TECHNICAL ARTICLE—PEER-REVIEWED

Simulation and Experimental Work on a Thin-Walled StructureUnder Crushing

L. S. Lee • Aidy Ali • A. B. Sanuddin •

Reza Afshar

Submitted: 15 September 2009 / in revised form: 19 January 2010

� ASM International 2010

Abstract This study investigates a thin-walled cylindri-

cal structure subjected to impact loading. Both simulations

and experiments were performed to predict the deformation

mode and its deformation rate. In the simulation, a three-

dimensional finite element model of a soft drink can was

developed and post-processed using ANSYS and

LS-DYNA commercial packages, respectively. The

experimental work was carried out by dropping a flat

weight from a certain height to crush the specimen, using a

drink can to represent the thin-walled structure. A high-

speed camera was employed to capture the entire process

of the crush. The simulation and experiment showed strong

agreement.

Keywords Thin-walled structure � Crush � FEM �ANSYS � LS-DYNA

Introduction

A thin-walled cylinder can be defined as a cylinder with a

ratio of outer diameter to thickness of at least 1:20. Many

containment tanks, such as natural gas vehicle (NGV) tanks

and domestic gas tanks, can be classified as thin-walled,

although their shells seem very thick. In engineering

applications, a wide range of thin-walled cylinder shells to

sustain axial compressive loads are employed, particularly

in relation to buckling due to their ease in plastic defor-

mation [1]. The studies of the deformation mode can be

very helpful in designing such devices for safe operation.

To improve the absorbed energy by crushing, many

structural elements with various shapes as well as uniform

cross-sections have been investigated, such as tapered

tubes (frusta), double skin tubes, grooved tubes, tubes

stiffened by stringers and rings [2]. Ezra and Fay classified

energy-absorbing devices into three groups; each one relies

either on material deformation, extrusion or friction [3].

This project emphasizes the material deformation.

The crushing of thin-walled cylinders can normally be

evaluated in many ways ranging from quasi-static testing

[4–8], dynamic testing [4, 6, 8] and impact testing [5, 9].

For this project, an emphasis is given to impact testing in

evaluating the deformation mode of thin-walled cylinders.

In previous studies conducted by other researchers, many

factors that affect the crushworthiness of thin-walled cyl-

inders have been investigated. Examples of the factors are

the ratio of radius to thickness [10], height to thickness,

initial imperfection due to preliminary defects during

manufacturing and to handling problems [11] and liquid- or

sand-filled cylinders [12]. However, in this work, other

factors such as loading directions and the amount of liquid

contained in the cylinder are explored.

In the simulation work, the finite element method (FEM)

was used as a tool to analyze both small- and large-scale

deflection under loading or applied displacement. The

simulation is crucially required in order to reduce the

astronomical number of calculations that are needed to

analyze a large structure. It is evident that the finite element

analysis (FEA) is a way to deal with structures that are

more complex than can be described by analytical solutions

using classical theories [13, 14]. Therefore, the objectives

of this work are (1) to develop an impact model of a soft

drink can and (2) to study the crush behavior of modeled

soft drink can under impact loading.

L. S. Lee � A. Ali (&) � A. B. Sanuddin � R. Afshar

Department of Mechanical and Manufacturing Engineering,

Universiti Putra Malaysia, 43400

Serdang, Selangor, Malaysia

e-mail: aidy@eng.upm.edu.my

123

J Fail. Anal. and Preven.

DOI 10.1007/s11668-010-9330-2

Methodology

In this work, the main studies were carried out by applying

an impact load on thin-walled cylinder specimens and

analyzed through experiments and simulation. The simu-

lation of the models was developed using ANSYS and

analyzed with LS-DYNA. The steps taken in methodology

are shown in Fig. 1.

Material Selection

The specimen used in this study was a standard soft drink

can with material properties listed in Table 1 and 2.

The materials are 3004-H19 for the body and 5182-H19 for

the top lid. The detailed specimen geometry and its

dimensions are shown in Fig. 2.

Simulation Work

The finite element (FE) software program ANSYS was

used for pre-processing and processing, while LS-DYNA

was used for post-processing the drink can model.

Experimental Work

Simulation Work

Preparing the specimens

Experiment design

Experiment procedure

Measure specimen dimension

Simulation procedure

Simulating result

Comparing the experiment and simulation result

Discussion and conclusion

Load holder design and fabrication

Fig. 1 The methodology flow chart

Table 1 Mechanical properties of 3004-H19

Density, q1 2720 kg/m3

Ultimate tensile strength, rUTS,1 295 MPa

Yield strength, ry,1 285 MPa

Modulus of elasticity, E1 69 GPa

Elongation at break 2.00%

Poison ratio, m1 0.35

Table 2 Mechanical properties of 5182-H19

Density, q2 2650 kg/m3

Ultimate tensile strength, rUTS,2 420 MPa

Yield strength, ry,2 395 MPa

Modulus of elasticity, E2 69.6 GPa

Elongation at break 4.00%

Poison ratio, m2 0.33

Fig. 2 Geometry of specimen in mm

J Fail. Anal. and Preven.

123

Geometry Modeling

The simulation work using ANSYS began with modeling

the specimen, load and support; the geometry for these

models is shown in Fig. 2–4, respectively. The specimen

was created by extruding the lines after defining appro-

priate key points using by axis from the ANSYS guide user

interface. However, for the support and the load, it was

created by constructing the volume from the boxes option.

Figure 5 shows the FEM model drawn by ANSYS for the

load, specimen and support.

Mesh Generation

In this section, quadrilateral meshing elements were uti-

lized for specimen area mesh, while hexahedral meshing

elements were used in the volume mesh for the load and

support to avoid degenerate shell and solid element shapes.

There were 15,837 elements created in total; 117 elements

made up the load and support and the other 15,720 ele-

ments were the specimen. The specimen was the only

structure that underwent large strain deformation; hence, a

large number of elements was required in order to obtain a

precise and reliable result. Figure 6 shows the meshed

FEM model developed by ANSYS.

Boundary Conditions

The constraint was applied to the load in the X and Y

directions, while the support was constraint in all degrees

of freedom (DOF). Since this is an explicit analysis, con-

tact conditions are unpredictable. In this work, automatic

surface to surface contact was chosen and the program

automatically adjusted for the changes during the

simulation.

Experimental Procedures

The results obtained from experiment are the data on the

specimen deformation shape and the strain of the specimen

Fig. 3 Geometry of load in mm

Fig. 4 Geometry of support in mm

Fig. 5 FEM model developed by ANSYS

Fig. 6 Meshed FEM model developed by ANSYS

J Fail. Anal. and Preven.

123

at a certain time after collision. These data can be extracted

from the movie captured by using the camera software that

was installed on the computer. In experimental work, the

specimen was loaded under a drop weight impact loading

machine, as shown in Fig. 7. Then, a mass of 7.5 kg was

left to drop freely from a height of 7 feet above the

specimen. A load holder was designed in order to allow the

load to free fall when released at a certain height, as

depicted in Fig. 8. The load holder ensured that the load

was released at the exact height in every test made. A high-

speed camera with a capability to capture the picture at up

to 3000 frames per second was used in the experiment,

which means the pictures were taken approximately every

0.00033 s, thus allowing a better comparison with the

simulation results.

Results and Discussion

The results from the finite element model were compared

with the experiment. The comparison was made in terms of

the deformation mode and deformation rate of both results.

In addition, the energy absorbed by the specimen was also

observed from the FEM analysis, which used a thin-walled

structure in contact with high-velocity impact loading.

Deformation Mode

The deformation mode is related to how the specimen is

crushed. A small variation in time between the experiment

and simulation might occur because of the differences in

the time step between the experiment and the simulation. In

comparison, buckling waves were first created before the

crush, as shown in Fig. 9. In the experiment, the first

buckling waves were created at the time 0.0003333 s;

however, the obvious waves were created in simulation

falls around 0.00165 s. Another comparison, as in Fig. 10

and 11, shows different results, where in the simulation the

Load holder

Specimen

Load

Adjusting handle

High speed camera

Support

Fig. 7 Impact machine with high camera set up

Fig. 8 Load holder

J Fail. Anal. and Preven.

123

specimen collapsed first and continued at the base and the

lid. However, the specimen starts and continues to collapse

at the body in the experiment. The sole reason for the

difference is due to the assumption made in the modeling

of the specimen, where the thickness throughout the FEM

model is constant. In fact, the thickness is thicker at the

base and lid.

Figures 12–14 show the deformed specimen for both the

experiment and simulation at a certain time after crushing

by the drop load. These figures indicated that the

Buckling Waves

(a) (b)

Fig. 9 Buckling waves created

on the surface of the specimen

for (a) experiment and

(b) simulation

Does not Collapse Collapsed

Fig. 10 Condition of the

specimen in the experiment at

0.000667 s and in the

simulation at 0.000775 s

J Fail. Anal. and Preven.

123

simulation and experiment behave differently. One of the

known reasons for the discrepancy between the simulation

and experiment is the coefficients keyed into the simulation

option, namely, the static friction coefficient, dynamic

friction coefficient, exponential decay coefficient, coeffi-

cient for viscous friction and viscous damping coefficient.

Those coefficients were not able to be determined properly

because they can only be obtained by research and

experiment. Moreover, in the experiment, the air is con-

tained inside the specimen under atmospheric pressure.

When the drop load hits the specimen, the air inside the

specimen is compressed, thus leading to higher pressure.

Therefore, it provides strength to counteract the impact

force.

In another case, the simulation was similar to the

experiment in terms of the pattern of buckling waves

Does Not Collapse Collapsed

Fig. 11 Condition of the

specimen in the experiment at

0.001333 s and in the

simulation at 0.0015 s

Fig. 12 Condition of the

specimen for both the

experiment and simulation at

0.003 s

J Fail. Anal. and Preven.

123

produced but at different times. The similarities occurred at

0.000667 s for the experiment and 0.00225 s for the sim-

ulation. The similarities of the wave are shown by the color

circle drawn in Fig. 15.

Deformation Rate

Figure 16 shows the graph of the Y-displacement versus

time for both the experiment and simulation. In the graph,

the two results are close at the start but continuously sep-

arate as the time increases. This divergence pattern may be

due to the air that is contained inside the specimen in the

experiment but not in the simulation. As the air is further

compressed, the higher pressure counteracts the force

exerted on it. Hence, less deformation is produced in the

experiment compared to the simulation.

Energy Absorption

The thin-walled cylinder structure plays an important role in

energy absorption, especially in the automotive industry.

The energy absorbed can reduce the amount of force

Fig. 13 Condition of the

specimen for both the

experiment and simulation at

0.009 s

Fig. 14 Condition of the

specimen for both the

experiment and simulation at

0.015 s

Fig. 15 Similarity shown in

both experiment and simulation

J Fail. Anal. and Preven.

123

transmitted to the whole system during a crush. Figure 17

shows the amount of energy in the drop load during free fall.

From the graph, there is a total of 157.93 J of energy at the

release of the drop load, which is maintained just before it

hits the specimen. The first slanting line is due to small

buckling waves created on the slanted surface at the speci-

men, which further reduce the velocity of the load. The large

slanting line, which is created at time 0.00165 s, is due to the

obvious buckling waves created on the surface of the

specimen. The horizontal line after that is due to the defor-

mation of the specimen body. This pattern is further move on

alternately between buckling waves forming and deforma-

tion occurred. Hence, it can be said that the buckling waves

created have a greater effect on reducing the load velocity,

which directly reflects the higher kinetic energy being

absorbed. The total energy absorbed was 23.3 J.

Conclusions

A three-dimensional finite element analysis was used to

examine the crushing under high-velocity impact loading

of a drink can, and comparisons were made with experi-

mental results. These comparisons include the deformation

mode and the deformation rate. The simulation and

experiment did not yield the same result, but the results are

similar. Hence, a FEM used to predict the crushing of a

thin-walled structure was successfully developed. There-

fore, the present simulation work can be used to predict the

crushing of thin-walled structures of other materials, load

and geometry. Several improvements could be made to

improve the accuracy of the simulation. In the future, the

experimental method should use several cameras posi-

tioned at different angles to obtain a clearer overview, thus

making the comparisons easier. In the simulation, the

thickness of the FEM model should vary according to the

experimental specimen, which is thicker at the base and lid.

Coefficients in contact analysis should be obtained to yield

a higher accuracy in simulation, and finally, the air pressure

should be incorporated in the simulation.

Fig. 17 Graph of kinetic

energy versus time for the drop

load

Fig. 16 Y-displacement versus time for both experiment and

simulation

J Fail. Anal. and Preven.

123

Acknowledgment The authors would like to thank the Ministry of

Higher Education of Malaysia for their financial support.

References

1. Hambly, E.T., Calladine, C.R.: Buckling experiment on dam-

aged cylindrical shells. Int. J. Solids Struct. 33, 3539–3548

(1996)

2. Adachi, T., Tomiyama, A., Araki, W., Yamaji, A.: Energy

absorption of a thin-walled cylinder with ribs subjected to axial

impact. Int. J. Impact Eng. 35, 65–79 (2008)

3. Mamalis, A.G., Johnson, W.: The quasi-static crumpling of thin-

walled circular cylinders and frusta under axial compression. Int.

J. Mech. Sci. 25, 713–732 (1983)

4. Reddy, T.Y., Wall, R.J.: Axial compression of foam-filled

thin-walled circular tubes. Int. J. Impact Eng. 7, 151–166

(1988)

5. Gupta, N.K.: Some aspects of axial collapse of cylindrical thin-

walled tubes. Thin-Walled Struct. 32, 111–126 (1998)

6. Mirfendereski, L., Salimi, M., Ziaei-Rad, S.: Parametric study

and numerical analysis of empty and foam-filled thin-walled

tubes under static and dynamic loadings. Int. J. Mech. Sci. 50,

1042–1057 (2008)

7. Zhang, X.W., Yu, T.X.: Energy absorption of pressurized thin-

walled circular tubes under axial crushing. Int. J. Mech. Sci. 51,

335–349 (2009)

8. Kormi, K., Webb, D.C., Johnson, W.: The crash response of

circular tubes under general applied loading. Int. J. Impact Eng.

13, 243–257 (1993)

9. Ujihashi, S., Yamanaka, T., Kuroda, H., Inou, N.: Energy-

absorption abilities of CFRP cylinders during impact crushing.

Thin-Walled Struct. 28, 297–307 (1997)

10. Guillow, S.R., Lu, G., Grzebieta, R.H.: Quasi-static compression

of thin-walled circular aluminium tubes. Int. J. Mech. Sci. 43,

2103–2123 (2001)

11. Al Galib, D., Limam, A.: Experimental and numerical investi-

gation of static and dynamic axial crushing of circular aluminum

tubes. Thin-Walled Struct. 42, 1103–1137 (2004)

12. Nahas, M.N.: Impact energy dissipation characteristics of thin-

walled cylinders. Thin-Walled Struct. 15, 81–93 (1993)

13. Chan, O.B., Elwi, A.E., Gilbert, G.Y.: Simulation of crack

propagation in steel plate with strain softening model. Structural

Engineering Report No. 266 (2006)

14. Hertzberg, R.: Deformation and Fracture Mechanics of Engi-

neering Materials, 4th edn. Wiley, New York (1996)

J Fail. Anal. and Preven.

123