittings.
2. The impact force produced by the transport vehicle is influenced by both the initial
clearance and initial velocity, i.e., its maximum value increases with the clearance or
velocity, which in turn directly affects the LNG impact force on the head, the tank
container axial acceleration at the mass center and the frame deformation and stress
distribution. This result is valuable for the design of the corner fittings and frames of
the container.
3. The largest average pressure brought on by the LNG impact force is 0.053 MPa, or
8.83% of the 0.6 MPa design pressure, so the inner vessel should be designed with a
thickness allowance. When the initial velocity is 8 km/h, the ratio of the maximum
LNG impact force on the back head to the static inertia force at each clearance is less
than 0.23, which means that the calculation method of LNG static inertia force is con-
servative. This result may be helpful for the design of the thickness of container.
4. When the initial velocity is 8 km/h and the initial clearance is 8 mm, the maximum
axial acceleration of the tank container is 63 m/s2, greater than 4g inertial acceleration
specified in the container design standard, meaning if assessed by the impact, the
specifications of the standard are not conservative. This conclusion may provide ref-
erence for the editing or revision of the container design standard.

Author Contributions: Conceptualization, Z.W., C.Q. and W.L.; data curation, Z.W.; formal analy-
sis, Z.W.; investigation, Z.W., C.Q. and W.L.; methodology, Z.W., C.Q. and W.L.; project admin-
istration, W.L.; resources, C.Q. and W.L.; supervision, W.L.; validation, Z.W., C.Q. and W.L.; visu-
alization, Z.W.; writing—original draft, Z.W.; writing—review and editing, C.Q. and W.L. All au-
thors have read and agreed to the published version of the manuscript.
Funding: This research received no external funding.
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement: Not applicable.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Goudarzi, M.A.; Sabbagh-Yazdi, S.R.; Marx, W. Investigation of sloshing damping in baffled rectangular tanks subjected to the
dynamic excitation. Bull. Earthq. Eng. 2009, 8, 1055–1072. https://doi.org/10.1007/s10518-009-9168-8.
2. Akyildiz, H.; Uenal, N.E. Sloshing in a three-dimensional rectangular tank: Numerical simulation and experimental validation.
Ocean. Eng. 2006, 33, 2135–2149. https://doi.org/10.1016/j.oceaneng. 2005.11.001.
3. Anghileri, M.; Castelletti, L.; Tirelli, M. Fluid–structure interaction of water filled tanks during the impact with the ground. Int.
J. Impact Eng. 2005, 31, 235–254. https://doi.org/10.1016/j.ijimpeng.2003.12.005.
4. Zhang, A.N.; Suzuki, K. A comparative study of numerical simulations for fluid–structure interaction of liquid-filled tank dur-
ing ship collision. Ocean. Eng. 2007, 34, 645–652. https://doi.org/10.1016/j.oceaneng.2006.06.001.
5. Aquelet, N.; Souli, M.; Gabrys, J.; Olovson, L. A new ALE formualtion for sloshing analysis. Struct. Eng. Mech. 2003, 16, 423–
440. https://doi.org/10.12989/sem.2003.16.4.423.
6. Nasar, T.; Sannasiraj, S.A.; Sundar, V. Experimental study of liquid sloshing dynamics in a barge carrying tank. Fluid Dyn. Res.
2008, 40, 427. https://doi.org/10.1016/j.fluiddyn.2008.02.001.
7. Liu, D.M.; Lin, P.Z. A numerical study of three-dimensional liquid sloshing in tanks. J. Comput. Phys. 2008, 227, 3921–3939.
https://doi.org/10.1016/j.jcp.2007.12.006.
8. Kang, N.; Liu, K. Influence of baffle position on liquid sloshing, during braking and turning of a tank truck. J. Zhejiang Univ. A
2010, 11, 317–324. https://doi.org/10.1631/jzus.A0900521.
Appl. Sci. 2023, 13, 1351 17 of 17

9. Cao, Y.; Jin, X.L. Dynamic response of flexible container during the impact with the ground. Int. J. Impact Eng. 2010, 37, 999–
1007. https://doi.org/10.1016/j.ijimpeng.2010.05.001.
10. Ibrahim, R.A. Assessment of breaking waves and liquid sloshing impact. Nonlinear Dyn. 2020, 100, 1837–1925.
https://doi.org/10.1007/s11071-020-05605-7.
11. Jung, J.H.; Yoon, H.S.; Lee, C.Y.; Shin, S.C. Effect of the vertical baffle height on the liquid sloshing in a three-dimensional
rectangular tank. Ocean Eng. 2012, 44, 79–89. https://doi.org/10.1016/j.oceaneng.2012.01.034.
12. Zhang, Z.L.; Khalid, M.S.U.; Long, T.; Chang, J.Z.; Liu, M.B. Investigations on sloshing mitigation using elastic baffles by cou-
pling smoothed finite element method and decoupled finite particle method. J. Fluids Struct. 2020, 94, 102942.
https://doi.org/10.1016/j.jfluidstructs.2020.102942.
13. Sauer, M. Simulation of high velocity impact in fluid-filled containers using finite elements with adaptive coupling to smoothed
particle hydrodynamics. Int. J. Impact Eng. 2011, 38, 511–520. https://doi.org/10.1016/j.ijimpeng.2010.10.023.
14. Pal, N.C.; Bhattacharyya, S.; Sinha, P.K. Non-linear coupled slosh dynamics of liquid-filled laminated composite containers: A
two dimensional finite element approach. J. Sound Vib. 2003, 261, 729–749. https://doi.org/10.1016/S0022-460X(02)01011-8.
15. Reed, P.E.; Breedveld, G.; Lim, B.C. Simulation of the drop impact test for moulded thermoplastic containers. Int. J. Impact Eng.
2000, 24, 133–153. https://doi.org/10.1016/S0734-743X(99)00148-7.
16. Wang, W.Y.; Peng, Y.; Zhang, Q.; Ren, L.; Jiang, Y. Sloshing of liquid in partially liquid filled toroidal tank with various baffles
under lateral excitation. Ocean Eng. 2017, 146, 434–456. https://doi.org/10.1016/j.oceaneng.2017.09.032.
17. Tiernan, S.; Fahy, M. Dynamic FEA modelling of ISO tank containers. J. Mater. Process. Technol. 2002, 124, 126–132.
https://doi.org/10.1016/S0924-0136(02)00196-6.
18. Dai, L.; Xu, L.; Setiawan, B. A new non-linear approach to analysing the dynamic behaviour of tank vehicles subjected to liquid
sloshing. Proc. Inst. Mech. Eng. Part K J. Multi-body Dyn. 2005, 219, 75–86. https://doi.org/10.1243/146441905X9944.
19. Li, J.G.; Hamamoto, Y.; Liu, Y.; Zhang, X. Sloshing impact simulation with material point method and its experimental valida-
tions. Comput. Fluids 2014, 103, 86–99. https://doi.org/10.1016/j.compfluid.2014.07.025.
20. Wu, W.F.; Zhen, C.W.; Lu, J.S.; Tu, J.Y.; Zhang, J.W.; Yang, Y.B.; Zhu, K.B.; Duan, J.X. Experimental study on characteristic of
sloshing impact load in elastic tank with low and partial filling under rolling coupled pitching. Int. J. Nav. Arch. Ocean Eng. 2020,
12, 178–183. https://doi.org/10.1016/j.ijnaoe.2019.10.003.
21. Kim, Y. Numerical simulation of sloshing flows with impact load. Appl. Ocean Res. 2001, 23, 53–62. https://doi.org/10.1016/S0141-
1187(00)00021-3.
22. Bendarma, A.; Jankowiak, T.; Rusinek, A.; Lodygowski, T.; Klosak, M. Perforation Tests of Aluminum Alloy Specimens for a
Wide Range of Temperatures Using High-Performance Thermal Chamber-Experimental and Numerical Analysis. In IOP Con-
ference Series: Materials Science and Engineering; IOP Publishing: Bristol, UK, 2019; Volume 491, p. 012027.
23. Kpenyigba, K.M.; Jankowiak, T.; Rusinek, A.; Pesci, R. Influence of projectile shape on dynamic behavior of steel sheet subjected
to impact and perforation. Thin-Walled Struct. 2013, 65, 93–104. https://doi.org/10.1016/j.tws.2013.01.003.
24. Bendarma, A.; Jankowiak, T.; Łodygowski, T.; Rusinek, A.; Klósak, M. Experimental and numerical analysis of the aluminum
alloy AW5005 behavior subjected to tension and perforation under dynamic loading. J. Theor. Appl. Mech. 2017, 55, 1219–1233.
https://doi.org/10.15632/jtam-pl.55, 1219-1233.
25. Song, T.T. Research on Q450NQR1 High Strength Sheet Metal forming Springback and Application of Shallow Drawing. Di-
ploma Thesis, Shandong University, Jinan, China, 2011.
26. Pan, J.H.; Chen, Z.; Hong, Z.Y. A novel method to estimate the fracture toughness of pressure vessel ferritic steels in the ductile
to brittle transition region using finite element analysis and Master Curve method. Int. J. Press. Vessel. Pip. 2019, 176, 103949.
https://doi.org/10.1016/j.ijpvp.2019.103949.
27. Han, Y. Study on Technique of Cold Stretched Austenitic Stainless Steel Pressure Vessle and Its Performance Evaluation in
Typical Media Environment. Diploma Thesis, Hefei University of Technology, Hefei, China, 2012.
28. Mi, L. Research on Numerical Simulation of Impact Test on Railway Vehicles. Diploma Thesis, Beijing Jiaotong University,
Beijing, China, 2014.

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s)
and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or
property resulting from any ideas, methods, instructions or products referred to in the content.

You might also like